Assessing the Restoration Potential of a Remnant American Shad Spawning Run with Hatchery Supplementation Michael M

This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 19:56 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 To Stock or Not to Stock? Assessing the Restoration Potential of a Remnant American Shad Spawning Run with Hatchery Supplementation Michael M. Bailey a b & Joseph D. Zydlewski a c a Department of Wildlife Ecology , University of Maine , Orono , Maine , 04469-5755 , USA b U.S. Fish and Wildlife Service , Central New England Fishery Resources Office , Nashua , New Hampshire , 03063 , USA c U.S. Geological Survey, Maine Cooperative Fish and Wildlife Research Unit , University of Maine , Orono , Maine , 04469 , USA Published online: 29 Apr 2013.

To cite this article: Michael M. Bailey & Joseph D. Zydlewski (2013): To Stock or Not to Stock? Assessing the Restoration Potential of a Remnant American Shad Spawning Run with Hatchery Supplementation, North American Journal of Fisheries Management, 33:3, 459-467 To link to this article: http://dx.doi.org/10.1080/02755947.2013.763874

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:459–467, 2013 American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.763874

ARTICLE

To Stock or Not to Stock? Assessing the Restoration Potential of a Remnant American Shad Spawning Run with Hatchery Supplementation

Michael M. Bailey* Department of Wildlife Ecology, University of Maine, Orono, Maine 04469-5755, USA; and U.S. Fish and Wildlife Service, Central New England Fishery Resources Ofﬁce, Nashua, New Hampshire 03063, USA Joseph D. Zydlewski Department of Wildlife Ecology, University of Maine, Orono, Maine 04469-5755, USA; and U.S. Geological Survey, Maine Cooperative Fish and Wildlife Research Unit, University of Maine, Orono, Maine 04469, USA

Abstract Hatchery supplementation has been widely used as a restoration technique for American Shad Alosa sapidissima on the East Coast of the USA, but results have been equivocal. In the Penobscot River, Maine, dam removals and other improvements to fish passage will likely reestablish access to the majority of this species’ historic spawning habitat. Additional efforts being considered include the stocking of larval American Shad. The decision about whether to stock a river system undergoing restoration should be made after evaluating the probability of natural recolonization and examining the costs and benefits of potentially accelerating recovery using a stocking program. However, appropriate evaluation can be confounded by a dearth of information about the starting population size and age structure of the remnant American Shad spawning run in the river. We used the Penobscot River as a case study to assess the theoretical sensitivity of recovery time to either scenario (stocking or not) by building a deterministic model of an American Shad population. This model is based on the best available estimates of size at age, fecundity, rate of iteroparity, and recruitment. Density dependence was imposed, such that the population reached a plateau at an arbitrary recovery goal of 633,000 spawning adults. Stocking had a strong accelerating effect on the time to modeled recovery (as measured by the time to reach 50% of the recovery goal) in the base model, but stocking had diminishing effects with larger population sizes. There is a diminishing return to stocking when the starting population is modestly increased. With a low starting population (a spawning run of 1,000), supplementation with 12 million larvae annually Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 accelerated modeled recovery by 12 years. Only a 2-year acceleration was observed if the starting population was 15,000. Such a heuristic model may aid managers in assessing the costs and benefits of stocking by incorporating a structured decision framework.

Stocking is a commonly used tool in fisheries management into other populations, for example, can cause unintended con- (Molony et al. 2003), particularly for the restoration of anadro- sequences (Emlen 1991; Waples 1991) and may necessitate mous fish stocks (Moring 1986). The decision to use hatchery management to reduce the potential risks of unintended nat- products is often contentious. Stocking requires the allocation ural straying (Hayes and Carmichael 2002). Other risks and of resources, has uncertain benefits, and, in some cases, may pitfalls of starting new hatchery rearing operations are reviewed have deleterious effects (Schramm and Piper 1995). Straying in Molony et al. (2003). Common problems include (1) poorly

*Corresponding author: michael [email protected] Received December 27, 2011; accepted December 19, 2012 459 460 BAILEY AND ZYDLEWSKI

defined objectives, (2) lack of a rigorous scientific approach, Penobscot River watershed following project completion and (3) lack of clearly defined recovery indicators by which to (Trinko Lake et al. 2012). In addition to its role in the PRRP, cease stocking operations. the Maine Department of Marine Resources (MDMR) has been American Shad Alosa sapidissima is an anadromous clupeid proactive in developing an operational plan for the restoration of native to the East Coast of North America that spawns in rivers diadromous fishes to the Penobscot River. This long-term plan from the St. Johns River, Florida, to the St. Lawrence River, includes a conceptual framework for shad restoration (MDMR Quebec (Liem 1966; Collette and Klein-MacPhee 2002). Since 2009). the late 1800s, a suite of anthropogenic factors has caused a Although the Penobscot River historically supported Amer- decline in anadromous stocks in the genus Alosa along the East ican Shad spawning runs that may have numbered as high as 2 Coast (Bilkovic et al. 2002). Most notably, the construction of million fish prior to the 1830s (Foster and Atkins 1869), the cur- dams with inadequate fish passage has greatly reduced access rent spawning run is presumed to be nearly extirpated (ASMFC to spawning and nursery grounds (Rulifson 1994). Although 2007). Habitat loss is likely a major factor behind shad declines some fishways were constructed when the dams were built, most in the Penobscot River. Currently shad are restricted to habitat were recognized to be ineffective within a short time (Stevenson below Veazie Dam, with virtually no upstream passage being 1899). American Shad have become of increasing conservation available for more than 130 years. There is an extant spawn- concern in recent decades, as the number of spawning runs has ing run below Veazie Dam, but it is poorly characterized, as declined to fewer than half their historic levels (Limburg et al. there is no commercial fishery, targeted recreational fishing is 2003). minimal, and the fishway at Veazie Dam is not conducive to alo- Hatchery supplementation efforts for American Shad spawn- sine passage (Haro et al. 1999). Only about 25 shad have been ing runs date back more than a century, with the species first recorded to have passed the current fishway since its installation being reared and stocked in the Connecticut River in 1867. in 1970 (O. Cox, MDMR, unpublished data). Historically, shad Hatchery rearing was common in many large coastal rivers by accessed 145 km of main-stem habitat in the Penobscot system the turn of the century. By 1949, billions of American Shad (Collette and Klein-MacPhee 2002). Beginning in 1771, dams larvae had been stocked by hatcheries, but spawning runs con- excluded shad from historic spawning grounds, and by the time tinued to decline rangewide. This was likely due to continued Bangor Dam (rkm 41; removed in 1995) was complete in 1877, dam construction, deteriorating water quality, and unregulated the spawning run could no longer support a commercial fishery fisheries. Due to a perceived lack of efficacy in these programs, (ASMFC 2007). most were halted. The Pennsylvania Fish and Boat Commis- While there have been no studies to obtain estimates of run sion’s Van Dyke Hatchery was the first modern shad hatchery, size, the MDMR has conservatively estimated the spawning run starting operations in the mid-1970s (M. Hendricks, Pennsylva- to comprise as few as 1,000 adults. With the anticipated change nia Fish and Boat Commission, personal communication). By in habitat accessibility in the Penobscot River, an area-based the 1990s, other state and federal hatcheries had come online to model predicts the potential for a sustained run size of over produce shad larvae for restoration efforts from Maine to South 633,300 fish (MDMR 2009). As with many restoration efforts Carolina. As with earlier efforts, these hatchery efforts have had coastwide, the uncertainty of the current spawning run size and equivocal success. Stocking often accompanies other restoration the efficacy of the tools at hand confound the decision-making efforts, making assessment difficult or worse (Aunins 2010). process. Restoration options include annually stocking 6–12 Poor utilization of fishways, commercial in-river fisheries, and million larvae (reared from 500 to 1,200 adult American Shad incidental bycatch in ocean-intercept fisheries, are now seen as from donor stock) into the Penobscot River until a restoration

Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 the main culprits for diminished shad spawning runs, even when target of 633,300 spawners is realized for a period of five con- bolstered by hatchery supplementation (ASMFC 2007). secutive years. This option is being weighed against natural The Penobscot River Restoration Project (PRRP) is an ambi- recolonization of newly accessible habitat, though a significant tious cooperative effort (including Pennsylvania Power & Light delay in the timeline of restoration is feared. Co., the Penobscot Indian Nation, six conservation groups, and Until recently few tools were available with which to as- state and federal agencies) to restore 11 diadromous fish species sess the theoretical potential for American Shad expansion into to the Penobscot River, while maintaining hydroelectric power new habitat, and the available models may be unsuitable (but see production (Day 2006). Proposed restoration efforts include Harris and Hightower 2012). Based on a simple matrix model, if the removal of the two most seaward dams, Veazie and Great the spawning run is indeed as low as 1,000, reaching the restora- Works dams (river kilometers [rkm] 48 and 60, respectively), tion goal through natural recolonization may take more than a and modification of a third, Milford Dam (rkm 62), with im- century (MDMR 2009). This model, however, does not include proved fish passage. Further upstream, a “nature-type” bypass life history complexity (i.e., iteroparity) and specific survival channel around a fourth dam, Howland Dam (rkm 100), will and fecundity parameters that are likely important to shad pop- be installed. All fish passage improvements are planned to be ulation dynamics. This management dilemma, while specific to complete by 2014. It is anticipated that American Shad will the Penobscot River, is symptomatic of the quandary of shad have access to 93% of their historic spawning habitat in the restoration coastwide. To stock or not to stock? Managers must RESTORATION POTENTIAL OF AMERICAN SHAD SPAWNING RUN 461

balance risks and beneﬁts to make informed decisions; however, a framework with which to assess sensitivity to assumptions (e.g., of starting spawning run size) and describe best-case projections of hatchery supplementation has been lacking. We sought to develop a model that would be instructive in assessing these questions. Such an approach is relevant not only to the decision of when (or if) to use hatchery supplementation but also when to cease stocking operations. We applied our shad population recovery model to the Penobscot River as a case study to probe the impacts of alternative management actions. Additionally, we assessed the relative sensitivity of a population to stocking, initial run size, and at-sea mortality.

METHODS Model construction.—We reviewed the published literature and ASMFC publications for reproductive and survival estimates for American Shad. These data were then used in a pop- FIGURE 2. Spawning run age distribution of American Shad in the Exeter ulation model designed to conduct sensitivity analyses of the River, New Hampshire, and modeled Penobscot River population stabilized effects of the parameters commonly measured in population re- after 100 years averaged over 100 runs. covery research, both with and without stocking. We used Stella (High Performance Systems, Inc., Hanover, New Hampshire) at 633,000. The specific inputs for this model are described modeling software to construct a deterministic age-structured below. model. Data drawn primarily from ASMFC (2007) were used Size and fecundity.—As there has been virtually no mon- to define an age-structured population model with a maximum itoring of the remnant spawning run in the Penobscot River, age of 9 years that was reflective of iteroparous spawning runs we used size-at-age information for fish captured in the Exeter in the northern extent of the shad range (Figure 1). All of the River in New Hampshire (New Hampshire Fish and Game De- processes modeled were based on annual time steps from the partment, unpublished data; Figure 2) and the Merrimack River initiation of spawning. Age-0 individuals were recruited in a in Massachusetts (ASMFC 2007). The modeled lengths of fish density-dependent fashion. Specific life history input variables (L) at age a were based on a normal curve with an increas- included length at age, critical life stage recruitment relations, ing mean from age 4 to age 9 (47–62 cm; Table 1). Length juvenile survival, adult survival, and the size–fecundity rela- affected the model only through fecundity. No data specific to tionship. Our model had the following simplistic assumptions: the fecundity of American Shad in Maine are available. There- (1) sex ratios were equal and (2) all shad return to their natal fore, we followed ASMFC (2007) in assuming that most Maine rivers and no straying from other rivers contributes to the Penob- shad spawn between 20,000 and 150,000 eggs per female (Liem scot River spawning run. The stock–recruitment parameters of and Scott 1966), similar to Canadian stocks. Fecundity (F)was the model were adjusted such that the spawning run stabilized

TABLE 1. Values of recruitment, length, and fecundity used in the population Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 model. Recruitment data are for New England recruitment of female American Shad and are from the ASMFC. Recruitment to first spawning is the year-class- specific probability of first spawning based on cumulative recruitment. Cumula- tive recruitment includes first and repeat spawners for the stability model. Length data were estimated from New England rivers and were used to determine the fecundity range (conservatively based on Canadian stocks, as described in Liem and Scott [1966] using the relationship described in Methods).

Age Recruit to Cumulative Fork length (years) ﬁrst spawn recruitment (cm) Fecundity 4 0.02 0.2 47 20,654 FIGURE 1. Schematic of the deterministic age-structured model of American 5 0.23 0.25 52 34,674 Shad population recovery using the Penobscot River population as a case study. 6 0.48 0.61 57 58,210 The letter S represents survival and the letter R represents recruitment to the 7 0.64 0.86 60 79,433 spawning population. The components of recruitment to age 1, the mortality 8 0.71 0.96 61 88,480 rate at sea, postspawn survival, and spawner recruitment are explained in the 9 1.00 1.00 62 97,724 text. All ﬁsh are removed from the model at age 9. 462 BAILEY AND ZYDLEWSKI

calculated as of the exponential growth constant (r), which range from 0.15 to 0.45 for the spawning run (not the population), were based on F = 10mL+b, (1) estimates from stock assessments presented in ASMFC (2007). We used a conservative value of r = 0.15 for spawning run where m (the slope) and b (the intercept) were determined so growth to set the base model α value. This was done by removing as to conservatively set the range up to about 100,000 (20,654– density dependence (setting β equal to 0) and running the model 97,724). Using this relationship, egg production from wild ﬁsh for 100 years at a range of α values. The natural log–transformed was calculated as a function of individual fork lengths (Table 1). spawning run values were regressed against year, and the slope Fecundity was calculated under the assumptions that one-half of of the linear regression (r) was recorded. This was repeated the spawning run was female (i.e., that there was a 1:1 sex ratio), for 10 α values from 0.002 to 0.011 and the resulting relation that shad mature at age 4, and that older ﬁsh spawn every year (R2 = 0.997) was used to solve for α when r was set to 0.15 − after their initial spawning. The total number of wild-spawned (α = 3.232 × 10 3): eggs (Ew) was derived by summing over all of the age-classes that produced eggs: r = 0.148 log e (α) + 0.992. (4)

9 β was parameterized and selected by running the model with a Ew = [pN (eF ) · 0.5] , (2) − a a series of values (from 1 to 10 × 10 10), averaging the run size = a 4 from year 75 to year 100, and fitting the following curve (R2 = where e is the proportion of mature (spawning) adults (i.e., 0.9998): recruitment; ASMFC 2007; Table 1), Fa is the fecundity of an = . × −4 β−1.000. individual of length La at age a, Na is the total number of age-a Run Avg (3 464 10 ) (5) fish, and p is migratory success (the probability of successfully spawning once having entered the river). We arbitrarily set p to The base-model value of β value was selected such that run size 0.9 to account for in-river mortality and other factors that might stabilized at the management target of 633,000 spawning shad. lead to failure to spawn. This resulted in β = 5.4737 × 10−10. Total number of larvae.—The total number of larvae was Mortality.—The mortality associated with the model includes derived from three components: (1) the number of eggs from “at-sea” natural mortality and “acute postspawn” mortality. At natural reproduction based on the fecundity of wild individuals each step of the model, all nonreproductive fish (ages 1–8) (Ew from equation 2), (2) the hatchability rate of eggs from incur a constant natural mortality (M) of 0.38, as determined natural reproduction (h; our starting assumption was that egg-to- by Hoenig’s methods (ASMFC 2007), such that survival was larval survival was 10%), and (3) the number of larvae produced calculated as from hatchery eggs (EH; included in stock recruitment without an additional hatchability or mortality factor). In this model, the −Mt N + = N e , (6) number of hatchery larvae is equal to the number of hatchery (t 1) t eggs, as stocking numbers are based on live larvae released. Stock–recruitment.—A density-dependent curve for alosines where N is population size and t is time. The maximum age has not been well documented. In the absence of a more ap- in the model is 9 (i.e., all fish attaining age 9 die). Iteroparity was included in this model by allowing spawning fish to spawn

Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 propriate relationship, we used a Ricker stock–recruitment relationship from spawned eggs to age-1 subadults. Recruitment and, for those that survived acute postspawning mortality, to from total number of larvae to age-0 juveniles was modeled by return to the ocean. For northern rivers the species is increas- using a Ricker (1975) stock–recruitment relationship, ingly iteroparous with latitude (Chittenden 1976), though this is poorly quantiﬁed in many rivers, including the Penobscot

−β(hEw +E ) River. Iteroparity has been described in the synthesis paper of R = α(hEw + E )e H , (3) H Leggett and Carscadden (1978), which reported clinal variation in spawning among populations. We used the reported data to where R is the recruitment of age-0 ﬁsh, α and β are parameters regress the incidence of repeat spawning (I) against latitude, determining the shape of the stock–recruitment relationship, and providing the relationship (R2 = 0.76) hEw + EH is the total number of larvae. In this relation, the value of α determines the rate of increase in recruitment while that = . − . of β (the capacity parameter) determines the strength of density I 5 08(Latitude) 165 (7) dependence resulting in a leveling off of the population with increased abundance. While recruitment is poorly characterized Given the latitude of the Penobscot River (44.5◦N), this relation- for American Shad, there are data with which to describe the ship predicts a 61% rate of iteroparity. Because the model was rate of spawning run increases in recovering stocks. The values not individual based, iteroparity was calculated as a summation RESTORATION POTENTIAL OF AMERICAN SHAD SPAWNING RUN 463

of probabilities of repeat spawning, i.e.,

9 600 I = (spawning at age i + n|spawned at age i)/ i=4 9 400 (spawning at age i). (8) 50% of maximum i=4

We forced all spawners to spawn in all successive years. The 200 degree of iteroparity was therefore controlled by survival in the Spawning Run (Number x 1000) Run (Number x Spawning year after spawning. Acute postspawn survival (Ss)wassetat 70%. Survival to agei+1 from spawning at agei is represented by ΔT the product of acute postspawn survival and at-sea survival for 0 0 20406080 9 months, namely, Years (−M 0.75) Si to i+1 = Ss e . (9) FIGURE 3. Example of modeled American Shad spawning run size increase over time assuming instantaneous access to habitat made available through The result is a calculated iteroparity rate of 48%. This is about planned restoration on the Penobscot River. The plots represent 100 averaged 80% of what was predicted by Leggett and Carscadden (1978). runs of the base model (solid line; e.g., a starting run size of 1,000) and the model with a shifted parameter (dashed line; e.g., a starting run size of 20,000). Model execution.—The model censuses the existing popu- The calculated shift in reaching 50% of the maximum spawning run size is lation in each age-class and calculates the required summary indicated by the dotted lines. statistics, including run size, population size, contributions of each age-class, total fecundity, and total larval production. At the next iteration, annual at-sea mortality and postspawning mor- parameter, and Pn is the nominal parameter (Haefner 2005). talities are incurred. All surviving individuals of each year-class Parameters were considered “highly sensitive” if |S| > 1.00. graduate to the next year-class as reproductive or nonreproduc- Effects of starting population and stocking.—In our model, tive individuals based on age and the probability of previous neither the starting population size nor hatchery stocking in- spawning. Starting numbers for each age-class at sea and as fluenced the stabilized population level. In order to assess the spawners were determined by running the base model using the sensitivity of modeled American Shad recovery to changes, in values to generate proportional representation within the popu- both initial run populations and hatchery stocking we used the lation. The base model was run using the parameters described change in the time to attain 50% of the maximum value as a above and a starting run size of 1,000 spawning American Shad. metric. As above, the model stabilized at an average spawning Exeter River data.—Data from the Exeter River, New Hamp- run of 633,000. Therefore, we used the year in which the model shire, were used to compare the age structure of the stabilized surpassed 316,500 spawning shad (Figure 3). The precise value model with data from a New England river with a small run (fraction of year) was calculated via linear regression of the of American Shad captured at a fish trapping facility (18–163 points that bracketed this value. annual run from 1995 to 2004, for a total of 529 fish aged; New To assess the impact of hatchery supplementation, we mod- Hampshire Fish and Game, unpublished data). The age distribu- eled the annual stocking of between 0 and 48 million American tion generated by the model and the age distribution (average) Shad larvae annually. This was done using the base model with Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 from the Exeter River were compared using a Kolmogorov– 1,000 shad in the spawning run. To assess the impact of starting Smirnov two-sample test. run size, we modeled the time to 50% recovery over a range of Local sensitivity.—The local sensitivity of two variables— starting run sizes (1,000–300,000) without stocking. Similarly, the modeled spawning run size and the estimated time to at- the model was run using a stocking rate of 12 million larvae tainment of 50% of the maximum population level in the base annually over the same range of starting populations. model—to the estimated life history parameters was evaluated. The parameters included stock–recruitment parameters, fecundity parameters, and survival values. Changes in run size and RESULTS time to 50% recovery were evaluated after a 1% increase in life Base Model Run history parameters. Sensitivity (S) was defined as Using the inputs from the base model, the adult spawn-

(Ra − Rn)/Rn ing distribution in this system was dominated by ﬁsh of ages S = , (10) 5–7, with very few above age 7 (Table 1; Figure 2). The age (Pa − Pn)/Pn distribution generated by the model and the age distribution where Ra is the model result for the altered parameter, Rn is (average) from the Exeter River were not signiﬁcantly different the model result for the unaltered parameter, Pa is the altered (Kolmogorov–Smirnov two sample test; P = 0.078). 464 BAILEY AND ZYDLEWSKI

TABLE 2. Sensitivity (S) to model parameters of the modeled American Shad stabilized population level and the rate of attainment of 50% of the target population level. Parameters include α (which determines the rate of increase in recruitment), β (which determines the strength of density dependence), h (hatch success), m (the slope of the fecundity relationship), b (the intercept of the fecundity relationship), the at-sea mortality rate, and acute postspawn survival. Values of |S| > 1.00 are indicated by bold italics.

Parameter Nominal value S of stabilized run size S of time to 50% α 0.003232 0.97 −0.99 β 5.2043 × 10−10 −0.99 0.11 h 0.1 −0.03 −0.88 m 0.0045 −0.34 −5.26 b 2.2 −0.26 −4.42 At-sea mortality rate 0.38 −1.98 2.35 Acute postspawn survival 0.7 0.75 −0.93

The stabilized run size and time to 50% recovery from this more rapid recovery, with diminishing effects at higher levels of model were predictably sensitive to the stock–recruitment pa- supplementation. With a starting spawning run of 1,000, the an- rameters α and β (Table 2). Run size was highly sensitive to nual supplementation of 3 million larvae accelerated the time to changes in β, while this parameter had little influence on re- recovery by 4 years. An additional 9 million larvae (12 million covery time. Recovery time had a greater sensitivity to α and total annually) accelerated recovery by only 8 additional years. stabilized run size a comparable sensitivity. Recovery time was The effect of stocking was greatly dependent upon run size. The sensitive to the parameters influencing survival (hatch success, effect of stocking 12 million larvae annually with a starting run at-sea mortality, and postspawn mortality) though except for size of 1,000 fish was to advance recovery by 12 years (Fig- at-sea mortality these parameters had little effect on stabilized ure 4). However, when 5,000 fish were present in the spawning run size. Fecundity estimators understandably influence some run, the additional gain in recovery time was less than half that– outputs. Time to recovery was sensitive to both fecundity esti- only 4 years (Figure 5). In the same vein, when the starting run mators, but stabilized run size was not. size was 15,000, the additional gain was a mere 2 years. The salient point here is that in even without stocking, the time to recovery is very sensitive to starting run size. Time to recovery Effect of Stocking and Population Size approximated a linear relationship with the log10 transformed Stocking had a strong effect on the time to recovery in value of run size, so that small differences in run size at low lev- the base model. Predictably, stocking more fish resulted in a els of stocking had great effects on time to recovery. Conversely,

Effect of stocking 12 M Fry Effect of stocking 50

No stocking (base model) 1,000 Stocking 40

Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 40

30 5,000

20 30 15,000 Years to 50% recovery Years to 50% recovery 10

0 0 1000 10000 100000 010203040 Starting Population Stocking Rate in Millions

FIGURE 4. Effects of the starting size of the American Shad spawning run on FIGURE 5. Years to 50% recovery with variable supplemental stocking rates the modeled number of years needed to reach 50% recovery with and without and initial starting populations of 1,000, 5,000, and 15,000, assuming instanta- supplemental stocking of 12 million larvae assuming instantaneous access to neous access to habitat made available through planned restoration of American habitat made available through planned restoration on the Penobscot River. Shad on the Penobscot River. RESTORATION POTENTIAL OF AMERICAN SHAD SPAWNING RUN 465

at higher run sizes, differences had a diminishing influence on In general, the efficacy of hatchery supplementation is by recovery time. no means a known quantity. There are known risks with stocking out-of-basin fish, including outbreeding depression (Lynch 1991), low effective population size (Waples and Do 1994), and DISCUSSION swamping of adaptive genetic variation (Hansen et al. 2001). In regions where American Shad restoration is the goal, Other restoration stocking projects with American Shad have the decision to stock or not is made on the basis of the best considered these risks in designing a conservation plan. An ini- available information. Our population model draws on data from tial goal for the James River, Virginia, hatchery program was to diverse sources with unknown accuracy and precision in order restrict the collection of broodstock to fish from within the river to serve as a platform for assessing the sensitivities of popula- to minimize the potential risks associated with transfers (Brown tion recovery. Given these limitations, it is important to point et al. 2000). This goal could not be met, so next-best alterna- out that these parameters generated an age structure that was tives were considered, including using fish from rivers that (1) not different than what was observed in the Exeter River, a sys- support large and viable stocks, (2) are nearest neighbors, and tem that usually has fewer than 100 returns of shad annually (3) are genetically less divergent from other stocks (Epifanio (Figure 2). This lends support for the model’s ability to evaluate et al. 1995). For our case study, managers in Maine would potential population outcomes under different stocking levels face similar challenges in collecting broodstock. A Penobscot and starting population sizes. River shad stocking program would likely use out-of-basin It is not surprising that the starting population size has a source stock, as Penobscot River shad will be “not easily cap- strong effect on the rate of (and years to) recovery (Figure 4), tured” until the Milford Dam fish lift is complete and operational as this model is assuming newly opened habitat, but this effect (MDMR 2009). The nearest reliable source for shad broodstock is biologically noteworthy. An increase in starting run size from is the Merrimack River, which is approximately 201 km from 1,000 to 5,000 fish is predicted to reduce the time to recovery by the mouth of Penobscot Bay and entails a more than 3-h transfer 11 years. The strong sensitivity of this model to starting run size via stocking truck. There are at least three other river systems highlights the importance of characterizing the extant popula- known to have shad runs that are closer, but all have small or tion size prior to restoration. In our case study, the starting run unknown population sizes (MDMR 2009). size of the Penobscot River remains a critical unknown. Only Our model is limited in that it does not take into account recently have biologists recognized that there appears to be a genetic stock structure or the potential for hatchery restora- self-sustaining population (A. Grote, University of Maine, per- tion to disrupt the genetic structure of a remnant population or sonal communication) with juveniles in the estuary (C. Lipsky, compromise any undetected adaptive potential that is currently NOAA–Fisheries, personal communication). present. However, genetic structure must be carefully considered If hatchery supplementation is chosen as a recovery tool, before a stocking program progresses, as effective restoration it is also important to understand the predicted interaction should attempt to recover representative diversity as far as is between starting population and stocking effectiveness. This practical (Hasselman and Limburg 2012). Previous studies have model shows the efficacy of stocking to be greatest at the lowest found that American Shad have a shallow but significant stock population levels. For example, at an assumed level of 1,000 fish structure (Bentzen et al. 1989; Brown et al. 2000). However, in the spawning run, stocking 12 million larvae annually is pre- these results were obtained with a less than ideal power to dif- dicted to accelerate anticipated recovery by 12 years. With 5,000 ferentiate stocks; the study used relatively few microsatellite fish in the spawning run, the same aggressive stocking program loci (five) and spawning populations within close geographic

Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 would result in greatly diminished returns and advance recovery proximity. A recent study in Canada with a broader geographic by fewer than 4 years. Note also that a difference in run size range and more statistical power (13 polymorphic microsatellite of only 4,000 fish has a comparable result in recovery. The di- loci) found more substantial population structure between rivers minishing effectiveness of stocking as natural or hatchery-aided (Hasselman et al. 2010). Even shallow stock structure is notable, recovery takes place is an important consideration before invest- as early shad restoration programs (1800s–1950) stocked over ing limited resources in hatchery-based supplementation. In our 1 billion larvae and often with mixed-river stocks (Hasselman case study, the current plan for the Penobscot River (MDMR and Limburg 2012). In the James River, the reproductive contri- 2009) calls for stocking hatchery-reared individuals until carry- bution of individual broodstock was clearly nonrandom and is ing capacity is reached for five consecutive years. These data cause for close hatchery management and evaluation. Although indicate that even at low run sizes supplementation may be there did not appear to be a significant decline in microsatellite ineffectual, and that at run sizes near the carrying capacity sup- variation, one male fathered more than half the progeny and plementation may be futile. Only at the lowest run sizes (less nearly half the progeny represented only three families (Brown than 10% of the carrying capacity) did hatchery supplementation et al. 2000). have the potential to noticeably increase run size. This indicates Hatchery restoration efforts have been deemed successful in that stocking American Shad is a better tool for reintroduction a number of rivers where a large percentage of the individuals than for supplementation. returning to spawn are of hatchery origin, reflecting population 466 BAILEY AND ZYDLEWSKI

increases comprised of donor stocks rather than native river tower 2012) showed that increased access to habitat is not a stock (Aunins 2010). In the Potomac River, however, the in- panacea for population recovery and may not increase Ameri- crease in adult returns is thought to be driven largely by a reduc- can Shad populations without increases in other factors in the tion in the at-sea fishery (Aunins 2010), and not supplementation newly available habitat, such as juvenile survival and spawning with hatchery fish. Increasing adult escapement and accessibil- success. ity to spawning grounds via adequate fish passage may be a more Due to the high variability inherent in natural biological sys- powerful driver of population recovery than hatchery stocking tems, many of the assumptions we used to construct our model (Aunins 2010) at all but the lowest run sizes. were conservative in order to prevent overestimation of popu- In using the restoration of the Penobscot River as a case lation trends. Reality may thus exceed the trends seen in this study, we reiterate that the study was not meant to predict the model. Our rate of iteroparity is low because it is based on the time course of recovery in the river once the anticipated dam predictions of Leggett and Carscadden (1978). The calculated removals are accomplished. A key assumption of this model rate of run size increase (r = 0.15) is low compared with that of is that the removal of the dams and the greater connectivity other recoveries involving American Shad due to increased habi- afforded by improvements to passage will allow for recolo- tat accessibility and natural population fluctuations. In northern nization that will be dominated by fecundity, survival, and the systems, the degree of iteroparity may increase the rate at which theoretical carrying capacity of the system. The model does new habitat is filled. In the context of this model, the degree to not attempt to quantify the quality of the habitat at the newly which hatchery supplementation might be effective is dependent accessible historical spawning grounds. There are simply too upon the intrinsic growth rate. many unknowns associated with the anticipated passage to It is not our intent to recommend stocking or not stocking base a recovery model on increased habitat accessibility. Such in the Penobscot River or any other river; such decision would assumptions of access and utilization of habitat after restoration be value judgments. Rather, our intent is to highlight the sen- represent important goals of the PRRP assessment. sitivities of this model in order to fill gaps in our knowledge In constructing this model, we used the best available data, so that management decisions can be based on the best sci- though we identified several key components for which the only ence available. In practice, active restoration is an integration data available are unsatisfactory. As a result, the specific values of both values and science (Hart et al. 2002). As such, the de- of many of the parameters could—and should—be critically cision to stock or not to stock could be informed by the use evaluated. Specifically, there appear to have been no attempts to of a structured decision-making approach. This process allows estimate the batch fecundity of Maine American Shad and few stakeholders to fully explore their fundamental objectives, ul- attempts for shad in their northern range as a whole (Collette timately focusing on the potential trade-offs of management and Klein-MacPhee 2002). Even if the fecundity estimates were action (Holling 1978). Irwin et al. (2011) suggest that linking precise and accurate, recent research has suggested that shad are management options to expected outcomes is most effectively not likely to spawn all available eggs (Olney and McBride 2003). accomplished through the use of quantitative systems models It is also unknown what the average fertilization rate is for shad as decision-support tools. Such a decision-making framework eggs released during a natural spawning event. Our model is also would be instructive not only for the decision when or if to stock based on the best local data on maturity schedules, but maturity but also for the decision of when to stop. schedules vary spatially (Tuckey and Olney 2010) and aging shad via scales is imprecise at best (McBride et al. 2005; Duffy et al. 2011). ACKNOWLEDGMENTS

Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 Our model is based on a Ricker-type recruitment curve. This Support for this work was through the U.S. Geological Sur- type of curve has not been described for American Shad, and vey, Maine Cooperative Fish and Wildlife Research Unit, North- there have been few successful attempts to apply any type of east Fisheries Science Center, U.S. Fish and Wildlife Service, stock–recruitment curve to alosines (Crecco et al. 1986). Early and the Nature Conservancy. This manuscript was greatly im- survival rates of wild shad larvae are difficult to assess and proved by comments from Chris Caudill, Kevin Dockendorf, rarely studied (see Crecco et al. 1983 for one of the few excep- Dan Hasselman, Jerre Mohler, and two anonymous reviewers. tions). However, these rates are needed to assess the advantage Mention of trade names does not imply endorsement by the U.S. of hatchery-produced larvae in shad recovery programs across Government. the species’ native range. The results of this model rely heavily on the density-dependent effects afforded by this recruitment model. Additionally, our assumption of a 10-fold increase REFERENCES in survival from eggs to larvae is based on what limited data ASMFC (Atlantic States Marine Fisheries Commission). 2007. American Shad are available and is likely an oversimplification. Survival rates stock assessment report for peer review, volumes I–III. ASMFC, Stock As- sessment Report 07–01 Supplement, Washington, D.C. among hatchery-raised shad can be known until stocking, but Aunins, A. W. 2010. Genetic evaluation of American Shad, Alosa sapidissima, poststocking survival to juvenile stages or seaward migration restoration success in James River, Virginia. Doctoral dissertation. Virginia is difficult to assess. Another recent model (Harris and High- Commonwealth University, Richmond. RESTORATION POTENTIAL OF AMERICAN SHAD SPAWNING RUN 467

Bentzen, P., G. C. Brown, and W. C. Leggett. 1989. Mitochondrial DNA poly- eries: Dynamics, Management, and Ecosystem Science [online serial] 4: morphism, population structure, and life history variation in American Shad 174–187. (Alosa sapidissima). Canadian Journal of Fisheries and Aquatic Sciences Hayes, C., and R. Carmichael. 2002. Salmon restoration in the Umatilla River: 46:1446–1454. a study of straying and risk containment. Fisheries 27(10):10–19. Bilkovic, D. M., C. H. Hershner, and J. E. Olney. 2002. Macroscale assessment Holling, C. S., editor. 1978. Adaptive environmental assessment and manage- of American Shad spawning and nursery habitat in the Mattaponi and Pa- ment. Wiley, London. munkey rivers, Virginia. North American Journal of Fisheries Management Irwin, B. J., M. J. Wilberg, M. L. Jones, and J. R. Bence. 2011. Applying 22:1176–1192. structured decision making to recreational fisheries management. Fisheries Brown, B. L., T. P Gunter, J. M. Waters, and J. M. Epifanio. 2000. Evaluating ge- 36:113–122. netic diversity associated with propagation-assisted restoration of American Leggett, W. C., and J. E. Carscadden. 1978. Latitudinal variation in reproductive Shad. Conservation Biology 14:294–303. characteristics of American Shad (Alosa sapidissima): evidence for popula- Collette, B. B., and G. Klein-MacPhee. 2002. Fishes of the Gulf of Maine. tion specific life history stages in fish. Journal of the Fisheries Research Board Smithsonian Press, Washington, D.C. of Canada 35:1469–1478. Chittenden, M. E., Jr. 1976. Weight loss, mortality feeding, and duration of Liem, A. H., and W. B. Scott. 1966. Fishes of the Atlantic Coast of Canada. residence of adult American Shad, Alosa sapidissima, in fresh water. U.S. Fisheries Research Board of Canada Bulletin 28:23–41. Fish and Wildlife Service Fishery Bulletin 74:151–157. Limburg, K. E., K. A. Hattala, and A. W. Kahnle. 2003. American Shad in Crecco, V., T. Savoy, and L. Gunn. 1983. Daily mortality rates of larval and its native range. Pages 125–140 in K. E. Limburg and J. R. Waldman, ed- juvenile American Shad (Alosa sapidissima) in the Connecticut River with itors. Biodiversity, status, and conservation of the world’s shads. American changes in year-class strength. Canadian Journal of Fisheries and Aquatic Fisheries Society, Symposium 35, Bethesda, Maryland. Sciences 40:1719–1728. Lynch, M. 1991. The genetic interpretation of inbreeding depression and out- Crecco, V., T. Savoy, and W. Whitworth. 1986. Effects of density-dependent breeding depression. Evolution 45:622–629. and climatic factors on American Shad, Alosa sapidissima, recruitment: a MDMR (Maine Department of Marine Resources). 2009. Operational plan for predictive approach. Canadian Journal of Fisheries and Aquatic Sciences the restoration of diadromous fishes to the Penobscot River. MDMR and 43:57–463. Department of Inland Fisheries and Wildlife, Augusta. Day, L. R. 2006. Restoring native fisheries to Maine’s largest watershed: the McBride, R. S., M. L. Hendricks, and J. E. Olney. 2005. Testing the validity Penobscot River Restoration Project. Journal of Contemporary Water Re- of Cating’s (1953) method for age determination of American Shad using search and Education 134:29–33. scales. Fisheries Research 30:10–18. Duffy, W. J., R. S. McBride, S. X. Cadrin, and K. Oliveira. 2011. Is Cating’s Molony, B. W., R. Lenanton, G. Jackson, and J. Norriss. 2003. Stock enhance- method of transverse groove count of annuli applicable for all stocks of ment as a fisheries management tool. Reviews in Fish Biology and Fisheries American Shad? Transactions of the American Fisheries Society 140:1023– 13:409–432. 1034. Moring, J. R. 1986. Stocking anadromous species to restore or enhance fisheries. Emlen, J. 1991. Heterosis and outbreeding depression: a multilocus Pages 59–74 in R. H. Stroud, editor. Fish culture in fisheries management: model and an application to salmon production. Fisheries Research 12: proceedings of a symposium on the role of fish culture in fisheries manage- 187–212. ment. American Fisheries Society, Bethesda, Maryland. Epifanio, J. M., P.E. Smouse, C. J. Kobak, and B. L. Brown. 1995. Mitochondrial Olney, J. E., and R. S. McBride. 2003. Intraspecific variation in batch fecundity DNA divergence among populations of American Shad (Alosa sapidissima): of American Shad: revisiting the paradigm of reciprocal latitudinal trends How much variation is enough for mixed-stock analysis? Canadian Journal in reproductive traits. Pages 185–192 in K. E. Limburg and J. R. Waldman, of Fisheries and Aquatic Sciences 52:1688–1702. editors. Biodiversity, status, and conservation of the world’s shads. American Foster, N. W., and G. C. Atkins. 1869. Second Report of the Commissioners of Fisheries Society, Symposium 35, Bethesda, Maryland. Fisheries of the state of Maine, 1868. Owen and Nash, Printers to the State, Ricker, W. E. 1975. Computation and interpretation of biological statistics of Augusta, Maine. fish populations. Fisheries Research Board of Canada Bulletin 191. Haefner, J. W. 2005. Modeling biological systems: principles and applications. Rulifson, R. A. 1994. Status of anadromous Alosa along the East Coast of North Springer Science, Business Media, New York. America. Pages 134–158 in J. E. Cooper, R. T. Eades, R. J. Klauda, and Hansen, M. M., D. E. Ruzzante, E. E. Nielsen, and K. L. D. Mensberg. 2001. J. G. Loesch, editors. Anadromous Alosa Symposium. American Fisheries Brown Trout (Salmo trutta) stocking impact assessment using microsatellite Society, Bethesda, Maryland. DNA markers. Ecological Applications 11:148–160. Schramm, H. L., Jr., and R. G. Piper, editors. 1995. Uses and effects of cultured Downloaded by [Department Of Fisheries] at 19:56 28 May 2013 Haro, A., M. Odeh, T. Castro-Santos, and J. Noreika. 1999. Effect of slope and fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, headpond on passage of American Shad and Blueback Herring through simple Bethesda, Maryland. Denil and deepened Alaska steeppass fishways. North American Journal of Stevenson, C. H. 1899. The shad fisheries of the Atlantic coast of the United Fisheries Management 19:51–58. States. Pages 101–269 in Report of the commissioner for 1899. U.S. Com- Harris, J. E., and J. E. Hightower. 2012. Demographic population model for mission of Fish and Fisheries, Washington, D.C. American Shad: will access to additional habitat upstream of dams increase Trinko Lake, T. R., K. R. Ravana, and R. Saunders. 2012. Evaluating changes population sizes? Marine and Coastal Fisheries: Dynamics, Management, and in diadromous species distributions and habitat accessibility following the Ecosystem Science [online serial] 4:262–283. Penobscot River Restoration Project. Marine and Coastal Fisheries: Dynam- Hart, D. D., T. E. Johnson, K. L. Bushaw-Newton, R. J. Horwitz, A. T. Bednarek, ics, Management, and Ecosystem Science [online serial] 4:284–293. D. F. Charles, D. A. Kreeger, and D. J. Velinsky. 2002. Dam removal: chal- Tuckey, T. D. and J. E. Olney. 2010. Maturity schedules of female American lenges and opportunities for ecological research and river restoration. Bio- Shad vary at small spatial scales in Chesapeake Bay. North American Journal science 52:669–681. of Fisheries Management 30:1020–1031. Hasselman, D. J., R. G. Bradford, and P. Bentzen. 2010. Taking stock: defining Waples, R. S. 1991. Genetic interactions between hatchery and wild salmonids: populations of American Shad (Alosa sapidissima) in Canada using neu- lessons from the Pacific Northwest. Canadian Journal of Fisheries and Aquatic tral genetic markers. Canadian Journal of Fisheries and Aquatic Sciences Sciences 48(Supplement 1):124–133. 67:1021–1039. Waples, R. S., and C. Do. 1994. Genetic risk associated with supplementa- Hasselman, D. J., and K. E. Limburg. 2012. Alosine restoration in the tion of Pacific salmonids: captive broodstock programs. Canadian Journal of 21st century: challenging the status quo. Marine and Coastal Fish- Fisheries and Aquatic Sciences 48:124–133. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 19:58 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Biological Reference Points for the Nutritional Status of Chesapeake Bay Striped Bass John M. Jacobs a , Reginal M. Harrell b , Jim Uphoff c , Howard Townsend d & Kyle Hartman e a National Oceanic and Atmospheric Administration, National Centers for Coastal Ocean Science, Cooperative Oxford Laboratory , 904 South Morris Street, Oxford , Maryland , 21654 , USA b Department of Environmental Science and Technology , University of Maryland , 2113 Animal Science/Agricultural Engineering Building, College Park , Maryland , 20742 , USA c Maryland Department of Natural Resources , Fisheries Service , 904 South Morris Street, Oxford , Maryland , 21654 , USA d National Marine Fisheries Service , Chesapeake Bay Office , 904 South Morris Street, Oxford , Maryland , 21654 , USA e Division of Forestry and Natural Resources , West Virginia University , 310A Percival Hall, Morgantown , West Virginia , 26506 , USA Published online: 28 Apr 2013.

To cite this article: John M. Jacobs , Reginal M. Harrell , Jim Uphoff , Howard Townsend & Kyle Hartman (2013): Biological Reference Points for the Nutritional Status of Chesapeake Bay Striped Bass, North American Journal of Fisheries Management, 33:3, 468-481 To link to this article: http://dx.doi.org/10.1080/02755947.2013.763876

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:468–481, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.763876

ARTICLE

Biological Reference Points for the Nutritional Status of Chesapeake Bay Striped Bass

John M. Jacobs* National Oceanic and Atmospheric Administration, National Centers for Coastal Ocean Science, Cooperative Oxford Laboratory, 904 South Morris Street, Oxford, Maryland 21654, USA Reginal M. Harrell Department of Environmental Science and Technology, University of Maryland, 2113 Animal Science/Agricultural Engineering Building, College Park, Maryland 20742, USA Jim Uphoff Maryland Department of Natural Resources, Fisheries Service, 904 South Morris Street, Oxford, Maryland 21654, USA Howard Townsend National Marine Fisheries Service, Chesapeake Bay Ofﬁce, 904 South Morris Street, Oxford, Maryland 21654, USA Kyle Hartman Division of Forestry and Natural Resources, West Virginia University, 310A Percival Hall, Morgantown, West Virginia 26506, USA

Abstract The assessment of the nutritional status of fish is a central requirement of fisheries management. However, there has been little consensus on the appropriate indicator to use, and even less effort toward defining biological thresholds and reference points. With current efforts to manage fisheries in an ecosystem context, environmental effects and trophic relationships need to be considered and appropriate indicators developed. To address this concern, we compiled five different studies in which multiple indicators of nutritional status were applied to Striped Bass Morone saxatilis of different age-classes, geographical origins, and environments (cultured and wild). Proximate composition analysis was used to compare measured lipid concentrations in the anterior dorsal muscle against selected indicators, including

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 Fulton’s condition factor, relative weight, percent moisture, and an index of visceral body fat. The results suggest that weight-at-length indices are less sensitive than proximate analysis and poorly related to lipid concentration. However, models developed for both moisture and the body fat index adequately represent tissue lipids and offer clear thresholds of lipid depletion. We propose the use of the proportion of Striped Bass with anterior dorsal muscle composed of more than 80% moisture, or classiﬁed as having no observable visceral body fat, as a working protocol for thresholds of poor condition in ecosystem approaches to Chesapeake Bay Striped Bass management. Further, based on historical data we propose an interim target condition of 75% of individuals containing less than 80% moisture as a management goal.

Biological reference points have long supported ﬁsheries stocks of commercial importance and aspects of production and management by providing targets and thresholds for decision removal (Sissenwine and Shepherd 1987; Mace 1994). More making. Traditionally, these have primarily focused on single recently, ecosystem approaches to ﬁsheries management have

*Corresponding author: [email protected] Received May 3, 2011; accepted December 19, 2012 468 NUTRITION OF STRIPED BASS 469

gained momentum, with a holistic goal of obtaining sustain- have been applied, including Fulton’s condition factor (Nash able ecosystems (Pikitch et al. 2004; Link 2005). In this light, et al. 2006) and relative weight (Brown and Murphy 1991a, identifying appropriate indicators that consider trophic relation- 1991b). Fulton’s condition factor is based on a proposed cubic ships and other processes independent of fisheries removals are relationship between weight and length and has been widely ap- paramount in informing management decisions. plied in fisheries science and management. It has also received A clear example of the need for ecosystem-based approaches considerable criticism, primarily due to its assumption of iso- to management and appropriate indicators lies in the plight of the metric growth and because it is a measure of physical robustness Atlantic coast Striped Bass Morone saxatilis population and, in rather than nutritional state (Jakob et al. 1996; Nash et al. 2006). particular, the challenges faced within Chesapeake Bay. The de- Relative weight relates the weight of an individual fish to a stan- cline and subsequent recovery of the Atlantic coast Striped Bass dard weight established for each species, thus providing easily have been well documented and this case is largely considered determined reference values for management (e.g., proportion a management success (Richards and Rago 1999). However, of standard weight). Standard weights have been developed for coinciding with the historically high abundance of the species a wide variety of species, and the index has been widely applied, by the mid to late 1990s are indications of additional impacts particularly in inland fisheries (Blackwell et al. 2000; Pope and throughout the ecosystem, such as a high prevalence of disease Kruse 2007). (Overton et al. 2003; Kaattari et al. 2005; Gauthier et al. 2008), While weight- and length-based indices have the benefit of increased natural mortality (Jiang et al. 2007; Sadler et al. 2010), ease of collection and readily available technology, their rela- a reduction in the abundance of principal prey sources (Uphoff tionship to the true nutritional status of fish has been inconsistent 2003), and shifts in foraging behavior (Hartman and Brandt (Niimi 1972; Brown and Murphy 1991a; Herbinger and Friars 1995b; Griffin and Margraf 2003; Pruell et al. 2003; Walter 1991; Plante et al. 2005; Davidson and Marshall 2010). In gen- et al. 2003; Overton et al. 2009). The energy reserves of indi- eral, fish starvation is a sequential process of glycogen and lipid vidual fish and populations relate strongly to foraging success, depletion coupled directly with tissue hydration. During pro- reproductive success, potential prey density, habitat conditions, longed starvation, protein is catabolized from tissue but gener- environmental stressors, and subsequent fish health and survival ally contributes less energetically than lipid (Love 1970; Niimi (Love 1970; Adams 1999; Schultz and Conover 1999; Biro et al. 1972; Jobling 1980; Jezierska et al. 1982; Black and Love 1986). 2004; Jacobs et al. 2009b). A reliable and easily applied indi- This negative relationship between lipid and water during starva- cator of nutritional state is critical for evaluating hypotheses tion serves to conserve weight and hamper morphometric-based related to nutrition, prey abundance, density, and the outcome indicators of nutritional status. For this reason, the direct mea- of the management measures that may follow. surement of the biochemical composition of fish (moisture, lipid, Nutritional indicators for management played an important protein, and ash), or proximate composition, is still considered role in developing alternative hypotheses to explain the collapse the standard method for comparing other techniques (Brown and of Atlantic Cod Gadus morhua. Cod populations from southern Murphy 1991a; Hartman and Margraf 2008; Jacobs et al. 2008). Labrador to eastern Nova Scotia collapsed in the late 1980s, and Proximate composition offers a precise measure of nutri- moratoria were declared by the early 1990s (Myers et al. 1996). tional state; however, the expense, extensive processing time, While the usual suspects of fishing mortality and environmental and lethality of the approach have limited its application degradation were initially implicated, a striking divergence be- (Walsberg 1988; Brown et al. 1993; Jacobs et al. 2008). The tween field data and stock assessment model results suggested inverse relationship between moisture and lipid as well as pro- a loss of fish beyond the capacity of the fishery (Fu et al. 2001). tein in fish (Love 1970) has been used as the basis for estimating

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 It was suggested that natural mortality had increased dramat- lipid content and energy density from moisture alone (Hartman ically and this was linked to reduced body condition indices and Brandt 1995a). Alternatively, direct observation of visceral (Lambert and Dutil 1997; Dutil and Lambert 2000; Shelton and lipid reserves may equally allow for a cost-effective means of Lilly 2000). Concurrent with the decline of the most northern assessing nutritional status (Goede and Barton 1990). It is also stocks has been the disappearance of Capelin Mallotus villo- possible that traditional weight- and length-based measures are sus, a lipid-rich, energy-dense prey that historically constituted appropriate for some species, which would facilitate historical the cod’s most important food source (Lilly 1994). It has been analysis. In all cases, indicators need to be evaluated for individ- suggested that the rebuilding of the northern stocks is depen- ual species and biological reference points established if they dent on Capelin production (Rose and O’Driscoll 2002). As are to have utility in resource management. demonstrated by the plight of the cod, the ability to elicit the In the Chesapeake Bay region, efforts are underway to tran- connection between nutritional state and ecosystem processes sition to an ecosystem-based fishery management approach can be enlightening. (Maryland Sea Grant 2009). As part of this process, Striped While a wide diversity of condition indices have been ap- Bass and four other species are being considered with an effort plied in resource management, weight and length indicators to identify indicators and thresholds at which management ac- have been most commonly employed to assess fish condition tion is needed. Clearly identified is the need for condition or due to their ease of measurement and low cost (Stevenson and nutritional health indicators for Striped Bass to assess the impli- Woods 2006). Several methods for relating weight to length cations of changes in ecosystem trophodynamics. In response 470 JACOBS ET AL.

to this management need, the objectives of this study were to where (1) compare morphometric (weight–length), observational (vis- ≡− . + . , ceral body fat), and indirect (moisture) indicators for assessing log10 Ws 4 924 3 007(log10 TL) muscle lipid content in Striped Bass and (2) establish biological reference points for the most appropriate indicators. Ws is length-specific standard weight and TL is total length (cm). Proximate composition.—For all data sets included in this analysis, the skinless anterior portion of the left fillet was used METHODS for proximate analysis. All muscle samples were wrapped in Data sources.—The data used in this analysis were derived both cellophane and freezer paper (outer covering) and grouped from five different experimental studies and field monitoring in freezer bags with all air removed. Field samples were placed efforts spanning a 10-year period from 1996 to 2005 (Table 1). immediately on dry ice, returned to the laboratory, and main- The first data source used (1996) was a strain evaluation consist- tained at –80◦C until analysis. Samples collected from cul- ing of the progeny of 19 single-parent crosses from the Chesa- tured fish were frozen directly at –80◦C. Proximate composition peake Bay, Hudson, Roanoke, Apalachicola, and St. Johns rivers methodology followed standard protocols (AOAC 2005; Jacobs (Jacobs et al. 1999). All fish were fed a commercial diet to satia- et al. 2008) with up to three replicates per sample. Briefly, sam- tion and reared in recirculating systems with the goal of optimiz- ples were weighed (nearest 0.01 g), homogenized, and dried ing growth. Full proximate composition data as well as weights to completion (3–4 g per replicate) with moisture being deter- and lengths were recorded at the termination of the growth trials; mined by the difference in sample weight. Neutral lipids were however, the body fat index described below was not computed. extracted from dried samples (2.5 g per replicate) over 8 h by The second data source represents wild-collected Striped Bass means of a Golfisch apparatus with petroleum ether as a solvent. from Maryland’s portion of the Chesapeake Bay during the fall Protein was determined by Kjeldahl nitrogen from fat-free dried (October–November) of 1998–2001 by the Maryland Depart- tissue (0.25 g per replicate, tissue nitrogen to protein conversion ment of Natural Resources (MDNR). Fish were captured by both factor = 6.2), followed by ash by combustion (1 g per repli- hook and line and pound net and processed immediately in the cate, 550◦C, 12 h). All values are reported as percentages of field. Complete proximate composition, weight-at-length, and wet weight (mg/g). Replicates were averaged prior to statistical body fat indices were available (Jacobs 2007). These fish were analysis. Measurement error among replicates from individual considered Chesapeake Bay residents and not part of the stock fish averaged ± 0.2% and 0.3% (SD) for lipid and moisture, that joined the coastal migration. The final three data sources respectively. were lipid depletion studies conducted to establish reference Body fat index.—The body fat index (BFI) was modified from values for field observations (depletion study 1, 2000; Jacobs that proposed by Goede and Barton (1990) to reduce the number 2007), evaluate serum chemistry indicators of nutritional status of categorical classifications. A relative score was assigned to (depletion study 2, 2004–2005; J. M. Jacobs, unpublished data), each fish based on the visual prevalence of visceral storage and determine lipid distribution within a Striped Bass (deple- lipids and scaled as follows: 0 = no detectable storage lipids; tion study 3, 2005; Jacobs et al. 2008). Depletion study 1 was 1 = lipids present but less than 25% of viscera covered; 2 = conducted with wild-collected Striped Bass, with one-half of 25–75% of viscera covered; and 3 = approximately 75% or the fish being fed Atlantic Menhaden Brevoortia tyrannus daily greater of viscera covered (Figure 1). Because of the subjective and the other half starved to establish a range of lipid concen- nature of the index, a minimum of two observers were used to trations over the course of 3 months. The latter two depletion obtain consensus on classification and reduce individual bias.

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 studies were conducted in a similar manner, but the fish were F1 All observers were trained through the use of photographs and progeny of Choptank River Striped Bass and were fed a com- guidance of senior field personnel experienced with the index. mercial diet. In all cases, fish were sampled at regular intervals Statistical analysis.—The first step of our analysis was to during starvation to capture a full range of lipid concentrations. determine alternative indicators reliably predicted by muscle Weight-at-length indices.—Two morphometric indices were lipid concentration and factors that may alter the nature of this considered for estimation of fish condition or robustness: Ful- relationship. The general approach was to fit candidate sets of ton’s condition factor (K) and relative weight (Wr). The former models and use an information-theoretic approach (Burnham was calculated as follows: et al. 2011) for model selection (i.e., the Akaike information criterion corrected for small sample size [AICc]). The models W were built for three outcome variables (dependent variables): K = × 10,000, 3 L BFI, moisture, and Wr. The predictor variables (independent variables) for the candidate models were linear combinations where W is wet weight (g) and L is TL (cm). Relative weight was of lipid, sex, length, source (cultured or wild), and the interac- calculated after Brown and Murphy (1991a, 1991b) as follows: tion of lipid and source (Table 2). Coding was used to include the discrete variables sex (female = 0, male = 1) and source W (cultured = 0, wild = 1). Moisture and Wr were fit using gen- Wr = × 100, Ws eralized linear models with a Gaussian distribution and identity Downloaded by [Department Of Fisheries] at 19:58 28 May 2013

TABLE 1. Descriptive statistics for the Striped Bass from the ﬁve collections used for model development, by year, study, origin, source, and sex. Source refers to whether Striped Bass were maintained in artiﬁcial settings (Culture) or removed directly from natural waters (Wild); Wr = relative weight, ww = wet weight, and length = TL.

Length (cm) Lipid (% ww) Moisture (%) Wr

Year Study Origin Source N Sex Mean (SD) Range Mean (SD) Range Mean (SD) Range Mean (SD) Range Reference

1996 Striped Bass strain Hudson River, Culture 33 M 37.72 (1.80) 32.10–41.50 4.27 (1.47) 0.64–6.83 71.90 (2.23) 66.82–76.80 118.55 (14.64) 92.72–175.47 Jacobs et al. evaluation New York, to 26 F 37.82 (1.73) 34.60–42.10 3.90 (1.29) 1.80–6.56 72.58 (1.84) 68.30–75.76 112.90 (13.24) 157.50–81.49 (1999) Apalachicola River, Florida 1998–2001 Striped Bass health Chesapeake Bay, Wild 83 M 47.01 (5.83) 32.50–63.80 0.37 (0.60) 0.00–3.90 79.72 (2.11) 75.59–88.91 75.96 (10.57) 47.97–114.10 Jacobs (2007) evaluation North of 23 F 47.01 (5.67) 34.50–58.80 0.19 (0.26) 0.03–0.92 79.16 (1.47) 77.40–82.18 71.39 (6.41) 57.81–87.58 Potomac River 2000 Depletion study 1 Chesapeake Bay Culture 12 M 50.39 (5.10) 43.70–59.50 1.41 (1.50) 0.04–4.48 77.18 (2.91) 74.00–82.62 81.02 (15.11) 63.82–105.94 Jacobs (2007) 1 F 54.00 0.96 75.63 89.68 2004–2005 Depletion study 2 Choptank River, Culture 14 M 32.46 (1.26) 30.50–34.50 2.47 (0.56) 1.54–3.28 75.09 (0.83) 73.06–76.13 82.79 (7.30) 72.09–95.55 Jacobs Chesapeake Bay 18 F 32.83 (2.89) 30.00–38.50 1.94 (0.58) 1.06–3.31 75.58 (3.66) 62.45–82.15 81.93 (6.97) 72.92–96.19 (unpublished) 2005 Depletion study 3 Choptank River, Culture 27 M 24.82 (1.88) 21.50–28.00 2.36 (0.96) 0.55–3.73 76.93 (1.53) 74.73–80.78 78.51 (7.20) 64.28–92.68 Jacobs et al. Chesapeake Bay 25 F 24.75 (1.82) 22.00–29.00 2.51 (0.86) 0.92–4.66 76.98 (1.36) 74.31–88.83 73.79–10.88 42.01–88.83 (2008) 471 472 JACOBS ET AL.

variance (ANOVA) followed by least-square means comparison was employed to test the discriminatory capability of each BFI class from the other based on tissue moisture and total lipid. This same analysis was used to examine the potential improvement in classification by reducing the number of BFI categories from four (as described above) to three (classes 2 and 3 were collapsed to represent >25% coverage of viscera) and two (presence or absence). Thresholds.—We used the point of tissue lipid depletion (lipid = 0[L0]) as the key threshold for risk and evaluated the utility of the surrogate indicators (moisture, BFI, and Wr)in predicting L0. Moisture threshold was calculated by first solving the best moisture regression model for L0 for all possible scenarios for wild fish (minimum and maximum length, male and female) and then determining the mean. Mean moisture and lipid concentrations were compared with those of fish given a BFI score of zero (no observable visceral lipid) to determine the utility of this indicator in predicting L0. Reference data and interim targets.—For reference conditions, mean proximate composition values for wild-collected Striped Bass previously published by Karahadian et al. (1995) were used. The study design in their 1990 collection was nearly identical to that used for the wild-collected fish in the present study in terms of the months of collection, size of fish, location of capture, and tissue processing; the notable exception lies in their FIGURE 1. Photographs illustrating the derivation of the body fat index for method of lipid determination. The authors used a chloroform : Striped Bass as modified from that proposed by Goede and Barton (1990). methanol extraction, which precludes direct comparison of to- Visceral lipid deposits are readily observable as off-white adipose tissue loosely tal lipids with our data. The authors present mean muscle tissue connected to the digestive tract (arrow). As lipid accumulates, visceral organs moisture data from fish (X¯ = 44.5 cm TL) collected in the upper are increasingly hidden from visual observation, offering a convenient reference = = for semiquantification. Each fish is scored based on the percent coverage of its Chesapeake Bay (n 22) and Potomac River (n 25) during viscera, with 0 = 0% (panel A), 1 = <25%, 2 = 25–75%, and 3 = 75% and October and November of 1990. Choptank River fish (n = 10) greater (panel B). [Figure available in color online.] are also listed as part of this seasonal collection but were not used for reference calculation because they were obtained in January link. Length was excluded from model building for Wr because of 1991 and represent overwintering rather than fall conditions it is part of the equation for calculating this index. The BFI (Karahadian et al. 1995). The reported mean moisture concen- was fit using an ordinal logistic regression model because of the trations were 77.88% (upper bay) and 78.86% (Potomac), with categorical nature of the data. an overall mean of 78.37%. This is the only chemical compo- Under the information-theoretic approach, a candidate set of sition data available for Chesapeake Bay Striped Bass prior to

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 models is fit and then compared using a model selection crite- our efforts in 1998; the collection offers insight into the condi- rion. The best model is determined by examining their relative tion of Striped Bass during a time when population levels were distance to the “truth.” We used AICc weight as a measure of approximately one-half that of current levels and predator–prey the relative plausibilities of the models within the candidate ratios were more favorable (Uphoff 2003). set. We interpret the AICc weight (wi) as evidence that model To enhance the utility of the thresholds determined from i is the best approximating model, given the data and set of this analysis and provide an initial, interim target to serve as candidate models (Burnham and Anderson 2002). Coefficients reference, we used the cumulative distribution function of the of determination (R2) were also calculated for each model to standard normal curve to define the proportion of the population demonstrate the proportion of variance explained. For the BFI equal to or exceeding the moisture threshold or that had mois- models, McFadden’s pseudo-R2 (Menard 2000) was calculated ture values greater than or equal to the point of lipid depletion from the ratio of intercept (IO) and intercept and covariate (IC) (L0). This approach allows for a simple comparison of the ob- –2 log likelihood as follows: served number of fish that are in poor condition with the number that would have been in this state in the reference population. 2 = − / . Rlogistic (IO IC) IO Because information on statistical variability was not obtain- able from Karahadian et al. (1995), the error associated with Contingency tables were further used to evaluate the errors in the each annual collection of wild-collected Chesapeake Bay fish classification rates for the BFI logistic models, and analysis of (1998–2001; Table 1) was used as a surrogate to calculate the NUTRITION OF STRIPED BASS 473

TABLE 2. Model selection based on Akaike’s information criterion corrected for small sample size (AICc), the change in AICc,AICc weight, and the coefﬁcient 2 2 of determination (R or McFadden’s pseudo-R for the ordinal logistic regression model for the body fat index [BFI]). Models are ranked according to AICc (lower is better). The BFI was not evaluated in the Striped Bass strain evaluation data set and thus is not included in model selection; –2 log(L) = –2 log likelihood or deviance and K = the number of estimated parameters.

Dependent 2 variable Predictor variables K −2log(L)AICc AICc AICc weight R

Wr Lipid + Source + Sex 4 −1,056.3 2,122.8 0.0 0.416 0.52 Lipid + Source 3 −1,057.7 2,123.6 0.8 0.279 0.52 Lipid + Source + Source × Lipid + Sex 5 −1,056.2 2,124.8 2.0 0.153 0.52 Lipid + Sex 3 −1,058.6 2,125.3 2.5 0.119 0.51 Lipid 2 −1,060.9 2,127.9 5.1 0.032 0.51 Intercept only 1 −1,153.8 2,311.6 188.8 0.000 Moisture Lipid + Source + Sex + Length 5 −536.5 1,085.3 0.0 0.731 0.76 Lipid + Source + Source × Lipid + 6 −537.6 1,085.5 0.2 0.256 0.76 Sex + Length Lipid + Source 3 −543.3 1,094.7 9.4 0.006 0.74 Lipid + Sex + Length 4 −543.2 1,096.7 11.4 0.002 0.74 Lipid + Length 3 −544.8 1,097.8 12.5 0.001 0.74 Lipid 2 −545.9 1,097.9 12.6 0.001 0.74 Lipid + Sex 3 −545.0 1,098.1 12.8 0.001 0.74 Intercept only 1 −702.0 1,408.1 322.8 0.000 BFI Lipid + Length 3 −147.2 304.8 0.0 0.404 0.40 Lipid + Sex + Length 4 −147.0 306.5 1.7 0.173 0.40 Lipid 2 −149.2 306.7 1.9 0.156 0.40 Lipid + Source 3 −148.7 307.8 3.0 0.090 0.40 Lipid + Source + Source × Lipid + 6 −145.9 308.5 3.7 0.064 0.40 Sex + Length Lipid + Source + Sex + Length 5 −147.0 308.7 3.8 0.059 0.40 Lipid + Sex 3 −149.2 308.8 4.0 0.055 0.40 Intercept only 1 −245.1 496.4 191.6 0.000

proportion of the 1990 population that exceeded the moisture possible for the species (0–6.83% lipid; Table 1). While this was threshold. The moisture data from this study most closely re- the purpose of combining data sources, differences are apparent sembled a normal distribution (Shapiro–Wilk W = 0.91, where in the distribution of the data relating to the purpose of the 1 is absolute normality). The standard deviations for these data study and whether the fish were wild collected or cultured. Wild Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 ranged from 1.08 to 2.33, with a mean of 1.81. Error distri- fish collected from 1998 to 2001 in Chesapeake Bay tended butions were evaluated for similarity by the modified Levene to be lower in lipid, higher in moisture, and larger than fish test (Brown and Forsythe test) conducted on absolute deviations considered cultured. Wild fish held in tanks and fed an Atlantic from the median of each sample (Boos and Brownie 2004). Menhaden diet (depletion study 1) were considered cultured in The variance in moisture was determined to be homogeneous this effort and demonstrated a full range of lipid (0.04–4.48%), among the sampling years (F = 2.09, P = 0.13), suggesting that similar to the other depletion studies with hatchery-reared fish. changes in mean moisture did not influence the error distribu- The strain evaluation fish had characteristically elevated lipid in tion. The proportions of fish exceeding the moisture threshold comparison to others (∼4%), relating to the intense husbandry calculated using these errors were used as target scenarios. strategy for optimal growth employed in this study. Moisture, BFI, and condition indices followed similar trends (Table 1).

RESULTS Weight-at-Length Indices Data Sources Both Fulton’s condition factor and relative weight were ini- The ﬁve data sources combined proved to offer a sufﬁcient tially included in the statistical analysis as simple measures of range of lipid concentrations to represent the physiological range robustness. However, the two morphometric indices were found 474 JACOBS ET AL.

200 only a slight improvement over a fully parameterized model (AIC = 0.2) but larger improvements over those not contain- 180 Wr = 70.52 + (8.12 × Lipid) c ing source as an explanatory variable (Table 2). The equation R2 = 0.51

) 160 best describing the moisture–lipid relationship is

Wr 140 Moisture = 80.95 − (1.50 × lipid) − (0.05 × length) 120 + (0.24 × sex) + (1.38 × source). 100 80 Lipid content clearly relates strongly to moisture content (Fig-

Relative Weight ( Weight Relative ure 3, upper panel), with only minor improvements in model 60 ﬁt being offered by more heavily parameterized models (2% 40 improvement in R2; Table 2). However, there is a difference in 20 0246895 A Lipid (mg/g ww) Moisture = 79.97 - (1.67 × Lipid) 90 R2 = 0.74 FIGURE 2. General relationship between relative weight (Wr) and anterior dorsal muscle lipid in Striped Bass. Model selection exercises demonstrated 85 only marginal improvement with the inclusion of source or sex as additional explanatory variables (Table 2). While Wr clearly relates to measured lipid, the model ﬁt is poor, with a high degree of variability in Wr at any given lipid level. 80

to be highly correlated (R2 = 0.99) over the range of sizes ex- (%) Moisture 75 amined (21.5–63.8 cm TL). Determining “standard condition” from 100% standard weight yields a “standard” Fulton’s condi- 70 tion factor for Striped Bass of 1.25. Because the two indices are so highly correlated, only relative weight was used for further 65 analysis. 02468 Of the candidate sets of models explored, the linear com- Lipid (mg/g ww) bination of lipid, sex, and source of fish proved to have the best AICc score (Table 2). However, the AICc weight for this 95 model suggests only a 42% chance of being better than the other B candidates. Akaike’s information criterion evidence ratios (the 90 ratios of the AICc weights) further suggest that this model is only 1.5 times more likely than the next highest scored (lipid 85 and source) to be the best-performing model (Table 2). The change in AICc between the full model and the model with lipid 80 only is 5.1, resulting in a 1% change in the explanation of vari- 2 Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 ance (R ). While this represents a significant change in AIC , c (%) Moisture 75 lipid is clearly the most important variable, with relatively mi- nor improvement being offered by more parameterized models 70 (Figure 2). However, none of the models fit particularly well, as demonstrated by the relatively low R2 values (Table 2). The 65 equation best describing the W –lipid relationship is r 10 20 30 40 50 60 70 TL (cm) Wr = 64.10 + (9.35 × lipid) + (2.95 × sex) + (5.42 × source). FIGURE 3. General relationships between (A) moisture and lipid and (B) moisture and length, by source of Striped Bass (squares = wild fish, circles = Tissue Moisture cultured fish) from this study. Panel (A) shows the strong relationship between As with relative weight, the more heavily parameterized mod- moisture and lipid expressed as a percentage of wet weight (ww). Model selection exercises suggested that those including a combination of source and length els ranked higher than those with lipid alone in terms of AICc are preferred in terms of their AICc weights (Table 2), but the improvement is (Table 2). A model containing lipid, sex, length, and source was marginal and offers little in practical terms. Panel (B) shows lower moisture and weighted as having a 73% chance of being better than the other lipid values were characteristic of the wild fish used in this study; wild fish also candidates. Evidence ratios suggest that this model provides generally had greater TL than cultured fish. NUTRITION OF STRIPED BASS 475

lipid and moisture content between wild and cultured fish and (79.97 ± 0.16%) (Figure 3, lower panel). Thus, we propose length in this data set. The larger fish tend to be wild collected a moisture threshold of 80% to indicate muscle lipid deple- and have lower lipid and higher moisture levels (Figure 3, tion (L0). lower panel). The equation demonstrates that while the model is pointing out these differences in length and source, the two variables have opposing beta weights that tend to diminish their Body Fat Index impact (source code = 1 for wild fish) and yield predictions As with relative weight, model selection for the BFI sug- nearly identical to those of the lipid-only model. Thus, little is gests only marginal improvement by more heavily parameter- gained from a practical standpoint by using the more heavily ized models over the one using lipid alone. A model containing parameterized model. lipid and length proved to rank highest but is only 40% more Solving the moisture equation for the point of lipid depletion likely to be the most appropriate model (Table 2). Evidence ra- (L0) for the various combinations of length, sex, and source tios suggest that the model is only 2.3 times more likely to be demonstrated a slightly lower moisture threshold for cultured the best-performing model with respect to the one containing than wild fish (–1%) and similar declines with increasing length lipid, length, and sex and 2.6 times more likely with respect to a (–1.6%; the difference in maximum versus minimum length in lipid-only model (Table 2). Interestingly, the classification sys- this data set), and being female (–0.24%). The mean moisture tem of 0–3 roughly approximates the mean lipid concentration threshold representing lipid depletion (L0) from the best model determined analytically for each class in milligrams per gram for wild fish was 80.05 ± 0.47%, which is virtually identical of wet weight (Table 3). However, the ability of the model to to the intercept of the moisture–lipid-only regression model correctly classify higher categories was marginal, with 54–57%

TABLE 3. Alternative approaches for categorization of observed visceral fat (BFI) and relationship to observed lipid and moisture in anterior dorsal muscle of Striped Bass. A relative score was assigned to each Striped Bass based on the visual prevalence of visceral storage lipids as follows: 0 = no detectable storage lipids, 1 = lipids present but less than 25% of viscera covered, 2 = 25–75% of viscera covered, and 3 = approximately 75% or more of viscera covered; Int = intercept. Three approaches are compared: (1) the original categorization, (2) fish with scores 0, 1, and 2 + 3 combined, and (3) presence or absence. Reduction of the number of categories results in reduced classification error and clear separation of lipid and moisture among classes. Within rows, values with the same letter not significantly different (P > 0.05).

Percent correct classiﬁcation and component means, by score Model Parameter Estimate SE 0 1 2 3 Full Int 0 1.62 0.25 80.61 64.71 57.45 53.85 Int 1 4.27 0.44 Int 2 7.33 0.68 Lipid 2.18 0.21 Mean moisture 80.05 z 77.77 y 76.53 x 75.63 x SD 0.189 0.245 0.264 0.398 Mean lipid 0.18 x 1.29 y 2.35 z 2.83 z SD 0.076 0.099 0.107 0.161

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 N 82 44 45 18 0, 1, and 2 + 3 Int 0 1.79 0.27 81.25 62.50 76.56 Int 1 4.63 0.52 Lipid −2.48 0.27 Mean moisture 80.05 z 77.77 y 76.25 x SD 0.19 0.247 0.22 Mean lipid 0.18 x 1.29 y 2.45 z SD 0.077 0.101 0.09 N 82 44 63 Presence/absence Int 0 −2.33 0.36 94.06 83.52 Lipid 4.02 0.67 Mean moisture 80.05 z 76.93 y SD 0.200 0.173 Mean lipid 0.18 y 1.96 z SD 0.092 0.080 N 82 107 476 JACOBS ET AL.

correct classification for classes 2 and 3 (Table 3). In addi- From our data, we clearly demonstrate that two common tion, classes 2 and 3 were not significantly different from each meristic measures of robustness, Fulton’s condition factor (K) other in total lipid or moisture levels (P > 0.05; Table 3). Re- and relative weight (Wr), are only coarsely related to lipid con- ducing the BFI to three classes by combining classes 2 and centration in Striped Bass. For nearly a century, Fulton’s condi- 3 (0, low lipid [<25%], and high lipid [>25%]) resulted in a tion factor has been a common means for comparing the robust- roughly 20% improvement in correct classification of high lipids ness of fish (Nash et al. 2006). However, it has received its share and little change in the other categories. Means comparison of of criticism due largely to the assumption of isometric growth associated lipid and moisture values also demonstrated clear and, to a lesser extent, the inability to compare species (Pope separation of BFI classes (Table 3). A simple presence/absence and Kruse 2007). Relative weight has been largely adopted by model was also constructed that resulted in 84% and 94% cor- fisheries management, especially in inland fisheries (Brown and rect classifications, respectively. Fish given a BFI score of 0 Murphy 1991a, 1991b; Blackwell et al. 2000; Pope and Kruse consisted of 0.18 ± 0.076 (mg/g ww; mean ± SE) tissue lipid 2007). However, the two indicators appear to provide identical and 80.05 ± 0.19% moisture, which is identical to the pro- information for Striped Bass on different scales. This strong posed moisture threshold. Thus, both the moisture and body fat correlation stems from the fact that the slope of the equation indicators appear capable of classifying lipid depletion. for Striped Bass derived by Brown and Murphy (1991b) is almost exactly 3 (3.007). Thus, the Ws for the species is a cubic Target function of length plus a constant. For the size range of fish we We propose an initial target for moisture derived from ref- used in this analysis, both indices describe the same isometric erence data collected in the late fall of 1990 from the Potomac relationship. River and upper Chesapeake Bay (n = 47; 78.37% moisture; While our model selection exercises demonstrated only a Karahadian et al. 1995). The proportion of fish in the 1990 data relatively minor improvement with the inclusion of source (cul- that would have exceeded the moisture threshold determined tured versus wild) and sex, external influences on the nature here as indicating lipid depletion (80%) was calculated based of the relationship between lipid and relative weight have been on the model intercept and using the minimum (1.08), mean noted. Copeland and Carline (2004) found the relationship be- (1.81), and maximum (2.33) standard deviations derived from tween lipid and condition indices to vary significantly by source, the 1998–2001 Chesapeake Bay data. The one-tailed propor- season, and lake among populations of Walleye Sander vit- tions of the data predicted to exceed a given Z for moisture, or reus fingerlings. In Bluegills Lepomis macrochirus, the lipid– Q score, were 7, 19, and 25%, respectively. Thus, a conserva- condition relationship may also change with respect to source as tive approximation of the lipid distribution in fish collected in well as feeding history and current bioenergetic state (Copeland 1990 would suggest that approximately 75% of those individ- et al. 2010). In principle, stressors (captivity or natural) or uals would have contained moisture below the threshold. For changes in energetic trajectory (feeding versus starving) can comparison, 64% of the wild-collected fish used in this study influence the allocation strategy of individuals between storage, had moisture values below the threshold, and 72% were classi- maintenance, and growth. This relationship enhances variabil- fied as having no observable body fat (BFI = 0). The BFI was not ity in weight and length within a population and contributes applied to our reference population, precluding the development to the poor correlation often seen between condition indices of a similar target for the index at this time. and proximate composition in wild fish (Copeland and Carline 2004; Trudel et al. 2005; Copeland et al. 2010), contradicting the results of controlled laboratory investigations (Love 1970;

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 DISCUSSION Brown and Murphy 1991a; Jacobs et al. 2008). We have demonstrated consistently strong relationships be- As with relative weight, the more heavily parameterized mod- tween two efficient and sensitive indices of nutritional health and els for moisture ranked higher in terms of AICc score and weight measured lipids in Striped Bass, tissue moisture and the BFI. than one using lipid alone. However, the models are likely point- Lipid depletion was determined to occur at approximately 80% ing out differences in the data sets used in this study rather than moisture and was highly correlated with the visual observation functional or physiological relationships. The data sets were in- of zero body fat (BFI = 0), which we propose as thresholds of tentionally combined in this effort in order to represent the full lipid depletion for the species. Further, we propose an interim range of physiological potential for lipid and moisture, which target of 75% of the population containing moisture values be- was not possible from any given study. It is not surprising that low the moisture threshold as a management goal. A similar tissue moisture provided a strong indicator of tissue lipid in target for BFI was not developed due to lack of reference data. Striped Bass. Interrelationships of chemical constituents in fish However, the BFI has been in continuous use by the MDNR have been well established in numerous fish species (Love 1970; since 1998, allowing for continuity and subsequent retrospec- Caulton and Bursell 1977; Jobling 1980; Jezierska et al. 1982; tive analysis with respect to benchmarks. To our knowledge, Black and Love 1986; Salam et al. 2000), including Striped Bass this effort represents the first attempt to develop nutrition-based (Hartman and Margraf 2008; Jacobs et al. 2008). As lipids are biological reference points for an estuarine fish. used during periods of fasting, they are replaced with water in a NUTRITION OF STRIPED BASS 477

linear fashion (Love 1970; Black and Love 1986) and metabolic individual goals. While less refined than measuring moisture, rate and activity decline to conserve energy reserves (Beamish the simple proportion of fish containing no lipid reserves could 1964; Ince and Thorpe 1976). During prolonged periods of star- be readily employed as another target for fisheries management. vation, white muscle is catabolized as a protein source for amino Based on a statistical estimation of index values for our 1990 acid generation and gluconeogenesis with a corresponding de- reference moisture data, we suggest that between 7% and 25% cline in tissue glycolytic enzyme activity (Lowery et al. 1987). of the fish would have been depleted of lipid reserves if concur- Finally, physiological limits are reached, organ function ceases, rent foraging conditions were considered adequate. From these and mortality ensues (Simpkins et al. 2003). observations, we conservatively propose an initial reference tar- Precise, direct measurement of lipids is time-consuming and get of 75% of fish containing tissue moisture concentrations of costly and requires dedicated equipment, laboratory space, and less than the lipid depletion threshold (L0) for moisture (80% staff for analysis (Jacobs et al. 2008). For this reason, analysis moisture). It is important to emphasize that the reference val- of moisture content is more attractive in that it can be conducted ues provided are for fall (October–November)-collected fish, as with simple equipment (e.g., a scale and drying oven) with feeding patterns change seasonally in Striped Bass (Hartman high throughput and good precision. In this analysis, we have and Brandt 1995b; Overton et al. 2009). Striped Bass are oppor- consistently used muscle tissue rather than whole-fish prepara- tunistic feeders, consuming a variety of prey items throughout tions. This protocol not only allows for direct comparison with the year and, depending on the water temperature, they may feed the work of Karahadian et al. (1995) but simplifies the process throughout the winter (Overton et al. 2009). However, the fall immensely by allowing for faster drying times to completion generally represents a period when Striped Bass forage heavily and ease of tissue homogenization. Previously, Jacobs et al. in preparation for overwintering and subsequent spring spawn- (2008) demonstrated strong linear relationships between whole ing. Future efforts to follow the nutritional status of Striped Bass fish and both muscle tissue and abdominal wall tissue with re- throughout the year are warranted and may provide needed inspect to proximate components, allowing for direct conversion sight into the variability of nutritional status and outcomes. when whole-fish energy estimates are required. The interre- It is also important to again note that these reference esti- lationship of proximate components in Striped Bass has also mates were derived using errors for Striped Bass moisture con- been described for whole Striped Bass by Hartman and Margraf tent from a different collection than the 1990 study. Although (2008). the 1998–2001 collections used for error estimation had similar The simple BFI modified from that proposed by Goede and error structures among those years and fish were collected from Barton (1990) provides another reliable index of measured tissue the same regions of the Chesapeake Bay at the same time of lipid. The visual observation and classification strongly relate year and were similar in size to the fish from the 1990 collec- to measured lipid concentration in milligrams per gram of wet tions, the true distributions of moisture and lipid values from weight and provide a simple, rapid means of evaluating lipid re- the 1990 fish are unknown. However, this is the only historic serves when fish are to be sacrificed. Although it is a subjective proximate composition data available from wild Chesapeake index, the presence or absence of visceral lipids is subject to Bay Striped Bass, and we submit that the 1990 sample and less error than discerning relative quantity. This is apparent in associated lipid presence represent an appropriate interim ref- our data, as the higher index classes did not differ significantly erence point. Our argument is based on the fact that bay waters yet classes 0–2 provided distinct separation. Alternatively, the were still under a moratorium for the harvest of Striped Bass index can be reduced to three classes representing the absence and the population was rebuilding and well below historic highs of lipid (0), low lipid (1; up to 25% coverage of viscera), and in the late 1990s (Richards and Rago 1999). During this time,

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 elevated lipid (2 + 3; approximately 25% or more coverage primary forage species were relatively abundant, resulting in of the viscera) if warranted by individual research and moni- improved predator–prey ratios for adult Striped Bass (Uphoff toring needs to allow for complete separation of categories. A 2003). While mycobacteriosis was detected in Chesapeake Bay model is also presented for the simple presence/absence of lipid Striped Bass as early as 1984 (Jacobs et al. 2009a), there is little reserves which offers exceptional discriminatory capability in indication of elevated prevalence until the late 1990s (Jacobs classiﬁcation but at the cost of reduced information. et al. 2009c). Finally, variance in weight at length and length The advantages of using BFI are numerous. The index pro- at age has increased since 1990 in adult Striped Bass (Uphoff vides an immediate assessment of nutritional status without 2003; Warner and Versak 2008). Given the relationship between the need of expensive laboratory equipment or technician time. lipid and weight and length-based condition indices, it is reason- Little technical expertise is required for its application, opening able to assume that, if anything, the variance in moisture would opportunities for nontraditional observers to aid in data col- be greater in our data than in that of the 1990 study. Thus, our lection. Most importantly, the index is a strong and validated proposed target based on maximum error estimates for moisture indicator of lipid concentration. While classiﬁcation error will from the 1998–2001 collection should represent a conservative increase with the number of classes employed, there is little estimate. incentive to alter the index a priori. Postprocessing exercises Whereas future efforts to apply these indicators to Chesa- such as those conducted here can be employed as needed for peake and other Striped Bass populations will aid in assessing 478 JACOBS ET AL.

the relative condition of stocks and adjusting targets, our ref- demonstrated to be negatively impacted by poor adult nutri- erence values provide a solid interim goal. The proportion of tion (Izquierdo et al. 2001). With fish such as Striped Bass that Striped Bass with body fat indices of 0 in the 1998–2001 fall undergo long seasonal migrations, ensuring adequate energy wild fish collection used in this study was 72%, with a moisture reserves is critical. Spawning stock biomass may also be influ- concentration of 79.52 ± 1.79% (mean + SD). Thus, dur- enced by poor condition through increased age of maturation ing this time period, lipid concentrations were low in Striped (Morgan 2004). This consideration is critical in the develop- Bass, with 36% of fish exceeding the moisture threshold. In ment of size and slot restrictions on the harvest of Striped Bass comparison, fish from our 1990 reference population had a and in protecting females through the first spawn. The preserva- mean moisture content of 78.37%, with a conservative esti- tion of the diversity in the age and behavioral structures of the mate of 25% exceeding the moisture threshold. The difference spawning stock is largely attributed to the successful restoration in moisture content (1.15%) represents a 1% change in mean of the Atlantic coast and Chesapeake Bay stocks (Secor 2000). lipid content—in this case, the difference between having lipid However, there is evidence in surveys conducted by the MDNR reserves and not. The range of expected moisture values is rel- (Warner and Versak 2008) that age at maturity in female Striped atively narrow, and minor changes in moisture percentage are Bass has increased since the early 1990s. significant. Thus, methods for determination need to be precise. With the exception of winter mortality, it is likely that few fish The method for moisture determination used in this study has a truly die of starvation but rather from associated processes. Thus, measurement error of 0.3%, which offers sufficient precision for our proposed threshold value for Striped Bass (80% moisture) determining changes in moisture content. The MDNR has been represents a state of vulnerability rather than lethality. Reduc- applying the BFI continuously since 1998, and full proximate tion in energy intake or starvation compromises both innate and composition for 2010–2011. This analysis is forthcoming and adaptive immune responses, leading to reductions in disease re- will lend great insight into annual variability in lipid content sistance or enhanced severity of infection (Blazer 1991; Lim and and the long-term nutritional status of Chesapeake Bay Striped Klesius 2003; Shoemaker et al. 2003; Jones et al. 2008; Jacobs Bass. et al. 2009b). This is of particular concern in Chesapeake Bay The proposal and application of interim target values for fish- Striped Bass due to their history of disease issues and the cur- eries management is common in the context of precautionary rently elevated prevalence of mycobacteriosis (Baya et al. 1990; approaches to fisheries management (FAO 1995). For example, Kaattari et al. 2005; Gauthier et al. 2008; Jacobs et al. 2009c). in recent years interim fishing mortality targets have been pro- Jacobs et al. (2009b) demonstrated that reductions in nutritional posed in the Chesapeake region for both blue crabs Callinectes state adversely affect the progression, severity, and reactivation sapidus and Atlantic Menhaden (CBSAC 2008; ASMFC 2011). of acute inflammation associated with mycobacteriosis in con- Their application acknowledges the need to reduce uncertainty trolled experimental studies with Striped Bass. Interestingly, the but concedes that management based upon the best available reactivation of disease and the onset of disease-associated mor- science is both appropriate and necessary as an interim, precau- tality occurred at tissue moisture values of 79.83 ± 0.36%, or tionary measure. approximately the point of lipid exhaustion.

Implications of Lipid Depletion Mortality associated with lipid depletion has been well doc- Management Considerations umented in association with overwintering, particularly in ﬁrst- Using nutritional reference points for Striped Bass manage- year ﬁsh in temperate environments (Hurst and Conover 1998, ment in Chesapeake Bay requires a broader management process

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 2001, 2003; Fullerton et al. 2000; Biro et al. 2004; Hurst 2007). and perspective than is currently encompassed in single-species Size and energy reserves are consistently identified as factors of- management (Maryland Sea Grant 2009). The current manage- fering a competitive advantage for surviving cold temperatures ment of Striped Bass in Chesapeake Bay is designed to control (Schultz and Conover 1999; Fullerton et al. 2000; Hurst 2007), fishing mortality in order to maintain spawning stock biomass and starvation is considered the primary source of overwinter and consists of interstate management coordinated by the At- mortality (Shuter and Post 1990; Hurst and Conover 2003). lantic States Marine Fisheries Commission. The Chesapeake While less effort has been taken with other age-classes, winter Bay stock is assessed and managed along with the Hudson and mortality can influence adult fish as well, altering the popula- Delaware River stocks as a single Atlantic coast stock. This tion and ecosystem dynamics (Hurst 2007). While the causes strategy does not address dynamics such as predation, compe- (disease, etc.) and the potential contribution of nutrition are un- tition, lack of forage, and disease, nor does it address regional known, the natural mortality rates of Chesapeake Bay Striped problems for the contingent of Striped Bass (i.e., residents) Bass have increased since the early 1990s (Jiang et al. 2007; that remains in the Chesapeake Bay after spawning (Secor and Gauthier et al. 2008; Sadler et al. 2010). Piccoli 2007). Confounding issues of migration and mortality Not only may nutritional effects lead to mortality, they can further complicate the assessment of resident Striped Bass using also influence subsequent reproductive potential. Fecundity, fer- a technique such as the statistical catch-at-age model used for tilization, embryo development, and larval quality have all been the Atlantic coast. NUTRITION OF STRIPED BASS 479

We envision a nutritional threshold for resident Striped Bass the Chesapeake Bay and its tributaries. Journal of Fish Diseases 13:251– as part of a framework of indicators, targets, and thresholds for 253. managing in an ecological context (Maryland Sea Grant 2009). Beamish, F. W. H. 1964. Influence of starvation on standard and routine oxygen consumption. Transactions of the American Fisheries Society 93:103–107. Combined with specific growth indicators, such as length or Biro, P. A., A. E. Morton, J. R. Post, and E. A. Parkinson. 2004. Overwinter weight at age and relative prey abundance, this would provide a lipid depletion and mortality of age-0 Rainbow Trout (Oncorhynchus mykiss). complete picture terms of short- and long-term foraging success Canadian Journal of Fisheries and Aquatic Sciences 61:1513–1519. in relation to available resources. Further, habitat suitability in- Black, D., and R. M. Love. 1986. The sequential mobilisation and restoration dicators exist for the Chesapeake region, along with estimates of energy reserves in tissues of Atlantic Cod during starvation and refeeding. Journal of Comparative Physiology B 156:469–479. of disease severity and impact that may offer further explana- Blackwell, B. G., M. L. Brown, and D. W. Willis. 2000. Relative weight (Wr) tion of current and long-term condition (Costantini et al. 2008; status and current use in fisheries assessment and management. Reviews in Gauthier et al. 2008). Sainsbury (1998) advocated formulat- Fisheries Science 8:1–44. ing multiple hypotheses about stock status and evaluating them Blazer, V. S. 1991. Piscine macrophage function and nutritional influences: a with empirical data, while Hilborn (2003) anticipated moving review. Journal of Aquatic Animal Health 3:77–86. Boos, D. D., and C. Brownie. 2004. Comparing variances and other measures towards rules specified ahead of time based directly on data or of dispersion. Statistical Science 19:571–578. simple models. Using either approach, the basic monitoring data Brown, M. L., D. M. Gatlin III, and B. R. Murphy. 1993. Nondestructive mea- exist for the Chesapeake Bay with which to begin employing surement of Sunshine Bass, Morone chrysops (Rafinesque) × Morone sax- holistic approaches to stock management. atilis (Walbaum), body composition using electrical conductivity. Aquacul- Within the context of ecosystem-based management of fish- ture Research 24:585–592. Brown, M. L., and B. R. Murphy. 1991a. Relationship of relative weight (Wr) eries, it is necessary to augment existing single-species indi- to proximate composition of juvenile Striped Bass and hybrid Striped Bass. cators with those that provide connectivity to other aspects of Transactions of the American Fisheries Society 120:509–518. the ecosystem (Pikitch et al. 2004; Link 2005). The proposed Brown, M. L., and B. R. Murphy. 1991b. Standard weights (Ws) for Striped Bass, indicators and reference values provide just such a mechanism White Bass, and hybrid Striped Bass. North American Journal of Fisheries for relating nutritional status to other indicators of stock health Management 11:451–467. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel and abundance. The functional management reality of using a inference: a practical information-theoretic approach, 2nd edition. Springer- moisture index as a biological reference point is that it is less Verlag, New York. time-consuming and expensive than other empirical reference Burnham, K. P., D. R. Anderson, and K. P. Huyvaert. 2011. AIC model selection points (e.g., proximate analysis) and it is more accurate than tra- and multimodel inference in behavioral ecology: some background, observa- ditional indicators (e.g., morphometric indicators). As we move tions, and comparisons. Behavioral Ecology and Sociobiology 65:23–35. Caulton, M. S., and E. Bursell. 1977. The relationship between changes in forward with holistic management approaches, employing tools condition and body composition in young Tilapia rendalli Boulenger. Journal such as this will allow not only for sustainable resources but for of Fish Biology 11:143–150. healthy, viable ecosystems. CBSAC (Chesapeake Bay Stock Assessment Committee). 2008. 2008 Blue crab advisory report. CBSAC, Annapolis, Maryland. Copeland, T., and R. F. Carline. 2004. Relationship of lipid content to size and ACKNOWLEDGMENTS condition in Walleye fingerlings from natural and aquacultural environments. Because of the 15-year span and variety of experiments rep- North American Journal of Aquaculture 66:237–242. resented by this effort, many individuals contributed to the col- Copeland, T., B. R. Murphy, and J. J. Ney. 2010. The effects of feeding history and environment on condition, body composition, and growth of Bluegills lection and processing of these fish. In no particular order, we Lepomis macrochirus. Journal of Fish Biology 76:538–555. are grateful for the efforts of Bart Rodgers, Matt Rhodes, Brett Costantini, M., S. A. Ludsin, D. M. Mason, X. Zhang, W. C. Boicourt, and S. B. Coakley, Mark Matsche, Larry Pieper, Yuan Tian, Yusaf Hafvez, Brandt. 2008. Effect of hypoxia on habitat quality of Striped Bass (Morone Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 Stephen Jordan, and William Van Heukelem. The manuscript saxatilis) in Chesapeake Bay. Canadian Journal of Fisheries and Aquatic was much improved by the critical reviews of Ed Martino, Sciences 65:989–1002. Davidson, D., and C. T. Marshall. 2010. Are morphometric indices accurate Lonnie Gonsalves, and the anonymous reviewers, editor, and indicators of stored energy in Herring Clupea harengus? Journal of Fish associate editor provided by this journal. Biology 76:913–929. Dutil, J. D., and Y. Lambert. 2000. Natural mortality from poor condition in Atlantic Cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic REFERENCES Sciences 57:826–836. Adams, S. M. 1999. Ecological role of lipids in the health and success of fish FAO (Food and Agriculture Organization of the United Nations). 1995. Pre- populations. Pages 132–160 in M. T. Arts and B. C. Wainman, editors. Lipids cautionary approaches to fisheries, part 1. Guidelines on the precautionary in freshwater ecosystems. Springer-Verlag, New York. approach to capture fisheries and species introductions. FAO, Rome. AOAC (Association of Official Analytical Chemists). 2005. Official methods of Fu, C., R. Mohn, and L. P. Fanning. 2001. Why the Atlantic Cod (Gadus analysis, 17th edition. AOAC, Gaithersburg, Maryland. morhua) stock off eastern Nova Scotia has not recovered. Canadian Journal ASMFC (Atlantic States Marine Fisheries Commision). 2011. Addendum V to of Fisheries and Aquatic Sciences 58:1613–1623. amendment 1 to the Atlantic Menhaden fishery management plan: alternative Fullerton, A. H., J. E. Garvey, R. A. Wright, and R. A. Stein. 2000. Overwinter reference points and fishery management tools. ASMFC, Arlington, Virginia. growth and survival of Largemouth Bass: interactions among size, food, Baya, A. M., B. Lupiani, F. Hetrick, B. Roberson, R. Lukacovic, E. May, and origin, and winter severity. Transactions of the American Fisheries Society C. Poukish. 1990. Association of Streptococcus spp. with fish mortalities in 129:1–12. 480 JACOBS ET AL.

Gauthier, D. T., R. J. Latour, D. M. Heisey, C. F. Bonzek, J. Gartland, E. J. Jiang, H., K. H. Pollock, C. Brownie, J. M. Hoenig, R. J. Latour, B. K. Wells, Burge, and W. K. Vogelbein. 2008. Mycobacteriosis-associated mortality in and J. E. Hightower. 2007. Tag return models allowing for harvest and catch wild Striped Bass (Morone saxatilis) from Chesapeake Bay, USA. Ecological and release: evidence of environmental and management impacts on Striped Applications 18:1718–1727. Bass fishing and natural mortality rates. North American Journal of Fisheries Goede, R. W., and B. A. Barton. 1990. Organismic indices and an autopsy-based Management 27:387–396. assessment as indicators of health and condition of fish. Pages 93–108 in Jobling, M. 1980. Effects of starvation on proximate chemical composition and S. M. Adams, editor. Biological indicators of stress in fish. Amererican Fish- energy utilization of Plaice, Pleuronectes platessa L. Journal of Fish Biology eries Society, Symposium 8, Bethesda, Maryland. 17:325–334. Griffin, J. C., and F. J. Margraf. 2003. The diet of Chesapeake Bay Striped Bass Jones, S. R. M., M. D. Fast, and S. C. Johnson. 2008. Influence of reduced feed in the late 1950s. Fisheries Management and Ecology 10:323–328. ration on Lepeophtheirus salmonis infestation and inflammatory gene ex- Hartman, K. J., and S. B. Brandt. 1995a. Estimating energy density of fish. pression in juvenile Pink Salmon. Journal of Aquatic Animal Health 20:103– Transactions of the American Fisheries Society 124:347–355. 109. Hartman, K. J., and S. B. Brandt. 1995b. Predatory demand and impact of Kaattari, I. M., M. W. Rhodes, H. Kator, and S. L. Kaattari. 2005. Comparative Striped Bass, Bluefish, and Weakfish in the Chesapeake Bay: applications of analysis of mycobacterial infections in wild Striped Bass Morone saxatilis bioenergetics models. Canadian Journal of Fisheries and Aquatic Sciences from Chesapeake Bay. Diseases of Aquatic Organisms 67:125–132. 52:1667–1687. Karahadian, C., K. P. Fowler, and D. H. Cox. 1995. Comparison of chemical Hartman, K. J., and F. J. Margraf. 2008. Common relationships among proximate composition of Striped Bass (Morone saxatilis) from three Chesapeake Bay composition components in fish. Journal of Fish Biology 73:2352–2360. tributaries with those of two aquaculture hybrid Striped Bass types. Food Herbinger, C. M., and G. W. Friars. 1991. Correlation between condition factor Chemistry 54:409–418. and total lipid content in Atlantic Salmon, Salmo salar L., parr. Aquaculture Lambert, Y., and J. D. Dutil. 1997. Condition and energy reserves of Atlantic Research 22:527–529. Cod (Gadus morhua) during the collapse of the northern Gulf of St. Lawrence Hilborn, R. 2003. The state of the art in stock assessment: where we are and stock. Canadian Journal of Fisheries and Aquatic Sciences 54:2388– where we are going. Scientia Marina 67:15–20. 2400. Hurst, T. P. 2007. Causes and consequences of winter mortality in fishes. Journal Lilly, G. R. 1994. Predation by Atlantic Cod on Capelin on the southern Labrador of Fish Biology 71:315–345. and Northeast Newfoundland shelves during a period of changing spatial Hurst, T. P., and D. O. Conover. 1998. Winter mortality of young-of-the-year distributions. ICES Marine Science Symposium 198:600–611. Hudson River Striped Bass (Morone saxatilis): size-dependent patterns and Lim, C., and P. H. Klesius. 2003. Influence of feed deprivation on hematology, effects on recruitment. Canadian Journal of Fisheries and Aquatic Sciences macrophage chemotaxis, and resistance to Edwardsiella ictaluri challenge of 55:1122–1130. Channel Catfish. Journal of Aquatic Animal Health 15:13–20. Hurst, T. P., and D. O. Conover. 2001. Activity-related constraints on overwin- Link, J. S. 2005. Translating ecosystem indicators into decision criteria. ICES tering young-of-the-year Striped Bass (Morone saxatilis). Canadian Journal Journal of Marine Science 62:569–576. of Zoology 79:129–136. Love, R. M. 1970. The chemical biology of fishes. Academic Press, London. Hurst, T. P., and D. O. Conover. 2003. Seasonal and interannual variation in the Lowery, M. S., S. J. Roberts, and G. N. Somero. 1987. Effects of starvation on allometry of energy allocation in juvenile Striped Bass. Ecology 84:3360– the activities and localization of glycolytic enzymes in the white muscle of the 3369. BarredSandBassParalabrax nebulifer. Physiological Zoology 60:538–549. Ince, B. W., and A. Thorpe. 1976. The effects of starvation and force feeding on Mace, P. M. 1994. Relationships between common biological reference points the metabolism of the Northern Pike, Esox lucius L. Journal of Fish Biology used as thresholds and targets of fisheries management strategies. Canadian 8:79–88. Journal of Fisheries and Aquatic Sciences 51:110–122. Izquierdo, M. S., H. Fernandez-Palacios, and A. G. J. Tacon. 2001. Effect Maryland Sea Grant. 2009. Ecosystem-based fisheries management for Chesa- of broodstock nutrition on reproductive performance of fish. Aquaculture peake Bay: Striped Bass species team background and issues briefs. Mary- 197:25–42. land Sea Grant, College Park, Maryland. Available: http://www.mdsg. Jacobs, J. M. 2007. Mycobacteriosis in Chesapeake Bay Striped Bass (Mo- umd.edu/store/reports/ebfm/. (March 2013). rone saxatilis): the interaction of nutrition and disease. Doctoral dissertation. Menard, S. 2000. Coefficients of determination for multiple logistic regression University of Maryland, College Park. analysis. Amercian Statistician 54:17–24. Jacobs, J. M., D. W. Howard, M. R. Rhodes, M. W. Newman, E. B. May, and Morgan, M. J. 2004. The relationship between fish condition and the probability R. M. Harrell. 2009a. Historical presence of mycobacteriosis in Chesapeake of being mature in American Plaice (Hippoglossoides platessoides). ICES Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 Bay Striped Bass (1975–1985). Diseases of Aquatic Organisms 85:181–186. Journal of Marine Science 61:64–70. Jacobs, J. M., S. Lindell, W. Van Heukelem, E. M. Hallerman, and R. M. Harrell. Myers, R. A., J. A. Hutchings, and N. J. Barrowman. 1996. Hypotheses for 1999. Strain evaluation of Striped Bass (Morone saxatilis) under controlled the decline of cod in the North Atlantic. Marine Ecology Progress Series conditions. Aquaculture 173:171–177. 138:293–308. Jacobs, J. M., M. R. Rhodes, A. Baya, R. Reimschuessel, H. Townsend, and Nash, R. D. M., A. H. Valencia, and A. J. Geffen. 2006. The origin of Fulton’s R. M. Harrell. 2009b. Influence of nutritional state on the progression and condition factor: setting the record straight. Fisheries 31:236–238. severity of mycobacteriosis in Striped Bass Morone saxatilis. Diseases of Niimi, A. J. 1972. Changes in the proximate body composition of Largemouth Aquatic Organisms 87:183–197. Bass (Micropterus salmoides) with starvation. Canadian Journal of Zoology Jacobs, J. M., M. R. Rhodes, B. Martin, D. McIntosh, W. F. Van Heukelem, and 50:815–819. R. M. Harrell. 2008. Tissue-specific sampling for the estimation of Striped Overton, A. S., F. J. Margraf, and E. B. May. 2009. Spatial and temporal patterns Bass whole-body proximate composition. Transactions of the American Fish- in the diet of Striped Bass in Chesapeake Bay. Transactions of the American eries Society 137:386–392. Fisheries Society 138:915–926. Jacobs, J. M., C. B. Stine, A. M. Baya, and M. L. Kent. 2009c. A review of Overton, A. S., F. J. Margraf, C. A. Weedon, L. H. Pieper, and E. B. May. 2003. mycobacteriosis in marine fish. Journal of Fish Diseases 32:119–130. The prevalence of mycobacterial infections in Striped Bass in Chesapeake Jakob, E. M., S. D. Marshall, and G. W. Uetz. 1996. Estimating fitness: a Bay. Fisheries Management and Ecology 10:301–308. comparison of body condition indices. Oikos 77:61–67. Pikitch, E. K., C. Santora, E. A. Babcock, A. Bakun, R. Bonfil, D. O. Conover, Jezierska, B., J. R. Hazel, and S. D. Gerking. 1982. Lipid mobilization during P. Dayton, P. Doukakis, D. Fluharty, B. Heneman, E. D. Houde, J. Link, P. starvation in the Rainbow Trout, Salmo gairdneri Richardson, with attention A. Liningston, M. Mangel, M. K. McAllister, J. Pope, and K. J. Sainsbury. to fatty acids. Journal of Fish Biology 21:681–692. 2004. Ecosystem-based fishery management. Science 305:346–347. NUTRITION OF STRIPED BASS 481

Plante, S., C. Audet, Y. Lambert, and J. de la Noue.¨ 2005. Alternative methods Shelton, P. A., and G. R. Lilly. 2000. Interpreting the collapse of the northern for measuring energy content in Winter Flounder. North American Journal cod stock from survey and catch data. Canadian Journal of Fisheries and of Fisheries Management 25:1–6. Aquatic Sciences 57:2230–2239. Pope, K. L., and C. G. Kruse. 2007. Condition. Pages 423–471 in C. S. Guy Shoemaker, C. A., P. H. Klesius, C. Lim, and M. Yildirim. 2003. Feed de- and M. L. Brown, editors. Analysis and interpretation of freshwater fisheries privation of Channel Catfish, Ictalurus punctatus (Rafinesque), influences data. American Fisheries Society, Bethesda, Maryland. organosomatic indices, chemical composition, and susceptibility to Flavobac- Pruell, R. J., B. K. Taplin, and K. Cicchelli. 2003. Stable isotope ratios in terium columnare. Journal of Fish Diseases 26:553–561. archived Striped Bass scales suggest changes in trophic structure. Fisheries Shuter, B. J., and J. R. Post. 1990. Climate, population viability, and the zoo- Management and Ecology 10:329–336. geography of temperate fishes. Transactions of the American Fisheries Soci- Richards, R. A., and P. J. Rago. 1999. A case history of effective fishery man- ety 119:314–336. agement: Chesapeake Bay Striped Bass. North American Journal of Fisheries Simpkins, D. G., W. A. Hubert, C. M. Del Rio, and D. C. Rule. 2003. Physiolog- Management 19:356–375. ical responses of juvenile Rainbow Trout to fasting and swimming activity: Rose, G. A., and R. O’Driscoll. 2002. Capelin are good for cod: can the northern effects on body composition and condition indices. Transactions of the Amer- stock rebuild without them? ICES Journal of Marine Science 59:1018–1026. ican Fisheries Society 132:576–589. Sadler, P. W., J. M. Hoenig, R. E. Harris, R. J. Wilk, and L. M. Goins. 2010. Sissenwine, M. P., and J. G. Shepherd. 1987. An alternative perspective on Evaluation of Striped Bass stocks in Virginia: monitoring and tagging studies, recruitment overfishing and biological reference points. Canadian Journal of 2005–2009. Virginia Institute of Marine Science, Gloucester Point. Fisheries and Aquatic Sciences 44:913–918. Sainsbury, K. 1998. Living marine resource assessment for the 21st century: Stevenson, R. D., and W. A. Woods. 2006. Condition indices for conservation: what will be needed and how will it be provided? Pages 1–40 in F. Funk, T. J. new uses for evolving tools. Integrative and Comparative Biology 46:1169– Quinn II, J. Heifetz, J. N. Ianelli, J. E. Powers, J. F. Schweigert, P. J. Sullivan, 1190. and C.-I. Zhang, editors. Fishery stock assessment models. Alaska Sea Grant Trudel, M., S. Tucker, J. F. T. Morris, D. A. Higgs, and D. W. Welch. 2005. College Program, AK-SG-98-01, University of Alaska, Fairbanks. Indicators of energetic status in juvenile Coho Salmon and Chinook Salmon. Salam, A., M. Ali, and S. Masud. 2000. Effects of various food deprivation North American Journal of Fisheries Management 25:374–390. regimes on body composition dynamics of Thaila, Catla catla. Journal of Uphoff, J. H. 2003. Predator–prey analysis of Striped Bass and Atlantic Men- Research Bahauddin Zakariya University 11:26–32. haden in upper Chesapeake Bay. Fisheries Management and Ecology 10:313– Schultz, E. T., and D. O. Conover. 1999. The allometry of energy reserve 322. depletion: test of a mechanism for size-dependent winter mortality. Oecologia Walsberg, G. E. 1988. Evaluation of a nondestructive method for determining 119:474–483. fat stores in small birds and mammals. Physiological Zoology 61:153–159. Secor, D. H. 2000. Spawning in the nick of time? Effect of adult demographics Walter, J. F., A. S. Overton, K. H. Ferry, and M. E. Mather. 2003. Atlantic coast on spawning behaviour and recruitment in Chesapeake Bay Striped Bass. feeding habits of Striped Bass: a synthesis supporting a coastwide understand- ICES Journal of Marine Science 57:403–411. ing of trophic biology. Fisheries Management and Ecology 10:349–360. Secor, D. H., and P. M. Piccoli. 2007. Oceanic migration rates of upper Chesa- Warner, L., and B. Versak. 2008. Characterizaton of Striped Bass spawning peake Bay Striped Bass (Morone saxatilis) determined by otolith micro- stocks in Maryland. Pages 159–206 in Chesapeake Bay finfish/habitat inves- chemical analysis. U.S. National Marine Fisheries Service Fishery Bulletin tigations. Maryland Department of Natural Resources, Federal Aid in Sport 105:62–73. Fish Restoration, Project F-61-R4, Annapolis. Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 19:58 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Effectiveness of Three Compounds to Anesthetize Rainbow Trout during PIT Tag Implantation Surgery Jacob L. Davis a , Michael E. Barnes b & Jerry W. Wilhite c a South Dakota Department of Game , Fish and Parks , 4130 Adventure Trail, Rapid City , South Dakota , 57702 , USA b South Dakota Department of Game , Fish and Parks , 19619 Trout Loop, Spearfish , South Dakota , 57783 , USA c Western Area Power Administration , Post Office Box 28213, Lakewood , Colorado , 80228 , USA Published online: 28 Apr 2013.

To cite this article: Jacob L. Davis , Michael E. Barnes & Jerry W. Wilhite (2013): Effectiveness of Three Compounds to Anesthetize Rainbow Trout during PIT Tag Implantation Surgery, North American Journal of Fisheries Management, 33:3, 482-487 To link to this article: http://dx.doi.org/10.1080/02755947.2013.768568

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:482–487, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.768568

MANAGEMENT BRIEF

Effectiveness of Three Compounds to Anesthetize Rainbow Trout during PIT Tag Implantation Surgery

Jacob L. Davis* South Dakota Department of Game, Fish and Parks, 4130 Adventure Trail, Rapid City, South Dakota 57702, USA Michael E. Barnes South Dakota Department of Game, Fish and Parks, 19619 Trout Loop, Spearﬁsh, South Dakota 57783, USA Jerry W. Wilhite Western Area Power Administration, Post Ofﬁce Box 28213, Lakewood, Colorado 80228, USA

Griffin 2004). Anesthesia is essential to those studies where tags Abstract need to be surgically implanted to understand fish movements, This study evaluated the efficacy of two potential zero- obtain population estimates, collect behavioral observations, or withdrawal anesthetics, Benzoak (20% benzocaine; 50, 60, and conduct any other studies where the tagged fish must survive and 75 mg/L) and Aqui-SE (50% eugenol; 50, 60, and 75 mg/L) compared with tricaine methanesulfonate (MS-222; 55, 80, and behave normally (Kelsch and Shields 1996). Field surgeries also 100 mg/L), to anesthetize Rainbow Trout Oncorhynchus mykiss for need anesthetics that do not require any withdrawal time prior to PIT tag implantation surgery. In general, higher doses resulted in release into aquatic environments where they could potentially faster induction time to stage 4 anesthesia (defined by the cessation be harvested for human consumption (Young 2009; Trushenski of reflex activity). At 204 s, the time to stage 4 anesthesia was slowest et al. 2012). using MS-222 at 55 mg/L, followed by 50 mg/L of either Benzoak and Aqui-SE, which in turn were significantly slower to induce this Many anesthetics used historically are now prohibited be- level of anesthesia than were Benzoak or Aqui-SE at 60 or 75 mg/L cause of carcinogenic or effectiveness concerns, as well as pos- or MS-222 at 80 mg/L. At 100 mg/L, MS-222 had the quickest time, sible undesirable physiological effects (Summerfelt and Smith 57 s, to stage 4 anesthesia. Time to recovery was longest for Rain- 1990; DeTolla et al. 1995; Trushenski et al. 2012). Currently, bow Trout exposed to any concentration of Aqui-SE and shortest tricaine methanesulfonate (MS-222) is the only anesthetic ap- for MS-222, and recovery times from Benzoak were intermediate. Although Rainbow Trout length and weight varied significantly proved by the U.S. Food and Drug Administration (FDA) for Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 among the treatments, time to anesthesia and recovery were more use on fish (Coyle et al. 2004; Trushenski et al. 2012). While dependent on the anesthetic and concentration used. In our opin- effective, MS-222 is relatively expensive and requires a 21-d ion, doses of either Benzoak or Aqui-SE of greater than 60 mg/L withdrawal period (Schnick et al. 1986; Trushenski et al. 2012). will induce rapid anesthesia and provide relatively quick recovery Carbon dioxide (CO ), while technically not approved by the times for adult Rainbow Trout. 2 FDA, has a been designated as a drug of low regulatory priority (LRP), which states that regulatory action is unlikely when the Anesthesia is typically required to humanely conduct sur- proper standards are met (Trushenski et al. 2012). Carbon diox- gical procedures on fish (Summerfelt and Smith 1990; Nickum ide does not require a withdrawal period, but its use has been et al. 2004). In addition to animal welfare considerations, proper problematic. While CO2 can be used as an anesthetic (Prince selection and use of an anesthetic is particularly necessary dur- et al. 1995; Gelwicks et al. 1998; Wagner et al. 2002), because of ing surgical tag implantation to alleviate stress effects from the insufficient efficacy (Marking and Meyer 1985; Gilderhus and procedure that would confound the results of the subsequent Marking 1987; Sanderson and Hubert 2007) and deleterious study (Summerfelt and Smith 1990; Mulcahy 2003; Davis and stressful effects (Marking and Meyer 1985; Iwama et al. 1989;

*Corresponding author: [email protected] Received July 27, 2012; accepted January 10, 2013 482 MANAGEMENT BRIEF 483

Bernier and Randall 1998; Taylor and Roberts 1999; Pirhonen (Western Chemical, Ferndale, Washington), and Benzoak (Fron- and Schreck 2003), the use of other anesthetics is recommended tier Scientific, Logan, Utah). (DeTolla et al. 1995). However, no other legal options for zero- Anesthetic evaluations occurred on three different dates dur- withdrawal anesthetics are available in the USA for fish that ing the summer of 2010, with a unique hatchery lot of Rainbow may be harvested and later consumed by humans. Trout used on each date. On the first sampling date, concentra- Recently, two commercially produced products have tions of 50 mg/L Aqui-SE and Benzoak and 55 mg/L MS-222 emerged as possible immediate-release anesthetics: Benzoak were used. On the second sampling date, the concentrations of (ACD Pharmaceuticals, Norway) and Aqui-SE (AQUI-S New Aqui-SE and Benzoak were increased to 60 mg/L, and MS-222 Zealand Ltd, New Zealand). Benzoak, containing 20% benzo- concentrations were increased to 80 mg/L. On the final sampling caine, can be an effective anesthetic for fish (Allen et al. 1994; date, 75 mg/L of both Aqui-SE and Benzoak and 100 mg/L of Trushenski et al. 2012) and acts by blocking sodium channels MS-222 were used. All anesthetics were mixed at a ratio of 30: and preventing the transmission of action potential (Burka et al. l with hatchery well water from the raceway where the fish were 1997; Kiessling et al. 2009). It is considered to be less soluble removed. than MS-222 (Ross and Ross 2008) and is more effective at Within each trial a total of 150 surgeries were performed, similar or lower concentrations (McFarland and Klontz 1969; and each anesthetic was used on 50 Rainbow Trout per trial. Ferreira et al. 1979; Hseu et al. 1998). Data collection began with the submersion of an individual fish Aqui-SE contains 50% eugenol as its active ingredient. in a bath of water containing anesthetic and the time to reach Eugenol makes up 90–95% of the compound clove oil (Briozzo stage 4 anesthesia was measured to the nearest second. Stage 4 et al. 1989), which is distilled from natural plants including anesthesia, defined by Hikasa et al. (1986) as the cessation of the leaves of clove trees Eugenia aromatica (Anderson et al. reflex activity, was determined when fish lost equilibrium and 1997). Additionally, clove oil has been shown to be an effective showed little to no movement, and could be easily caught by fish anesthetic (Taylor and Roberts 1999; Sladky et al. 2001). hand. Rainbow Trout were then removed from the anesthetic Eugenol can also be an effective anesthetic, and compared with bath, measured to nearest millimeter, weighed to the nearest MS-222 it has a shorter induction time to anesthesia and program, and placed ventral side up in a wooden tagging trough for duces a decreased stress response in fish at similar doses (Keene PIT tag implantation. The duration of the surgical procedure was et al. 1998; Sladky et al. 2001; Kildea et al. 2004; Ross and Ross measured to the nearest second and after the completion of the 2008). Its relatively long recovery time makes eugenol particu- surgery, individuals were placed in a holding cage in freshwater larly well suited for aquaculture facilities where time constraints in the raceway. Time to recovery from anesthesia was recorded often do not exist (Keene et al. 1998; Prince and Powell 2000). to the nearest second. In trials 2 and 3, sudden movement or Eugenol is listed in the FDA category of materials as “generally thrashing by Rainbow Trout during the surgical procedure, in- regarded as safe” (Ross and Ross 1999) and has been approved dicating possible loss of stage 4 anesthesia, was recorded. All as a food additive (WHO 1982). However, neither clove oil Rainbow Trout were observed for 21 d postanesthesia to deter- nor eugenol is approved for fisheries anesthetic use in the USA mine mortality. (USFDA 2002; Young 2009). Data were compiled and assessed for normality with a This study was undertaken because of the need for an alter- Kolmogorov–Smirnov test and homogeneity of variance using a native immediate-release anesthetic and the lack of information Folded F test. Differences in time to induction, time of surgery, using alternatives to MS-222 during fish surgery. The objective time to recovery, length, and weight of fish between the three of this study was to evaluate the effectiveness of two commer- drugs within individual trials and the different concentrations

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 cially produced, potential zero-withdrawal anesthetics, Benzoak of the same drug among the three separate trials were analyzed and Aqui-SE, during tag-implantation surgeries. using the SPSS (9.0) statistical analysis program (SPSS 1998), in which significance was predetermined at P < 0.05. One-way ANOVA was conducted and if the treatments were significantly different, pairwise mean comparisons were performed using the METHODS Tukey honestly significant difference means comparison proce- All experiments were carried out at McNenny State Fish dure (Kuehl 2000). Hatchery, Spearfish, South Dakota, using well water (11◦C; total hardness as CaCO3, 360 mg/L; alkalinity as CaCO3, 210 mg/L; pH, 7.6; total dissolved solids, 390 mg/L) and RESULTS hatchery-reared Rainbow Trout Oncorhynchus mykiss. As part Time to anesthesia of Rainbow Trout was significantly differ- of another project, these fish were undergoing surgical implan- ent among the treatments (Table 1). At 204 s, MS-222 at 55 mg/L tation of 23-mm PIT tags using methods described by Roussel took nearly twice as long to induce anesthesia than did 50 mg/L et al. (2000). During this procedure, we evaluated the effec- of Benzoak or Aqui-SE. All three anesthetics took significantly tiveness of three different chemical anesthetics, MS-222 (Ar- longer to produce stage 4 anesthesia at their lowest concentra- gent Chemical Laboratories, Ferndale, Washington), Aqui-SE tions. Rainbow Trout exposed to either 75 mg/L Aqui-SE or 484 DAVIS ET AL.

TABLE 1. Mean (SE) time to stage 4 anesthesia, time of surgery, time to full 60 mg/L of Benzoak or Aqui-SE or 80 mg/L of MS-222. In recovery, lengths, and weights of Rainbow Trout exposed to three different doses contrast, less than half of that number moved when anesthetized of Benzoak, Aqui-SE, and MS-222. Numbers of fish that moved during surgery are also listed. For each variable, values within a row and column with different with Benzoak or Aqui-SE at 75 mg/L, and only two Rainbow letters are significantly different (P < 0.05, n = 50); NA = not available. Trout exposed to 100 mg/L MS-222 moved. Time of surgery was significantly longer during the first trial Anesthetic compared with subsequent trials (P = 0.018). The mean time of surgery was 45–55 s for all treatments in the first trial. Trial Benzoak Aqui-SE MS-222 Rainbow Trout in trial 2 (Benzoak and Aqui-SE at 60 mg/L Concentration (mg/L) and MS-222 at 80 mg/L) were significantly shorter (P < 15050550.001) with a total length of 205 ± 1.97 mm (mean ± SE) 2606080compared with fish in trial 1 (281 ± 2.73 mm) and trial 3 3 75 75 100 (285 ± 2.40 mm). In addition, Rainbow Trout exposed to Time to stage 4 anesthesia (s) 50 mg/L Benzoak were of intermediate size compared with 1 104 (3.8) y 119 (5.2) y 204 (7.7) z the smaller Rainbow Trout in trial 2 and larger Rainbow Trout 2 74 (2.7) x 76 (3.4) x 78 (14.6) x in the other treatments. 3 79 (2.8) x 62 (2.6) xw 57 (1.6) w No mortalities were observed during the 21-d observation period after exposure to any concentrations of the three anesthetics Time to recovery (s) evaluated. 1 141 (7.3) y 220 (8.9) z 74 (4.6) w 2 91 (4.0) xw 189 (11.3) z 78 (4.0) w 3 114 (8.2) yx 202 (7.8) z 110 (2.9) x DISCUSSION Number moved Marking and Meyer (1985) suggested that the ideal anesthetic 1NANANAwould have an induction time of 3 min or less and a recovery 2221216time of less than 5 min. All three anesthetics tested in this 3872study would, by that definition, qualify as “ideal anesthetics” Time of surgery (s) despite statistical differences between trials and concentrations. 1 55 (2.7) z 45 (2.2) yx 51 (2.8) zy Although significant differences existed among the drugs and 2 38 (2.1) xw 38 (1.5) xw 37 (1.8) xw concentrations tested, these differences would have little impact 3 33 (2.6) w 37 (2.2) xw 31(1.5) w on surgical or other management activities. Except for the low- Length (mm) est concentration of MS-222, time to anesthesia for all of the 1 269 (4.6) y 285 (5.4) zy 290 (3.6) z other treatments was less than 2 min and exhibited very little 2 204 (3.2) x 206 (3.0) x 206 (3.1) x variation at treatment concentrations of 60 mg/L or more. Time 3 279 (4.4) zy 287 (3.8) z 291 (4.2) z to recovery showed more variation; fish exposed to Aqui-SE typically recovered shortly after 3 min, and less time was re- Weight (g) quired for fish anesthetized with Benzoak or MS-222. Although 1 210 (8.9) y 257 (14.3) z 253 (10.6) z significant differences were detected, they would probably be 2 96 (3.7) x 97 (4.1) x 98 (3.8) x of low biological importance as no mortalities were observed in 3 227 (11.2) zy 235 (10.3) zy 246 (11.8) zy the 21-d observational period. However, increased differences

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 (e.g., several minutes) in time to anesthesia may be important, as extended exposure of salmonids to anesthetics probably results 100 mg/L MS-222 experienced the shortest induction times at in increased mortality (Sanderson and Hubert 2007). approximately 60 s. All of the concentrations of Benzoak used in this study were Time to recovery was also significantly influenced by the greater than the 30 mg/L used by Iversen et al. (2003) to induce anesthetic and dose used (Table 1). The longest recovery times, stage 4 anesthesia in Atlantic Salmon Salmo salar smolts. They ranging from 189 to 220 s, were observed in the Rainbow Trout were also greater than the 25–30 mg/L used by Gilderhus (1990) exposed to Aqui-SE. These times were significantly longer than to reach a level of effective handling with Chinook Salmon O. that observed in Rainbow Trout exposed to 50 mg/L Benzoak, tshawytscha or the 40 mg/L needed to caudal-fin-clip Rain- which was significantly longer than for either 60 or 75 mg/L bow Trout (Gilderhus and Marking 1987). Other authors have Benzoak and 100 mg/L MS-222. Recovery from MS-222 con- noted that concentrations greater than 50 mg/L, similar to those centrations of 55 or 80 mg/L was significantly shorter than from used in this study, were required for induction of anesthesia in any of the other treatments except Benzoak at 60 mg/L. Striped Bass Morone saxatilis, Common Carp Cyprinus carpio, The number of Rainbow Trout that moved during surgery and Mozambique Tilapia Oreochromis mossambicus (Ferreira decreased with increasing anesthetic concentrations. From 12 et al. 1979; Gilderhus et al. 1991). In addition, Ferreira et al. to 22 Rainbow Trout moved during surgery when exposed to (1979) observed benzocaine to be more effective than MS-222 MANAGEMENT BRIEF 485

for induction of anesthesia. In contrast, McErlean and Kennedy An evaluation of temperature effects was not completed dur- (1968) reported that benzocaine possessed the same anesthetic ing this study as water temperature was constant (11◦C). Water qualities as MS-222, but required longer induction times when temperature can affect anesthesia induction time for clove oil- used on White Perch M. americana. based compounds (Park et al. 2009) and MS-222 (Sylvester The induction times to reach stage 4 anesthesia using Aqui- and Holland 1982) and recovery time from benzocaine-induced SE observed in this study were considerably shorter than that anesthesia (McErlean and Kennedy 1968). reported for anesthesia using eugenol-based Aqui-S by Iversen Although surgical times were different among the treatments, et al. (2003), Woods et al. (2008), and Young (2009). Over they were less than 60 s and probably had little or no effect on the range of concentrations tested for Aqui-SE, induction times our results. Much longer times for tag-related surgeries have decreased as concentrations increased, similar to what Bowker been reported in the literature, including surgeries exceeding (2006) observed in Aqui-S-exposed Rainbow Trout. Keene et al. 2–3 min while implanting radio transmitters (Sanderson and (1998) noted longer induction and recovery times for juve- Hubert 2007). Additionally, although total lengths and weights nile Rainbow Trout exposed to eugenol concentrations of 40– of fish exposed to all three of the anesthetics on the second 60 mg/L. Exposure to concentrations of Aqui-SE resulted in date were significantly smaller than those of fish used on the the longest recovery times observed in this study. These results other two trial dates, fish size probably had little to no effect are consistent with previous research documenting that recov- on anesthetic efficacy. The Rainbow Trout used in this study ery times from anesthesia involving eugenol-based compounds were shorter than the 330–600-mm (FL) adult Rainbow Trout are generally much longer than those from other fish anesthetics anesthetized by Prince and Powell (2000) with clove oil-based (Munday and Wilson 1997). anesthetics; Prince and Powell (2000) also did not observe any All concentrations of Aqui-SE evaluated in this study were relationship between fish size and induction or recovery times in greater than the 30 mg/L of clove oil used by Prince and Powell adult Rainbow Trout. Time to anesthesia and recovery appeared (2000) to anesthetize adult Rainbow Trout to stage 4 anesthesia. more dependent on concentration of anesthetics rather than fish Additionally, Prince and Powell (2000) observed considerably size. These results have implications for field work where a longer induction and recovery times (up to 18 min) than what variety of size-classes are likely to be sampled and subsequently was observed in this study. Similarly, Iversen et al. (2003) found anesthetized. concentrations of 30 mg/L eugenol to be efficient to induce stage Sanderson and Hubert (2007) defined five criteria for a suit- 4 anesthesia in Atlantic Salmon smolts and noted stress-reducing able anesthetic when surgically implanting transmitters into fish: effects associated with exposure at lower concentrations. (1) fast induction, (2) a deep anesthesia level, (3) fast surgical Gilderhus and Marking (1987) determined 60 mg/L of MS- recovery time, (4) high postsurgery survival, and (5) low postsur- 222 to be effective at inducing adequate anesthesia in Rainbow gical delayed mortality. Based only on these criteria, the use of Trout, with an induction time of 4 min, which was similar to the Benzoak, Aqui-SE, and MS-222 were viable anesthetic options 3.5 min required for 55 mg/L of MS-222 in this study. The most when performing surgical procedures on Rainbow Trout. How- rapid induction time observed with Rainbow Trout exposed to ever, other aspects of anesthetic use must be considered. There 100 mg/L of MS-222 in this study was similar to that reported can be a narrow margin between effective and toxic doses using by Flostrand and Schweigert (2005) who used the same MS-222 MS-222 during anesthesia (Gilderhus and Marking 1987), and concentration in slightly colder water temperatures with Pacific it is likely to never be approved as a zero-withdrawal anesthetic. Herring Clupea pallasii. Also similar to the results of this study, While benzocaine is eliminated from Rainbow Trout relatively Flostrand and Schweigert (2005) noted recovery times for MS- rapidly, with a mean half-life of 46.4 min (Allen 1988), there

Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 222 were shorter and less variable than those for eugenol and may be human health issues with the use and handling of this isoeugenol at 150 mg/L. Previous research has also suggested compound (Ferraro-Borgida et al. 1996; Walsh and Pease 2002; that specific water chemistry conditions, such as pH, can affect Moore et al. 2004; Basketter 2008). In addition, benzocaine MS-222 sedation (Iwama and Ackerman 1994). Relatively low may depress fish immune responses (Ortuno˜ et al. 2002). The or high pH levels can reduce induction times (Black and Connor results from this study indicate that Aqui-SE, at concentrations 1964; Gilderhus et al. 1973). Often a buffering agent, such as as low as 60 mg/L, was an effective anesthetic for use during tag- sodium bicarbonate, is used to raise pH levels and potentially implantation surgeries. It also probably poses little risk to human reduce induction times (Smit and Hattingh 1979; Smit et al. health (Moore et al. 2004), probably has minimal environmental 1978). However, other research has suggested that pH is not a concerns (Cho and Heath 2000), and has the potential for pos- factor for moderate to rapid anesthesia (Schoettger et al. 1967) sible future approval as a zero-withdrawal anesthetic for fish. and that a buffering does not affect MS-222 efficacy (Welker et al. 2007). Other influences, such as increasing the number of fish per application, may increase resistance to MS-222, but ACKNOWLEDGMENTS the mechanisms driving this response are unclear (Sylvester and We thank Dave Erdahl, Jim Bowker, and the staff of the Holland 1982). Aquatic Animal Drug Approval Partnership Program for their 486 DAVIS ET AL.

assistance with this study. In addition, we thank Michelle Gilderhus, P. A., B. L. Berger, J. B. Sills, and P. D. Harman. 1973. The efficacy Bucholz, Dylan Jones, Eric Krebs, Patrick Nero, Greg Simpson, of quinaldine sulfate: MS-222 mixtures for the anesthetization of freshwater Luke Schultz, and Keith Wintersteen. This study was conducted fish. U.S. Fish and Wildlife Service Investigations in Fish Control 54. Gilderhus, P. A., C. A. Lemm, and L. C. Woods III. 1991. Benzocaine as an under INAD numbers 11–740 and 11–741. Any views expressed anesthetic for Striped Bass. Progressive Fish-Culturist 53:105–107. in this article do not necessarily represent the views of the West- Gilderhus, P. A., and L. L. Marking. 1987. Comparative efficacy of 16 anes- ern Area Power Administration or the U.S. Government. thetic chemicals on Rainbow Trout. North American Journal of Fisheries Management 7:288–292. Hikasa, Y., K. Takase, T. Ogasawara, and S. Ogasawara. 1986. Anesthesia REFERENCES and recovery with tricaine methanesulfonate, eugenol and thiopental sodium in the carp, Cyprinus carpio. Japanese Journal of Veterinary Science 48: Allen, J. L. 1988. Residues of benzocaine in Rainbow Trout, Largemouth Bass, 341–351. and fish meal. Progressive Fish-Culturist 50:59–60. Hseu, J. R., S. L. Yeh, Y. T. Chu, and Y. Y. Ting. 1998. Comparison of efficacy Allen, J. L., G. Vang, S. Steege, and S. Xiong. 1994. Solubility of benzocaine of five anesthetics in Goldlined Sea Bream, Sparus sarba. Acta Zoologica in freshwater. Progressive Fish-Culturist 56:145–146. Taiwanica 9:11–18. Anderson, W.G., R. S. McKinley, and M. Colavecchia. 1997. The use of clove oil Iversen, M., B. Finstad, R. S. McKinley, and R. A. Eliassen. 2003. The efficacy as an anesthetic for Rainbow Trout and its effects on swimming performance. TM R North American Journal of Fisheries Management 17:301–307. of metomidate, clove oil, Aqui-S and Benzoak as anaesthetics in Atlantic Basketter, D. A. 2008. Skin sensitization: strategies for the assessment and Salmon (Salmo salar L.) smolts, and their potential stress-reducing capacity. management of risk. British Journal of Dermatology 159:267–273. Aquaculture 221:549–566. Bernier, N. J., and D. J. Randall. 1998. Carbon dioxide anaesthesia in Rainbow Iwama, G. K., and P. A. Ackerman. 1994. Anesthetics. Pages 1–15 in Trout: effects of hypercapnic level and stress on induction and recovery from P. W. Hochachka and T. P. Mommsen, editors. Biochemistry and molecu- anaesthetic treatment. Journal of Fish Biology 52:621–637. lar biology of fishes. Volume 3: analytical techniques. Elsevier, Amsterdam. Black, E. C., and A. R. Connor. 1964. Effects of MS 222 on glycogen and lactate Iwama, G. K., J. C. McGeer, and M. P. Pawluk. 1989. The effects of five levels in Rainbow Trout (Salmo gairdneri). Journal of the Fisheries Research fish anaesthetics on acid–base balance, hematocrit, blood gases, cortisol, Board of Canada 21:1539–1542. and adrenaline in Rainbow Trout. Canadian Journal of Zoology 67:2065– Bowker, J. D. 2006. The safety of AQUI-SR as an anesthetic on Rainbow 2073. Trout Oncorhynchus mykiss. U.S. Fish and Wildlife Service, Aquatic An- Keene, J. L., D. L. G. Noakes, R. D. Moccia, and C. G. Soto. 1998. The imal Drug Approval Partnership Program, Study AQUIS-06-TAS-RBT.1, efficacy of clove oil as an anaesthetic for Rainbow Trout, Oncorhynchus Bozeman, Montana. mykiss (Walbaum). Aquaculture Research 29:89–101. Briozzo, J., L. Nuncez,´ J. Chirife, L. Herszage, and M. D’Aquino. 1989. An- Kelsch, S. W., and B. Shields. 1996. Care and handling of sampled organ- timicrobial activity of clove oil dispersed in a concentrated sugar solution. isms. Pages 121–155 in B. R. Murphy and D. W. Willis, editors. Fisheries Journal of Applied Microbiology 66:69–75. techniques, 2nd edition. American Fisheries Society, Bethesda, Maryland. Burka, J. F., K. L. Hammell, T. E. Horsberg, G. R. Johnson, D. J. Rainnie, Kiessling, A., D. Johansson, I. H. Zahl, and O. B. Samuelsen. 2009. Phar- and D. J. Speare. 1997. Drugs in salmonid aquaculture: a review. Journal of macokinetics, plasma cortisol and effectiveness of benzocaine, MS-222 Veterinary Pharmacology and Therapeutics 20:333–349. and isoeugenol measured in individual dorsal aorta-cannulated Atlantic Cho, G. K., and D. D. Heath. 2000. Comparison of tricaine methanesulphonate Salmon (Salmo salar) following bath administration. Aquaculture 286:301– (MS222) and clove oil anaesthesia effects on the physiology of juvenile Chi- 308. Kildea, M. A., G. L. Allan, and R. E. Kearney. 2004. Accumulation and clearance nook Salmon Oncorhynchus tshawytscha (Walbaum). Aquaculture Research TM 31:537–546. of the anaesthetics clove oil and AQUI-S from the edible tissue of Silver Coyle, S. D., R. M. Durborow, and J. H. Tidwell. 2004. Anesthetics in aqua- Perch (Bidyanus bidyanus). Aquaculture 232:265–277. culture. Southern Regional Aquaculture Center, Publication 3900, Stoneville, Kuehl, R. O. 2000. Design of experiments: statistical principles of research Mississippi. design and analysis, 2nd edition. Duxbury Press, Pacific Grove, California. Davis, K. B., and B. R. Griffin. 2004. Physiological responses of hybrid Marking, L. L., and F. P. Meyer. 1985. Are better anesthetics needed in fisheries? Striped Bass under sedation by several anesthetics. Aquaculture 233:531– Fisheries 10(6):2–5. 548. McErlean, A. J., and V. S. Kennedy. 1968. Comparison of some anesthetic properties of benzocaine and MS-222. Transactions of the American Fisheries Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 DeTolla, L. J., S. Srinivas, B. R. Whitaker, C. Andrews, B. Hecker, A. S. Kane, and R. Reimschuessel. 1995. Guidelines for the care and use of fish in Society 97:496–498. research. ILAR (Institute for Laboratory Animal Research) Journal 37:159– McFarland, W. N., and G. W. Klontz. 1969. Anesthesia in fishes. Federation 173. Proceedings 28:1535–1540. Ferraro-Borgida, M. J., S. A. Mulhern, M. O. DeMeo, and M. J. Bayer. 1996. Moore, T. J., C. S. Walsh, and M. R. Cohen. 2004. Reported adverse event Methemoglobinemia from perineal application of an anesthetic cream. Annals cases of methemoglobinemia associated with benzocaine products. Archives of Emergency Medicine 27:785–788. of Internal Medicine 164:1192–1196. Ferreira, J. T., G. L. Smit, H. J. Schoonbee, and C. W. Holzapfel. 1979. Com- Mulcahy, D. M. 2003. Surgical implantation of transmitters into fish. ILAR parison of anesthetic potency of benzocaine hydrochloride and MS-222 in (Institute for Laboratory Animal Research) Journal 44:295–306. two freshwater fish species. Progressive Fish-Culturist 41:161–163. Munday, P. L., and S. K. Wilson. 1997. Comparative efficacy of clove oil and Flostrand, L., and J. F. Schweigert. 2005. Pacific Herring anaesthetic trials other chemicals in anaesthetization of Pomacentrus amboinensis, a coral reef with eugenol, isoeugenol, and MS-222 in association with a coded wire fish. Journal of Fish Biology 51:931–938. tagging study. Canadian Technical Report of Fisheries and Aquatic Sciences Nickum, J. G., H. L. Bart Jr., P. R. Bowser, I. E. Greer, C. Hubbs, J. A. Jenkins, 2578. J. R. MacMillan, J. W. Rachlin, J. D. Rose, P. W. Sorensen, and J. R. Tomasso. Gelwicks, K. R., D. J. Zafft, and J. P. Bobbitt. 1998. Efficacy of carbonic acid 2004. Guidelines for the use of fishes in research. American Fisheries Society, as an anesthetic for Rainbow Trout. North American Journal of Fisheries Bethesda, Maryland. Management 18:432–438. Ortuno,˜ J., M. A. Esteban, and J. Meseguer. 2002. Effects of four anaesthetics Gilderhus, P. A. 1990. Benzocaine as a fish anesthetic: efficacy and safety for on the innate immune response of Gilthead Seabream (Sparus aurata L.). spawning-phase salmon. Progressive Fish-Culturist 52:189–191. Fish and Shellfish Immunology 12:49–59. MANAGEMENT BRIEF 487

Park, M. O., S. Y.Im, D. W. Seol, and I. S. Park. 2009. Efficacy and physiological SPSS (Statistical Product and Service Solutions). 1998. SPSS 9.0 for Windows. responses of Rock Bream, Oplegnathus fasciatus, to anesthetization with SPSS, Chicago. clove oil. Aquaculture 287:427–430. Summerfelt, R. C., and L. S. Smith. 1990. Anesthesia, surgery, and re- Pirhonen, J., and C. B. Schreck. 2003. Effects of anaesthesia with MS-222, lated techniques. Pages 213–272 in C. B. Schreck and P. B. Moyle, ed- clove oil and CO2 on feed intake and plasma cortisol in steelhead trout itors. Methods for fish biology. American Fisheries Society, Bethesda, (Oncorhynchus mykiss). Aquaculture 220:507–514. Maryland. Prince, A., S. E. Low, T. J. Lissimore, R. E. Diewert, and S. G. Hinch. 1995. Sylvester, J. R., and L. E. Holland. 1982. Influence of temperature, water hard- Sodium bicarbonate and acetic acid: an effective anesthetic for field use. ness, and stocking density on MS-222 response in three species of fish. North American Journal of Fisheries Management 15:170–172. Progressive Fish-Culturist 44:138–141. Prince, A., and C. Powell. 2000. Clove oil as an anesthetic for invasive field Taylor, P. W., and S. D. Roberts. 1999. Clove oil: an alternative anaes- procedures on adult Rainbow Trout. North American Journal of Fisheries thetic for aquaculture. North American Journal of Aquaculture 61:150– Management 20:1029–1032. 155. Ross, L. G., and B. Ross. 1999. Anaesthetic and sedative techniques for aquatic Trushenski, J. T., J. D. Bowker, B. R. Gause, and B. L. Mulligan. 2012. Chemical animals, 2nd edition. Wiley, New York. and electrical approaches to sedation of hybrid Striped Bass: induction, recov- Ross, L. G., and B. Ross. 2008. Anaesthetic and sedative techniques for aquatic ery, and physiological responses to sedation. Transactions of the American animals, 3rd edition. Blackwell Scientific Publications, Oxford, UK. Fisheries Society 141:455–467. Roussel, J. M., A. Haro, and R. A. Canjuk. 2000. Field-test of a new method USFDA (U.S. Food and Drug Administration). 2002. CVM issues guidance on for tracking small fishes in shallow rivers using passive integrated transpon- the use of clove oil and eugenol for fish. USFDA, Veterinarian Newsletter 16, der (PIT) technology. Canadian Journal of Fisheries and Aquatic Sciences Washington, D.C. 57:1326–1329. Wagner, E., R. Arndt, and B. Hilton. 2002. Physiological stress responses, egg Sanderson, T. B., and W. A. Hubert. 2007. Assessment of gaseous CO2 and survival and sperm mobility for Rainbow Trout broodstock anesthetized with AQUI-S as anesthetics when surgically implanting radio transmitters into clove oil, tricaine methanesulfonate or carbon dioxide. Aquaculture 211:353– Cutthroat Trout. North American Journal of Fisheries Management 27:1053– 366. 1057. Walsh, C. T., and B. C. Pease. 2002. The use of clove oil as an anaesthetic for the Schnick, R. A., F. P. Meyer, and D. F. Walsh. 1986. Status of fishery chemicals Longfinned Eel, Anguilla reinhardtii (Steindachner). Aquaculture Research in 1985. Progressive Fish-Culturist 48:1–17. 33:627–635. Schoettger, R. A., and A. M. Julin. 1967. Efficacy of MS-222 as an anesthetic Welker, T. L., C. Lim, M. Yildirim-Aksoy, and P. H. Klesius. 2007. Effect on four salmonids. Investigations in Fish Control 13. of buffered and unbuffered tricaine methanesulfonate (MS-222) at different Sladky, K. K., C. R. Swanson, M. K. Stoskopf, M. R. Loomis, and G. A. concentrations on the stress responses of Channel Catfish, Ictalurus punctatus Lewbart. 2001. Comparative efficacy of tricaine methanesulfonate and clove Rafinesque. Journal of Applied Aquaculture 19(3):1–18. oil for use as anesthetics in Red Pacu (Piaractus brachypomus). American WHO (World Health Organization). 1982. Evaluation of certain food additives Journal of Veterinary Research 62:337–342. and contaminants. WHO Technical Report Series 20. Smit, G. L., P. S. J. Halliday, and J. Hattingh. 1978. The effect of neutralization Woods, L. C., III, D. D. Theisen, and S. Y. He. 2008. Efficacy of Aqui-S of the anaesthetic MS 222 on freshwater quality. Journal of the Limnological as an anesthetic for market-sized Striped Bass. North American Journal of Society of Southern Africa 4:31–34. Aquaculture 70:219–222. Smit, G. L., and J. Hattingh. 1979. Anaesthetic potency of MS 222 and neu- Young, M. J. 2009. The efficacy of the aquatic anaesthetic AQUI-SR for anaes- tralized MS 222 as studied in three freshwater fish species. Comparative thesia of a small freshwater fish, Melanotaenia australis. Journal of Fish Biochemistry and Physiology 62C:237–241. Biology 75:1888–1894. Downloaded by [Department Of Fisheries] at 19:58 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 19:59 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Hydrilla Management in Piedmont Reservoirs Using Herbicides and Triploid Grass Carp: A Case Study Kenneth L. Manuel a , James P. Kirk b , D. Hugh Barwick a & Tommy W. Bowen a a Duke Energy Corporation, Water Strategy, Hydro Licensing and Lake Services , Environmental Center , MG03A3, 13339 Hagers Ferry Road, Huntersville , North Carolina , 28078 , USA b Environmental Laboratory , Engineer Research and Development Center , 3909 Halls Ferry Road, Vicksburg , Mississippi , 39180 , USA Published online: 28 Apr 2013.

To cite this article: Kenneth L. Manuel , James P. Kirk , D. Hugh Barwick & Tommy W. Bowen (2013): Hydrilla Management in Piedmont Reservoirs Using Herbicides and Triploid Grass Carp: A Case Study, North American Journal of Fisheries Management, 33:3, 488-492 To link to this article: http://dx.doi.org/10.1080/02755947.2013.768570

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:488–492, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.768570

MANAGEMENT BRIEF

Hydrilla Management in Piedmont Reservoirs Using Herbicides and Triploid Grass Carp: A Case Study

Kenneth L. Manuel Duke Energy Corporation, Water Strategy, Hydro Licensing and Lake Services, Environmental Center MG03A3, 13339 Hagers Ferry Road, Huntersville, North Carolina 28078, USA James P. Kirk* Environmental Laboratory, Engineer Research and Development Center, 3909 Halls Ferry Road, Vicksburg, Mississippi 39180, USA D. Hugh Barwick and Tommy W. Bowen Duke Energy Corporation, Water Strategy, Hydro Licensing and Lake Services, Environmental Center MG03A3, 13339 Hagers Ferry Road, Huntersville, North Carolina 28078, USA

native aquatic vegetation, use of diploid and triploid Grass Carp Abstract Ctenopharyngodon idella, or some combination of these ap- We developed a three-step management strategy for hydrilla proaches. A minimalist approach of doing nothing, except for Hydrilla verticillata in ﬁve Piedmont reservoirs operated by Duke herbicide spot treatments for lake access and navigation, is an- Energy Corporation. This strategy involves (1) early detection of hydrilla, (2) use of registered herbicides for plant suppression along other possible management alternative. with stocking 20 triploid Grass Carp Ctenopharyngodon idella per Symposia held during 1994 (Grass Carp Symposium) and surface acre of infestation, and (3) maintenance stocking of triploid 2004 (Hydrilla Management in Florida) further summarized Grass Carp to prevent hydrilla regrowth from tubers. Following hydrilla management strategies as well as the potential of us- this strategy, hydrilla in the water column was eliminated within ing Grass Carp as a management tool (Haller 1994a; Hoyer one calendar year after Grass Carp introduction in four out of ﬁve reservoirs. This suggests that integrating herbicide applica- et al. 2005). Additional studies that evaluated methodologies, tions with stocking Grass Carp largely eliminates the multiyear lag impacts, or controversies associated with Grass Carp were con- effect normally associated with using Grass Carp alone. A mainte- ducted in Lake Conroe, Texas (Klussmann et al. 1988; M. A. nance density of at least one triploid Grass Carp per eight surface Webb et al. 1994; Elder and Murphy 1997), Lake Guntersville, acres of the reservoir prevented hydrilla regrowth except for a Alabama (Bain et al. 1990; D. H. Webb et al. 1994; Morrow and

Downloaded by [Department Of Fisheries] at 19:59 28 May 2013 brief and minor reinfestation in one of ﬁve study reservoirs. This management approach proved successful when hydrilla coverage Kirk 1997), and the Santee Cooper reservoirs, South Carolina was as little as 1–3% of the reservoir’s surface area. Detecting and (Morrow et al. 1997; Kirk et al. 2000, 2001; Henderson et al. controlling hydrilla early during the infestation should reduce the 2003; Kirk and Socha 2003; Kirk and Henderson 2006). Most of cost of management and perhaps minimize some adverse effects these studies determined that introducing triploid Grass Carp is associated with the introduction and use of triploid Grass Carp. the most cost-effective method for long-term control of hydrilla but with a major limitation: triploid Grass Carp are best used Hydrilla Hydrilla verticillata appeared in Florida during the where loss of all palatable submersed vegetation is acceptable early 1950s (Blackburn et al. 1969; Schmitz et al. 1991; Hoyer for an extended period of time (Allen and Wattendorf 1987; et al. 2005) and has continually spread to areas such as Washing- Wattendorf and Anderson 1987; Hoyer et al. 2005). ton and Maine where the infestation was not expected (USACE An effective hydrilla management strategy that attempts to 2005). Management tools to control hydrilla include the ap- minimize herbicide and triploid Grass Carp use by interven- plication of registered herbicides, mechanical harvesting, win- ing before hydrilla becomes widespread has evolved. This ap- ter drawdowns, hydrilla-speciﬁc insect pests, establishment of proach relies on early detection of hydrilla infestations, prompt

*Corresponding author: [email protected] Received November 14, 2012; accepted January 15, 2013 488 MANAGEMENT BRIEF 489

suppression with registered herbicides, triploid Grass Carp reservoirs to locate and control mosquito breeding areas. These stocked at normal rates (20 fish per vegetated acre), and low- crews, who were also trained to identify aquatic vegetation, level maintenance stockings of Grass Carp (for at least a decade) sought invasive aquatic vegetation during their surveys. Once to prevent hydrilla regrowth from tubers in the hydrosoil. nuisance vegetation, especially hydrilla, was located, control activities were started. The extent of hydrilla infestation was METHODS determined by pulling a rake, which was attached to a rope, Study area.—Five Piedmont reservoirs (in North Carolina through the vegetation to detect hydrilla. Corresponding coordi- and South Carolina) operated by Duke Energy Corporation and nates determined from a GPS were plotted to determine the area managed in cooperation with local and state agencies were used of infestation. Then the infestation was treated with registered to develop the management strategy examined in this study aquatic herbicides (usually Komeen) according to label instruc- (Figure 1). Four of the reservoirs impound the Catawba River tions. In most cases the infestation was treated and suppressed and (from north to south) are Lake James, Lake Norman, Moun- for several years using herbicides (Table 1) until stakeholder tain Island Lake, and Lake Wylie. Belews Lake is an impound- groups (i.e., state agencies, Duke Energy Corporation, marine ment of Belews Creek in the Roanoke River drainage. commissions, and interested citizens) could be assembled to de- Each reservoir and the area that could potentially be infested velop, approve, and fund an integrated approach. This approach with hydrilla based upon the reservoir’s 20-ft depth contour combined herbicide applications, stocking Grass Carp at a rate is described in Table 1. The 20-ft depth contour was chosen of 20 triploid fish per surface acre of hydrilla, and maintaining because, in our experience, it represented the maximum depth a density of one fish per every eight surface acres once hydrilla subject to hydrilla infestation. The five reservoirs total 59,674 had been eliminated in the water column. surface acres of which 14,920 acres are vulnerable to infestation. Triploid Grass Carp were usually stocked during the spring Reservoir size ranged from 3,281 to 32,475 acres (Table 1). The or early summer. Fish were at least 10 in TL, carefully tem- first infestations of hydrilla were documented to have occurred pered to reduce temperature differences between the fish hauler in each of the study reservoirs between 1999 and 2006 (Table 1). and the reservoir, and inspected by a biologist for possible in- Management approach.—During the growing season, Duke juries incurred during transport. After hydrilla was controlled, Energy mosquito control crews routinely surveyed the study (i.e., hydrilla could not be located by mosquito control crews in Downloaded by [Department Of Fisheries] at 19:59 28 May 2013

FIGURE 1. The study sites including Belews Lake, Lake James, Mountain Island Lake, Lake Norman, and Lake Wylie. [Figure available in color online.] 490 MANUEL ET AL.

TABLE 1. Characteristics and hydrilla management chronology of ﬁve Piedmont reservoirs in North Carolina and South Carolina operated by Duke Energy Corporation. Reservoir characteristics include total surface area, surface area potentially infested by hydrilla (the 20-ft contour), the year hydrilla was ﬁrst detected, the year triploid Grass Carp were stocked, and the year that hydrilla was eliminated in the water column.

Surface area Area infested Potential Year Grass Carp Control Reservoir (acres) (acres) infestation (acres) infested stocked achieved Lake Norman 32,475 400 8,000 2000 2004 2005 Mountain Island Lake 3,281 1,000 1,200 2000 2000 2003 Lake James 6,812 1,050 1,400 1999 2002 2003 Belews Lake 3,663 106 920 1999 2005 2006 Lake Wylie 13,443 90 3,400 2006 2009 2009

systematic surveys using rakes attached to a rope) triploid Grass a half years (C. Page, South Carolina Department of Natural Carp were stocked as needed to maintain a minimum density Resources, personal communication). of at least one fish per every eight surface acres (of the entire We speculate that rapid control, except in Mountain Island reservoir) in order to prevent regrowth of hydrilla from the tu- Lake, probably resulted from Grass Carp consuming mostly hy- ber banks. Maintenance stockings were based upon replacing an drilla that had regrown from tubers rather than both hydrilla in annual loss of 32% (Kirk et al. 2000; Kirk and Henderson 2006). the water column and regrowth. While integrated use of Grass Carp and herbicides has been proposed for years (Sutton et al. 1986), few published studies detail this management approach RESULTS in reservoirs (Jaggers 1994). Hydrilla can also be rapidly con- With the exception of Mountain Island Lake, mosquito controlled by using high stocking rates of Grass Carp. For example, trol crews confirmed that hydrilla in the water column was elim- high stocking rates (50 fish per vegetated acre) rapidly elimi- inated within one calendar year after stocking triploid Grass nated hydrilla in Lake Conroe, Texas, twice between the 1980s Carp. During 2003, a total of 1,000 and 1,050 acres of hydrilla and 2008 (Klussmann et al. 1988; Chilton et al. 2008). These were controlled in Mountain Island Lake and Lake James, re- high stocking rates, while successful, were apparently necessary spectively (Table 1). A 400-acre infestation was controlled in to control rapidly expanding hydrilla. Lake Norman during 2005, and a 106-acre infestation in Belews During this study, workers were able to detect increasingly Lake was controlled the following year. Small infestations (es- smaller infestations, which were swiftly controlled. After con- timated at 5 acres) in Lake Wylie were treated with Komeen trol was achieved in both Lake James and Mountain Island beginning in 2006. Despite herbicide treatment, hydrilla contin- Lake with hydrilla infestations of approximately 1,000 acres, ued to spread to multiple sites within the reservoir and covered smaller infestations ranging from 90 to 400 acres were at- 90 acres by November 2008. A total of 1,800 triploid Grass Carp tempted. Control in Lake Norman, Lake Wylie, and Belews (20 fish/acre) were stocked in April–May 2009 and hydrilla in- Lake was achieved when widely spread hydrilla ranged from festation was controlled by October 2009. 1% to 3% of the impoundment’s surface area. Maintenance densities of at least one fish per every eight surface acres of each reservoir controlled hydrilla regrowth except Management Implications in Lake James. During October 2009, a floating patch of hydrilla Downloaded by [Department Of Fisheries] at 19:59 28 May 2013 The use of triploid Grass Carp for hydrilla control in reser- approximately 100 ft2 in size was found (and eliminated using voirs remains contentious, and the tradeoffs in using this man- herbicides) during a routine survey 6 years after achieving ini- agement tool have long been discussed in the literature (Noble tial control. It was not known whether the infestation was from et al. 1986; Bain 1993). Usual stocking densities in the south- regrowth or introduction. eastern USA are commonly 10–20 fish per vegetated acre, but these stocking densities can be confounded by weather and DISCUSSION nutrient-related factors (Canfield et al. 1983; Maceina et al. We demonstrated the potential to rapidly eliminate hydrilla in 1992), poorly understood migration (Bain et al. 1990; Foltz and the water column using a combination of herbicides and triploid Kirk 1994; Kirk et al. 2001), or unexpected mortality (Kirk Grass Carp. Often there is a multiyear lag period before Grass 1992; Clapp et al. 1994). Despite stocking models (Miller and Carp alone control hydrilla growth (Sutton et al. 1986; Leslie Decell 1984; Swanson and Bergerson 1988) and population as- et al. 1987). In the Santee Cooper system, triploid Grass Carp sessment developed in large systems (Morrow and Kirk 1997; were stocked beginning in 1989 but did not achieve control Morrow et al. 1997; Kirk et al. 2000), regulating the degree of until 1997 (Kirk and Socha 2003; Kirk and Henderson 2006). control has been poor. Usually, the use of Grass Carp has resulted In nearby Lake Murray, South Carolina, which had an infes- in either no response or near total elimination of submersed tation covering 6,600 acres, the lag period was about two and aquatic vegetation (Sutton 1977; Leslie et al. 1987; Hanlon MANAGEMENT BRIEF 491

et al. 2000). Kirk and Socha (2003) found most triploid Grass improving this article. The support of the Aquatic Plant Control Carp in the Santee Cooper reservoirs died before age 10 and Research Program is also appreciated. overstocking may be reversible over time. The Grass Carp mortality encountered in the Santee Cooper reservoirs has resulted in a substantial rebound (approximately 10% coverage) in sub- REFERENCES mersed native vegetation in the Santee Cooper system during Allen, S. K., Jr., and R. J. Wattendorf. 1987. Triploid Grass Carp: status and the last decade (S. Lamprecht, South Carolina Department of management implications. Fisheries 12(4):20–24. Natural Resources, personal communication). As a note of cau- Bain, M. B. 1993. Assessing impacts of introduced aquatic species: Grass Carp tion, the experience in Florida has been just the opposite with in large systems. Environmental Management 17:211–224. long-term depletions of submersed native vegetation after the Bain, M. B., D. H. Webb, M. D. Tangedal, and L. N. Mangum. 1990. Movements introduction of Grass Carp (Haller 1994b; Cassani et al. 2008). and habitat use by Grass Carp in a large mainstream reservoir. Transactions of the American Fisheries Society 119:553–561. In our case study, hydrilla control became easier with ex- Blackburn, R. D., L. W. Weldon, R. R. Yeo, and T. M. Taylor. 1969. Identification perience. With practice, mosquito control crews were able to and distribution of certain similar-appearing submersed aquatic weeds in accurately locate and map small and widespread hydrilla in- Florida. Hyacinth Control Journal 8:17–23. festations. Using a combination of herbicide applications and Canfield, D. E., Jr., M. J. Maceina, and J. V. Shireman. 1983. Effects of hydrilla triploid Grass Carp before the infestation becomes widespread and Grass Carp on water quality in a Florida lake. Journal of the American Water Resources Association 19:773–778. may obviate some of the drawbacks often encountered in man- Cassani, J., S. Hardin, V.Mudrak, and P.Zajicek. 2008. A risk analysis pertaining aging hydrilla (e.g., difficulty in controlling rapidly expanding to the use of triploid Grass Carp for the biological control of aquatic plants. hydrilla, elimination of palatable native vegetation, and long- Florida Department of Environmental Protection and Florida Department of term vegetation loss). Additionally, the cost of hydrilla control Agriculture and Consumer Services, Tallahassee. would be substantially less than was experienced in the Santee Chilton, E. W., II, M. A. Webb, and R. A. Ott Jr. 2008. Hydrilla management in Lake Conroe, Texas: a case history. Pages 247–257 in M. S. Allen, S. Sam- Cooper reservoirs. In the Santee Cooper system, hydrilla was mons, and M. J. Maceina, editors. Balancing fisheries management and water not controlled early in the infestation and expanded to fill most uses for impounded river systems. American Fisheries Society, Symposium available habitat, despite triploid Grass Carp introductions and 62, Bethesda, Maryland. early use of herbicides, until about 48,000 surface acres had Clapp, D. F., R. S. Hestand III, and B. Z. Thompson. 1994. Hauling and post- been infested. A total of 768,500 triploid Grass Carp stocked stocking mortality of triploid Grass Carp. Journal of Aquatic Plant Manage- ment 32:41–43. between 1989 and 1996 were able to control hydrilla growth Elder, H. S., and B. R. Murphy. 1997. Grass Carp (Ctenopharyngodon idella) during 1996–1997 (Kirk and Henderson 2006). in the Trinity River, Texas. Journal of Freshwater Ecology 12:281–289. A number of factors may have contributed to successful hy- Foltz, J. W., and J. P. Kirk. 1994. Aquatic vegetation and water quality in Lake drilla management in the study reservoirs. Native submersed Marion, South Carolina. Pages 93–107 in W. T. Haller, editor. Proceedings aquatic vegetation was sparse in the five study reservoirs. Ad- of the Grass Carp symposium. U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Mississippi. ditionally, stakeholders were concerned about the potential ad- Haller, W. T., editor. 1994a. Proceedings of the Grass Carp symposium, March verse effects of hydrilla to water use, recreation, and property 7–9, 1994, Gainesville, Florida. U.S. Army Corps of Engineers, Waterways values. As a consequence, the use of triploid Grass Carp as Experiment Station, Vicksburg, Mississippi. a management tool was less controversial than it might have Haller, W. T. 1994b. Probable Grass Carp stocking scenarios. Pages 236–238 in been. However, because part of this management strategy in- W. T. Haller, editor. Proceedings of the Grass Carp symposium. U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Mississippi. volves triploid Grass Carp use, we suggest three areas of inves- Hanlon, S. G., M. V.Hoyer, C. E. Cichra, and D. E. Canfield Jr. 2000. Evaluation tigation to complement this study. The first is to determine the of macrophyte control in 38 Florida lakes using triploid Grass Carp. Journal

Downloaded by [Department Of Fisheries] at 19:59 28 May 2013 smallest area of coverage that is susceptible to this management of Aquatic Plant Management 38:48–54. strategy. A second is to better understand maintenance stocking Henderson, J. E., J. P. Kirk, S. D. Lamprecht, and W. E. Hayes. 2003. Economic approaches by monitoring maintenance stockings for longer pe- impacts of aquatic vegetation to angling in two South Carolina reservoirs. Journal of Aquatic Plant Management 41:53–56. riods. The third involves testing this approach in other regions of Hoyer, M. V., M. D. Netherland, M. S. Allen, and D. E. Canfield Jr. 2005. the country. Grass Carp use in Florida has been extensively tried Hydrilla management in Florida: a summary and discussion of issues identi- and generally found wanting (Cassani et al. 2008). However, fied by professionals with future management recommendations—final docu- hydrilla is continually spreading and the integrated use of herbi- ment. Florida LAKEWATCH, Department of Fisheries and Aquatic Sciences, cides and triploid Grass Carp early during an infestation may be University of Florida/Institute of Food and Agricultural Sciences, Gainesville. Jaggers, B. V. 1994. Economic considerations of integrated hydrilla manage- a feasible management approach in other regions of the country. ment: a case history of Johns Lake, Florida. Pages 151–163 in W. T. Haller, editor. Proceedings of the Grass Carp symposium. U.S. Army Corps of En- gineers, Waterways Experiment Station, Vicksburg, Mississippi. ACKNOWLEDGMENTS Kirk, J. P. 1992. Efficacy of triploid Grass Carp in controlling nuisance aquatic We thank Duke Energy Corporation mosquito control crews vegetation in South Carolina farm ponds. North American Journal of Fisheries Management 12:581–584. for their diligence in locating hydrilla infestations. We also Kirk, J. P., and J. E. Henderson. 2006. Management of hydrilla in the Santee acknowledge the contributions of K. J. Killgore, W.T. Slack, Cooper reservoirs, South Carolina: experiences from 1982 to 2004. Journal and A. Harrison-Lewis as well as two reviewers in editing and of Aquatic Plant Management 44:98–103. 492 MANUEL ET AL.

Kirk, J. P., K. J. Killgore, J. V. Morrow Jr., S. D. Lamprecht, and D. W. Cooke. Noble, R. L., P. W. Bettoli, and R. K. Betsill. 1986. Considerations for the use of 2001. Movements of triploid Grass Carp in the Cooper River, South Carolina. Grass Carp in large, open systems. Lake and Reservoir Management 2:46–48. Journal of Aquatic Plant Management 39:59–62. Schmitz, D. C., B. V. Nelson, L. E. Nall, and J. D. Schardt. 1991. Exotic Kirk, J. P., J. V. Morrow Jr., K. J. Killgore, S. J. de Kozlowski, and J. W. aquatic plants in Florida: a historical perspective and review of the present Preacher. 2000. Population response of triploid Grass Carp to declining levels aquatic plant regulation program. Pages 303–326 in T. D. Center, R. F. Doren, of hydrilla in the Santee Cooper reservoirs, South Carolina. Journal of Aquatic R. L. Hofstetter, R. L. Myers, and L. D. Whiteaker, editors. Proceedings of the Plant Management 38:14–17. symposium on exotic pest plants, Miami, November 1988. U.S. Department of Kirk, J. P., and R. C. Socha. 2003. Longevity and persistence of triploid Grass the Interior, National Parks Service Technical Report NPS/NREVER/NRTR- Carp stocked into the Santee Cooper reservoirs of South Carolina. Journal of 91/06, Washington, D.C. Aquatic Plant Management 41:90–92. Sutton, D. L. 1977. Grass Carp (Ctenopharyngodon idella Val.) in North Amer- Klussmann, W. G., R. L. Noble, R. D. Martyn, W. J. Clark, R. K. Betsill, ica. Aquatic Botany 3:157–164. P. W. Bettoli, M. F. Cichra, and J. M. Campbell. 1988. Control of aquatic Sutton, D. L., V. V. Vandiver Jr., and J. E. Hill. 1986. Grass Carp: a ﬁsh macrophytes by Grass Carp in Lake Conroe, Texas, and the effects on the for biological management of hydrilla and other aquatic weeds in Florida. reservoir ecosystem. Texas Agricultural Experiment Station Miscellaneous University of Florida, Institute of Food and Agricultural Sciences Extension, Publication 1664. Bulletin 867, Gainesville. Leslie, A. J., Jr., J. M. Van Dyke, R. S. Hestand III, and B. Z. Thompson. Swanson, E. D., and E. P. Bergersen. 1988. Grass Carp stocking model for 1987. Management of aquatic plants in multi-use lakes with Grass Carp coldwater lakes. North American Journal of Fisheries Management 8:284– (Ctenopharyngodon idella). Lake and Reservoir Management 3:266–276. 291. Maceina, M. J., M. F. Cichra, R. K. Betsill, and P.W. Bettoli. 1992. Limnological USACE (U.S. Army Corps of Engineers). 2005. APIS: aquatic plant manage- changes in a large reservoir following vegetation removal by Grass Carp. ment information system. USACE, Washington, D.C. Journal of Freshwater Ecology 7:81–95. Wattendorf, R. J., and R. S. Anderson. 1987. Hydrilla consumption by triploid Miller, A. C., and J. L. Decell. 1984. Use of the White Amur for aquatic Grass Carp. Proceedings of the Annual Conference Southeastern Association plant management. U.S. Army Corps of Engineers, Waterways Experiment of Fish and Wildlife Agencies 38(1984):319–326. Station, Aquatic Plant Control Research Program, Instruction Report A-84-1, Webb, D. H., L. N. Mangum, A. L. Bates, and H. D. Murphy. 1994. Aquatic Vicksburg, Mississippi. vegetation in Guntersville Reservoir following Grass Carp stocking. Pages Morrow, J. V., Jr., and J. P. Kirk. 1997. Age and growth of Grass Carp in Lake 199–209 in W. T. Haller, editor. Proceedings of the Grass Carp symposium. Guntersville, Alabama. Proceedings of the Annual Conference Southeastern U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Association of Fish and Wildlife Agencies 49(1995):187–194. Mississippi. Morrow, J. V.Jr., J. P.Kirk, and K. J. Killgore. 1997. Collection, age, growth, and Webb, M. A., J. C. Henson, and M. S. Reed. 1994. Lake Conroe ﬁsheries: population attributes of triploid Grass Carp stocked into the Santee–Cooper population trends following macrophyte removal. Pages 169–185 in W. T. reservoirs, South Carolina. North American Journal of Fisheries Management Haller, editor. Proceedings of the Grass Carp symposium. U.S. Army Corps 17:38–43. of Engineers, Waterways Experiment Station, Vicksburg, Mississippi. Downloaded by [Department Of Fisheries] at 19:59 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:00 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Capture Efficiency of Barbed versus Barbless Artificial Flies for Trout Roger K. Bloom a a California Department of Fish and Wildlife , 1701 Nimbus Road, Rancho Cordova , California , 95670 , USA Published online: 28 Apr 2013.

To cite this article: Roger K. Bloom (2013): Capture Efficiency of Barbed versus Barbless Artificial Flies for Trout, North American Journal of Fisheries Management, 33:3, 493-498 To link to this article: http://dx.doi.org/10.1080/02755947.2013.769920

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:493–498, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.769920

ARTICLE

Capture Efﬁciency of Barbed versus Barbless Artiﬁcial Flies for Trout

Roger K. Bloom* California Department of Fish and Wildlife, 1701 Nimbus Road, Rancho Cordova, California 95670, USA

Abstract I examined the capture efficiency of artificial flies fished with barbed and barbless hooks in various coldwater fisheries throughout California. Capture efficiency was defined as the proportion of trout (family Salmonidae) landed to the total number of trout hooked while angling. Waters were selected based on high catch per unit effort along with trout species present in an effort to increase the probability of encounters and the species represented. Artificial flies were standardized by J-style hooks and three artificial fly types (dry, nymph, and streamer). In an effort to reduce bias, anglers were not told what hook type (i.e., barbed or barbless) they were using and were not allowed to handle or visually inspect flies. A total of 1,617 trout were landed with a mean total length of 213 mm and a range of 64–660 mm. Mean capture efficiency (and ranges) was 76% (38–100%) for anglers using barbed flies and 63% (0–100%) for anglers using barbless flies. Results show that anglers using barbless flies landed proportionately less trout than when they used barbed flies. Fisheries managers must weigh any perceived benefits from barbless regulations with potential reductions in catch rates and associated angler satisfaction.

Fisheries managers are frequently tasked with developing location, hooking duration, trout size, and water temperature. and maintaining quality sportfishing opportunities while bal- Hook type (barbed versus barbless) was never addressed as a ancing a need to protect and monitor aquatic resources. Using contributing factor in his summary. Although the effects on fish sportfishing regulations as a management tool can have a sub- captured using barbed hooks have shown increased injury and stantial effect on fisheries and, if used appropriately, can enhance handling times (Meka 2004), this may not have an impact at angling opportunities. Currently, California has various fresh- the population level. Some studies and fisheries experts have water fishing regulations that require the use of barbless hooks. questioned the efficacy of using barbless hook regulations as a These regulations were proposed and adopted based on a per- management tool (Schaeffer and Hoffman 2002; DuBois and Downloaded by [Department Of Fisheries] at 20:00 28 May 2013 ception that barbless hooks would decrease hooking mortality Dubielzig 2004; Cooke and Schramm 2007). Regulating fish- by reducing handling time, stress, and trauma. eries through the use of fishing gear, such as barbless hooks, Although previous studies and assessments have referenced may be ineffective in reducing injury and mortality (Cooke quicker unhooking times when using barbless hooks (Barnhart and Schramm 2007). Perceptions relating to benefits and ef- 1990; Schill and Scarpella 1997; Schaeffer and Hoffman 2002; fects of barbless hooks and their use through regulations as a Meka 2004), effects from barbless hooks on postrelease sur- management tool arise from issues that are largely social rather vival have shown mixed results (Wydoski 1977; Mongillo 1984; biological (Schill and Scarpella 1997; Radomski et al. 2001). Taylor and White 1992; DuBois and Dubielzig 2004). Faragher Although the benefits to trout survival arising from the use of (2004) provided an overview of existing studies on hook- single hooks, barbless hooks, or flies may be relatively small, ing mortalities for trout (family Salmonidae). He distilled his the use of these restrictions could be very important in situations findings into a summary of selected literature that showed when fish are hooked many times (Wright 1992). Evaluating the hooking mortality of trout is variable depending on hooking effects of barbless hooks on fish, both positive and negative,

*E-mail: [email protected] Received May 24, 2012; accepted January 17, 2013 493 494 BLOOM

should be a consideration when establishing sportfishing METHODS regulations. The study was conducted on both public and private wa- Substantial interest and research have focused on hooking ters throughout California from 2005 to 2009. To increase the mortality based on gear, hook type, and fish species; however, probability that sufficient data were acquired, high catch-per- less effort has been put into evaluating the probability of capture unit-effort waters were selected a priori. These waters were associated with these variables. The proportion of fish landed also chosen to diversify locations and trout species represented. to those not landed while angling is often referred to as capture I utilized volunteers and CDFW personnel as anglers. Only efficiency (CE). Although prior studies compared CE of barbed CDFW personnel were used to observe and assist anglers. These and barbless hooks (Barnhart 1990; DuBois and Dubielzig 2004; “observers” were responsible for making the study “blind” by Meka 2004; Ostrand et al. 2006), only Barnhart (1990) and tying-on and switching flies for anglers, removing flies from Meka (2004) assessed the CE of artificial flies. Species targeted trout, assessing injury, measuring length, and releasing. This in these two studies were relatively large migratory (anadro- process eliminated the ability of anglers to see whether he mous, fluvial, adfluvial) Coastal Rainbow Trout Oncorhynchus or she was using a barbed or barbless fly. Anglers and ob- mykiss irideus and may not be directly applicable to other inland servers were trained, prior to sampling, in study protocols and coldwater fisheries managed with barbless regulations. In addi- were given necessary field gear. Flies were separated by hook tion, these studies allowed anglers to know what hook type they type and fly type into labeled tackle boxes. Observers main- were using. Anglers that are aware of a certain type of hook may tained possession of tackle boxes at all times and were the only not fish with the same level of intensity (Schaeffer and Hoffman ones to handle, switch, and tie-on flies during sampling peri- 2002). This bias could affect angler ability to fight and land fish ods. Anglers were given a choice of what fly type they wanted with equal effort while using different hook types. to begin with. Anglers were allowed to switch fly types until The rationale for choosing only artificial flies was based on an encounter occurred. The initiation of an encounter was de- information from angler surveys. California Department Fish fined as when (1) a trout had volitionally taken the fly, (2) an and Wildlife (CDFW) angler survey data collected from 1999 angler had set the hook, and (3) there was resistance on the to 2003 were analyzed to evaluate gear preference (fly, lure, bait, rod from a trout for a period no shorter than 2 s. These crite- or a combination) by trout anglers fishing both streams and lakes. ria were established to reduce false hooking and missed strike None of the waters chosen for analysis had an artificial-fly-only data. regulation. When provided the opportunity to use either bait, Once the first encounter occurred, anglers had to stay with lures, or flies, anglers strongly preferred using flies only, in both that fly type for the duration of a sampling period. To allow streams (79%) and lakes (78%) compared with anglers that only flexibility, anglers could switch fly patterns or colors within that used lures (17% for both streams and lakes). A small percentage fly type during a sampling period. The fly patterns and associ- (5%) of the remaining anglers used both lures and flies, bait, ated hook sizes selected were based on top-selling fly patterns or their choice was unknown. The vast majority of special- for 2005, both in California and nationally. Ten patterns were regulation waters within California have a barbless lure or flies selected in each fly type; however, pattern selection was limited regulation; however, given the strong data supporting angler to only straight shank hook types. Hook style was limited to preferences for using artificial flies in these waters, focus was straight shank J-style 1–3 X long, 1–2 X wide for all fly types. limited to artificial flies. The objective of the study was to test CE Hook size ranged from 6 to 8 for streamers and 12 to 16 for of three artificial fly types fished with barbed and barbless hooks. both dry and wet flies. Although hook size differed for stream- Additionally, I also evaluated CE based on angler experience. ers versus the other fly types, it was assumed that hook size

Downloaded by [Department Of Fisheries] at 20:00 28 May 2013 Associated data on injury rates and release times were compared had no effect. Since all flies were initially barbed, flies used in by hook type. the barbless hook type were debarbed prior to sampling peri- It is critical fisheries managers evaluate, adopt, and monitor ods. Debarbing was done by CDFW personnel using specially special regulations with specific strategies and objectives. This designed debarbing pliers. approach will allow for assessment of regulations and associated Each sampling period was 4-h in duration, randomly divided responses within a fishery. Adopting special regulations without into eight, 30-min sessions. During four sessions, anglers used specific justification, realistic goals, and measurable objectives a barbed fly and during the other four sessions anglers used a can lead to conflicts and poor results. Many perceive special reg- barbless fly. Prior to each sampling period, observers would ulations as a panacea for all existing fisheries problems (AFS randomly select one of 64 scenario cards, which provided the 2009). Unfortunately, improper use of an otherwise effective pattern of 30-min sessions during a sampling period. These tool can result in negative angler perceptions, continued decline scenario cards represented all different potential combinations of fishing quality, loss of agency and professional credibility, of session patterns. During each sampling period, the observer and unrealistic angler expectations (Behnke 1987). Understand- would change hook type based on preselected session rotation ing the CE of artificial flies, both barbed and barbless, will on the card. Randomization of hook type throughout a sampling assist fisheries managers in making informed decisions when period was designed to reduce the possible effects of bias in assessing or establishing special regulations. angling success. This approach also made it difficult for anglers BARBED VERSUS BARBLESS ARTIFICIAL FLIES 495

to decipher a pattern relating to the hook type they were using performed Tukey multiple comparisons. Due to the RBD, I was during each sampling period. not able to test for interaction between angler experience and Anglers were classified as either advanced (>200 d expe- hook type, nor for interaction between fly type and hook type. rience), intermediate (30–200 d experience), or novice (<30 d Since the joint distribution of the CE and the number of fish experience) based on the total number of days they had fly encounters per 4-h sampling period was not bivariate normal, I fished in their life. In addition to switching and tying-on flies calculated a nonparametric Spearman correlation coefficient. I for anglers, observers also kept track of session rotation time, also regressed the CE on the number of fish encounters. Using hook type, duration of encounter, species identification, injury, logistic regression by hook type, I modeled the probability of and handling time. The various timed events were recorded in landing a trout on duration of fish encounter. For the logistic seconds using handheld stopwatches. The timing of encounters model, the usual R2 was not used, since the magnitude of R2 is by observers began when he or she confirmed an encounter a misleading measure of explained variability when the outcome (per the criteria stated previously) and continued until anglers variable is binomial (Ryan 1997). Hence, I did not report R2 for landed the trout. Anglers landed all trout by use of a soft-mesh the logistic regression. I used a paired t-test to evaluate the dif- net. After anglers landed their trout, they passed netted fish to ference between the two hook types for CE, handling times, and the observer for hook removal, injury assessment, measurement injury rates. I performed statistical analysis with SAS version in total length (TL), and notation of hooking location. Injury 9.1.3 (SAS 2006). The level of significance for tests was set at was defined as torn tissue, bleeding, or external hooking. α = 0.10. Only injuries resulting from an encounter or hook removal process were noted. The severity of injury was not assessed or ranked. RESULTS I used a randomized block design (RBD) to evaluate the A total of 32 different anglers participated in one to seven effects on CE for hook type, fly type, injury rate, and angler ex- separate sampling periods. Although some anglers participated perience. The 4-h sampling period served as a block or sampling in multiple sampling periods, each period was treated as unique unit. Only sampling periods that had at least two encounters for in the analysis. Eighteen different CDFW personnel served as each hook type were used in the analysis. Capture efficiency observers. Seventy-eight sampling periods out of 98 total were represented the percentage of trout landed for all trout encoun- used in the analysis. Twenty sampling periods were removed tered during a sampling period. Encounters in which a trout from the analysis due to the low (less than two) number of broke the line and was not landed were excluded from anal- encounters per hook type. Within the 78 sampling periods, 12 ysis. For each sampling period, I calculated CE for the hook encounters were removed from the analysis due to line breakage types, the fly type used, and the experience level of the angler. I during the encounter. Thus, a total of 2,258 encounters qualified tested fly types and angler experience with one-factor analysis for analysis with 48% (n = 1,077) in the barbed hook type and of variance (ANOVA). If the overall F-test was significant, I 52% (n = 1,181) in the barbless hook type (Table 1).

TABLE 1. Summary of trout landed and those not landed in association with capture efﬁciency, angler experience, hook type, and ﬂy type.

Barbed flies Barbless flies Angler experience & Trout Trout not Number of Capture Trout Trout not Number of Capture Downloaded by [Department Of Fisheries] at 20:00 28 May 2013 fly type landed landed encounters efficiencya landed landed encounters efficiencya Advanced angler 79% 68% Dry fly 308 69 377 82% 297 114 411 72% Nymph 91 40 131 69% 73 38 111 66% Streamer 72 18 90 80% 72 34 106 68% Intermediate angler 75% 61% Dry fly 200 62 262 76% 219 116 335 65% Nymph 52 28 80 65% 42 31 73 58% Streamer 21 7 28 75% 20 16 36 56% Novice angler 77% 53% Dry fly 66 23 89 74% 56 27 83 67% Nymph 2 2 4 50% 1 2 3 33% Streamer 14 2 16 88% 11 12 23 48% Totals 826 251 1,077 791 390 1,181

aCapture efﬁciency for ﬂy types was generated from the overall percentage of trout landed by the number of encounters, not by individual sampling periods. 496 BLOOM

A total of 1,617 trout were landed by anglers with a mean intermediate anglers (61%), and novice anglers with the lowest ( ± SE) TL of 213 mm ( ± 2.54 mm) and a range of 64 mm to mean CE (53%). 660 mm. The trout species caught consisted of Coastal Rainbow Since two of the sampling periods had unusually high num- Trout, Lahontan Cutthroat Trout O. clarkii henshawi, California bers of encounters (over 100), I omitted them in calculating the Golden Trout O. mykiss aguabonita, Brown Trout Salmo trutta, correlation between the number of encounters per sampling pe- and Brook Trout Salvelinus fontinalis. The make-up of angler riod and the CE. With 76 periods, the nonparametric Spearman experience used in the analysis by sampling period consisted of correlation (rs = 0.30) showed a significant (P < 0.01) positive 34 advanced anglers (44%), 32 intermediate anglers (41%), and relationship. I then fitted a linear regression model. The regres- 12 novice anglers (15%). Sampling periods conducted in lotic sion slope of 0.002 (t = 2.74, df = 74, P < 0.01) for the number habitats made up the majority of the study (83%) with surveys of trout encounters was very small, suggesting that increasing in lentic habitats comprising a smaller portion (17%). Anglers trout encounters increased the CE very gradually. chose to use dry flies for the majority of the sampling periods In the logistic regression model for each hook type, duration (57%), followed by nymphs (33%), and streamers (10%). The of the encounter significantly affected the probability of landing number of encounters (mean ± SE) per sampling period was a trout for barbed (chi-square = 158.3, df = 1, P = <0.0001) and 28.9 ± 3.5, with a range of 4–192. Advanced anglers had the barbless (chi-square = 137.8, df = 1, P = <0.0001) flies. Sam- highest number of encounters per sampling period with 36 ± ple sizes for the regression models were n = 1,077 for barbed 5.3, followed by intermediate anglers with 25 ± 6.1, and novice and n = 1,181 for barbless flies. Generally, a longer encounter anglers with 18 ± 3.5. time was associated with a higher probability of landing a trout. Capture efficiency differed significantly by fly type However, the regression slopes for the fight time in both hook (ANOVA: F = 3.24; df = 2, 75; P = 0.04). The CE (mean ± types were small (0.26 for barbed and 0.14 for barbless). Thus, SE) for dry flies was the highest at 72 ± 2%, followed by the relationship of encounter time to the probability of landing streamers at 70 ± 5%, and nymphs at 63 ± 3%. In the Tukey a trout was marginal. multiple comparisons, dry flies had significantly higher CE than Anglers fishing with barbed flies landed significantly (t = nymph flies. No other comparisons of fly types were significant. 4.50, df = 77, P < 0.0001) more trout than those using barbless When CE is derived from overall trout landed by number of flies. The CE (mean ± SE) for barbed flies was higher at 76 ± encounters and fly type, barbed streamers showed the highest 2%, as opposed to the CE for barbless flies of 63 ± 3%. The CE (88%), with barbless nymphs representing the lowest CE range of CE for barbless flies was 0–100%, compared with a (33%) (Table 1). Sample sizes in some categories were very low smaller range for barbed flies at 38–100%. Since handling time and data were combined for all sample periods, hence caution was only recorded for landed trout, the two periods without should be used in interpreting CE generated in this way. landed trout out of the 78 sampling periods were removed Angler experience did not significantly affect CE (ANOVA: from handling time analysis. Handling time (mean ± SE) for F = 2.25; df = 2, 75; P = 0.11). No Tukey multiple com- trout landed with barbed flies (35.5 ± 1.8 s) was significantly parisons were significant at α = 0.10; however, for barbless longer (t = 6.31, df = 75, P < 0.0001) than for trout landed flies, there was a decreasing trend in the observed CE means with barbless flies (28.4 ± 1.5 s). Observers recorded no direct by angler experience (Figure 1). With barbless flies, advanced mortality for any trout landed. Since observers only inspected anglers experienced the highest mean CE (68%), followed by landed trout for injury, again I removed two sampling periods without injury data from the injury analysis. Trout landed by anglers using barbed flies had significantly more (t = 4.71, df =

Downloaded by [Department Of Fisheries] at 20:00 28 May 2013 75, P < 0.0001) injuries than trout landed with barbless flies. The proportion of injuries, on average, for trout landed with barbed flies was 0.24 compared with barbless flies at 0.12.

DISCUSSION Results showed that, on average, barbed flies effectively landed more trout than barbless flies, regardless of fly type and angler experience. Although this may seem intuitive for both fisheries managers and anglers, the significance in difference highlights a need for fisheries managers to justify the efficacy of barbless regulations based on biological rationale and management objectives. Fisheries managers likely adopt barbless hook regulations in an attempt to reduce hooking mortality through FIGURE 1. Mean capture efficiencies for barbed and barbless flies in associ- decreased handling times and injury. There are over 80 wa- ation with angler experience shown as percentages with one SE. ters throughout California, not including one entire county (San BARBED VERSUS BARBLESS ARTIFICIAL FLIES 497

Diego), that have barbless regulations, which allow for harvest single-hook spinners was unlikely to improve survival chances of one or more trout. These regulations reflect both anadromous nor reduce sublethal injuries of released fish. All trout captured and inland trout waters and range from seasonal to year-round were landed and processed using a soft-mesh landing net. This restrictions. Based on my results, anglers fishing these waters probably reduced both encounter and handling times. In com- will face a significantly reduced CE using barbless flies. Fly parison to Meka (2004), mean handling times were nearly half anglers interested in harvesting trout in these barbless-regulated of what she reported. These reduced times may be due to the waters will also likely have to spend more time fishing in order small mean size (213 mm; SD, 2.54) of captured trout or because to attain their bag limit, thus increasing the number of anglers on handling fish was conducted by experienced CDFW personnel, the water at any given time. Although anglers may experience not the anglers. Handling and release times for an average angler slightly reduced catch rates when using barbless flies, the poten- would likely be longer than the handling times experienced dur- tial benefits would be that more fish are retained in a population ing the study. The exposure of fish to air was minimized through over a longer period of time along with reductions in sublethal the use of landing nets and associated handling protocol. Fish injuries. These may be important components to consider for caught by average anglers, especially ones not using landing managing fisheries with high angling pressure. nets, would likely experience increased air exposure through Reduced catch rates from barbless flies may also affect di- increased handling times when using barbed flies. Trout held rected management objectives. In some cases, fisheries man- out of the water for long periods of time (>60 s) run the risk agers may be interested in specific harvest on portions of a of performance impairments, displacement downstream, or re- fishery. This is the case for a number of steelhead (anadromous lated predation (Schreer et al. 2005). Injury rates were found to Rainbow Trout) fisheries in California that are supplemented be relatively high for both hook types; however, my criteria for with hatchery fish. All hatchery steelhead stocked in California qualifying injury were very liberal. Although injuries were not are marked with an adipose clip to allow differentiation from ranked by severity, no injuries were noted in vital organs or in wild steelhead. There is currently no allowable harvest on wild areas that would likely lead to postrelease mortality. Injury rates steelhead in California; however, harvest of hatchery steelhead for trout landed with barbed flies were double that of barbless is promoted by fisheries managers. Recent regulatory changes flies. This difference highlights the need for further research have increased bag limits for hatchery steelhead throughout to assess sublethal injuries associated with barbed flies and the many California coastal waters based on this approach. Catch relationship with hook removal, direct effects of the barb, and rates in steelhead waters can be very low and any reduction in different fly angling techniques. catch rates, especially for hatchery fish, will likely have an effect Some angler groups may gauge their satisfaction with a given on management objectives (reduced average harvest on hatch- angling experience by how many trout they catch. This may be in ery fish) and angler satisfaction (less fish landed on average). the form of fish caught and released or ones that are harvested. However, caution should be used with assuming harvest will Reducing catch rates through use of barbless regulations will have any effect at the population level. Wright (1992) asserts likely affect all anglers, regardless of experience, in their ability that seasons and daily bag limits, when used by themselves, are to land and harvest trout if they are using artificial flies. The abil- probably not effective management tools because they do not ity of anglers to notice a reduction in CE is probably dependent apply to each fish that is captured. The reduction in CE from on individual angler expectation, existing catch rates, and angler barbless regulations also needs to be balanced with increased experience. Even if anglers are aware of a reduced CE, it may handling times and injury rates associated with barbed flies. be off-set by high catch rates or philosophical support of barb- The use of barbless hooks is often an important mitigating fac- less hooks. Reduced catch rates from barbless hook regulations

Downloaded by [Department Of Fisheries] at 20:00 28 May 2013 tor when assessing the efficacy of allowable take on imperiled or may not have the same effect on all angling groups in relation special-status fish. These issues are further complicated when to their overall angling experience. Novice anglers showed both there are mixed stocks of fish, with some fish that are feder- the lowest mean encounter rate and lowest mean CE (53%) of ally or state listed and others that are targeted for harvest. The all angler experience levels. Providing the best possible angling trade-off of reduced CE from barbless regulations may be small experience for novice anglers may prove essential in maintain- when the alternative could be a complete closure to allowable ing their interest in angling. In evaluating the effects of catch sportfishing. rates and associated satisfaction within specific user groups, it Along with reduced CE from a barbless hook type, results may prove very important for young anglers (<15 years) and also clearly demonstrated a reduction in both injury and han- consumptive anglers (interested in harvest) to obtain their bag dling times for trout landed with barbless flies. This informa- limit during a fishing outing (Sanyal and McLaughlin 1993). tion could prove important for managers needing justification Capture efficiency for advanced anglers was highest overall, for using barbless hooks on sensitive species or fisheries with along with having the smallest absolute difference (11%) in CE high angling pressure. However, the difference in mean han- between the two hook types. Advanced anglers, given their ex- dling times between the two hook treatments was only 7.2 s, perience and related skill, have more encounters than novice which would not likely lead to increased mortality. This assess- and intermediate anglers, which could mitigate a reduction in ment is similar to DuBois and Dubielzig (2004), who found CE. A reduction in CE may also be off-set by fisheries with high the slight decrease in unhooking times gained by using barbless catch rates, which allow for numerous encounters regardless of 498 BLOOM

experience levels. Additionally, anglers not interested in harvest Maryland. Available: fisheries.org/docs/policy statements/policy 28f.pdf. may have less of a concern regarding slight reductions in CE. (January 2009). The hook shape and size associated with the fly patterns Barnhart, R. A. 1990. Comparison of steelhead caught and lost by anglers using flies with barbed or barbless hooks in the Klamath River, California. that were selected represented just a portion of the potential California Fish and Game 76:43–45. hooks used for artificial flies by anglers. It should be noted that Behnke, R. J. 1987. Catch and release: the last word. Pages 291–298 in CE could be affected by different hook shapes and sizes. As R. A. Barnhart and T. D. Roelofs, editors. Proceedings of the symposium an example, Meka (2004) found that J-style hooks were more on catch-and-release fishing: a decade of experience. California Cooperative efficient at landing fish than circle hooks regardless of being Fish Research Unit, Humboldt State University, Arcata. Cooke, S. J., and H. L. Schramm. 2007. Catch-and-release science and its ap- barbed or barbless. Hook size and associated CE could also be plication to conservation and management of recreational fisheries. Fisheries affected by fish size. The physical ability of a fish to take a Management and Ecology 14:73–79. hook into the mouth, in relation to hook size, may play a role DuBois, R. B., and R. R. Dubielzig. 2004. Effect of hook type on mortality, in CE. Otway and Craig (1993) found that, although there was trauma, and capture efficiency of wild stream trout caught by angling with a trend toward fewer catches as hook size increased, catch rates spinners. North American Journal of Fisheries Management 24:609–616. Faragher, R. A. 2004. Hooking mortality of trout: a summary of scientific (number/time) for fish were not significantly different among studies. Australian Department of the Environment and Water Resources, size 12, 10, and 8 hooks. Additionally, the use of curved shank NSW (New South Wales) Fisheries, Fisheries Research Report Series 9, hooks, multiple flies, trailing hooks, and other terminal gear are Cronulla, Australia. other important factors that need further research in relation to Meka, J. M. 2004. The influence of hook type, angler experience, and fish size CE and injuries. on injury rates and the duration of capture in an Alaskan catch-and-release Rainbow Trout fishery. North American Journal of Fisheries Management Given the strong preference of anglers to use artificial flies in 24:1309–1321. California coldwater fisheries, specific attention should be paid Mongillo, P. E. 1984. A summary of salmonid hooking mortality. Washington to assess the effects of this angling technique and gear along Department of Game, Fisheries Management Division, Olympia. with other traditional angling approaches. Understanding and Ostrand, K. G., M. J. Siepker, and S. J. Cooke. 2006. Capture efficiencies of researching the effects of angling with associated regulations is two hook types and associated injury and mortality of juvenile muskellunge angled with live baitfish. North American Journal of Fisheries Management paramount to responsible fishery management. Having updated 26:622–627. angler preference information will also prove invaluable in es- Otway, N. M., and J. R. Craig. 1993. Effects of hook size on the catches tablishing and justifying management objectives. Maintaining of undersized snapper Pagrus auratus. Marine Ecology Progress Series 93: and managing our coldwater fisheries in California is an ever- 9–15. evolving, challenging task. As fisheries managers we have a lim- Radomski, P. J., G. C. Grant, P. C. Jacobson, and M. F. Cook. 2001. Visions for recreational fishing regulations. Fisheries 26(5):7–18. ited assortment of tools to use in maintaining, conserving, and in Ryan, T. P. 1997. Modern regression methods. Wiley, New York. many cases recovering our fisheries. Special regulations, along Sanyal, N., and W. J. McLaughlin. 1993. Angler market segmentation, angler with associated gear and seasons, are sometimes all we have to satisfaction, and activity persistence among Idahoans. Idaho Fish and Game, work with. Understanding how these approaches and techniques Department of Resource Recreation and Tourism, University of Idaho, Boise. affect managed stocks and anglers cannot be overstated. SAS (Statistical Analysis Systems). 2006. Base SAS 9.1.3 procedures guide, 2nd edition, volumes 1, 2, 3, and 4. SAS Institute, Cary, North Carolina. Schaeffer, J. S., and E. M. Hoffman. 2002. Performance of barbed and barbless ACKNOWLEDGMENTS hooks in a marine recreational fishery. North American Journal of Fisheries Management 22:229–235. I want to thank the many volunteer anglers who fished Schill, D. J., and R. L. Scarpella. 1997. Barbed hook restrictions in catch-and- long and hard along with the CDFW Heritage and Wild Trout release trout fisheries: a social issue. North American Journal of Fisheries Statewide Crew for all their support, both in angling and serv- Management 17:873–881. Downloaded by [Department Of Fisheries] at 20:00 28 May 2013 ing as observers. I also want to thank Ed Pert for his support in Schreer, J. F., D. M. Resch, M. L. Gately, and S. J. Cooke. 2005. Swimming per- initiating the study. Jeff Weaver and Stephanie Mehalick pro- formance of Brook Trout after simulated catch-and-release angling: looking for air exposure thresholds. North American Journal of Fisheries Management vided critical assistance in planning and implementing the field- 25:1513–1517. work. Bruce Olson (Umpqua Feather Merchants) and Bill Kiene Taylor, M. J., and K. R. White. 1992. A meta-analysis of hooking mortality (Kiene’s Fly Shop) provided critical assistance on fly selection of nonanadromous trout. North American Journal of Fisheries Management from the national and local level. I want to thank Donn Burton 12:760–767. and Calvin Chun for data entry, statistical analysis, and editorial Wright, S. 1992. Guidelines for selecting regulations to manage open-access fisheries for natural populations of anadromous and resident trout in stream assistance. habitats. North American Journal of Fisheries Management 12:517–527. Wydoski, R. S. 1977. Relation of hooking mortality and sublethal hooking stress to quality fishery management. Pages 43–87 in R. A. Barnhart and REFERENCES T. D. Roelofs, editors. Catch-and-release fishing as a management tool: a AFS (American Fisheries Society). 2009. Special fishing regulations for national sport fishing symposium. California Cooperative Fish Research Unit, managing freshwater sport fisheries. AFS, Policy Statement 28, Bethesda, Humboldt State University, Arcata. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:01 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Utility of Restrictive Harvest Regulations for Trophy Largemouth Bass Management Jason R. Dotson a , Micheal S. Allen b , Janice A. Kerns b & William F. Pouder c a Florida Fish and Wildlife Conservation Commission , Gainesville Fisheries Research Laboratory , 7922 Northwest 71st Street, Gainesville , Florida , 32653 , USA b School of Forest Resources and Conservation , University of Florida , 7922 Northwest 71st Street, Gainesville , Florida , 32653 , USA c Florida Fish and Wildlife Conservation Commission , Southwest Region Headquarters, Division of Freshwater Fisheries Management , 3900 Drane Field Road, Lakeland , Florida , 33811 , USA Published online: 28 Apr 2013.

To cite this article: Jason R. Dotson , Micheal S. Allen , Janice A. Kerns & William F. Pouder (2013): Utility of Restrictive Harvest Regulations for Trophy Largemouth Bass Management, North American Journal of Fisheries Management, 33:3, 499-507 To link to this article: http://dx.doi.org/10.1080/02755947.2013.769921

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:499–507, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.769921

ARTICLE

Utility of Restrictive Harvest Regulations for Trophy Largemouth Bass Management

Jason R. Dotson* Florida Fish and Wildlife Conservation Commission, Gainesville Fisheries Research Laboratory, 7922 Northwest 71st Street, Gainesville, Florida 32653, USA Micheal S. Allen and Janice A. Kerns School of Forest Resources and Conservation, University of Florida, 7922 Northwest 71st Street, Gainesville, Florida 32653, USA William F. Pouder Florida Fish and Wildlife Conservation Commission, Southwest Region Headquarters, Division of Freshwater Fisheries Management, 3900 Drane Field Road, Lakeland, Florida 33811, USA

Abstract Trophy-size fish are a critical component of recreational Largemouth Bass Micropterus salmoides floridanus fisheries; therefore, many agencies have prioritized management actions to improve catches of large fish. Length- based harvest regulations are commonly used to increase the abundance of trophy-size fish, but the rarity of large fish in sampling programs makes it difficult to use field data to evaluate the effectiveness of those regulations. We used an age-structured simulation model parameterized for a trophy Largemouth Bass fishery to evaluate the potential for a range of size limits to increase abundance and angler catches of trophy Largemouth Bass (>610 mm TL). We compiled creel information from four Florida lakes with varying harvest regulations that were known to have high-quality trophy fisheries in order to assess the performance of the model. Model results were scaled to represent trips per trophy catch for a range of size limits. The model predicted that the average number of angler trips required to catch a trophy fish were expected to decline from 83 under a 350-mm minimum length limit (e.g., baseline model that represents the standard length limit in the peninsula of Florida) to 47 for a 600-mm minimum length limit if exploitation rates were 0.2. Maximum size limits and protective slot limits also showed potential to substantially improve trophy catches. The model results and creel estimates showed similar trends for the predicted number of angler trips required to catch a trophy fish on lakes managed for trophy Largemouth Bass in Florida. Our model

Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 could be combined with fish population data to forecast the effectiveness of regulation changes on trophy fish catches. This could provide insight into trophy fisheries, where field measurements of trophy abundance and angler catches are difficult to obtain with traditional sampling programs.

Trophy-size Largemouth Bass Micropterus salmoides flori- Martell 2004; Birkeland and Dayton 2005). In recent years, the danus are a critical component of recreational fisheries from a social dynamics of anglers have included a greater emphasis on biological, sociological, and economic standpoint. Largemouth opportunities for trophy-size fish (Wilson and Dicenzo 2002). Bass are apex predators in freshwater systems that can influ- Fisheries management policies that create or enhance the pro- ence trophic dynamics through top-down processes (Sass et al. duction of trophy-size fish can have significant economic ben- 2006), and trophy-size fish are important for their reproductive efits. For example, the trophy Largemouth Bass fishery at Lake potential because fecundity increases with fish size (Walters and Fork, Texas, was estimated to be worth approximately US$27.5

*Corresponding author: [email protected] Received August 31, 2012; accepted January 17, 2013 499 500 DOTSON ET AL.

million in 1994, and anglers outside the local area accounted for The model used survivorship curves to calculate the survivors 92% of the total expenditures (Chen et al. 2003). Thus, many per recruit to each age. Survivorship to age a in the absence of fisheries agencies have attempted to improve catches of trophy- fishing was found as size fish to meet ecological targets, angler expectations, and improve local economies. la = Sala−1, (1) Length-based regulations are commonly used to prevent overfishing and create trophy fishing opportunities; however, where Sa is the age-specific finite annual natural survival rate the success of length limits to increase abundance and influence −M size structure has been inconsistent. Length-based harvest regu- (e ). The discard mortality of fish caught and released by lations are typically initiated under the assumption that exploita- anglers is an important consideration in recreational fisheries tion is reducing fish abundance or skewing the size structure where length limits can cause large numbers of fish to be released toward smaller fish. The success of length limits in increasing (Coggins et al. 2007). In Largemouth Bass fisheries, this is abundance and influencing size structure depends on the fishing particularly important because many anglers voluntarily release mortality rate, as well as the natural mortality rate, recruitment, fish that may be legally harvested (Myers et al. 2008). The model and growth (Wilde 1997; Homans and Ruliffson 1999; Allen survivorship schedule incorporated natural mortality, harvest, et al. 2002). Voluntary release rates of legal-size fish can be and discard mortality as high in some recreational Largemouth Bass fisheries, which can = − − − , result in declining fishing and total mortality rates, lessen the lfa lfa−1 Sa(1 UVa−1)(1 (UoVa−1 UVa−1)D) (2) response of fisheries to regulations, and make it more difficult to detect the effects of regulation changes (Allen et al. 2008). where lfa is the survivorship in fished condition, U is the finite Myers et al. (2008) documented large increases in the voluntary annual exploitation rate (i.e., fish that are harvested), U0 is the release rate of legal-size Largemouth Bass from the late 1970s finite annual capture rate by anglers (i.e., fraction of the fish to early 2000s. stock that is caught by anglers), Va and Va are age-specific Despite increased voluntary catch and release by anglers, vulnerabilities to harvest and capture, respectively, and D is exploitation can still influence the size structure of fish popula- the discard (catch-and-release) mortality rate. The first term, tions (Henry 2003). Restrictive regulations such as high mini- UVa−1, describes deaths due to harvest, and the second term, mum length limits, low maximum length limits, large protective − (UoVa−1 UVa−1)D, models discard mortality for fish caught slots, and mandatory catch and release can increase the number and voluntarily released by anglers. Age-specific abundance, of trophy-size fish for some fisheries (Wilson and Dicenzo 2002; Na, was estimated as the product of the number of age-1 recruits Myers and Allen 2005; Carlson and Isermann 2010). However, (Req) and the age-specific survivorship schedule. Recruitment it is often difficult to determine whether regulations are effective was modeled with a Beverton and Holt stock recruit curve with at increasing trophy fish catches. a Goodyear (1980) recruitment compensation ratio of 15. Details Gathering information about trophy-size fish is especially of the recruitment function are described in Allen et al. (2009). difficult due to their rarity in fish populations and because their The model used length-specific natural mortality rates. We habitat use may preclude collection via conventional sampling allowed survival from natural mortality (Sa) to increase with techniques (Bayley and Austen 2002). Because it is so difficult age, as per Lorenzen (2000), as follows: to use field measurements to assess the success of length-based regulations on abundance and angler catches of trophy fish, −M( TLr )c = TLa , Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 population models provide an alternative to evaluate the effi- Sa e (3) cacy of management strategies that could improve angler catch

of trophy fish (Arlinghaus et al. 2010). Our objective was to use where M is the instantaneous natural mortality rate, TLa is the an age-structured simulation model parameterized for a trophy mean total length at age, TLr is a reference length, and c is the Florida Largemouth Bass fishery to evaluate the potential of allometric exponent modifying the relationship between natural using a variety of size limits (e.g., minimum length limits, max- mortality and length. Mean total length at age, TLa, was calcu- imum length limits, protective slot limits, and mandatory catch lated from the von Bertalanffy growth model described below, and release) to increase abundance and angler catches of trophy and TLr (666 mm TL) was set at the mean total length of the Largemouth Bass. We used creel survey information from four terminal age (age 15) in the model. Florida lakes managed for trophy Largemouth Bass with varying We specified the proportion of fish vulnerable to harvest un- harvest regulations to evaluate the performance of the model. der minimum length limits and capture (Va and Va, respectively) using a logistic model. The model was METHODS We adapted an age-structured population model from Allen = 1 , Va(orV ) − (4) et al. (2009) and parameterized the model to represent a trophy a − (TLa Llow) Largemouth Bass (>610 mm TL; 3.63 kg) fishery (Table 1). 1 + e SDlow HARVESTING TROPHY LARGEMOUTH BASS 501

TABLE 1. Deﬁnitions and values of parameters used in the simulation model describing a trophy Largemouth Bass ﬁshery. The baseline model parameters are shown as values, and ranges represent the range for all simulations considered.

Parameter Definition Value Natural mortality M Instantaneous natural mortality rate 0.3 Sa Finite annual natural survival rate 0.36–0.74 c Natural mortality exponent 1 TLy Reference length (mm) for natural mortality 666 Fishing mortality U Finite annual exploitation rate 0.1–0.2 Uo Finite annual capture rate 0.4 D Recreational discard mortality rate 0.1 Capture vulnerability Llow Total length (mm) at 50% capture vulnerability 250 SDlow Standard deviation of 50% capture vulnerability 12.5 Harvest vulnerability Llow Lower total length (mm) at 50% capture vulnerability 300–700 SDlow Standard deviation of 50% capture vulnerability 15–35 Lhigh Upper total length (mm) at 50% capture vulnerability 350–600 SDhigh Standard deviation of 50% capture vulnerability 17.5–30.0 Growth L∞ Asymptotic length (mm) 670 K Growth coefficient 0.35 t0 Time at zero length (yr) 0 Length–weight a Length–weight coefficient 3.39E-06 b Length–weight exponent 3.24 Recruitment R0 Average annual unfished recruitment 1,000 CR Goodyear recruitment compensation ratio 15 Wmat Weight (kg) at maturity 0.86

where Va is the vulnerability schedule (either Va or Va), Llow is where Lhigh is the upper total length at 50% vulnerability to the lower total length at 50% vulnerability to harvest or capture, harvest, and SDhigh is the standard deviation for Lhigh. Values and SDlow is the standard deviation of the logistic distribution for of SDlow and SDhigh specify the steepness of each side of the Llow. Capture vulnerability (Va) was set with Llow of 250 mm, vulnerability curve and were set at 5% of the respective length Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 as fish below this size are seldom captured by anglers. Min- at 50% vulnerability for all simulations. imum length limits were modeled by setting Llow for harvest We modeled several length limits to explore their impact on vulnerability (Va) equal to the lower minimum length limit. the abundance and angler catches of trophy fish. An example of We specified the proportion of fish vulnerable to harvest different vulnerability functions is shown in Figure 1. Minimum under maximum length limits as the minimum of Va or 1 – length limits were simulated from 300 to 700 mm TL, and equation (4), such that fish larger than the maximum size limit maximum size limits were considered from 350 to 600 mm TL. (Llow) could not be harvested but would still be vulnerable to A 300-mm minimum length limit is essentially no regulation, as capture by anglers. Vulnerability to harvest under protective slot anglers generally harvest only fish above this size. A 700-mm limits was simulated using a double logistic function: minimum length limit reflected a catch-and-release-only fishery, since our growth parameters did not allow fish to reach that size. ⎛ ⎞ Protective slot limits were evaluated with lower limits from 300 to 400 mm TL and upper limits from 500 to 600 mm TL. ⎜ ⎟ = , − 1 − 1 , We populated the model with growth rates and mortality Va Min ⎝V 1 − − ⎠ a − (TL Llow) − (TL Lhigh) functions that would be expected for trophy Largemouth Bass 1 + e SDlow 1 + e SDhigh fisheries in Florida. All fish in the model were female Large- (5) mouth Bass (which represents the gender capable of reaching 502 DOTSON ET AL.

et al. 2008). Some Largemouth Bass fisheries have exploitation rates lower than 0.2 due to increases in the voluntary release rate of Largemouth Bass (Myers et al. 2008). Exploitation rates can have significant impacts on the efficacy of regulations because lower exploitation rates typically lessen the response of the fishery to regulation changes (Wilde 1997). Thus, we conducted additional simulations with an exploitation rate of 0.1 (all other parameters held constant) to investigate the potential to increase trophy fish abundance with more restrictive regulations on a fishery with very low exploitation. Traditional model outputs such as yield per recruit or total catch are not meaningful metrics to anglers when evaluating trophy fisheries, because they do not provide any inference about the catch occurrence of trophy-sized fish. Wilde and Pope (2004) evaluated the probability of anglers catching record fish in Texas based on catch records. We adopted a similar approach and used the model to predict the average number of angler trips required to catch a trophy fish (i.e., reciprocal of average angler catch of trophy fish per trip) as our output metric for trophy catches. We standardized the baseline model to repre- FIGURE 1. Examples of vulnerability to harvest (Va) and capture (Va ) schedules used in model simulations. The vulnerability to capture and three harvest sent a 350-mm minimum length limit (represents the standard vulnerabilities (Va for a 350-mm minimum length limit, 500-mm maximum length limit in the peninsula of Florida) and compared the other length limit, and a protective slot limit of 400–600 mm) are shown. length limit simulations with this regulation. The population model had an arbitrary 1,000 recruits to age-1 in the unfished trophy size) with length at age (La; mm) represented by La = condition, resulting in 26 trophy fish caught for the baseline −0.35(age) 670(1 – e ). Growth parameters (e.g., Linf and k)were model parameters (Table 1). To provide a realistic scale for an- set such that an age 10 female (L10 = 650 mm) would weigh gler catch rates of trophy fish for each regulation, we tuned approximately 4.5 kg (Crawford et al. 2002). Many studies that fishing effort to represent 0.012 trophy fish caught per angler evaluate growth of Largemouth Bass (e.g., Allen et al. 2002) trip for the baseline model (i.e., 26 trophy fish caught would have small sample sizes of older fish and may poorly represent require 2,165 angler trips) based on trophy fish catch per trip growth rates of trophy fish. Crawford et al. (2002) collected 810 data and Largemouth Bass angler effort from creel surveys at trophy fish from taxidermists and determined that mean age of Garcia Reservoir, Florida. Garcia Reservoir is managed with 4.5-kg fish was 9.8 years. a 350-mm minimum length limit and has a history of trophy Density-dependent growth rates can lessen the response of fish catches. Angler trips average about 4 h in Florida based regulations and thus should be considered. We ran alternate simon creel survey data (Florida Fish and Wildlife Conservation ulations to determine the impacts of density-dependent growth Commission, unpublished data). The baseline model was set to on the number of trips to catch a trophy using the Lorenzen and mimic angler catch rates of trophy fish at Garcia Reservoir, and Enberg (2002) relationship: alternate regulations were compared with this baseline based on

Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 how they would influence the number of trips required to catch Linf B = LinfL −gB, (6) a trophy fish, assuming constant fishing effort. Alternate length-based regulations (i.e., alterations to the vul- where LinfB is a linear declining function of population biomass nerability schedule) were simulated under the baseline total an- density (B), LinfL represents the asymptotic length as biomass gler effort for Garcia Reservoir (2,165 total angler trips per approaches 0, and g describes the decline in asymptotic length year) to evaluate changes in angler catch rates of trophy fish. In- per unit of biomass density. We set LinfL to 700 mm and used a creases in catch rates of trophy fish reflected increases in trophy g of 1.1, which represented the average g for nine populations fish abundance because the average capture rate Uo remained of fish estimated in Lorenzen and Enberg (2002). There were no constant during all simulations. estimates of g available for Largemouth Bass, but the estimates We then compared the model output with creel survey data from Lorenzen and Enberg (2002) allowed us to explore the from trophy fisheries at four Florida lakes with varying harvest impact of density-dependent growth on model output. regulations to assess whether the model trends were reason- The average natural mortality (M) was set at 0.3 from Allen able relative to observed creel survey data. Garcia Reservoir, et al. (1998) but was higher for small fish and decreased with fish Stick Marsh, Lake Istokpoga, and Lake Weohyakapka all have size, as per equation (3). We set the average capture rate (Uo) Largemouth Bass fisheries that receive high angler effort, have at 0.4 (Henry 2003) and the average exploitation rate (U)at0.2 factors that favor rapid fish growth, and have produced nu- based on a recent literature review for Largemouth Bass (Allen merous catches of trophy-size Largemouth Bass (Florida Fish HARVESTING TROPHY LARGEMOUTH BASS 503

and Wildlife Conservation Commission, unpublished data). We compared model predictions to the observed number of angler trips required to catch a trophy ﬁsh at these systems.

RESULTS The model showed that harvest regulations could substantially improve the probability that an angler would catch a trophy Largemouth Bass under average exploitation rates (U = 0.2). The number of trips required to catch a trophy fish was expected to decline from 83 for the baseline 350-mm minimum length limit to 47 for a 600-mm minimum length limit (Fig- ure 2). Maximum length limits were expected to decrease the number of trips required to catch a trophy fish from 84 under a 600-mm maximum length limit to 45 for a 350-mm maximum length limit (Figure 2). Protective slot limits showed similar results, with large size ranges of protected fish predicted to reduce the number of angler trips required for a trophy catch. For example, a protective slot limit of 400–500 mm was predicted to require 76 trips to catch a trophy fish, whereas a slot limit

FIGURE 3. Model-predicted number of angler trips required to catch a trophy (>610 mm TL) Largemouth Bass under protective slot limits. The lower slot length (x-axis) and upper slot length (y-axis) are shown, and isopleths represent the trips per catch.

of 300–600 mm would require only 47 trips to catch a trophy fish (Figure 3). A mandatory catch-and-release regulation was expected to produce a trophy catch occurrence every 40 trips. Thus, more restrictive length limits were expected to increase trophy fish abundance and reduce the number of angler trips required to catch a trophy Largemouth Bass. Harvest regulations had less impact on trophy fish abundance under low exploitation rates (U = 0.1). The number of trips required to catch a trophy fish were expected to decline from 57 for the baseline 350-mm minimum length limit to 43 for a 600-mm minimum length limit, a more modest decline than the simulations for average exploitation rates (U = 0.2). The

Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 baseline 350-mm minimum length limit requires fewer trips to catch a trophy under low exploitation (U = 0.1) than average exploitation (U = 0.2) because abundance of trophy-sized fish is higher when they are harvested at a low rate. Similarly, the average trips per trophy for U = 0.1 were expected to decline from 56–42 from a 600-mm maximum length limit to a 350-mm maximum length limit, and from 55 to 43 from a protective slot limit of 400–500 mm to 300–600 mm. Density-dependent growth rates also influenced the number of trips required to catch a trophy. We ran simulations with density-dependent growth for a 500-mm minimum length limit and a catch-and-release-only regulation to evaluate the impacts on trips to catch a trophy. A 500-mm minimum length limit re- FIGURE 2. Model-predicted number of angler trips required to catch a trophy sulted in a modest decrease of Linf from 670 mm to 660 mm TL (>610 mm TL) Largemouth Bass under a range of minimum length limits (top and increased the number of trips required to catch a trophy fish panel) and maximum length limits (bottom panel). by 7%. A catch-and-release-only regulation resulted in a greater 504 DOTSON ET AL.

TABLE 2. Average trophy catch per trip, trophy catch, and number of trips required to catch a trophy Largemouth Bass from creel data estimates on four Florida lakes under various harvest regulations. Range (min–max) indicates the minimum and maximum value for each category. The numbers following the lake name denote the years in which data were collected.

Trophy catch/trip Trophy catch Trips/trophy catch Lake (min–max) (min–max) (min–max) Regulation Garcia1 0.012 (0.001–0.020) 79 (8–131) 81 (49–736) 350-mm minimum Istokpoga2 0.043 (0.021–0.076) 1,098 (561–1,895) 23 (13–47) 381–610 mm protective slot Weohyakapka3 0.026 (0.016–0.037) 167 (141–193) 48 (27–63) 381–610 mm protective slot Stick Marsh4 0.043 (0.021–0.077) 314 (129–717) 21 (15–36) Catch and release

12002, 2004, 2006, 2008. 22006–2009. 32006, 2009. 4 2005, 2007–2009.

decline of Linf from 670 mm to 645 mm TL and increased the However, our study showed that the probability of an angler number of trips required to catch a trophy fish by 18%. catching a trophy fish can be improved with length limits despite The model results for the number of angler trips required to relatively low exploitation rates. This resulted because large catch a trophy fish showed trends similar to those of creel esti- minimum length limits, low maximum size limits, and large mates on Florida lakes with highly restrictive Largemouth Bass slot limits protect fish over multiple ages, which saves relatively regulations (Table 2). Creel survey estimates for Lake Stick large numbers of fish. Myers and Allen (2005) determined that Marsh, which has a mandatory catch-and-release regulation, lakes with restrictive regulations in Texas were more likely to averaged 21 trips per trophy catch compared with the model produce a trophy Largemouth Bass catch occurrence (≥5.9 kg) estimate of 40 angler trips under this regulation. The model pre- than lakes with the standard statewide minimum length limit. dicted approximately 57 angler trips per trophy catch under a Carlson and Isermann (2010) suggested that restrictive regula- protective slot limit of 380–600 mm, whereas creel estimates tions (e.g., catch and release only and maximum length limits) for a protective slot limit of 381–610 mm on lakes Istokpoga were four times more likely to improve the size structure of and Weohyakapka averaged 23 and 48 trips per trophy catch, Largemouth Bass populations than standard regulations in Min- respectively. We did not have creel survey data available on nesota lakes. Wilde (1997) found that minimum length limits did trophy fish catches for lakes with maximum size limits for com- not alter size structure in Largemouth Bass fisheries, whereas parison. The model showed a higher number of angler trips were slot limits achieved changes in size structure. Our model pre- required to catch a trophy than actual creel estimates; however, dicted that trophy catches could be altered with both protective trends in creel estimates and model simulations were similar, slot limits and minimum length limits, but effects were most with the number of angler trips required to catch a trophy fish drastic when large size ranges of fish were protected with either declining with more restrictive regulations. Our model was pa- regulation type. Most of the regulation evaluations Wilde (1997) rameterized to represent an average trophy Largemouth Bass reviewed were not as protective as the large minimum length fishery in Florida, and thus it is not surprising that individual limits and protective slot limits we simulated, and thus, differ- lakes performed differently than model simulations, likely ow- ences between our results and Wilde (1997) may result from a ing to differences in rates of growth, mortality, or recruitment. relatively low number of studies that evaluated wide protective Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 Nevertheless, all of the lakes in Table 2 are considered trophy slot limits and large minimum length limits. Largemouth Bass fisheries, and the model simulations approxi- It is not surprising that more restrictive harvest regulations mated the trend in trophy fish abundance and angler catch rates will decrease mortality and increase abundance and catch of across lakes that were under different harvest regulations. trophy-sized fish, but we sought to estimate the degree to which trophy catches could be improved under the various harvest restrictions and used a unique output metric that is useful to DISCUSSION both fisheries managers and anglers. Lower exploitation rates Our simulations suggested that length-based regulations (U = 0.1) had less of an impact on trophy fish abundance but (e.g., high minimum, low maximum, catch and release only, and still reduced the number of trips required to catch a trophy fish. large protective slot limits) could substantially increase abun- The abundance of trophy fish is higher under low exploitation dance and angler catch of trophy fish at average exploitation scenarios (with standard regulations), which makes it difficult rates. Allen et al. (2008) showed that annual exploitation rates to increase abundance with more restrictive harvest regulations. for Largemouth Bass had declined through time from about Additionally, it would be much more difficult to detect changes 0.4–0.2 (as per our simulations), which could reduce the effec- in abundance resulting from a harvest regulation change when tiveness of length limits to improve Largemouth Bass fisheries. exploitation rates are low. HARVESTING TROPHY LARGEMOUTH BASS 505

Restrictive regulations are typically not applied randomly to suggested that length limits improved population size structure lakes, making it difficult to infer a cause and effect relation- statewide compared with reference lakes and that maximum ship between harvest regulations and trophy fish abundance. length limits produced significant long-term increases in the Restrictive regulations are commonly implemented on systems percentages of fish over the length limit. Abundance of trophy that favor rapid fish growth and already have a high potential Muskellunge Esox masquinongy (≥965 mm TL) increased more of producing trophy fish (Myers and Allen 2005). Similarly, than 250% at Bone Lake, Wisconsin, following increases in the our creel survey data from four Florida lakes included popu- minimum length limit between 1982 and 1995 (Cornelius and lar trophy fisheries and nonrandom use of regulations, but the Margenau 1999). These studies utilized long-term data sets with simulation model showed similar trends to the field estimates. treatment and reference lakes, which required a significant in- Creel survey estimates of the number of angler trips per trophy vestment of time and effort. The efficacy of a model simulation catch approximated trends in model simulations, indicating that to predict responses in trophy fish production to a regulation the model performed reasonably well at predicting how a trophy change is relatively inexpensive compared with field investi- Largemouth Bass population would respond to implementation gations. Estimates of critical population metrics necessary for of a more restrictive regulation than the standard regulation (e.g., model simulations are often available for major fisheries or 350-mm minimum length limit in peninsular Florida). can be approximated using published data for similar fisheries. Our results were based on model parameters specific to tro- Fishery responses can be compared with simulated estimates phy Florida Largemouth Bass populations, and thus the results to evaluate performance of the model with a creel survey or would vary if growth, mortality, or recruitment compensation sampling program (Wilson and Dicenzo 2002; De Jesus et al. varied substantially from our parameters. Thus, the model re- 2009). sults are relevant to locations that have relatively fast growth Although expectations can vary among anglers, bodies of rates, where Florida Largemouth Bass regularly obtain trophy water, and regions, the potential to catch a trophy-size fish can sizes of 610 mm TL. Benefits of making regulations more re- be important to anglers for a variety of fisheries. For exam- strictive would be lower if growth rates and fishing mortality are ple, Briery Creek Lake, Virginia, was created with the primary lower than we considered. We used a moderate recruitment com- management objective of producing trophy Largemouth Bass, pensation of 15 based on reviews of maximum fish reproductive and 81% of Largemouth Bass anglers fishing the reservoir in- rates at low population sizes (Myers et al. 1999; Goodwin et al. dicated that the opportunity to catch a trophy was their primary 2006), and 15 represented average values for many short-lived reason for fishing (Wilson and Dicenzo 2002). Most Muskel- predators (e.g., Walleye Sander vitreus and European Whiting lunge fisheries are managed for trophy fish production (Hanson Merlangius merlangus). The Goodyear compensation ratio for et al. 1986). Angler perception surveys have indicated that the Largemouth Bass has not been estimated, but Allen et al. (2011) trophy component of catfish (Flathead Catfish Pylodictis oli- found evidence for substantial compensation for Florida Large- varis, Channel Catfish Ictalurus punctatus, and Blue Catfish mouth Bass in a pond experiment. In separate analyses, we I. furcatus; Arterburn et al. 2002), Northern Pike (Arlinghaus found that compensation ratios higher than 15 did not substan- et al. 2010), and trout (Rainbow Trout Oncorhynchus mykiss tially influence model results. However, if the compensation and Brown Trout Salmo trutta; Hutt and Bettoli 2007) fisheries ratio for Largemouth Bass was set very low (e.g., <5), then is important. Thus, fishery managers should consider strategies more restrictive harvest regulations would provide more benefit that improve trophy catches, and modeling efforts provide a way than our results showed because protecting adult fish would in- to assess outcomes. crease recruitment. Thus, there is a need to estimate recruitment Highly restrictive regulations used in our model showed a

Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 compensation for Largemouth Bass populations. common trend in decreasing the average number of trips re- Further, model simulations revealed that density-dependent quired to produce a trophy fish catch. However, all alternative growth could at least partially negate the benefits of more re- regulation simulations were conducted under the assumption strictive regulations, and managers must consider this point be- that fishing effort would not change from the baseline model fore implementing regulation changes. We used the generalized (i.e., 350-mm minimum size limit). Angling effort is influenced empirical relationship by Lorenzen and Enberg (2002) to as- by catch rates (Cox et al. 2003), indicating that fishing effort sess impacts of density-dependent growth, but more studies are could increase under alternative regulations if anglers placed a needed to quantify how changes in Largemouth Bass abundance high value on catching trophy-size fish, even if they could not influences growth rates. Quantifying how growth changes with harvest them (e.g., Wilson and Dicenzo 2002). Large increases density, both within and among populations, would further eluci- in fishing effort could lessen the effectiveness of alternative date how density-dependent growth could influence Largemouth regulations to increase catch rates of trophy-size fish because Bass population responses to regulations. angler catch rates typically diminish as effort increases, even Previous studies have shown the utility of length limits for if exploitation is low (van Poorten and Post 2005; Askey et al. improving trophy catches for other species. Pierce (2010) con- 2006). Thus, our model predictions of increased angler catch ducted a long-term evaluation (21–37 years) of length limit reg- rates could be inaccurate if angler effort changed substantially ulations for Northern Pike Esox lucius in Minnesota lakes and after a regulation was enacted. However, increases in angler 506 DOTSON ET AL.

effort could still represent fishery benefit from an economic per- fish populations. Lakes and Reservoirs: Research and Management 3: spective, and increased angler effort could be an objective of a 67–79. trophy fish regulation. More work is needed to understand how Allen, M. S., M. W. Rogers, M. J. Catalano, D. G. Gwinn, and S. J. Walsh. 2011. Evaluating the potential for stock size to limit recruitment in Largemouth angler effort responds to changes in fish abundance and other Bass. Transactions of the American Fisheries Society 140:1093–1100. fishery attributes including harvest regulations (Johnston et al. Allen, M. S., W. Sheaffer, W. F. Porak, and S. Crawford. 2002. Growth and 2010). The model presented here could be used to make predic- mortality of Largemouth Bass in Florida waters: implications for use of tions of expected angler trophy catch rates following regulation length limits. Pages 559–566 in D. P. Philipp and M. S. Ridgway, editors. changes, and creel surveys could assess whether angler effort Black Bass: ecology, conservation, and management. American Fisheries Society, Symposium 31, Bethesda, Maryland. changes in response. Use of such a model when exploring har- Allen, M. S., C. J. Walters, and R. Myers. 2008. Temporal trends in Largemouth vest regulations could allow more informed regulation choices, Bass mortality, with fishery implications. North American Journal of Fisheries particularly for trophy fisheries where data are often lacking Management 28:418–427. because of the relative rarity of trophy-size fish. Arlinghaus, R., S. Matsumura, and U. Dieckmann. 2010. The conservation and Additionally, each alternative regulation would present trade- fishery benefits of protecting large pike (Esox lucius L.) by harvest regulations in recreational fishing. Biological Conservation 143:1444–1459. offs that must be considered because we did not evaluate changes Arterburn, J. E., D. J. Kirby, and C. R. Berry Jr. 2002. A survey of angler atti- in total catch or harvest yield. A mandatory catch-and-release- tudes and biologist opinions regarding trophy catfish and their management. only regulation was the best management policy for trophy Fisheries 27(5):10–21. fish catches, but anglers must sacrifice the option to harvest Askey, P. J., S. A. Richards, J. R. Post, and E. A. Parkinson. 2006. Linking any fish. High minimum length limits can increase trophy fish angling catch rates and fish learning under catch-and-release regulations. North American Journal of Fisheries Management 26:1020–1029. catches while still allowing harvest of the largest trophy fish for Bayley, P. B., and D. J. Austen. 2002. Capture efficiency of a boat electrofisher. consumption, taxidermy, tournament participation, or weight Transactions of the American Fisheries Society 131:435–451. records. Low maximum length limits can increase trophy fish Birkeland, C., and P. K. Dayton. 2005. The importance in fishery management production while allowing for harvest of small fish from the pop- of leaving the big ones. Trends in Ecology and Evolution 20:356–358. ulation but prevents harvest of trophy catches for consumption, Carlson, A. J., and D. A. Isermann. 2010. Mandatory catch and release and maximum length limits for Largemouth Bass in Minnesota: is exploitation taxidermy, tournament participation, or record weights. Large still a relevant concern? North American Journal of Fisheries Management protective slot limits can increase trophy fish production while 30:209–220. allowing for harvest of small fish and the largest trophy fish Chen, R. J., K. M. Hunt, and R. B. Ditton. 2003. Estimating the economic but prevents the harvest of quality-sized fish for consumption, impacts of a trophy Largemouth Bass fishery: issues and applications. North taxidermy, or tournament participation. Tournament angling is American Journal of Fisheries Management 23:835–844. Coggins, L. G., Jr., M. J. Catalano, M. S. Allen, W. E. Pine III, and C. J. Walters. economically important to many fisheries and the implemen- 2007. Effects of cryptic mortality and the hidden costs of using length limits tation of highly restrictive regulations could negatively impact in fishery management. Fish and Fisheries 8:196–210. tournament anglers. Some state agencies (e.g., Florida) issue Cornelius, R. R., and T. L. Margenau. 1999. Effects of length limits on muskel- exemptions to tournament anglers to possess fish that are not lunge in Bone Lake, Wisconsin. North American Journal of Fisheries Man- legal sized for weighing but require the immediate release of agement 19:300–308. Cox, S. P., C. J. Walters, and J. R. Post. 2003. A model-based evaluation of fish after weigh-in. However, this practice could diminish the active management of recreational fishing effort. North American Journal of benefits of highly restrictive harvest regulations if tournament Fisheries Management 23:1294–1302. angling resulted in higher capture rates than we accounted for Crawford, S., W. F. Porak, D. J. Renfro, and R. L. Cailteux. 2002. Character- (40%) or if tournament-associated mortality is higher than we istics of trophy Largemouth Bass populations in Florida. Pages 567–581 in accounted for (10%). Understanding the trade-offs among reg- D. P. Philipp and M. S. Ridgway, editors. Black Bass: ecology, conservation, and management. American Fisheries Society, Symposium 31, Bethesda, Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 ulations is a critical component of fisheries management, and Maryland. regulations to improve trophy catches should be considered as De Jesus, M. J., S. J. Magnelia, and C. C. Bonds. 2009. Use of a volunteer options to benefit this component of recreational fisheries. angler survey for assessing length distribution and seasonal catch trends of trophy Largemouth Bass. North American Journal of Fisheries Management 29:1225–1231. ACKNOWLEDGMENTS Goodwin, N. B., A. Grant, A. L. Perry, N. K. Dulvy, and J. D. Reynolds. 2006. Bob Eisenhauer and Beacham Furse provided data that aided Life history correlates of density-dependent recruitment in marine fishes. Canadian Journal of Fisheries and Aquatic Sciences 63:494–509. our analyses. We would like to thank Travis Tuten, Katie Wood- Goodyear, C. P. 1980. Compensation in fish populations. Pages 253–280 in side, and Stephanie Shaw for providing reviews on a previous C. H. Hocutt and J. R. Stauffer Jr., editors. Biological monitoring of fish. draft of this manuscript. Heath, Lexington, Massachussetts. Hanson, D. A., J. R. Axon, J. M. Casselman, R. C. Haas, A. Schiavone, and M. R. Smith. 1986. Improving musky management: a review of management REFERENCES and research needs. Pages 335–341 in G. E. Hall, editor. Managing muskies: Allen, M. S., P. Brown, J. Douglas, W. Fulton, and M. Catalano. 2009. An as- a treatise on the biology and propagation of muskellunge in North America. sessment of recreational fishery harvest policies for Murray Cod in southeast American Fisheries Society, Special Publication 15, Bethesda, Maryland. Australia. Fisheries Research 95:260–267. Henry, K. R. 2003. Evaluation of Largemouth Bass exploitation and potential Allen, M. S., L. E. Miranda, and R. E. Brock. 1998. Implications of com- harvest restrictions at Rodman Reservoir, Florida. Master’s thesis. University pensatory and additive mortality to the management of selected sport- of Florida, Gainesville. HARVESTING TROPHY LARGEMOUTH BASS 507

Homans, F. R., and J. A. Ruliffson. 1999. The effects of minimum size limits Myers, R. A., J. Taylor, M. Allen, and T. F. Bonvechio. 2008. Temporal trends in on recreational fishing. Marine Resource Economics 14:1–14. voluntary release of Largemouth Bass. North American Journal of Fisheries Hutt, C. P., and P.W. Bettoli. 2007. Preferences, specialization, and management Management 28:428–433. attitudes of trout anglers fishing in Tennessee tailwaters. North American Pierce, R. B. 2010. Long-term evaluations of length limit regulations for Journal of Fisheries Management 27:1257–1267. Northern Pike in Minnesota. North American Journal of Fisheries Manage- Johnston, F. D., R. Arlinghaus, and U. Dieckmann. 2010. Diversity and com- ment 30:412–432. plexity of angler behaviour drive socially optimal input and output regulations Sass, G. G., J. F. Kitchell, S. R. Carpenter, T. R. Hrabik, A. E. Marburg, and in a bioeconomic recreational-fisheries model. Canadian Journal of Fisheries M. G. Turner. 2006. Fish community and food web responses to a whole-lake and Aquatic Sciences 67:1507–1531. removal of coarse woody habitat. Fisheries 31:321–330. Lorenzen, K. 2000. Allometry of natural mortality as a basis for assessing opti- van Poorten, B. T., and J. R. Post. 2005. Seasonal fishery dynamics of a pre- mal release size in fish-stocking programmes. Canadian Journal of Fisheries viously unexploited Rainbow Trout population with contrasts to established and Aquatic Sciences 57:2374–2381. fisheries. North American Journal of Fisheries Management 25:329–345. Lorenzen, K., and K. Enberg. 2002. Density-dependent growth as a key Walters, C. J., and S. J. D. Martell. 2004. Fisheries ecology and management. mechanism in the regulation of fish populations: evidence from among- Princeton University Press, Princeton, New Jersey. population comparisons. Proceedings of the Royal Society of London B 269: Wilde, G. R. 1997. Largemouth Bass fishery responses to length limits. Fisheries 49–54. 22(6):14–23. Myers, R. A., and M. S. Allen. 2005. Factors related to angler catch of tro- Wilde, G. R., and K. L. Pope. 2004. Anglers’ probabilities of catching record- phy Largemouth Bass in Texas reservoirs. Lake and Reservoir Management size fish. North American Journal of Fisheries Management 24:1046–1049. 21:309–315. Wilson, D. M., and V. J. Dicenzo. 2002. Profile of a trophy Largemouth Bass Myers, R. A., K. G. Bowen, and N. J. Barrowman. 1999. Maximum reproductive fishery in Briery Creek Lake, Virginia. Pages 583–592 in D. P. Philipp and rate of fish at low population sizes. Canadian Journal of Fisheries and Aquatic M. S. Ridgway, editors. Black Bass: ecology, conservation, and management. Sciences 56:2404–2419. American Fisheries Society, Symposium 31, Bethesda, Maryland. Downloaded by [Department Of Fisheries] at 20:01 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:02 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Spatial Consistency of Chinook Salmon Redd Distribution within and among Years in the Cowlitz River, Washington Katherine J. C. Klett a b , Christian E. Torgersen b , Julie A. Henning c & Christopher J. Murray d a School of Environmental and Forest Sciences , University of Washington , Box 352100, Seattle , Washington , 98195 , USA b U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, Cascadia Field Station, University of Washington , School of Environmental and Forest Sciences , Box 352100, Seattle , Washington , 98195 , USA c Washington Department of Fish and Wildlife , 600 Capitol Way North , Olympia , Washington , 98501 , USA d Pacific Northwest National Laboratory , Post Office Box 999 , Richland , Washington , 99352 , USA Published online: 28 Apr 2013.

To cite this article: Katherine J. C. Klett , Christian E. Torgersen , Julie A. Henning & Christopher J. Murray (2013): Spatial Consistency of Chinook Salmon Redd Distribution within and among Years in the Cowlitz River, Washington, North American Journal of Fisheries Management, 33:3, 508-518 To link to this article: http://dx.doi.org/10.1080/02755947.2013.778924

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:508–518, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.778924

ARTICLE

Spatial Consistency of Chinook Salmon Redd Distribution within and among Years in the Cowlitz River, Washington

Katherine J. C. Klett* School of Environmental and Forest Sciences, University of Washington, Box 352100, Seattle, Washington 98195, USA; and U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, Cascadia Field Station, University of Washington, School of Environmental and Forest Sciences, Box 352100, Seattle, Washington 98195, USA Christian E. Torgersen U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, Cascadia Field Station, University of Washington, School of Environmental and Forest Sciences, Box 352100, Seattle, Washington 98195, USA Julie A. Henning Washington Department of Fish and Wildlife, 600 Capitol Way North, Olympia, Washington 98501, USA Christopher J. Murray Paciﬁc Northwest National Laboratory, Post Ofﬁce Box 999, Richland, Washington 99352, USA

Abstract We investigated the spawning patterns of Chinook Salmon Oncorhynchus tshawytscha on the lower Cowlitz River, Washington, using a unique set of fine- and coarse-scale temporal and spatial data collected during biweekly aerial surveys conducted in 1991–2009 (500 m to 28 km resolution) and 2008–2009 (100–500 m resolution). Redd locations were mapped from a helicopter during 2008 and 2009 with a hand-held GPS synchronized with in-flight audio recordings. We examined spatial patterns of Chinook Salmon redd reoccupation among and within years in relation to segment-scale geomorphic features. Chinook Salmon spawned in the same sections each year with little variation among years. On a coarse scale, 5 years (1993, 1998, 2000, 2002, and 2009) were compared for reoccupation. Redd locations were highly correlated among years. Comparisons on a fine scale (500 m) between 2008 and 2009 also revealed a high degree of consistency among redd locations. On a finer temporal scale, we observed that Chinook Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 Salmon spawned in the same sections during the first and last week. Redds were clustered in both 2008 and 2009. Regression analysis with a generalized linear model at the 500-m scale indicated that river kilometer and channel bifurcation were positively associated with redd density, whereas sinuosity was negatively associated with redd density. Collecting data on specific redd locations with a GPS during aerial surveys was logistically feasible and cost effective and greatly enhanced the spatial precision of Chinook Salmon spawning surveys.

*Corresponding author: [email protected] Received August 10, 2012; accepted February 14, 2013

508 SPATIAL CONSISTENCY OF CHINOOK SALMON REDD DISTRIBUTION 509

Chinook Salmon Oncorhynchus tshawytscha are known to restoration. Additional information about where salmon are spawn consistently in the same areas year after year, and yet spawning and why they choose certain areas allows for more there have been few published papers that have statistically efficient management of the fishery. Fisheries managers can use evaluated spatiotemporal consistency in spawning patterns of this information to identify the locations and geomorphic char- salmon (Oncorhynchus spp.) over decades. Furthermore, previ- acteristics of reaches that are occupied repeatedly by salmon ous research on spawning distributions has been conducted at and to more effectively protect these areas. The objectives of the reach scale (∼101 m; Geist et al. 2000) and at the basin scale this study were to (1) examine temporal consistency (i.e., reoc- (∼103 m or larger; Isaak and Thurow 2006; Isaak et al. 2007 cupation) of spawning locations within a single year and among [see a comprehensive review of previous research on salmonid years, (2) determine if Chinook Salmon redds were distributed spawning distribution at various spatial scales by Beechie et al. in distinct aggregations throughout the Cowlitz River, and 2008]). To fully understand spawning patterns, it is important (3) investigate the segment-scale habitat features that may be to examine patterns at multiple scales, both spatially and tem- affecting redd distribution. porally (Geist and Dauble 1998; Fausch et al. 2002). On a reach scale, Geist et al. (2000) investigated the spatial and temporal patterns of fall Chinook Salmon spawning sites in a spatial METHODS × grid with cells that were 20 20 m wide in 1994 and 1995 in Study Area the Hanford Reach, Columbia River (Washington). Isaak and The Cowlitz River is located in southwestern Washington Thurow (2006) examined basin-scale patterns over an entire (Figure 1), and the lower Cowlitz basin encompasses approx- watershed and among tributaries in central Idaho, using 28- imately 1,140 km2. The 82-km study section is located on the km sections for analysis. In contrast to these previous studies, we evaluated spawning patterns of Chinook Salmon in the lower 80 km of the Cowlitz River at the segment scale (∼102 m), which is intermediate to the reach (∼101 m) and basin (>103 m) scales (Frissell et al. 1986). Previous studies of redd distribution at a reach scale have emphasized patterns of sediment and water quality parameters in and adjacent to individual redds (Bjornn and Reiser 1991; Geist et al. 2000; Geist et al. 2002; Malcolm et al. 2003; Moir et al. 2004). To complement previous work at finer spatial scales, we investigated the effects of geomorphic features at a segment scale. Channel bifurcation (Dauble et al. 2003), sinuosity (Dauble and Geist 2000), tributaries (Martin et al. 2004; Rice et al. 2008), depth discontinuities (Brunke and Gonser 1997), and channel gradient (Dauble and Geist 2000) have been identified as potential segment-scale controls on redd distribution. Fall and spring Chinook Salmon in the Cowlitz River are part of the Lower Columbia River Evolutionarily Significant Unit and are listed as threatened under the Endangered Species

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 Act (Good et al. 2005). The number of wild adult fall Chinook Salmon spawning in the Cowlitz River was estimated at 100,000 adults historically but has dropped to less than 2,000 individuals recently (LCFRB 2004). There were 1,620 redds counted during aerial surveys in 2007, including both wild and hatchery spring and fall Chinook Salmon (Henning 2008). This decline in Chi- nook Salmon has been attributed to numerous causes but is most likely due to overﬁshing and habitat degradation. In the lower Cowlitz basin, there have been many human-caused changes such as dams, dredging, diking, and straightening of the main channel, which have had a negative effect on salmon habitat (LCFRB 2004). This habitat is crucial for juvenile salmon, but it also has a large impact on spawning adults and on the survival of their eggs (Sear and DeVries 2008). FIGURE 1. Study area on the lower Cowlitz River in southwestern Wash- ington. Study sections 1–8 are demarcated by black bar symbols labeled with Determining the patterns of Chinook Salmon spawning the section number in a circle. River kilometer (rkm) markers at 5-km intervals within the Cowlitz River is necessary for conservation and indicate the location upstream from the mouth of the Cowlitz River. 510 KLETT ET AL.

main-stem Cowlitz River between Kelso and the Barrier Dam, Spring and fall Chinook Salmon both spawn in the main stem which is a diversion dam that allows fish to be captured and of the Cowlitz River, and we did not differentiate between the released in the upper river. Above the Barrier Dam, the Cowlitz two runs when redds are counted during aerial surveys. There River has three hydroelectric dams: Mayfield Dam, Mossyrock is evidence from carcass surveys that spring Chinook Salmon Dam, and Cowlitz Falls Dam. The Cowlitz River Salmon Hatch- typically spawn in late September, during the time of the first ery is located at the Barrier Dam and has been operated by the aerial flight. For fish management purposes, WDFW uses a run Washington Department of Fish and Wildlife (WDFW) with timing date of September 30 to differentiate spring and fall adult support from Tacoma Power since 1968. Spring and fall Chinook salmon to the hatcheries. Carcass surveys show that the number Salmon and Coho Salmon O. kisutch are produced in the hatch- of fall Chinook Salmon is consistently much higher than the ery but also spawn naturally in the Cowlitz River and upstream number of spring Chinook Salmon, even during the time pe- tributaries (Tipping and Busack 2004). There is no volitional riod when the two runs overlap. For example, in 2008 and 2009 fish passage through the Barrier Dam, and all fish that spawn the population estimates for spring Chinook Salmon were 425 in historic spawning areas above the Barrier Dam have been and 763, respectively. Estimates for fall Chinook Salmon during transported (Fulton 1968). The geologic setting of the lower these two years were 2,100 and 2,800, respectively (Q. Daugh- Cowlitz valley consists of Eocene basalt flows and flow brec- erty, Washington Department of Fish and Wildlife, unpublished cias and has a typical maritime climate with warm (16–24◦C), data). Because the relative abundance of spring Chinook Salmon dry (17–89 mm precipitation) summers and cool (8–17◦C), wet was so low in comparison to fall Chinook Salmon, we refer to (100–200 mm precipitation) winters. The majority of precipi- Chinook Salmon redds in general for the purposes of this study. tation occurs as rain between October and March, resulting in We do, however, refer to the two different runs in our analysis peak flows occurring during these months; however, peak flows of the spatial distribution of the first and last redds. occasionally occur in the spring due to snowmelt. The flows be- In 2008 and 2009, a digital audio recorder (Olympus Digi- low the Barrier Dam are regulated by hydroelectric dams, and tal Voice Recorder WS-2105) and a global positioning system any extremes in the hydrograph are moderated by flow regula- (GPS; Garmin GPSmap 60CS) with a track log of geographic tion. Below Mayfield Dam, 80% of the land use within the basin positions recorded at 1-s intervals were used to map the loca- is commercial timber harvest. Human population increases in tions of each redd or cluster of redds. The recorded accuracy Cowlitz and Lewis counties have also resulted in increased resi- specified by the GPS at the time of measurement was approxi- dential, industrial, and agricultural development along the river. mately 15 m. All observations were made while the helicopter Forest climax species are western hemlock Tsuga heterophylla, was in flight, but we were not able to hover over each individual Douglas fir Psuedotsuga menziesii, and western red cedar Thuja redd. The entire flight was recorded in the GPS track log, and plicata, while red alder Alnus rubra, black cottonwood Popu- the times when redds were observed were noted on the audio lus trichocarpa, bigleaf maple Acer macrophyllum, and willow recorder. When there were too many redds to identify individ- Salix spp. dominate the riparian areas (LCFRB 2004). ually, we attempted to count all of the redds in that area and then associated one GPS point with the aggregation of redds. Data Collection This was often done in the section between Mill Creek Boat Field surveys.—Aerial surveys of redds on the lower Cowlitz Launch and the Barrier Dam (Section 8) because there were River have been conducted by WDFW for Tacoma Power since high densities of redds and it was difficult to map each redd the 1940s (Mark LaRiviere, Tacoma Power, personal communi- accurately with single GPS points. Visser et al. (2002) docu- cation) to evaluate population trends. We used data from 1991 mented that redd counts are often underestimated due to high

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 to 2009 in this study because they were easily accessible and densities; therefore, these estimates may be slightly lower than the methods from these years were most consistent. Four to six the actual number. The length of Section 8 was approximately helicopter flights a year were conducted on a biweekly basis, de- 500 m, and this distance was used as the minimum bin length pending on weather and river conditions, from mid-September (i.e., spatial resolution) for spatial analysis. Redd locations were through December. The accuracy and availability of redd data georeferenced in the laboratory by synchronizing the GPS track were dependent on weather and river conditions, which could log with the digital audio recordings from the flight. This in- cause very poor visibility due to high runoff during heavy rain- formation was then transferred into a geographic information fall. Chinook Salmon redds were easily observed from the air system (GIS) database (ArcMap 9.3; Environmental Systems due to their large size and high visibility (Isaak and Thurow Research Institute, Redlands, California). 2006). The river was divided into eight sections that ranged in To characterize the depth patterns throughout the river, a length from 0.5 to 28.0 km and were demarcated by landmarks Solinst LTC Levelogger Junior pressure transducer (vertical easily identified from the air, such as boat ramps and bridges accuracy = 0.01 m; Model 3001; Solinst, Georgetown, On- (Figure 1). For each survey, Chinook Salmon redds were counted tario) was towed behind a drift boat, near the river bottom, to and recorded in each section. Each individual redd was counted record barometric pressure measurements every 2 s (Vaccaro whenever possible, but when densities were too high to count and Maloy 2006). The drift boat floated with the current or accurately, the number of redds was estimated. was rowed at approximately 4 km/h in slow-moving sections SPATIAL CONSISTENCY OF CHINOOK SALMON REDD DISTRIBUTION 511

during the 3-d float of the river. Field measurements of river depth were conducted in August 2009. A GPS track log with 1-s time stamps was used to georeference the barometric pressure measurements. The GPS and the Solinst Levelogger were synchronized to an atomic clock prior to deployment. Spatial analysis and GIS.—Geomorphic variables at the 500- m segment scale that were evaluated in relation to redd distribution included distance upstream from the river mouth (river kilometer), channel bifurcation, tributary junctions, sinuosity, channel gradient, and depth discontinuities. To quantify these geomorphic variables, we used 1:24,000 U.S. Geological Sur- vey (USGS) topographical maps, 2009 National Agriculture Imagery Program aerial photos (1 m resolution), and 10-m digi- FIGURE 2. Schematic diagram of a bifurcated river channel with an island tal elevation models in ArcMap. A stream line was created using and two river channels. The black line indicates the length of side channel used the National Agriculture Imagery Program photos. Linear ref- in calculations of channel characteristics per 500-m bin. erencing was used to assign route measures to the entire river, redds, and geomorphic features. Schulte 2009; Yan et al. 2010) for each segment. To calculate Segment-scale depth discontinuities are similar to transition channel gradient, we used a method similar to Dauble and Geist areas between pools and riffles but on a larger scale (e.g., 500 m). (2000) based on USGS 1:24,000 topographical maps. The average relative depth over each segment was calculated from the barometric pressure measurements. We identified rela- Statistical Analysis tive differences in the depth profile, where the river transitioned Five different statistical analyses were conducted to examine from a decreasing depth to an increasing depth. The absolute the reoccupation, randomness, and habitat associations of the amount of change in depth was not calculated. Segments with redds (Table 1). Fine-scale (500-m) spatial data were only avail- depth discontinuities may have increased hyporheic–surface wa- able in 2008 and 2009. The coarse-scale spatial data were used ter interactions, which are known to affect redd site selection by only to investigate reoccupation historically. Reoccupation was spawning salmon (Brunke and Gonser 1997). examined at coarse and fine spatial and temporal scales. The River characteristics that were evaluated included river kilo- fine-scale spatial data from 2008 and 2009 were used to de- meter, channel bifurcation, tributary junctions, sinuosity, and termine whether redds were distributed randomly or in distinct channel gradient. To examine channel morphology, we mea- aggregations. Regression analysis was performed to assess as- sured the length of multiple channels associated with islands in sociations between habitat features and the number of redds; each section (Figure 2); the total length of channels by section this analysis incorporated the fine-scale spatial data from 2009. provided a relative measure of the degree of channel bifurcation. Reoccupation and spatial aggregation.—Reoccupation Channel bifurcation was defined as the splitting of a channel between the locations of historic redds (starting in 1991) into two or more active channels. To identify tributary junc- and current redds in 2009 was examined by comparing redd tions, we used a USGS topographical map and a 10-m digital counts in each section of the river over time (Figure 1). Due to elevation model to calculate flow accumulation (ArcMap Spa- availability of redd data, statistical analysis was performed for tial Analyst toolbox). We included tributaries that appeared in five nonconsecutive years: 1993, 1998, 2000, 2002, and 2009

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 both the flow accumulation analysis and were indicated on the (Table 1). We chose survey years that had at least five flights USGS topographical maps. To calculate sinuosity, we divided over the entire length of the river. There were five flights in 2004, the main-stem river route into 500-m segments and then used but because spawning distributions in 2002 and 2004 were very Hawth’s Tools (Beyer 2004) to measure sinuosity (Rayburn and similar, data from 2002 was used as that year had more redds. We

TABLE 1. Spatial and temporal scales of data and survey years used in statistical analyses.

Statistical analysis Spatial scale Years Weeks Reoccupation Coarse-scale redd data 1993, 1998, 2000, All 2002, and 2009 Reoccupation Fine-scale redd data 2008, 2009 All Reoccupation Fine-scale redd data 2008, 2009 First and last Spatial aggregation Fine-scale redd data 2008, 2009 All Generalized linear model (GLM) Fine-scale redd data and 2009 All segment-scale habitat data 512 KLETT ET AL.

did not use data from 2008 because those data were very similar 1 year. Multicollinearity was addressed using variance inflation to data from 2009 (as shown by the fine-scale comparison factor analysis and correlation matrices (Mansfield and Helms between 2008 and 2009). These 5 years represented a large range 1982). To investigate autocorrelation among residuals, we gen- in the annual number of redds (714–5,643 between 2000 and erated an autocorrelation function plot for the final model resid- 2002). Some years had fewer flights due to weather conditions uals. Variables were removed using a backwards-elimination or water visibility (i.e., high flows). To statistically analyze re- stepwise regression with an elimination value of P = 0.05 (Kut- occupation, we ranked the sections by redd density (number/m2) ner et al. 2004). Distributions that were examined as possibilities in each year. We used R (R Development Core Team 2009) to for the GLM included the Poisson distribution and the nega- calculate a correlation matrix with associated P-values and ad- tive binomial distribution or quasi-Poisson distribution, which justed P-values, using Holm’s method, to account for multiple are both statistically appropriate when data are overdispersed correlations (Wright 1992). To quantify the linear trend that was (Lawless 1987; Ver Hoef and Boveng 2007). Akaike informa- observed, we also calculated a correlation coefficient between tion criterion (AIC) values and weights were used to validate the section number and the rank for each of the 5 years analyzed. stepwise variable selection (Buckland et al. 1997; Burnham and To evaluate reoccupation on a finer spatial scale between Anderson 2002; Torgersen and Close 2004). Alignment between 2008 and 2009 (Table 1), we divided the river into 500-m sec- peaks from observed data and fitted model values was visually tions and determined whether each section was occupied during evaluated as a metric of model fit. any week in 2008 or 2009. We used R to create 2 × 2 contingency tables and calculate the G-statistic, which is an analog RESULTS for the χ2 test of independence (Zar 1999). Contingency tables were calculated based on whether a section was (1) occupied Reoccupation and Spatial Aggregation in both 2008 and 2009, (2) never occupied, or (3) occupied in Chinook Salmon consistently spawned in the same propor- one year or the other. The G-statistic test compared the observed tions in the same sections of the Cowlitz River during the 5 years distribution to a completely random distribution and determined that were examined (Figure 3). The two sections closest to the whether there was more reoccupation than would have been ex- Barrier Dam consistently had the highest density of redds. There pected, given a random distribution of spawning locations in were two reaches in which sections alternated in rank (i.e., sec- the 2 years. We also examined reoccupation of redd locations tions 2 and 3 and sections 5 and 6). The lowest correlation coeffi- within years for survey data in 2008 and 2009 (Table 1) with con- cient between sections was 0.90, and the adjusted P-values were tingency tables and the G-statistic (Zar 1999). Because spring all less than 0.002. This indicated that (1) the ranks of sections Chinook Salmon typically spawn during the first week and fall by redd density for all of the years were highly correlated and (2) Chinook Salmon spawn during the later weeks of our survey pe- there was little variation in the ranks of the sections among years. riod, we examined whether the potential combination of spring The longitudinal trend in redd density with increasing density and fall runs would result in different distributions in the first upstream was generally consistent among sections (Figure 3). versus the last weeks of the survey period. The lowest correlation coefficient between section number and We compared the spatial pattern of redd distribution in 2008 rank was 0.93, thus supporting the observation of the linear trend and 2009 to a random distribution using standard and median in Figure 3. runs analysis (α = 0.05; Turechek and Madden 1999; Borradaile For the fine-scale reoccupation analysis of redd distributions 2003; Gent et al. 2006). For the standard runs analysis, a section in 2008 and 2009, the G-statistic was 17.14 (P < 0.001), indi- was coded with either “0” if there were no redds, or “1” if there cating that redds occurred in the same locations in these 2 years

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 were redds present. The median runs analysis calculated the (Figure 4). Spatial patterns of redd density were very similar median number of redds per year and then coded a section “0” in 2008 and 2009, even though the number of redds was 120% if n ≤ median or “1” if n > median. To ensure that the sampling higher in 2009 (n = 2,728) than it was in 2008 (n = 1,247). distribution of the total number of runs followed a normal dis- Thus, we used data from 2009 for the habitat association study. tribution, we counted the zeroes and ones and verified that there The redds from the first and last weeks of the survey period were at least nine of each (Borradaile 2003). Redds were de- occurred in the same sections in 2008 (P < 0.02) and 2009 termined to be clustered if the number of runs was significantly (P < 0.001). Redds were clustered in 2008 and 2009 based lower than what would be expected given a random distribution. on both the standard and median runs analyses (P < 0.001; Segment-scale habitat associations.—We used multiple re- Figure 4). gression with a generalized linear model (GLM) to assess the relationships between redd locations from 2009 and the six ge- Segment-Scale Habitat Associations omorphic variables (river kilometer, channel bifurcation, sinu- Multicollinearity among explanatory variables in the 500-m osity, tributary junction presence, depth discontinuity presence, model was not detected, and variance inflation factor values were and channel gradient; Kutner et al. 2004). The analysis was com- slightly above 1 and none were larger than 2.0. Autocorrelation pleted at the 500-m scale for redd counts in 2009, which had values were not larger than the 95% confidence interval lines, higher redd counts than in 2008. We used the reoccupation re- nor were there significant longitudinal trends; therefore, we as- sults to determine whether it would be sufficient to evaluate only sumed that there was no significant autocorrelation (Neumann SPATIAL CONSISTENCY OF CHINOOK SALMON REDD DISTRIBUTION 513 Downloaded by [Department Of Fisheries] at 20:02 28 May 2013

FIGURE 3. Ranks of redds/m2 by study section for 1993, 1998, 2000, 2002, and 2009. A rank of 8 indicates the greatest number of redds/m2. Section locations are depicted in Figure 1.

et al. 2003). When using a Poisson distribution for the GLM, Stepwise model selection and AIC values and weights in- the data were overdispersed (Ver Hoef and Boveng 2007). The dicated that the best model for predicting redd distribution in quasi-Poisson and negative binomial regressions both resulted 2009 included river kilometer, channel bifurcation, and sinuosity in the same signiﬁcant variables. The results from the negative (Table 2; Figure 5). Redd density was positively associated with binomial regression were used because (1) they included AIC river kilometer and channel bifurcation and negatively associ- values, which we used for model comparison purposes, and ated with sinuosity (Table 3; Figure 5). River kilometer was the (2) quasi-Poisson regressions cannot calculate AIC values (Ver most signiﬁcant variable but not the only important variable, as Hoef and Boveng 2007). indicated by changes in AIC values. The 500-m model correctly 514 KLETT ET AL. Downloaded by [Department Of Fisheries] at 20:02 28 May 2013

FIGURE 4. Spatial distribution of Chinook Salmon redds in 2008 and 2009 in the Cowlitz River downstream of the Barrier Dam. Pointer arrows indicate section breaks and the numbers indicate the section label.

predicted the locations of 74% of the peaks in the observed data, from previous studies. For example, our analyses were at the seg- but the model often underpredicted the total number of redds ment scale (80 km) and incorporated relatively high-resolution expected in those peaks (Figure 6). spatial data (500 m). Isaak and Thurow (2006) used cumulative curves and the Shannon–Wiener diversity index to demonstrate that redds were distributed nonrandomly in the Salmon River DISCUSSION watershed in central Idaho, but these analyses were conducted Chinook Salmon in the Cowlitz River spawned in clusters at a basin scale. Neville et al. (2006) also examined random- and reoccupied the same areas at different spatial and temporal ness in the Salmon River redd data at a number of spatial scales scales. To assess redd distribution at these different scales, we (1, 2, 5, 10, and 20 km) and found that redds were distributed in used data and analytical methods at scales that were different clusters; however, they used autocorrelation function plots for SPATIAL CONSISTENCY OF CHINOOK SALMON REDD DISTRIBUTION 515

TABLE 2. Candidate GLMs and corresponding AIC values and weights. The TABLE 3. Estimated coefficients from a GLM of segment-scale stream habi- top model was selected using backwards-elimination stepwise regression and tat variables explaining the distribution of Chinook Salmon redds in the Cowlitz AIC values and weights. The model with the lowest AIC was identified as River, Washington. Abbreviations are as follows: RKM = river kilometer, the best model. Greater differences between respective models based on AIC CB = channel bifurcation, and SIN = sinuosity. weights indicate a better fit. Abbreviations are as follows: RKM = river kilometer, CB = channel bifurcation, SIN = sinuosity, TJP = tributary junction Parameter presence, DDP = depth discontinuity presence, and CG = channel gradient. Variable estimate SE Z-value P-value AIC AIC Intercept 4.00 2.01 Model value weight RKM 7.06 × 10−5 8.25 × 10−6 8.56 <0.01 CB 1.86 × 10−3 8.29 × 10−4 2.24 0.03 RKM + CB + SIN 795.43 0.52 SIN −5.16 1.88 −2.75 0.01 RKM + CB + SIN + TJP 796.45 0.31 RKM + CB + SIN + TJP + CG 798.26 0.13 RKM + CB + SIN + TJP + DDP + CG 800.24 0.05 generally understood that salmon spawn in clusters in the same areas each year; however, it is important to statistically examine these assumptions at different spatial and temporal scales to their analysis. Geist et al. (2000) investigated Chinook Salmon confirm that they are valid. spawning at a reach scale using a nearest neighbor analysis and By comparing redd distribution between the first versus the contingency tables and found that spawning occurred in clus- last weeks of the spawning period, we were able to deter- tered patterns in the same areas for two different years. It is mine whether spring and fall Chinook Salmon were consistently spawning in the same areas in the Cowlitz River. The first redds of the year are built by spring Chinook Salmon, whereas the later redds are fall Chinook Salmon redds (Henning 2008). The redds from the first and last weeks of the survey period occurred in the same areas in both 2008 and 2009 (Figure 4). This is unexpected because typically in the Columbia River basin there is spatial separation between spring and fall Chinook Salmon spawning grounds, with spring Chinook Salmon spawning in smaller tributaries and upper reaches of principal tributaries and fall Chinook Salmon spawning in the lower river tributaries and main stem (Fulton 1968). For example, fall Chinook Salmon on the Yakima River only spawn in the lower river, typically below Granger, Washington, while spring Chinook Salmon spawn at least 100 km farther upstream, between the Easton Dam and El- lensburg, Washington, and in connecting tributaries (Major and Mighell 1969). On the Cowlitz River, spring Chinook Salmon spawn in the same areas as fall Chinook Salmon. Spring Chi- nook Salmon cannot migrate past the Barrier Dam and have

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 to be transported in order to spawn in historic spawning areas (Fulton 1968). Therefore, some spring Chinook Salmon spawn in the lower portions of the Cowlitz River, which are preferred by fall Chinook Salmon. Our analysis used 500-m sections; therefore, we cannot conclude that redd superimposition is occurring in these highly used areas. However, the inability of spring Chinook Salmon to fully access their historic spawning areas suggests that migration barriers could cause competition between fall and spring Chinook Salmon for quality spawning habitat. River kilometer was the most signiﬁcant predictor of the spawning locations for the 500-m model predicting redd dis- FIGURE 5. Longitudinal proﬁles of bifurcated channel length, sinuosity, chan- tribution in the Cowlitz River. The linear pattern of increasing nel gradient (%), and redds (number/500 m) from 2009 in the lower Cowlitz River. Gray dashed vertical lines demarcate peaks in the 2009 Chinook Salmon spawning density up to the Barrier Dam is partially explained by redd distribution relative to spatial patterns of bifurcated channel length, sinu- the location of the Cowlitz River Salmon Hatchery adjacent to osity, and channel gradient. the Barrier Dam. This pattern also could be caused by the barrier 516 KLETT ET AL.

able for spawning (Dauble and Geist 2000). High densities of salmon redds in the Cowlitz River were more likely related to islands and associated increased hyporheic flow as opposed to the increased area of the riverbed available for spawning. For example, a single split channel on the Cowlitz River was designated by fisheries managers in WDFW as an entire section for aerial surveys because of the high density of redds in the area. Dauble and Geist (2000) found that Chinook Salmon spawning in the Hanford Reach, Washington, were concentrated in FIGURE 6. Longitudinal spatial patterns of observed redds (number/500 m) braided river sections and areas that had complex channel for- from 2009 and GLM fitted values. Black circles indicate locations where peaks mations. Coulombe-Pontbriand and Lapointe (2004) also found were observed in both the observed redds and those predicted by the model. that large numbers of redds of Atlantic Salmon Salmo salar occurred near the upstream margins of channel islands in the Pe- to the migration of salmon to their historic spawning grounds. tite Cascapedia´ and Bonaventure rivers on the Gaspe Peninsula, Historically, all spring Chinook Salmon spawned above the Bar- Quebec. rier Dam. There were also high-quality spawning grounds for Sinuosity was negatively associated with spawning locations fall Chinook Salmon directly above the Barrier Dam and farther in the Cowlitz River. The entire Cowlitz River had a very low upstream in a 13-km section that was inundated by Mayfield sinuosity (1.6) and was often confined by levees and develop- Reservoir (Fulton 1968). On the Yakima River, there is a similar ment, particularly on the lower river. Previous studies on the linear trend in redd distribution up to the Easton Dam, which Columbia and Snake rivers have shown that areas with higher cuts off access to historic spring Chinook Salmon spawning ar- sinuosity have higher Chinook Salmon redd density at a 1.6-km eas (Dittman et al. 2010). There is fish passage at Easton Dam, scale (Dauble and Geist 2000). Additionally, Fukushima (2001) but it is only operational during certain flows and most fish now found that Sakhalin Taimen Hucho perryi preferred to spawn in do not pass the dam. Martin et al. (2004) also observed large sites located below highly sinuous reaches. However, sinuosity numbers of Chinook Salmon returning to spawn directly below was only important as a predictor variable at a small scale a fish barrier on the Green River that prevented migration to (50 m). In higher-sinuosity streams that are not bounded by historical spawning areas. channel constraints, sinuosity has been associated with channel Another factor that may contribute to the strong association morphology and pool–riffle complexes that are associated of redd counts and river kilometer is that habitat quality in- with higher redd densities (McKean et al. 2008). However, on creases on the Cowlitz River from rkm 30–82. In the first 30 km the Cowlitz River, sinuosity was not correlated with channel of the Cowlitz River, no redds were observed in 2008 and 2009; bifurcation or depth discontinuities from pool–riffle complexes. this area did not appear to have high-quality spawning areas, This is a potential explanation for the negative association presumably due to high concentrations of silt and residential de- between sinuosity and redd density in our study. velopment in the lower Cowlitz River. Upstream of these highly Tributary junctions, depth discontinuities, and channel gradi- developed residential areas, there is substantially less channel ent were not significant predictors of redd density on the Cowlitz modification (e.g., levees and bank armoring), and the overall River. These variables were positively associated with redd loca- quality of spawning habitat is greater. Furthermore, sediment tions in previous studies at a reach and larger spatial scales. Trib- inputs in downstream reaches of the Cowlitz River have been utaries can affect sediment inputs and create optimal locations

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 high historically. For example, the Cowlitz River received large for spawning (Rice et al. 2008). On the North Fork Stillaguamish amounts of fine sediment from the Toutle River, which enters the River, Rice et al. (2008) found that patterns of Chinook Salmon Cowlitz River at rkm 25, due to the eruption of Mt. St. Helens in redd distribution were not associated with tributaries at 1.1- 1980. The volcanic eruption moved large amounts of sediment km scale. However, tributaries were associated with spawning into both rivers, and recovery to preeruption sediment yields in patterns at smaller spatial scales. Discontinuities in depth can the Cowlitz basin is not predicted for another 10 years (Major also indicate areas of increased hyporheic–surface water ex- et al. 2000). change. Previous studies found redds in tailouts of pools and Channel bifurcation was positively associated with the oc- at the boundaries between pools and riffles (Bjornn and Reiser currence of spawning in reaches in the Cowlitz River. Mul- 1991), where a change in depth was associated with increased tiple channels are associated with intragravel flow on a large hyporheic–surface water exchange. These patterns have been scale that is critical for incubating salmon (Brunke and Gonser described at multiple spatial scales (Baxter and Hauer 2000). 1997; Geist 2000). Increased intragravel flow often occurs at We investigated these associations at a 500-m scale and found the upstream and downstream ends of channel bars and is- that depth discontinuities were not associated with redd loca- lands, where the river is slower and shallower (Brunke and tions in the Cowlitz River. Although we observed no association Gonser 1997; Dauble and Geist 2000). Multiple channels also with channel gradient, lower-gradient areas typically have well- can be associated with an increased area of the riverbed avail- developed floodplains and gravel bars, and these areas have been SPATIAL CONSISTENCY OF CHINOOK SALMON REDD DISTRIBUTION 517

shown to have high densities of Chinook Salmon redds in the ulation viability of Chinook Salmon in the Cowlitz River and in Columbia River basin (Fulton 1968; Dauble and Geist 2000; other similar rivers. Dauble et al. 2003). The GLM regression approach that we used effectively pre- ACKNOWLEDGMENTS dicted areas of peak redd density (74% accuracy; Figure 6), We thank Ryan Klett, Matt Groce, and Jeremy Cram for but it underpredicted the number of redds at these peaks. The assistance with field work, Ethan Welty for GIS expertise and model also predicted that redds would be present in some lo- programming in R statistical software (linbin package), and cations where no redds were observed. The underprediction by Mark LaRiviere, from Tacoma Power, for providing logistical the model of the number of redds at the peaks may be attributed support and background information on the Cowlitz River. We to other factors and the scale(s) at which these factors were would also like to thank Gino Lucchetti, Susan Bolton, and measured (Torgersen et al. 2012). For example, at the basin two anonymous reviewers for their constructive comments and scale, the number of returning fish depends on ocean condi- recommendations. The School of Environmental and Forest Sci- tions, which were not incorporated into the model. Conversely, ences at the University of Washington and the U.S. Geological at small spatial scales, information such as sediment grain size, Survey, Forest and Rangeland Ecosystem Science Center pro- water chemistry, and velocity measurements, may be needed vided partial funding for this work. Any use of trade, product, or to obtain more accurate predictions. We found that segment- firm names is for descriptive purposes only and does not imply scale features accurately predicted where redds occurred in the endorsement by the U.S. Government. Cowlitz River; however, to fully understand the observed patterns in redd distribution, a more detailed assessment at reach and basin scales is needed (Lapointe 2012). Further work on REFERENCES the Cowlitz River could involve spatially continuous surveys Baxter, C. V., and F. R. Hauer. 2000. Geomorphology, hyporheic exchange, and selection of spawning habitat by Bull Trout (Salvelinus confluentus). of sediment size and other reach-scale variables that could be Canadian Journal of Fisheries and Aquatic Sciences 57:1470–1481. included in predictive models (Brenkman et al. 2012). Beechie, T., H. Moir, and G. Pess. 2008. Hierarchical physical controls on salmonid spawning location and timing. Pages 83–101 in D. A. Sear and Management Implications P. DeVries, editors. Salmonid spawning habitat in rivers: physical controls, biological responses, and approaches to remediation. American Fisheries Collecting location data with a GPS and digital audio Society, Symposium 65, Bethesda, Maryland. recorder during aerial surveys of redd distribution is cost- Beyer, H. L. 2004. Hawth’s analysis tools for ArcGIS. Spatial Ecology, Toronto. effective and allows much greater flexibility for analysis of redd Available: www.spatialecology.com. (April 2010). distribution across spatial scales. Many aerial redd surveys still Bjornn, T. C., and D. W. Reiser. 1991. Habitat requirements of salmonids in rely on paper maps and require redd counts to be tallied in streams. Pages 83–138 in W. R. Meehan, editor. Influences of forest and rangeland management on salmonid fishes and their habitats. American Fish- flight between landmarks over long sections (0.5–28.0 km) of eries Society, Special Publication 19, Bethesda, Maryland. river. With such low-precision techniques, important informa- Borradaile, G. 2003. Statistics of earth science data: their distribution in time, tion about redd distribution patterns is lost, and analyses can space and orientation. Springer-Verlag, Berlin. only be conducted at coarse spatial scales. We found that rel- Brenkman, S. J., J. J. Duda, C. E. Torgersen, E. Welty, G. R. Pess, R. Peters, atively inexpensive GPS and digital audio technology can be and M. L. McHenry. 2012. A riverscape perspective of Pacific salmonids and aquatic habitats prior to large-scale dam removal in the Elwha River, used effectively in an aircraft and requires minimal postpro- Washington, USA. Fisheries Management and Ecology 19:36–53. cessing to analyze in a GIS. Furthermore, this approach can be Brunke, M., and T. Gonser. 1997. The ecological significance of exchange used by fisheries managers who may have minimal training or processes between rivers and groundwater. Freshwater Biology 37:1–33.

Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 experience with GPS or GIS. Buckland, S. T., K. P. Burnham, and N. H. Augustin. 1997. Model selection: an Maps of salmon redd distribution and aquatic habitat are im- integral part of inference. Biometrics 53:603–618. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel portant tools in riverine fisheries management because salmon inference: a practical information-theoretic approach, 2nd edition. Springer- may spawn in the same sections of the river year after year, Verlag, New York. and this information can be used to prioritize habitat conser- Coulombe-Pontbriand, M., and M. Lapointe. 2004. Geomorphic controls, riffle vation and restoration efforts. In many rivers with anadromous substrate quality, and spawning site selection in two semi-alluvial salmon salmonids in North America, river reaches with multiple chan- rivers in the Gaspe´ Peninsula, Canada. River Research and Applications 20:577–590. nels provide more suitable areas for conservation and restora- Dauble, D. D., and D. R. Geist. 2000. Comparison of mainstem spawning tion. Knowledge of where salmon redds occur in relation to habitats for two populations of fall Chinook Salmon in the Columbia River channel features that can be mapped using the publically avail- basin. Regulated Rivers: Research and Management 16:345–361. able digital maps and data sources described in this study can Dauble, D. D., T. P.Hanrahan, D. R. Geist, and M. J. Parsley. 2003. Impacts of the help river managers plan reach- and segment-scale restoration Columbia River hydroelectric system on main-stem habitats of fall Chinook Salmon. North American Journal of Fisheries Management 23:641–659. efforts with greater confidence and potentially use engineered Dittman, A. H., D. May, D. A. Larsen, M. L. Moser, M. Johnston, and D. logjams to increase channel complexity (Roni et al. 2002). Fast. 2010. Homing and spawning site selection by supplemented hatchery- Segment-scale information of the type described by this study and natural-origin Yakima River spring Chinook Salmon. Transactions of the can be used in fisheries management to promote long-term pop- American Fisheries Society 139:1014–1028. 518 KLETT ET AL.

Fausch, K. D., C. E. Torgersen, C. V. Baxter, and H. W. Li. 2002. Landscapes ment of Natural Resources and Parks, Water and Land Resources Division, to riverscapes: bridging the gap between research and conservation of stream Seattle. fishes. BioScience 52:483–498. McKean, J. A., D. J. Isaak, and C. W. Wright. 2008. Geomorphic controls on Frissell, C. A., W. J. Liss, C. E. Warren, and M. D. Hurley. 1986. A hierarchical salmon nesting patterns described by a new, narrow-beam terrestrial–aquatic framework for stream habitat classification: viewing streams in a watershed lidar. Frontiers in Ecology and the Environment 6:125–130. context. Environmental Management 10:199–214. Moir, H. J., C. N. Gibbins, C. Soulsby, and J. Webb. 2004. Linking chan- Fukushima, M. 2001. Salmonid habitat–geomorphology relationships in low- nel geomorphic characteristics to spatial patterns of spawning activity and gradient streams. Ecology 82:1238–1246. discharge use by Atlantic Salmon (Salmo salar L.). Geomorphology 60: Fulton, L. A. 1968. Spawning areas and abundance of Chinook Salmon (On- 21–35. corhynchus tshawytscha) in the Columbia River basin: past and present. U.S. Neumann, D. W., B. Rajagopalan, and E. A. Zagona. 2003. Regression model for Fish and Wildlife Service Special Scientific Report Fisheries 571. daily maximum stream temperature. Journal of Environmental Engineering Geist, D. R., and D. D. Dauble. 1998. Redd site selection and spawning habitat 129:667–674. use by fall Chinook Salmon: the importance of geomorphic features in large Neville, H. M., D. J. Isaak, J. B. Dunham, R. F. Thurow, and B. E. Rieman. rivers. Environmental Management 22:655–669. 2006. Fine-scale natal homing and localized movement as shaped by sex and Geist, D. R., T. P. Hanrahan, E. V. Arntzen, G. A. McMichael, C. J. Murray, spawning habitat in Chinook Salmon: insights from spatial autocorrelation and Y. J. Chien. 2002. Physicochemical characteristics of the hyporheic zone analysis of individual genotypes. Molecular Ecology 15:4589–4602. affect redd site selection by Chum Salmon and fall Chinook Salmon in the R Development Core Team. 2009. R: a language and environment for statisti- Columbia River. North American Journal of Fisheries Management 22:1077– cal computing. R Foundation for Statistical Computing, Vienna. Available: 1085. www.R-project.org. (April 2010). Geist, D. R., J. Jones, C. J. Murray, and D. D. Dauble. 2000. Suitability crite- Rayburn, A. P., and L. A. Schulte. 2009. Landscape change in an agricultural ria analyzed at the spatial scale of redd clusters improved estimates of fall watershed in the U.S. Midwest. Landscape and Urban Planning 93:132–141. Chinook Salmon (Oncorhynchus tshawytscha) spawning habitat use in the Rice, S. P., P. Kiffney, C. Greene, and G. R. Pess. 2008. The ecological im- Hanford Reach, Columbia River. Canadian Journal of Fisheries and Aquatic portance of tributaries and confluences. Pages 209–242 in S. Rice, A. Roy, Sciences 57:1636–1646. and B. Rhoads, editors. River confluences, tributaries and the fluvial network. Gent, D. H., W. F. Mahaffee, and W. W. Turechek. 2006. Spatial heterogene- Wiley, Chichester, UK. ity of the incidence of powdery mildew on hop cones. Plant Disease 90: Roni, P., T. J. Beechie, R. E. Bilby, F. E. Leonetti, M. M. Pollock, and G. R. Pess. 1433–1440. 2002. A review of stream restoration techniques and a hierarchical strategy Good, T. P., R. S. Waples, and P. Adams, editors. 2005. Updated status of for prioritizing restoration in Pacific Northwest watersheds. North American federally listed ESUs of West Coast salmon and steelhead. NOAA Technical Journal of Fisheries Management 22:1–20. Memorandum NMFS-NWFSC-66. Sear, D. A., and P. DeVries, editors. 2008. Salmonid spawning habitat in rivers: Henning, J. A. 2008. Cowlitz River evaluation program annual report for 2007. physical controls, biological responses, and approaches to remediation. Amer- Washington Department of Fish and Wildlife, Olympia. ican Fisheries Society, Symposium 65, Bethesda, Maryland. Isaak, D. J., and R. F. Thurow. 2006. Network-scale spatial and temporal vari- Tipping, J. M., and C. A. Busack. 2004. The effect of hatchery spawning proto- ation in Chinook Salmon (Oncorhynchus tshawytscha) redd distributions: cols on Coho Salmon return timing in the Cowlitz River, Washington. North patterns inferred from spatially continuous replicate surveys. Canadian Jour- American Journal of Aquaculture 66:293–298. nal of Fisheries and Aquatic Sciences 63:285–296. Torgersen, C. E., C. V. Baxter, J. L. Ebersole, and R. E. Gresswell. 2012. Isaak, D. J., R. F. Thurow, B. E. Rieman, and J. B. Dunham. 2007. Chinook Incorporating spatial context into the analysis of salmonid–habitat relations. Salmon use of spawning patches: relative roles of habitat quality, size, and Pages 216–224 in M. Church, P. M. Biron, and A. G. Roy, editors. Gravel-bed connectivity. Ecological Applications 17:352–364. rivers: processes, tools, environments. Wiley, Chichester, UK. Kutner, M. H., C. J. Nachtsheim, and J. Neter. 2004. Applied linear regression Torgersen, C. E., and D. A. Close. 2004. Influence of habitat heterogeneity on models, 4th edition. McGraw-Hill, New York. the distribution of larval Pacific Lamprey (Lampetra tridentata) at two spatial Lapointe, M. 2012. River geomorphology and salmonid habitat: some examples scales. Freshwater Biology 49:614–630. illustrating their complex association, from redd to riverscape scales. Pages Turechek, W. W., and L. V. Madden. 1999. Spatial pattern analysis of straw- 191–215 in M. Church, P. M. Biron, and A. G. Roy, editors. Gravel-bed rivers: berry leaf blight in perennial production systems. Phytopathology 89: processes, tools, and environments. Wiley, Chichester, UK. 421–433. Lawless, J. F. 1987. Negative binomial and mixed Poisson regression. Canadian Vaccaro, J. J., and K. J. Maloy. 2006. A thermal profile method to identify Downloaded by [Department Of Fisheries] at 20:02 28 May 2013 Journal of Statistics 15:209–225. potential ground-water discharge areas and preferred salmonid habitats for LCFRB (Lower Columbia Fish Recovery Board). 2004. Lower Columbia long river reaches. U.S. Geological Survey, Scientific Investigations Report salmon recovery and fish and wildlife subbasin plan, volume IIE. Cowlitz 2006-5136, Reston, Virginia. River subbasin plan. LCFRB, Longview, Washington. Ver Hoef, J. M., and P. L. Boveng. 2007. Quasi-Poisson vs. negative bino- Major, J. J., T. C. Pierson, R. L. Dinehart, and J. E. Costa. 2000. Sediment yield mial regression: how should we model overdispersed count data? Ecology following severe volcanic disturbance: a two-decade perspective from Mount 88:2766–2772. St. Helens. Geology 28:819–822. Visser, R., D. D. Dauble, and D. R. Geist. 2002. Use of aerial pho- Major, R. L., and J. L. Mighell. 1969. Egg-to-migrant survival of spring Chinook tography to monitor fall Chinook Salmon spawning in the Columbia Salmon (Oncorhynchus tshawytscha) in the Yakima River, Washington. U.S. River. Transactions of the American Fisheries Society 131:1173– National Marine Fisheries Service Fishery Bulletin 67:347–359. 1179. Malcolm, I. A., A. F. Youngson, and C. Soulsby. 2003. Survival of salmonid Wright, S. P. 1992. Adjusted P-values for simultaneous inference. Biometrics eggs in a degraded gravel-bed stream: effects of groundwater–surface water 48:1005–1013. interactions. River Research and Applications 19:303–316. Yan, B., M. D. Tomer, and D. E. James. 2010. Historical channel movement and Mansfield, E. R., and B. P. Helms. 1982. Detecting multicollinearity. American sediment accretion along the South Fork of the Iowa River. Journal of Soil Statistician 36:158–160. and Water Conservation 65:1–8. Martin, D., L. Benda, and D. Shreffler. 2004. Core areas: a framework for Zar, J. H. 1999. Biostatistical analysis, 4th edition. Prentice-Hall, Upper Saddle identifying critical habitat for salmon. Report to King County Depart- River, New Jersey. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:05 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Acclimation Enhances Postrelease Performance of Hatchery Fall Chinook Salmon Subyearlings While Reducing the Potential for Interaction with Natural Fish Stuart J. Rosenberger a , William P. Connor b , Christopher A. Peery b , Deborah J. Milks c , Mark L. Schuck c , Jay A. Hesse d & Steven G. Smith e a Idaho Power Company , 1221 West Idaho Street, Boise , Idaho , 83702 , USA b U.S. Fish and Wildlife Service , Idaho Fishery Resource Office , 276 Dworshak Complex Drive, Orofino , Idaho , 83544 , USA c Washington Department of Fish and Wildlife , Snake River Laboratory , 401 South Cottonwood Street, Dayton , Washington , 99328-1277 , USA d Nez Perce Tribe Department of Fisheries Resources Management , Post Office Box 365, Lapwai , Idaho , 83540 , USA e National Marine Fisheries Service , Northwest Fisheries Science Center , 2725 Montlake Boulevard East, Seattle , Washington , 98112-2097 , USA Published online: 29 Apr 2013.

To cite this article: Stuart J. Rosenberger , William P. Connor , Christopher A. Peery , Deborah J. Milks , Mark L. Schuck , Jay A. Hesse & Steven G. Smith (2013): Acclimation Enhances Postrelease Performance of Hatchery Fall Chinook Salmon Subyearlings While Reducing the Potential for Interaction with Natural Fish, North American Journal of Fisheries Management, 33:3, 519-528 To link to this article: http://dx.doi.org/10.1080/02755947.2013.768567

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:519–528, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.768567

MANAGEMENT BRIEF

Acclimation Enhances Postrelease Performance of Hatchery Fall Chinook Salmon Subyearlings While Reducing the Potential for Interaction with Natural Fish

Stuart J. Rosenberger* Idaho Power Company, 1221 West Idaho Street, Boise, Idaho 83702, USA William P. Connor and Christopher A. Peery U.S. Fish and Wildlife Service, Idaho Fishery Resource Office, 276 Dworshak Complex Drive, Orofino, Idaho 83544, USA Deborah J. Milks and Mark L. Schuck Washington Department of Fish and Wildlife, Snake River Laboratory, 401 South Cottonwood Street, Dayton, Washington 99328-1277, USA Jay A. Hesse Nez Perce Tribe Department of Fisheries Resources Management, Post Office Box 365, Lapwai, Idaho 83540, USA Steven G. Smith National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, Washington 98112-2097, USA

with natural fish while in transit through the reservoirs associated Abstract with Lower Granite, Little Goose, and Lower Monumental dams; One form of prerelease acclimation of hatchery anadromous and (2) confinement with natural fish at those three dams, where salmonid Oncorhynchus spp. juveniles is to truck the fish to re- fish collection and raceway holding were followed by transport in mote points for extended holding at low densities in rearing vessels tanker trucks. Our findings support acclimation as a method for en- (e.g., tanks, raceways, or in-ground ponds) supplied with river hancing postrelease performance of hatchery fall Chinook Salmon water. We conducted a 3-year study to determine whether such subyearlings and reducing their potential interactions with natural Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 acclimation enhanced the postrelease performance of hatchery fall conspecifics. Chinook Salmon O. tshawytscha subyearlings and reduced the potential for interaction with natural fall Chinook Salmon subyear- A critical component of an effective fish culture program lings. In comparison with hatchery subyearlings that were released is to provide time for the fish to recover after transport and directly into the lower Snake River just downstream of the accli- acclimate to a new environment. Acclimation is a strategy com- mation facility, acclimated hatchery subyearlings (1) passed downstream to Lower Monumental Dam (the third dam encountered monly used to allow hatchery juvenile anadromous salmonids during seaward migration) faster, (2) passed the dam earlier, and Oncorhynchus spp. to recover from transport and imprint to re- (3) survived from release to the dam tailrace at higher rates. The lease water, thus facilitating their return to that location as adults differences in downstream passage rate and dam passage timing (Dittman et al. 2010). Extended acclimation practices involve were also much greater between acclimated hatchery subyearlings transport to remote sites where fish are held, for days to weeks, and natural subyearlings than between directly released hatchery subyearlings and natural subyearlings. Thus, acclimation pro- at low densities in tanks, raceways, or in-ground ponds supplied vided a survival advantage to the hatchery fish while reducing the with river water (Bugert 1998). The effect of acclimation on sur- potential for (1) aggressive and nonaggressive social interactions vival to adulthood and straying of anadromous salmonids has

*Corresponding author: [email protected] Received July 25, 2012; accepted January 9, 2013 519 520 ROSENBERGER ET AL.

been evaluated for Coho Salmon O. kisutch, steelhead (anadro- over time; presently, six acclimation facilities are operated, and mous Rainbow Trout O. mykiss), and spring Chinook Salmon fish are also transported for direct release at three other locations O. tshawytscha (Johnson et al. 1990; Whitesel et al. 1994; Ke- (Figure 1). The number of hatchery subyearlings released by the naston et al. 2001; Appleby et al. 2002; Clarke et al. 2010). How- program increased from roughly 333,000 fish in 1997 to roughly ever, studies of acclimation effects on juvenile salmon perfor- 5,000,000 fish after 2008. The program focuses on returning mance have produced mixed results. For example, Whitesel et al. hatchery adults to river stretches that support natural spawning. (1994) found that acclimation did not influence the downstream Two of the four remaining major spawning aggregates of natural passage rate but did increase survival of juvenile steelhead rela- fall Chinook Salmon (and the associated acclimation facilities tive to non-acclimated (directly released, the typical operation) for these fish) are found along the lower Snake River. The lower juveniles, whereas Clarke et al. (2010) found the opposite for Snake River aggregates spawn (1) between Hells Canyon Dam the same species. Acclimation studies of juvenile salmon also and the Salmon River confluence and (2) between the Salmon tend to focus on the performance of acclimated versus directly River confluence and the upper end of Lower Granite Reservoir, released hatchery fish. We found no studies that have explored which is formed by Lower Granite Dam (Figure 1). Our in- acclimation as a tool for moderating interactions between hatch- vestigation focused on hatchery subyearlings released from the ery and natural fish, but most literature reviews support the con- acclimation facility at Captain John Rapids (Figure 1) during servative view of limiting such interactions (e.g., Jonsson and peak spring runoff in late May, coincident with the dispersal Jonsson 2006; Kostow 2009; but see Tatara et al. 2011). of a large portion of natural subyearlings from natal riverine We studied the influence of acclimation on the postrelease habitats into Lower Granite Reservoir. performance of Snake River basin hatchery fall Chinook Salmon After hatchery subyearlings are released from acclimation (age 0; hereafter, subyearlings) during 2008–2010, with an em- facilities, the potential for their interaction with natural phasis on evaluating the potential for interaction with natural subyearlings is spatially and temporally dependent. Postrelease conspecifics. The hatchery fish originated from Lyons Ferry dispersal of hatchery subyearlings into Lower Granite Reser- Hatchery (Figure 1), which began operation in 1984 to mitigate voir is rapid and facilitated by high riverine velocities and for the loss of natural production of fall Chinook Salmon at- turbulence (Smith et al. 2003; Tiffan et al. 2009). Thus, natural tributed to dams constructed on the lower Snake River (Bugert subyearlings that rear along the shorelines prior to dispersal et al. 1995; Bugert 1998). To supplement production in the wild, into the reservoir (Connor et al. 2003) spend little time with a portion of the hatchery production was released as acclimated hatchery subyearlings in riverine habitat. After reservoir entry, subyearling smolts in 1997. The program evolved and expanded the marked decline in velocity that delays both natural and Downloaded by [Department Of Fisheries] at 20:05 28 May 2013

FIGURE 1. Map of the Columbia River and lower Snake River, including the area (lower Snake River from Hells Canyon Dam to the upper end of Lower Granite Reservoir) where natural fall Chinook Salmon subyearlings were seined, PIT-tagged, and released at the point of capture. Other relevant locations include Lyons Ferry Hatchery (black square), where the hatchery fall Chinook Salmon subyearlings were cultured; the Captain John Rapids acclimation facility (gray circle), where hatchery subyearlings were acclimated prior to release; and the Couse Creek boat launch (open circle), where hatchery subyearlings were directly released as part of this study. The Nez Perce Tribal Hatchery (open square) also produces hatchery ﬁsh for acclimation and on-station release. Five additional acclimation facilities (black circles) are operated in the Snake River basin, and direct releases are made at two additional locations (cross-hatched circles). Asterisks denote dams that are equipped with PIT tag monitoring systems. MANAGEMENT BRIEF 521

hatchery subyearling migrants (e.g., Tiffan et al. 2009) also Hells Canyon Dam. To increase the number of natural subyear- concentrates them and increases the potential for aggressive and lings that were tagged with PIT tags, supplemental stations were nonaggressive interactions; however, most of the information sampled during the last 2 weeks of May and first week of June, on these types of interaction comes from studies of low-order coincident with peak dispersal into Lower Granite Reservoir. streams or from laboratory studies (e.g., review by Weber Sampling ended in early July once the catch was near zero. and Fausch 2003). Aggressive interaction can prematurely Natural subyearlings were distinguished morphologically from displace natural fish from preferred feeding habitats (Chapman hatchery subyearlings, received an implanted PIT tag (8.5-mm 1962), increase physical contact (McMichael et al. 1999), tags in 50–59-mm FL fish; 12-mm tags in fish > 59 mm FL), increase energy expenditure (Peery and Bjornn 2004), and and were released at the collection site as described by Tiffan decrease growth (McMichael et al. 1997). Nonaggressive and Connor (2011). mutual attraction and schooling of natural and hatchery fish It was not possible to culture the different Lyons Ferry can result in premature downstream movement of natural fish Hatchery release groups by using identical methods because that otherwise would linger and continue to grow (Hansen and spawn dates either were unknown or differed between years. In Jonsson 1985; Hillman and Mullan 1989). Although little is 2008, the spawn date was not recorded for either of the hatchery known about density-dependent interactions between natural subyearling groups. In 2009, the spawn date for the acclimated and hatchery fish in large water bodies, including Lower Granite release group was 2 weeks earlier than the spawn date for the Reservoir, it is intuitive that the speed at which hatchery fish direct-release group. In 2010, the spawn date was identical pass downstream would affect the duration of their interaction for the two hatchery subyearling groups. Fertilized eggs were with natural fish, provided that migration timing overlaps. incubated in 12◦C well water at Lyons Ferry Hatchery. In As migration continues, the subyearlings must pass eight February, fry from both groups were transferred to raceways main-stem dams on the Snake and Columbia rivers to reach the (3 raceways per release strategy; 30 m long × 3mwide × Pacific Ocean (Figure 1). Fish primarily pass the dams via spill- 1 m deep) at a rearing density of approximately 1 kg/m3 (i.e., ways, through a bypass after being guided away from turbine density index = 0.05 lb·ft−3·in−1; Piper et al. 1986). Flow intakes by screens (e.g., Giorgi et al. 1988), or under the screens (single-pass well water at a rate of 3,028 L/min), temperature and through the powerhouse. Fish that pass the Lower Granite, (12◦C), and feed ration were identical for all raceways. Little Goose, Lower Monumental, and McNary dams via the In mid-April of each year (37–43 d prior to release), a random bypasses are routed to raceways, from which they are loaded subsample of the hatchery subyearlings from each of the six onto either a barge or a tanker truck for transport and release raceways was routed to a tagging trailer, and the fish in the downstream of Bonneville Dam on the Columbia River (e.g., sample were measured to the nearest millimeter. Fish longer Ward et al. 1997; Figure 1). The potential for physical aggres- than 59 mm FL in the random samples received 12-mm PIT tags sion, bodily contact, elevated stress, and disease transmission that were implanted via the techniques described by Prentice between natural and hatchery subyearlings is highest when the et al. (1990a; as modified by McCutcheon and Richmond 2010). fish are confined together in the raceways, barges, and trucks The overall percentages of sampled fish that were longer than (e.g., Congleton et al. 2000; Van Gaest et al. 2011). Thus, the 59 mm were 92% in 2008, 97% in 2009, and 87% in 2010, degree of similarity in dam passage timing between hatchery indicating that the PIT-taggedfish represented the majority of the and natural subyearlings will influence the level of interaction untagged population of hatchery fish. Posttagging mortality and at this stage of migration. PIT tag retention were not consistently or accurately monitored In this paper, we compare (1) the downstream passage rate prior to release. Thus, we assumed that there was no difference

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 (as measured between release into the lower Snake River and in mortality or tag retention between acclimated and directly dam passage) among acclimated hatchery subyearlings, directly released hatchery subyearlings. released hatchery subyearlings, and natural subyearlings; (2) the On 5 and 6 May 2008, 4 and 5 May 2009, and 4 May 2010, timing of dam passage among the three subyearling groups; and hatchery subyearlings that were slated for acclimated release (3) survival from release to dam passage between acclimated were loaded onto aerated tanker trucks at an approximate den- and directly released hatchery subyearlings. The premise of the sity of 10 kg/m3 (i.e., loading density = 0.74 lb/gal; Piper et al. survival analysis is that a survival advantage for either of the 1986) and were transported for 2 h to the acclimation facility at hatchery subyearling groups will persist through adult return. Captain John Rapids. Upon arrival at the facility, ﬁsh were piped into an in-ground pond (38 m long × 24 m wide × 1 m deep) that was supplied with river water (∼9,000 L/min). Fish were METHODS not tempered before release into the pond because temperatures Data collection.—We used a beach seine to capture natural were similar between the tanker truck (12.0◦C in all years) and subyearlings at sites in the lower Snake River during 2008– the pond (10.6◦C in 2008, 10.9◦C in 2009, and 10.3◦C in 2010). 2010. Sampling began at the onset of fry emergence in late Initial rearing density in the pond across years was approxi- March and was conducted 3 d/week at a total of 15 permanent mately 2–3 kg/m3 (density index = 0.04–0.06 lb·ft−3·in−1). The stations between the upper end of Lower Granite Reservoir and duration of acclimation was 23 d in 2008, 22 d in 2009, and 20 d 522 ROSENBERGER ET AL.

in 2010. Daily mean temperature during acclimation across detected. We downloaded PIT tag release and detection data years ranged from 9.4◦C to 14.4◦C. Feed ration was approx- from the regional PIT tag database (PTAGIS 2012) for imately 2% of body weight per day. Fish in the direct-release analyses. group remained at Lyons Ferry Hatchery, where rearing density Preliminary analyses.—Our study was one of several stud- in the raceways was maintained at approximately 7.5 kg/m3 ies being conducted with fish from Lyons Ferry Hatchery. When (density index = 0.12 lb·ft−3·in−1), water temperature was the hatchery subyearlings averaged about 55 mm FL during late 12◦C, and feed ration was approximately 2% of body weight March to early April, predetermined proportions of fish in each per day. raceway were marked with either an adipose fin clip or both an In mid-May of each year, 60 subyearlings at the acclima- adipose fin clip and a coded wire tag (Jefferts et al. 1963) to aid tion facility and in the direct-release group at the hatchery were in the management and monitoring of harvest and the estimation randomly selected for prerelease disease testing. An enzyme- of adult return rates. Due to differences in hatchery evaluation linked immunosorbent assay was used to test for Renibacterium objectives that were beyond our control, the mark rate was lower salmoninarum antigen (Elliott et al. 1989). Gill, kidney, and for subyearlings in the acclimated group (roughly 33%) than for spleen tissue was also examined for viruses associated with in- subyearlings in the direct-release group (100%). To ensure that fectious pancreatic necrosis, infectious hematopoietic necrosis, differences in mark rate between acclimated and directly re- and viral hemorrhagic septicemia. The optical densities from leased hatchery subyearlings did not affect our conclusions, we all assays were less than 0.09 (i.e., low), and viral tests were conducted preliminary analyses (all statistical tests of differ- all negative. Mortality was low and commensurate with routine ences were performed at α = 0.05) by using the methods de- fish culture practices. scribed below to compare the 2008–2010 downstream passage The Couse Creek boat launch, which is located along rates, dam passage timing, and survival of marked and unmarked the Washington shore of the Snake River 9 km downstream acclimated hatchery subyearlings that were PIT-tagged for our of the Captain John Rapids acclimation facility, was chosen as study. We found no significant or consistently directional effect the direct-release site because it was close to the acclimation of marking. Accordingly, we formed the final PIT tag data set facility and was accessible to the tanker truck (Figure 1). Fish for acclimated fish by pooling data from marked and unmarked were trucked to the boat launch under the same conditions as acclimated hatchery subyearlings. described for the transfer of fish to the acclimation facility. Fish Final analyses.—We selected Lower Monumental Dam as were not tempered before release because temperatures were the point for calculations of downstream passage rate, dam pas- similar between the tanker truck (12.0◦C in all years) and the sage timing, and survival. One reason for this selection was Snake River (11.2◦C in 2008, 12.1◦C in 2009, and 11.2◦Cin that Lower Monumental Dam was the furthest downstream 2010). Acclimated fish were forced into the river (i.e., not vo- point where adequate numbers of fish were detected to main- litionally released) on the same days that direct releases were tain precise estimates of the response variables. A second rea- made from the tanker truck (28 May 2008, 26 May 2009, and son was that results obtained from analyses of data collected 24 May 2010). Length and weight were not measured prior to at Lower Granite, Little Goose, and Lower Monumental dams release because it was not possible to collect representative sam- were largely consistent and redundant (e.g., 2005 and 2006 pilot ples from the in-ground pond at the acclimation facility without results; Rosenberger 2008). stressing the fish. Thus, the final rearing density of fish at the We calculated the downstream passage rate (km/d) for each acclimation facility was estimated based on growth projections detected hatchery and natural subyearling as the channel dis- and observed feeding levels. The estimated rearing density in tance between the release point and Lower Monumental Dam

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 the acclimation pond immediately prior to forced release was divided by the elapsed time between release and detection. We 2.7 kg/m3 (density index = 0.05 lb·ft−3·in−1) in 2008; 3.2 kg/m3 used a square-root transformation of the downstream passage (density index = 0.06 lb·ft−3·in−1) in 2009; and 3.9 km/m3 (den- rates to meet the normality assumption. Annual mean ± SD sity index = 0.07 lb·ft−3·in−1) in 2010. Final rearing density in downstream passage rates for each group of fish were calcu- the Lyons Ferry Hatchery raceways that contained the direct- lated from the square-root-transformed individual downstream release fish was approximately 7.5 kg/m3 (density index = passage rates. To test the null hypothesis that there was no differ- 0.12 lb·ft−3·in−1) during all 3 years of the study. ence in the square-root-transformed mean annual downstream As PIT-tagged natural and hatchery subyearlings migrated passage rates between the two hatchery subyearling groups, we downstream, a portion of the fish was diverted to the bypasses used a one-way ANCOVA (two-sided; with length at tagging and detected at Lower Granite, Little Goose, Lower Monu- as the covariate). We did not statistically test for differences in mental, Ice Harbor, McNary, John Day, and Bonneville dams downstream passage rates (or detection timing) between either (Figure 1; Prentice et al. 1990b). Of the fish that were di- of the hatchery subyearling groups and the natural subyearlings verted and detected at Lower Granite, Little Goose, and Lower because such tests would not be informative if differences in the Monumental dams, some were transported and the remaining means of the response variables compared between both of the fish were routed back to the river to continue migration with hatchery subyearling groups and the natural subyearlings were the population of PIT-tagged fish that were not bypassed and consistently significant. To determine the hatchery subyearling MANAGEMENT BRIEF 523

group to which natural fish were most similar in terms of their released fish were classified as 74 mm (74.0–74.9 mm long). downstream passage rates, we expressed the difference in annual We then used a single-release release–recapture model a second means between each hatchery group and the natural subyear- time to estimate annual survival from release to the tailrace of lings as a percent difference: (annual mean downstream passage Lower Monumental Dam but with estimates made separately rate for the hatchery group – annual mean downstream pass- for the fish in the paired 74-mm bins of hatchery subyearlings. age rate of natural subyearlings)/(annual mean downstream pas- If the detected number of PIT-tagged fish in the 74-mm bin sage rate of natural subyearlings). As the percent differences in was sufficient to estimate survival for both hatchery subyearling annual mean passage rates increased, the potential for competi- groups, then that bin was retained. If there were insufficient data tion for food and space between the hatchery subyearling group in a bin, then the bin was combined with an adjacent bin. Bins and the natural subyearlings would decrease. were combined for both groups of hatchery subyearlings even We used the PIT tag detection data collected at Lower Monu- if data were insufficient for only one of the groups; thus, all mental Dam to calculate the annual mean ( ± SD) detection date bins remained matched between the two groups. We applied the as an index of passage timing at the dam. A one-way ANCOVA combining rule until we were left with a set of bins from which (as described for downstream passage rate) was applied to the survival could be estimated for both groups (N = 20 bins in square-root-transformed annual mean detection dates to test the 2008; 28 bins in 2009; and 18 bins in 2010). null hypothesis that there was no difference in mean annual We calculated a survival ratio for each bin by dividing the es- detection timing between the two hatchery subyearling groups. timated annual survival of acclimated hatchery subyearlings by To compare the overlap in passage timing between a hatchery the corresponding estimated annual survival of directly released subyearling group and natural subyearlings, we used the daily hatchery subyearlings in the same bin. To test the null hypothe- detection data to calculate annual percent overlap in detection sis that there was no difference in mean annual survival between timing (a measure of niche overlap; Krebs 1999). The poten- the two hatchery subyearling groups, we conducted a weighted tial for physical aggression, bodily contact, elevated stress, and t-test (two-sided) on natural-logarithm-transformed annual sur- pathogenic interaction within the holding raceways at dams and vival ratios calculated for each bin. The sample size for this test within transport vehicles would increase between hatchery and for calculating weighted mean ( ± SE) annual survival was and natural subyearlings as percent overlap increased. equal to the total number of bins. The weight for each bin was We used a single-release release–recapture model (Cormack based on the total number of acclimated and directly released 1964; Skalski et al. 1998) to estimate annual survival (%; ± SE) hatchery subyearlings in the bin. for each hatchery subyearling group from release to the tailrace of Lower Monumental Dam. To account for fish length in our RESULTS survival analysis, we classified fish from each hatchery subyear- Overall, 43,349 acclimated hatchery subyearlings, 44,819 ling group into bins according to their FLs at the time of tagging. directly released hatchery subyearlings, and 21,803 natural sub- We maximized the number of bins by first binning the fish in yearlings were PIT-tagged for release into the Snake River 1-mm increments. For example, in the 2008 release year, 798 (Table 1). Acclimated hatchery subyearlings were smaller at of 14,757 acclimated subyearlings and 1,714 of 15,807 direct- tagging than directly released hatchery subyearlings in 2008,

TABLE 1. Total number (N) of acclimated hatchery fall Chinook Salmon subyearlings that were released from the Captain John Rapids acclimation facility and number of hatchery subyearlings that were directly released into the Snake River at the Couse Creek boat launch; the number of ﬁsh in each subyearling group that were PIT-tagged (n); the date(s) of release; and the mean ( ± SD) FL (mm) at tagging, 2008–2010. The hatchery subyearlings were tagged 37–43 d prior to Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 release, and their FLs at release were not measured. Natural subyearlings were captured with a beach seine, tagged, and released on the same day (thus, FL at tagging was equal to FL at release).

Release dates Year Group NnEarliest Latest FL 2008 Acclimated 512,745 14,757 28 May 28 May 69 ± 6 Direct 230,401 15,807 28 May 28 May 72 ± 6 Natural 6,755 22 Apr 16 Jul 64 ± 11 2009 Acclimated 524,910 13,726 26 May 26 May 80 ± 6 Direct 200,744 13,665 26 May 26 May 76 ± 6 Natural 6,880 8 Apr 7 Jul 61 ± 9 2010 Acclimated 528,777 14,866 24 May 24 May 75 ± 5 Direct 203,162 15,347 24 May 24 May 71 ± 4 Natural 8,168 30 Mar 22 Jul 61 ± 10 524 ROSENBERGER ET AL.

whereas acclimated hatchery subyearlings were larger at tag- 2009: 90 ± 4% versus 77 ± 4%; 2010: 71 ± 2% versus 67 ± ging than directly released hatchery subyearlings in 2009 and 2%). Weighted mean annual survival to Lower Monumental 2010 (Table 1). In all years, hatchery subyearlings were larger Dam calculated for the FL bins differed significantly between at tagging than natural subyearlings (Table 1). acclimated and directly released hatchery subyearlings in 2008 Square-root-transformed annual mean downstream passage and 2009 but not in 2010 (Table 3). Mean annual percent rates differed significantly (2008–2010, all P-values < 0.0001) difference in weighted mean annual survival was greatest in between acclimated and directly released hatchery subyearlings. 2008, intermediate in 2009, and lowest in 2010 (Table 3). Acclimated hatchery subyearlings passed downstream from release to Lower Monumental Dam faster than directly released hatchery subyearlings during each of the study years (Table 2). DISCUSSION Annual percent differences in mean annual downstream passage Hatchery fall Chinook Salmon subyearlings that were accli- rates were also consistently greater between acclimated hatchery mated at the Captain John Rapids facility for 20–23 d passed subyearlings and natural subyearlings than between directly re- downstream faster, passed through the lower Snake River reser- leased hatchery subyearlings and natural subyearlings (Table 2). voirs earlier, and survived at higher rates postrelease than sub- Owing to the similarities in release date and the differences in yearlings from the same hatchery that were released directly downstream passage rate, square-root-transformed annual mean into the river. For an unknown reason, FL differed between detection dates differed significantly (2008–2010, all P-values acclimated and directly released hatchery subyearlings during < 0.0001) between acclimated and directly released hatchery 2008–2010, as would possibly be the case during any year in subyearlings. During each study year, acclimated hatchery sub- which managers schedule the release of both acclimated fish yearlings were detected at Lower Monumental Dam earlier than and directly released fish. Regardless of the cause, the length directly released hatchery subyearlings (Table 2). The annual differences at tagging might have persisted to the time of release detection distributions at Lower Monumental Dam were also into the river, thus potentially influencing our evaluation of ac- less protracted for acclimated hatchery subyearlings than for climation. Length of subyearling Chinook Salmon is directly directly released hatchery subyearlings (Figure 2). The annual proportional to downstream passage rate, inversely propor- detection distributions for natural subyearlings were the most tional to detection timing, and directly proportional to survival protracted (Figure 2). Consequently, annual percent overlap was (Connor et al. 2000, 2003, 2004). We statistically controlled for consistently lower between acclimated hatchery subyearlings fish size in all three sets of annual analyses to account for the and natural subyearlings than between directly released hatch- length differences. With the exception of weighted mean annual ery subyearlings and natural subyearlings (Table 2). survival in 2010, we found significant and consistent differ- Estimated survival from release to the tailrace of Lower ences in all comparisons, even for 2008, when acclimated fish Monumental Dam was consistently higher for acclimated were smaller than directly released fish. These findings demon- hatchery subyearlings than for directly released hatchery strate that size alone cannot fully explain the enhanced postre- subyearlings (2008: mean ± SE = 65 ± 3% versus 50 ± 3%; lease performance of acclimated hatchery subyearlings. The two

TABLE 2. Number of PIT tag detections (n) at Lower Monumental Dam, mean ( ± SD) annual downstream passage rate (km/d) calculated from release in the lower Snake River to detection at the dam, and mean ( ± SD) annual detection date at the dam, 2008–2010, for the three groups of PIT-tagged fall Chinook Salmon: (1) hatchery subyearlings that were acclimated at the Captain John Rapids acclimation facility prior to release, (2) hatchery subyearlings that were released directly into the Snake River at the Couse Creek boat launch, and (3) natural subyearlings. Percent differences in mean annual passage rate and percent overlap in detection timing between each hatchery subyearling group and the natural subyearlings are also given. All statistical tests for differences between the hatchery subyearling Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 groups were signiﬁcant (P ≤ 0.0001) after accounting for differences in FL.

Passage rate Detection date Percent Percent Year Group n Mean difference Mean overlap 2008 Acclimated 1,388 13.3 ± 5.5 67 14 Jun ± 8d 22 Direct 682 8.0 ± 3.7 0 25 Jun ± 14 d 45 Natural 144 8.0 ± 7.2 14 Jul ± 36 d 2009 Acclimated 1,533 20.6 ± 6.9 249 6 Jun ± 4d 18 Direct 973 11.7 ± 4.7 97 13 Jun ± 7d 33 Natural 233 5.9 ± 4.3 3 Jul ± 22 d 2010 Acclimated 2,236 12.2 ± 2.3 96 10 Jun ± 4d 28 Direct 2,513 9.8 ± 1.9 58 13 Jun ± 5d 45 Natural 420 6.2 ± 4.1 4 Jul ± 32 d MANAGEMENT BRIEF 525

TABLE 3. Number of matched length-bins (N bins) that were used to calculate the weighted mean ( ± SE) annual survival (%) and mean percent difference in weighted mean annual survival (2008–2010) for PIT-tagged hatchery fall Chinook Salmon subyearlings that either were acclimated at the Captain John Rapids acclimation facility prior to release or were released directly into the Snake River at the Couse Creek boat launch. The P-values are from weighted t-tests (two-sided) conducted to determine whether weighted mean annual survival differed signiﬁcantly (α = 0.05) between acclimated and directly released hatchery subyearlings.

Acclimated ﬁsh Directly released ﬁsh Year N bins Survival N bins Survival Percent difference P 2008 20 65 ± 32050± 331± 10 0.003 2009 28 90 ± 42877± 418± 8 0.017 2010 18 71 ± 21867± 25± 4 0.201

hatchery groups were reared and handled under similar conditions up to the point at which the acclimated fish were moved off-site, further supporting acclimation as a factor in the enhanced postrelease performance. There are three plausible explanations for enhanced performance by acclimated subyearlings. The first is that fish grew faster in the pond at the acclimation facility than in the raceways at the hatchery. Although we were unable to measure prerelease growth and although preliminary work conducted at Lyons Ferry Hatchery and Captain John Rapids in 2005 and 2006 did not show a large or consistent growth benefit of acclimation (Rosen- berger 2008), spring growth has been correlated to the degree of smoltification (Dickhoff et al. 1995, 1997; Beckman et al. 1999, 2000). If acclimated hatchery subyearlings grew faster prior to release, they might have been more willing—as well as better able—to migrate downstream than directly released hatchery subyearlings. Secondly, the acclimation pond was ar- guably a more natural rearing environment than the hatchery raceways. It is possible that the relatively natural features of the acclimation pond prepared the acclimated fish for exposure to natural food items and habitat, thereby influencing their postrelease performance (Suboski and Templeton 1989). A third, more commonly referenced explanation relates to stress. Fish experience chronic stress during hatchery rearing (Whitesel et al. 1994). Subsequent handling and confinement during transport have been shown to subject juvenile salmonids to acute stress (Chinook Salmon: Strange et al. 1978; Rainbow Trout: Barton Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 et al. 1980; Coho Salmon: Schreck et al. 1989; Atlantic Salmon Salmo salar: Iversen et al. 1998). Acute stress can impair respi- ration during swimming (Barton and Schreck 1987) and hence can affect swimming speed and stamina. Acute stress can also reduce predator avoidance capabilities (Olla et al. 1992) while depressing immune functions and increasing susceptibility to disease (Maule et al. 1989). The acclimation pond at the Captain John Rapids facility was a low-density, low-velocity, food-rich, predator-free environment that likely enhanced postrelease performance by giving the hatchery subyearlings time to recover FIGURE 2. Cumulative PIT tag detection distributions at Lower Monumental from cumulative stressors before they were subjected to the Dam, 2008–2010, for three groups of fall Chinook Salmon: (1) hatchery subyearlings that were released from the Captain John Rapids acclimation facility rigors of the river. (Acclimated); (2) hatchery subyearlings that were directly released at the Couse Regardless of the explanation for the enhanced performance Creek boat launch (Direct); and (3) natural subyearlings that were collected by of acclimated subyearlings, acclimation reduced the potential seining, PIT-tagged, and released at the point of capture along the lower Snake for interactions between hatchery and natural fish by reducing River (Natural). the time spent by hatchery fish in Lower Granite Reservoir. 526 ROSENBERGER ET AL.

This finding was important because for natural subyearlings, and Van Gaest et al. (2011) were the first to document the the period spent lingering and feeding in the reservoir prior to adverse physiological, immunological, and pathogenic re- becoming smolts (see Connor et al. 2003) is critical to maintain- sponses of yearling Chinook Salmon to confinement during ing the high growth rates associated with smoltification (e.g., transport; it would be useful to conduct similar studies of sub- Beckman and Dickhoff 1998), which were observed prior to yearling Chinook Salmon. supplementation (Connor and Burge 2003). At the simplest and We conclude that acclimation enhanced the postrelease per- least debatable level, reducing the time spent by hatchery sub- formance of hatchery fall Chinook Salmon subyearlings and yearlings in the reservoir would reduce the amount of food reduced the potential for interaction of these hatchery fish with consumed by the hatchery fish, thereby reducing the potential natural subyearlings, which are listed for protection under the for a density-dependent growth effect on natural subyearlings. U.S. Endangered Species Act (NMFS 1992). The ability to en- Faster downstream passage of hatchery fish might also reduce hance production of fish populations is the global measure of the potential for aggressive and nonaggressive social interac- the success of a given fish culture practice. In the case of actions with natural subyearlings during their reservoir rearing climation, adult return rate analyses on groups of acclimated period. Little is known about the frequency and consequences and directly released hatchery juveniles have yielded results of such interactions in large reservoirs, but displacement can that vary geographically as well as among and within species be substantial when hatchery and natural juveniles interact in (Johnson et al. 1990; Whitesel et al. 1994; Kenaston et al. 2001; small streams (e.g., a decrease in juvenile Coho Salmon density Appleby et al. 2002; Clarke et al. 2010; Dittman et al. 2010). of 41%: Nickelson et al. 1986; displacement of natural steel- This study represents the first component of a long-term effort head juveniles in 79% of encounters: McMichael et al. 1999). to assess the effect of acclimation on hatchery fall Chinook Although the relation between displacement level and stream Salmon adult returns to the lower Snake River. The adult re- order has not been established, the conservative view is that any turns for the 2008–2010 hatchery releases will be complete in level of displacement dependent on hatchery abundance could 2014, and the coded wire tag recovery data will be available by alter the migration timing and growth phenotypes of natural about 2016. Analyses of the coded wire tag data will supple- subyearlings. Understanding the influence of expanding hatch- ment the existing understanding of the efficacy of acclimation ery programs on phenotypic changes in natural fishes within while testing our premise that the juvenile survival advantage to large water bodies, including Lower Granite Reservoir, is an Lower Monumental Dam will persist through adult return. The important area of future research. present results may provide some guidance to others attempting Acclimation also reduced overlap in passage timing at Lower to reduce conspecific interactions when supplementing natural Monumental Dam as well as at Lower Granite and Little Goose populations with hatchery fish. dams, where juveniles were also collected and transported. This reduction in overlap undoubtedly reduced the interaction between natural and hatchery subyearlings, as the two groups ACKNOWLEDGMENTS cannot be separated prior to being routed to the holding race- We thank our colleagues from BioMark, the Nez Perce ways at the dams. Delayed juvenile mortality associated with Tribe Department of Fisheries Resources, National Oceanic and the stress induced by the collection and transportation process Atmospheric Administration Fisheries, U.S. Army Corps of En- has been hypothesized as a factor contributing to decreases in gineers, U.S. Fish and Wildlife Service, and Washington Depart- productivity of anadromous salmonids in the Columbia River ment of Fish and Wildlife, whose efforts contributed to the com- basin (e.g., Budy et al. 2002; but see Muir et al. 2006). The pletion of this study. This study (and many other studies we have

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 potential for elevated stress, physical aggression, bodily con- conducted) would not have been possible without assistance tact, and pathogenic interaction between natural and hatchery from Pacific States Marine Fisheries Commission personnel, subyearlings in the dam raceways and in transport barges and including D. Marvin and N. Tancreto, who helped operate and trucks varies seasonally. The level of interaction that is solely maintain the Columbia River Basin PIT Tag Information Sys- attributable to hatchery subyearlings is low during the early tem. Anonymous peer review greatly improved this manuscript. portion of the migration season, as passage abundance of other This study was funded in part by the Bonneville Power Admin- numerically dominant species and age-classes of anadromous istration (Project 199102900) administered by D. Docherty and juveniles is high at the dams during that period (FPC 2012), J. George. The majority of the tagging costs were covered by the and raceway densities and barge loadings are regulated accord- U.S. Army Corps of Engineers under a contract administered ing to passage abundance. However, by mid-June to early July, by S. Dunmire, D. Holecek, and J. George. The use of trade the fish passing the dams are primarily fall Chinook Salmon names does not imply endorsement by the U.S. Government or subyearlings (FPC 2012); hence, any increase in hatchery fish other collaborating agencies. The findings and conclusions in abundance will be directly reflected in the raceways and the this article are those of the authors and do not necessarily rep- tanker trucks, which are used in lieu of barges to transport fish as resent the views of the U.S. Fish and Wildlife Service or other passage abundance declines. Studies by Congleton et al. (2000) collaborating agencies. MANAGEMENT BRIEF 527

REFERENCES Dickhoff, W.W., B. R. Beckman, D. A. Larsen, C. V.W.Mahnken, C. B. Schreck, Appleby, A. E., J. M. Tipping, and G. E. VanderHaegen. 2002. Effects of surface C. Sharpe, and W. S. Zaugg. 1995. Quality assessment of hatchery-reared water acclimatization on postrelease survival of yearling spring Chinook spring Chinook Salmon smolts in the Columbia River basin. Pages 292–302 Salmon. North American Journal of Aquaculture 64:301–304. in H. L. Schramm Jr. and R. G. Piper, editors. Uses and effects of cultured Barton, B. A., R. E. Peter, and C. R. Paulencu. 1980. Plasma cortisol levels fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, of fingerling Rainbow Trout (Salmo gairdneri) at rest, and subjected to han- Bethesda, Maryland. dling, confinement, transport, and stocking. Canadian Journal of Fisheries Dittman, A. H., D. May, D. A. Larsen, M. L. Moser, M. Johnston, and D. and Aquatic Sciences 37:805–811. Fast. 2010. Homing and spawning site selection by supplemented hatchery- Barton, B. A., and C. B. Schreck. 1987. Metabolic cost of acute physical stress in and natural-origin Yakima River spring Chinook Salmon. Transactions of the juvenile steelhead. Transactions of the American Fisheries Society 116:257– American Fisheries Society 139:1014–1028. 263. Elliott, D. G., R. J. Pascho, and G. L. Bullock. 1989. Developments in the Beckman, B. R., and W. W. Dickhoff. 1998. Plasticity of smolting in spring control of bacterial kidney disease of salmonid fishes. Diseases of Aquatic Chinook Salmon: relation to growth and insulin-like growth factor-I. Journal Organisms 6:201–215. of Fish Biology 53:808–826. FPC (Fish Passage Center). 2012. Fish passage center data. FPC, Portland, Beckman, B. R., W. W. Dickhoff, W. S. Zaugg, C. Sharpe, S. Hirtzel, R. Schrock, Oregon. Available: www.fpc.org. (June 2012). D. A. Larsen, R. D. Ewing, A. Palmisano, C. B. Schreck, and C. V. W. Giorgi, A. E., G. A. Swan, W. S. Zaugg, T. Coley, and T. Y. Barila. 1988. Mahnken. 1999. Growth, smoltification, and smolt-to-adult return of spring Susceptibility of Chinook Salmon smolts to bypass systems at hydroelectric Chinook Salmon from hatcheries on the Deschutes River, Oregon. Transac- dams. North American Journal of Fisheries Management 8:25–29. tions of the American Fisheries Society 128:1125–1150. Hansen, L. P., and B. Jonsson. 1985. Downstream migration of hatchery-reared Beckman, B. R., D. A. Larsen, C. Sharpe, B. Lee-Pawlak, C. B. Schreck, smolts of Atlantic Salmon (Salmo salar L.) in the River Imsa, Norway. and W. W. Dickhoff. 2000. Physiological status of naturally reared juvenile Aquaculture 45:237–248. spring Chinook Salmon in the Yakima River: seasonal dynamics and changes Hillman, T. W., and J. W. Mullan. 1989. Effect of hatchery releases on the associated with smolting. Transactions of the American Fisheries Society abundance and behavior of wild juvenile salmonids. Pages 266–285 in Don 129:727–753. Chapman Consultants, editors. Summer and winter ecology of juvenile Chi- Budy, P., G. P. Thiede, N. Bouwes, C. E. Petrosky, and H. Schaller. 2002. nook Salmon and steelhead trout in the Wenatchee River. Report to Chelan Evidence linking delayed mortality of Snake River salmon to their earlier County Public Utility District, Wenatchee, Washington, Don Chapman Con- hydrosystem experience. North American Journal of Fisheries Management sultants, Boise, Idaho. 22:35–51. Iversen, M., B. Finstad, and K. J. Nilssen. 1998. Recovery from loading and Bugert, R. M. 1998. Mechanics of supplementation in the Columbia River. transport stress in Atlantic Salmon (Salmo salar L.) smolts. Aquaculture Fisheries 23(1):11–20. 168:387–394. Bugert, R. M., C. W. Hopley, C. A. Busack, and G. W. Mendel. 1995. Mainte- Jefferts, K. B., P.K. Bergman, and H. F. Fiscus. 1963. A coded wire identification nance of stock integrity in Snake River fall Chinook Salmon. Pages 267–276 system for macro-organisms. Nature (London) 198:460–462. in H. L. Schramm Jr. and R. G. Piper, editors. Uses and effects of cultured Johnson, S. L., M. F. Solazzi, and T. E. Nickelson. 1990. Effects on survival and fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, homing of trucking hatchery yearling Coho Salmon to release sites. North Bethesda, Maryland. American Journal of Fisheries Management 10:427–433. Chapman, D. W. 1962. Aggressive behavior in juvenile Coho Salmon as a cause Jonsson, B., and N. Jonsson. 2006. Cultured Atlantic Salmon in nature: a review of emigration. Journal of the Fisheries Research Board of Canada 19:1047– of their ecology and interaction with wild fish. ICES Journal of Marine 1080. Science 63:1162–1181. Clarke, L. R., M. W. Flesher, T. A. Whitesel, G. R. Vonderohe, and R. W. Kenaston, K. R., R. B. Lindsay, and R. K. Schroeder. 2001. Effect of acclimation Carmichael. 2010. Postrelease performance of acclimated and directly re- on the homing and survival of hatchery winter steelhead. North American leased hatchery summer steelhead into Oregon tributaries of the Snake River. Journal of Fisheries Management 21:765–773. North American Journal of Fisheries Management 30:1098–1109. Kostow, K. 2009. Factors that contribute to the ecological risks of salmon and Congleton, J. L., W. J. LaVoie, C. B. Schreck, and L. E. Davis. 2000. Stress steelhead hatchery programs and some mitigating strategies. Reviews in Fish indices in migrating juvenile Chinook Salmon and steelhead of wild and Biology and Fisheries 19:9–31. hatchery origin before and after barge transportation. Transactions of the Krebs, C. J. 1999. Ecological methodology, 2nd edition. Addison-Wesley, American Fisheries Society 129:946–961. Menlo Park, California. Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 Connor, W. P., and H. L. Burge. 2003. Growth of wild subyearling fall Chinook Maule, A. G., R. A. Tripp, S. L. Kaattari, and C. B. Schreck. 1989. Stress alters Salmon in the Snake River. North American Journal of Fisheries Management immune function and disease resistance in Chinook Salmon (Oncorhynchus 23:594–599. tshawytscha). Journal of Endocrinology 120:135–142. Connor, W. P., S. G. Smith, T. Andersen, S. M. Bradbury, D. C. Burum, McCutcheon, S., and R. Richmond. 2010. PIT-tagging year production, subyear- E. E. Hockersmith, M. L. Schuck, G. W. Mendel, and R. M. Bugert. 2004. ling production and surrogate groups of Snake River fall Chinook Salmon Postrelease performance of hatchery yearling and subyearling fall Chinook to evaluate the responses of Snake and Columbia river basin fall Chinook Salmon released into the Snake River. North American Journal of Fisheries Salmon to dam passage strategies and experiences. Annual Job Completion Management 24:545–560. Report to the U.S. Army Corps of Engineers, Walla Walla, Washington. Connor, W. P., R. K. Steinhorst, and H. L. Burge. 2000. Forecasting survival McMichael, G. A., T. N. Pearsons, and S. A. Leider. 1999. Behavioral in- and passage of migratory juvenile salmonids. North American Journal of teractions among hatchery-reared steelhead smolts and wild Oncorhynchus Fisheries Management 20:651–660. mykiss in natural streams. North American Journal of Fisheries Management Connor, W. P., R. K. Steinhorst, and H. L. Burge. 2003. Migrational behavior 19:948–956. and seaward movement of wild subyearling fall Chinook Salmon in the Snake McMichael, G. A., C. S. Sharpe, and T. N. Pearsons. 1997. Effects of residual River. North American Journal of Fisheries Management 23:414–430. hatchery-reared steelhead on growth of wild Rainbow Trout and spring Chi- Cormack, R. M. 1964. Estimates of survival from the sighting of marked animals. nook Salmon. Transactions of the American Fisheries Society 126:230–239. Biometrika 51:429–438. Muir, W. D., D. M. Marsh, B. P. Sandford, S. G. Smith, and J. G. Williams. Dickhoff, W. W., B. R. Beckman, D. A. Larsen, C. Duan, and S. Moriyama. 2006. Post-hydropower system delayed mortality of transported Snake River 1997. The role of growth in endocrine regulation of salmon smoltification. stream-type Chinook Salmon: unraveling the mystery. Transactions of the Fish Physiology and Biochemistry 17:231–236. American Fisheries Society 135:1523–1534. 528 ROSENBERGER ET AL.

Nickelson, T. E., M. F. Solazzi, and S. L. Johnson. 1986. Use of hatchery Coho migrant juvenile salmonids in the Snake and Columbia rivers. Canadian Jour- Salmon (Oncorhynchus kisutch) presmolts to rebuild wild populations in nal of Fisheries and Aquatic Sciences 55:1484–1493. Oregon coastal streams. Canadian Journal of Fisheries and Aquatic Sciences Smith, S. G., W. D. Muir, E. E. Hockersmith, R. W. Zabel, R. J. Graves, C. V. 43:2443–2449. Ross, W. P. Connor, and B. D. Arnsberg. 2003. Influence of river conditions NMFS (National Marine Fisheries Service). 1992. Threatened status for Snake on survival and travel time of Snake River subyearling fall Chinook Salmon. River spring/summer Chinook Salmon, threatened status for Snake River fall North American Journal of Fisheries Management 23:939–961. Chinook Salmon. Federal Register 57:78(22 April 1992):14653–14663. Strange, R. J., C. B. Schreck, and R. D. Ewing. 1978. Cortisol concentrations Olla, B. L., M. W. Davis, and C. B. Schreck. 1992. Comparison of preda- in confined juvenile Chinook Salmon (Oncorhynchus tshawytscha). Transac- tor avoidance capabilities with corticosteroid levels induced by stress in ju- tions of the American Fisheries Society 107:812–819. venile Coho Salmon. Transactions of the American Fisheries Society 121: Suboski, M. D., and J. J. Templeton. 1989. Life skills training for hatchery fish: 544–547. social learning and survival. Fisheries Research 7:343–352. Peery, C. A., and T. C. Bjornn. 2004. Interactions between natural and hatchery Tatara, C. P., S. C. Riley, and B. A. Berejikian. 2011. Effects of hatchery Chinook Salmon parr in a laboratory stream channel. Fisheries Research fish density on emigration, growth, survival, and predation risk of natural 66:311–324. steelhead parr in an experimental stream channel. North American Journal of Piper, R. G., I. B. McElwain, L. E. Orme, J. P. McCraren, L. G. Fowler, and J. R. Fisheries Management 31:224–235. Leonard. 1986. Fish hatchery management. U.S. Fish and Wildlife Service, Tiffan, K. F., and W. P. Connor. 2011. Distinguishing between natural and Washington, D.C. hatchery Snake River fall Chinook Salmon subyearlings in the field us- Prentice, E. F., T. A. Flagg, and C. S. McCutcheon. 1990a. Feasibility of using ing body morphology. Transactions of the American Fisheries Society 140: implantable passive integrated transponder (PIT) tags in salmonids. Pages 21–30. 317–322 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. Jester Jr., E. Tiffan, K. F., T. J. Kock, C. A. Haskell, W. P. Connor, and R. K. Steinhorst. 2009. D. Prince, and G. A. Winans, editors. Fish-marking techniques. American Water velocity, turbulence, and migration rate of subyearling fall Chinook Fisheries Society, Symposium 7, Bethesda, Maryland. Salmon in the free-flowing and impounded Snake River. Transactions of the Prentice, E. F., T. A. Flagg, C. S. McCutcheon, and D. F. Brastow. 1990b. American Fisheries Society 138:373–384. PIT-tag monitoring systems for hydroelectric dams and fish hatcheries. Pages Van Gaest, A. L., J. P. Dietrich, D. E. Thompson, D. A. Boylen, S. A. Strickland, 323–334 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. Jester Jr., E. T. K. Collier, F. J. Loge, and M. R. Arkoosh. 2011. Survey of pathogens in D. Prince, and G. A. Winans, editors. Fish-marking techniques. American hatchery Chinook Salmon with different out-migration histories through the Fisheries Society, Symposium 7, Bethesda, Maryland. Snake and Columbia rivers. Journal of Aquatic Animal Health 23:62–77. PTAGIS (PIT Tag Information System). 2012. The source for information about Ward, D. L., R. R. Boyce, F. R. Young, and F. E. Olney. 1997. A review and PIT-tagged fish in the Columbia River basin. Pacific States Marine Fisheries assessment of transportation studies for juvenile Chinook Salmon in the Snake Commission, Portland, Oregon. Available: www.ptagis.org. (March 2012). River. North American Journal of Fisheries Management 17:652–662. Rosenberger, S. J. 2008. Postrelease performance of acclimated and di- Weber, E. D., and K. D. Fausch. 2003. Interactions between hatchery and wild rectly released subyearling hatchery fall Chinook Salmon (Oncorhynchus salmonids in streams: differences in biology and evidence for competition. tshawytscha). Master’s thesis. University of Idaho, Moscow. Canadian Journal of Fisheries and Aquatic Sciences 60:1018–1036. Schreck, C. B., M. F. Solazzi, S. L. Johnson, and T. E. Nickelson. 1989. Trans- Whitesel, T. A., P. T. Lofy, R. W. Carmichael, R. T. Messmer, M. W. Flesher, portation stress affects performance of Coho Salmon, Oncorhynchus kisutch. and D. W. Rondorf. 1994. A comparison of the performance of acclimated Aquaculture 82:15–20. and direct stream released, hatchery-reared steelhead smolts in northeast Skalski, J. R., S. G. Smith, R. N. Iwamoto, J. G. Williams, and A. Hoff- Oregon. Pages 87–92 in D. D. MacKinlay, editor. High performance fish. mann. 1998. Use of passive integrated transponder tags to estimate survival of Fish Physiology Association, Vancouver. Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:05 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Disinfection of Three Wading Boot Surfaces Infested with New Zealand Mudsnails Kelly A. Stockton a c & Christine M. Moffitt b a Idaho Cooperative Fish and Wildlife Research Unit , University of Idaho, Perimeter Drive MS 1141 , Moscow , Idaho , 83844-1141 , USA b U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit , University of Idaho, Perimeter Drive MS 1141 , Moscow , Idaho , 83844-1141 , USA c Marrone Bio Innovations , Post Office Box 50743, Henderson , Nevada , 89016 , USA Published online: 29 Apr 2013.

To cite this article: Kelly A. Stockton & Christine M. Moffitt (2013): Disinfection of Three Wading Boot Surfaces Infested with New Zealand Mudsnails, North American Journal of Fisheries Management, 33:3, 529-538 To link to this article: http://dx.doi.org/10.1080/02755947.2013.768569

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:529–538, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.768569

ARTICLE

Disinfection of Three Wading Boot Surfaces Infested with New Zealand Mudsnails

Kelly A. Stockton1 Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, Perimeter Drive MS 1141, Moscow, Idaho 83844-1141, USA Christine M. Mofﬁtt* U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, Perimeter Drive MS 1141, Moscow, Idaho 83844-1141, USA

Abstract New Zealand mudsnails Potamopyrgus antipodarum (NZMS) have been introduced into many continents and are easily transported live while attached to wading and other field gear. We quantified the relative attachment by different life stages of NZMS to felt, neoprene, and rubber-soled boots exposed to two densities of NZMS in experimental exposure totes. Attachment by NZMS occurred on boots of all surfaces, but the highest numbers of all life stages occurred on boots with felt surfaces. We found a 15–20-min bath application of 20 g/L Virkon Aquatic was a reliable tool to disinfect boot surfaces infested with NZMS and other aquatic invertebrates. Our studies support that spray application of this disinfectant was not reliable to provide complete mortality of attached adult NZMS or neonates. Wading gear surfaces exposed to repeated bath disinfections showed little deterioration. Our results provide strong evidence that bath disinfections with Virkon Aquatic are helpful to assure biosecurity in field and hatchery settings, but applications should be coupled with cleaning procedures to remove organic materials that can deactivate the reagent.

Many nonindigenous invertebrates and microorganisms have the community (Brenneis et al. 2011; Mofﬁtt and James 2012), been introduced throughout the world via ships and boats, wad- others show population density effects on native snails (Lysne ing gear, and human activities (Mills et al. 1993; Ricciardi and and Koetsier 2008), and others result in major trophic shifts

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 Rasmussen 1998; Johnson et al. 2001; Briski et al. 2012; Root (Hall et al. 2006; Riley et al. 2008). and O’Reilly 2012). New Zealand mudsnails Potamopyrgus an- Chemical reagents are useful tools to reduce the spread of tipodarum (NZMS) have been introduced and become estab- NZMS from infested areas on wading gear, but chemical treat- lished in many continents, and these transplants are speculated to ments must be effective, safe, and practical for the proposed be associated with human activities such as recreational fishing, applications (Dwyer et al. 2003; Hosea and Finlayson 2005). boating, and aquaculture operations (Bowler 1991; Loo et al. In the USA several chemical disinfectants and biocides have 2007; Bruce and Moffitt 2010; Alonso and Castro-D´ıez 2012). been evaluated for their effectiveness in killing NZMS on con- Studies of the effects of NZMS on the aquatic environment after taminated surfaces and gear in field and hatchery operations. their introductions show a range of outcomes, probably due to Hosea and Finlayson (2005) examined the effectiveness of a the specifics of the aquatic community and the wide habitat tol- large suite of compounds on snail mortality: grapefruit seed erance of the snails (reviewed in Alonso and Castro-D´ıez 2008, extract (GSE), benzethonium chloride (BZCl, Alpha Aesar), 2012). Some introductions appear to have few consequences on Clorox, copper sulfate pentahydrate, Formula 409 degreaser and

*Corresponding author: cmofﬁ[email protected] 1Present address: Marrone Bio Innovations, Post Ofﬁce Box 50743, Henderson, Nevada 89016, USA. Received November 11, 2012; accepted January 17, 2013 529 530 STOCKTON AND MOFFITT

disinfectant, potassium permanganate, isopropyl alcohol, Pine- studies, and Johnson et al. (2003) reported its efficacy against Sol, and household ammonia. All substances tested by Hosea the amphibian chytrid fungus Batrachochytrium dendrobatidis. and Finlayson (2005), except for copper sulfate and Formula Vertebrates may be more tolerant to exposure, as Schmidt et al. 409, caused damage to wading gear. Schisler et al. (2008) ex- (2009) found tadpoles of the European common frog Rana plored the efficacy of formulations of quaternary ammonium temporaria and the common toad Bufo bufo survive limited compounds (QACs), including common Formula 409 and Spar- exposures to 4 g/L Virkon even after a 1-week exposure. quat 256, but did not test their effects on gear. Oplinger and International, national, and state and provincial agencies have Wagner (2009a, 2010) tested several compounds and explored increased their attention to developing guidance documents for their use in the context of fish safety and regulatory concerns. international trade of livestock and commodities and their po- Any approved use must consider both human and environmental tential effects on human health (Firestone and Corbett 2005; safety. Copper sulfate has potentially harmful effects on aquatic Perrings et al. 2010), but there are few, if any, harmonized pro- species (McGeer et al. 2000; Shaw et al. 2012), and residues of tocols for disinfecting contaminated gear in field and hatchery QACs, which have wide use in disinfectants and other industrial settings (Root and O’Reilly 2012). When measures are in place, applications, have come under increasing scrutiny by scientists instructions generally include cleaning and drying of equipment and regulators (Boethling 1984; Garc´ıa et al. 2001; EPA 2010; and wading gear, but this step may not be completely reliable. Sarkar et al. 2010). Researchers stress the need for additional As a precaution, several states, provinces, and countries have testing of disinfecting protocols and chemicals to consider a va- banned or are considering prohibitions on the use of felt-covered riety of environmental conditions and target species to improve boots for field and hatchery use because of the risks of transport tools that can be used safely to reduce the risks of transport of of invasive species and pathogens (CANS 2012). viable invasive invertebrates, microorganisms, and pathogens The objectives of our study were to test felt, neoprene, or in field and hatchery settings (Schisler et al. 2008; Britton and rubber-surfaced wading boots to determine (1) the likelihood of Dingman 2011). infestation by NZMS, (2) the effectiveness of disinfecting with Highly secure facilities such as hatcheries routinely use dis- spray or bath applications of Virkon Aquatic, (3) the effect of infectants on equipment and wading gear to limit the spread of the proposed disinfection procedures on the durability of wading diseases caused by bacteria and viral agents. The disinfectant, gear, and (4) the effectiveness of solutions of Virkon Aquatic Virkon Aquatic (reformulated from Virkon S in 2007; Mainous when contaminated with organics. et al. 2010) is registered by the U.S. Environmental Protection Agency (USEPA) and labeled specifically for use in aquacul- METHODS ture facilities as a surface disinfectant against bacterial, fungal, and viral pathogens (DAHS 2011a). Virkon Aquatic is com- Experimental Organisms posed of a triple salt of potassium monopersulphate acting as New Zealand mudsnails were collected from springs at an oxidizing agent, sulfamic acid and malic acid, sodium hex- Hagerman National Fish Hatchery (HNFH), Idaho, and shipped amethaphosphate buffer, and sodium alkyl benzene sulphonate in coolers to the University of Idaho fisheries wet laboratory. as a surfactant (Western Chemical 2012), and its breakdown Upon arrival, the snails were washed through a 2.0-mm and products are nontoxic salts. The chemical oxidizes proteins and 0.85-mm sieve to separate snails from sediments. The NZMS other components of cell protoplasm, resulting in inhibition of samples, which included other aquatic invertebrates from the enzyme systems and loss of cell-wall integrity (Curry et al. springs (additional snails and insects), were transferred into 2-L 2005). In unpublished studies in our laboratory, we found that containers with dechlorinated, aged, well water equilibrated to ◦ Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 concentrations of Virkon Aquatic at 20 g/L were effective in 15 C. A portion of the water in each container was changed killing all life stages of NZMS when target snails were placed every other day. The NZMS were retained in the laboratory for in glass beakers and immersed for 20 min (see Stockton 2011). no more than 3 weeks. The culture containers also included The efficacy of Virkon S and Virkon Aquatic as disinfectants some algae and vascular plants such as pondweed Potamogeton has been evaluated in several studies reported by Mainous et al. spp. as food sources for the invertebrate cultures in the lab- (2010). As a dry powder it is easily stored and transported, oratory. Temperature in the room was maintained at a target and Virkon S was used in the Antarctic for disinfecting boots level of 15◦C throughout trials and recorded at 15-min inter- of visitors to reduce the risk of translocation of microbial vals with HOBO data loggers (Onset Computer Corporation, pathogens to the sensitive ecosystem (Curry et al. 2005). Other Bourne, Massachusetts). A natural photoperiod for the latitude studies demonstrated a reduction in numbers and species of of HNFH was maintained in the test room. zooplankton (Schmidt et al. 2009) and killing of mollusks, including Sydney rock oyster Saccostrea glomerata veligers Test Substance (Dove and O’Connor 2007), red-rim melania Melanoides Virkon Aquatic (lot 2258523 or 2258515; Western Chemical, tuberculata (Mitchell et al. 2007), and faucet snails Bithynia Ferndale, Washington) was used for testing. We weighed Virkon tentaculata (Mitchell and Cole 2008). Root and O’Reilly Aquatic (0.01 g) to make 10- or 20-g/L solutions (1% and 2%) (2012) evaluated the efficacy of 10-g/L Virkon solutions to kill in deionized water. Test reagents were then placed into acid- the freshwater diatom Didymosphenia geminata in laboratory washed totes or flasks until tested. At least 1 h was allowed for DISINFECTING WADING BOOTS 531

all solutions to acclimate to test temperature and to activate fully. NZMS before and after the test duration with a YSI 556 MPS The concentrations were verified with Virkon Aquatic test strips multiprobe (YSI, Yellow Springs, Ohio). (Western Chemical, Ferndale, Washington) to ensure chemical Disinfection of three wader surfaces infested with NZMS.— was mixed and active. To determine the efficacy of disinfections and to evaluate whether the multiple bath disinfections of infested surfaces Experimental Design affected wader durability, we tested felt, neoprene, and rubber Colonization of three wading-boot surfaces.—A controlled wader boot or wading bootie surfaces in laboratory simulations. 3 × 2 factorial designed study was conducted to determine the We obtained 18 waders of each type from regional field or colonization likelihood of NZMS on three different wading- hatchery operators (54 total). All waders tested were in good boot surfaces: neoprene, felt, and rubber. We evaluated the rate condition at the start of testing, but the equipment had been of infestation onto the wading surfaces in exposures to two used (to simulate a normal field or hatchery setting). Waders densities, 50 or 100 g wet weight per tote (55 × 37 × 43 cm were cut at the knee to leave only the boot or bootie. They deep; volume, 114 L), of mixed culture of NZMS with small were numbered with unique codes prior to the experiments and numbers of other aquatic invertebrates. washed to ensure that there was no contamination of the NZMS To test the rate of colonization, we placed one foot from container. Before and at the end of each experiment, all boots each of three different wading-boot surfaces (three boots per were inspected and any damage, such as cracks, holes, stains, tote) into a tote with live NZMS. The arrangement of test sur- tears, and discoloration, was recorded. At the end of the testing faces was rotated (from left to right) for each replicate test to the boots were inspected for change in condition and the damage account for placement variation in our analysis. The boots were was ranked as (1) none if there were no cracks, discoloration, left in the colonization tote for 30 min and were then removed loss of flexibility, tearing, or stitching failure, and the gear carefully and placed into a cleaning bath tote where all organ- could still repel water, (2) mild if the wading gear had slight isms attached to the boots were carefully removed, recovered, discoloration, small cracks, or tearing, and the gear still repelled and enumerated. The numbers of adult, juvenile, and neonate water, or (3) severe if the wading gear showed signs of high NZMS and other aquatic invertebrates were recorded, and the discoloration, flexibility loss, large cracks and tears, or stitching organisms were identified and then destroyed. The colonization failure, and the boots did not repel water or leaked significantly. testing was replicated three times at both densities of NZMS. To allow for a volitional laboratory infestation of the wader Efficacy of spray versus bath application.—To compare the boots, we placed adult NZMS (>3 mm) into a 114-L tote mortality of NZMS after a bath or spray exposure we con- (55 × 37 × 43 cm) 2 d before testing. Before each trial, we ducted controlled experiments at 15◦C. Two test solutions of removed any snails from the sides of the tote and swirled Virkon Aquatic (10 and 20 g/L) were used. A bath exposure the contents to spread out the snails on the bottom surface. was conducted with 100 mL of test substance introduced into a Three replicate boots or booties of the same surface were 150-mL acid-washed glass beaker. A spray test was conducted then placed into each tote for a trial. The inside of boots on 10 adult-sized NZMS in a 30-mm-diameter glass petri dish. was weighted down with plastic bags containing bricks or A spray application consisted of two sprays (∼2mL)froma1-L gravel to ensure that the boots were touching the bottom of spray bottle of test solution onto the snails in the petri dish. To the experimental tote. The wader boots were allowed to be serve as controls, we treated snails in beakers and petri dishes colonized by NZMS for 20–60 min until at least 10 or more with water and no chemical. The test durations were 15, 20, or snails could be observed on the surface of each one. 30 min for bath applications and 20, 30, or 40 min for spray After colonization, the boots or booties were transferred di-

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 applications. We increased the exposure time for the spray ap- rectly into a disinfection bath with 20 g/L Virkon Aquatic or plications to determine whether the added time would result in with tap water (as control). The boots were then retained for 15, additional mortality. Each trial was repeated three times for each 20, or 30 min in the disinfection (or control) bath and removed. concentration and exposure time. Boots were then placed into a rinse sequence of three baths At the end of each exposure, the snails were removed from each containing clean tap water. Rinses lasted approximately the test system by pouring the test chemical and snails through a 5–10 s. Boots were inspected to ensure all snails were removed. sieve. Snails were rinsed three times with clean, aged, laboratory Each rinse bath was then sieved, and any NZMS that fell off in water and placed into small 250-mL plastic cups containing aged the rinse system were placed into a recovery cup. The disinfec- water for recovery. Mortality of test organisms was observed tion baths were also emptied, and all organisms were sieved to immediately and at 24 h, and ﬁnal assessment was conducted collect NZMS that had dropped off from the boots. All snails after 48 h by counting the number of live and dead NZMS in recovered from test or rinse systems were again rinsed in three each cup with the aid of a dissecting microscope. We observed separate baths and placed into recovery cups with aged labora- snails for movement or probed individuals to elicit movement or tory water. We repeated our trials until we tested at least 100 tactile response. Neonates released from test snails were counted NZMS infested on each boot type and test interval. As in other and assessed for mortality. We recorded temperature, pH, and experiments, we assessed mortality and enumerated all NZMS conductivity in replicate test and control beakers containing in recovery cups immediately, 24 h, and 48 h postexposure. Any 532 STOCKTON AND MOFFITT

neonates released from test snails were counted at each time iodometric method is a titration that uses the principle that chlo- interval and assessed for mortality. rine will liberate free iodine from potassium iodide solutions Deterioration of waders following repeated disinfecting.— at pH 8 or less, where the liberated iodine is titrated with a We conducted a brief experiment to examine the effect of re- standard solution of sodium thiosulfate with starch as the indi- peated disinfection exposures without rinsing on waterproof cator. To accomplish these tests, a 15-g sample of the 20-g/L characteristics of new neoprene and nylon breathable waders. solution or a 30-g sample of the 10-g/L Virkon Aquatic solu- Two pairs of nylon breathable and one pair of neoprene waders tion was added to a test beaker. Approximately 1 g potassium (CADDIS Systems, La Pine, Oregon) were exposed to a 20-g/L iodide (lot 143188, Sigma Aldrich) and 10 mL of a 200-mL/L Virkon Aquatic solution repeatedly for a 20-min soak at 15◦C sulfuric acid solution was added and mixed. The sample was to follow the deterioration. All material below the front chest immediately titrated with 0.1 N sodium thiosulfate (lot 100705, pocket was immersed in the experimental tote. After each ex- Fisher Scientific) until the yellow color was almost gone. Then posure, waders were examined for holes, discoloration, tears, 2–4 mL of a 1% starch indicator (lot 0155-03, Fisher Scientific) failure of water proofing, or any other signs of wear. One pair was added to make a blue color and titration continued until the of nylon breathable waders was rinsed for comparison. The solution remained clear for at least 30 s. The percent available waders were hung up in the laboratory to dry fully between active oxidizer (%AO) was calculated with the following equa- trials. Waders were then worn by subjects in a large tank with tion provided by Thomas P. Tufano, DuPont Chemical Solutions water for at least 5 min to assess leakage, and disinfections were Enterprise (personal communications): repeated until waders exhibited damage. Effect of organic contaminants on oxidizing activity.—To de- %AO termine the effect of organic contaminants on the effectiveness × × × . / × / = V1 Nthio 100 152 17 g mole KHSO5 16 g mole O2 , of the reagent, we followed the decay of the Virkon Aquatic − 1,000 mL × Ws (g) × 2e × 152.17 g/mole KHSO5 activity with additions of Sphagnum peat moss, stream mud, and (1) NZMS. The peat moss was obtained from a garden supply com- = = pany, the stream mud was collected from a nearby creek, and the where V1 volume of sodium thiosulfate (mL) Nthio normal- = NZMS were from our laboratory We determined the proportion ity of sodium thiosulfate, and Ws weight of Virkon Aquatic of organic content of the peat moss and stream mud by drying sample tested. representative samples at 105◦C to a constant weight. We used the loss-on-ignition method to determine the amount of organic Statistical Analysis material (ASTM 2007). The proportion of organic material was All statistical analyses were conducted using SAS version calculated as the difference between the initial dry weight and 9.2 (SAS Institute, Cary North Carolina) and significant dif- final sample ash weight divided by the initial sample dry weight ferences were reported as P-values. We used a general linear to determine the percentage of organics (Schumacher 2002). model to test for significant differences in the number of adult, To study the deterioration of the test solution (active Virkon juvenile, and neonate NZMS, other aquatic invertebrates, and Aquatic) with and without contamination, we prepared solutions total invertebrates colonizing each of the three boot surfaces that of 10 and 20 g/L of Virkon Aquatic. For tests of peat moss and could be attributed to the boot surface, the two test densities, and stream mud, we placed 1 L of the 10- or 20-g/L concentration their interaction. When we detected a significant effect of boot into a glass beaker with 10 g of dried peat moss or stream surface, we used Tukey’s honestly significant difference test to mud (1% addition). Solutions were stirred with a magnetic stir separate the treatment means. We used a general linear model to

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 plate to assure continuous mixing. At selected intervals (0.5, test the proportion of total to live NZMS remaining on the three 2, 4, 24, and 48 h), we removed triplicate 15- or 30-g samples boot surfaces after the three periods of disinfection to determine from each test system and placed them into separate beakers for differences over time. We used repeated measures ANOVA to analysis of activity. As a control, we placed a solution of 10 and determine the associations between the dependent variable of 20 g/L Virkon Aquatic without organic materials and measured percent active oxidizer over time and organic constituents with similarly at longer intervals (0, 4, 24, 48, 72, 120, and 168 h). concentration of Virkon Aquatic combined as treatments as fol- Tests were repeated three times over 3 weeks. lows: stream mud in 10 g/L, stream mud in 20 g/L, peat moss To study the effect of NZMS as an organic contaminant on in 10 g/L, and peat moss in 20 g/L. The time intervals were 0.5, active chemical, 100-mL samples of a 20-g/L Virkon Aquatic 2, 4, 24, and 48 h. A polynomial transformation was used to solution were poured into individual 150-mL glass beakers for account for the uneven time intervals. two test treatments (with or without 10 NZMS). We removed samples and tested two replicate samples from these beakers, as RESULTS above, at daily intervals for 7 d. We determined the available oxidizing agent (active Virkon Colonization of Three Wader Surfaces Aquatic) in all experiments with the chlorine (residual) iodo- We observed NZMS of all sizes, including neonates, and metric method I (Method 4500-ClB; Clesceri et al. 1996). The other aquatic invertebrates on the wader surfaces after 30 min DISINFECTING WADING BOOTS 533

TABLE 1. Summary of ANOVA of counts of invertebrates colonizing three TABLE 2. Summary of single-degree comparisons of least-squares means types of boot surfaces after 30-min exposure in totes with cultures of New (Table 1) of New Zealand mudsnails (NZMS) and other invertebrates by boot Zealand mudsnails (NZMS) and other invertebrates at two densities. Each test surface and invertebrate grouping. Means with signiﬁcant differences at P < of boots and density was repeated nine times. 0.05 are accompanied by different letters.

Source df Sum of squares F-value P Surface Least-squares mean Separation All invertebrates including NZMS All Invertebrates including NZMS Density 1 930.454 17.82 <0.001 Felt 12.722 z Surface 2 777.463 7.45 0.001 Neoprene 8.111 y Density × Surface 2 71.352 0.68 0.507 Rubber 6.361 y All NZMS All NZMS Density 1 56.333 1.85 0.177 Felt 6.639 z Surface 2 357.389 5.86 0.004 Neoprene 3.778 y Density × Surface 2 18.722 0.31 0.736 Rubber 2.250 y Adult NZMS Adult NZMS Density 1 14.815 0.95 0.333 Felt 4.417 z Surface 2 242.130 7.75 <0.001 Neoprene 3.167 z Density × Surface 2 2.796 0.09 0.915 Rubber 0.806 y Juvenile and neonate NZMS Juvenile and neonate NZMS Density 1 13.370 1.47 0.227 Felt 2.222 Surface 2 46.741 2.58 0.081 Neoprene 0.611 Density × Surface 2 7.407 0.41 0.666 Rubber 1.444 Other invertebrates excluding NZMS Other invertebrates excluding NZMS Density 1 528.898 30.68 <0.001 Felt 6.083 Surface 2 84.019 2.44 0.093 Neoprene 4.333 Density × Surface 2 19.241 0.56 0.574 Rubber 4.111

of exposure. We found significant differences in the numbers at- Disinfection of Three Wader Surfaces Infested with NZMS tributable to type of wader surface (Table 1), but no significant All NZMS on wader surfaces exposed to baths with 20 g/L interactions between the density of NZMS at exposure and sur- Virkon Aquatic were killed regardless of exposure time. No > > face. The effect of wader surface was 0.10 P 0.05 for models neonates were found in recovery baths (Table 4). The pH, tem- of numbers of juvenile and neonate NZMS and numbers of in- perature, and conductivity of the test solutions remained consis- vertebrates excluding NZMS (Table 1). We found counts were tent throughout all trials. Some mortality of NZMS was observed significantly higher on felt surfaces for totals of all invertebrates and all NZMS (P < 0.05). We found that the least-squares mean number of organisms on rubber surfaces was lowest, except for juvenile and neonate NZMS (Table 2). Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 Spray versus Bath A bath of 20 g/L Virkon Aquatic for 30 min was effective in killing all stages of NZMS (Figure 1). In shorter exposure times, we observed some snails surviving or live neonates released during recovery. Live neonates were observed released from adult NZMS in control and 10-g/L bath applications, and live neonates were found in recovery containers of all concentration and exposure times tested with sprays (Table 3). The pH, temperature, and conductivity of the test solutions were consistent and varied little over the trials. The temperature of all solutions averaged 15.4◦C before and after testing, and pH averaged 7.16 in controls and 2.4 in solutions of 10 and 20 g/L Virkon Aquatic, respectively. Total conductivity increased with increasing concentrations of Virkon and averaged 8,123 mS/cm FIGURE 1. The percent mortality (±SE) of New Zealand mudsnails after in 10 g/L and 14,500 mS/cm in 20 g/L. Conductivity of control simulated spray or bath exposures with 10 and 20 g/L of Virkon Aquatic. water was 268 µS/cm. Control mortality ranged from 0% to 4%. 534 STOCKTON AND MOFFITT

TABLE 3. Summary of mortality of adult New Zealand mudsnails and number of live neonates released after spray versus bath applications of Virkon Aquatic using three exposure times and two concentrations (10 and 20 g/L).

Concentration Time Number Total number Percent Number of Test (g/L) (min) dead tested mortality (SE) live neonates Bath Control (0) 15 2 90 2 (0.04) 19 20 0 90 0 (0) 19 30 1 90 1 (0.03) 22 10 15 81 90 90 (0.15) 5 20 85 92 92 (0.09) 7 30 89 92 97 (0.03) 5 20 15 89 90 99 (0.03) 0 20 89 90 99 (0.03) 0 30 90 90 100 (0) 0 Spray Control (0) 20 0 90 0 (0) 22 30 0 89 0 (0) 26 40 0 90 0 (0) 23 10 20 67 91 74 (0.17) 27 30 84 91 92 (0.09) 9 40 84 90 93 (0.07) 3 20 20 86 91 95 (0.07) 6 30 89 90 99 (0.03) 2 40 89 90 99 (0.03) 1

TABLE 4. Summary of mortality in disinfections of boots naturally infested with New Zealand mudsnails (NZMS). The total mortality of NZMS and number of live neonates recovered from soles of boots by concentration and time in bath exposure to Virkon Aquatic is provided. TNTC = too numerous to count (>300).

Boot Concentration Exposure Number of Total number Number Percent Number of surface (g/L) (min) replicates NZMS tested dead mortality (SE) live neonates Felt Control 15 3 150 9 6 (3.4) 6 20 2 116 3 3 (1.9) 5 30 3 108 5 5 (2.7) 8 20 15 3 174 174 100 (0) 0 20 2 233 233 100 (0) 0 Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 30 3 247 247 100 (0) 0 Neoprene Control 15 1 967 5 0.5 TNTC 20 2 267 4 1.5 (0.2) TNTC 30 1 235 4 2 TNTC 20 15 3 152 152 100 (0) 0 20 1 216 216 100 (0) 0 30 3 730 730 100 (0) 0 Rubber Control 15 4 56 0 0 (0) 13 20 4 48 2 4 (5.9) 30 30 4 113 7 6 (3.8) 31 20 15 4 233 233 100 (0) 0 20 4 207 207 100 (0) 0 30 4 532 532 100 (0) 0 DISINFECTING WADING BOOTS 535

TABLE 5. Percent active oxidizer (AO) (and SE) for each concentration of Virkon Aquatic with no organic material, 10 g/L peat moss, or 10 g/L Paradise Creek stream mud for each time tested. NT = not tested.

Mean% AO (SE) 10 g/L Virkon Aquatic 20 g/L Virkon Aquatic Time (h) No organic material Peat moss Stream mud No organic material Peat moss Stream mud 0 12.81 (1.85) NT NT 12.37 (3.52) NT NT 0.5 NT 12.20 (3.49) 11.73 (0.87) NT 9.31 (0.78) 10.44 (0.11) 2 NT 10.42 (1.15) 9.66 (0.16) NT 8.87 (0.46) 9.67 (0.28) 4 10.98 (0.39) 10.10 (0.81) 10.01 (0.42) 10.28 (0.06) 8.34 (0.08) 9.39 (0.24) 24 10.65 (0.52) 7.27 (0.38) 8.35 (0.41) 9.96 (0.22) 6.76 (0.31) 8.51 (0.16) 48 10.79 (0.26) 6.00 (0.27) 4.10 (0.16) 9.30 (0.07) 4.85 (0.12) 7.56 (0.19) 72 11.21 (0.20) NT NT 9.29 (0.07) NT NT 120 11.23 (0.19) NT NT 9.00 (0.01) NT NT 168 11.10 (0.53) NT NT 9.01 (0.19) NT NT

in control solutions. We detected no damage to felt, rubber, or for the sphericity. We found that the linear part of the polyno- neoprene wading gear over repeated disinfections with 20 g/L mial model was significant (P < 0.05), but the quadratic, cubic, Virkon Aquatic. Preobservation and postobservation tests on or quartic trends were not significant (P > 0.10). The oxidizing damages revealed that the holes, cracks, stains, discoloration, activity of the 20-g/L solution of snails and no snails in Virkon and tears did not increase in severity throughout the study. We Aquatic over 0–144 h of exposure remained above 9.0%; the found no waterproofing failure in any of the tested boots or means (SE) were 9.72% (SE = 0.322) and 9.70% (SE = 0.599) booties. in solutions with and without NZMS, respectively.

Deterioration of Waders with Repeated Disinfecting The crotch of the neoprene waders started to leak after 65 DISCUSSION exposures to Virkon Aquatic with rinsing. Testing continued Our studies validated the already published account that felt with the neoprene waders until the legs leaked water, and after surfaces on wading gear are likely to transport aquatic in- 77 exposures, both booties leaked around the seams. The ny- vertebrates and other microorganisms including ﬁsh disease lon breathable waders that were not rinsed leaked behind both pathogens (Gates et al. 2008, 2009; Bothwell et al. 2009; knees after 29 exposures to Virkon Aquatic, and waterprooﬁng sealant appeared to disintegrate. After 30 exposures, the breathable waders that were rinsed leaked at the crotch. We repeated testing with these waders until the legs leaked, at 43 exposures, with disintegration of the glue occurring around the seams. We detected no discoloration, tears, or holes in the waders, but dam-

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 age occurred in the seam or waterprooﬁng glue.

Effect of Organic Contaminants on Oxidizing Activity Organic material caused a decrease in oxidizing activity of Virkon Aquatic solutions over time of exposure (Table 5; Figure 2). Tests above 9.0% were considered active compound (Jeffery Odle, DuPont Animal Health Solutions, Wilmington, Delaware, personal communications). The organic fractions of peat moss and creek mud were 92.9% and 3.9%, respectively. The peat moss deactivated the 10- and 20-g/L solutions rapidly (Figure 2). The stream mud deactivated the 10-g/L Virkon Aquatic solution more than the 20-g/L solution, and the differences were apparent at 48 h. FIGURE 2. The percent available oxidizer over time in solutions of 10 and We found a signiﬁcant time and treatment effect on the ox- 20 g/L Virkon Aquatic with 10 g peat moss or stream mud. Virkon Aquatic idizing activity (Time F = 56.22, P < 0.001; Treatment F = is considered active at percent available oxidizer above 9.0 (horizontal dashed 3.39, P = 0.018). The data ﬁt the model, satisfying the criteria line). 536 STOCKTON AND MOFFITT

Waterkeyn et al. 2010). We found NZMS infested all boot sur- after approximately 30 exposures of 20 min each to a 20-g/L faces including rubber soles, although numbers were lower on Virkon Aquatic solution. Neoprene seam glue lasted longest, the rubber surfaces. The need for regulations and informed be- failing after 67 exposures. We recommend additional testing of havior regarding wading gear transport cannot be overestimated. equipment to evaluate seam glues and the effects on different Gates et al. (2009) found that anglers wearing boots and waders equipment manufacturers’ products. potentially transported an average of 16.78 g of sediments dur- The 20-g/L concentration that was effective in our tests is in ing angling trips in Montana. Gates et al. (2008) reported that line with concentrations recommended for disinfecting surfaces the interstitial spaces of felt boots were the largest when com- and gear for target bacteria, fungi, and parasites of 10 or 20 g/L pared with lightweight nylon, neoprene, and rubber. Waterkeyn for at least 5 min (DAHS 2011a; Western Chemical 2012). For et al. (2010) found a suite of organisms were contained in mud aquaculture facilities already using Virkon Aquatic disinfectant on footwear and wading suits of 25 field workers and biolo- protocols to protect against pathogens, the expansion of this tool gists but did not relate the gear type. They found 18 taxa on the to reduce risks of infestation with NZMS can be easy. However, footwear and wading suits, including protozoans, brine shrimp additional tests will probably be needed to expand the label Artemia sp., ostracods, rotifers, cladocerans, large freshwater claim to meet the USEPA criteria for invasive mollusks. branchiopods, bryozoans, turbellarians, and nematodes. We documented that Virkon Aquatic solutions were deacti- In our small-scale tests comparing spray applications of vated by organic material after 4–24 h of exposure. We found NZMS in petri dishes with submersion in laboratory beakers, stream mud deactivated 10 g/L Virkon Aquatic at a faster rate we found a disinfection bath of Virkon Aquatic for 30 min than the peat moss. Although this was not expected, stream was needed to kill all NZMS and neonates. Spray applica- mud is a heterogeneous mixture containing metals and organic tions provided only 99% mortality after 30–40 min before rins- material, and thorough mixing did not make the mixture ho- ing, and live neonates were present in the recovery containers. mogenous. Since Virkon Aquatic is broken down by metal salts Oplinger and Wagner (2009b) found that 15-min spray appli- (DAHS 2011b), there could have been a different composition cations of copper sulfate, hydrogen peroxide, Clorox Commer- of organic material and metal salts in the mud. Another expla- cial Solutions Formula 409 cleaner–degreaser–disinfectant, and nation for the different decomposition rates between 10- and Hyamine 1622 were 100% effective on NZMS, but they did not 20-g/L solutions was that a concentration of 20 g/L had a higher report examination of recovery systems for neonates. In other amount of active ingredients to prolong the reaction than did the studies of NZMS, live neonates were often released rapidly in concentration of 10 g/L. The peat moss had a high percent or- the recovery water systems from dying adults (Bruce and Moffitt ganic material and deactivated the 20-g/L solution faster than the 2010). Schisler et al. (2008) reported that compromised NZMS stream mud. Disinfecting protocols recommend removal of or- that survived exposure often released live neonates in the re- ganics and debris during or before decontamination procedures covery bath. Bath disinfection is recommended for most other (Hosea and Finlayson 2005; Schisler et al. 2008; CDFG 2012). reagents tested on NZMS (Hosea and Finlayson 2005; Schisler We did not test the effectiveness of the solutions below the sug- et al. 2008). Spray applications do not cover all surfaces with gested limit of activity of 9%. Further tests could be conducted the same amount or concentration of disinfectant for the re- to explore the safety limits of compromised solutions. quired time, and there are more likely to be inconsistencies in Banning different gear types has been proposed as one so- the individual delivery methods and spray apparatus. lution for controlling the spread of aquatic invasive species, We validated that the concentration and durations of exposure specifically Didymosphenia geminata and Myxobolus cerebralis to Virkon Aquatic in our small laboratory test systems were ex- (Spaulding and Elwell 2007; Gates et al. 2008; Bothwell et al.

Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 ceptionally effective on boots and booties infested with NZMS. 2009; Kilroy and Unwin 2011). Wading gear provides a path- Mortality was rapid, and no snails were observed crawling in the way for NZMS and other nonindigenous organisms to spread disinfectant bath. Instead, we observed NZMS falling off when (Dwyer et al. 2003; Richards et al. 2004; Hosea and Finlayson the boots or booties were placed into test solutions. Very few 2005). However, our research demonstrates that banning felt- NZMS were found in the rinse tubs and those found were dead, soled wading boots will not eliminate the risk of introduction. even after a 15-min exposure. The effectiveness of the shorter Moveover, precaution must be exercised when using chemical exposure time in tests with infested surfaces could be attributed disinfectants to ensure efﬁcacy and safety of the environment to the fact that most snails are attached with their foot to gear and worker (Oplinger and Wagner 2009a, 2010). Many chem- and the chemical reached them rapidly. In the laboratory bath icals tested are not safe near the water and have high aquatic tests, the NSMS had closed opercula. toxicity. Although other studies have documented the efﬁcacy Although our tests of deterioration of boots and waders were of a range of QACs as disinfectants (Hosea and Finlayson 2005; limited, our studies support that Virkon Aquatic was not as harsh Schisler et al. 2008; Oplinger and Wagner 2009b), there are on waders and boots as observed with some other chemicals. concerns about the widespread use of QACs and their poten- Hosea and Finlayson (2005) found deterioration of boots and tial effect on aquatic and soil systems (Garc´ıa et al. 2001; Li waders after seven exposures to chemicals such as bleach and and Brownawell 2010). The QACs are used in numerous indus- Pine-Sol. We observed disintegration in the glue of wader seams trial and household applications, including in fabric softeners, DISINFECTING WADING BOOTS 537

detergents, disinfectants, preservatives, and personal care prod- Britton, D., and S. Dingman. 2011. Use of quaternary ammonium to control ucts. Their persistence has increased concerns regarding their the spread of aquatic invasive species by wildland fire equipment. Aquatic genotoxic effects (Ferk et al. 2007). We found that Virkon Invasions 6:169–173. Bruce, R. L., and C. M. Moffitt. 2010. Quantifying risks of volitional consump- Aquatic was effective on boots after a short duration of bath tion of New Zealand mudsnails by steelhead and Rainbow Trout. Aquaculture exposure. The chemical is deactivated quickly by organic mate- Research 41:552–558. rial and is minimally corrosive on gear. However, prerinsing gear CANS (Center for Aquatic Nuisance Species). 2012. Status of felt restrictions to remove soil, excess organic material, and large organisms is in the United States. Invasive Species Action Network, CANS, Livingston, an essential step before using any disinfectant. Montana. Available: www.stopans.org/Felt Bans.htm. (June 2012). CDFG (California Department of Fish and Game). 2012. Aquatic invasive species decontamination protocol. CDFG, Sacramento. Available: nrm.dfg.ca.gov/FileHandler.ashx?DocumentID=43333. (December 2012). ACKNOWLEDGMENTS Clesceri, L. S., A. E. Greenberg, and A. D. Eaton, editors. 1996. Standard We thank B. Sun, T. Britton, A. Christensen, and B. Win- methods for the examination of water and wastewater, 20th edition. American ston for assistance with laboratory testing and preparation at the Public Health Association, American Water Works Association, and Water University of Idaho. We are especially grateful to A. Barenberg Environment Federation, Washington, D.C. Curry, C. H., J. S. McCarthy, H. M. Darragh, R. A. Wake, S. E. Churchill, for conducting experiments as part of her Environmental Sci- A. M. Robins, and R. J. Lowen. 2005. Identification of an agent suitable for ence senior thesis program. Staff at Hagerman National Fish disinfecting boots of visitors to the Antarctic. Polar Record 41:39–45. Hatchery (HNFH) provided snails and waders for testing. Other DAHS (DuPont Animal Health Solutions). 2011a. VirkonR Aquatic effi- waders were provided by staff at Dworshak National Fish Hatch- cacy against specific fish pathogens. DAHS, E. I. Du Pont de Nemours ery, Hagerman State Fish Hatchery, Willow Beach National and Company, Wilmington, Delaware. Available: www2.dupont.com/DAHS EMEA/en GB/ahb/fish/efficacy data.html. (April 2011). Fish Hatchery, Clearwater Fish Hatchery, Niagara Springs Fish DAHS (DuPont Animal Health Solutions). 2011b. Environmental decompo- Hatchery, Magic Valley Fish Hatchery, and Utah Division of sition of VirkonR Aquatic. DAHS, E. I. Du Pont de Nemours and Com- Wildlife Resources. Tri-State and Sportsman’s Warehouse do- pany, Suffolk, UK. Available: www.syndel.com/Assets/File/environmental nated new waders for tests. Funding was provided by the U.S. decomposition of virkon aquatic.pdf. (January 2011). Fish and Wildlife Service (P. Heimowitz, project officer). Ad- Dove, M. C., and W. A. O’Connor. 2007. Ecotoxicological evaluations of common hatchery substances and procedures used in the production of Sydney ditional support was provided by the U.S. Geological Survey rock oysters Saccostrea glomerata (Gould 1850). Journal of Shellfish Re- and the National Science Foundation, CRISSP Research Expe- search 26:501–508. rience for Undergraduates. We are grateful to A. Sepulveda and Dwyer, W. P., B. L. Kerans, and M. M. Gangloff. 2003. Effect of acute exposure two anonymous reviewers for their assistance reviewing earlier to chlorine, copper sulfate, and heat on survival of New Zealand mud snails. drafts of this manuscript. The use of trade names or products Intermountain Journal of Sciences 9:53–58. EPA (Environmental Protection Agency). 2010. Residues of quaternary ammo- does not constitute endorsement by the U.S. Government. nium compounds, N-alkyl (C12-14) dimethyl ethylbenzyl ammonium chloride; exemption from the requirement of a tolerance, final rule. Federal Register 75:134(14 July 2010):40729–40736. REFERENCES Ferk, F., M. Misˇ´ık, C. Hoelzl, M. Uhl, M. Fuerhacker, B. Grillitsch, W. Alonso, A., and P. Castro-D´ıez. 2008. What explains the invading success of Parzefall, A. Nersesyan, K. Micieta,ˇ T. Grummt, V. Ehrlich, and S. the aquatic mud snail Potamopyrgus antipodarum (Hydrobiidae, Mollusca)? Knasmuller.¨ 2007. Benzalkonium chloride (BAC) and dimethyldioctadecyl- Hydrobiologia 614:107–116. ammonium bromide (DDAB), two common quaternary ammonium com- Alonso, A., and P. Castro-D´ıez. 2012. The exotic aquatic mud snail Potamopy- pounds, cause genotoxic effects in mammalian and plant cells at environ- rgus antipodarum (Hydrobiidae, Mollusca): state of the art of a worldwide mentally relevant concentrations. Mutagenesis 22:363–370. invasion. Aquatic Sciences 74:375–383. Firestone, J., and J. J. Corbett. 2005. Coastal and port environments: inter- ASTM (American Society for Testing and Materials). 2007. ASTM D 2974-07a: national legal and policy responses to reduce ballast water introductions Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 standard test methods for moisture, ash, and organic matter of peat and other of potentially invasive species. Ocean Development and International Law organic soils. ASTM, West Conshohocken, Pennsylvania. 36:291–316. Boethling, R. S. 1984. Environmental fate and toxicity in wastewater treatment Garc´ıa, M. T., I. Ribosa, T. Guindulain, J. Sanchez-Leal,´ and J. Vives-Rego. of quaternary ammonium surfactants. Water Research 18:1061–1076. 2001. Fate and effect of monoalkyl quaternary ammonium surfactants in the Bothwell, M. L., D. R. Lynch, H. Wright, and J. Deniseger. 2009. On the boots aquatic environment. Environmental Pollution 111:169–175. of fishermen: the history of didymo blooms on Vancouver Island, British Gates, K. K., C. S. Guy, A. V. Zale, and T. B. Horton. 2008. Adherence of Columbia. Fisheries 34:382–388. Myxobolus cerebralis myxospores to waders: implications for disease dis- Bowler, P. A. 1991. The rapid spread of the freshwater hydrobiid snail Pota- semination. North American Journal of Fisheries Management 28:1453– mopyrgus antipodarum (Gray) in the middle Snake River, southern Idaho. 1458. Pages 173–182 in E. P. Pister, editor. Proceedings of the desert fishes council, Gates, K. K., C. S. Guy, A. V. Zale, and T. B. Horton. 2009. Angler awareness volume 21. Desert Fishes Council, Bishop, California. of aquatic nuisance species and potential transport mechanisms. Fisheries Brenneis, V. E. F., A. Sih, and C. E. de Rivera. 2011. Integration of an invasive Management and Ecology 16:448–456. consumer into an estuarine food web: direct and indirect effects of the New Hall, R. O., Jr., M. F. Dybdahl, and M. C. VanderLoop. 2006. Extremely high Zealand mud snail. Oecologia 167:169–179. secondary production of introduced snails in rivers. Ecological Applications Briski, E., C. J. Wiley, and S. A. Bailey. 2012. Role of domestic shipping in 16:1121–1131. the introduction or secondary spread of nonindigenous species: biological Hosea, R. C., and B. Finlayson. 2005. Controlling the spread of New Zealand invasions within the Laurentian Great Lakes. Journal of Applied Ecology mud snails on wading gear. California Department of Fish and Game, Office of 49:1124–1130. Spill Prevention and Response, Administrative Report 2005-02, Sacramento. 538 STOCKTON AND MOFFITT

Johnson, L. E., A. Ricciardi, and J. T. Carlton. 2001. Overland dispersal of Trout growth and condition. North American Journal of Aquaculture 72: aquatic invasive species: a risk assessment of transient recreational boating. 207–212. Ecological Applications 11:1789–1799. Perrings, C., S. Burgiel, M. Lonsdale, H. Mooney, and M. Williamson. 2010. Johnson, M. L., L. Berger, L. Philips, and R. Speare. 2003. Fungicidal effects International cooperation in the solution to trade-related invasive species of chemical disinfectants, UV light, desiccation and heat on the amphib- risks. Annals of the New York Academy of Sciences 1195:198–212. ian chytrid Batrachochytrium dendrobatidis. Diseases of Aquatic Organisms Ricciardi, A., and J. B. Rasmussen. 1998. Predicting the identity and impact 57:255–260. of future biological invaders: a priority for aquatic resource management. Kilroy, C., and M. Unwin. 2011. The arrival and spread of the bloom-forming, Canadian Journal of Fisheries and Aquatic Sciences 55:1759–1765. freshwater diatom, Didymosphenia geminata, in New Zealand. Aquatic Inva- Richards, D. C., P.O’Connell, and D. Cazier Shinn. 2004. Simple control method sions 6:249–262. to limit the spread of the New Zealand mudsnail Potamopyrgus antipodarum. Li, X., and B. J. Brownawell. 2010. Quaternary ammonium compounds in urban North American Journal of Fisheries Management 24:114–117. estuarine sediment environments: a class of contaminants in need of increased Riley, L. A., M. F. Dybdahl, and R. O. Hall Jr. 2008. Invasive species impact: attention? Environmental Science and Technology 44:7561–7568. asymmetric interactions between invasive and endemic freshwater snails. Loo, S. E., R. P. Keller, and B. Leung. 2007. Freshwater invasions: using Journal of the North American Benthological Society 27:509–520. historical data to analyse spread. Diversity and Distributions 13:23–32. Root, S., and C. M. O’Reilly. 2012. Didymo control: increasing the effectiveness Lysne, S., and P. Koetsier. 2008. Comparison of desert valvata snail growth of decontamination strategies and reducing spread. Fisheries 37:440–448. at three densities of the invasive New Zealand mudsnail. Western North Sarkar, B., M. Megharaj, Y. Xi, G. S. R. Krishnamurti, and R. Naidu. 2010. American Naturalist 68:103–106. Sorption of quaternary ammonium compounds in soils: implications to the Mainous, M. E., S. A. Smith, and D. D. Kuhn. 2010. Effect of common aquacul- soil microbial activities. Journal of Hazardous Materials 184:448–456. ture chemicals against Edwardsiella ictaluri and E. tarda. Journal of Aquatic Schisler, G. J., N. K. M. Vieira, and P.G. Walker. 2008. Application of household Animal Health 22:224–228. disinfectants to control New Zealand mudsnails. North American Journal of McGeer, J. C., C. Szebedinszky, D. G. McDonald, and C. M. Wood. 2000. Fisheries Management 28:1172–1176. Effects of chronic sublethal exposure to waterborne Cu, Cd or Zn in Rainbow Schmidt, B. R., C. Geiser, N. Peyer, N. Keller, and M. von Rutte.¨ 2009. As- Trout 2: tissue specific metal accumulation. Aquatic Toxicology 50:245– sessing whether disinfectants against the fungus Batrachochytrium dendro- 256. batidis have negative effects on tadpoles and zooplankton. Amphibia-Reptilia Mills, E. L., J. H. Leach, J. T. Carlton, and C. L. Secor. 1993. Exotic species in 30:313–319. the Great Lakes: a history of biotic crises and anthropogenic introductions. Schumacher, B. A. 2002. Methods for the determination of total organic carbon Journal of Great Lakes Research 19:1–54. (TOC) in soils and sediments. U.S. Environmental Protection Agency, Office Mitchell, A. J., and R. A. Cole. 2008. Survival of the faucet snail after chemical of Research and Development, Ecological Risk Assessment Support Center, disinfection, pH extremes, and heated water bath treatments. North American Washington, D.C. Journal of Fisheries Management 28:1597–1600. Shaw, B. J., G. Al-Bairuty, and R. D. Handy. 2012. Effects of waterborne copper Mitchell, A. J., M. S. Hobbs, and T. M. Brandt. 2007. The effect of chemical nanoparticles and copper sulphate on Rainbow Trout (Oncorhynchus mykiss): treatments on red-rim melania Melanoides tuberculata, an exotic aquatic snail physiology and accumulation. Aquatic Toxicology 116–117:90–101. that serves as a vector of trematodes to fish and other species in the USA. Spaulding, S. A., and L. Elwell. 2007. Increase in nuisance blooms and geo- North American Journal of Fisheries Management 27:1287–1293. graphic expansion of the freshwater diatom Didymosphenia geminata. U.S. Moffitt, C. M., and C. A. James. 2012. Dynamics of Potamopyrgus antipodarum Geological Survey, Open File Report 2007-1425, Reston, Virginia. infestations and seasonal water temperatures in a heavily used recreational Stockton, K. A. 2011. Methods to assess, control, and manage risks for two watershed in intermountain North America. Aquatic Invasions 7:193–202. invasive mollusks in fish hatcheries. Master’s thesis. University of Idaho, Oplinger, R. W., and E. J. Wagner. 2009a. Toxicity of common aquaculture Moscow. disinfectants to New Zealand mud snails and mud snail toxicants to Rainbow Waterkeyn, A., B. Vanschoenwinkel, S. Elsen, M. Anton-Pardo, P. Grillas, Trout eggs. North American Journal of Aquaculture 71:229–237. and L. Brendonck. 2010. Unintentional dispersal of aquatic invertebrates Oplinger, R. W., and E. J. Wagner. 2009b. New Zealand mudsnail contain- via footwear and motor vehicles in a Mediterranean wetland area. Aquatic ment: tests of chemicals, copper strips, fluorochrome marking and terrestrial Conservation: Marine and Freshwater Ecosystems 20:580–587. movement. Intermountain Journal of Sciences 15:41–53. Western Chemical. 2012. VirkonR Aquatic brochure. Western Chemi- Oplinger, R. W., and E. J. Wagner. 2010. Effect of potassium permanganate cal, Ferndale, Washington. Available: www.wchemical.com/Webpage.aspx? treatments on New Zealand mud snail behavior and survival and Rainbow WebpageId=19. (October 2012). Downloaded by [Department Of Fisheries] at 20:05 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:06 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Longitudinal Variation in the Ohio River Fish Community Greg Seegert a , Joe Vondruska a , Elgin Perry b & Doug Dixon c a EA Engineering, Science, and Technology , 444 Lake Cook Road, Suite 18, Deerfield , Illinois , 60015 , USA b Statistics Consultant , 2000 Kings Landing Road, Huntingtown , Maryland , 20639 , USA c Electric Power Research Institute , 3420 Hillview Avenue, Palo Alto , California , 94304 , USA Published online: 16 May 2013.

To cite this article: Greg Seegert , Joe Vondruska , Elgin Perry & Doug Dixon (2013): Longitudinal Variation in the Ohio River Fish Community, North American Journal of Fisheries Management, 33:3, 539-548 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785990

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:539–548, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785990

ARTICLE

Longitudinal Variation in the Ohio River Fish Community

Greg Seegert* and Joe Vondruska EA Engineering, Science, and Technology, 444 Lake Cook Road, Suite 18, Deerﬁeld, Illinois 60015, USA Elgin Perry Statistics Consultant, 2000 Kings Landing Road, Huntingtown, Maryland 20639, USA Doug Dixon Electric Power Research Institute, 3420 Hillview Avenue, Palo Alto, California 94304, USA

Abstract The Ohio River fish community was sampled three times (June, August, and October) in 2005 by electrofishing near 17 power plants encompassing nearly the entire length of the Ohio River (river kilometer 82–1,527). Six 500-m zones were electrofished at each plant. Using a generalized additive model, we examined this data set of 306 electrofishing samples to determine how the abundance of 31 taxa varied over the length of the river. We also looked for trends in several measures of community health (catch rate, species richness, index of biotic integrity, and modified index of well-being). Of the 31 taxa examined, all but three showed longitudinal differences over the course of the river. Based on the patterns for the 31 taxa and 5 community measures, we assigned each taxon or measure to one of seven classes. The largest class included 10 taxa that declined steadily in abundance from upriver to downriver. All community measures followed a similar pattern except that each measure increased near the downstream-most plant. Conversely, only two taxa (Striped Bass Morone saxatilis and River Carpsucker Car- piodes carpio) showed the opposite trend (i.e., increasing in a downstream direction). Some taxa were restricted to the upper or lower end of the river or to certain reaches along the river. Species that were restricted to the upper river were primarily stream fishes (certain minnows Notropis spp. and several darters Etheostoma spp.). Conversely, fish species that were restricted to the lower river were primarily large-river fishes as well as species that were typical of the Mississippi River fauna. Site-specific differences in water temperature, Secchi depth, specific conductance, and dissolved oxygen were generally unimportant in explaining the observed patterns. Substrate type (percent cobble/boulder and percent silt) occasionally was important. The observed longitudinal trends appeared to be primarily in response to natural temperature and habitat differences between the upper and lower sections of the Ohio River. Downloaded by [Department Of Fisheries] at 20:06 28 May 2013

Studies of the Ohio River fish community span well over two during his 1818 trip (Pearson and Krumholz 1984). Agriculture centuries. The first reasonably detailed account of Ohio River and deforestation in the 19th century and dam building in the fishes was provided by Rafinesque (1820), who in 1818 traveled 20th century transformed the Ohio River from a clear, relatively the length of the river and described over 20 new species. At shallow river dominated by hard substrates to a turbid, slug- that time, the river was fairly clear and was dominated by hard gish river with predominately silt and sand substrates (Trautman substrates except in its lowermost reaches, with shallow-water 1981; Pearson and Krumholz 1984); most of the other large U.S. areas being common in the summer (Trautman 1981). The most rivers also share this modification. Until fairly recently, the fish famous shallow-water area was the “Falls of the Ohio” (near community in the upper portion of the river received little at- Louisville, Kentucky), where Rafinesque spent much of his time tention from resource agencies because of pollution problems

*Corresponding author: [email protected] Received March 16, 2012; accepted March 10, 2013 539 540 SEEGERT ET AL.

caused by various point sources and acid-mine drainage into the frequency of sampling (e.g., Ohio River: Pearson et al. 2011; Monongahela River basin, which forms the headwaters of the Tennessee River: Dennis Baxter, Tennessee Valley Authority, Ohio River (Pearson and Krumholz 1984). personal communication). The participation of 17 power plants During the past 50 years, two programs have contributed in 2005 allowed us the unique opportunity to assess the distri- greatly to our understanding of Ohio River fishes. Beginning bution of fishes over the course of the river in a quantitative in 1957, the Ohio River Valley Water Sanitation Commission fashion. (ORSANCO) conducted studies of various lock chambers; these By any measure, the Ohio River is a “large” river, and it studies involved rotenone application to the chambers and tab- is characterized as a “great” river by some authors (Simon and ulation of the fishes collected. The Ohio River Valley Water Lyons 1995; Emery et al. 2003). Given its consistently large size Sanitation Commission is a consortium of states bordering the and the fact that most large-river species are widespread and river and has the responsibility of establishing water quality highly mobile, one might expect the fauna to vary little over the standards on the river. Because of concerns that the lock cham- length of the river. Thus, one of our objectives was to determine ber samples did not represent a complete cross section of the quantitatively whether this is the case for the Ohio River. As fish community, ORSANCO began an electrofishing program indicated above, we are not aware of any previous effort on in 1990 to intensively sample some pools during each year, with such an expanse of the Ohio River or any other river. A second the goal of sampling all pools in the river over about a 5-year objective was to determine what factors might be responsible period. for longitudinal differences if those differences were observed. The other major, ongoing monitoring program in the system We chose the 2005 data set (i.e., with data from 17 power plants) is the Ohio River Ecological Research Program (ORERP), a because it covered the greatest area (river kilometer [rkm] 82– collaborative effort in which various power plants located along 1,527) and had the fewest spatial gaps. the Ohio River collectively fund fish studies near each plant. We believe this analysis will be useful to resource agencies The ORERP has been conducted annually since 1970. Over the in making management decisions. Knowledge of how the abun- years, the ORERP has funded studies employing a multigear dance of certain species changes over the course of the Ohio approach to collect adult and juvenile fishes (EPRI 2007, 2008, River should also aid agencies in making assessments as to 2010, 2012). Previously, larval fish were also studied regularly whether the abundance levels they observe are the result of nat- (ESE 1991), and in 2005–2007, an intensive fish impingement ural longitudinal variation or possibly the result of point source study was conducted (King et al. 2010). The number of plants or non-point-source inputs. These data should be particularly participating in the ORERP varies annually; in 2005, a record useful to those developing or refining iterations of the index number of plants (17) participated in the adult fish surveys. of biotic integrity (IBI) for the Ohio River. One of our goals Resource agencies bordering the Ohio River also conduct was to determine whether the abundance of various fish species various fish monitoring activities on the river; however, the pro- or community measures change over the course of the river. grams conducted by the resource agencies typically target one If abundance does not change, then an IBI metric using that or two species of management concern and are usually limited species or measure (e.g., overall catch rate and species richness) to the political boundaries of each agency. These studies typi- can be uniformly applied throughout the river. If abundance cally involve only limited sections of the river in a given year, does change longitudinally, then the metric will likely have to often during only a portion of the field season. The Ohio River be adjusted to account for river kilometer. Valley Water Sanitation Commission studies the entire river but generally only samples a few pools in a given year, making it

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 difficult to determine whether abundance differences in different pools are real or are the result of interyear variability. Other METHODS authors (Pearson and Krumholz 1984; Burr and Page 1986) Study area.—The Ohio River is 1,579 km long and drains have described the geographic distribution of fishes in the river, 490,601 km2; average annual discharge at Cairo, Illinois, is but these descriptions are primarily qualitative, essentially us- about 8,000 m3/s. The river begins in Pittsburgh, Pennsylva- ing presence/absence (rather than quantitative data) to describe nia, at the confluence of the Monongahela and Allegheny rivers distribution. In 2004–2006, Pearson et al. (2011) collected 147 and flows in a west–southwesterly direction, joining the Mis- electrofishing samples over the length of the Ohio River. To our sissippi River near Cairo, Illinois (Figure 1). The Ohio River knowledge, the 2005 ORERP is the first study of a large river flows through three ecoregions: the Western Allegheny Plateau, that covered nearly the entire length of the river in a single year the Interior Plateau, and the Interior River Lowlands (Omernik with three sets of seasonal data (late spring, summer, and fall) and Gallant 1988; Woods et al. 1998, 2002). The Western Al- and that involved analysis of the data in a quantitative man- legheny Plateau ecoregion extends from near Pittsburgh (rkm 0) ner. Studies of other large rivers have covered shorter distances to approximately rkm 646; the Interior Plateau ecoregion ranges (e.g., Willamette River, Oregon: Hughes and Gammon 1987), approximately from rkm 646 to rkm 1,207; and the Interior River have not included a seasonal component (e.g., Missouri River: Lowlands ecoregion spans the reach between rkm 1,207 and the Berry et al. 2004), or have had greater spatial gaps and a lower Ohio River’s confluence with the Mississippi River. LONGITUDINAL VARIATION IN THE FISH COMMUNITY 541

FIGURE 1. Power plants that were studied during the 2005 Ohio River Ecological Research Program. See Table 1 for river kilometer locations of the power plants. [Figure available in color online.]

The mainstream Ohio River fish fauna has been affected by TABLE 1. Power plants that participated in the 2005 Ohio River Ecological the system of locks and dams that provides commercial naviga- Research Program (rkm = river kilometer). tion through maintenance of a 2.7-m-deep channel for the entire Plant rkm length of the Ohio River. These dams have transformed a free- flowing river into a continuous series of impoundments, thereby W. H. Sammis (FirstEnergy) 86.7 altering fish community structure. There are 20 lock-and-dam Cardinal (Buckeye Power) 123.4 structures on the Ohio River to aid navigation and control flood- R. E. Burger (FirstEnergy) 164.1 ing (Pearson and Krumholz 1984). Although these structures Kammer (American Electric Power) 178.8 Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 undoubtedly affect fish movement on a short-term basis, most Willow Island (Monongahela Energy) 258.2 of them have been in place for 40–50 years, allowing the fish Gorsuch (American Municipal Power–Ohio) 284.0 community to disperse over the course of the river. For exam- Philip Sporn (American Electric Power) 388.6 ple, White Perch Morone americana did not appear in the Ohio Kyger Creek (Ohio Valley Electric Corp.) 418.3 River until 1992 (ORERP, unpublished field data). However, by J. M. Stuart (Dayton Power and Light) 652.8 2005, White Perch were found riverwide, sometimes in high W. C. Beckjord (Duke Energy) 728.9 numbers (King et al. 2010). Tanners Creek (American Electric Power) 794.9 The large number of power plants participating in the 2005 Clifty Creek (Indiana–Kentucky Electric Corp.) 901.0 ORERP allowed nearly complete longitudinal coverage of the Gallagher (Duke Energy) 981.5 river (Table 1; Figure 1). Sampling began at rkm 82 (upstream Cane Run (E.ON US [Louisville Gas and Electric]) 992.4 of the W. H. Sammis Plant) and extended to rkm 1,527 (down- Coleman (E.ON US [Western Kentucky Energy 1,172.0 stream of the Shawnee Plant). This represents 92% of the length Corp.]) of the river. Sampling near each of the plants was conducted Elmer Smith (Owensboro Municipal Utilities) 1,212.2 seasonally in June, August, and October. Six 500-m-long elec- Shawnee (Tennessee Valley Authority) 1,522.1 trofishing zones were established near each plant. For all plants 542 SEEGERT ET AL.

except the Shawnee Plant, three electrofishing zones were lo- calls for the exclusion of 13 highly tolerant species, all hybrids, cated upstream of a plant’s thermal discharge and three were and all exotic species from the number and weight calculations. located downstream of the thermal discharge. At the Shawnee However, these taxa are included in the diversity index calcula- Plant, two zones were established upstream of the thermal dis- tions that are part of the IWB. charge and four were established downstream of it. Thus, 18 We also used the Ohio River fish index (ORFIn) to assess electrofishing samples were collected near each plant, comfort- community health. The ORFIn is an Ohio River-specific ver- ably above the sample size of 15 that has been recommended to sion of the IBI developed by Emery et al. (2003) and consists of adequately characterize pools in the Ohio River (Blocksom et al. 13 metrics. Because of their irruptive nature (Seegert 2000), Giz- 2009). Although the thermal discharge from each plant might zard Shad Dorosoma cepedianum and Emerald Shiner Notropis have affected how fish were distributed on a local (i.e., plant- atherinoides were excluded from the calculation of the propor- specific) scale, we would not expect any effect on a riverwide tional metrics but were included in the species richness metrics. scale, which was the focus of this study. Furthermore, previous Fish sampling.—Pulsed-DC electrofishing was conducted ORERP studies have shown that even these local effects are along the shoreline at night using a boat-mounted electrofishing typically minor, and when they do occur, it is only during the system powered by a 5,000-W generator, with the output con- summer (August) surveys. Thus, data from areas both upstream trolled by a Coffelt Model VVP-15 electrofisher (or equivalent). and downstream of each plant were included in our assessment. Output settings were maintained at 100 pulses/s, 25–40% pulse Water quality and instream habitat variables.—Water tem- width, 300 V or greater, and 6–8 A. The six sampling zones were perature, dissolved oxygen (DO), and specific conductance were all within 5 km of the plant being sampled. The sampling crew measured by using a YSI Model 85 meter at all fish sampling consisted of one driver and two dipnetters. A 4.76-mm-mesh locations during each survey. Water clarity (i.e., Secchi depth) dip net was used to collect stunned fish. Sampling began at the was measured at each electrofishing location by using a standard upstream end of each 500-m zone and continued in a down- Secchi disk. stream direction. All fish that were stunned during a sampling At each electrofishing location (zone), habitat was assessed run were placed in a holding tub until processed. Consistent with by using the method of ORSANCO (2003), which consists of ORSANCO guidance, electrofishing was conducted when river collecting water depth and substrate data along six evenly spaced stage was within 1 m of normal pool and when Secchi depths transects within each 500-m electrofishing zone. The transects were greater than 300 mm. began at the upstream end of each electrofishing zone and then Fish processing.—Up to 30 specimens of each game fish and were placed at each subsequent 100-m mark, perpendicular to large-bodied species (e.g., suckers, gars, and carps) collected at shore. Water depth, substrate types, and instream cover types each location were measured and weighed individually. If more were recorded at 3-m intervals along each transect, beginning than 30 individuals of such species were collected at a given at the onshore (0-m) starting point and ending at a point up to location, the remaining individuals were counted and batch- 30 m from shore. Thus, within each electrofishing zone, habitat weighed. Prey species (mostly native minnows and darters) were data were collected from 66 points (6 transects × 11 points per counted and batch-weighed. To assess recruitment, age-0 fish transect). For each zone, the average depth, percent composition were noted as such on field data sheets. A gross external exami- of substrate types, and percent instream cover were determined. nation of each fish was performed to determine the incidence of The Ohio River Valley Water Sanitation Commission classifies disease, parasites, or physical abnormalities because this is one sites with at least 14% cobble/boulder as class A (best habi- of the ORFIn metrics. The classification of external anomalies tat); sites with less than 14% cobble/boulder and less than 70% followed Ohio EPA (1989, 1996) guidelines. A voucher col-

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 sand/silt as class B; and sites with at least 70% sand/silt as lection containing representatives of each species collected was class C (worst habitat). The qualitative habitat evaluation index compiled and is maintained at the Deerfield, Illinois, office of (QHEI; Rankin 1989) was also used to evaluate habitat. The EA Engineering, Science, and Technology, Inc. QHEI uses a series of seven metrics to score habitat on a scale Data analysis.—We determined longitudinal trends for 31 of 0–100. The habitat characterization was conducted to ensure widespread, common taxa by using a four-parameter spline in that habitats in the zones sampled upstream and downstream of a generalized additive model (GAM; Wood 2006). The GAM each plant were similar. estimation procedure was introduced in the mid-1980s (Hastie Community measures.—The index of well-being (IWB) was and Tibshirani 1990) as a tool to estimate functions of unknown developed by Gammon (1976) and uses the number, weight, and nonlinear form and is commonly used for spatial analysis of diversity of fish to assess the health of a fish community. We used fisheries data (Swartzman et al. 1992; O’Brien and Rago 1996; a modified version of the IWB (IWBmod); the Ohio Environmen- Stoner et al. 2001). This free-form approach is well suited to the tal Protection Agency (Ohio EPA) applied this modification to task of discovering the nature of longitudinal trends. A GAM make the index more sensitive to a wider array of environmental based on a four-knot spline was sufficient to capture the non- disturbances, particularly those that result in shifts in commu- linearity of longitudinal trends, whereas a greater number of nity composition without large reductions in species richness, knots resulted in excessive nonlinearity of the trend function, numbers, and/or biomass (Ohio EPA 1987). This modification indicating overfitting of the data. While seasonal differences LONGITUDINAL VARIATION IN THE FISH COMMUNITY 543

and thermal effects were not the focus of this analysis, season GAM showed that both longitude and the covariate were signif- and upstream/downstream terms were included in the model icant but the F-value for longitude was reduced by 5.0 or more as categorical variables. Inclusion of these terms increased the relative to the first GAM, then these cases were coded PC to in- power of statistical inference by reducing the error variance dicate that the covariate partially, but not completely, explained of the model fit. The dependent variable, electrofishing catch the longitudinal trend, and the longitudinal trend remained imper effort (CPE; number of fish/km), was log transformed as portant. The value of 5.0 was chosen as a benchmark because an loge(CPE) = log10(CPE + 1). For graphical presentation, the F-value of 5.0 would generally be considered significant for the model predictions were back-transformed to the original CPE error degrees of freedom available in these models. If the second units. To emphasize the longitudinal aspect of the model results, GAM showed that both longitude and the covariate were sig- a single curve was produced to represent the longitudinal trend nificant and the F-value for longitude was reduced by less than for summer. Given the length of the study area (>1,400 km), 5.0 or increased, then the case was coded BI (both important) trends should not differ among the seasons. Thus, the trend to indicate that both longitude and the covariate were important lines for other seasons were constrained to be parallel to the and relatively independent in explaining the dependent variable. trend line for summer. Plotted observations were adjusted for We used covariate analysis to determine whether six physic- the season (Dagum 1978) to the expected observation for the ochemical variables (water temperature, DO, specific conduc- summer season, and similarly plotted observations were ad- tance, Secchi depth, percent silt, and percent cobble/boulder) justed for upstream/downstream effects to the upstream con- explained or contributed to the observed longitudinal patterns dition. For example, the spring data were adjusted by adding for the 31 taxa and the 5 measures of community health. the mean difference between summer and spring curves to each spring observation. The adjustment was made on the logarith- mic scale, and the resulting value was back-transformed for RESULTS plotting. The model constrained the longitudinal pattern in the Longitudinal Variation in Common Taxa logarithm metric to be the same for each season in order to Based on statistical comparisons and graphical results (e.g., reduce the number of model parameters. In addition to assess- Figures 2–4), the distributional patterns of the 31 fish taxa and ment of trends for 31 taxa, we also determined whether any 5 community measures fell into seven categories: (1) similar of the broader measures of community health—namely CPE abundance throughout the river; (2) most abundant in the middle (both with and without irruptive species: Seegert 2000), species segment of the river; (3) noticeably more common in the upper richness, IWB, and ORFIn—varied longitudinally. half of the river; (4) noticeably more common in the upper third To test whether longitudinal trends could be attributed to of the river; (5) exhibiting a nearly linear decline from upstream one of the measured covariates, we used the following process. to downstream; (6) exhibiting a general decline from upstream First, we used a smoothing function to assess whether the data to downstream (i.e., like category 5) but with an increase at rkm for each dependent variable exhibited a longitudinal trend. The 1,522; and (7) an increase from upstream to downstream. smoothing function has the flexibility to identify increasing, decreasing, and nonmonotonic trends. If a longitudinal trend was found, then additional models were fitted using each covariate. The predictive potential of each covariate was assessed relative to the predictive potential of the longitudinal trend smoothing function. A set of rules was implemented to categorize each out-

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 come. The details of the statistical algorithm that was used to characterize each case are as follows. First, a GAM was fitted to each dependent variable by using a four-knot spline function to estimate the longitudinal trend. If the first GAM showed that the longitudinal trend was not significant, then this dependent variable was coded NT (no trend), and no further investigation was done. If the longitudinal trend was significant but none of the covariates was statistically significant, these cases were coded NC (no covariate statistically significant). If the longitudinal trend was significant and one or more of the covariates was statistically significant, then each of the covariates was assessed as follows. A second GAM was fitted that included the spline FIGURE 2. Longitudinal distribution of Logperch electrofishing catch rates smooth for longitudinal trend and the covariate. If the partial (fish/km) in the Ohio River, illustrating a category 4 distribution (i.e., more com- statistic for longitudinal trend adjusted for the covariate was not mon in the upper third of the river). River kilometer zero represents the beginning (upper end) of the Ohio River. The plotted line represents the longitudinal trend significant, then it was inferred that the covariate explained the for the summer season and the upstream condition; data observations were longitudinal trend, and these cases were coded CE. If the second adjusted for seasonal and area effects to match that condition (see Methods). 544 SEEGERT ET AL.

the electrofishing catch) than Gizzard Shad, were also collected at similar rates over the length of the river. Category 2 (most abundant in the middle reach of the river).—Five species (Sauger Sander canadensis, Longear Sun- fish Lepomis megalotis, Longnose Gar Lepisosteus osseus, Skip- jack Herring Alosa chrysochloris, and River Shiner Notropis blennius) were most abundant in the middle river segment (rkm 483–965). For example, Longear Sunfish showed a distinct peak around rkm 800 and were less abundant in both the upper and lower reaches of the river. Category 3 (more common in the upper half of the river).— Five species (White Perch, Spotted Bass Micropterus punctu- latus, Silver Chub Macrhybopsis storeriana, Channel Shiner Notropis wickliffi, and Bluegill Lepomis macrochirus)were much more common in the upper half of the river. For example, FIGURE 3. Longitudinal distribution of Emerald Shiner electrofishing catch catches of Spotted Bass from rkm 82 to rkm 724 were typically rates (fish/km) in the Ohio River, illustrating a category 5 distribution (i.e., 7–8 fish/km, but in the lower half of the river catch rates were consistent decline from upstream to downstream). River kilometer zero repre- only 1–2 fish/km. sents the beginning (upper end) of the Ohio River. The plotted line represents Category 4 (more common in the upper third of the river).— the longitudinal trend for the summer season and the upstream condition; data observations were adjusted for seasonal and area effects to match that condition Five species (Logperch Percina caprodes, Smallmouth Bass (see Methods). Micropterus dolomieu, Bluntnose Minnow Pimephales nota- tus, Northern Hog Sucker Hypentelium nigricans, and Spotfin Shiner Cyprinella spiloptera) were significantly more common Category 1 (no trend).—Three species (Common Carp Cypri- in the upper third of the river (Figure 2). nus carpio, Gizzard Shad, and Largemouth Bass Micropterus Category 5 (consistent decline from upstream to salmoides) exhibited no longitudinal trend. The Gizzard Shad, downstream).—This was the largest of the seven longitudi- which is perhaps the most common fish in the river (Pearson nal distribution categories and included 10 taxa: the Golden and Krumholz 1984; Pearson and Pearson 1989), comprised Redhorse Moxostoma erythrurum, Emerald Shiner, Quillback 40% of our electrofishing catch and was most commonly col- Carpiodes cyprinus, Smallmouth Redhorse Moxostoma bre- lected at CPEs of about 100 fish/km throughout the length of viceps, Flathead Catfish Pylodictis olivaris, Freshwater Drum the river. Common Carp, though much less abundant (0.2% of Aplodinotus grunniens, Smallmouth Buffalo Ictiobus bubalus, White Bass Morone chrysops, hybrid striper (hybrid Morone spp.), and Silver Redhorse Moxostoma anisurum. For example, Emerald Shiner declined from being very common (CPE of about 100 fish/km) in the upper river to common (30–50 fish/km) in the middle river and then moderately common (10– 20 fish/km) in the lower river (Figure 3). Category 6 (like category 5 but with an increase near

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 rkm 1,522).—This category included Channel Catfish Ictalurus punctatus and all community measures (Figure 4). Channel Cat- fish catch rates were 5–15 fish/km upstream of rkm 400, declined to a minimum of 1–2 fish/km at rkm 1,200, and then increased to about 3 fish/km at rkm 1,522. The community measures all peaked at the upstream end of the study area, reached their min- imums at rkm 1,200, and rebounded at rkm 1,522 (Figure 4). Category 7 (increase from upstream to downstream).—This category included only two species: Striped Bass Morone saxatilis and River Carpsucker Carpiodes carpio. River Carpsucker FIGURE 4. Longitudinal distribution of species richness in the Ohio River, catch rates were about 5 fish/km at rkm 1,522 but only about 2 illustrating a category 6 distribution (i.e., general decline from upstream to fish/km over most of the rest of the river. downstream but with an increase near rkm 1,522). River kilometer zero repre- Uncommon and local taxa.—In addition to comparing the sents the beginning (upper end) of the Ohio River. The plotted line represents abundance of 31 common taxa along the length of the river, the longitudinal trend for the summer season and the upstream condition; data observations are adjusted for seasonal and area effects to match that condition we also found that several species had restricted distributions, (see Methods). typically at the upper or lower end of the river. Exclusive of LONGITUDINAL VARIATION IN THE FISH COMMUNITY 545

species that have been reported only once or twice as occurring portant but did not explain the longitudinal pattern (Table 2). in the river (Pearson and Krumholz 1984), whose presence in Conversely, in only 3 of the 216 cases did the covariate by itself any area might be due to chance alone, several species were explain the longitudinal pattern. In 19 cases (9%), the covariate restricted to the lower river. The Mississippi Silvery Minnow was statistically significant but explained only a portion of the Hybognathus nuchalis, Threadfin Shad Dorosoma petenense, longitudinal trend. With regard to these 19 cases, DO, Secchi Yellow Bass Morone mississippiensis, and Inland Silverside depth, percent cobble/boulder, and specific conductance only Menidia beryllina were common to abundant near the Shawnee were important in explaining a portion of the pattern in one to Plant at rkm 1,522 but were rare or absent upriver. The Spotted three cases (1%) each. Water temperature was important in 5 of Gar Lepisosteus oculatus,BowfinAmia calva, Goldeye Hiodon the 216 cases (2% of the total cases), and percent silt was impor- alosoides, three Asian carp species (Grass Carp Ctenopharyn- tant in six cases (3%). Thus, in most cases (90%), none of the six godon idella, Silver Carp Hypophthalmichthys molitrix, and physicochemical variables explained the observed longitudinal Bighead Carp Hypophthalmichthys nobilis), and Red Shiner pattern. In 9% of the cases (19 of 216), one of the six dependent Cyprinella lutrensis were also restricted to the lower river variables explained some but not all of the longitudinal pattern, but were uncommon to rare during 2005. The Blue Catfish and in only three cases (1%) did one of the physicochemical Ictalurus furcatus and Shortnose Gar Lepisosteus platostomus variables fully explain the observed longitudinal pattern. were fairly common near rkm 1,522, rare to common in the lower half of the river, and essentially absent from the upper half of the river (EPRI 2007, 2008; and present data). DISCUSSION Of the rare to uncommon species that were restricted to or The longitudinal patterns observed were consistent with noticeably more common in the upper third of the river, all those reported by others (Pearson and Krumholz 1984; Reash were cyprinids and percids. Cyprinids that were restricted to and Van Hassel 1988; Van Hassel et al. 1988). Pearson and the upper river included the Spottail Shiner Notropis hudso- Krumholz (1984) provided the most extensive previous discus- nius and Rosyface Shiner Notropis rubellus. Percids that were sion regarding the zoogeography of Ohio River fishes. They restricted to the upper river or nearly so were the Greenside found that the fish community of the lower Ohio River is sim- Darter Etheostoma blennioides, Fantail Darter Etheostoma fla- ilar to that of the adjacent Mississippi River. In addition to the bellare, Banded Darter Etheostoma zonale, Bluebreast Darter species with restricted ranges previously discussed, Pearson and Etheostoma camurum, Channel Darter Percina copelandi, and Krumholz (1984) indicated that the Sicklefin Chub Macrhy- Yellow Perch Perca flavescens. bopsis meeki, Silverband Shiner Notropis shumardi, Pugnose Minnow Opsopoeodus emiliae, Blacktail Shiner Cyprinella Influence of Six Physicochemical Variables venusta, Flier Centrarchus macropterus, and Bluntnose Darter Based on covariate analyses, the six physicochemical vari- Etheostoma chlorosoma have been found only in the lower por- ables we measured explained little of the longitudinal patterns tion of the river. Sicklefin Chub have never been collected dur- documented herein. Of the 216 cases (6 water quality variables ing the 40 years of ORERP sampling; the other five species × 36 longitudinal patterns [31 fish taxa and 5 community health have all been collected by the ORERP within the past 10 years measures]) considered, six (3%) showed no longitudinal trend. but only rarely (EPRI 2007, 2008; King et al. 2010). Although In 51% of the 216 cases, there was no statistically significant re- Pearson and Krumholz (1984) considered the riverwide distribu- lationship between the longitudinal trends we documented and tion of Ohio River fishes, their assessment was based solely on any of the independent variables (Table 2). In 36% of the cases, presence/absence data. This approach is reasonable for species

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 one of the physicochemical covariates was independently im- whose distribution is restricted to one end of the river or the

TABLE 2. Summary of the inﬂuence of six physicochemical variables on the observed distributional patterns of 31 ﬁsh taxa and 5 community-level measures in the Ohio River.

Number of combinations Percent represented Category (code) in each category by each category No longitudinal trend (NT) 6 3 Longitudinal trend is present, but no covariate is statistically signiﬁcant (NC) 110 51 Both the covariate and the longitudinal trend are independently important 78 36 (i.e., covariate does not explain the longitudinal pattern) (BI) Covariate explains the longitudinal trend (CE) 3 1 Both the longitudinal trend and the covariate are important, and the covariate 19 9 partially explains the longitudinal trend (PC) Total 216 100 546 SEEGERT ET AL.

other. However, it does not reveal possible clinal differences in and Red Shiner) are thermally tolerant or have primarily south- abundance over the course of the river for the 31 widely dis- ern distributions (Talmage and Opresko 1981; Page and Burr tributed species that were quantitatively assessed in this study. 1991). Our categories 4 and 5 contained several thermally sen- Except in the extreme lower river, where the presence of sitive species (e.g., Golden Redhorse, Smallmouth Redhorse, several fishes can be explained by the close proximity of the Logperch, and Northern Hog Sucker). Thus, it is reasonable to Mississippi River, the 2005 longitudinal patterns appear to be conclude that the clinal, natural change in water temperature be- based on a combination of physicochemical variables and habi- tween the upper and lower river contributed to the longitudinal tat conditions. Reash and Van Hassel (1988) and Van Hassel patterns observed in 2005, at least for these categories. et al. (1988) compared fish distributions between the upper and In addition to differences in water temperature, there are middle reaches of the Ohio River. They found that species prefer- habitat differences between the upper and lower river that likely ring smaller, free-flowing streams dominated upper Ohio River account for the differences in distribution of the aforementioned sampling sites, whereas species typically found in slow-moving species. The upper river contains proportionally more hard sub- waters dominated at middle river sites. This finding is consistent strates (i.e., gravel, rubble, and cobble), has more habitat di- with our category 4, which was dominated by small-river forms versity (e.g., islands and point bars), and in general is more like the Logperch and Northern Hog Sucker. Reash and Van riverine in nature relative to the lower river (Trautman 1981; Hassel (1988) and Van Hassel et al. (1988) also reported that Pearson and Krumholz 1984; ORERP, unpublished field data). proximity to tributary streams was important. Those authors did The percids and round-bodied suckers that were proportionately not find optimal temperature to be an important factor affecting more abundant in the upper river require hard substrates for var- the distribution of fishes in the upper and middle Ohio River. ious life history needs, such as feeding, escape from predators, Because they considered only the upper half of the river, they and spawning (Trautman 1981; Page 1983). Most likely, these could not assess the distributional pattern of fishes over the entire groups of species are particularly limited by spawning substrate, length of the river as was done in our assessment. Furthermore, as most are simple lithophils that need clean, medium to large the latitude change in the portion of the river they considered is hard substrates in order to spawn successfully (Balon 1975; less than the latitude change that occurs in the lower half of the Ohio EPA 1989). The fact that such substrates are rare or ab- river. Field data collected during the ORERP have consistently sent in most of the lower Ohio River suggests that this group of shown that temperatures vary by 1◦C to as much as 3◦C between species would not be successful in the lower river. the extreme upper and lower ends of the river, particularly dur- Although habitat differences on a riverwide scale (as de- ing the spring and fall (e.g., see Figure 5). The groups that are scribed above) likely affected fish distribution, overall habitat disproportionately abundant in the upper river—that is, percids quality on a local scale does not appear to have been a factor. (except Saugers), round-bodied suckers (especially redhorses), At the upper 14 power plants, habitat quality as measured by and certain minnows—are all near the “sensitive” end of the ORSANCO habitat class was rather variable (mean = 1.0–2.1; temperature tolerance scale (Talmage and Opresko 1981; Reash Table 3) but showed no longitudinal pattern. The uniformly low et al. 2000; Seegert 2000; Yoder et al. 2006). Conversely, many scores at the lower four plants are attributable to the lack of hard of the species that are restricted to or most abundant in the lower substrate and the more depositional nature of the lower river as Ohio River (e.g., Shortnose Gar, Blue Catfish, Threadfin Shad, discussed previously. Poorer habitat quality in the lower river helps to explain the patterns seen for the 10 common taxa in our category 5 but does not explain why all of the community measures exhibited an improvement near the Shawnee Plant (i.e.,

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 category 6). Although the six physicochemical variables we considered usually did not explain the observed longitudinal patterns, several of the catch measures responded in an expected fashion to two of these variables: percent cobble/boulder and percent silt. In almost half (44%) of the 36 cases, taxon-specific catch rates and community health measures were significantly and directly related to percent cobble/boulder. Conversely, 44% of the 36 measures were significantly but inversely related to percent silt. In most cases, the same measures were affected but in opposite directions. For example, IWBmod scores, ORFIn scores, and species richness were all directly related to percent cobble/boulder and were inversely related to percent silt. Catch rates of many species responded similarly: the CPEs of FIGURE 5. Water temperature in relation to river kilometer on the Ohio River Logperch, Northern Hog Suckers, Smallmouth Bass, Small- during the spring, summer, and fall of 2005. mouth Redhorses, Smallmouth Buffalo, and Spotted Bass were LONGITUDINAL VARIATION IN THE FISH COMMUNITY 547

TABLE 3. Mean habitat scores for sampling locations near 17 power plants on cluded catch rate and species richness, two metrics that Emery the Ohio River, 2005. Power plants are presented in order from upstream (W. H. et al. (2003) also reported as varying inversely with river kilome- Sammis) to downstream (Shawnee). Qualitative habitat evaluation index (QHEI) means are the means of the actual QHEI scores calculated for six locations near ter. Pearson and Krumholtz (1984) noted that gravel substrates each plant (QHEI scores range from 0 to 100; scores > 60 indicate good habitat, are more common in the upper river, which likely explains the scores of 45–60 indicate fair habitat, and scores < 45 indicate poor habitat). greater abundance of redhorses in the upper river, as this group Ohio River Valley Water Sanitation Commission (ORSANCO) habitat classes needs clean gravel/cobble substrates for successful spawning. were assigned the following values: class A = 3 (best), class B = 2, and class The results of our covariate analysis confirmed that site- C = 1 (worst). The resultant numerical values were then averaged. specific differences in physicochemical variables explained lit- Plant ORSANCO habitat class QHEI tle of the longitudinal differences we observed. Thus, we believe that the Ohio River fish community is responding to broad W. H. Sammis 2.0 50.2 changes in substrate composition, water temperature, and river Cardinal 1.5 41.3 size that occur gradually from the upper portion to the lower por- R. E. Burger 1.3 42.0 tion of the river. The noticeable difference in the fish community Kammer 2.0 45.0 at our downstream-most site was also affected by the site’s close Willow Island 1.0 36.1 proximity to the Mississippi River and its faunal assemblage. Gorsuch 1.9 40.0 One of our goals was to determine whether the abundance Philip Sporn 1.5 41.1 of species, taxa, or selected community measures varied over Kyger Creek 1.9 45.3 the course of the Ohio River in a consistent manner. We were J. M. Stuart 1.3 40.9 able to show that a large majority of the species and measures W. C. Beckjord 2.0 45.1 we examined followed predictable patterns. Given our findings, Tanners Creek 1.8 48.0 we believe that resource agencies should consider these patterns Clifty Creek 1.9 45.9 when determining expectations for any such species. Similarly, Gallagher 2.1 48.5 those wanting to develop an IBI that covers all or a substan- Cane Run 1.0 39.5 tial portion of the river should account for these differences in Coleman 1.0 39.3 abundance (and sometimes even presence/absence) when set- Elmer Smith 1.0 43.3 ting their metric expectations. Shawnee 1.0 42.0

ACKNOWLEDGMENTS all directly related to percent cobble/boulder but were inversely This study was conducted under the direction of the Elec- related to percent silt. Thus, at the local level, the responses tric Power Research Institute project manager, Doug Dixon; we of catch measures to changes in substrate quality were as one thank him for his support and guidance throughout the project. might expect, but the longitudinal differences we documented We also thank the following companies, both for their ﬁnancial were not explained by changes in substrate over the course of support and for coordinating sampling at their respective facil- the river or by any of the other four physicochemical variables ities: Allegheny Energy (now FirstEnergy); American Electric we measured. Power Company; American Municipal Power–Ohio; Buckeye Longitudinal differences in community structure were also Power, Inc.; Dayton Power and Light; Duke Energy; E.ON U.S. reported by Emery et al. (2003), who found that 8 of the 13 (Louisville Gas and Electric and Western Kentucky Energy);

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 metrics they developed for the ORFIn varied according to river FirstEnergy; Indiana–Kentucky Electric Corp.; Ohio Valley kilometer. They reported that the number of total species, num- Electric Corp.; Owensboro Municipal Utilities; and Tennessee ber of sucker species, and number of intolerant species all varied Valley Authority. We acknowledge the numerous field personnel inversely with river kilometer (i.e., numbers of species in these who collected and processed over 600 samples in 2005. Lastly, groups decreased from upstream to downstream). They also we are grateful to an anonymous reviewer and Richard Eades found inverse relationships between river kilometer and overall for their comments on a previous draft of this paper. catch rate as well as inverse relationships with the percentages of simple lithophils, detritivores, invertivores, and piscivores. With REFERENCES regard to the number of sucker species, Emery et al. (2003) Balon, E. K. 1975. Reproductive guilds of fishes: a proposal and definition. noted that the number of round-bodied suckers (e.g., Moxos- Journal of the Fisheries Research Board of Canada 32:821–864. toma spp.) decreased in the lower river, whereas the number of Berry, C. R. Jr., M. Wildhaber, and D. L. Galat. 2004. Population structure deep-bodied sucker species (i.e., Carpiodes spp. and Ictiobus and habitat use of benthic fishes along the Missouri and lower Yellowstone spp.) increased. Although Emery et al. (2003) looked primarily rivers, volume 3: fish distribution and abundance. U.S. Geological Survey, Cooperative Research Units, South Dakota State University, Brookings. at groups of species, the longitudinal patterns they saw for these Blocksom, K., E. Emery, and J. Thomas. 2009. Sampling effort needed to esti- groups are consistent with our findings for many species (i.e., mate condition and species richness in the Ohio River, USA. Environmental category 5, our largest category) and for category 6, which in- Monitoring and Assessment 155:157–167. 548 SEEGERT ET AL.

Burr, B. M., and L. M. Page. 1986. Zoogeography of fishes of the lower Ohio– Pearson, M. S., T. R. Angradi, D. W. Bolgrien, T. M. Jicha, D. L. Taylor, M. upper Mississippi basin. Pages 287–324 in C. H. Hocutt and E. O. Wiley, F. Moffett, and B. H. Hill. 2011. Multimetric fish indices for midcontinent editors. The zoogeography of North American freshwater fishes. Wiley, New (USA) great rivers. Transactions of the American Fisheries Society 140: York. 1547–1564. Dagum, E. B. 1978. Modelling, forecasting and seasonally adjusting economic Pearson, W. D., and L. A. Krumholz. 1984. Distribution and status of Ohio River time series with the X-11 ARIMA method. Statistician 27:203–216. fishes. Oak Ridge National Laboratory, Publication ORNL/Sub/79-7831/1, Emery, E. B., T. P. Simon, F. H. McCormick, P. L. Angermeier, J. E. Deshon, Oak Ridge, Tennessee. C. O. Yoder, R. E. Sanders, W. D. Pearson, G. D. Hickman, R. J. Reash, Pearson, W. D., and B. J. Pearson. 1989. Fishes of the Ohio River. Ohio Journal and J. A. Thomas. 2003. Development of a multimetric index for assessing of Science 89:181–187. the biological condition of the Ohio River. Transactions of the American Rafinesque, C. S. 1820. Ichthyologia Ohiensis, or natural history of the fishes Fisheries Society 132:791–808. inhabiting the River Ohio and its tributary streams, preceded by phys- EPRI (Electric Power Research Institute). 2007. Ohio River ecological re- ical description of the Ohio and its branches. W. G. Hunt, Lexington, search program (ORERP): 2005 Ohio River monitoring results. EPRI, Report Kentucky. 1015422, Palo Alto, California. Rankin, E. T. 1989. The qualitative habitat evaluation index (QHEI): rationale, EPRI (Electric Power Research Institute). 2008. Ohio River ecological re- methods, and application. Ohio Environmental Protection Agency, Division search program (ORERP): 2006 Ohio River monitoring results. EPRI, Report of Water Quality Planning and Assessment, Ecological Assessment Section, 1017002, Palo Alto, California. Columbus. EPRI (Electric Power Research Institute). 2010. Ohio River ecological re- Reash, R. J., G. L. Seegert, and W. L. Goodfellow. 2000. Experimentally de- search program (ORERP): 2008 Ohio River monitoring results. EPRI, Report rived upper thermal tolerances for Redhorse Suckers: revised 316(a) variance 1021540, Palo Alto, California. conditions at two generating facilities in Ohio. Environmental Science and EPRI (Electric Power Research Institute). 2012. Ohio River ecological re- Policy 3(Supplement 1):191–196. search program (ORERP): 2009 Ohio River monitoring results. EPRI, Report Reash, R. J., and J. H. Van Hassel. 1988. Distribution of upper and middle Ohio 1024821, Palo Alto, California. River fishes, 1973–1985: II. influence of zoogeographic and physicochemical ESE (Environmental Science and Engineering). 1991. Ohio River ecological tolerance factors. Journal of Freshwater Ecology 4:459–476. research program: analysis of long-term larval fish data. ESE, St. Louis, Seegert, G. 2000. Considerations regarding development of index of biotic Missouri. integrity metrics for large rivers. Environmental Science and Policy 3(Sup- Gammon, J. R. 1976. The fish populations of the middle 340 km of the Wabash plement 1):99–106. River. Water Resources Research Center, Purdue University, Technical Report Simon, T. P., and J. Lyons. 1995. Application of the index of biotic integrity to 86, West Lafayette, Indiana. evaluate water resource integrity in freshwater ecosystems. Pages 245–262 in Hastie, T. J., and R. J. Tibshirani. 1990. Generalized additive models. Chapman W. S. Davis and T. P. Simon, editors. Biological assessment and criteria: tools and Hall, London. for water resource planning and decision making. CRC Press, Boca Raton, Hughes, R. M., and J. R. Gammon. 1987. Longitudinal changes in fish assem- Florida. blages and water quality in the Willamette River, Oregon. Transactions of the Stoner, A. W., J. P.Manderson, and J. P.Pessutti. 2001. Spatially explicit analysis American Fisheries Society 116:196–209. of estuarine habitat for juvenile Winter Flounder: combining generalized ad- King, R. G., G. Seegert, J. Vondruska,E. S. Perry, and D. A. Dixon. 2010. Factors ditive models and geographic information systems. Marine Ecology Progress influencing impingement at 15 Ohio River power plants. North American Series 213:253–271. Journal of Fisheries Management 30:1149–1175. Swartzman, G., C. Huang, and S. Kaluzny. 1992. Spatial analysis of Bering Sea O’Brien, L., and P. Rago. 1996. An application of the generalized additive groundfish survey data using generalized additive models. Canadian Journal model to groundfish survey data with Atlantic Cod off the northeast coast of Fisheries and Aquatic Sciences 49:1366–1378. of the United States as an example. NAFO (Northwest Atlantic Fisheries Talmage, S. S., and D. M. Opresko. 1981. Literature review: response of fish Organization) Scientific Council Studies 28:79–95. to thermal discharges. Electric Power Research Institute, EPRI EA-1840, Ohio EPA (Environmental Protection Agency). 1987. Biological criteria for Project 877, Palo Alto, California. the protection of aquatic life, volume II: users manual for biological field Trautman, M. B. 1981. The fishes of Ohio, 2nd edition. Ohio State University assessment of Ohio surface waters, updated January 1, 1988. Ohio EPA, Press, Columbus. Division of Water Quality Planning and Assessment, Columbus. Van Hassel, J. H., R. J. Reash, H. W. Brown, J. L. Thomas, and R. C. Mathews Jr. Ohio EPA (Environmental Protection Agency). 1989. Biological criteria for 1988. Distribution of upper and middle Ohio River fishes, 1973–1985: I. as- Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 the protection of aquatic life, volume III: standardized biological field sam- sociations with water quality and ecological variables. Journal of Freshwater pling and laboratory methods for assessing fish and macroinvertebrate com- Ecology 4:441–458. munities. Ohio EPA, Division of Water Quality Planning and Assessment, Wood, S. N. 2006. Generalized additive models: an introduction with R. Chap- Columbus. man and Hall/CRC, Boca Raton, Florida. Ohio EPA (Environmental Protection Agency). 1996. Ohio EPA’s guide to DELT Woods, A. J., J. M. Omernik, C. S. Brockman, T. D. Gerber, W. D. Hosteter, anomalies. Ohio EPA, Working Copy DSW-MAS-EAU, Columbus. and S. H. Azevedo. 1998. Ecoregions of Indiana and Ohio (2-sided color Omernik, J. M., and A. L. Gallant. 1988. Ecoregions of the upper Midwest states. poster with map, descriptive text, summary tables, and photographs). U.S. U.S. Environmental Protection Agency, Environmental Research Laboratory, Geological Survey, Reston, Virginia. EPA/600/3-88/037, Corvallis, Oregon. Woods, A. J., J. M. Omernik, W. H. Martin, G. J. Pond, W. M. Andrews, S. ORSANCO (Ohio River Valley Water Sanitation Commission). 2003. Stan- M. Call, J. A. Comstock, and D. D. Taylor. 2002. Ecoregions of Kentucky dard operating procedures for the habitat data collection for fish population (2-sided color poster with map, descriptive text, summary tables, and pho- surveys. ORSANCO, Cincinnati. tographs). U.S. Geological Survey, Reston, Virginia. Page, L. M. 1983. Handbook of darters. T.F.H. Publications, Neptune City, New Yoder, C. O., B. J. Armitage, and E. T. Rankin. 2006. Re-evaluation of the tech- Jersey. nical justification for the existing Ohio River mainstem temperature criteria: Page, L. M., and B. M. Burr. 1991. A field guide to freshwater fishes of North report to the Ohio River Valley Water Sanitation Commission (ORSANCO) America north of Mexico: Peterson field guide series 42. Houghton Mifflin, ad hoc committee on temperature criteria re-evaluation. MBI (Midwest Bio- Boston. diversity Institute), Technical Report MBI/05-05-2, Columbus, Ohio. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:06 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Experimental Examination of Surgical Procedures for Implanting Sonic Transmitters in Juvenile Shortnose Sturgeon and Atlantic Sturgeon James A. Crossman a d , K. Larry Hammell b & Matthew K. Litvak c a Department of Biology and Centre for Coastal Studies and Aquaculture , University of New Brunswick , Post Office Box 5050, Saint John , New Brunswick , E2L 4L5 , Canada b Department of Health Management and Centre for Veterinary Epidemiological Research, Atlantic Veterinary College , University of Prince Edward Island , 550 University Avenue, Charlottetown , Prince Edward Island , C1A 4P3 , Canada c Department of Biology , Mount Allison University , 63B York Street, Sackville , New Brunswick , E4L 1G7 , Canada d BC Hydro , 601 18th Street, Castlegar , British Columbia , V1N 2N1 , Canada Published online: 16 May 2013.

To cite this article: James A. Crossman , K. Larry Hammell & Matthew K. Litvak (2013): Experimental Examination of Surgical Procedures for Implanting Sonic Transmitters in Juvenile Shortnose Sturgeon and Atlantic Sturgeon, North American Journal of Fisheries Management, 33:3, 549-556 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785994

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:549–556, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785994

ARTICLE

Experimental Examination of Surgical Procedures for Implanting Sonic Transmitters in Juvenile Shortnose Sturgeon and Atlantic Sturgeon

James A. Crossman*1 Department of Biology and Centre for Coastal Studies and Aquaculture, University of New Brunswick, Post Ofﬁce Box 5050, Saint John, New Brunswick E2L 4L5, Canada K. Larry Hammell Department of Health Management and Centre for Veterinary Epidemiological Research, Atlantic Veterinary College, University of Prince Edward Island, 550 University Avenue, Charlottetown, Prince Edward Island C1A 4P3, Canada Matthew K. Litvak Department of Biology, Mount Allison University, 63B York Street, Sackville, New Brunswick E4L 1G7, Canada

Abstract Acoustic telemetry has become a leading tool for monitoring the movements of and habitat use by many sturgeon species worldwide; however, procedures for internal tagging of small juvenile sturgeon (<55 cm TL) are lacking. We examined effects of implantation technique on growth, tag retention, and survival of juvenile (<55 cm) Shortnose Sturgeon Acipenser brevirostrum and Atlantic Sturgeon A. oxyrinchus by using dummy acoustic tags. Two implantation techniques were used: (1) anchoring the tag to the wall of the peritoneal cavity and (2) no internal anchoring of the tag. These treatment groups were compared with two control groups: fish that received anesthetic only and fish that received anesthetic and an incision. Retention rate was significantly higher for anchored tags (88%) than for nonanchored tags (25%) in Shortnose Sturgeon juveniles, while Atlantic Sturgeon juveniles retained 100% of tags regardless of treatment. Nonanchored tags that were lost during the second week were expelled through the incision site, whereas later tag expulsions (during weeks 7 and 8) occurred through the anus. There was no significant difference in absolute growth and specific growth rates between the treatment groups throughout the 8-week study for either species. Growth of both species was significantly lower in the first week after surgery but increased and remained Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 constant for the remainder of the experiment. Use of nonanchored tags significantly increased incision healing times for both species; however, Atlantic Sturgeon healed significantly faster (35 d) than Shortnose Sturgeon (42 d). No mortality occurred in any of the treatment groups. Results of this study suggest that juvenile Shortnose Sturgeon and Atlantic Sturgeon can undergo internal tag implantation resulting in long retention times with largely unaffected growth and no mortality.

Many river systems along the eastern coast of North Amer- Horgan 2002; COSEWIC 2011). Both species have dramati- ica historically supported populations of Shortnose Sturgeon cally declined in numbers due to habitat degradation, pollution, Acipenser brevirostrum and Atlantic Sturgeon A. oxyrinchus and overﬁshing (Gross et al. 2002; Smith et al. 2002). Many of (Dadswell et al. 1984; Smith and Clugston 1997; Kynard and these populations are now extirpated. The Shortnose Sturgeon

*Corresponding author: [email protected] 1Present address: BC Hydro, 601 18th Street, Castlegar, British Columbia V1N 2N1, Canada. Received May 21, 2012; accepted March 6, 2013 549 550 CROSSMAN ET AL.

is protected throughout its range; it has been a species of special were unsuccessful when used with juvenile Shortnose Sturgeon concern in Canada (COSEWIC 2011) since 1980, was listed by and Atlantic Sturgeon smaller than 55 cm TL, and low retention the International Union for Conservation of Nature (IUCN) as rates were likely due to tag movement within the peritoneal cav- vulnerable in 1984, and has been listed as endangered in the ity (M. K. Litvak, unpublished data). Schramm and Black (1984) USA since 1967. The Atlantic Sturgeon sustained an extensive suggested the use of internal suturing as a form of attachment commercial fishery until dramatic declines led to a moratorium to prevent tag movement; we set out to evaluate this method. in the USA in 1999 (Collins et al. 2000; Secor et al. 2000; The objective of this study was to test two different proce- Savoy and Pacileo 2003). The Atlantic Sturgeon is currently dures for the surgical implantation of sonic tags into the ab- listed by the IUCN as near threatened and is listed as threatened domens of juvenile Shortnose Sturgeon and Atlantic Sturgeon: in Canada (COSEWIC 2011); five distinct population segments (1) anchoring the tag to the wall of the peritoneal cavity and (2) are currently listed (April 2012) as endangered or threatened tag implantation without the use of internal anchoring. We quan- under the U.S. Endangered Species Act. tified the effects of surgical implantation technique on survival, Telemetry studies have been successfully used to describe tag retention, incision healing, and growth. the distribution of and habitat use by adults of the Shortnose Sturgeon (Buckley and Kynard 1985; Moser and Ross 1995; Collins et al. 2003; Li et al. 2007; Kynard et al. 2009; Usvyatsov METHODS et al. 2012) and Atlantic Sturgeon (Kynard et al. 2000; Hatin Experimental fish.—The juvenile Shortnose Sturgeon and At- et al. 2002; Fernandes et al. 2010). Several telemetry studies lantic Sturgeon used in this study were reared from eggs at the have focused on the juvenile stage (Hall et al. 1991; Kieffer and University of New Brunswick, Saint John. Fish were held in Kynard 1993; Trested et al. 2011), but despite such work, little circular holding tanks that were part of a partial recirculating is known about juvenile distribution and activity patterns for (1,000-L) system; tanks were 70 cm deep × 40 cm in diameter these species. This lack of knowledge is particularly important (volume = 90 L; water turnover rate of twice per hour). Tank in light of findings that for both Atlantic Sturgeon and Shortnose temperature remained constant at 13.0 ± 0.2◦C for the entire Sturgeon, the juvenile stage is the most sensitive to perturbation experiment. Photoperiod was set at 16 h light : 8 h dark. Each (Gross et al. 2002). Thus, more information on the distribution fish was tagged with a passive integrated transponder (PIT) tag and requirements of this life history stage is needed to further our (AVID) 3 weeks prior to surgery to allow for individual identi- understanding of recruitment mechanisms (Gross et al. 2002; fication. The PIT tags were inserted with a 16-gauge needle just Secor et al. 2002) and to develop conservation and management below the skin along the dorsal midline from the posterior edge plans for both species. of the third dorsal scute to the anterior edge of the fourth dorsal The success of telemetry studies requires adequate trans- scute (Moser et al. 2000). mitter retention. Transmitter attachment on adult sturgeon has Experimental design.—Fish of the two species were kept in been studied, but little work has been conducted on juveniles separate tanks due to findings by Giberson (2004) that juve- smaller than 55 cm TL (Moser et al. 2000; Collins et al. 2002). nile Shortnose Sturgeon exhibit suppressed growth rates when External attachment is acceptable for short-term studies in envi- in the presence of juvenile Atlantic Sturgeon. One month prior ronments that lack high velocities or physical obstructions; use to the experiment, 32 juveniles of each species were randomly of externally attached transmitters depends greatly upon the as- assigned to one of four holding tanks for purposes of acclima- sumption that the transmitter does not negatively affect normal tion. Fish were then randomly assigned to one of four treatment physiology or behavior (Bridger and Booth 2003). Counihan and groups, which were dispersed evenly among all tanks to ac-

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 Frost (1999) showed that juvenile White Sturgeon A. transmon- count for any possible tank effects. Treatments were applied in tanus with external transmitters exhibited significantly reduced random order to minimize bias. Treatments consisted of (1) ex- swimming speeds. Sutton and Benson (2003) found that exter- posure to anesthetic only (control), (2) anesthetic exposure and nal transmitters exceeding 1.25% of fish body weight resulted application of an incision but no tag, (3) anesthetic exposure and in significantly lower tag retention and growth rates in juve- implantation of a nonanchored tag, and (4) anesthetic exposure nile Lake Sturgeon A. fulvescens relative to control groups. In and implantation of a tag anchored to the peritoneal cavity wall. Shortnose Sturgeon and Atlantic Sturgeon, proper implantation The experiment was a completely crossed design with two fish of internal sonic transmitters provides a retention rate much of each treatment type placed in each tank and four replicates higher than that of externally attached tags (Moser et al. 2000). of each tank per species. Transmitters (∼16 × 110 mm) dipped in a biologically inert Dummy tags.—Dummy tags were fabricated by Vemco compound were found to have high internal retention in juvenile (Amirix Ltd.) to replicate one of the smaller transmitters avail- Atlantic Sturgeon and adult Shortnose Sturgeon ranging from able from that company at the time of the study (i.e., Model 55 to 100 cm TL (B. Kynard, BK-Riverfish, personal commu- V8SC-2 L-R256). Tags were cylindrical, with rounded ends nication). Furthermore, Collins et al. (2002) found that ventral (2.7 ± 0.01 cm long; 0.9 ± 0.001 cm in diameter). We added insertion of free-floating radio transmitters into adult Shortnose 0.5 cm of 48-h epoxy resin to one end of each tag used in Sturgeon yielded the best retention. However, these techniques the treatments. Once the epoxy hardened, a 0.16-cm hole was TRANSMITTER IMPLANTATION IN JUVENILE STURGEON 551

drilled through the epoxy to accommodate the 2–0-gauge su- with a reference scale after surgery to allow quantification of ture. The tags had a mass of 4.63 ± 0.02 g (mean ± SE) in incision length by using image analysis (ImageJ version 1.19z; air and 2.83 ± 0.027 g in water. These dimensions were well freeware). within the generally accepted rule that tags should not exceed Postsurgical examination.—Fish were monitored every 2% of fish body weight (1.5 ± 0.3% of the body weight of the 30 min for the first 8 h after surgery and then twice daily for 8 experimental fish; Winter 1996). weeks. All fish were digitally imaged and measured for TL (cm), Surgical treatments and procedures.—Food was withheld FL (cm), and weight (g) weekly. Fish were fed a commercially from all experimental fish for 36 h prior to surgery. Fish were prepared diet (Corey Hi-Pro 3.0 pellets) at 1.5% of body weight individually anesthetized with tricaine methanesulfonate (MS- per day. The feed ration was adjusted based on weekly weight 222; 125 mg of MS-222/L of water, following Summerfelt and measurements. Excess food was siphoned daily, and tanks were Smith 1990) in an aerated container containing 13◦C water. checked for lost tags. Daily visual assessments of recovery were Anesthetic exposure was monitored until opercular rates de- recorded and included observations of the fish’s activity level creased and the individual lost reactivity to external stimuli and feeding. Growth rates (weekly) and tag retention (daily) (Summerfelt and Smith 1990). Average time to sedation was were followed as the experiment progressed. For individuals 8 min. Fish were then measured (mm FL and TL) and weighed that lost transmitters during the course of the experiment, the (g). Prior to surgery, the mean mass ± SD of juvenile Short- expulsion pathways were noted and measurements on these fish nose Sturgeon was 318.1 ± 12.7 g and that of juvenile Atlantic were continued until the end of the 8 weeks. Wounds resulting Sturgeon was 334.8 ± 11.9 g. For Shortnose Sturgeon, ini- from transmitter expulsion were left to heal. Recording of mea- tial TL was 41.7 ± 0.6 cm (mean ± SD) and initial FL was surements was terminated after 8 weeks; however, transmitters 35.7 ± 0.5 cm. For Atlantic Sturgeon, initial TL was 44.6 ± remained in the fish for a total of 6 months to mimic the battery 0.5 cm (mean ± SD) and initial FL was 37.0 ± 0.4 cm. After life of the sonic transmitters used in field studies. Tag retention measurements were recorded, fish were placed on their backs was monitored daily throughout this 6-month period. in a V-shaped, portable surgical table that was constructed via Data analysis.—Statistical analyses were completed by using the methods described by LaVigne (2002). For the duration of the Statistical Analysis Systems (SAS) version 8.2. Type I error the surgery, the gills were continuously perfused with filtered was set at 0.05. All data were tested for normality by using the recirculated water (13◦C) containing a maintenance dosage of UNIVARIATE procedure in SAS (SAS 1990) and were tested MS-222 (50 mg/L); this process also kept the skin moist. Sur- for homogeneity of variances by using an Fmax test (Rohlf and gical equipment and dummy tags were immersed and sterilized Sokal 1981). Data were log10 transformed to achieve normality in a 70% solution of alcohol for 10 min prior to usage. The and homogeneity of variances. A one-way ANOVA was used to handling time for each individual was recorded, and fish were test for differences in tag retention time between the anchored returned to their corresponding tanks for recovery. All incisions and nonanchored tag treatment groups of both species. Two- were made using a scalpel (number-21 blade) on the ventral sur- factor, nested, repeated-measures ANOVAs were performed to face of the fish just left of the abdominal midline and anterior to examine the effects of treatment group on absolute growth, spe- the pelvic girdle. Incision sites were quickly closed with three cific growth rate, and TL for both species. For the null hypothesis sutures that were applied in a simple interrupted pattern (1 stitch of no significant difference between treatment groups, the error per 0.75 cm), as this has been shown to be the strongest and most term was specified as fish nested within treatment. The null hy- suitable method for incision closure on the skin of fish (Sum- pothesis of no significant difference of treatment over time had merfelt and Smith 1990; Wagner et al. 2000). Material used for a specified error term of time interacting with fish nested within

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 suturing consisted of an FSLX cutting needle attached to 3–0, treatment. Specific growth rate (%/week) was calculated as black, nonabsorbable monofilament nylon (Ethicon). Incisions were closed with a simple square knot. For nonanchored tags, log (Weight ) − log (Weight ) Specific growth rate = e f e i × 100, the tag transmitter was inserted through the incision into the t peritoneal cavity and was positioned just posterior to the liver and adjacent to the spiral valve. A closed hemostat was used where Weightf is final weight, Weighti is initial weight, and t as a blunt probe to shift and protect the internal organs during is the duration of the experiment (Ricker 1975). tag insertion. For anchored tags, suture material was threaded Repeated-measures ANOVAs were also used to compare be- through the fabricated hole at the anterior end of the tag. Forceps tween the two sturgeon species by using the same variables were used to hold the incision open and provide an unobstructed as mentioned above. Multiple a posteriori comparisons were view of the peritoneal cavity wall. The cutting needle was then completed with a least-squares means test that adjusts for the used to secure the suture to the dorsolateral abdominal, with care experimentwise error rate using Tukey’s correction. Growth of not to penetrate the dermis. The tag was then inserted through individuals that lost tags was compared with that of the popu- the incision and positioned in the same location, and the suture lation by standardizing their growth values each week relative material loop was closed and knotted to anchor the tag. The ven- to the average growth rate to identify any changes through- tral surface of each fish was photographed (Nikon Coolpix 990) out the study. For analysis of healing, incision lengths were 552 CROSSMAN ET AL.

measured using ImageJ version 1.19z. A generalized linear model was used to examine the change in incision length over time and to identify any differences among treatments. Re- gressions of incision lengths were then run separately for each species, and homogeneity of slopes was evaluated (Zar 1999) to identify differences in healing. A regression was also performed to examine the effects of surgical order on average absolute growth for both species to determine whether there was any surgical bias (i.e., associated with initial surgeries conducted by the inexperienced surgeon). Reviews of tagging studies completed to date (Bridger and Booth 2003; Cooke et al. 2011) indicated that many studies fail to document negative effects associated with tagging application, whether due to low sample sizes or improper statistical design (Cooke et al. 2011). In instances where the null hypothesis was not rejected, we conducted a retrospec- tive power analysis to determine the chance of committing a type II error. Power analyses were completed in GPOWER version 2.0 (freeware) for an a posteriori two-way ANOVA (Erdfelder et al. 1996).

RESULTS Tagging The time to complete surgery did not differ between anchored and nonanchored tag treatments for Shortnose Sturgeon (mean ± SD = 5.2 ± 0.8 min; F = 0.21, df = 1, 14, P = 0.656) or for Atlantic Sturgeon (4.9 ± 0.7 min; F = 0.14, df = 1, 14, P = 0.710); there was no difference in surgery time between species (F = 0.07; df = 1, 29; P = 0.798). The mean surgery time ± SD was 5.10 ± 0.80 min. Recovery from sedation took approximately 10 min for fish in all treatments. For Shortnose Sturgeon juveniles, a significant difference in retention time was found between anchored tags (88%) and FIGURE 1. Weekly incision lengths (cm) for juvenile (A) Shortnose Sturgeon nonanchored tags (25%; F = 6.94, df = 1, 14, P = 0.02) dur- and (B) Atlantic Sturgeon assigned to one of three treatment types over an ing the 8-week experimental period. With the exception of two 8-week period. tags, Shortnose Sturgeon tags that were not anchored were all expelled. Two nonanchored tags were expelled through the incision site during week 2 (days 8 and 12); the remaining tags plained much of the variation in healing for both Shortnose 2 Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 were expelled near the anus, and these tag expulsions occurred Sturgeon (r = 0.88; t = 61.89; df = 1, 166; P < 0.001) and in later weeks (days 31, 38, 47, and 54). Anchored tags had high Atlantic Sturgeon (r2 = 0.945; t = 76.51; df = 1, 142; P < retention in Shortnose Sturgeon, with only one tag expelled 0.001). Comparison of regression slopes revealed that Atlantic during the study; that tag was lost on day 54 through intesti- Sturgeon juveniles experienced a significantly higher rate of in- nal absorption. Atlantic Sturgeon juveniles in both treatment cision healing over the experiment (F = 56.07; df = 2, 328; groups retained 100% of their tags. All tags that were retained P < 0.001) than did Shortnose Sturgeon. Wounds that formed for the initial 8-week period were also retained for the next due to transmitter expulsion had fungal infections for a short 4 months. period (days) but healed promptly thereafter. Among Shortnose Sturgeon juveniles, those with nonanchored tags healed signif- Survival and Incision Healing icantly more slowly (F = 6.32; df = 2, 21; P = 0.007) than No fish died during the experiment. Initial incision lengths the other treatment groups. Similarly, Atlantic Sturgeon with (i.e., measured immediately after surgery) did not differ among nonanchored tags also had a significantly slower rate of healing treatments for Shortnose Sturgeon (F = 0.39; df = 2, 21; (F = 4.62; df = 2, 21; P = 0.0095) than the other treatment P = 0.684) or Atlantic Sturgeon (F = 0.36; df = 2, 21; P = groups. Incision lengths decreased during each week of the ex- 0.702). Mean incision length ( ± SD) was 2.55 ± 0.03 cm. periment for all individuals, indicating constant healing in both Both species healed rapidly, and time (in days; Figure 1) ex- species. TRANSMITTER IMPLANTATION IN JUVENILE STURGEON 553

TABLE 1. Average ( ± SE) weekly absolute growth (g) of Shortnose Sturgeon and Atlantic Sturgeon juveniles in four treatment groups over an 8-week period. Treatments were (1) exposure to anesthetic only, with no surgery; (2) anesthetic exposure and application of an incision with no tag; (3) anesthetic exposure and implantation of a nonanchored tag; and (4) anesthetic exposure and implantation of an anchored tag. Lowercase letters (z, y, x, w, and v) correspond to statistical differences ( P < 0.05) between species and between treatment groups over time.

Treatment Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8

Shortnose Sturgeon 19.00± 1.52 z 12.48 ± 1.91 w 12.23 ± 1.84 w 13.30 ± 1.66 w 12.99 ± 1.46 w 13.04 ± 1.41 w 13.52 ± 1.48 w 13.07 ± 1.36 w 28.60± 2.13 z 10.87 ± 2.17 w 11.38 ± 2.50 w 11.21 ± 2.50 w 11.07 ± 2.17 w 11.33 ± 1.94 w 11.92 ± 1.78 w 12.00 ± 1.71 w 35.80± 1.43 y 7.90 ± 1.51 v 7.77 ± 1.54 v 7.78 ± 1.68 v 7.63 ± 1.55 v 7.72 ± 1.40 v 8.26 ± 1.47 v 8.19 ± 1.31 v 47.40± 1.72 z 10.06 ± 2.05 w 9.66 ± 1.92 w 10.32 ± 2.02 w 9.80 ± 1.80 w 10.00 ± 1.73 w 10.40 ± 1.74 w 10.15 ± 1.69 w Atlantic Sturgeon 113.14± 2.34 x 15.85 ± 1.84 w 14.16 ± 1.82 w 13.74 ± 1.62 w 13.57 ± 1.29 w 13.40 ± 1.50 w 12.85 ± 1.38 w 12.94 ± 1.48 w 28.54± 2.84 x 9.31 ± 2.22 w 8.24 ± 1.85 w 8.60 ± 1.74 w 8.22 ± 1.54 w 8.27 ± 1.52 w 8.26 ± 1.43 w 8.65 ± 1.50 w 311.42± 1.92 x 13.89 ± 2.32 w 12.16 ± 2.27 w 11.70 ± 2.06 w 10.94 ± 2.08 w 10.79 ± 2.19 w 10.90 ± 1.94 w 11.20 ± 1.96 w 49.01± 2.04 x 13.26 ± 1.72 w 11.73 ± 1.49 w 11.83 ± 1.56 w 10.99 ± 1.51 w 11.01 ± 1.61 w 10.84 ± 1.57 w 11.30 ± 1.56 w

Growth respectively, for treatment groups 1–4. For Shortnose Sturgeon, Shortnose Sturgeon and Atlantic Sturgeon juveniles in all there was a significant difference in specific growth rate among treatment groups exhibited positive growth for the entirety of the treatment groups (F = 7.48; df = 3, 224; P < 0.001), but the the experiment (Table 1). A comparison of Shortnose Sturgeon interaction was not significant over time (F = 0.56, df = 21, 224, and Atlantic Sturgeon juveniles indicated no significant differ- P = 0.9553; power = 0.7461, f = 0.59, F3, 28 = 2.95). However, ence between species in the total amount of weight gained (F = as with absolute growth, the first week of the experiment was 0.08; df = 1, 62; P = 0.777). Atlantic Sturgeon exhibited higher characterized by significantly lower specific growth rates across absolute growth for the first week of the experiment (F = 2.89; all treatment groups (F = 66.78; df = 8, 224; P < 0.001). The df = 8, 448; P = 0.0038) but not afterwards (Table 1). For Short- specific growth rate of Atlantic Sturgeon was significantly lower nose Sturgeon juveniles, there was no significant difference in for the first 2 weeks (F = 59.59; df = 8, 224; P < 0.001) but absolute growth between treatments on a week-to-week basis was not affected across treatments (F = 1.01; df = 3, 28; P = throughout the 8-week period (F = 0.73, df = 24, 224, P = 0.4029) or across treatments over time (F = 0.89, df = 24, 224, = = = = 0.8171; power = 0.7768, effect size [ f ] = 0.61, F3, 28 = 2.95). P 0.6213; power 0.8438, f 0.66, F3, 28 2.95). However, absolute growth of Shortnose Sturgeon was reduced Shortnose Sturgeon juveniles that lost anchored tags experi- for all treatment groups during the first week of the experiment enced no negative effects on absolute growth; growth was, on only ( P < 0.001), as was observed with the species comparison. average, 1.3 standard deviations above the weekly mean abso- Despite fairly constant weekly growth rates over time, absolute lute growth for the full 8 weeks. The two Shortnose Sturgeon growth of Shortnose Sturgeon in the nonanchored tag treatment that expelled nonanchored tags during week 2 had absolute was significantly lower than that of juveniles in the other three growth values that were well below (i.e., 1.13 and 0.94 g lower treatments (F = 12.13; df = 3, 224; P < 0.001). For Atlantic than) the mean for all weeks of the experiment; the two fish that Sturgeon juveniles, absolute growth was not significantly dif- lost nonanchored tags during the last 2 weeks exhibited abso-

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 ferent among treatment groups (F = 2.22; df = 3, 224; P = lute growth that was 0.53 and 1.44 g lower than the mean. In 0.1075) or between treatment groups over time (F = 1.24, df = contrast, the two Shortnose Sturgeon from which nonanchored 24, 224, P = 0.2134; power = 0.8188, f = 0.64, F3, 28 = 2.95). tags were expelled during weeks 4 and 5 had absolute growth As with the Shortnose Sturgeon, absolute growth of Atlantic values that were above (i.e., 0.99 and 0.48 g greater than) the Sturgeon was affected during the first week of the experiment population mean for the entire experiment. Surgical order did (F = 289.63; df = 8, 224; P < 0.0001). not affect average absolute growth (r2 = 0.005; t = 0.466; df = Analysis of specific growth rate also revealed that growth 1, 46; P = 0.644). during the first week of the experiment was significantly different between Shortnose Sturgeon and Atlantic Sturgeon (F = 2.60; df = 8, 448; P = 0.009). Specific growth rate was not DISCUSSION significantly different (F = 0.86; df = 3, 49; P = 0.4674) be- Evaluation of new tagging techniques in the laboratory prior tween the species for the remainder of the experiment. Average to field trials helps to decrease uncertainty about the reliability specific growth per week for Shortnose Sturgeon in treatment of field data (Cooke et al. 2011). In our laboratory trials with groups 1–4 was 3.15, 2.83, 2.37, and 2.26%, respectively, over juvenile Shortnose Sturgeon and Atlantic Sturgeon (<55 cm the experimental period. Average specific growth per week for TL), anchoring the surgically implanted tag to the peritoneal Atlantic Sturgeon juveniles was 3.21, 3.17, 2.74, and 3.76%, cavity wall provided a significantly higher retention rate than 554 CROSSMAN ET AL.

nonanchored tags over a 6-month period. Anchoring the tag tag was not a result of any bias due to the surgeon’s experience. was quick, did not cause mortality, and can likely be conducted We are confident that for our experiment, prior training with on small research vessels with portable surgical equipment. a veterinary professional was instrumental in reducing surgical Shortnose Sturgeon with nonanchored tags had lower tag re- bias among the treatments. tention, lower initial growth, and slower rates of incision heal- Most studies that have evaluated surgical tag implantations ing than Shortnose Sturgeon that received anchored tags or no have not commented on the performance and suitability of the tags. Initial tag expulsions occurred through the incision and chosen suture material (Bridger and Booth 2003; Cooke et al. were likely due to transmitter pressure on the sutures. Smith 2011). The nonabsorbable monofilament proved to be favorable and King (2005) reported a 40% tag loss rate in juvenile Lake for use on the tough skin of juvenile Shortnose Sturgeon and At- Sturgeon within 1 week, and the expulsions were attributed to lantic Sturgeon. It held knots and tension extremely well and did the transmitters being physically pushed through the incision not degrade until after the wounds were healed. Monofilament site. Similarly, Walsh et al. (2000) attributed transmitter loss in sutures have also been shown to result in less inflammation Striped Bass Morone saxatilis × White Bass Morone chrysops and improved wound healing (Cooke et al. 2003). Although hybrids to pressure on the incision site rather than suture failure. monofilament sutures tend to take longer when suturing, their Furthermore, in a study by Wagner et al. (2000), radio transmit- long-lasting durability proved to be crucial in suture selection. ters that were implanted freely in the peritoneal cavity of Rain- The control for our study was an abdominal incision; how- bow Trout Oncorhynchus mykiss caused increased pressure and ever, future studies would benefit from using controls that in- wear on incision sites in comparison with fish that did not re- clude aspects of the main treatments they are testing. For ex- ceive transmitters. Although Collins et al. (2002) reported that ample, in our study, applying an internal suture without implan- Shortnose Sturgeon rubbed their incision sites, we did not ob- tation of a tag would have provided more insight into results serve this behavior. For Shortnose Sturgeon used in the present from the anchored tag treatment. Several studies have evaluated study, nonanchored tags that were not initially expelled through different suture materials (see review by Cooke et al. 2011), and the incision site moved posterior of the pelvic region and were future studies on sturgeon would benefit from examining the eventually expelled near the anus. A simple procedure for an- efficacy of different suture materials and application patterns choring the tag with nonabsorbable sutures helped to prevent and including those elements in the controls. tag movement and associated complications. The small peritoneal cavity of juvenile sturgeon makes it dif- Shortnose Sturgeon juveniles in all treatments exhibited ficult to implant even small acoustic tags. The slim body shape lower absolute growth than Atlantic Sturgeon juveniles during and narrow peritoneal cavity increase the probability that the the first week. This may be attributed to the Atlantic Sturgeon’s tag will press on internal organs and apply pressure to the inci- higher feeding efficiencies (Giberson and Litvak 2003). Atlantic sion site. Hence, when selecting the appropriate tags for use in Sturgeon juveniles were much less affected by the surgical pro- telemetry studies, the volume and shape of the tags are important cedures, as evidenced by their consistent growth rates and 100% factors and should be considered along with the species’ biol- transmitter retention. It was noted that the abdominal wall was ogy. Additional, species-level work in differing environments is much thinner in Atlantic Sturgeon juveniles than in Shortnose needed to relate the physical size of the transmitter (e.g., weight Sturgeon juveniles, and this difference may have influenced the and volume) to fish body weight and size before telemetry can ability of Atlantic Sturgeon to heal more quickly (J. A. Cross- be adapted to even smaller fish (Jepsen et al. 2002). The tags man, unpublished data). The two Shortnose Sturgeon that lost used in our study were light, but their volume could limit use tags via expulsion through the incision site healed very rapidly. in smaller fish. More recent advances in telemetry tags have

Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 The Shortnose Sturgeon that expelled nonanchored tags during produced further decreases in tag size relative to that used in weeks 7 and 8 were small, and the pelvic region may not have this study and may allow the tagging of smaller sturgeon than been sufficiently wide for the tag to pass during the initial weeks were examined here. of the experiment. In a small juvenile sturgeon, the movement Results from our study indicate that the anchoring of tags of the tag to a position posterior of the pelvic girdle may al- provides a reliable technique that can be used to tag juvenile ter the fish’s ability to achieve a maximum and efficient thrust Shortnose Sturgeon and Atlantic Sturgeon in the field. Although from its heterocercal tail (Liao and Lauder 2000). The behavior Atlantic Sturgeon in this study retained all of their implanted of our experimental fish did not appear to be altered; however, tags (whether anchored or not), we recommend the anchoring of further studies should be undertaken to examine swimming per- tags in field applications. In a preliminary laboratory study, we formance over long retention times (>50 d). did observe tag expulsion from Atlantic Sturgeon juveniles (M. Fish surgeries are frequently conducted by individuals that K. Litvak, unpublished data), and that preliminary result was the have used literature, mentoring, or trial and error as surgical justification for conducting this larger, controlled tagging study. teaching methods (Cooke et al. 2003, 2011). For the present ex- Ultimately, reducing the uncertainty in telemetry data will ad- periment, the single anchored tag that was lost from a Shortnose vance our ability to study juveniles of both Shortnose Sturgeon Sturgeon was the first anchoring surgery conducted by the sur- and Atlantic Sturgeon in the wild, which is important for imper- geon. However, regression analysis indicated that the loss of this iled populations with limited juvenile age-classes present. TRANSMITTER IMPLANTATION IN JUVENILE STURGEON 555

ACKNOWLEDGMENTS Giberson, A. V., and M. K. Litvak. 2003. Effect of feeding frequency on growth, We thank C. Davis for her help in rearing and maintaining food conversion efficiency, and meal size of juvenile Atlantic Sturgeon and the juvenile sturgeon used in this study, and we are grateful Shortnose Sturgeon. North American Journal of Aquaculture 65:99–105. Gross, M. R., J. Repka, C. T. Robertson, D. H. Secor, and W. Van Winkle. 2002. to I. Butts for his contributions to the analysis. Comments re- Sturgeon conservation: insights from elasticity analysis. Pages 13–30 in W. ceived from three anonymous reviewers greatly improved this Van Winkle, P. J. Anders, D. H. Secor, and D. A. Dixon, editors. Biology, paper. We also thank M. Collins and B. Kynard, who provided management, and protection of North American sturgeon. American Fisheries comments on an earlier version of the manuscript. Funding Society, Symposium 28, Bethesda, Maryland. for this work was provided by grants from the World Wildlife Hall, J. W., T. I. J. Smith, and S. D. Lamprecht. 1991. Movements and habitats of Shortnose Sturgeon, Acipenser brevirostrum, in the Savannah River. Copeia Fund Canada, New Brunswick Trust Fund, MEC Environmental 1991:695–702. Fund, and a Natural Sciences and Engineering Research Council Hatin, D., R. Fortin, and F. Caron. 2002. Movements and aggregation areas Discovery Grant to M.K.L. of adult Atlantic Sturgeon (Acipenser oxyrinchus) in the St. Lawrence River estuary, Quebec,´ Canada. Journal of Applied Ichthyology 18:586–594. Jepsen, N., A. Koed, E. B. Thorstad, and E. Baras. 2002. Surgical implantation REFERENCES of telemetry transmitters in fish: how much have we learned? Hydrobiologia Bridger, C. J., and R. K. Booth. 2003. The effects of biotelemetry transmit- 483:239–248. ter presence and attachment procedures on fish physiology and behavior. Kieffer, M. C., and B. Kynard. 1993. Annual movements of Shortnose and Reviews in Fisheries Science 11:13–34. Atlantic sturgeons in the Merrimack River, Massachusetts. Transactions of Buckley, J., and B. Kynard. 1985. Habitat use and behavior of pre-spawning the American Fisheries Society 122:1088–1103. and spawning Shortnose Sturgeon, Acipenser brevirostrum, in the Connecti- Kynard, B., M. Breece, M. Atcheson, M. Kieffer, and M. Mangold. 2009. Life cut River. Pages 111–117 in F. P. Binkowski and S. I. Doroshov, editors. history and status of Shortnose Sturgeon (Acipenser brevirostrum)inthe North American sturgeons: biology and aquaculture potential. Dr W. Junk, Potomac River. Journal of Applied Ichthyology 25(Supplement 2):34–38. Dordrecht, The Netherlands. Kynard, B., and M. Horgan. 2002. Ontogenetic behavior and migration of At- Collins, M. R., D. Cooke, B. Post, J. Crane, J. Bulak, T. I. J. Smith, T. W. lantic Sturgeon, Acipenser oxyrinchus oxyrinchus, and Shortnose Sturgeon, Greig, and J. M. Quattro. 2003. Shortnose Sturgeon in the Santee–Cooper A. brevirostrum, with notes on social behavior. Environmental Biology of reservoir system, South Carolina. Transactions of the American Fisheries Fishes 63:137–150. Society 132:1244–1250. Kynard, B., M. Horgan, M. Kieffer, and D. Seibel. 2000. Habitats used by Collins, M. R., D. W. Cooke, T. I. J. Smith, W. C. Post, D. C. Russ, and D. Shortnose Sturgeon in two Massachusetts rivers, with notes on estuarine C. Walling. 2002. Evaluation of four methods of transmitter attachment on Atlantic Sturgeon: a hierarchical approach. Transactions of the American Shortnose Sturgeon, Acipenser brevirostrum. Journal of Applied Ichthyology Fisheries Society 129:487–503. 18:491–494. LaVigne, H. R. 2002. An improved portable surgical table for the field and Collins, M. R., S. G. Rogers, T. I. J. Smith, and M. L. Moser. 2000. Primary fac- laboratory. North American Journal of Fisheries Management 22:571–572. tors affecting sturgeon populations in the southeastern United States: fishing Li, X. H., M. K. Litvak, and J. E. Hughes Clarke. 2007. Overwintering habitat mortality and degradation of essential habitats. Bulletin of Marine Science use of Shortnose Sturgeon (Acipenser brevirostrum): defining critical habitat 66:917–928. using a novel underwater video survey and modeling approach. Canadian Cooke, S. J., B. D. S. Graeb, C. D. Suski, and K. G. Ostrand. 2003. Effects of su- Journal of Fisheries and Aquatic Sciences 64:1248–1257. ture material on incision healing, growth and survival of juvenile Largemouth Liao, J., and G. V. Lauder. 2000. Function of the heterocercal tail in White Stur- Bass implanted with miniature radio transmitters: case study of a novice and geon: flow visualization during steady swimming and vertical maneuvering. experienced fish surgeon. Journal of Fish Biology 62:1366–1380. Journal of Experimental Biology 203:3585–3594. Cooke, S. J., C. M. Woodley, M. B. Eppard, R. S. Brown, and J. L. Nielsen. Moser, M. L., M. Bain, M. R. Collins, N. Haley, B. Kynard, J. C. O’Herron 2011. Advancing the surgical implantation of electronic tags in fish: a gap II, G. Rogers, and T. S. Squiers. 2000. A protocol for use of Shortnose and analysis and research agenda based on a review of trends in intracoelomic Atlantic sturgeons. NOAA Technical Memorandum NMFS-OPR-18. tagging effects studies. Reviews in Fish Biology and Fisheries 21:127–151. Moser, M. L., and S. W. Ross. 1995. Habitat use and movements of Short- COSEWIC (Committee on the Status of Endangered Wildlife in Canada). nose and Atlantic sturgeons in the lower Cape Fear River, North Carolina. 2011. COSEWIC assessment and update status report on the Shortnose Stur- Transactions of the American Fisheries Society 124:225–234. Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 geon Acipenser brevirostrum in Canada. COSEWIC, Environment Canada, Ricker, W. E. 1975. Computation and interpretation of biological statistics of Ottawa. fish populations. Fisheries Research Board of Canada Bulletin 191. Counihan, T. D., and C. N. Frost. 1999. Influence of externally attached transmit- Rohlf, F. J., and R. R. Sokal. 1981. Statistical tables, 2nd edition. Freeman, New ters on the swimming performance of juvenile White Sturgeon. Transactions York. of the American Fisheries Society 128:965–970. SAS (Statistical Analysis Systems). 1990. SAS/STAT user’s guide, version 6, Dadswell, M. J., B. D. Taubert, T. S. Squiers, D. Marchette, and J. Buckley. 1984. edition 4, volume 2. SAS Institute, Cary, North Carolina. Synopsis of biological data on Shortnose Sturgeon, Acipenser brevirostrum Savoy, T., and D. Pacileo. 2003. Movements and important habitats of subadult LeSueur 1818. NOAA Technical Report NMFS 14. Atlantic Sturgeon in Connecticut waters. Transactions of the American Fish- Erdfelder, E., F. Faul, and A. Buchner. 1996. GPOWER: a general power analysis eries Society 132:1–8. program. Behavior Research Methods, Instruments, and Computers 28:1–11. Schramm, H. L., Jr., and D. J. Black. 1984. Anesthesia and surgical procedures Fernandes, S. J., G. B. Zydlewski, J. D. Zydlewski, G. S. Wippelhauser, and M. T. for implanting radio transmitters into Grass Carp. Progressive Fish-Culturist Kinnison. 2010. Seasonal distribution and movements of Shortnose Sturgeon 46:185–190. and Atlantic Sturgeon in the Penobscot River estuary, Maine. Transactions of Secor, D. H., P. J. Anders, W. Van Winkle, and D. A. Dixon. 2002. Can we study the American Fisheries Society 139:1436–1449. sturgeons to extinction? what we do and don’t know about the conservation Giberson, A. V. 2004. Social behaviour, competition and interactions of ju- of North American sturgeons. Pages 3–10 in W. Van Winkle, P. J. Anders, venile Shortnose Sturgeon (Acipenser brevirostrum) and Atlantic Sturgeon D. H. Secor, and D. A. Dixon, editors. Biology, management, and protection (Acipenser oxyrinchus). Master’s thesis. University of New Brunswick, Saint of North American sturgeon. American Fisheries Society, Symposium 28, John. Bethesda, Maryland. 556 CROSSMAN ET AL.

Secor, D. H., E. J. Niklitschek, J. T. Stevenson, T. E. Gunderson, S. P. Sturgeon. Transactions of the American Fisheries Society 132:1257– Minkkinen, B. Richardson, B. Florence, M. Mangold, J. Skjeveland, and A. 1263. Henderson-Arzapalo. 2000. Dispersal and growth of yearling Atlantic Stur- Trested, D. G., K. Ware, R. Bakal, and J. J. Isely. 2011. Microhabitat use and geon, Acipenser oxyrinchus, released into Chesapeake Bay. U.S. National seasonal movements of hatchery-reared and wild Shortnose Sturgeon in the Marine Fisheries Service Fishery Bulletin 98:800–810. Savannah River, South Carolina–Georgia. Journal of Applied Ichthyology Smith, K. M., and D. K. King. 2005. Movement and habitat use of yearling 27:454–461. and juvenile Lake Sturgeon in Black Lake, Michigan. Transactions of the Usvyatsov, S., J. Watmough, and M. K. Litvak. 2012. Age and population American Fisheries Society 134:1159–1172. size estimates of overwintering Shortnose Sturgeon in the Saint John River, Smith, T. I. J., and J. P. Clugston. 1997. Status and management of Atlantic New Brunswick, Canada. Transactions of the American Fisheries Society Sturgeon, Acipenser oxyrinchus, in North America. Environmental Biology 141:1126–1136. of Fishes 48:335–346. Wagner, G. N., E. D. Stevens, and P. Byrne. 2000. Effects of suture type and Smith, T. I. J., M. Collins, W. C. Post, and J. W. McCord. 2002. Stock enhance- patterns on surgical wound healing in Rainbow Trout. Transactions of the ment of Shortnose Sturgeon: a case study. Pages 31–44 in W. Van Winkle, P. American Fisheries Society 129:1196–1205. J. Anders, D. H. Secor, and D. A. Dixon, editors. Biology, management, and Walsh, M. G., K. A. Bjorgo, and J. J. Isely. 2000. Effects of implantation method protection of North American sturgeon. American Fisheries Society, Sympo- and temperature on mortality and loss of simulated transmitters in hybrid sium 28, Bethesda, Maryland. Striped Bass. Transactions of the American Fisheries Society 129:539–544. Summerfelt, R. C., and L. S. Smith. 1990. Anesthesia, surgery, and related Winter, J. 1996. Advances in underwater biotelemetry. Pages 555–590 in B. R. techniques. Pages 213–272 in C. B. Schreck and P.B. Moyle, editors. Methods Murphy and D. W.Willis, editors. Fisheries techniques, 2nd edition. American for ﬁsh biology. American Fisheries Society, Bethesda, Maryland. Fisheries Society, Bethesda, Maryland. Sutton, T. M., and A. C. Benson. 2003. Inﬂuence of external trans- Zar, J. H. 1999. Biostatistical analysis, 4th edition. Prentice-Hall, Englewood mitter shape and size on tag retention and growth of juvenile Lake Cliffs, New Jersey. Downloaded by [Department Of Fisheries] at 20:06 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:07 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Linking Fish and Angler Dynamics to Assess Stocking Strategies for Hatchery-Dependent, Open-Access Recreational Fisheries Paul J. Askey a , Eric A. Parkinson b & John R. Post c a British Columbia Ministry of Forests, Lands, and Natural Resource Operations, Fish and Wildlife Section , 102 Industrial Place, Penticton , British Columbia , V2A 7C8 , Canada b British Columbia Ministry of Environment, Fisheries Center , University of British Columbia , 2202 Main Mall, Vancouver , British Columbia , V6T 1Z4 , Canada c Department of Biological Sciences , University of Calgary , 2500 University Drive, Calgary , Alberta , T3L 1N4 , Canada Published online: 16 May 2013.

To cite this article: Paul J. Askey , Eric A. Parkinson & John R. Post (2013): Linking Fish and Angler Dynamics to Assess Stocking Strategies for Hatchery-Dependent, Open-Access Recreational Fisheries, North American Journal of Fisheries Management, 33:3, 557-568 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785996

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:557–568, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785996

ARTICLE

Linking Fish and Angler Dynamics to Assess Stocking Strategies for Hatchery-Dependent, Open-Access Recreational Fisheries

Paul J. Askey* British Columbia Ministry of Forests, Lands, and Natural Resource Operations, Fish and Wildlife Section, 102 Industrial Place, Penticton, British Columbia V2A 7C8, Canada Eric A. Parkinson British Columbia Ministry of Environment, Fisheries Center, University of British Columbia, 2202 Main Mall, Vancouver, British Columbia V6T 1Z4, Canada John R. Post Department of Biological Sciences, University of Calgary, 2500 University Drive, Calgary, Alberta T3L 1N4, Canada

Abstract Optimization of stocking practices regarding release size and density requires an understanding of how dynamic ecological and angler effort processes interact. We used experimental data on size and density-dependent fish recruitment processes and combined these with empirical fishery data to model the outcome of different stocking strategies. The model is based on the British Columbia Rainbow Trout Oncorhynchus mykiss fishery, which is an open-access recreational fishery where fish recruitment is entirely derived from hatchery production. Under this scenario, changes to stocking practices primarily influence angler effort densities, whereas angling quality, defined as a function of catch rate and fish size, remain relatively constant. In light of this, we suggest that the primary performance measure for open-access recreational fisheries management be sustainable angler effort. We modeled the effort response to stocking changes under contrasting biological (lake productivity) and fishery (remoteness, harvest regulations) characteristics. Lake-specific effort is maximized by stocking larger fish into productive lakes that are harvested. However, the strategy to maximize regional effort across multiple lakes was dependent on the ratio of hatchery production to total lake area and the mix of individual lake productivities within the total lake area. Downloaded by [Department Of Fisheries] at 20:07 28 May 2013

Many inland fisheries rely on the stocking of hatchery-reared Post et al. 1999, Lorenzen 2000; Leber et al. 2005), and timing fish to sustain viable recreational angling opportunities where of release (Madenjian et al. 1991; Donovan et al. 1997) on the natural reproduction is limited or absent (Ellison and Franzin numbers and size of fish after a given time interval (usually the 1992; Cowx et al. 1998; Kozfkay et al. 2006). This has motivated first growing season). These studies show that ecological pro- research on how to maximize fish production given a variety of cesses such as density-dependent growth and size-dependent potential stocking strategies. Previous studies have assessed the mortality influence the efficiency of stocking programs. How- impact of release densities (Walters and Post 1993; Whalen and ever, angler harvest will also affect the density and size structure LaBar 1994; Lorenzen 1995; Post et al. 1999; Fayram et al. of stocked populations. Furthermore, angler effort is not inde- 2005), size at release (Santucci and Wahl 1993; Lorenzen 1995; pendent of fish production, and it is expected that effort will

*Corresponding author: [email protected] Received July 10, 2012; accepted March 10, 2013 557 558 ASKEY ET AL.

respond to changes in stocking practices (Moring 1985, 1993; angling quality is equivalent to the regional average (equal- Andrews and Wilen 1988; Johnson and Carpenter 1994; Cowley quality isopleth). et al. 2003). This dynamic interaction between stocking prac- Given the context of effort dynamics in recreational fisheries tices, ecology, and fishing effort has not been well integrated outlined above, it is time to reconsider how, and at what scale, the to a format useful to fisheries managers in previous work con- performance of stocking programs intended to enhance recre- sidering stocking strategies for recreational fisheries (but see ational fisheries should be measured. We suggest that angler Fenichel et al. 2010). effort is the appropriate performance measure for management The ability of fishers to track variation in fish production success when the angler population is mobile across a region across spatial scales has been recognized in many fisheries with multiple bodies of water. Standard measures of angling (Gillis et al. 1993; Cowley et al. 2003; Gillis 2003). The common quality such as fish size, density, condition, or angler catch rates null hypothesis for such behavior is an ideal free distribution are really only representative of what is left over by the anglers (Fretwell and Lucas 1970), whereby fishing effort concentrates and not what is being produced (stocked). Furthermore, we must on fish aggregations and equalizes catch rates across fishing consider trends in angler effort at a regional scale across individ- “patches.” In commercial fisheries this process is recognized as ual bodies of water. A relevant fisheries management “region” creating a serious bias in the use of catch rates as an index of is a geographic area containing multiple bodies of water that are abundance (Hilborn and Walters 1992). These ideas also apply equally accessible to a common pool of anglers (and therefore to recreational fisheries (Post et al. 2002, 2008; Beard et al. 2003; well described by a single-quality isopleth). Whereas, a gover- Cox et al. 2003), which are often composed of many individual nance region (e.g., province of British Columbia) may exhibit bodies of water that may vary in productivity but are subject to a a wide range in the distance and accessibility of lakes from common pool of anglers. Ideal free distribution theory predicts population centers (i.e., available angler pool), which leads to a that angling quality on a set of lakes with equivalent cost of fish- series of smaller fisheries management regions with corresponding (i.e., equally accessible) should be equivalent, and changes ing equal-quality isopleths (Parkinson et al. 2004). Thus, man- in fish production between lakes should manifest as differences agers should measure performance of actions on a single body in angler effort density. of water as changes to angler effort given the regional context. Application of ideal free distribution theory to recreational Ultimately, a stocking program (where no natural recruitment fisheries requires the definition of two primary components: exists) would aim to maximize long-term, region-wide effort for (1) cost of fishing, and (2) angling quality. The precise cost of a given set of lakes and hatchery capacity. fishing is probably a complex function of many factors; however, Our objective was to develop a conceptual framework and distance from population centers appears to be a key driving fac- model to integrate (1) the ecological processes that drive fish tor (Parkinson et al. 2004; Post et al. 2008; Carson et al. 2009; population abundance and size structure, and (2) angler effort Hunt et al. 2011). Similarly, recreational anglers may assess an- and catch statistics (i.e., angling quality) in order to determine gling quality using a variety of catch and noncatch attributes of optimal stocking strategies. We applied the approach to the fishing sites (Hunt 2005; Arlinghaus 2006). Noncatch-related British Columbia Rainbow Trout Oncorhynchus mykiss fishery, attributes (e.g., aesthetics, amenities) affect angler effort, but which is an open-access, hatchery-dependent, recreational fish- (like travel costs) are typically unrelated to fish populations and ery. This fishery is an ideal example because of the availability fishery management decisions. Thus, once noncatch constants, of a large experimental data set on the ecological processes of such as amenities and travel costs, are accounted for, angler ef- growth and survival of lentic Rainbow Trout in British Columbia fort dynamics are primarily a function of individual site choice (Post et al. 1999; Askey 2007) and a large creel data set that de-

Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 based on characteristics of the catch (Post et al. 2008; Hunt et al. fines the size–number tradeoffs of the population of anglers in 2011). Parkinson et al. (2004) showed that a strong relationship this region (Parkinson et al. 2004). We also generalize the results exists between the number and size of retained fish from an- across an ecological productivity gradient, across regions that gler surveys in the southern interior of British Columbia. They differ in the size of the pool of anglers, and for fisheries that use suggested that this relationship represented the reduction of bio- restrictive regulations to control harvest. logically feasible fish population size and density combinations to a more narrowly defined range of alternatives between numerous small fish and few large fish. The empirical curve was METHODS termed an equal-quality isopleth, which accounts for the aggre- Study site and model structure.—The stocking optimization gate preferences of anglers and the necessary trade-off between model was based on the small lakes (<1,000 ha) Rainbow numbers of fish captured and size of the fish when selecting Trout fishery in British Columbia. This is the largest sport fishing sites. The strong relationship implies that changes to fishery in British Columbia, and it is heavily supplemented stocking practices on a given body of water will have little with hatchery production (over 800 lakes are stocked annually). effect on perceived angling quality when alternative angling The fishery is divided among many biologically independent opportunities are available. Rather it is expected that angler stocks (lakes) that are fished by a common pool of anglers. effort will simply move to or from the body of water until In order to simulate this process we created a single Rainbow ASSESSING STOCKING STRATEGIES FOR RECREATIONAL FISHERIES 559 Downloaded by [Department Of Fisheries] at 20:07 28 May 2013

FIGURE 1. (a) The structure of the stocking optimization model with interactions among hatchery production, fish ecology, the fishery, and anglers with outside influences of budgets, environment, regulations, and distance and travel costs. (b) Age-structured interactions of stocked Rainbow Trout. The thickness of the lines represents the relative proportion of individuals from a compartment following that path. Mathematical descriptions of all model interactions and transitions are given in the text.

Trout population that is subject to regional effort dynamics. considered were size and stage, density-dependent and The age-structured model considered the simple case of a lake environment-dependent growth, and survival of fish. Key fish- without natural reproduction. The model was structured as a ery processes included size-dependent vulnerability to capture series of ecological, fishery, and angler processes designed to and harvest and dynamic angler effort. Angler behavior was assess the dynamic outcome of stocking practices, environ- modeled assuming an ideal free distribution (Fretwell and Lucas mental drivers, and regulations and how they affect fishing 1970) based on the integrated expression of fishing quality from quality and effort in a lake embedded in a landscape of alternate Parkinson et al. (2004), which implicitly includes noncatch fishing opportunities (Figure 1a). Given there is no natural attributes such as travel distance or costs. An explicit age- (and reproduction in stocked lakes, the key ecological processes size-) structured approach was used to link the ecological to 560 ASKEY ET AL.

fishery processes to follow the demography of stocked density 0, and α describes the rate at which growth decreases fish throughout their development and exposure to harvest with effective population density (D). Population density was (Figure 1b). estimated as the sum of lengths squared for all individuals in Simulations involved following sequential cohorts of stocked the population divided by the lake area, which is a better metric fish throughout their lifespan as they grew and were subjected of exploitative competition based on the assumption that con- to size-selective natural and fishing mortality, resulting in a sta- sumption rate varies with body length squared (Walters and Post ble age distribution at equilibrium (100-year simulations). We 1993; Post et al. 1999). The rate at which growth declines with simulated over broad but realistic ranges of size-at-release and effective density (α) was dependent on the regional lake envi- stocking-density strategies. We conducted four types of simula- ronment (warm or cold: Askey 2007), which is denoted by the tions. First, we simulated the expected size and number of fish subscript R. that would be caught under pristine conditions. Second, we cal- The annual growth increment of immature fish (h) outlined culated the angler effort necessary to drive the fishery down to above was also used as one of the parameters to describe the the empirically observed equal-quality isopleth under all combi- growth trajectory of fish after maturation. Lester et al. (2004) nations of size and density at release. Third, we generalized our reparameterized the von Bertalanffy growth model in terms of single lake model results by assessing the effects of (1) variation juvenile growth (h), reproductive investment (g), and age at in ecological productivity, (2) distance of the fishery from the maturity (T) so that the standard parameters (L∞, k, t0) can be point of origin of anglers, and (3) restrictive regulations. Fourth, derived as L∞ = 3h/g; k = ln(1 + g/3); and t0 = T + ln(1 − we demonstrated how the single-lake effort response surfaces g(T − t1)/3)/ ln(1 + g/3). Thus size-at-age data combined with integrate into the maximization of total effort across a region the data on immature growth were fit to the Lester et al. (2004) (containing multiple lakes). model by maximum likelihood. The model fit is presented in Ecological processes.—All data on biological processes out- Figure 2 and the maximum likelihood parameter estimates are lined in the model structure were derived from 17 lakes (41 lake- presented in Table 1. years of data) from two different groups of experimental lakes Juvenile survival over the growing season is a function of (Post et al. 1999; Askey 2007). The two groups (18 warm and initial size, growth rate, and population density (Post et al. 1999; 21 cold) of experimental lakes represented two discrete samples Askey 2007). The optimal model fit from 41 lake-years of data of lake environmental conditions occurring over the larger scale led to the following model describing the change in numbers (mountainous) landscape of the interior of British Columbia. We (N) for a given age-cohort (a) over a yearly time step (t): refer to these subgroups of lakes as “warm” (altitude ≈ 1,000 m, = µ mean epilimnetic total phosphorus 27 g/L) or “cold” (alti- p1 Na+ ,t+ = Na,t exp(−s¯a,t p D ). (3) tude ≈ 1,500 m, mean epilimnetic total phosphorus = 12 µg/L) 1 1 0 lakes for simplicity. Growth and survival rates were measured across a range of manipulated fish densities and size-classes in In the above equation p0 and p1 are empirical parameters de- all of the experimental lakes. scribing how mortality varies with effective population density Growth rates of immature Rainbow Trout are a function of (D) (Figure 3) and s¯ is a parameter describing the size-dependent initial size, population density, and the lake environmental con- susceptibility to predation by gape-limited predators. The rela- ditions (Post et al. 1999; Askey 2007). However, for purposes tive susceptibility to predation was modeled as a negative expo- of the present model we were interested in growth in length nential function of fish length: over all life stages. Lester et al. (2004) provided a theoretical

Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 link between juvenile growth rate and asymptotic length. They s(l) = exp[−β(l − lmin)], (4) pointed out that growth in length for immature fish is expected to be geometric, and could be of the form: where lmin is the smallest size at release (2.5 cm) and assumed fully vulnerable [i.e., s(0) = 1], and β is the rate at which l + = hl , (1) t 1 t susceptibility to predation declines with size. The value of s may change substantially within the first season as fish grow where h is the annual growth increment of immature fish, l is out of the size range at which they are susceptible to predation. fork length (FL), and t is time in years. In order to incorporate However, the model used an annual time step and thus it was affects of density and lake environmental conditions, the growth necessary to simplify susceptibility over the season to a single increment was fit to growth data from a range of densities in value (s¯). Given that the rate of change in l is constant for two lake environments as: immature fish equation (1), we estimated s¯ as the average value of the function s(l) over the interval between the initial and final h1 h = + h0, (2) predicted lengths (denoted lt and lt + 1) as follows: 1 + αR D where h is a minimal annual growth increment at high density, 1 lt+1 0 s¯ = s(l)dl. (5) h + h is the maximum annual growth increment at population − 1 0 (lt+1 lt ) lt ASSESSING STOCKING STRATEGIES FOR RECREATIONAL FISHERIES 561

TABLE 1. List of parameter values used in the stocking optimization model.

Parameter Description Value Growth T Age at maturity 1.50 g Reproductive investment 0.54 h0 Minimum growth increment (cm) 2.17 h1 Maximum additional growth at low density (cm) 19.88 αR Density-dependent decline in growth 0.64(warm), 1.67(cold) Natural mortality p0 Density-dependent predation risk 0.023 p1 Density-dependent predation risk 0.21 lmin Size where fully vulnerable to predators (cm) 2.50 β Size-dependence 0.12 Maturation mortality rate (proportion per year) 0.71 Fishery l50 Length at 50% vulnerability to the fishery Variable (age-2 length) m Slope of angling size-selectivity function 9.00 q Catchability coefficient (ha swept/angler-day) 0.10 umax Maximum exploitation rate 0.85 k Rate of fish turnover into a vulnerable state 1.90 =−ln(1 − umax) l* Minimum length of fish valued by anglers (cm) 20.00 a Quality isopleth shape parameter 9.11 b Quality isopleth shape parameter −0.89 r Relative value of released fish 0.16 cr Angling release mortality 0.10

Simplifying the above equation gives an approximation for the where la,t is predicted FL of age-class a at time t, l50 is length at natural survival rate of age-class a over one full season starting which fish are 50% vulnerable to angling, and m is proportional in year t: to the slope at l50. Interestingly the relative vulnerability-to- angling function is not constant across lakes and appears to be −β − − −β − related to the size of fish within the lake (Cox 2000). Thus l50 = exp[ (la,t lmin)] exp[ (la+1,t+1 lmin)]. s¯a,t (6) was set equal to length at age 2, which allows the l50 parameter to β(l + , + − l , ) a 1 t 1 a t increase with increasing growth and size of fish in the lake. The parameter m was set to a value of 9 based on angling experiments We used the above approximation for all age-classes, but large (Askey et al. 2006). fish typically have very low s¯ and therefore negligible predation Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 The selectivity function in equation (7) represents a size mortality. However, large mature Rainbow Trout do suffer mat- refuge from anglers, and fish may also limit their exposure to uration mortality (increased mortality rate due to the maturation anglers through behavior. If fish move to areas inaccessible to process even when spawning is not possible). Maturation mor- angling gear or are simply nonreactive for some proportion of tality was estimated by comparing the ratio of age-2 males to time, then exploitation rates will necessarily be below unity. age-2 females given known maturation rate of males in the first This concept was investigated in detail in Cox (2000) and Cox year and zero maturation rate for females in the first year. The and Walters (2002) and resulted in the following equation for maturation mortality estimate (Table 1) was the last necessary instantaneous annual fishing mortality (f ): parameter to define the biological model structure in Figure 1. Fishery processes.—Angling is size selective and can be de- = qkEt , scribed by a sigmoidal relative vulnerability function (Cox 2000; ft (8) 2k + qEt Askey et al. 2006). The relative vulnerability of fish to angling was modeled as where q is the catchability coefficient, k is the rate at which fish turnover between behaviorally invulnerable and vulnerable m la,t states, and E is effort. The units for E used in the model were v , = , (7) a t m + m la,t l50 angler-days per hectare (angler-days/ha), where an angler-day 562 ASKEY ET AL.

300 0.050 2.5 cm (a) 5 cm 0.040 10 cm 250 20 cm 0.030 30 cm 200 0.020 150

Mortality rate 0.010

100 0.000

Growth increment 012345 50 Population density

0 FIGURE 3. Density-dependent natural mortality rates for Rainbow Trout 0123456given ﬁve different initial fork lengths from 2.5 to 30 cm. The functions were derived from 42 lake-years of data of age-0 and age-1 ﬁsh in experimental Population density lakes in two regions of the southern interior of British Columbia. Density was estimated as cm2·ha−1·10−5 and the mortality rate is %/d.

400 (b) 350 the above equations were taken from Parkinson et al. (2004) 300 and are presented in Table 1. Note that the model structure 250 was such that angler harvest was applied before the density- 200 dependent natural mortality and growth equations. This was chosen as the best option for implementing ﬁshing mortality 150 because the majority of angler effort is observed to occur in the

Fork length (mm) 100 spring (Tredger 1992). Catch-and-release regulations were also 50 simulated. It was assumed that all anglers were compliant with regulations, but a release mortality rate of 10% was implemented 0 based on the mode of angling release mortality for salmonids in 01234567 Age Bartholomew and Bohnsack (2005). Angler behavior.—Dynamic fishing effort was modeled by FIGURE 2. Components of the parameterization of Rainbow Trout growth assuming that anglers will conform to an ideal free distribu- rates. (a) Growth in length (mm) over 165 d as a function of population density tion and all lakes within a region will be driven to an equiva- − − (cm2·ha 1·10 5) for age-0 and age-1 fish in experimental lakes in two re- lent angling quality. Parkinson et al. (2004) showed that there gions of British Columbia characterized by warm and nutrient-rich lakes (solid is a relationship between size and number of fish retained by line, filled symbols) and cold, nutrient-poor lakes (dashed line, open symbols). (b) Size at age (FL, mm) and a fitted growth curve for both immature and adult anglers. They suggested that this relationship represented an Rainbow Trout from three, cold, nutrient-poor lakes. The growth model was fit equal-quality isopleth, which accounted for the trade-off be- simultaneously with data from immature and mature fish. The immature fish tween size and number of fish harvested in angler assessment growth rates provide information on the density-dependence of growth, whereas of quality. However, the data used for their isopleths ignored the size-at-age data provide information on the rate at which growth decreases released fish, and thus they implicitly assumed that released fish Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 with reproductive investment later in life. have no utility value for anglers. In order to assess the effect of regulations that may force anglers to release some proportion is equivalent to 4 h of fishing time by a single angler (Tredger of the catch it was necessary to establish a measure of angling 1992). Given the above estimate of instantaneous fishing mor- quality that included released fish. Intuitively, released fish must tality, the annual exploitation (proportion of fish harvested for a have a relative value (compared with harvested fish) between 0 fully vulnerable age-class; denoted as u)is and 1, and this parameter was denoted as r. If this value is 1 then anglers assess the quality of a fishery solely on the size u = 1 − exp( f ), (9) t t and number of fish caught, and have no added utility value for harvested fish. However, if anglers place value on retaining fish and the number of fish removed from year-class a over a single then r must be below 1, and an assessment of quality must stan- season is dardize catch rates for the proportion of catch harvested. An effective catch rate (ECUE) that weights released fish by their Catch , = v , u N , , (10) a t a t t a t estimated relative value r can be expressed as recalling that v accounts for size differences among age-classes and angler effort is incorporated within u. Parameter values for ECUE = rpCUE + (1 − p)CUE = [1 + p(r − 1)]CUE (11) ASSESSING STOCKING STRATEGIES FOR RECREATIONAL FISHERIES 563

where p is the proportion of the total catch that is released. Note this assumption is independent of the average size retained. It is that equation (11) states that two different lakes with the same likely that the released fish are smaller than the killed catch, and absolute catch rates may have different effective catch rates if this leads to potential bias. We assessed the potential influence regulations force the anglers to return a different proportion of these biases by varying both the isopleth and relative value of the catch on each lake. Conversely, different absolute catch of released fish during simulations. rates (CUE) may lead to equivalent effective catch rates (ECUE) The quality-equalizing process described above was incor- if the difference in CUE between lakes is proportional to 1/r. porated into the model by making angler effort a function of Finally, ECUE should be standardized by the average size of angling quality at the previous time step as follows: fish captured to create an equal-quality isopleth as follows:

Et+1 = Qt Et . (14) ECUE(l) = a(l − l*)b. (12) A spatially structured recreational fishery provides a mechanism The above relationship predicts the expected ECUE (denoted for E to be highly responsive to changes in Q (e.g., British ECUE) given the average size of retained fish (l), where a and Columbia Rainbow Trout anglers can just move between lakes if b are empirical shape parameters and l* is the minimal length Q varies). However, the spatial component may not be essential. of fish “useful” to anglers (set at 20 cm). If the above relation- The key assumption is not where the demand comes from, but ship depicts a regional equal-quality isopleth, then the relative that the demand is elastic (i.e., small changes in Q will result in angling quality of a single lake can be represented as large changes in E). We assessed the expected size and number of fish that could ECUE be captured when lakes are pristine (i.e., fish population density Q = . (13) ECUE(l) and size structure has not been altered by harvest) by running the model with a constant but negligible angler effort (0.00001 However, if anglers move among lakes to equalize quality then angler-days/ha) for warm and cold lake types. Simulations were Q should quickly be driven to a value of 1. We minimized the done with a size at stocking of 6.5 cm and stocking density sums of squares (Q − 1)2 to fit the three parameters: a, b, and r, ranging from 10 to 2,010 fry/ha to produce the fish size–catch using the same angler catch data from Parkinson et al. (2004). rate quality isopleths for pristine conditions. The resulting fit is presented in Figure 4 and the estimated Selection of optimal stocking rates.—The biological and an- parameter values are presented in Table 1. Overall, the model gler effort processes outlined above will predict angler effort for fit indicated that anglers generally place high value on large fish any given combination of stocking rate and size at release. How- (observed steep decline in catch rates for lakes with increasing ever, fisheries managers must make stocking decisions based on fish size) and relatively low value on released fish (r = 0.16). a fixed hatchery production capacity that may be well below However, since there are no data on lengths of released fish, the production capacity needed to optimize angler effort across the expected trade-off of catch versus size implicitly assumes all lakes available for stocking in their region. In that case, the that released fish are of the same size as retained fish and that primary decision is how to optimally allocate fish supply to maximize total effort across all lakes. We simulated an example scenario of this challenge by assuming a region with equal proportions of warm and cold lakes and by maximizing the ratio of total effort to fish supply for different ratios of fish supply to Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 lake area.

RESULTS The empirical decline of catch rate with average fish size documented by Parkinson et al. (2004) was derived from 38 popular angling lakes that had on average an altitude of approximately 1,100 m, which is more similar, on average, to the environmental conditions of our warm than to those of our cold experimental lakes (Figure 5, upper panel). The observed decline of catch rate with average fish size observed in the fished FIGURE 4. The relationship between catch rates and the average size of lakes was much larger than would be expected in warm lake fish captured for lakes in the southern interior of British Columbia with open pristine conditions (Figure 5, lower panel). This implies that fisheries. Open diamonds are total catch per angler-day, solid diamonds are fish harvested per angler-day, and black dashes represent the effective catch rate the observed trade-off between catch rate and size is not due (ECUE), which weights released fish as having less value. The solid black line solely to an ecological density-dependent effect on population is the best fit line to the effective catch rate data. size structure but rather to the current angler harvest and are, 564 ASKEY ET AL.

doubling of stocking density for 2.5-cm fry from 1,250 to 2,500 fish/ha led to an increase of less than 6 angler-days/ha. How- ever, as the size at release increased, the effort response surface became more defined with large decreases in effort at the extremes where either very few fish were stocked or so many fish were stocked that fish in the population were becoming stunted (average size near or below 20 cm). There was a positive influence of size at release on angler effort over the range of sizes considered. This was expected since there is competition for a common food resource across all considered sizes. Therefore (given no natural reproduction), stocking bigger fish “rests” the lake from production of the early growth (prefishery) biomass, and lakes can sustain higher densities of the bigger, catchable fish. However, this does not account for potential infrastructure constraints or high marginal cost of production. In order to generalize results to other fisheries we repeated the simulations for different lake types, available effort pools, and regulations. Not surprisingly, the predicted optimal stocking densities were lower for cold lakes (Figure 6, lower left panel). However, the decline in optimal stocking rates was much greater than the decline in juvenile growth between the two lake types. For example the optimal stocking rate for 3.5-cm fry was 15% of the warm lakes; however, the annual growth increment of juvenile fish for cold lakes is approximately 50% of that for warm lakes. The disproportionately large decrease in stocking rates arises because the FIGURE 5. The frequency distribution of lake elevations (upper panel) from cold lakes have a smaller scope for fishery production above the creel census data from lakes presented in Figure 4. The solid grey line indicates equal-quality isopleth (see Figure 5). Therefore, higher stocking the altitude (±10 m) of the “warm” experimental lakes and the dashed black rates can lead to stunted average fish sizes (≤20 cm), which are ± lines indicates the elevation ( 10 m) of the “cold” experimental lakes. The not valued by anglers at any catch rate. This is readily apparent lower panel shows the empirical relationship between catch rates and size of fish in the catch observed in the southern interior of British Columbia (solid when comparing absolute effort versus stocking rate (at a given black line) compared with the expected size and catch of fish under pristine release size) for the two lake types (Figure 6). If fish are stocked conditions for warm lakes (solid grey line) and cold lakes (dashed black line) into regions where there is a smaller available angler pool from from the population model. The pristine conditions were generated by stocking which to draw effort, then the equal-quality isopleth rises. We 6.5-cm fry at densities ranging from 10 to 2,010 fish/ha and imposing a trivial considered this by running the model with an isopleth derived and constant annual angler effort of 0.00001 angler-days/ ha. from catch rates from a management region in north-central British Columbia with a smaller pool of anglers (isopleths pa- therefore, well below pristine conditions for most lakes in the rameters: a = 238.73, b =−1.76). We found that stocking rates southern interior of British Columbia (Figure 5). However, the should be reduced in this low effort region to allow for the

Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 fish growth rate in the pristine “cold” lakes led to conditions decrease in harvest rates (Figure 6, lower right panel). Finally, that were only slightly better than the regional average angling stocking rates may also need to be adjusted downward if harvest quality. This indicates that such high altitude, low productivity is decreased for other reasons such as catch-and-release regu- lakes are not likely to attract much angler effort. lations (Figure 6, upper right panel). However, the adjustment The next stage of model simulations proceeded with a dy- is conditioned on the relative value anglers place on released namic effort response to angling quality. Under this scenario, a fish (r) and the release mortality rate. The r estimate obtained broad range of stocking strategies resulted in equivalent angling from assuming released fish were the same size as harvested fish quality (Q always stabilized at a value of 1 within 10 years), appears to be biased low (as expected), since negligible effort is but variable angler effort densities (Figure 6). As expected, generated by catch and release fisheries with the estimated r = density-dependent processes led to an inverse relationship be- 0.16 (results shown for r = 0.25). tween optimal size at release and optimal numbers released Scaling results from the single lake model up to regional allo- (Figure 6). However, density dependence in growth and survival cation of fish among lakes can be done by maximizing the total combined with the angler trade-off of size versus numbers led regional effort for a given supply of fish. We maximized the ratio to a fairly broad region of equally effective stocking policies for of total effort to fish for three different ratios of fish supply to warm lakes. The effort response was least sensitive to changes lake area. When the ratio is low, only productive “warm” lakes in stocking rates for the smallest size-classes. For example, a would be stocked; however, as supply increases a proportion of ASSESSING STOCKING STRATEGIES FOR RECREATIONAL FISHERIES 565

FIGURE 6. The annual angler effort density (angler-days/ha) for a given lake is predicted to vary with size at release, stocking density, and the speciﬁc lake characteristics. Individual panels show predicted angler density for the following: warm lake with harvest (upper left panel); warm lake with catch and release =

Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 (r 0.25) (upper right panel); cold lake with harvest (lower left panel); and warm lake in more remote region (lower right panel).

fish are allocated to cold lakes (Figure 7). In an applied sense, a common feature of spatially structured recreational fisheries the manager must search for the highest marginal gain in angler where catch-related attributes are important in driving angler effort per additional “truckload” of available stock given the behavior (Hunt et al. 2011). Classical studies of heavily fished known effort response surfaces in Figure 6. However, dimin- streams in California illustrated that large changes to stocking ishing returns are expected given the effort response functions, rates led to almost proportional changes in angler effort and un- and the ratio of total effort to total fish stocked decreased from changed catch rates (Vestal 1954; Butler and Borgeson 1965). 0.11 to 0.06 and 0.03 for 250, 750, and 1,500 total fish per total Other studies that have measured effort under different stocking hectare), respectively. densities also find effort responses (Moring 1985, 1993; Cox 2000). Effort responses are often included in recreational fisheries models (Johnson and Carpenter 1994; Beard et al. 2003; DISCUSSION Cox et al. 2003; Parkinson et al. 2004; Post et al. 2008; Carson This analysis was built on the premise that changes in man- et al. 2009; Hunt et al. 2011) but have rarely been incorporated agement actions will often lead to a response in angler effort into stocking models (but see Cowley et al. 2003; Fenichel et al. as opposed to angling quality. Such angler dynamics should be 2010). We present an approach to optimize stocking practices 566 ASKEY ET AL.

sales, maximizing total angler effort). The first step is to define optimal stocking strategies under the current regime and compare this with current stocking practices. However, the present model’s focus on optimization of a single lake allows for two key assumptions: (1) that anglers redistributing to or from a single lake will not have an appreciable effect on the regional angler pool (i.e., no change to the regional isopleth); (2) one lake does not significantly change the range of opportunities available in the overall fishery (given individual anglers may prefer fisheries at different points along the equal-quality isopleth). Conversely, if a manager simultaneously optimizes stocking strategies in all of the lakes in a region then the regional isopleth is expected to change. This violation of the single-lake model assumptions would feed back to invalidate the previous model parameterization and require a reanalysis of the regional isopleth. Thus, regional management is clearly a complex and iterative adaptive FIGURE 7. Optimal allocation of stocked fish between lakes is dependent on management exercise that will be facilitated by modeling and the ratio of total available fish to total available lake area and the productive future research. capacity of different lakes. Lines represent predicted effort response when 6-cm We suggest the initial step of defining an optimal stocking fry are stocked into warm (solid line) or cold (dashed line) lakes. The optimal allocation (maximum total effort) of fish (6 cm at release) between lakes is strategy at a regional scale is to maximize total effort, given the given for different hatchery capacities and assumes equivalent delivery costs, regional suite of lakes (with varying productive capacities) and lake size, and access characteristics of all individual lakes. a finite fish supply. Some qualitative aspects to the solution are apparent from the observed effort response surfaces in Figure 6. that incorporates this intuitive link between the “put” and “take” In the case of highly productive lakes, it is advantageous to of stocked fisheries. spread fish among as many lakes as possible (given equivalent The fundamental shift in assessing optimal stocking strate- delivery costs and lake size), because there is almost an instanta- gies by angler effort as opposed to statistics related to the fish neous response of effort to the presence of fish, and the marginal population (e.g., fish condition, size, or numbers) leads to quite increase in effort per density of stocked fish is greatest near the different conclusions. This logic suggests that managers should origin. Conversely, cold (unproductive) lakes require a substan- not hope to control the size or number of fish captured by tial minimal investment in stocking before effort is generated; anglers through manipulation of stocking rates and sizes alone. thus, stocking low numbers in many lakes could generate less Although, variation in mean size and catch rates may occur with total effort than focusing fish into fewer lakes at higher densities. changes, the end result will always fall somewhere along the Ultimately, even if a manager had the potential to saturate all equal-quality isopleth. Thus, managers cannot hope to improve available hectares of lake to the point of maximum effort, this the angling quality (perceived at the individual level) on a would not be optimal when considering diminishing returns. In single lake by changing stocking practices on that lake because all lake types, our model predicted effort to be a saturating func- of the effort response. Since many anglers are simultaneously tion of stocking rate due to density-dependent ecological pro- and continuously removing fish, one individual can only cesses (we put no limitations on potential angler effort). Thus,

Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 capture what is left over from the others (i.e., catch success at increasing stocking rates toward maximum effort will result in the individual level is rapidly depleted to regional equilibrium an increasing marginal cost. Fully addressing this issue requires even if the lake is yielding many more fish). However, using data for variable costs associated with the different stocking optimal stocking practices is still crucial because (1) the strategies (variable fish production and delivery costs), which number of anglers sustained by that lake will vary, and (2) are not available. However, it should be recognized the most when management policies are adopted on a regional scale, as robust optimal stocking strategy would be derived from an eco- opposed to single lakes, then benefits will arise in the form of nomic analysis to maximize social welfare (e.g., Fenichel et al. either a new isopleth (higher regional quality) or recruitment of 2010) and would incorporate hatchery capacity as a management new anglers into the population. variable (as opposed to a fixed quantity as in our example). These last points highlight the broader regional-scale deci- Further research incorporating economics and angler typol- sions that managers must confront. In the context of stocking, ogy would provide the opportunity to increase the realism managers must decide not only the size and number of fish re- of this model format and potentially reach an absolute opti- leased in a specific lake, but how fish should be spread over the mization scheme for regional management of hatchery opera- landscape among many lakes. An ideal situation would be to tions. The empirical quality isopleths presented by Parkinson create a regional management plan with an optimized stocking et al. (2004), and modified here, are related to the concept of program in order to meet a stated goal (e.g., increase license indifference curves from economics. Indifference curves show ASSESSING STOCKING STRATEGIES FOR RECREATIONAL FISHERIES 567

the different bundles of goods (can be likened to catch size population density and size structure is primarily defined by the and numbers here) to which consumers are indifferent (i.e., angler population abundance and their preferences. Biological derive the same utility). However, the quality isopleths repre- production parameters (including stocking practices) are not sent an emergent property of the interactions among ecological likely to influence angling quality but will affect angler effort. processes, noncatch attributes of the individual fisheries, and Therefore, we must consider management actions in the full the current angler population in the southern interior of British context of ecological and human dynamics. Columbia. It is likely that heterogeneity in angler preferences (different individual-level indifference curves for size and num- REFERENCES bers) within the population of anglers results in differential use Andrews, E. J., and J. E. Wilen. 1988. Angler response to success in the of lakes along the equal-quality isopleth (Johnston et al. 2010). California salmon sportfishery: evidence and management implications. Management initiatives that alter stocking strategies, regula- Marine Resource Economics 5:125–138. Arlinghaus, R. 2006. On the apparently striking disconnect between motivation tions, access, or other noncatch attributes may change the mix and satisfaction in recreational fishing: the case of catch orientation of German of anglers in the population through differential recruitment or anglers. North American Journal of Fisheries Management 26:592–605. exit from the fishery driven by this preference variability (e.g., Askey, P. J. 2007. Towards optimal management of spatially structured, recre- Scrogin et al. 2004). Our model implicitly assumed that anglers ational fisheries: linking ecology and angler dynamics in British Columbia were homogenous in their preferences, and predictions may have Rainbow Trout (Oncorhynchus mykiss) lakes. Doctoral dissertation. Univer- sity of Calgary, Calgary, Alberta. unknown biases given our approach. An alternative, one argued Askey, P. J., S. A. Richards, J. R. Post, and E. A. Parkinson. 2006. Linking for by Fenichel et al. (2013), is to develop an individual-based angling catch rates and fish learning under catch-and-release regulations. mechanistic model that explicitly assesses heterogeneity in pref- North American Journal of Fisheries Management 26:1020–1029. erences among anglers and their recruitment and exit dynamics. Bartholomew, A., and J. A. Bohnsack. 2005. A review of catch-and-release Such an approach should lead to more robust predictions of ag- angling mortality with implications for no-take reserves. Reviews in Fish Biology and Fisheries 15:129–154. gregate angler behavior in response to management initiatives if Beard, T. D., Jr., S. P. Cox, and S. R. Carpenter. 2003. Impacts of daily bag they alter the preference structure of the angler population. Un- limit reductions on angler effort in Wisconsin Walleye lakes. North American fortunately the data do not exist at the present time in this fishery, Journal of Fisheries Management 23:1283–1293. or most others, to adopt this approach (see Hunt et al. 2011 for Butler, R. L., and D. P. Borgeson. 1965. California “catchable” trout fisheries. an exception). We are currently conducting an adaptive man- California Department of Fish and Game, Fish Bulletin 127. Carson, R. T., W. M. Hanemann, and T. C. Wegge. 2009. A nested logit model agement experiment in this fishery in which we have included of recreational fishing demand in Alaska. Marine Resource Economics 24: the type of angler surveys necessary to allow us to contrast our 101–129. current approach with that promoted by Fenichel et al. (2013). Cowley, D. E., F. A. Ward, R. Deitner, and M. D. Hatch. 2003. Optimizing the The importance of angler dynamics does not invalidate the allocation of hatchery-produced fish among multiple stocking sites. North need for good biological understanding. For example, the bio- American Journal of Fisheries Management 23:216–229. Cowx, I. G., K. T. O’Grady, R. L. Welcomme, and D. M. Bartley. 1998. Cur- logical productive capacity of lakes is a key parameter for opti- rent approaches to the enhancement of fisheries. Fisheries Management and mizing stocking strategies. Lakes with low productive capacity Ecology 5:351–382. will have a reduced scope for diversity in size and numbers of Cox, S. P.2000. Angling quality, effort response, and exploitation in recreational catch because they are limited by the maximum size and density fisheries: field and modeling studies on British Columbia Rainbow Trout of fish that they can produce. Ultimately this means that stock- (Oncorhynchus mykiss) lakes. Doctoral dissertation. University of British Columbia, Vancouver. ing rates do not vary proportionally with the individual growth Cox, S. P., and C. Walters. 2002. Modeling exploitation in recreational fisheries rates of fish within lakes. Low productivity lakes must com- and implications for effort management on British Columbia Rainbow Trout

Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 pete with high productivity lakes for angler effort, and therefore lakes. North American Journal of Fisheries Management 22:21–34. stocking rates must be reduced substantially in these lakes to Cox, S. P., C. J. Walters, and J. R. Post. 2003. A model-based evaluation of produce fish sizes that are acceptable to anglers who judge fish active management of recreational fishing effort. North American Journal of Fisheries Management 23:1294–1302. quality on a regional scale. Ecological processes also greatly Donovan, N. S., R. A. Stein, and M. M. White. 1997. Enhancing percid stocking influence effort responses through size and density-dependent success by understanding age-0 piscivore–prey interactions in reservoirs. growth and survival. For example, it may be difficult to observe Ecological Applications 7:1311–1329. effort responses when small size-classes of fish are stocked be- Ellison, D. G., and W. G. Franzin. 1992. Overview of the symposium on Wall- cause density-dependent growth and mortality compensate for eye stocks and stocking. North American Journal of Fisheries Management 12:271–275. changes in stocking rates before fish recruit into the catchable Fayram, A. H., M. J. Hansen, and N. A. Nate. 2005. Determining optimal size-classes (Post et al. 1999; Askey 2007). On the other hand, stocking rates using a stock–recruitment model: an example using Walleye stocking larger juveniles or catchable-sized fish greatly reduces in northern Wisconsin. North American Journal of Fisheries Management the potential for compensatory natural mortality and should re- 25:1215–1225. sult in a much stronger effort response. Fenichel, E. P., J. K. Abbott, and B. Huang. 2013. Modelling angler behaviour as part of the management system: synthesizing a multi-disciplinary literature. Recreational fisheries are similar to commercial fisheries in Fish and Fisheries 14:137–157. that they heavily manipulate the mortality rates and size structure Fenichel, E. P., R. D. Horan, and J. R. Bence. 2010. Indirect management of fish populations (e.g., Sullivan 2003). We argue that fish of invasive species through bio-controls: a bioeconomic model of salmon 568 ASKEY ET AL.

and Alewife in Lake Michigan. Resource and Energy Economics 32:500– Moring, J. R. 1985. Relation of angler catch rates and effort to reductions in 518. numbers of trout stocked in five Oregon streams. North American Journal of Fretwell, S. D., and H. L. Lucas Jr. 1970. On territorial behavior and other Fisheries Management 5:575–579. factors influencing habitat distributions in birds: I. theoretical development. Moring, J. R. 1993. Effect of angling effort on catch rate of wild salmonids Acta Biotheoretica 19:16–36. in streams stocked with catchable-size trout. North American Journal of Gillis, D. M. 2003. Ideal free distributions in fleet dynamics: a behavioral Fisheries Management 13:234–237. perspective on vessel movement in fisheries analysis. Canadian Journal of Parkinson, E. A., J. R. Post, and S. P. Cox. 2004. Linking the dynamics Zoology 81:177–187. of harvest effort to recruitment dynamics in a multistock, spatially struc- Gillis, D. M., R. M. Peterman, and A. V. Tyler. 1993. Movement dynamics in a tured fishery. Canadian Journal of Fisheries and Aquatic Sciences 61:1658– fishery: application of the ideal free distribution to spatial allocation of effort. 1670. Canadian Journal of Fisheries and Aquatic Sciences 50:323–333. Post, J. R., E. A. Parkinson, and N. T. Johnston. 1999. Density-dependent Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: processes in structured fish populations: interaction strengths in whole-lake choice, dynamics and uncertainty. Routledge, Chapman and Hall, New York. experiments. Ecological Monographs 69:155–175. Hunt, L. M. 2005. Recreational fishing site choice models: insights and future Post, J. R., L. Persson, E. A. Parkinson, and T. van Kooten. 2008. Angler opportunities. Human Dimensions of Wildlife 10:153–172. numerical response across landscapes and the collapse of freshwater fisheries. Hunt, L. M., R. Arlinghaus, N. Lester, and R. Kushneriuk. 2011. The effects Ecological Applications 18:1038–1049. of regional angling effort, angler behavior, and harvesting efficiency on land- Post, J. R., M. Sullivan, S. Cox, N. P. Lester, C. J. Walters, E. A. Parkinson, scape patterns of overfishing. Ecological Applications 21:2555–2575. A. J. Paul, L. Jackson, and B. J. Shuter. 2002. Canada’s recreational fisheries: Johnson, B. M., and S. R. Carpenter. 1994. Functional and numerical re- the invisible collapse? Fisheries 27(1):6–17. sponses: a framework for fish-angler interactions? Ecological Applications 4: Santucci, V. J., Jr., and D. H. Wahl. 1993. Factors influencing survival and 808–821. growth of stocked Walleye (Stizostedion vitreum) in a centrarchid-dominated Johnston, F. D., R. Arlinghaus, and U. Dieckmann. 2010. Diversity and com- impoundment. Canadian Journal of Fisheries and Aquatic Sciences 50:1548– plexity of angler behaviour drive socially optimal input and output regulations 1558. in a bioeconomic recreational-fisheries model. Canadian Journal of Fisheries Scrogin, D., K. Boyle, G. Parsons, and A. J. Plantinga. 2004. Effects of regu- and Aquatic Sciences 67:1507–1531. lations on expected catch, expected harvest, and site choice of recreational Kozfkay, J. R., J. C. Dillon, and D. J. Schill. 2006. Routine use of sterile fish in anglers. American Journal of Agricultural Economics 86:963–974. salmonid sport fisheries: are we there yet? Fisheries 31:392–401. Sullivan, M. G. 2003. Active management of Walleye fisheries in Alberta: dilem- Leber, K. M., R. N. Cantrell, and P. S. Leung. 2005. Optimizing cost- mas of managing recovering fisheries. North American Journal of Fisheries effectiveness of size at release in stock enhancement programs. North Management 23:1343–1358. American Journal of Fisheries Management 25:1596–1608. Tredger, C. D. 1992. Estimation of angler effort using index boat counts. British Lester, N. P., B. J. Shuter, and P. A. Abrams. 2004. Interpreting the von Berta- Columbia Ministry of Environment, Lands and Parks, Fisheries Branch, Fish- lanffy model of somatic growth in fishes: the cost of reproduction. Proceed- eries Technical Circular 94, Victoria. ings of the Royal Society of London B 271:1625–1631. Vestal, E. H. 1954. Creel returns from Rush Creek test stream, Mono County, Lorenzen, K. 1995. Population dynamics and management of culture-based California, 1947–1951. California Fish and Game 40:89–104. fisheries. Fisheries Management and Ecology 2:61–73. Walters, C. J., and J. R. Post. 1993. Density-dependent growth and competitive Lorenzen, K. 2000. Allometry of natural mortality as a basis for assessing opti- asymmetries in size-structured fish populations: a theoretical model and rec- mal release size in fish-stocking programmes. Canadian Journal of Fisheries ommendations for field experiments. Transactions of the American Fisheries and Aquatic Sciences 57:2374–2381. Society 122:34–45. Madenjian, C. P., B. M. Johnson, and S. R. Carpenter. 1991. Stocking strate- Whalen, K. G., and G. W. LaBar. 1994. Survival and growth of Atlantic Salmon gies for fingerling Walleyes: an individual-based model approach. Ecological (Salmo salar) fry stocked at varying densities in the White River, Vermont. Applications 1:280–288. Canadian Journal of Fisheries and Aquatic Sciences 51:2164–2169. Downloaded by [Department Of Fisheries] at 20:07 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:08 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Migratory Movements of American Shad in the James River Fall Zone, Virginia Aaron W. Aunins a , Bonnie L. Brown a , Matt Balazik b & Greg C. Garman b a Department of Biology , Virginia Commonwealth University , Life Sciences Building, 1000 West Cary Street, Post Office Box 842012, Richmond , Virginia , 23284-2012 , USA b Center for Environmental Studies , Virginia Commonwealth University , Life Sciences Building, 1000 West Cary Street, Post Office Box 842012, Richmond , Virginia , 23284-2012 , USA Published online: 24 May 2013.

To cite this article: Aaron W. Aunins , Bonnie L. Brown , Matt Balazik & Greg C. Garman (2013): Migratory Movements of American Shad in the James River Fall Zone, Virginia, North American Journal of Fisheries Management, 33:3, 569-575 To link to this article: http://dx.doi.org/10.1080/02755947.2013.768564

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:569–575, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.768564

MANAGEMENT BRIEF

Migratory Movements of American Shad in the James River Fall Zone, Virginia

Aaron W. Aunins and Bonnie L. Brown Department of Biology, Virginia Commonwealth University, Life Sciences Building, 1000 West Cary Street, Post Ofﬁce Box 842012, Richmond, Virginia 23284-2012, USA Matt Balazik and Greg C. Garman* Center for Environmental Studies, Virginia Commonwealth University, Life Sciences Building, 1000 West Cary Street, Post Ofﬁce Box 842012, Richmond, Virginia 23284-2012, USA

American Shad in 1994 in its portion of Chesapeake Bay Abstract (ASMFC 1999). In spite of improved water quality, enforce- The installation of Bosher’s Dam fishway and notches or ment of state and federal moratoria on commercial harvest, a breaches to all dams downstream in the James River, Virginia, long-term supplementation program, and the modification of were completed during 1989–1999 to help restore the river’s alosine populations by providing access to over 400 river kilometers (rkm) some migration barriers, American Shad recovery in the James of historical spawning habitat that had been blocked for about River basin remains stagnant (ASMFC 2007; Hilton et al. 2011). 175 years. We used stationary radiotelemetry receivers in April– Historically, James River American Shad migrated over 530 July 2009 to assess the passage of tagged adult American Shad river kilometers (rkm) upstream (measuring from the Hampton Alosa sapidissima through the fall zone and Bosher’s Dam fishway. Roads Bridge Tunnel near the confluence of Chesapeake Bay Three receivers encompassed 25 rkm from below head of tide to 3 rkm above Bosher’s Dam. Ninety-four American Shad were radio- and the James River mouth in Hampton, Virginia), as far west tagged over 30 d, either at the head of tide (n = 64) or upstream be- as Covington, Virginia (Stevenson 1899; Garman 1992). How- low the Bosher’s Dam fishway (n = 30). No American Shad tagged ever, construction of Bosher’s Dam at rkm 173 in 1823 and the at the head of tide were detected at the base of Bosher’s Dam, and subsequent construction of five other fall zone dams within 14 none were detected above Bosher’s Dam fishway. However, several rkm downstream of Bosher’s Dam, confined American Shad tagged fish released at the base of Bosher’s Dam remained there for days (x¯ = 4.0 d, SD = 5.9 d) without negotiating the fishway. spawning habitat to tidal waters (Weaver et al. 2003; Figure 1). These results suggest that passage at Bosher’s Dam and through Approximately 175 years later (between 1989 and 1993), nat- other fall zone dams needs improvement and that American Shad ural or constructed breaches to the five dams below Bosher’s access to historical spawning habitat remains thwarted. and the construction of Bosher’s Dam Fishway in 1999 restored Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 access to the historical spawning habitat above Bosher’s Dam Chesapeake Bay historically supported important commer- (Virginia Department of Game and Inland Fisheries [VDGIF] cial fisheries for American Shad Alosa sapidissima; over 5,200 2010; Figure 1). Bosher’s Dam fishway was designed with metric tons were harvested in 1897 among all major Virginia American Shad as the target species for passage and has the the- rivers, including the James River (Stevenson 1899; Walburg oretical capacity to pass up to 250,000 American Shad annually and Sykes 1957). However by 1982, Virginia catches had de- (Weaver et al. 2003). Since 1999, a limited number of American clined to less than 450 metric tons (Hilton et al. 2011). Con- Shad are observed annually at the viewing window near the exit current with decreasing commercial catches for the region, of the Bosher’s Dam fishway (Weaver et al. 2003; Fisher 2007), abundance of James River American Shad declined severely but no adults have been documented from the James River above due to the combined effects of overfishing, impediments to up- the Bosher’s Dam fishway since its construction. stream migration, and pollution (Olney et al. 2003; Weaver et al. One hypothesis for the nonrecovery of American Shad 2003). In response, Virginia imposed a harvest moratorium for in the James River basin is that ineffective modifications

*Corresponding author: [email protected] Received April 18, 2012; accepted January 10, 2013 569 570 AUNINS ET AL.

FIGURE 1. Map of the James River study area showing the locations of the land-based telemetry receivers for monitoring the movements of radio-tagged American Shad. The release locations (filled stars) were at the head of tide (base of the fall zone) and the base of Bosher’s Dam. Dams upstream from the base of the fall zone were as follows: 1 = Manchester, 2 = Brown’s Island, 3 = Bell Isle, 4 = Williams Dam South, 5 = Williams Dam North, 6 = Bosher’s Dam. The escapement receiver was located 7.9 rkm below the head of tide (37◦2740.42N, 77◦2512.88W), Bosher’s receiver just below Bosher’s Dam fishway (37◦3335.94N, 77◦3431.00W), and the above-Bosher’s receiver above the Bosher’s Dam fishway (37◦3322.93N, 77◦3618.39W).

to migration barriers may limit the efficient movement of pre- most years (Hilton et al. 2011). The lack of juvenile recruitment reproductive American Shad through the fall zone and into his- downstream of Bosher’s Dam, despite consistent recruitment of torical spawning habitat within upper (nontidal) river reaches. hatchery fish stocked above the dam, suggests that access to From 1994 to 2008, millions (x¯ = 6.94 million, SD = 2.05 upstream habitat is critical if American Shad in James River million) of oxytetracycline (OTC)-tagged American Shad lar- are to become a self-sustaining population. We hypothesize vae derived from broodstock collected from the neighboring that the poor juvenile recruitment downstream of Bosher’s Dam Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 Pamunkey River were annually stocked above Bosher’s Dam by is due to extensive urbanization, channelization, and pollution VDGIF (Olney et al. 2003; VDGIF 2009). Current monitoring over the last several decades that have contributed to the de- data suggest that the American Shad population in James River cline in spawning habitat quality and the collapse of the James is dependent upon these hatchery stockings (Hilton et al. 2011). River American Shad population, while the habitat upstream of Annual monitoring efforts above Bosher’s Dam by VDGIF tar- Bosher’s Dam is capable of supporting self-sustaining levels of geting American Shad juveniles from July–August using elec- juvenile recruitment. Until such time as habitat below Bosher’s trofishing and push-netting consistently collect OTC tagged and other fall zone dams is vastly improved, providing unhin- juveniles; however, no wild juvenile American Shad have been dered passage to higher quality habitat appears to be the best collected since modifications to impediments were completed. short-term goal for successful restoration of American Shad in The lack of wild spawning above Bosher’s Dam could be due James River. to low passage efficiency at Bosher’s Dam fishway or through The main objectives of this study were to (1) evaluate the other modified dams in the fall zone (Figure 1). Furthermore, passage of American Shad through the Bosher’s Dam vertical- the juvenile abundance index for American Shad calculated slot fishway and other breached barriers between the head of from seine haul data by Virginia Institute of Marine Science tide and Bosher’s Dam (Figure 1) and (2) describe the migratory during July–September below the dam is at or close to zero in behavior of adult American Shad on the spawning grounds using MANAGEMENT BRIEF 571

radiotelemetry. We hypothesized that because adult American Shad are routinely collected at the base of Bosher’s Dam during spring months, ﬁsh passage to Bosher’s Dam is not restricted by breached low-head dams within the fall zone and a majority of American Shad radio-tagged at the base of the fall zone would reach the base of Bosher’s Dam. In addition, we hypothesized that some portion of the American Shad we tagged would ascend Bosher’s Dam ﬁshway.

METHODS Study area.—The James River watershed drains an area of 26,000 km2; the river flows approximately 536 rkm from the junction of the Jackson and Cowpasture rivers in Bote- FIGURE 2. James River discharge and temperature data from U.S. Geological tourt County, Virginia, eastward, emptying into Chesapeake Bay Survey gauges 02037500 and 02035000, respectively, from 28 March to 6 June (Garman and Nielsen 1992). The fall zone descends 30 m over 2009. The numbers of American Shad released by date at Bosher’s Dam and the head of tide combined are indicated by the filled circles along the x-axis. 15 km in Richmond (Figure 1) and is characterized by extensive rapids and riffle–pool habitat (Garman and Nielsen 1992). Substrate within the fall zone consists primarily of boulder and tly into the stomach using a hollow plastic tube fitted over the gravel and is interspersed with numerous small and large is- antenna, leaving the antenna trailing out of the mouth. During lands. Tidal influence extends approximately 158 km upstream tagging, total length (mm) and fork length (mm) were recorded to head of tide at Richmond. The area of the James River chosen and individuals were released immediately within 100 m of the for this study is within the primary spawning grounds of Amer- collection site at both sampling areas. Spawning condition and ican Shad, which as identified by Aunins and Olney (2009) sex were visually estimated, but expression of gametes was not extends from the base of the fall zone to 33.5 rkm downstream. done to avoid additional stress during the tagging process. Although Aunins and Olney (2009) did not sample the 14 rkm American Shad movement monitoring.—Radiotelemetry reach between the base of the fall zone and Bosher’s Dam, Amer- equipment (Advanced Telemetry Systems, Isanti, Minnesota) ican Shad ichthyoplankton were collected at the base of Bosher’s included receivers (ATS model R4520C System 2 coded re- Dam in the early 2000s (G. Garman, unpublished data), suggest- ceivers), and radio tags (ATS model F1815, 12 × 36 mm, 7 g). ing spawning habitat extends upstream to the base of Bosher’s Pings per minute were set at 55, resulting in a battery life of ap- Dam. proximately 140 d. Continuous records of radio tag detections Six low-head dams, five of which have been notched or nat- were obtained using stationary, land-based receivers installed urally breached, exist within the fall line of the James River in weatherproof boxes with external antennae. Receivers were (Figure 1; VDGIF 2010). The lowermost dams are the Manch- installed at three locations (Figure 1). The escapement receiver ester and Brown’s Island dams, which both had 30-m breaches was located 7.9 rkm downstream of head of tide to detect Amer- completed in 1989. Upstream is the Belle Isle Dam, which was ican Shad leaving the study area. The Bosher’s receiver was naturally breached by storm events in 1989. Two dams are lo- located 22.5 rkm upstream of the escapement receiver to de- cated 6.2 km further upstream on both sides of Williams Island. tect American Shad that migrated upstream through the fall

Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 The dam in the southernmost channel of Williams Island is zone and associated dams and reached the base of Bosher’s notched for fish passage. A vertical-slot fishway adjacent to the Dam. The Bosher’s receiver was oriented to detect transmitters largest dam within the fall zone (Bosher’s Dam; 272 m × 3m) across the full length of the base of Bosher’s Dam but could has been operated each spring since 1999. not differentiate occurrence in the fishway tailrace versus other Capture and tagging.—Adult American Shad were captured points along the dam. The above-Bosher’s receiver was located by boat electrofishing (Smith-Root SR16 electrofishing boat, 2.8 rkm upstream of the dam to detect tagged shad that migrated Vancouver, Washington), tagged, and released at the head of successfully through the fishway. At each site, range tests were tide (n = 64) or immediately below the Bosher’s Dam fishway performed to ensure tags could be detected across the full width (n = 30) on eight dates from 2 to 29 April 2009 (Figure 2). of the river. The period of receiver deployment was 28 March to Immobilized shad were dipnetted directly into a rectangular on- 1 July 2009, which included the period of upstream migration board holding tank (1.25 × 0.7 × 0.7 m) supplied with fresh by adult (prespawn) American Shad in this river reach (based on ambient river water (temperature range: 11.8–24.6◦C; Figure 2). Fisher 2007; Aunins and Olney 2009). Every 3–7 d, telemetry Holding times for individual shad prior to tagging and release receiver batteries were replaced and detection data were down- did not exceed 10 min. Shad were removed from the holding loaded onto a laptop computer. tank and placed onto a wet measuring board. Radio tags were Data analyses.—Temporal and spatial plots of receiver de- coated with glycerin, placed in the esophagus, and pressed gen- tection data were generated for all tagged American Shad to 572 AUNINS ET AL.

depict patterns of movement among the three receivers. The pe- base of the dam; it was detected continuously at the Bosher’s riod of detection within the receiver array of tagged shad was receiver for the entire study period. No shad tagged and released calculated as the time between release and last detection at the at Bosher’s Dam were detected at the receiver above the dam, escapement receiver. Shad that were not detected at the escape- resulting in zero passage of tagged shad through Bosher’s Dam ment receiver did not have a period of detection calculated and fishway. were assumed to have expired or been removed from within the Although no shad were detected above Bosher’s Dam, 29 study area. Because it was not possible to quantify time spent tagged shad were detected for variable periods at the base of within the study area prior to tagging, estimates of periods of the dam, suggesting that some fish were searching for a means detection probably underestimate actual residence times within of passage. The period between the first and last detection by the James River system. the Bosher’s Dam receiver ranged from 0.0 to 29.0 d (x¯ = From James River discharge data (Figure 2) we calculated 4.0 d, SD = 5.9, n = 29). Detection patterns at the Bosher’s the 25th and 75th percentile of river discharge over the period Dam receiver could be generally categorized into three groups. of 1934–2008 at the same station for the period of our receiver The first group consisted of American Shad (n = 13) that pro- deployment. We considered flows between the 25th and 75th ceeded downstream within 2 d after tagging, of which 31% percentiles as normal, above the 75th percentile as high flow, were detected at the escapement receiver (Figure 3a). The sec- and below the 25th percentile as low. Water temperature data ond group (n = 9) remained at the base of Bosher’s Dam for (Figure 2) were obtained from a recording site upstream of more than 2 d with no breaks in detection lasting longer than Bosher’s Dam. 1 d, before emigrating downstream (Figure 3b); 44% of these shad were detected at the escapement receiver. The third group (n = 7) made downstream forays evidenced by gaps in detection RESULTS at the Bosher’s receiver greater than1dwithsubsequent return River Discharge and Temperature trips to the base of Bosher’s Dam, of which 57% eventually River discharge was high (exceeded the 75th percentile) dur- were detected at the escapement receiver (Figure 3c). ing numerous occasions throughout the study period (Figure 2), although flow was never so great as to submerge Bosher’s Dam (i.e., >990 m3/s according to Fisher 2007). Average discharge DISCUSSION during the period was 305 m3/s (SD = 155, range = 136– No American Shad tagged and released at Bosher’s Dam 807 m3/s). The mean water temperature from the first release were detected at the receiver above the dam. Similarly, no shad until the last detection of any tagged American Shad was 17.9◦C tagged and released at the head of tide traversed 22 rkm upstream (SD = 3.5, range = 11.8–24.6◦C; Figure 2). to the base of Bosher’s Dam. These results suggest that breached dams within the fall zone as well as Bosher’s Dam still impede Post-Tagging Movements passage of migrating American Shad, despite provisions for fish Of the 64 American Shad released at the head of tide, 63 passage. This finding does not conflict with the annual detection (97%) were eventually detected downstream at the escapement of untagged American Shad at the base of Bosher’s Dam, or in receiver. The period of detection within the receiver array ranged the Bosher’s Dam fishway. Rather, our results suggest that the from 0.1 to 22.4 d (x¯ = 3.0 d, SD = 4.5). The number of days number of American Shad detected at the base of Bosher’s Dam to first detection at the escapement receiver ranged from 0.1– would be greater if there were no downstream dams. In light of 5.6 d (x¯ = 0.8 d, SD = 0.9). A majority (45/63) of tagged our results, continued hatchery supplementation above Bosher’s

Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 shad were detected at the escapement receiver within 24 h after Dam in the absence of improving upstream passage through the release. Fourteen of those detected at the escapement receiver fall zone is not likely to establish a self-sustaining population, continued to remain in the system for 1.0–22.4 d (x¯ = 7.8 d, a focal objective of the Virginia American Shad Restoration SD = 6.1) as evidenced by multiple subsequent detections at Program. the escapement receiver, whereas the remaining 31 were not The reasons for the lack of passage of tagged shad through detected again and therefore assumed to have exited the study Bosher’s Dam fishway are speculative. The river flows and water area. No shad released at the head of tide location were detected temperatures we observed over the study period were within the upstream at Bosher’s Dam. ranges (44–894 m3/s and 14–30◦C) observed for American Shad Of the 30 American Shad released at the base of Bosher’s passage through Bosher’s Dam fishway over the 5 years exam- Dam, 12 (40%) were detected at the escapement receiver, two ined by Fisher (2007), and therefore should not have prevented within 24 h after tagging. The period of detection within the passage. One reason for zero passage may be that American receiver array for these 12 fish ranged from 0.6 to 14.5 d (x¯ = 7.2 Shad have difficulty locating the attraction flow and entrance to d, SD = 4.7). The remaining 18 were not detected downstream at the fishway. Barry and Kynard (1986) monitored the movements the escapement receiver during the study period and are assumed of 18 American Shad at the entrances to fish lifts on Connecticut to have expired, or been removed from within the study area. River and found a delay of 2–7 d before successful passage. They One shad tagged and released at Bosher’s Dam likely died at the concluded that turbulent water from one of the hydroelectric MANAGEMENT BRIEF 573 Downloaded by [Department Of Fisheries] at 20:08 28 May 2013

FIGURE 3. Representative detection patterns of radio-tagged American Shad released on 27 April 2009 at the base of Bosher’s Dam in the James River, as derived from three stationary radiotelemetry receivers: the escapement (E), Bosher’s (B), and above-Bosher’s (A) receivers (the spacing along the vertical axis indicates the approximate distances between them; see Figure 1). Adjacent left and right panels pertain to two representative American Shad chosen to illustrate individuals that (a) migrated downstream from Bosher’s Dam within 2 d of tagging (n = 13), (b) remained below the dam for two or more days before emigration (n = 9), and (c) remained in the vicinity of Bosher’s Dam for >2 d with periodic downstream forays (n = 7). Each open circle represents a single detection event, and the thin lines represent periods of no detection. 574 AUNINS ET AL.

turbines and spillage over the Holyoke Dam was repelling fish external appearance, some fish we selected for tagging may have from the lift entrances. Although the Connecticut River fish lifts been spent or partially spent. We hypothesize that spent or par- are of a different design than the Bosher’s Dam vertical-slot fish- tially spent fish would be less likely to continue upstream or stay way, turbulent water coming over Bosher’s Dam is a frequent on the spawning grounds after tagging than would prespawn- occurrence during high spring flows and may overwhelm the ing fish; however, because of the difficulty in reliably assessing fishway’s attraction flow. Future telemetry studies of passage maturation stage by visual examination alone, this possibility at Bosher’s Dam should strive for more fine-scale resolution of remains untested. movements below the dam to assess whether American Shad can We anticipated that some shad tagged at the head of tide locate the attraction flow and to determine retention time within would be detected upstream at Bosher’s Dam. James River has the area for those that do. This behavioral information could then been stocked annually since 1994 with millions of hatchery- be used to potentially modify the attraction flow at Bosher’s Dam reared American Shad larvae from the neighboring Pamunkey fishway to improve passage. While Bosher’s Dam vertical-slot River, 90% of which are released above Bosher’s Dam (Brown fishway incorporates many of the design features presumed to et al. 2000; Olney et al. 2003; VDGIF 2009). Managers of the maximize American Shad passage, there have been few quanti- Virginia American Shad Restoration Program have reasoned tative studies of the suitability of vertical-slot fishways for their that American Shad stocked above the dam as well as any nat- passage (Haro and Castro-Santos 2012). Therefore, research not urally spawned progeny would use Bosher’s Dam Fishway in only into optimal attraction flow, but other factors, such as tran- subsequent years (VDGIF 1994). The proportion of American sit time through the fishway, should also be the focus of future Shad with otolith oxytetracycline hatchery marks at head of tide investigations. from 2000 to 2008 collected by VDGIF in annual monitoring av- A majority (71%) of the shad we tagged and released at the eraged 73% (VDGIF 2009). Of the 64 shad tagged and released head of tide were detected at the escapement receiver within in this same area, we therefore expected that more than half 24 h of release, most of which were not detected again within would be hatchery returns and would have continued upstream the study area. This “fallback” movement downstream after to Bosher’s Dam to pass through the fishway. Our expectation tagging has been observed repeatedly in alosine telemetry stud- is supported by the findings of Hendricks et al. (2002), who ies (Beasley and Hightower 2000; Sprankle 2005; Aunins and observed that predominantly hatchery American Shad stocked Olney 2009). It is generally assumed that fallback behavior is a above fish ladders in the Lehigh River, Pennsylvania, used the consequence of tagging and handling stress, although a recent ladders in subsequent years. In addition, Burdick and Hightower study by Frank et al. (2009) suggests downstream movements (2006) found that wild American Shad readily moved past a after handling are not necessarily a stress response. Frank et al. removed dam in the Neuse River, North Carolina, suggesting (2009) observed radio-tagged Alewives A. pseudoharengus that wild American Shad in James River should move upstream to made short upstream movements (about 1–2 km) coupled with Bosher’s Dam fishway. Reasons why none of the radio-tagged downstream movements. Taken in context, these movements shad were detected upstream at Bosher’s Dam could include suggested active searching behavior as opposed to rapid down- difficulty in passage through the fall zone dams, tagging and stream movement out of the study area. In our study, the lack of handling stress, or selection of habitat at the head of tide for an upstream receiver within 1–3 km to fish tagged and released spawning. The use of additional receivers in future telemetry at head of tide, as was used by Frank et al. (2009), prevented our studies could help to more accurately characterize movements interpretation of posttagging movements as probable searching at the head of tide and through the fall zone. behavior. We were only able to detect upstream movement if Given recent evidence that hatchery supplementation of

Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 a shad released at head of tide traveled 22 km upstream to the American Shad populations may not result in self-sustaining Bosher’s Dam receiver, so if any made short upstream excur- populations (Hasselman and Limburg 2012), there is revived sions coupled with downstream movements, we were unable to impetus to maximize the efficiency of fishways and to improve detect them. habitat quality. Our study results suggest that improving passage The lack of upstream movements of shad tagged at the head through the fall zone and at Bosher’s Dam fishway, in partic- of tide could have been due to some fish spawning at the head ular, would be beneficial for American Shad in James River of tide region. Aunins and Olney (2009) documented American by increasing the probability of natural reproduction in more Shad spawning activity at the head of tide in the James River suitable upstream spawning habitat, thereby mitigating reliance and extending 33.5 rkm downstream. A related explanation for of the James River population on hatchery supplementation. In lack of upstream movement could be inadvertent tagging of addition, our results highlight the need in any anadromous fish postspawn fish. American Shad are batch spawners (Olney and restoration program to assess fish passage structure efficacy (as McBride 2003), and those captured at head of tide may com- opposed to merely documenting the passage success of the tar- prise a mixture of maturing, spawning, and postspawning fish, get species) prior to extensive supplementation, so as to ensure any of which could have been captured by our electrofishing. Al- that both hatchery and wild recruits have sufficient access to though we strove to tag gravid females and large males, based on historical spawning habitat. MANAGEMENT BRIEF 575

ACKNOWLEDGMENTS Martin, editors. Biodiversity of the southeastern United States: aquatic com- This was a cooperative project among Virginia Common- munities. Wiley, New York. wealth University, the U.S. Fish and Wildlife Service (Grant Haro, A., and T. Castro-Santos. 2012. Passage of American Shad: paradigms and realities. Marine and Coastal Fisheries: Dynamics, Management, and FONS 52330-2007-014), and the Fish America Foundation. We Ecosystem Science 4:252–261. are grateful to Alan Weaver and his field crew (Virginia De- Hasselman, D. J., and K. E. Limburg. 2012. Alosine restoration in the 21st partment of Game and Inland Fisheries) for assisting with the century: challenging the status quo. Marine and Coastal Fisheries: Dynamics, capture and tagging of American Shad throughout the study Management, and Ecosystem Science 4:174–187. area, especially at the base of Bosher’s Dam. Comments from Hendricks, M. L., R. L. Hoopes, D. A. Arnold, and M. L. Kaufmann. 2002. Homing of hatchery-reared American Shad to the Lehigh River, a tributary two anonymous reviewers and Joe Hightower significantly im- to the Delaware River. North American Journal of Fisheries Management proved the quality of this paper. The following landowners pro- 22:243–248. vided access to their property for the placement of radioteleme- Hilton, E. J., R. J. Latour, B. E. Watkins, and A. M. Rhea. 2011. Monitoring try receivers: Mr. and Mrs. Dean Williams (HMU, Inc.) and relative abundance of American Shad in Virginia rivers. 2010 Annual Report Chris Klotz (Commodore, Virginia Power Boat Association). to the Virginia Marine Resources Commission, Contract F116-R-13, Virginia Institute of Marine Science, Gloucester Point. Olney, J. E., D. A. Hopler Jr., T. P. Gunter Jr., K. L. Maki, and J. M. Hoenig. REFERENCES 2003. Signs of recovery of American Shad in the James River, Virginia. ASMFC (Atlantic States Marine Fisheries Commission). 1999. Amendment 1 to Pages 323–329 in K. E. Limburg and J. R. Waldman, editors. Biodiversity, the Interstate Fishery Management Plan for shad and river herring. ASMFC, status, and conservation of the world’s shads. American Fisheries Society, Fishery Management Report 35, Washington, D.C. Symposium 35, Bethesda, Maryland. ASMFC (Atlantic States Marine Fisheries Commission). 2007. American Shad Olney, J. E., and R. S. McBride. 2003. Intraspecific variation in batch fecundity stock assessment report for peer review, volume 1. ASMFC, Stock Assess- of American Shad: revisiting the paradigm of reciprocal latitudinal trends ment Report 07-1, Washington, D.C. in reproductive traits. Pages 185–192 in K. E. Limburg and J. R. Waldman, Aunins, A., and J. E. Olney. 2009. Migration and spawning of American Shad editors. Biodiversity, status, and conservation of the world’s shads. American in the James River, Virginia. Transactions of the American Fisheries Society Fisheries Society, Symposium 35, Bethesda, Maryland. 138:1392–1404. Sprankle, K. 2005. Interdam movements and passage attraction of American Barry, T., and B. Kynard. 1986. Attraction of adult American Shad to fish lifts Shad in the lower Merrimack River main stem. North American Journal of at Holyoke Dam, Connecticut River. North American Journal of Fisheries Fisheries Management 25:1456–1466. Management 6:233–241. Stevenson, C. H. 1899. The shad fisheries of the Atlantic coast of the United Beasley, C. A., and J. E. Hightower. 2000. Effects of a low-head dam on the States. Pages 101–269 in Report of the Commissioner for 1898. U.S. Com- distribution and characteristics of spawning habitat used by Striped Bass and mission of Fish and Fisheries, Washington, D.C. American Shad. Transactions of the American Fisheries Society 129:1316– VDGIF (Virginia Department of Game and Inland Fisheries). 1994. Virginia’s 1330. American Shad restoration project, 1994. VDGIF, Federal Aid in Fish Brown, B. L., T. P.Gunter, J. M. Waters, and J. M. Epifanio. 2000. Evaluating ge- Restoration, Project F-119-R2, Final Report, Richmond. netic diversity associated with propagation-assisted restoration of American VDGIF (Virginia Department of Game and Inland Fisheries). 2009. Virginia’s Shad. Conservation Biology 14:294–303. American Shad restoration project, January 1, 2008–December 31, 2008. Burdick, S. M., and J. E. Hightower. 2006. Distribution of spawning activity by VDGIF, Federal Aid in Sport Fish Restoration, Project F-123-R7, Final Re- anadromous fishes in an Atlantic Slope drainage after removal of a low-head port, Richmond. dam. Transactions of the American Fisheries Society 135:1290–1300. VDGIF (Virginia Department of Game and Inland Fisheries). 2010. Fish passage Fisher, M. T. 2007. Temporal and spatial patterns of anadromous fish passage at program. VDGIF, Richmond. Available: www.dgif.virginia.gov/fishing/fish- Boshers Dam vertical slot fishway on the James River, Richmond, Virginia. passage/#james. (December 2011). Master’s thesis. Virginia Commonwealth University, Richmond. Walburg, C. H., and J. E. Sykes. 1957. Shad fishery of Chesapeake Bay with Frank,H.J.,M.E.Mather,J.M.Smith,R.M.Muth,J.T.Finn,andS.D. special emphasis on the fishery of Virginia. U.S. Fish and Wildlife Service McCormick. 2009. What is “fallback”?: metrics needed to assess telemetry Research Report 48. tag effects on anadromous fish behavior. Hydrobiologia 635:237–249. Weaver, L. A., M. T. Fisher, B. T. Bosher, M. L. Claud, and L. J. Koth. Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 Garman, G. C. 1992. Fate and potential significance of postspawning anadro- 2003. Boshers Dam vertical slot fishway: a useful tool to evaluate Ameri- mous fish carcasses in an Atlantic coastal river. Transactions of the American can Shad recovery efforts in the upper James River. Pages 339–347 in K. E. Fisheries Society 121:390–394. Limburg and J. R. Waldman, editors. Biodiversity, status, and conservation Garman, G. C., and L. A. Nielsen. 1992. Medium-sized rivers of the Atlantic of the world’s shads. American Fisheries Society, Symposium 35, Bethesda, coastal plain. Pages 315–349 in C. T. Hackney, S. M. Adams, and W. H. Maryland. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:08 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Angler Characteristics and Management Implications in a Large, Multistock, Spatially Structured Recreational Fishery Hillary G. M. Ward a , Michael S. Quinn b & John R. Post a a Department of Biological Sciences , University of Calgary , 2500 University Drive Northwest, Calgary , Alberta , T2N 1N4 , Canada b Institute for Environmental Sustainability , Mount Royal University , 4825 Mount Royal Gate Southwest, Calgary, Alberta , T3E 6K6 , Canada Published online: 24 May 2013.

To cite this article: Hillary G. M. Ward , Michael S. Quinn & John R. Post (2013): Angler Characteristics and Management Implications in a Large, Multistock, Spatially Structured Recreational Fishery, North American Journal of Fisheries Management, 33:3, 576-584 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785991

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:576–584, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785991

ARTICLE

Angler Characteristics and Management Implications in a Large, Multistock, Spatially Structured Recreational Fishery

Hillary G. M. Ward* Department of Biological Sciences, University of Calgary, 2500 University Drive Northwest, Calgary, Alberta T2N 1N4, Canada Michael S. Quinn Institute for Environmental Sustainability, Mount Royal University, 4825 Mount Royal Gate Southwest, Calgary, Alberta T3E 6K6, Canada John R. Post Department of Biological Sciences, University of Calgary, 2500 University Drive Northwest, Calgary, Alberta T2N 1N4, Canada

Abstract Management of recreational fisheries involves understanding how anglers interact with the fishery resource. Managers must understand the source (spatial distribution), efficiency, and behavior of angler effort in order to develop optimal management strategies in a social–ecological framework. We interviewed anglers (n = 1,956) and assessed fish populations in 21 lakes that are part of a multistock, spatially structured fishery for Rainbow Trout Oncorhynchus mykiss in the interior of British Columbia, Canada. Our objective was to assess the spatial behavior, harvest behavior, and catch efficiency of anglers and to understand the strengths of interactions between anglers and fish populations in three regions within this large fishery. Our results suggest a diverse angler population that varied in its behavior and its impact on the fishery. Using a hierarchical cluster analysis, we identified four distinct angler groups based on three variables that directly described how anglers interacted with the fishery (spatial distribution, catchability, and harvest behavior). Angler characteristics varied between groups, and the relative proportions of the four discrete angler groups varied among management regions. Substantial variation in angler characteristics across groups and variation in the relative distribution of the groups across regions imply that a “one size fits all” management approach is not optimal for this fishery. Instead, strategies that are attuned to angler characteristics would constitute a better approach for managing this large, spatially structured fishery. Downloaded by [Department Of Fisheries] at 20:08 28 May 2013

Recreational fisheries are complex systems governed by the 2004), and these concepts have been more recently applied interaction of ecological and social processes. Ecological pro- to the management of recreational fisheries (Radomski 2003; cesses, such as recruitment dynamics, growth, and survival, are Allen et al. 2009). Similarly, human dimensions research on responsible for determining fish supply and production (Shuter recreational fisheries often aims to answer questions regarding et al. 1998; Parkinson et al. 2004), whereas social processes angling behavior—specifically to understand where, when, influence angler behavior and corresponding impacts on the and how much anglers fish and the factors influencing harvest fishery (Hunt et al. 2011). A substantial amount of research decisions among anglers (Fedler and Ditton 1994; Hunt 2005). has been conducted on modeling the ecological component However, these social and ecological processes do not operate of fisheries (Hilborn and Walters 1992; Walters and Martell independently, and several authors have recently suggested

*Corresponding author: [email protected] Received April 18, 2012; accepted March 10, 2013

576 ANGLER CHARACTERISTICS IN A RECREATIONAL FISHERY 577

that the management of recreational fisheries needs to focus on understanding and quantifying the dynamic interactions within this social–ecological system (Johnston et al. 2010; Hunt et al. 2011; van Poorten et al. 2011; Fenichel et al. 2013). Strategies for the management of recreational fisheries are moving away from the conventional single-stock paradigm and shifting toward the idea that individual fisheries are nested within a broader geographical landscape of angling opportunities (Shuter et al. 1998; Lester et al. 2003; Carpenter and Brock 2004; Post et al. 2008; Post and Parkinson 2012). Across a spatially structured fishery, anglers are likely to vary in skill level and harvest behaviors, and these anglers are not randomly distributed across the landscape. Hunt (2005) suggested that fishing site and participation choices vary across the angler population and involve six general attributes: costs, fishing quality, environmental quality, facility development, encounters with other anglers, and regulations. Similarly, it is well understood that anglers’ impacts on stocks vary greatly (Jones et al. 1995); angler experience varies substantially across the angler population, and angler skill level is strongly correlated with catch rates (Bannerot and Austin 1983; Fisher 1997). Exploitation results not only from the intensity of angling effort but also from the efficiency of the effort. Additionally, the propensity of anglers FIGURE 1. Locations of the three British Columbia fishery management re- to harvest fish varies across the population of anglers and is gions included in this study (region 3 = Thompson; region 7 = Omineca; region often correlated with angling specialization (Bryan 1977; Dit- 8 = Okanagan). ton et al. 1992; Johnston et al. 2010; Beardmore et al. 2011). This heterogeneity in angler characteristics results in the need to quantify the source (spatial distribution), efficiency, and behav- 1 fish. Cobb, Eena, and Vivian lakes were also stocked annually ior of effort in order to develop empirically based management with Brook Trout Salvelinus fontinalis. strategies for a fishery. Angler surveys were conducted at each lake between 1000 The Rainbow Trout Oncorhynchus mykiss fishery in British and 1800 hours from April 1 to August 31 in either 2010 or Columbia is an example of a multistock, spatially structured, 2011. Survey days were randomly stratified among lakes be- open-access fishery (Cox and Walters 2002; Parkinson et al. tween weekends and weekdays; more intense survey effort oc- 2004; Post et al. 2008). There is a substantial amount of incurred in the spring, when angling pressure was greatest. All formation on ecological processes that delimit the availability study lakes had single access points and no private housing of resources in this system (Walters and Post 1993; Post et al. (with the exception of Vivian and Eena lakes), ensuring that 1999; Parkinson et al. 2004), but the understanding of the spatial all anglers had an equal opportunity of being interviewed. An- distribution of anglers, efficiency of angling effort, and resultant glers that were 18 years of age and older were interviewed before

Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 harvest mortality for this fishery is more limited. Our goal for their trip and were asked questions relating to demographics and this research was to assess the spatial behavior, harvest behav- catch expectations. The same anglers were then interviewed af- ior, and catch efficiency of anglers across this large, spatially ter their trip and were asked to report on catch success (Table 2). structured fishery to understand the strengths of interactions in All anglers that completed a fishing trip within the duration of this social–ecological system. the survey day were approached for an interview, and if there was more than one angler in a party, each angler was interviewed separately. Fish density in each lake was estimated by using a standard METHODS gill-net sampling protocol in the fall after the angler survey. Study design and data collection.—We examined anglers and Two multimesh gangs of gill nets were set overnight in the fish populations from 21 lakes within three of the management littoral and pelagic habitat of each lake. This gill-net design is regions in the interior of British Columbia, Canada (Figure 1). highly size-selective against small fish and is essentially non- We selected lakes to represent the range of physical character- size-selective for larger fish (Askey et al. 2007). Fork lengths of istics and management conditions (i.e., bag limits) that exist all captured fish were measured and recorded. Estimates of fish within this fishery (Table 1). All study lakes were monocultures density were adjusted to account for angling vulnerability. Cox of Rainbow Trout and were annually stocked with age-0 or age- (2000) demonstrated that Rainbow Trout in similar lakes within 578 WARD ET AL.

TABLE 1. Physical characteristics, locations, bag limits, and sampling years for 21 study lakes in the British Columbia Rainbow Trout ﬁshery.

Management Year Lake Latitude (N) Longitude (W) Area (ha) Elevation (m) region Bag limit sampled Burnell 49◦1227.482 119◦3634.683 12.7 730 Okanagan 0 2011 Cobb 53◦5717.437 123◦3344.378 215.0 771 Omineca 5 2011 Crown 50◦4958.552 121◦4136.028 7.6 807 Thompson 5 2010 Doreen 50◦0705.081 119◦0807.919 44.7 1,358 Okanagan 5 2011 Eena 54◦0312.848 123◦0112.038 51.0 762 Omineca 5 2011 Flyﬁsh 50◦0529.820 119◦0825.125 29.2 1,354 Okanagan 5 2011 Gypsum 50◦2101.971 120◦5204.135 14.0 1,458 Thompson 5 2010 Idleback 49◦3034.120 119◦1727.906 11.6 1,440 Okanagan 1 2011 Jackpine 49◦5508.298 119◦4816.438 42.9 1,200 Okanagan 5 2011 Kentucky 49◦5423.028 120◦3353.146 36.0 1,000 Okanagan 5 2011 Kidd 49◦5512.227 120◦3736.824 18.8 1,058 Okanagan 0 2011 Leonard 49◦4921.101 120◦2600.272 11.9 1,344 Okanagan 2 2011 Loon 50◦0613.894 119◦0809.166 8.5 1,355 Okanagan 5 2011 McConnell 50◦3134.580 120◦2716.660 32.4 1,285 Thompson 5 2011 Pat 50◦4413.694 120◦4418.819 8.1 602 Thompson 2 2011 Ripley 49◦1429.589 119◦3758.634 5.7 923 Okanagan 5 2011 Stake 50◦3053.623 120◦2850.222 23.1 1,320 Thompson 5 2011 Turquoise 50◦4943.095 121◦419.622 6.5 808 Thompson 5 2010 Tyner 50◦1717.201 120◦5548.987 18.1 1,332 Thompson 5 2010 Vinson 49◦4802.286 120◦2621.253 20.5 1,374 Okanagan 1 2011 Vivian 54◦0134.861 123◦1257.503 45.0 779 Omineca 5 2011

this same area exhibited a length at 50% vulnerability (L50)of where m, the steepness of the curve at L50, is equal to 7 (Cox between 214 and 330 mm; we used an average value of 264 mm 2000). Fish density for each lake (DL) was calculated as for L50 in this analysis. Fish captured in gill nets were grouped into 10-mm size-classes, and vulnerability at length (vl)forthe lmax midpoint of the size-bin (l) was calculated as DL = vl Nl , (2) l lm v = , (1) where Nl is the population estimate for length-bin l. Additional l m + m l L50 sampling details are described by Ward et al. (2012). Angler characteristics.—It is well known that recreational

Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 anglers have diverse characteristics and behaviors that affect TABLE 2. Survey questions that were used to determine characteristics of optimal management strategies (Fedler and Ditton 1994; Fisher anglers participating in the Rainbow Trout fishery. 1997; Johnston et al. 2010). We conducted an exploratory analysis of angler characteristics in this fishery. Specifically, we chose Question type Question to focus the majority of our analysis on angler characteristics Pre-trip 1. How many days did you go fishing in the that directly impact the dynamic interaction of anglers with the previous 2 years? fishery: the source (spatial distribution of anglers), efficiency 2. What percentage of days fished in the (angler catchability [q]), and behavior (propensity of anglers previous 2 years were overnight trips? to harvest fish) of effort. The spatial distribution of anglers in 3. What is your postal code? this fishery was explored by calculating the distance traveled 4. How many fish do you expect to catch? by the interviewed anglers between their home (as indicated 5. How many fish do you expect to keep? by postal code) and the lake. Catchability is analogous to the Post-trip 1. How long did you fish for? capture efficiency of the angler, and we used q as a measure of 2. How many fish did you catch? an individual’s success because it provides a better assessment 3. How many fish did you keep? than CPUE across lakes that vary in fish density. Catchability is the parameter that links CPUE to fish abundance (N)inthe ANGLER CHARACTERISTICS IN A RECREATIONAL FISHERY 579

form CPUE = qN. Catchability for an individual angler was we used ANOVA followed by Tukey’s test for multiple compar- calculated as a measure of that angler’s success rate, isons to assess between-group differences in the distance traveled to the lake, q, and the proportion of catch that was harvested. CPUEi,L In addition to comparing differences in the three variables used qi,L = , (3) DL in the cluster analysis, we also compared three other variables that were descriptive of anglers. We used two variables to de- where qi,L is the relative catchability for angler i in lake L.Har- scribe the frequency of fishing: angler avidity (number of days vest behavior of anglers was explored by calculating the ratio fished per year) and the proportion of days fished in the previous between observed harvest and catch. In addition to understand- year that were overnight trips. Additionally, we compared an- ing how harvest behavior varies with other angler characteristics gler CPUE among groups. We chose not to include these three in this fishery, we also developed a linear model to relate the variables (days fished per year, proportion of overnight trips, proportion of catch that is harvested to the bag limit and CPUE and CPUE) in the cluster analysis because they are descriptive across lakes. variables that do not directly affect the interaction of anglers We used multivariate techniques to assign individual anglers with the fishery. to groups (clusters) based on three variables: distance traveled We also compared angler characteristics and the distribution to the lake, the proportion of catch that was harvested, and q.As- of anglers within the identified groups among the three man- signing anglers to specific groups by using this method creates agement regions. A chi-square analysis was used to test the hy- clusters of individuals that are more similar to each other than pothesis that the proportions of anglers belonging to the discrete to individuals in other clusters. We used an analysis similar to groups differed among the management regions. All statistical Chipman and Helfrich (1988) to assess the existence of discrete analyses were conducted in R version 2.13.1 and associated angler groups. Each of the three variables was transformed to a packages. value between 0 and 1 scaled across variables to ensure similar weighting in the analysis. These variables were then used in a RESULTS hierarchical cluster analysis based on a Euclidean distance ma- We interviewed 2,498 anglers across 21 lakes in British trix and Ward’s method to classify anglers (hclust package in R Columbia to quantitatively assess the interaction of anglers with version 2.13.1). The optimal number of clusters to extract was the Rainbow Trout fishery in the three management regions. In- determined by comparing the Dunn index, normalized gamma complete interviews were discarded from the data set, and the coefficient, average silhouette width, and within-cluster sum of analysis was conducted on the remaining 1,956 interviews. Uni- squares across a range of cluster sizes (Halkidi et al. 2001; variate analyses of the data showed that the spatial distribution Meila˘ 2007). Optimal solutions for the Dunn index, normalized (distance traveled to the lake), harvest behavior (proportion of gamma coefficient, and average silhouette width are considered catch that was harvested), and catch efficiency (catchability) to be the number of clusters that corresponds to the maximum varied substantially across the interviewed anglers (Figure 2). value. The optimal number of clusters for the within-cluster sum The distribution of the distance traveled to the lake by anglers of squares is identified by a rapid decrease in the slope. was bimodal (Figure 2a), suggesting that the angler population We explored how the mean and statistical significance of an- was composed of both local and nonlocal anglers. A split of gler characteristics varied among these discrete angler groups; the angler population into nonlocal and local groups (defined Downloaded by [Department Of Fisheries] at 20:08 28 May 2013

FIGURE 2. Frequency distributions (expressed as a proportion of interviewed anglers) for characteristics of anglers in the British Columbia Rainbow Trout ﬁshery: (a) distance traveled from the angler’s home to the study lake, (b) the proportion of catch that was harvested, and (c) catchability. The x-axis on graph (a) is on a log scale, and the axis labels have been back-transformed for interpretation. 580 WARD ET AL.

TABLE 3. Characteristics of the four angler groups identified by the cluster analysis of Rainbow Trout fishery survey data. Means are presented for each angler group. Within a given row, means with the same lowercase letter are not significantly different at P < 0.05 as indicated by Tukey’s test for multiple comparisons.

Angler group ANOVA results Variable 1 2 3 4 df F-value P Cluster variables Distance traveled to lake (km) 69.1 w 381.1 z 165.3 y 102.6 x 3 739.91 <0.05 Proportion of catch that was harvested 0.0 y 0.02 y 0.04 y 0.66 z 3 1,654.30 <0.05 Catchability (ha/h; × 10−3) 3.22 x 6.43 y 75.35 z 6.45 y 3 964.67 <0.05 Other variables Days ﬁshed per year 26.8 yx 30.4 zy 37.2 z 24.4 x 3 7.43 <0.05 Proportion of overnight trips 0.34 yx 0.44 z 0.40 zy 0.30 x 3 13.14 <0.05 CPUE (ﬁsh/h) 0.49 w 0.69 x 2.48 z 1.23 y 3 147.81 <0.05 Number of anglers 945 371 171 469 Percent of sample 48.3 19.0 8.7 24.0

arbitrarily as anglers traveling over 200 km or less than 200 km groups of anglers were identified by the cluster analysis based to the lake, respectively) revealed dramatic differences in catch on the four cluster validation statistics we examined. efficiency and harvest behavior between these subsets of anglers. The four angler groups identified by the cluster analysis var- Catchability differed between local and nonlocal anglers (ts = ied substantially in their quantitative characteristics and differed −3.68, df = 789, P = 2.48 × 10−4), with local anglers having in their proportional representation in the sample of interviewed lower mean q (mean [µ] = 9.41 × 10−3 ha/h; standard deviation anglers (Table 3). Among the four groups, angler group 1 was [σ] = 0.024) than nonlocal anglers (µ = 14.83 × 10−3 ha/h; characterized as having the lowest average travel distance to σ = 0.031). The proportion of the catch that was harvested also the lake (69.1 km), the lowest q (3.22 × 10−3 ha/h), and the varied between local and nonlocal anglers (t =−2.98, df = lowest proportion of catch harvested (0.00). Angler group 2 1,105, P = 2.95 × 10−3); local anglers harvested a greater was characterized as having the greatest travel distance to the proportion of their catch (µ = 0.18; σ = 0.34) than did nonlocal lake where they were interviewed (381.1 km); angler group 3 anglers (µ = 0.13; σ = 0.30). was characterized as having a 10-fold higher q (75.35 × 10−3 The distribution of the proportion of catch that was harvested ha/h); and angler group 4 harvested the highest proportion of (Figure 2b) suggested that most anglers harvested a low propor- their catch (0.66) in comparison with the other groups. The four tion of their catch (µ = 0.17; σ = 0.33). Across lakes, the discrete groups also differed in terms of other descriptive char- proportion of catch that was harvested (ρ) was found to be a acteristics that were not used in the cluster analysis (Table 3). linear function of the mean CPUE and the bag limit (BL)fora For example, angler group 3 was composed of the most avid lake (R2 = 0.9249), anglers (fishing an average of 37.2 d/year), and angler group 2 took the highest proportion of overnight trips for the purpose of −3 −2 fishing (0.44). The CPUE varied significantly among the angler ρ = [(−2.023 × 10 )CPUE]+ [(5.18 × 10 )BL ], (4) Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 groups and was strongly correlated with q. Angler groups 2 and 4 had nonstatistically different q-values, but the average CPUE suggesting that decreases in the CPUE and the bag limit cor- for angler group 4 was approximately two times higher than that responded to an increase in the proportion of catch that was for angler group 2, suggesting that angler group 4 fished most harvested. Catch efficiency (q) also varied among anglers (Fig- often in lakes with approximately double the fish density. ure 2c); approximately 88% of anglers had q less than 0.02 ha/h The variables that described the quantitative interaction of (or less than 2 fish/h in a lake with 100 fish/ha). anglers with the resource, as well as the other descriptive vari- We used multivariate techniques to determine whether sub- ables we considered, also differed among the three management classes of anglers existed within the total sample of interviewed regions stratified within the survey design (Table 4). Anglers in anglers based on variables that describe how anglers interact the Omineca region traveled the shortest distances to fish; an- with the fishery. Cluster validation statistics suggested that the glers in the Okanagan region harvested the lowest proportion of optimal number of clusters to extract from the data was four. their catch and had the highest q-values. Anglers in the Thomp- The Dunn index and average silhouette width were maximized son region were the most avid anglers (fished the most days at four clusters, and the normalized gamma coefficient was max- on average per year) and had the highest CPUE, and anglers in imized at eight clusters. The within-groups sum of squares had the Okanagan region took the highest proportion of overnight a rapid decrease in slope at four clusters. Thus, four discrete trips relative to anglers in the other regions. The proportions ANGLER CHARACTERISTICS IN A RECREATIONAL FISHERY 581

TABLE 4. Mean values for angler characteristics and the percentages of anglers that belonged to each angler group within the three management regions (Figure 1) of the Rainbow Trout ﬁshery. Within a given row, means with the same lowercase letter are not signiﬁcantly different at P < 0.05 as indicated by Tukey’s test for multiple comparisons.

Management region ANOVA results Okanagan Thompson Omineca Variable (region 8) (region 3) (region 7) df F-value P Cluster variables Distance traveled to lake (km) 138.9 y 175.7 z 86.3 x 2 24.52 <0.05 Proportion of catch that was harvested 0.13 y 0.20 z 0.25 z 2 14.52 <0.05 Catchability (ha/h; × 10−3) 13.0 z 9.7 y 1.9 x 2 16.22 <0.05 Other variables Days ﬁshed per year 26.6 y 32.7 z 19.7 x 2 15.12 <0.05 Proportion of overnight trips 0.42 z 0.26 y 0.26 y 2 53.03 <0.05 CPUE (ﬁsh/h) 0.83 zy 1.01 z 0.76 y 2 6.87 <0.05 Angler groups Group 1 percentage 49.8 39.8 64.9 Group 2 percentage 17.5 26.8 4.2 Group 3 percentage 12.3 4.3 0.0 Group 4 percentage 20.4 29.1 30.9 Number of anglers 1,187 578 191 Percentage of sample 60.7 29.5 9.8

of anglers belonging to the four angler groups identified by the Recreational specialization is now thought to include three main cluster analysis varied significantly among management regions dimensions: (1) a behavioral dimension that is measured by the (χ2 = 125.4, df = 6, P < 2.2 × 10−16). Angler group 1 was the frequency of participation, (2) a cognitive dimension that is most numerous in all regions. In the Okanagan region, anglers measured by knowledge and skill, and (3) an affective com- were distributed roughly evenly among angler groups 2, 3, and ponent that is measured by commitment to the activity (Scott 4. In contrast, anglers in the Omineca region mainly belonged and Shafer 2001). In North American-based studies of angler to groups 1 and 4. specialization, the least specialized anglers emphasized the importance of non-catch-related factors as motivations for fishing DISCUSSION but were still likely to harvest fish, whereas the most specialized Anglers who fished the stocked Rainbow Trout lakes across anglers were motivated by resource-related factors (trophy fish) a broad landscape in British Columbia varied substantially in and were most likely to demonstrate catch-and-release behavior characteristics that determine quantitatively how anglers interact (Bryan 1977; Chipman and Helfrich 1988; Fisher 1997; Hutt with the fishery. Spatial distribution, efficiency, and propensity and Bettoli 2007). Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 to harvest varied among anglers. Angler efficiency (q) and har- The characteristics of the four groups identified by the clus- vest behavior were strongly correlated with distance traveled to ter analysis align well with the results of other studies that have the lake. Anglers who traveled short distances to the lake (i.e., identified several groups of anglers representing the continuum local anglers) had lower q-values and harvested a higher propor- of specialization (Graefe 1980; Chipman and Helfrich 1988; tion of their catch than did anglers who traveled long distances Hutt and Bettoli 2007). For example, Chipman and Helfrich to fish (i.e., nonlocal anglers). These results suggest that the (1988) identified six groups of anglers and found that the fre- angler population does not interact homogeneously within this quency of fishing, investment in the resource, and consumptive spatially structured fishery, and therefore it is useful to under- habits best described the variation among anglers in their study. stand and quantify how different angler types affect the fishery. Similarly, Hutt and Bettoli (2007) identified five angler groups Our observation of the existence of multiple discrete groups that differed in their attitudes toward harvesting trout and catch- of anglers complements other studies that have described diver- ing trophy fish. The variation in behavior of anglers belonging sity in angler characteristics by using similar multidimensional to the discrete groups identified in our study is consistent with approaches (Chipman and Helfrich 1988; Fisher 1997). The the literature surrounding the theory of recreational specializa- theory of recreational specialization, first proposed by Bryan tion. For example, anglers in group 2 could be characterized (1977), provides a conceptual framework for understanding the as being the most specialized: these anglers traveled the great- multidimensional aspects of anglers’ behaviors and attitudes. est distance to fish, harvested a low proportion of their catch, 582 WARD ET AL.

and fished frequently. Similarly, several authors have found that management actions (Bryan 1977; Chipman and Helfrich 1988; the most specialized anglers travel greater distances, fish more Fisher 1997; Hutt and Bettoli 2007). In contrast, the Omineca frequently, and attach a greater emphasis on angling as a recre- region was almost entirely composed of anglers with low spe- ational activity than anglers with lower levels of specialization cialization, indicating that most lakes in this region could be (Salz et al. 2001; Oh et al. 2005; Hutt and Bettoli 2007; Carlin managed to maximize utility for these less-specialized anglers et al. 2012). In contrast, anglers in group 4 share similarities by having high stocking densities and liberal bag limits. The with the low-specialization groups described by Chipman and proportions of anglers belonging to the discrete angler groups Helfrich (1988). Those authors identified that the least special- in the Okanagan region were more uniform than in the other ized anglers were motivated by family-oriented recreation, were regions. Thus, lakes in the Okanagan region could be managed satisfied with catching small fish, and placed greater emphasis for both highly specialized anglers and least specialized anglers. on harvesting fish. In our study, anglers in group 4 harvested By including the diversity in angler characteristics in predictive the highest proportion of their catch, fished the lowest number models, managers’ ability to predict the ecological impacts of of days per year, and traveled a relatively short distance to fish. fishing and strategies for optimizing management will be im- The characteristics of anglers in group 1 tended to align best proved (Johnston et al. 2010). with less-specialized anglers except that the group 1 anglers did Anglers in group 4 represented 24% of all anglers inter- not harvest fish. Similarly, anglers in group 3 appeared to be viewed. These anglers were characterized by traveling short similar to the most specialized anglers, as they fished the most distances to the lake, harvesting a high proportion of the catch, frequently (and had the highest q), but they traveled shorter and having moderate levels of q and CPUE. Across all an- distances on average than anglers in group 2. glers, the propensity to harvest fish at a lake was negatively The relative proportions of the four angler groups differed related to mean CPUE and was positively related to the bag among regions within the British Columbia Rainbow Trout fish- limit. This suggests that as resources become more scarce, an- ery. This implies that region-specific management strategies glers are more likely to harvest fish; similar trends have been could be developed to better reflect the distribution of angler noted for a Walleye Sander vitreus fishery in Alberta, Canada types. For example, anglers in group 2 appeared to be the most (Sullivan 2002). In stocked fisheries, population collapse is not specialized and traveled the furthest to reach fishing opportuni- a concern, but neighboring wild populations may be at risk for ties. Arlinghaus and Mehner (2004) found that more-specialized overfishing. Specifically, if wild Rainbow Trout populations in anglers were willing to travel further to access fishing oppor- close proximity to population centers attract the least special- tunities in comparison with less-specialized anglers. Manage- ized anglers, then these anglers are more likely to harvest fish ment strategies aimed at maximizing utility across all anglers and can have impacts on sustainability. Therefore, it is useful to could consider optimizing management for highly specialized understand the impacts and relative effort of such anglers across anglers by using restrictive regulations that are aimed at creating the landscape to highlight potential management concerns for trophy-type fisheries in the least accessible lakes (see Johnston neighboring wild populations. et al. 2010 for an example). In general, more-specialized an- We recognize that our study was conducted on a relatively glers are often more receptive to restrictive regulations than are small and nonrandom subset of lakes within three management less-specialized anglers, and highly specialized anglers are more regions across the broad landscape of the Rainbow Trout fishery likely to follow restrictive regulations and even to impose volun- in British Columbia. Although these lakes were nonrandomly tary regulations to preserve fish stocks with high angling quality selected, they do provide fisheries that are reasonably represen- (Ditton et al. 1992; Oh and Ditton 2006; Dorow et al. 2010). tative of those in regions with moderately high effort, variable

Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 In contrast to the management of lakes for highly specialized travel distances from primary residences, and a range of fish anglers, easily accessible lakes would likely derive the maxi- abundances; therefore, we suggest that the inferences we have mum utility by optimizing management based on the demands drawn effectively describe how anglers interact with the fishery of the least specialized anglers (e.g., family-oriented settings through their spatial distribution, efficiency, and harvest behav- with liberal bag limits). ior. Our approach differs from that used by researchers who Anglers in group 1 (low-specialization anglers who did not simply want to study specialization, such as Oh and Ditton tend to harvest fish) were the most common across all regions. (2006); our goal was to understand and quantify the processes Since the least specialized anglers are motivated by family- that translate angler behavior into harvest impacts over a spa- oriented recreation and are satisfied with catching small fish, all tially structured recreational fishery. regions could focus on managing easily accessible lakes with The occurrence of discrete angler groups, variation in the high stocking densities and liberal bag limits for these anglers. behavioral characteristics of these groups, and variation in the The proportions of anglers within groups 2, 3, and 4 varied proportion of these groups across regions demonstrates the po- among regions. The Thompson region had the highest propor- tential for complex interplay of the social–ecological systems of tion of the most specialized anglers (angler group 2); fisheries recreational fisheries (Fenichel et al. 2013). Management that with restrictive regulations and low stocking densities would be integrates both social and ecological components of recreational optimal for this region, since highly specialized anglers tend fisheries has been called for repeatedly (Arlinghaus 2006; Hunt to prefer fishing at trophy fisheries that are enhanced by these et al. 2007; Post et al. 2008) but is not often carried out at the ANGLER CHARACTERISTICS IN A RECREATIONAL FISHERY 583

scale of spatially complex fisheries (for an exception, see Hunt Beardmore, B., W. Haider, L. M. Hunt, and R. Arlinghaus. 2011. The importance et al. 2011). Effective integration requires quantitative relation- of trip context for determining primary angler motivations: are more special- ships that describe both social and ecological processes over ized anglers more catch-oriented than previously believed? North American Journal of Fisheries Management 31:861–879. space. Our work represents the first attempt to simultaneously Bryan, H. 1977. Leisure value systems and recreational specialization: the case quantify variation in angler characteristics among groups, char- of trout fishermen. Journal of Leisure Research 9:174–187. acteristics that directly link to interactions with fish populations Carlin, C., S. A. Schroeder, and D. C. Fulton. 2012. Site choice among Min- (e.g., catchability and propensity to harvest), and spatial behav- nesota Walleye anglers: the influence of resource conditions, regulations and ior over a spatially distributed fishery. This understanding will catch orientation on lake preference. North American Journal of Fisheries Management 32:299–312. provide the processes that are necessary to develop predictive Carpenter, S. R., and W. A. Brock. 2004. Spatial complexity, resilience, and process models with which to examine optimal management policy diversity: fishing on lake-rich landscapes. Ecology and Society [online actions for this valuable fishery. serial] 9(1):article 8. Chipman, B. D., and L. A. Helfrich. 1988. Recreational specializations and motivations of Virginia river anglers. North American Journal of Fisheries ACKNOWLEDGMENTS Management 8:390–398. This study was funded by Discovery and Collaborative Re- Cox, S. P. 2000. Angling quality, effort response and exploitation in recreational search and Development Grants from the Natural Sciences and fisheries: field and modeling studies on British Columbia Rainbow Trout (Oncorhynchus mykiss) lakes. Doctoral dissertaton. University of British Engineering Research Council of Canada (NSERC) and the Columbia, Vancouver. Freshwater Fisheries Society of British Columbia (FFSBC) to Cox, S. P., and C. Walters. 2002. Modeling exploitation in recreational fisheries John R. Post. We recognize the Habitat Conservation Trust and implications for effort management on British Columbia Rainbow Trout Foundation (HCTF), as well as the anglers, hunters, trappers, lakes. North American Journal of Fisheries Management 22:21–34. and guides who contribute to the HCTF, for making a significant Ditton, R. B., D. K. Loomis, and S. Choi. 1992. Recreation specialization: re-conceptualization from a social worlds perspective. Journal of Leisure financial contribution to support this project. Hillary Ward was Research 24:33–51. supported by an Industrial NSERC Postgraduate Scholarship. Dorow, M., B. Beardmore, W. Haider, and R. Arlinghaus. 2010. Winners and We thank the staff of FFSBC and the British Columbia Ministry losers of conservation policies for European Eel, Anguilla anguilla: an eco- of Environment for the outstanding support they provided for nomic welfare analysis for differently specialised eel anglers. Fisheries Man- this project. This research was granted ethical approval by the agement and Ecology 17:106–125. Fedler, A. J., and R. B. Ditton. 1994. Understanding angler motivations in Conjoint Faculties Research Ethics Board (Certification 6546) fisheries management. Fisheries 19(4):6–13. and was approved by the Canadian Council on Animal Care Fenichel, E. P., J. K. Abbott, and B. Huang. 2013. Modelling angler behaviour as (Protocol BI08R-21) at the University of Calgary. We thank Paul part of the management system: synthesizing a multi-disciplinary literature. Askey, Eric Parkinson, Adrian Clarke, and Wolfgang Haider for Fish and Fisheries 14:137–157. discussions that led to the development of this manuscript, and Fisher, M. R. 1997. Segmentation of the angler population by catch preference, participation, and experience: a management-oriented application of we are grateful to Robert Arlinghaus for his insightful comments recreation specialization. North American Journal of Fisheries Management throughout the editorial process, which substantially improved 17:1–10. the paper. Field support was provided by Ryan Ashley, Kyera Graefe, A. R. 1980. The relationship between level of participation and selected Cook, Jessica Courtier, Travis Desy, Jon Fearns, Anne-Marie aspects of specialization in recreational fishing. Doctoral dissertation. Texas Flores, Brian France, Thamara Hendricks, Collin Middleton, A&M University, College Station. Halkidi, M., Y. Batistakis, and M. Vazirgiannis. 2001. On clustering validation Tabatha Nedokus, Eric Newton, Michelle Phillips, and Daniel techniques. Journal of Intelligent Information Systems 17:107–145. Swainson. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics, and uncertainty. Chapman and Hall, New York. Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 Hunt, L. M. 2005. Recreational fishing site choice models: insights and future REFERENCES opportunities. Human Dimensions of Wildlife 10:153–172. Allen, M. S., P. Brown, J. Douglas, W. Fulton, and M. Catalano. 2009. An as- Hunt, L. M., R. Arlinghaus, N. Lester, and R. Kushneriuk. 2011. The effects sessment of recreational fishery harvest policies for Murray Cod in southeast of regional angling effort, angler behavior, and harvesting efficiency on land- Australia. Fisheries Research 95:260–267. scape patterns of overfishing. Ecological Applications 21:2555–2575. Arlinghaus, R. 2006. On the apparently striking disconnect between motivation Hunt, L. M., B. N. Boots, and P. C. Boxall. 2007. Predicting fishing participation and satisfaction in recreational fishing: the case of catch orientation of German and site choice while accounting for spatial substitution, trip timing, and trip anglers. North American Journal of Fisheries Management 26:592–605. context. North American Journal of Fisheries Management 27:832–847. Arlinghaus, R., and T. Mehner. 2004. A management-orientated comparative Hutt, C. P., and P.W. Bettoli. 2007. Preferences, specialization, and management analysis of urban and rural anglers living in a metropolis (Berlin, Germany). attitudes of trout anglers fishing in Tennessee tailwaters. North American Environmental Management 33:331–344. Journal of Fisheries Management 27:1257–1267. Askey, P. J., J. R. Post, E. A. Parkinson, E. Rivot, A. J. Paul, and P. A. Biro. Johnston, F. D., R. Arlinghaus, and U. Dieckmann. 2010. Diversity and com- 2007. Estimation of gillnet efficiency and selectivity across multiple sampling plexity of angler behaviour drive socially optimal input and output regulations units: a hierarchical Bayesian analysis using mark–recapture data. Fisheries in a bioeconomic recreational-fisheries model. Canadian Journal of Fisheries Research 83:162–174. and Aquatic Sciences 67:1507–1531. Bannerot, S. P., and C. B. Austin. 1983. Using frequency distributions of catch Jones, C. M., D. S. Robson, H. D. Lakkis, and J. Kressel. 1995. Properties of per unit effort to measure fish-stock abundance. Transactions of the American catch rates used in analysis of angler surveys. Transactions of the American Fisheries Society 112:608–617. Fisheries Society 124:911–928. 584 WARD ET AL.

Lester, N. P., T. R. Marshall, K. Armstrong, W. I. Dunlop, and B. Ritchie. 2003. Salz, R. J., D. K. Loomis, and K. L. Finn. 2001. Development and validation of a A broad-scale approach to management of Ontario’s recreational fisheries. specialization index and testing of specialization theory. Human Dimensions North American Journal of Fisheries Management 23:1312–1328. of Wildlife 6:239–258. Meila,˘ M. 2007. Comparing clusterings—an information based distance. Journal Scott, D., and C. S. Shafer. 2001. Recreational specialization: a critical look at of Multivariate Analysis 98:873–895. the construct. Journal of Leisure Research 33:319–343. Oh, C., and R. B. Ditton. 2006. Using recreation specialization to understand Shuter, B. J., M. L. Jones, R. M. Korver, and N. P. Lester. 1998. A general, multi-attribute management preferences. Leisure Sciences 28:369–384. life history based model for regional management of fish stocks: the inland Oh, C., R. B. Ditton, D. K. Anderson, D. Scott, and J. R. Stoll. 2005. Under- Lake Trout (Salvelinus namaycush) fisheries of Ontario. Canadian Journal of standing differences in nonmarket valuation by angler specialization level. Fisheries and Aquatic Sciences 55:2161–2177. Leisure Science 27:263–277. Sullivan, M. G. 2002. Illegal angling harvest of Walleyes protected by length Parkinson, E. A., J. R. Post, and S. P. Cox. 2004. Linking the dynamics limits in Alberta. North American Journal of Fisheries Management 22:1053– of harvest effort to recruitment dynamics in a multistock, spatially struc- 1063. tured fishery. Canadian Journal of Fisheries and Aquatic Sciences 61: van Poorten, B. T., R. Arlinghaus, K. Daedlow, and S. S. Haertel-Borer. 2011. 1658–1670. Social-ecological interactions, management panaceas, and the future of wild Post, J. R., and E. A. Parkinson. 2012. Temporal and spatial patterns of angler fish populations. Proceedings of the National Academy of Sciences of the effort across lake districts and policy options to sustain recreational fisheries. USA 108:12554–12559. Canadian Journal of Fisheries and Aquatic Sciences 69:321–329. Walters, C. J., and S. J. D. Martell. 2004. Fisheries ecology and management. Post, J. R., E. A. Parkinson, and N. T. Johnston. 1999. Density-dependent Princeton University Press, Princeton, New Jersey. processes in structured fish populations: interaction strengths in whole-lake Walters, C. J., and J. R. Post. 1993. Density-dependent growth and competitive experiments. Ecological Monographs 69:155–175. asymmetries in size-structured fish populations: a theoretical model and rec- Post, J. R., L. Persson, E. A. Parkinson, and T. van Kooten. 2008. Angler ommendations for field experiments. Transactions of the American Fisheries numerical response across landscapes and the collapse of freshwater fisheries. Society 122:34–45. Ecological Applications 18:1038–1049. Ward, H. G. M., P. J. Askey, J. R. Post, D. A. Varkey, and M. K. McAllister. Radomski, P. 2003. Initial attempts to actively manage recreational fishery 2012. Basin characteristics and temperature improve abundance estimates harvest in Minnesota. North American Journal of Fisheries Management from standard index netting of Rainbow Trout (Oncorhynchus mykiss)in 23:1329–1342. small lakes. Fisheries Research 131–133:52–59. Downloaded by [Department Of Fisheries] at 20:08 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:09 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Frequent Usage of Tributaries by the Endangered Fishes of the Upper Colorado River Basin: Observations from the San Rafael River, Utah Jared L. Bottcher a d , Timothy E. Walsworth a e , Gary P. Thiede a , Phaedra Budy b & David W. Speas c a Department of Watershed Sciences , Utah State University , 5210 Old Main Hill, Logan , Utah , 84322 , USA b U.S. Geological Survey, Utah Cooperative Fish and Wildlife Research Unit, Department of Watershed Sciences , Utah State University , 5210 Old Main Hill, Logan , Utah , 84322 , USA c U.S. Bureau of Reclamation , Upper Colorado Regional Office , 125 South State Street, Salt Lake City , Utah , 84138 , USA d Klamath Basin Rangeland Trust , 700 Main Street, Suite 201A, Klamath Falls , Oregon , 97601 , USA e School of Aquatic and Fishery Sciences , University of Washington , Box 355020, Seattle, Washington , 98195 , USA Published online: 24 May 2013.

To cite this article: Jared L. Bottcher , Timothy E. Walsworth , Gary P. Thiede , Phaedra Budy & David W. Speas (2013): Frequent Usage of Tributaries by the Endangered Fishes of the Upper Colorado River Basin: Observations from the San Rafael River, Utah, North American Journal of Fisheries Management, 33:3, 585-594 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785993

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:585–594, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785993

MANAGEMENT BRIEF

Frequent Usage of Tributaries by the Endangered Fishes of the Upper Colorado River Basin: Observations from the San Rafael River, Utah

Jared L. Bottcher,*1 Timothy E. Walsworth,2 and Gary P. Thiede Department of Watershed Sciences, Utah State University, 5210 Old Main Hill, Logan, Utah 84322, USA Phaedra Budy U.S. Geological Survey, Utah Cooperative Fish and Wildlife Research Unit, Department of Watershed Sciences, Utah State University, 5210 Old Main Hill, Logan, Utah 84322, USA David W. Speas U.S. Bureau of Reclamation, Upper Colorado Regional Ofﬁce, 125 South State Street, Salt Lake City, Utah 84138, USA

the basin (Colorado Pikeminnow Ptychocheilus lucius, Bony- Abstract tail Gila elegans, Humpback Chub G. cypha, and Razorback The importance of main-stem rivers and major tributaries to Sucker Xyrauchen texanus) are listed as endangered under the endangered Colorado River fishes is well documented, but the use U.S. Endangered Species Act of 1973 (ESA) (ESA 1973; US- and significance of small tributary streams remains poorly understood. Historically, these fishes probably used smaller tributaries FWS 2002a, 2002b, 2002c, 2002d). In the upper Colorado River for spawning, rearing, feeding, and refuge. Currently, the prolif- basin, critical habitat has been designated for each of these eration of nonnative species and altered flows may have affected species in the San Juan, Green, Colorado, Gunnison, Duchesne, tributary use by endangered fishes. In February 2008 and 2009, Yampa, and White rivers, all of which are main-stem rivers or we installed a PIT-tag passive interrogation array (PIA) in the major tributaries (USFWS 1994; Figure 1). With a few excep- San Rafael River, Utah, approximately 2 km upstream from the confluence with the Green River, and another PIA approximately tions, recovery actions for these four endangered species, such 60 km upstream from the Green River confluence. Using passive as instream flow protection, research, monitoring, nonnative detections and active captures in the San Rafael River from 2008 fish removal, stocking, and habitat restoration efforts, have been to 2010, we detected 15 Colorado Pikeminnow Ptychocheilus lu- concentrated in critical habitat areas (USFWS 1994; UCREFRP cius, 16 Bonytails Gila elegans, 20 Razorback Suckers Xyrauchen 2011). However, despite over two decades of recovery efforts Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 texanus, and five “undocumented” fish. Several endangered fishes were detected on multiple occasions and across years, often moving and some isolated examples of positive biological response, en- into and out of the San Rafael River from distances up to 360 km dangered fish populations in the Colorado River have not recov- away (range, 6–360 km). Our findings demonstrate the use and the ered sufficiently to warrant their downlisting or delisting from potential importance of small tributaries and their fragile habitats the ESA. In the Colorado River basin, areas within the range to endangered fishes. of a species that are not categorized as critical habitat may still be vital to the health and recovery of discrete populations of endangered species (Finney 2006). These areas outside of des- The Colorado River basin, historically home to an endemic, ignated critical habitat are probably important for maintenance depauperate fish fauna, has experienced extensive alterations of genetic and life history diversity, or as a source of immigrants from anthropogenic activities and native fish populations have for depleted populations (Hanski and Gilpin 1991; Hanski and declined as a result (Minckley and Deacon 1991). Four fishes in Thomas 1994).

*Corresponding author: [email protected] 1Present address: Klamath Basin Rangeland Trust, 700 Main Street, Suite 201A, Klamath Falls, Oregon 97601, USA. 2Present address: School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington 98195, USA. Received May 13, 2012; accepted March 10, 2013 585 586 BOTTCHER ET AL. Downloaded by [Department Of Fisheries] at 20:09 28 May 2013

FIGURE 1. The upper Colorado River basin, including locations of the PIA systems in the San Rafael River, Utah, hatchery-release locations and similar tributaries throughout the basin, and river mile (RM) points for reference. MANAGEMENT BRIEF 587

Small tributary systems to the Green and Colorado rivers METHODS might have been historically important to endangered fishes of Study area.—The San Rafael River drains 4,500 km2 in the Colorado River basin (Tyus and Saunders 2001). Herein central Utah, and is formed by the merging of Cottonwood, we define “small tributary streams” both in terms of hydroge- Ferron, and Huntington creeks. It is a snowmelt- (in spring) and omorphology and in terms of their conservation status. These monsoon-driven (in autumn) system that flows approximately small tributary streams include fifth- to sixth-order rivers such 175 km from its headwaters in the Manti–LaSal National Forest as the Muddy, Fremont, Price, San Rafael, and Dirty Devil to its confluence with the Green River, near the town of Green rivers, with drainage areas of 2,175–6,277 km2, mean annual River, Utah (Figure 1). Before large water development projects discharges (MAQs) of 36.5–103.1 ft3/s, and 2-year flood (post- were initiated (1909–1918), peak streamflows averaged 4,905 dam) discharges of 343–3,882 ft3/s, based on a GIS analysis ft3/s (range, 1,300–12,000 ft3/s). With over 800 surface points of the U.S. Geological Survey (USGS) National Hydrography of diversion and 360 dams in the drainage, the San Rafael River Dataset (NHD) at their confluences (Figure 1). For compar- is one of the most overallocated rivers in Utah (Walker and ison, the Green River (at the Utah USGS gauge site) has a Hudson 2004), and minimum instream flows have not been MAQ of 5,317 ft3/s and a 2-year flood (postdam) discharge of established for this river. As such, the lower 64 km of the river 20,397 ft3/s. In terms of their conservation status, such small are frequently dewatered in the summer. tributaries are frequently not designated as critical habitat un- This tributary is home to several native fishes, including the der the ESA and thus do not assume a prominent role in the Flannelmouth Sucker, Bluehead Sucker, and Roundtail Chub, recovery process for endangered fishes in the Colorado River (three endemic sensitive species protected under a conservation basin. In addition, it is rare for consultations under the ESA to af- agreement: UDWR 2006), and Speckled Dace Rhinichthys ford protection for fish in nondesignated systems. Consequently, osculus. The system has been invaded by many nonnative vulnerability of small tributaries in unprotected systems to con- species, including Channel Catfish Ictalurus punctatus,Red tinued water development, associated habitat degradation, and Shiner Cyprinella lutrensis, Sand Shiner Notropis stramineus, other anthropogenic impacts tends to be greater than for small Common Carp Cyprinus carpio, Green Sunfish Lepomis tributaries in areas designated as critical habitat, in which re- cyanellus, Fathead Minnow Pimephales promelas, and virile covery actions are taking place or that are otherwise legally crayfish Orconectes virilis (Bottcher 2009). Additionally, the ri- protected. parian zone has been invaded by tamarisk (Tamarix spp.), which In the last few decades, there have been several documented has led to stream bank stabilization, increased channelization, occurrences of endangered species inhabiting small tributaries and decreased floodplain access, thus inhibiting the formation within the Colorado River basin (McAda et al. 1980; Tyus of complex instream habitat (Bottcher 2009; Fortney et al. 1987; Marsh et al. 1991; Wick et al. 1991). Concentrations of 2011). Razorback Sucker larvae in and near the mouth of the San Rafael Data collection.—As part of a larger capture–mark– River (lower Green River) suggest that such tributaries provide recapture study aimed at examining metapopulation dynamics spawning and rearing habitat for that species (Chart et al. 1999; and identifying limiting factors for Bluehead Sucker, Flannel- Bestgen et al. 2002). Both adult and juvenile Colorado mouth Sucker, and Roundtail Chub in the San Rafael River from Pikeminnow have been documented in small tributary systems 2008 to 2010 (Bottcher 2009; Walsworth 2011), we installed two (Cavalli 1999; Fresques et al. 2013) as have large numbers solar-powered, full-duplex (134.2 kHz), PIT-tag passive interro- of other, nonlisted native species (e.g., Flannelmouth Sucker gation array (PIA) systems. We installed one system in February Catostomus latipinnis, Bluehead Sucker C. discobolus, and 2008 at Chaffin Ranch, approximately 2 km upstream from the

Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 Roundtail Chub Gila robusta; McAda et al. 1980; Bottcher San Rafael River’s confluence with the Green River (Figure 1). 2009), suggesting that these systems provide important feeding In April 2009, we installed a second PIA approximately 3.5 km and rearing habitats for native fish fauna. Lastly, because the downstream from an impassible diversion dam at Hatt Ranch native Colorado River fish community would have probably (62 km upstream from the confluence with the Green River; comprised multiple life history expressions (e.g., migratory, Figure 1). Each PIA consisted of a set of two antenna loops potamodromous, resident), tributary systems would have 4 m upstream from another set of two antenna loops; each set consistently provided habitat for at least one life history of two antennas spans the bankfull width. The four antennas are expression or life stage. connected to a multiplexor unit, which records the date, time, We examined recapture data secondarily available from our and individual-specific PIT-tag number each time a tagged fish own comprehensive local study of sensitive fishes in the San swims over the antennas (e.g., Zydlewski et al. 2006). Within Rafael River, as well as from dispersed, basin-wide, endangered the San Rafael River, and at each PIA, we assumed that if a fish monitoring efforts throughout the upper Colorado River fish was detected by the downstream antennas and subsequently basin. Our objective was to evaluate the use of the San Rafael detected by the upstream antennas, it was moving upstream, and River, a small tributary to the Green River, by sensitive and vice versa. The PIA systems were operated nearly continuously endangered fishes. from installation date until 2010 with the exception of short-term 588 BOTTCHER ET AL.

TABLE 1. Number of unique (i.e., only first detection is counted) detections of endangered and undocumented fishes at both PIAs and unique number collected in trammel nets and electrofishing surveys near the mouth of the San Rafael River, 2007–2010. Undocumented fish may include non-ESA listed species.

Unique number Unique number Endangered species detected handled Size (TL, mm) Sources of fish Colorado Pikeminnow 11 4 358–648 Wild fish from the Green and White rivers. Bonytail 15 1 163–308 From Wahweap State Fish Hatchery (UDWR), stocked in Colorado or Green rivers. Razorback Sucker 16 4 262–560 From Ouray National Fish Hatchery, Green River; some stocked near Green River city. Undocumented fishes 5 0 Unknown No information available. Two PIT tags were distributed to UDWR, Salt Lake City office, and potentially implanted into nonendangered, sensitive species.

(less than 1 month) outages associated with maintenance to ad- RESULTS dress tuning and high flow or debris events. We qualitatively Using data from PIA systems and active recapture surveys estimated detection efficiency at each system by examining dif- in the San Rafael River from 2008 to 2010, we documented ferential detections between the upstream and downstream an- the presence of 51 individual fish (15 Colorado Pikeminnow, tennas (both within sites and among the Chaffin and Hatt Ranch 16 Bonytails, 20 Razorback Suckers) and five “undocumented” PIA systems) and by manually passing PIT tags over the anten- fishes (possible endangered fishes or other sensitive native fishes nas throughout the water column. implanted with PIT tags). For one of the five “undocumented” We were able to determine the timing and duration of tribu- PIT-tagged fish, we found the date when it was implanted tary use and movement information for individually PIT-tagged and location where it was released, but we could not locate fish through both active and passive recaptures in the San Rafael any history for the other four (even after contacting multiple River, coupled with database information on endangered fishes agencies in the upper Colorado River basin; Table 1). Nine previously tagged and released in the upper Colorado River of the 10 actively captured fish were collected with trammel basin by participants in the Recovery Program (mostly Utah nets or by electrofishing in the lower 300 m of the San Rafael Division of Wildlife Resources [UDWR], Colorado Division of River upstream from the confluence with the Green River (the Parks and Wildlife, U.S. Fish and Wildlife Service, and Col- most downstream sample site). The remaining actively cap- orado State University). The Recovery Program has stocked tured fish was an age-0 Colorado Pikeminnow collected by the large numbers of PIT-tagged Razorback Suckers and Bonytails UDWR approximately 4.5 km downstream from the Hatt Ranch annually over the past decade into the Colorado and Green rivers diversion. (Nesler et al. 2003; Zelasko et al. 2010), whereas wild Colorado Several individuals of the endangered fishes were detected Pikeminnow have been actively captured and tagged in the same on multiple occasions, often moving into and out of the San areas over the past 21 years, usually between April and Octo- Rafael River, for a total of 100 detections since 2008. At the ber (Osmundson and Burnham 1998; Bestgen et al. 2007). We Chaffin Ranch PIA, some fish were only detected on either the

Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 calculated distance moved as the minimum stream-network dis- downstream or upstream antennas, which limited our ability to tance between the site of release after initial tagging and the site determine the direction of movement. Three of the four fish de- of capture or detection in the San Rafael River. tected at the Hatt Ranch PIA were previously detected at the We also actively captured fish by electrofishing and seining downstream Chaffin Ranch PIA (three fish were also “undocu- sample sites (300-m reaches) throughout the San Rafael River mented,” see above). and trammel-netting at the mouth of the river. We selected 22 Under normal flow conditions, we estimated detection effi- sample sites in the lower 64 km of the river using a random strat- ciency to be a maximum of 80%, although the efficiency was ified sampling design and actively sampled each reach multiple probably reduced during periods of high flow. High flow events times between April and October 2007–2010. We measured (TL, were also associated with peaks in fish movement (Figure 2). In mm) and weighed (g) all handled fish and examined them for tu- addition to regular use, particularly by Colorado Pikeminnow, bercles and ripeness, scanned them for the presence of a PIT tag our detection data indicated that many of the endangered fishes (if not previously tagged, we implanted fish with a 23-mm PIT have migrated or dispersed great distances from the initial tag- tag), and released fish near the point of capture. Collectively, ging locations into the San Rafael River; these fish moved up to these approaches allowed us to determine the timing, direction, 360 km from initial tagging locations and a minimum of 6 km and extent of fish movement. into the San Rafael River (Table 2). MANAGEMENT BRIEF 589

TABLE 2. Minimum distance moved and time spent in the San Rafael River (SRR), Utah, including source and initial tagging or release locations for a highlighted subset of endangered species detected in the SRR. Multiple years detected in the SRR indicate repeat usage. RM = river mile.

First date Last date Minimum Minimum Initial tagging or Date tagged detected in detected in PIA location distance time in release location or released SRR SRR in SRR moved (km) SRR (d) Source of fish Colorado Pikeminnow Green River, RM 78 6 Mar 06 28 Jun 10 15 Jul 10 Chaffin 33 18 Wild Green River, RM 93 24 Apr 08 7 Jun 08 11 Jun 08 Chaffin 9 5 Wild 20 Jun 10 2 Jul 10 Chaffin 9 12 Wild Green River, RM 79 16 Jun 06 8 Jul 09 8 Jul 09 Chaffin 30 17 Wild 10 Jun 10 30 Jun 10 Chaffin and 90 17 Wild Hatt White River, RM 20 15 May 07 22 Jun 08 3 Jul 08 Chaffin 282 13 Wild Green River, RM 99 24 May 07 6 Jun 08 17 Jun 08 Chaffin 6 12 Wild 10 Jun 10 26 Jun 10 Chaffin 6 17 Wild Bonytail Colorado River, 16 Nov 07 27 May 08 27 May 08 Chaffin 333 1 Stocked, 2005 RM 111 year-class Green River, RM 120 11 Nov 07 25 May 08 24 Jul 08 Chaffin 39 60 Stocked, 2005 year-class Razorback Sucker Green River, RM 56 28 May 06 8 Jun 08 8 Jun 08 Chaffin 68 1 Stocked, October 2005 Green River, RM 262 29 Aug 08 5 May 09 5 May 09 Chaffin 268 1 Stocked Green River, RM 120 2 Sep 09 11 Jun 10 12 Jun 10 Chaffin 39 2 Stocked Green River, RM 319 29 Aug 06 18 Apr 10 18 Apr 10 Chaffin 360 1 Stocked

Our collections indicate that Colorado Pikeminnow are re- the San Rafael River also appears to be highly variable, from 1 peat visitors and enter the San Rafael River in May, June, d to 2 months (Table 2). and July (Figure 2). One passively detected adult Colorado The majority of Razorback Suckers appeared in the San Pikeminnow traveled a minimum of 60 km upstream within Rafael River in April, May, and June, corresponding with the the San Rafael River. Colorado Pikeminnow traveled between ascending limb of the hydrograph (Figure 2). Those Razorback 6 and 282 km to reach the San Rafael River. A single Colorado Suckers detected during both upstream and downstream move- Pikeminnow captured and handled near the conﬂuence with the ments spent very little time (maximum of 2 d) upstream of the Green River in late May 2008 was tuberculated, but did not Chafﬁn Ranch PIA, though many individuals were only detected express gametes. once and at only one antenna. Two Razorback Suckers moved

Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 Our detections also indicate that tagged Bonytails stocked considerable distances (268 km and 360 km; total range, 5– in the Green River near the town of Green River, Utah, (Green 360 km) downstream from their initial release locations in the River, river mile [RM] 120) appear to travel relatively short Green River (Table 2). For all three endangered fish species de- distances downstream (i.e., <40 km; total range, 39–333 km; tected, the number of passive detections increased with variation Table 2) and select the San Rafael River. Four individuals re- in discharge (i.e., spring and monsoonal spates; Figure 2). leased at RM 120 by federal biologists in autumn 2007 entered the San Rafael River in May, June, and July 2008; another seven Bonytails released at RM 120 in November 2008 moved into DISCUSSION the San Rafael River in May, June, and July 2009; and three The detection of 51 endangered fish from three species in Bonytails released at RM 120 in February 2009 entered the the San Rafael River from 2008 to 2010 highlights the need for San Rafael River in June 2010. One Bonytail was detected in further investigation into the role and importance of tributary November 2009, although it is unclear whether this fish was habitats, especially comparatively small systems that are not moving into or out of the San Rafael River. A single Bonytail emphasized in recovery plans or otherwise protected under stocked in the Colorado River near Cisco, Colorado, in autumn the ESA. While main-stem river reaches and large tributaries 2007 was detected in the San Rafael River in May 2008, having provide the physical and biological conditions necessary for traveled a distance of over 330 km (Table 2). Length of stay in conservation of endangered fishes (USFWS 1994, 2002a, 590 BOTTCHER ET AL.

FIGURE 2. Number of detections of PIT-tagged endangered and “undocumented” ﬁsh species at PIA systems in the San Rafael River, Utah, in relation to date and river discharge (cfs = ft3/s).

2002b, 2002c, 2002d), the designation of critical habitat is difficult to describe their spatial distribution within the lower focuses monitoring and conservation activities in these areas San Rafael River, the detection of one Colorado Pikeminnow (USFWS 1994; Hoekstra et al. 2002), and much less protection 60 km upstream from the Green River suggests they may travel is afforded for a wider variety of systems used by these fishes. extensively upstream in tributary systems to reach areas that Detections of endangered fishes in the San Rafael River and are known to provide spawning habitat for other native fishes

Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 other small tributary systems offer insight into why tributaries (Tyus and Karp 1990; Bottcher 2009). are important for the native fish assemblage and suggest that Not surprisingly, all Bonytails detected in the San Rafael managers reexamine the role tributary systems may play in River were from hatchery sources. Observations of Bonytails species recovery. The timing and duration of pikeminnow usually occur immediately after stocking events and are within detections, coupled with the presence of a large tuberculated a few tens of kilometers downstream from stocking locations adult (Bottcher 2009) and historical and contemporary records (Bestgen et al. 2006; Recovery Program, unpublished data; see of age-0 pikeminnow presence (McAda et al. 1980), collectively also Badame 2008). Thus, while it is not surprising that most suggest that Colorado Pikeminnow probably use the San Rafael Bonytails were detected downstream from stocking locations River for spawning. Homing behavior and high spawning-site in Green River, Utah, the detection of 16 individuals (some fidelity are characteristic behaviors of Colorado Pikeminnow released up to 3 years previously) in a small tributary was (Irving and Modde 2000). Accordingly, repeated detections of unexpected given their overall rarity in the upper Colorado three, large, adult pikeminnow in the San Rafael River in June River basin. Entry timing and duration of stay suggests that and July of 2008 and 2010 indicate the San Rafael River may be Bonytails are also selecting the San Rafael River to spawn a natal stream for these fish. The absence of detections in spring and feed (Minckley 1973). In addition, the single observation 2009 may be explained by nonannual spawning behavior (e.g., of movement downstream from Westwater Canyon of the Tyus 1990) or an inefficient PIA due to high flows. Although it Colorado River to the Green River and upstream to the San MANAGEMENT BRIEF 591

Rafael River is unprecedented in terms of both direction and 2010 outnumbered active recaptures in a ratio greater than five magnitude (over 300 km), as is their repeated “selection” for to one. Deployment of this technology allowed us to begin de- the San Rafael River in general. scribing the timing and duration of tributary usage and the ex- Although it is clear Razorback Suckers use the San Rafael tent moved by fish, and to highlight the potential importance of River, the lack of multiple detections for individuals limits our these tributaries as habitat for endangered fishes. Other tribu- inference. The majority of Razorback Suckers were detected in taries are also likely to be selected by endangered fishes, and late May and early June, coinciding with the ascending limb PIA systems provide a cost-effective and nearly constant mon- of the hydrograph. The timing of these migrations is indicative itoring option for data delivery. One obstacle that will need to of spawning movements, although the duration of use upstream be addressed, however, is the lack of a central, regularly up- from the Chaffin Ranch PIA appears to be minimal. Notably, our dated, PIT-tag database for all of the endangered and nonlisted detection efficiency was not 100% (maximum, 80%), especially sensitive species within the basin (e.g., “undocumented” fish during periods of high flows; hence, our estimates of tributary herein). A framework for such a database currently exists and use are conservative. Nonetheless, past studies indicate it is is maintained by the Recovery Program, but incorporating data likely that Razorback Suckers spawn and possibly rear in the from all institutions that mark fish in the upper Colorado River lower San Rafael River (Chart et al. 1999; Bestgen et al. 2002) basin (e.g., universities and other federal and state collabora- and possibly other tributaries (Tyus and Saunders 2001). tors), similar to the structure of the Columbia River database Based on available stocking, survival, and population-size (PTAGIS 2011), will provide a more robust system and will information, use of the San Rafael River by endangered fishes facilitate quick and accurate queries. Nonetheless, we suggest represents a modest fraction of their extant populations in the these technologies, ideal for describing population dynamics Green River basin. Based on the most recent estimate of abun- of highly marked populations with limited natural reproduction dance of Colorado Pikeminnow in 2008 (Bestgen et al. 2010), (Hewitt et al. 2010), should be widely applied in similar tribu- we documented, during our study period, about 1.9% of the taries throughout the basin to fill knowledge gaps regarding the lower Green River subpopulation (adults and juveniles found role of tributary habitats for various life history stages of these between RM 120 and RM 0) using the San Rafael River (15 of endangered fishes. 787 fish), or about 0.3% of the entire Green River population (15 Small tributaries in the upper Colorado River basin may of 4,788 fish). Population estimates for Razorback Sucker are provide spawning and feeding habitat in addition to their large, not available at this time. However, since this population is sus- main-stem counterparts. Spawning success in large systems tained almost entirely by stocking, we are limited to assessing controlled by hydroelectric dams is negatively affected by numbers stocked in the lower Green River from 2004 through modifications from peak and summer and winter base flows, 2009 in relation to published survival rates (5% to the first year blockage of migration routes, and cooler-than-average water at large, 75% in subsequent years; Zelasko et al. 2010; USFWS, temperatures (Vanicek et al. 1970; Weiss et al. 1998; USFWS unpublished stocking data). Based on these data, roughly 1,200 2002a, 2002b, 2002c, 2002d). These alterations also reduce fish would be expected to inhabit the lower Green River during survival of larvae and juvenile fish (Robinson and Childs 2001; 2010 and to be vulnerable to detection in the San Rafael River. Ward and Bonar 2003). Conversely, fish inhabiting tributaries Thus, while these estimates must be viewed with great caution may benefit from more natural temperature regimes in the ab- (i.e., emigration and immigration are ignored), observed use of sence of large dams (Sabo et al. 2012). Furthermore, some other the San Rafael River over the 3-year study could be about 1.7% investigators have argued that stocking tributaries may actually of the lower Green River subpopulation (20 of 1,200 fish), or improve survival of hatchery-reared fish and lead to augmen-

Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 about 0.6% of the entire Green River population (20 of 3,100 tation of existing populations (Irving and Modde 2000) more fish). No population size or survival data are available for Bony- so than stocking larger main-stem rivers (Chart and Cranney tail; however, detections of Bonytails in the San Rafael River 1992; Zelasko et al. 2010). However, associated and additional represent 41% of all captures reported during 2007 through conditions, such as those resulting from water withdrawals (i.e., 2010 in this reach of the Green River (U.S. Fish and Wildlife reduced cues to undergo movements, dewatering, stranding) Service surveys were only conducted in this reach in 2007 and and the negative effects of high densities of nonnative fishes in 2008; USFWS, unpublished data). In terms of average unique the tributaries, also probably limit survival (Ruppert et al. 1993; sightings per year, use of the San Rafael River by Bonytails Walsworth 2011), especially in drought years (Fagan et al. and Razorback Suckers (about 6 fish/year for each species) ex- 2002). As such, in the absence of restoration to a more natural ceeds that observed at the Redlands fish ladder on the Gunnison flow regime (i.e., providing large spring spates and minimum River, a large tributary to the Colorado River, in designated crit- instream flows) and removal of threats to larval and juvenile ical habitat (about two Razorback Suckers and less than one fish, the benefits of inhabiting intermittent tributary streams Bonytail per year; Burdick 2010). may be outweighed by the costs. The successful recovery of the Without the PIA systems, the presence of endangered species endangered fishes will require the expansion of current pop- in the San Rafael River would be underestimated or unknown. ulations and ranges (Maddux et al. 1993), potentially making Remote, passive detections of endangered fish from 2008 to restored tributary habitats vital to future recovery efforts. 592 BOTTCHER ET AL.

Riverine ecosystems are inherently dynamic (Stanford et al. Breidinger provided support and advice. Wally Macfarlane and 2005) and the most productive habitats can change over time Nira Salant provided GIS assistance and computed hydrogeo- scales ranging from days to decades (Stanford et al. 2005; morphic statistics. We thank the many technicians who assisted Schindler et al. 2010; Rypel et al. 2012). Ignoring the diver- this project in the field and laboratory. This research was funded sity of habitats available to and used by a species can result in by the U.S. Bureau of Reclamation, Conservation Activities the loss of “portfolio effects” buffering against poor local envi- for Sensitive Non-Game Native Fish Activities to Avoid Jeop- ronmental conditions (Schindler et al. 2010). As noted above, ardy Program, a S.E. and Jessie E. Quinney Fellowship, and at present, the San Rafael River and other small tributaries not the Ecology Center at Utah State University. Additional support designated as critical habitat or otherwise unprotected under the was provided by the Utah Division of Wildlife Resources, the ESA do not assume a prominent role in the recovery process U.S. Bureau of Land Management, and the U.S. Geological Sur- for endangered fishes in the Colorado River basin. Very little vey, Utah Cooperative Fish and Wildlife Research Unit at Utah observational data on endangered fish occurrence in the San State University (in-kind support). This study was performed Rafael River existed when recovery plans were developed for under the auspices of Utah State University’s IACUC protocol each species (USFWS 1990, 1991, 1998). Thus, even though the number 1310. Mention of brand names in this manuscript does San Rafael River and similar systems can provide one or more not imply endorsement by the U.S. Government. of the same elements that were considered in the designation of critical habitat (i.e., water, physical habitat, and biological REFERENCES attributes such as prey availability; USFWS 1994; Tyus and Badame, P. V. 2008. Population estimates for Humpback Chub (Gila cypha)in Saunders 2001), these areas were not proposed for critical habi- Cataract Canyon, Colorado River, Utah, 2003–2005. Final Report to Upper tat designation. While management actions for the San Rafael Colorado River Endangered Fish Recovery Program, Project 22L, Utah Di- River do appear in the recovery plan (UCREFRP 2011), efforts vision of Wildlife Resources, Salt Lake City. Available: www.coloradoriver recovery.org/documents-publications/technical-reports/research-monitoring. toward their implementation have largely taken place outside html. (November 2011). of the Recovery Program, and formal implementation of in- Baron, J. S., L. Gunderson, C. D. Allen, E. Fleishman, D. McKenzie, L. A. stream minimum flows for native fishes has yet to take place Meyerson, J. Oropeza, and N. Stephenson. 2009. Options for national parks (Bottcher 2009; Fortney et al. 2011; Walsworth 2011; UDWR, and reserves for adapting to climate change. Environmental Management unpublished report; J. Jimenez, Bureau of Land Management, 44:1033–1042. Bestgen, K. R., G. B. Haines, R. Brunson, T. Chart, M. Trammell, R. T. Muth, G. personal communication). Lack of critical habitat designation Birchell, K. Christopherson, and J. M. Bundy. 2002. Status of wild Razorback and other protections under the ESA leave populations with life Sucker in the Green River basin, Utah and Colorado, determined from bas- history strategies that take advantage of small tributary habitats inwide monitoring and other sampling programs. Final Report to Colorado vulnerable to continued anthropogenic impacts. Lastly, in the River Recovery Implementation Program, Project 22D, Larval Fish Labora- face of highly uncertain future environmental conditions, main- tory, Colorado State University, Fort Collins. Available: www.coloradoriver recovery.org/documents-publications/technical-reports/research-monitoring. taining population diversity may be increasingly important for html. (November 2011). these endangered fishes (Luck et al. 2003; Millar et al. 2007; Bestgen, K. R., J. A. Hawkins, G. C. White, K. D. Christopherson, J. M. Baron et al. 2009). Hudson, M. H. Fuller, D. C. Kitcheyan, R. Brunson, P. Badame, G. B. Haines, For the foregoing reasons, we believe that managers may J. A. Jackson, C. D. Walford, and T. A. Sorensen. 2007. Population status of need to reassess the importance of tributaries and the role they Colorado Pikeminnow in the Green River basin, Utah and Colorado. Trans- actions of the American Fisheries Society 136:1356–1380. play to endangered fishes of the Colorado River basin. Investi- Bestgen, K. R., J. A. Hawkins, G. C. White, C. D. Walford, P. Badame, and gating and documenting tributary habitat use should be a priority L. Monroe. 2010. Population status of Colorado Pikeminnow in the Green

Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 for research and monitoring. For example, larval drift surveys River basin, Utah and Colorado, 2006–2008. Final Report to Colorado could be employed to confirm or refute spawning activity for River Recovery Implementation Program, Project 128, Larval Fish Labora- all three endangered species in the lower river and identify im- tory, Contribution 161, Denver. Available: http://www.coloradoriverrecovery. org/documents-publications/technical-reports/research-monitoring.html. portant geomorphic characteristics of spawning reaches. Lastly, (May 2013). some level of protection (e.g., minimum instream flows, habitat Bestgen, K. R., K. A. Zelasko, R. I. Compton, and T. Chart. 2006. Response of restoration) may be needed for these small and highly suscep- the Green River fish community to changes in flow and temperature regimes tible systems, as they are potentially important not only for the from Flaming Gorge Dam since 1996 based on sampling conducted from sensitive fishes but also for the recovery of endangered fishes of 2002 to 2004. Final Report to the Colorado River Recovery Implementation Program, Project 115, Larval Fish Laboratory, Colorado State University, Fort the upper Colorado River basin. Collins. Available: www.coloradoriverrecovery.org/documents-publications/ technical-reports/instream-flow-identification-protection.html#flow. (Nov- ember 2011). ACKNOWLEDGMENTS Bottcher, J. L. 2009. Maintaining population persistence in the face of an ex- We thank Krissy Wilson (UDWR) and three anonymous tremely altered hydrograph: implications for three sensitive fishes in a tribu- reviewers for their thoughtful comments provided on previ- tary of the Green River, Utah. Master’s thesis. Utah State University, Logan. Available: digitalcommons.usu.edu/etd/496. (November 2011). ous drafts of this manuscript. Peter MacKinnon designed, set Burdick, B. D. 2010. Annual operation and maintenance of the fish passage up, and maintained the PIA systems. Craig Walker and Kenny structure at the Redlands diversion dam on the Gunnison River. Report MANAGEMENT BRIEF 593

to the Upper Colorado River Endangered Fish Recovery Program, Lake- tail, and Colorado Pikeminnow for the upper Colorado River endangered wood. Available: www.coloradoriverrecovery.org/documents-publications/ fish recovery program. Upper Colorado River Endangered Fish Recovery work-plan-documents/project-annual-reports.html#II. (November 2011). Program, Lakewood. Cavalli, P. A. 1999. Fish community investigations in the lower Price River, Osmundson, D. B., and K. P. Burnham. 1998. Status and trends of the endan- 1996–1997. Utah Division of Wildlife Resources, Publication 99–21, Salt gered Colorado Squawfish in the upper Colorado River. Transactions of the Lake City. American Fisheries Society 127:957–970. Chart, T. E., and J. S. Cranney. 1992. An evaluation of hatchery reared Bonytail PTAGIS (PIT Tag Information System). 2011. Pit tag information system for the Chub (Gila elegans) released in the Green River of Utah: 1988–1989. Utah Columbia River basin. Pacific States Marine Fisheries Commission, Portland, Division of Wildlife Resources, Salt Lake City. Oregon. Available: www.ptagis.org. (March 2011). Chart, T. E., D. P. Svendson, and L. D. Lentsch. 1999. Investigation of po- Robinson, A. T., and M. R. Childs. 2001. Juvenile growth of native fishes in the tential Razorback Sucker (Xyrauchen texanus) and Colorado Pikeminnow Little Colorado River and in a thermally modified portion of the Colorado (Ptychocheilus lucius) spawning in the lower Green River, 1994 and 1995: River. North American Journal of Fisheries Management 21:809–815. final report. Utah Division of Wildlife Resources, Publication 99-32, Salt Ruppert, J. B., R. T. Muth, and T. P. Nesler. 1993. Predation on fish larvae by Lake City. adult Red Shiner, Yampa and Green rivers, Colorado. Southwestern Naturalist ESA (Endangered Species Act). 1973. U.S. Code, volume 16, sections 1531– 38:397–399. 1544. Rypel, A. L., K. M. Pounds, and R. H. Findlay. 2012. Spatial and temporal Fagan, W. F., P. J. Unmack, C. Burgess, and W. L. Minckley. 2002. Rarity, trade-offs by Bluegills in floodplain river ecosystems. Ecosystems 15:555– fragmentation, and extinction risk in desert fishes. Ecology 83:3250–3256. 563. Finney, S. T. 2006. Colorado Pikeminnow (Ptychocheilus lucius) upstream Sabo, J. L., K. Bestgen, W. Graf, T. Sinha, and E. E. Wohl. 2012. Dams in the of critical habitat in the Yampa River, Colorado. Southwestern Naturalist Cadillac Desert: downstream effects in a geomorphic context. Annals of the 51:262–263. New York Academy of Sciences 1249:227–246. Fortney, S. T., J. C. Schmidt, and D. J. Dean. 2011. Establishing the geomor- Schindler, D. E., R. Hilborn, B. Chasco, C. P. Boatright, T. P. Quinn, L. A. phic context for wetland and riverine restoration of the San Rafael River: final Rogers, and M. S. Webster. 2010. Population diversity and the portfolio report. Intermountain Center for River Rehabilitation and Restoration, Utah effect in an exploited species. Nature 465:609–612. State University, NRCS (Natural Resources Conservation Service) Coopera- Stanford, J. A., M. S. Lorang, and F. R. Hauer. 2005. The shifting habitat tive Agreement 68-3A75-4-155, Logan. mosaic of river ecosystems. Internationale Vereinigung fur¨ theoretische und Fresques, T. D., R. C. Ramey, and G. J. Dekleva. 2013. Use of small tributary angewandte Limnologie Verhandlungen 29:123–136. streams by subadult Colorado Pikeminnows (Ptychocheilus lucius) in Yellow Tyus, H. M. 1987. Distribution, reproduction, and habitat use of the Razorback Jacket Canyon, Colorado. Southwestern Naturalist 58:104–107. Sucker in the Green River, Utah, 1979–1986. Transactions of the American Hanski, I., and M. Gilpin. 1991. Metapopulation dynamics: brief history and Fisheries Society 116:111–116. conceptual domain. Biological Journal of the Linnean Society 42:3–16. Tyus, H. M. 1990. Potamodromy and reproduction of Colorado Squawfish in Hanski, I., and C. D. Thomas. 1994. Metapopulation dynamics and conserva- the Green River basin, Colorado and Utah. Transactions of the American tion: a spatially explicit model applied to butterflies. Biological Conservation Fisheries Society 119:1035–1047. 68:167–180. Tyus, H. M., and C. A. Karp. 1990. Spawning and movements of Razorback Hewitt, D. A., E. C. Janney, B. S. Hayes, and R. S. Shively. 2010. Improving Sucker, Xyrauchen texanus, in the Green River basin of Colorado and Utah. inferences from fisheries capture–recapture studies through remote detection Southwestern Naturalist 35:427–433. of PIT tags. Fisheries 35:217–231. Tyus, H. M., and J. F. Saunders III. 2001. An evaluation of the role of tributary Hoekstra, J. M., W. F. Fagan, and J. E. Bradley. 2002. A critical role for critical streams for recovery of endangered fishes in the upper Colorado River basin, habitat in the recovery planning process? not yet. Ecological Applications with recommendations for future recovery plans. Final Report to the Upper 12:701–707. Colorado Endangered Fish Recovery Program, Project 101, Center for Irving, D. B., and T. Modde. 2000. Home-range fidelity and use of historic Limnology, University of Colorado, Boulder. Available: www.coloradoriver habitat by adult Colorado Pikeminnow (Ptychocheilus lucius) in the White recovery.org/documents-publications/technical-reports/research-monitoring. River, Colorado and Utah. Western North American Naturalist 60:16–25. html. (October 2011). Luck, G. W., G. C. Daily, and P. R. Ehrlich. 2003. Population diversity and UCREFRP (Upper Colorado River Endangered Fish Recovery Program). 2011. ecosystem services. Trends in Ecology and Evolution 18:331–336. Recovery implementation program recovery action plan (RIPRAP). Maddux, H. R., L. A. Fitzpatrick, and W. R. Noonan. 1993. Colorado River UCREFRP, Lakewood. Available: www.coloradoriverrecovery.org/ Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 endangered fishes critical habitat: biological support document. U.S. Fish documents-publications/foundational-documents/recovery-action-plan.html. and Wildlife Service, Salt Lake City, Utah. (November 2011). Marsh, P. C., M. E. Douglas, W. L. Minckley, and R. J. Timmons. 1991. UDWR (Utah Division of Wildlife Resources). 2006. Conservation and man- Rediscovery of Colorado Squawfish, Ptychocheilus lucius (Cyprinidae), in agement plan for three fish species in Utah: addressing needs for Roundtail Wyoming. Copeia 1991:1091–1092. Chub (Gila robusta), Bluehead Sucker (Catostomus discobolus), and Flan- McAda, C. W., C. R. Berry Jr., and C. E. Phillips. 1980. Distribution of fishes in nelmouth Sucker (Catostomus latipinnis). UDWR, Publication 06-17, Salt the San Rafael River system of the upper Colorado River basin. Southwestern Lake City. Naturalist 25:41–49. USFWS (U.S. Fish and Wildlife Service). 1990. Bonytail Chub revised recovery Millar, C. I., N. L. Stephenson, and S. L. Stephens. 2007. Climate change plan. USFWS, Region 6, Denver. and forests of the future: managing in the face of uncertainty. Ecological USFWS (U.S. Fish and Wildlife Service). 1991. Colorado Squawfish revised Applications 17:2145–2151. recovery plan. USFWS, Region 6, Denver. Minckley, W. L. 1973. Fishes of Arizona. Arizona Game and Fish Department, USFWS (U.S. Fish and Wildlife Service). 1994. Endangered and threatened Phoenix. wildlife and plants; determination of critical habitat for the Colorado River Minckley, W. L., and J. E. Deacon, editors. 1991. Battle against extinction: endangered fishes: Razorback Sucker, Colorado Squawfish, Humpback Chub, native fish management in the American West. University of Arizona Press, and Bonytail Chub, final rule. Federal Register 59:54(21 March 1994):13374– Tucson. 13400. Nesler, T. P., K. Christopherson, J. M. Hudson, C. W. McAda, F. Pfeifer, and USFWS (U.S. Fish and Wildlife Service). 1998. Razorback Sucker (Xyrauchen T. E. Czapla. 2003. An integrated stocking plan for Razorback Sucker, Bony- texanus) recovery plan. USFWS, Region 6, Denver. 594 BOTTCHER ET AL.

USFWS (U.S. Fish and Wildlife Service). 2002a. Colorado Pikeminnow Walsworth, T. E. 2011. Analysis of food web effects of non-native fishes and (Ptychocheilus lucius) recovery goals: amendment and supplement to the evaluation of restoration potential for the San Rafael River, Utah. Master’s Colorado Squawfish recovery plan. USFWS, Mountain-Prairie Region 6, thesis. Utah State University, Logan. Denver. Ward, D. L., and S. A. Bonar. 2003. Effects of cold water on susceptibility of USFWS (U.S. Fish and Wildlife Service). 2002b. Bonytail (Gila elegans) recov- age-0 Flannelmouth Sucker to predation by Rainbow Trout. Southwestern ery goals: amendment and supplement to the Bonytail Chub recovery plan. Naturalist 48:43–46. USFWS, Mountain-Prairie Region 6, Denver. Weiss, S. J., E. O. Otis, and O. E. Maughan. 1998. Spawning ecology of Flan- USFWS (U.S. Fish and Wildlife Service). 2002c. Humpback Chub (Gila cypha) nelmouth Sucker, Catostomus lattipinnis (Catostomidae), in two small trib- recovery goals: amendment and supplement to the Humpback Chub recovery utaries of the lower Colorado River. Environmental Biology of Fishes 52: plan. USFWS, Mountain-Prairie Region 6, Denver. 419–433. USFWS (U.S. Fish and Wildlife Service). 2002d. Razorback Sucker (Xyrauchen Wick, E. J., J. A. Hawkins, and T. P.Nesler. 1991. Occurrence of two endangered texanus) recovery goals: amendment and supplement to the Humpback Chub fishes in the Little Snake River, Colorado. Southwestern Naturalist 36:251– recovery plan. USFWS, Mountain-Prairie Region 6, Denver. 254. Vanicek, C. D., R. H. Kramer, and D. R. Franklin. 1970. Distribution of Green Zelasko, K. A., K. R. Bestgen, and G. C. White. 2010. Survival rates and River fishes in Utah and Colorado following closure of Flaming Gorge Dam. movement of hatchery-reared Razorback Suckers in the upper Colorado River Southwestern Naturalist 14:297–315. basin, Utah and Colorado. Transactions of the American Fisheries Society Walker, C. A., and M. Hudson. 2004. Surveys to determine the current distri- 139:1478–1499. bution of the Roundtail Chub, Flannelmouth Sucker, and Bluehead Sucker in Zydlewski, G. B., G. Horton, T. Dubreuil, B. Letcher, S. Casey, and J. Zydlewski. the San Rafael drainage, during 2003. Utah Division of Wildlife Resources, 2006. Remote monitoring of fish in small streams: a unified approach using Salt Lake City. PIT tags. Fisheries 31:492–502. Downloaded by [Department Of Fisheries] at 20:09 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:10 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Differentiating between Sampling- and Environment- Related Mortality in Shortnose Sturgeon Larvae Collected Using Anchored D-Frame Nets Sima Usvyatsov a , James Watmough b & Matthew K. Litvak a a Department of Biology , Mount Allison University , 63B York Street, Sackville , New Brunswick , E4L 1G7 , Canada b Department of Mathematics and Statistics , University of New Brunswick , Post Office Box 4400, Fredericton , New Brunswick , E3B 5A3 , Canada Published online: 24 May 2013.

To cite this article: Sima Usvyatsov , James Watmough & Matthew K. Litvak (2013): Differentiating between Sampling- and Environment-Related Mortality in Shortnose Sturgeon Larvae Collected Using Anchored D-Frame Nets, North American Journal of Fisheries Management, 33:3, 595-605 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785995

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:595–605, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785995

ARTICLE

Differentiating between Sampling- and Environment-Related Mortality in Shortnose Sturgeon Larvae Collected Using Anchored D-Frame Nets

Sima Usvyatsov* Department of Biology, Mount Allison University, 63B York Street, Sackville, New Brunswick E4L 1G7, Canada James Watmough Department of Mathematics and Statistics, University of New Brunswick, Post Ofﬁce Box 4400, Fredericton, New Brunswick E3B 5A3, Canada Matthew K. Litvak Department of Biology, Mount Allison University, 63B York Street, Sackville, New Brunswick E4L 1G7, Canada

Abstract This study provided the ﬁrst attempt to differentiate between environmental- and sampling-related mortality in larval Shortnose Sturgeon Acipenser brevirostrum. Anchored D-frame nets are widely used for sampling sturgeon larvae, but the associated larval mortality rates are not known. We assessed larval mortality (1) by examining the effect of net deployment period on mortality rate; (2) directly, by deploying live larvae in nets and observing survival over time; and (3) indirectly, by establishing the time of larval death based on body decomposition. Larval mortality in nets deployed for 12 and 24 h was similar: 50% of the nets contained no live larvae. After a 6-h deployment, half of the nets contained 60% or more live larvae. After placement of live larvae in nets for 12 h, the nets retained 60.0 ± 20.4% (mean ± SE) of the deployed larvae. Of the retained larvae, only 6.5 ± 2.3% (mean ± SE) survived 12 h inside the net. Decomposition rates of dead larvae were examined both in a controlled environment and in the nets. Dead larvae that were deployed in nets decomposed faster than those in a controlled environment, with high decay after only 12 h of deployment. A discriminant function analysis differentiated well between larvae that had been dead for 9 h and larvae that had been dead for 21 or 32 h. The assessment of time of death based on decomposition was used to identify the cause of mortality for Shortnose Sturgeon larvae collected in the Saint John River, New Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 Brunswick, during 2008–2010. We propose that in short to medium deployment periods (up to overnight), pristine and partly decayed larvae can be assigned to sampling mortality, while highly decayed larvae can be assigned to environment-related mortality. Using this distinction, sampling and environmental sources of mortality accounted for at least 20–56% and 4–25%, respectively, of the sampled Shortnose Sturgeon larvae.

Sturgeons (family Acipenseridae) are found throughout the sturgeon populations (Birstein et al. 1997). The Shortnose Stur- North Temperate Zone of Asia, Europe, and North America. geon Acipenser brevirostrum is currently listed as endangered Although sturgeons were once abundant, the heavy fishing for under the U.S. Endangered Species Act and has been designated caviar and flesh, combined with river fragmentation and flow a species of special concern under the Species at Risk Act in regulation, led to a dramatic reduction in numbers for most Canada (COSEWIC 2005). The habitat range of the Shortnose

*Corresponding author: [email protected] Received June 13, 2012; accepted March 7, 2013 595 596 USVYATSOV ET AL.

Sturgeon extends from Florida, USA, to New Brunswick, mortality rates. Moreover, most studies involving larval collec- Canada. The Saint John River in New Brunswick is the northern- tions have not reported mortality levels; the exception was a most occurrence of the Shortnose Sturgeon and represents the study by Caroffino et al. (2009), who reported 16% mortality in species’ only known habitat in Canada (Dadswell et al. 1984; Lake Sturgeon larvae that were collected in hourly checked nets. Kynard 1997). The Saint John River population of Shortnose The lack of information on mortality rates and the variability Sturgeon is one of the largest throughout the species’ habitat in sampling make it harder to identify whether the mortality of range (Kynard 1997), making this population a useful study captured larvae results from sampling design and gear use or model, as larvae can be readily obtained in the wild. from the effects of dams and natural mortality. Spawning and larval migration strategies are similar among We aimed to establish the cause of mortality among Short- many sturgeon species. Reproductively active adults may mi- nose Sturgeon larvae that were collected in D-frame nets during grate up to 100–200 km to reach spawning grounds in their natal larval migration in the Saint John River, New Brunswick. To rivers. This behavior was described for several species, including achieve this, we used three approaches. First, we assessed the Shortnose Sturgeon (Kynard 1997; Kynard and Horgan 2002), effect of net deployment duration on the mortality of captured Lake Sturgeon A. fulvescens (Auer 1996), and Chinese Stur- wild larvae. Second, we determined mortality rates of marked geon A. sinensis (Wei et al. 2009). Once spawned, the adhesive live larvae that were placed in the nets. Lastly, we determined eggs remain attached to the substrate until they hatch (Dadswell larval mortality indirectly by establishing the time of death via et al. 1984; Kynard 1997; Bruch and Binkowski 2002). During analysis of decomposition stage (i.e., taphonomy). In the case of the larval stage, age-0 sturgeon migrate downstream toward the larval decay in nets, changes to soft tissues occur within hours nursery grounds (Kynard 1997; Auer and Baker 2002; Kynard to days; hence, the analysis has great potential to help us de- and Horgan 2002). In Shortnose Sturgeon, two types of migra- termine the time of death and therefore its possible cause. The tion may occur: (1) if disturbed, young yolk sac larvae leave the taphonomic approach was used here to estimate time of death substrate, swim up into the water column, and passively drift for (1) marked and unmarked Shortnose Sturgeon larvae that downstream with the water flow; and (2) older larvae actively were held in a controlled environment, (2) marked larvae that swim downstream (Kynard and Horgan 2002). were subjected to field conditions, and (3) wild larvae that were Since most sturgeon rivers are now fragmented by dams collected during 2008–2010. The results of our study will allow (Beamesderfer and Farr 1997; Kynard 1997; NMFS 1998), for a better understanding of the causes of mortality in larval spawning often occurs immediately downstream of the lower- sturgeon catches, assist in providing guidelines for gear use, and most dam on the river (Kynard 1997). Due to the dams’ vicinity facilitate assessment of mortality rates in the wild. to spawning grounds, larvae are often subjected to highly fluctuating flow regimes and to fast changes in water levels. Rapid reductions in flow rate can easily result in stranding (Clarke METHODS et al. 2008), which leads to desiccation and death, and high Study site.—The Saint John River is New Brunswick’s largest turbulence can cause larval mortality as well (Killgore et al. river, flowing for 673 km through Maine, Quebec, and New 1987, 2001). Other potential sources of larval mortality include Brunswick (Benke and Cushing 2005). The main stem of the gas bubble trauma due to gas supersaturation downstream of lower Saint John River spans approximately 150 km between dams (Counihan et al. 1998; Clarke et al. 2008) as well as Mactaquac Dam and the river’s mouth at Reversing Falls, where failure to initiate feeding (Conte et al. 1988). Levels of larval it flows into the Bay of Fundy (Figure 1). The only known mortality in the wild are technically hard to assess. Sampling spawning site of the Shortnose Sturgeon population in the Saint

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 gear can be difficult to deploy and retrieve, and this difficulty John River is immediately downstream of Mactaquac Dam is exacerbated by the scale of sampling required to assess lar- (COSEWIC 2005). Apart from spring freshet and rainfall floods, val drift in many of the larger rivers in which sturgeon species Mactaquac Dam operates in peak-load mode. The discharge occur. levels that change according to electricity demand (Jessop and Anchored drift nets (D-frame nets) are widely used for the Harvie 2003) subject the larvae to altered and fluctuating flow sampling of sturgeon larvae (e.g., Benson et al. 2006; Caroffino regimes. We focused our efforts on larval collection at two tran- et al. 2009; Wei et al. 2009). The nets are set downstream of sects, which were positioned 12.5 and 17.0 km downstream the spawning location and capture larvae during their migratory from Mactaquac Dam (Figure 1). drift. Depending on project goals and the size of the sampled Shortnose Sturgeon larval sampling in the field.—Ten to river, various periods of D-frame net deployment have been fifteen anchored drift (D-frame) nets were used to collect Short- used (e.g., from 5 min to 2 h: Wei et al. 2009; 1 h: Auer 1996 nose Sturgeon larvae during their downstream migration from and Caroffino et al. 2009; overnight: Auer and Baker 2002; and the spawning grounds. The nets consisted of a 3.5-m-long mesh 6–11 h: Sykes 2009). The deployment period that minimizes cone made of 1.6-mm, knotless delta mesh. The cone’s mouth mortality was suggested to be 10 min to 1 h (Moser et al. 2000) (0.85 m in diameter) was held open by a D-shaped stainless-steel or, more recently, 1–3 h (Kahn and Mohead 2010). However, to hoop (0.9 m high; 1 m wide). The net’s end (0.3 m in diameter) our knowledge, this gear type has not been tested for associated opened to a cod end that collected the sample. Each net was MORTALITY IN LARVAL SHORTNOSE STURGEON 597

FIGURE 1. (A) The Atlantic coast, showing the Canadian Maritime Provinces, Quebec, and Maine; (B) the lower Saint John River, extending from Mactaquac Dam (upstream to Fredericton) to Saint John Harbour; and (C) the study site, with the dam and the two net transects indicated by arrows.

held in place on the river bottom by a Danforth anchor, which of Mactaquac Dam throughout 2008–2010 by Fisheries and was attached to the net via a short bridle that connected to a 15- Oceans Canada. Dam discharge levels (hourly averages; m3/s) m-long rope. The nets were positioned in two transects across were provided by NB Power, which recorded both generation the river channel (5 stations/transect) at 12.5 and 17.0 km below and spill flows. Mactaquac Dam (Figure 1). The number of stations per transect Effect of net deployment period on mortality and decompo- was chosen to represent shore, channel, and midpoint habitats. sition.—To assess the effect of net deployment period on the In 2008, each transect consisted of five nets that were evenly survival of Shortnose Sturgeon larvae, we used the 2009–2010 spaced across the channel. In 2009 and 2010, the downstream upstream transect design of 10 nets deployed at 5 stations as de- transect had five nets, while the upstream transect had 10 nets scribed above. The two nets within each station were set as close positioned at five stations that were evenly spaced across the as possible in order to minimize differences in environmental channel; each station consisted of two nets that were set 3–5 m characteristics. During 4–23 June 2009, one of the two nets apart. This design allowed us to deploy the two nets within each deployed at each station was checked twice per day (12-h de- station for different periods of time to compare mortality rates ployments), while the other net was checked once per day (24-h (see below). deployment). On 3–4 June 2010, we used the same setup to com- In 2008, the samples were examined for the presence of pare 6- and 12-h net deployments. The 6-h nets were checked Shortnose Sturgeon larvae immediately after each net was lifted. in the morning, at noon, in the evening, and at midnight. In 2009 and 2010, the cod ends of the nets were rinsed into in- We examined mortality rates by summing the catches from dividual 2-L pails and were taken to shore after all nets had nets with short deployments and comparing them with catches

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 been checked. The contents of the pails were gently transferred from the longer deployments. In 2009, all 12-h (morning and into a shallow tray for easier examination. During all 3 years of evening) deployments were summed to compare the resulting sampling, the mesh in the first 70 cm of the net adjacent to the larval survival with that from the 24-h (morning) deployments; cod end was examined for the presence of larvae. Although the in 2010, every two deployments of 6 h were summed to compare terms “free embryo” and “eleutheroembryo” are often used for survival with that from the 12-h deployments. The resulting describing newly hatched sturgeon, for the purpose of this study proportions (i.e., number of dead larvae out of the total number we considered hatched sturgeon to be “larvae” until they became of larvae in the catch) were compared by using randomization juveniles, which are characterized by the possession of fully de- tests. For each study year, we calculated the F-statistic associ- veloped fins and scutes. All sturgeon larvae were separated into ated with the effect of deployment period on larval mortality by live and dead groups, counted, and digitally imaged by using a using a one-way ANOVA. Following Manly (2007), we then 10-megapixel digital camera (Pentax Optio W60). Larvae that randomized the mortality rates and recalculated the F-statistic were alive but damaged were included in the live group, although of the effect of deployment period on the observed larval mortal- such larvae were rare. In the laboratory, larval images were an- ity. This process was repeated 10,000 times, creating an array of alyzed for SL by using ImageJ software (Abramoff` et al. 2004). F-values that would reasonably occur under the null hypothesis Daily averages of water temperature were determined by of no significant deployment effect. We then compared our orig- using a temperature logger that was deployed 1 km downstream inal F-value against the randomization-based distribution of 598 USVYATSOV ET AL.

F-values and calculated the percentage of replications in which the resampled F-values exceeded the original F.We used randomization tests instead of parametric tests because randomization-based approaches are free of assumptions about the distribution and therefore are considered more robust (Manly 2007). Deployment of live Shortnose Sturgeon larvae in nets.— Mortality of Shortnose Sturgeon larvae was directly assessed by deploying live, wild-captured larvae in anchored D-frame nets and monitoring their survival. Once the larvae were deployed in the nets, we had to be able to distinguish them from new larvae that had drifted into the net during the experiment. It was not possible for us to avoid this problem by performing the experiment upstream of the spawning site since (1) the spawning activity of Shortnose Sturgeon extends to the tailrace of Mactaquac Dam (M. K. Litvak, unpublished data) and (2) the environment in the Mactaquac Dam head pond may not represent the downstream riverine conditions. In addition, the alternative of performing these experiments in the laboratory was not feasible because the naturally experienced flows (at or above 1 m/s) are difficult to replicate in a flume. Therefore, we performed the experiments in situ and used vital staining to separate deployed and newly captured larvae. Calcein is a fluorescent chromophore that binds to calcium and can be visually detected without sacrificing fish (Mohler 1997; T. Bell, J. Bowker, M. Bowman, U.S. Fish and FIGURE 2. Shortnose Sturgeon larvae: pristine larva under white light; a Wildlife Service Aquatic Animal Drug Approval Partnership calcein-stained, pristine larva under ultraviolet light; a partly decomposed (al- Program, and G. Kindschi, Bozeman Fish Technology Center, tered) larva; and a decayed larva. Note changes in color, myomeres, notochord unpublished). A total of 25 live Shortnose Sturgeon larvae were integrity, eye shape and size, and fin fraying. Decomposition characters (see text for details) are (left to right) the head (coded as h), barbels (ba), eyes (e), stained in a 1-g/L solution of calcein (MP Biomedicals, LLC, mouth (m), pectoral fins (p), belly (bel), notochord integrity (n), myomeres (m), Solon, Ohio) for 12 h. This concentration was based on the firmness (firm), fin fold (fin), and caudal fin (c). results of a preliminary study, which indicated no delayed mortality in calcein-treated larvae (S. Usvyatsov, unpublished data). After staining, the larvae were rinsed in river water and were decomposed or altered state, or 3 for the worst possible condi- assessed for fluorescence under an ultraviolet lamp (Blak Ray tion (e.g., disintegrated or missing tissues). These scores were UVL-22). Five of the live, fluorescing larvae were then deployed then analyzed as integer values. Notochord integrity was scored in each of the five D-frame nets in the upstream transect for 12 h only as 1 or 3 because it was a binary variable distinguishing to assess mortality. The larvae were gently placed into the cod whether the notochord was nonsevered or severed. end, which contained approximately 200 mL of water, and the Taphonomic assessment: net decomposition.—Twenty-five

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 cod end was then attached to the net. The cod end was held live Shortnose Sturgeon larvae captured in the wild were stained upright during deployment to keep the larvae submerged and to with calcein as described above. Only larvae that appeared prevent them from escaping. The net was allowed to fully open healthy and active were used for these trials. The larvae were to the water flow, and only then was slowly lowered to the bottom euthanized in a 100-mg/L solution of tricaine methanesulfonate of the river. At net check, all larvae were separated from debris (MS-222); to monitor decomposition under field conditions, and examined under ultraviolet light. The numbers of dead and we deployed five dead larvae for 12 h in each of the five D- live calcein-stained larvae were recorded. The experiment was frame nets constituting the upstream transect. At net check, the performed twice (at morning and evening deployments). decomposition level of all fluorescent larvae was assessed by Taphonomic assessment.—The degree of decay was assessed using the characters described above. To reduce the risk of dam- by assigning decomposition scores. We followed Sansom et al. aging the examined structures, we used a large baster rather than (2010) by listing body characters that could be subjected to forceps to move and turn the larvae. The retrieved fluorescent, decomposition: eyes, head, mouth, barbels, fin fold, pectoral dead larvae were then redeployed for an additional 12-h period, fins, caudal fin, myomeres, firmness, belly, coloration, body followed by another assessment of decomposition. The experi- curvature, notochord integrity, and overall state (Figure 2). Each ment was repeated with 25 more dead larvae. We also analyzed character was evaluated by assigning a score of 1 for pristine data from larvae that were deployed live but died during the net condition (i.e., indistinguishable from a live larva), 2 for a partly mortality experiment. Their exact time of death was not known MORTALITY IN LARVAL SHORTNOSE STURGEON 599

but would have been between 0 and 12 h. River water tempera- SL × sampling year interaction. We calculated the z-statistic ture at the time of capture and deployment was 16–18◦C. associated with each variable. Following Manly (2007), we then Taphonomic assessment: controlled environment.—We as- randomized the larval states and calculated a new z-statistic. This sembled a constant-flow apparatus with individual chambers; it was repeated 10,000 times, creating an array of z-values that consisted of a closed-circle pipe that branched into 10 chambers. would reasonably occur under the null hypothesis of no signifi- The system was connected to a submersible pump and was set cant SL or year effect. We then calculated the percentage of repli- over a sheet of Styrofoam to ensure flotation and equal water cations in which the resampled z exceeded the original z-value. flow to all chambers. Each chamber was enclosed with 0.5-mm mesh on both ends, providing a secure individual environment RESULTS and allowing flow-through. The system was deployed in the Saint John River to provide water composition and temperature Shortnose Sturgeon Larval Sampling in the Field similar to those experienced by larvae in the wild. Water temperature during Shortnose Sturgeon larval drift To account for the possible effect of staining on decompo- ranged between 14.6◦C and 20.1◦C in 2008, between 12.9◦C sition, we included both stained and unstained larvae in the and 18.1◦C in 2009, and between 10.9◦C and 18.3◦C in 2010 controlled-environment trial. Ten Shortnose Sturgeon larvae (Figure 3). Although water temperature at the beginning of larval were stained with calcein as described above. The 10 stained drift was lower in 2010 than in the other years, the majority of larvae and additional 10 unstained larvae were euthanized in larval migration occurred after water temperature increased to 100-mg/L MS-222. Two dead larvae—one stained and one unstained—were placed into each tube. Larval decay was examined at 9, 21, and 32 h as described above. To test for differences between postmortem periods (9, 21, and 32 h), we performed a principal components analysis (PCA) using the package vegan (Oksanen et al. 2009) in R (R Develop- ment Core Team 2010). The principal component (PC) scores were calculated by using a correlation matrix of the decomposition scores of individual characters. The scores associated with each PC axis were then input as predictors in discriminant function analysis (DFA) by using the package MASS (Venables and Ripley 2002) in R (R Development Core Team 2010). Separate PC-DFAs were performed on data describing stained and unstained larvae from the controlled environment. Each DFA was performed on the scores of the first seven PCs for each group because the PCA captured over 90% of the variance in the first seven axes and because the seventh and subsequent eigenval- ues were below the eigenvalue average (Legendre and Legendre 1998). The discriminant model was tested using a leave-one-out cross validation. Decomposition of wild Shortnose Sturgeon larvae in 2008–

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 2010.—The 2008–2010 larval collections resulted in a large database of images of Shortnose Sturgeon larvae. The images could not be used to inspect myomeres, ﬁn structures, or body ﬁrmness. Therefore, instead of a suite of decomposition values, we assigned a single “overall state” score based on the overall integrity of each larva (Figure 2). For the 2008–2010 image data set, the scores were applied only to catches from nets that were deployed for less than 16 h. For longer deployment periods, even the larvae that died due to sampling would become highly decomposed, thus reducing our ability to differentiate between the two sources of mortality. The effects of Shortnose Sturgeon larval size and year on larval mortality were examined by using randomization tests in R FIGURE 3. Water temperature (solid line, secondary y-axis), daily cumulative (R Development Core Team 2010). We used generalized linear dam discharge (dotted line, secondary y-axis), and daily total counts of Shortnose models to perform logistic regression of larval state (live or pris- Sturgeon larvae (primary y-axis) in the Saint John River, 2008–2010. Note that tine) against larval SL (cm), sampling year (2008–2010), and the the scale of the primary y-axis differs among the panels. 600 USVYATSOV ET AL.

a daily average of 15.8◦C. Dam discharge differed between the than in the 12-h deployments (25th, 50th, and 75th quantiles = 3 years (Figure 3), with high-flow events resulting from heavy 0, 0, and 100% [24 h]; 0, 17, and 50% [12 h]). rainfall. The total number of captured Shortnose Sturgeon larvae was Deployment of Live Shortnose Sturgeon Larvae in Nets 2,251 in 2008; 460 in 2009; and 2,100 in 2010 (Figure 3). Of the 25 live Shortnose Sturgeon larvae that were deployed ± ± Dead larvae constituted 29% of the collections in 2008, 69% in in nets during the morning, 64 18.3% (mean SE for 2009, and 73% in 2010. The difference in percentage of dead the five nets) were retrieved after 12 h. Of the retrieved larvae, ± ± larvae between 2008 and 2009–2010 is probably attributable to 87 8.3% (mean SE for the five nets) were dead. Of the the change in sampling technique, since in 2009 and 2010 the 25 live larvae that were deployed in nets during the evening, ± ± samples were transferred into containers and moved to shore for 56 13.3% (mean SE for the five nets) were retrieved after processing. We found no effect of transect on mortality when 12 h, and all of the retrieved larvae were dead. Retrieval success analyzing net deployments of 16 h or less. Dead larvae in the varied between the nets, ranging from a single larva (i.e., 20%) upstream and downstream transects constituted 23% and 30% to all five larvae (i.e., 100%). The overall retrieval success was ± ± ± ± of the catch, respectively, in 2008; 76% and 52% in 2009; and 60 10.7% (mean SE), and 93.5 4.5% (mean SE) 65% and 81% in 2010. of the retrieved larvae were dead. The SLs of the live Shortnose Sturgeon larvae and the dead Taphonomic Assessment: Net Decomposition but pristine larvae ranged between 9.49 and 22.8 mm. The length In the first deployment of dead Shortnose Sturgeon larvae, distribution of larvae changed throughout the drift period, with 40 ± 24.5% (mean ± SE for all five nets) of the 25 larvae smaller larvae being prevalent early in the drift period and larger that were placed in the nets were retrieved after 12 h. In the larvae predominating in the middle and end of the drift period. second deployment of dead larvae, 60 ± 17.9% (mean ± SE) However, newly hatched larvae were captured throughout the of the larvae were retrieved from the nets after 12 h. Overall re- drift period (Usvyatsov et al. 2012), indicating an overlap in trieval success for larvae in both deployments was 50 ± 14.7% hatching and drifting periods. We found no significant effect of (mean ± SE). × larval SL, sampling year, or the SL sampling year interaction The decomposition of net-deployed, calcein-stained Short- on larval mortality (randomization tests of generalized linear nose Sturgeon larvae progressed with time. In general, firmness = models: P 0.063, 0.426, and 0.063, respectively). and coloration were the fastest-decomposing characters, as they were recorded as pristine in only 12% and 28% of the larvae, Effect of Net Deployment Period on Mortality respectively, after 12 h and in 0% of the larvae after 24 h. De- and Decomposition terioration of firmness and coloration was followed by changes Six-hour net deployments had the lowest mortality rates (me- in the barbels, head, fin fold, and pectoral fins; these characters = = dian 40%; 25th and 75th quantiles of mortality 20% and were recorded as pristine in 40, 56, 64, and 64% of larvae, re- 75%, respectively), while mortality rates for the 12-h deploy- spectively, after 12 h and in 5, 43, 38, and 29% of larvae after = ments in 2010 were higher and more variable (median 50%; 24 h. Firmness was the only character for which a score of 3 was = 25th and 75th quantiles 0% and 83%). Mortality from the present at a markedly higher percentage among dead-deployed = 12- and 24-h deployments in 2009 was higher still (median larvae at 24 h, increasing from 24% of larvae after 12 h of de- = 100% for both; 25th and 75th quantiles 67% and 100% [12 h]; ployment to 90% of larvae after 24 h. After 12 h of deployment, 63% and 100% [24 h]). Randomization tests found no signifi- 20% of the larvae were assigned an overall state of pristine, and cant difference between the 12- and 24-h deployments in 2009 76% were assessed as having an overall state of partly decayed.

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 = (P 0.651) or between the 6- and 12-h deployments in 2010 After 24 h of deployment, 0% of the larvae were assessed to be = (P 0.809). However, a significant difference was found pristine overall, and 90% were considered to be partly decayed between the 12-h deployments in 2009 and those in 2010 overall. The “live” subset, which consisted of live larvae that = (P 0.001). were placed in the nets as part of the mortality experiment but In addition to differences in Shortnose Sturgeon larval mor- were dead at retrieval, had decomposition values very similar tality rate between the 6-, 12-, and 24-h net deployments, we also to those of the dead-deployed larvae at 12 h. However, live- observed differences in larval decomposition level. The lowest deployed larvae had higher decay in fins (fin fold, caudal fin, level of decomposition was recorded for the 6-h deployments: and pectoral fins) than dead-deployed larvae at 12 h: 59, 56, in 75% of these samples, at least 70% of the dead larvae were in and 41% of larvae in the live subset retained pristine condition pristine condition. In the remaining 25% of samples, 50–70% of of the respective characters, whereas 64, 76, and 64% of dead- the dead larvae were pristine. In the 2009 comparative deploy- deployed larvae retained pristine condition of these characters ments, the proportion of collected dead larvae that were pristine after 12 h. was lower in the 24-h deployments than in the 12-h deployments (25th, 50th, and 75th quantiles = 0, 0, and 55% [24 h]; 0, 37, and Taphonomic Assessment: Controlled Environment 100% [12 h]). As expected, highly decomposed larvae consti- In the controlled environment, unstained Shortnose Sturgeon tuted a larger proportion of dead larvae in the 24-h deployments larvae decomposed faster than calcein-stained larvae. Within MORTALITY IN LARVAL SHORTNOSE STURGEON 601

each subset (stained and unstained), decomposition progressed Shortnose Sturgeon larvae in the controlled environment de- with increasing time postmortem (9, 21, and 32 h). However, composed more slowly than net-deployed larvae. For example, relative to the stained larvae, the unstained larvae had a higher after 12 h of deployment in the nets, pristine scores for firmness percentage of 3 scores and a lower percentage of 1 scores at and coloration were recorded in 12% and 28% of the stained each checkpoint. The fastest-degrading characters in unstained larvae, respectively. In comparison, after 9 h in the controlled larvae were the barbels, fin fold, pectoral fins, and caudal fin, environment, pristine scores for firmness and coloration were as they remained pristine in 0, 10, 0, and 30% of larvae, re- recorded in 70% and 60% of the stained larvae, respectively. spectively, at 9 h postmortem. Between 9 and 21 h postmortem, After 24 h in the nets, 90% of the larvae had highly deteriorated the head, myomeres, and coloration proceeded to degrade, as firmness, whereas after 21 h in the controlled environment, only the percentage of larvae with a score of 2 increased from 30, 30% of the larvae had a score of 3 for firmness. However, 40, and 60%, respectively, at 9 h to 67, 67, and 100% at 21 h. the overall state scores assessed for controlled-environment lar- By 32 h postmortem, only the eyes and notochord integrity re- vae at 21 h postmortem were similar to those assigned to net- tained a score of pristine in the majority of larvae (56% for both deployed larvae after 24 h. characters). The overall larval state was assessed as pristine and The discriminant model for unstained Shortnose Sturgeon partly decomposed in 10% and 80% of the unstained larvae, larvae was significant (Wilks’ lambda: P = 0.0014). The first lin- respectively, at 9 h postmortem; in 0% and 78% of the larvae at ear discriminant axis (LD1) explained 90% of the variation and 21 h; and in 0% and 67% of the larvae at 32 h. separated the 9-h postmortem cases from the 21- and 32-h cases In stained larvae, the fastest-degrading characters were the (Figure 4). The second axis (LD2) explained the remaining 10% pectoral fins and caudal fin, which remained pristine in 30% of variation and discriminated between the 21-h cases and the and 60% of the larvae, respectively, at 9 h postmortem; in 0% 32-h cases. The model correctly classified most of the cases in all and 50% of the larvae at 21 h; and in 0% and 10% of the larvae three categories (9, 21, and 32 h postmortem); however, most of at 32 h. A higher degree of degradation was seen at the 21-h the 21- and 32-h cases were misclassified during cross validation checkpoint, with low percentages of pristine scores for the fin (Table 1). That said, based on our review of the literature, D- fold, pectoral fin, firmness, and coloration; percentages of larvae frame nets are not usually set for longer than 12 h. Therefore, if with a score of 1 for these characters decreased from 100, 30, we combine the results for 21 and 32 h, the model correctly clas- 70, and 60%, respectively, at 9 h postmortem to 70, 0, 0, and sified 80% of the 9-h cases, correctly classified 100% of the 21- 0% at 21 h. By 32 h postmortem, the head, barbels, fin fold, and 32-h cases, and misclassified a single cross validation case. pectoral fins, caudal fin, firmness, and coloration all had high The discriminant model for stained Shortnose Sturgeon lar- decomposition scores (20, 30, 50, 10, 40, 70, and 10%, respec- vae was also significant (Wilks’ lambda: P < 0.001). The LD1 tively). The overall state of the stained larvae was assessed as explained 95% of the variation and discriminated among the partly decomposed and pristine in 40% and 50% of the larvae, 9-, 21-, and 32-h postmortem cases (Figure 4). The LD2 ex- respectively, at 9 h postmortem; in 90% and 0% of the larvae at plained 5% of the variation and separated the 21-h cases from 21 h; and in 80% and 0% of the larvae at 32 h. the cluster of 9- and 32-h cases. The model correctly classified all Downloaded by [Department Of Fisheries] at 20:10 28 May 2013

FIGURE 4. Plots of discriminant function analyses of Shortnose Sturgeon larval decomposition in a controlled environment. Unstained and stained larvae after a deployment of 9 h (open circles), 21 h (black shaded circles), or 32 h (open triangles) are plotted individually along the ﬁrst and second linear discriminant axes (LD1 and LD2, respectively). In addition, group centroids are plotted (“ + ” symbols). The percentage of variation explained by each axis is given in the axis label. 602 USVYATSOV ET AL.

TABLE 1. Classification (%) of decomposition cases of Shortnose Sturgeon TABLE 2. Distribution (%) of overall decomposition scores for Shortnose larvae in a controlled environment after the discriminant model was fitted and Sturgeon larvae in a controlled environment; net-decomposing larvae; and wild after a leave-one-out cross validation was performed. Cases were classified larvae collected in the Saint John River during 2008–2010. Cases are classified according to the length of the decomposition period (9, 21, or 32 h postmortem). according to decomposition level (1 = pristine; 2 = somewhat decomposed; In perfect discrimination, all data would be found in the table’s diagonals, as no 3 = highly decomposed) and the length of the decomposition period (9, 21, or cases would be misclassified. 32 h postmortem in the controlled environment; 12 or 24 h in the nets). Wild larvae (2008–2010 collections) are classified as live or according to their state Fitted classification Cross validation of decomposition if dead. Original classification 9 h 21 h 32 h 9 h 21 h 32 h Unstained larvae Calcein-stained larvae Period Live Unstained larvae or year (%) 1 2 3 N 123N 9 h 80 10 10 80 10 10 21 h 0 89 11 0 44 56 Controlled environment 32 h 0 11 89 11 67 22 9 h 10 80 10 10 50 40 10 10 21 h 0 78 22 9 0 90 10 10 Calcein-stained larvae 32 h 0 67 33 9 0 80 20 10 9 h 100 0 0 80 20 0 21 h 0 100 0 20 50 30 Net decomposition 32 h 0 20 80 0 30 70 Live 15 81 4 27 12 h 20 76 4 25 24 h 0 90 10 21 9- and 21-h cases and most of the 32-h cases; the cross validation Wild larvae misclassified several more cases than the original discrimination 2008 76 11 9 4 937 (Table 1). Using the 12-h cutoff, the model correctly classified 2009 27 30 18 25 239 80% of the 9-h cases and 60% of the 21- and 32-h cases. 2010 29 38 18 14 1,707 The DFA graphs (Figure 4) show the difference in decomposition between unstained and stained Shortnose Sturgeon larvae. The stained larvae are spread across the LD1, whereas for the state of live-deployed larvae was comparable to that of the unstained larvae the 21- and 32-h cases are clumped. This is to be dead-deployed larvae at 12 h postmortem. Decomposition of expected, considering the faster decomposition of the unstained stained larvae in the controlled environment was slower than larvae. that of stained larvae deployed in the nets, as detailed above. In addition, decomposition of unstained larvae proceeded faster Decomposition of Wild Shortnose Sturgeon Larvae than that of stained larvae in the controlled environment. There- in 2008–2010 fore, wild, unstained larvae that die during sampling in D-frame The 2008–2010 data set of captured Shortnose Sturgeon nets are expected to exhibit faster decay than was observed for larvae showed the yearly differences in catches of live or dead both stained and unstained larvae in the controlled environment. larvae and in their decomposition scores (Table 2). In 2008, 76% of the captured larvae were alive, while in 2009 and 2010 TABLE 3. Distribution (%) of overall decomposition scores for dead Short- the percentage dropped to 27–29%. The percentage of larvae nose Sturgeon larvae during each sampling year and in each transect based on with a score of 1 was much lower in 2008, reflecting the lower net deployments of 16 h or less.

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 rate of sampling mortality. The percentage of larvae with a score of 2 ranged between 11% in 2008 and 38% in 2010; the Decomposition percentage with a score of 3 ﬂuctuated between 4% in 2008 Partly Total number and 25% in 2009 (Table 2). In both 2009 and 2010, the nets Pristine decayed Decomposed of dead in the downstream transect contained more partly decomposed Transect (%) (%) (%) larvae and highly decomposed larvae than the nets in the upstream transect (Table 3). On the other hand, in 2008, the upstream nets 2008 contained more partly decomposed larvae than the downstream Upstream 44 39 17 196 nets, but upstream and downstream nets had similar numbers Downstream 52 30 19 27 of highly decomposed larvae (Table 3). 2009 To identify the source of mortality in wild Shortnose Upstream 43 24 33 152 Sturgeon larvae based on decomposition scores, we ﬁrst have Downstream 23 36 41 22 to consider the overall decomposition scores of stained versus 2010 unstained larvae in the controlled environment as well as com- Upstream 59 23 18 761 pare the decomposition of larvae in the controlled environment Downstream 46 31 23 449 with the decomposition of net-deployed larvae. The overall MORTALITY IN LARVAL SHORTNOSE STURGEON 603

However, it is likely to take considerably longer than 12 h of net samples prior to processing was the reason for the difference deployment for high numbers of larvae to reach an overall highly in mortality levels observed between 2008 and 2009–2010. In decomposed state. Therefore, pristine larvae in net collections 2008, most of the captured larvae were live (76%), whereas in most likely represent sampling mortality, whereas highly both 2009 and 2010, live larvae constituted less than 30% of the decomposed larvae are probably the result of environmental total catch. This difference was also reflected in the percentages mortality. A conservative approach would be to assign partly of larval decomposition states, as higher percentages of pristine decomposed larvae to sampling mortality, especially when and partly decomposed larvae were associated with higher analyzing data from nets that are deployed for medium to long gear-related mortality in 2009 and 2010. periods (overnight or longer). Applying this approach to our data for 2008–2010, environment- and sampling-related mortality Deployment of Live Shortnose Sturgeon Larvae in Nets accounted for 4–25% and 20–56%, respectively, of the sampled Only 50–60% of the dead and live Shortnose Sturgeon larvae larval population depending on the sampling year (Table 2). deployed in D-frame nets were recovered after 12-h deployments. This finding raises questions about the interpretation of DISCUSSION catch per unit effort and the accuracy of estimating larval abun- The analysis of mortality rates combined with time-of-death dance based on catches. A possible reason for the low retention estimates based on decomposition allowed us to extract im- of larvae in this study could be that the long period of deploy- portant information from dead Shortnose Sturgeon larvae in ment led to flow reduction within the net due to net clogging, collections. We used the examination of decomposition stages thereby resulting in larval escapement. Further research is re- as a fast, intuitive, and cost-effective approach to determining quired to estimate the extent of larval escapement at various the cause of mortality. Moreover, the ability to determine the lengths of net deployment. source of larval mortality based on an overall decomposition The high mortality rates of Shortnose Sturgeon larvae that score means that larval images taken during sampling in pre- were deployed live in the nets may be conservative. In our vious years can now be analyzed to assess historical rates of preliminary experiment of calcein staining, survival of calcein- sampling mortality and environmental mortality. treated larvae (80–100%) was higher than that of control fish (65–70%), albeit not significantly so (mixed-effects ANOVA: Effect of Net Deployment Period on Mortality P > 0.05; S. Usvyatsov, unpublished data). In addition, Crook and Decomposition et al. (2009) reported that Golden Perch Macquaria ambigua The lower mortality levels recorded at 6-h net deployments fingerlings receiving a treatment of salt and calcein (5% and in comparison with 12- and 24-h deployments suggest that if 1% solutions, respectively) exhibited significantly better growth the sampled Shortnose Sturgeon larvae are to be released alive, rates than control fish, although this result may be due to the short periods of deployment should be used. However, it must salt treatment rather than to calcein treatment. We have not be noted that the survival of larvae after sampling is the product found an explanation for this effect in the literature; however, of sampling mortality and delayed mortality, and both factors these findings indicate that mortality may be even higher than must be assessed to fully describe the impact of sampling on reported here. larvae. In addition, if the live release of larvae is not part of a study’s methodology, then sampling-related mortality becomes Decomposition irrelevant and the length of net deployment is then chosen solely The decomposition-based approach developed in this study based on desired effort. distinguished between Shortnose Sturgeon larvae that had been

Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 The difference in mortality rates between the 12-h net deploy- dead for 9 h and larvae that had been dead for 21 h or longer. ments in 2009 and 2010 may have been due to differences in However, it was difficult to differentiate between larvae that discharge from Mactaquac Dam. During the 2010 deployments, were dead for longer periods of time (i.e., the 21- and 32-h the daily discharge was 23 × 106 m3/d on 3 June and 51 × cases). Hence, this approach would probably be most useful 106 m3/d on 4 June. On the other hand, during the 2009 deploy- for discriminating between short and long postmortem periods. ments, discharge was 52 ± 17 × 106 m3/d (mean ± SD during This means that taphonomic analysis may be informative for 4–23 June) and the maximum discharge was 88 × 106 m3/d. short to medium periods of deployment (e.g., up to overnight), High river discharge may influence both the survival and but identification of mortality sources based on decomposition the decomposition of Shortnose Sturgeon larvae in the nets. In levels may not be reliable when net deployments are long (up addition to directly damaging the larvae inside the net, discharge to 24 h). is also correlated with higher debris loads. Debris may suffocate As expected, different tissues had different rates of decom- the larvae within the cod end during deployment and increase position. For example, barbels, fin structures, firmness, and the opportunity for physical strikes or crushing. In addition, high coloration decomposed fastest in both stained and unstained debris loads may increase mortality if samples are transferred Shortnose Sturgeon larvae. Therefore, these characters may into pails prior to examination instead of being sorted as soon be most important for determining the time of death. On the as the nets are lifted. We believe that using the pails for storing other hand, the eyes, belly, myomeres, and notochord integrity 604 USVYATSOV ET AL.

decomposed more slowly, retaining pristine values in some aquaculture (e.g., Chinese Sturgeon: Wei et al. 2009), which larvae even after 32 h. Both stained and unstained groups had suggests that delayed mortality is low; however, delayed mor- characters that decreased in decomposition score with time, tality rates must be quantified to better understand the impact of which indicates that the scores were somewhat subjective. sampling on larvae. Furthermore, in this study, environmental Hence, examination of the overall trend should be preferred characteristics were treated as a suite of variables rather than be- over individual characters. ing individually considered. To develop guidelines for optimal The faster decomposition of net-deployed Shortnose Stur- use of the sampling gear, we must understand how larval mor- geon larvae relative to larvae in the controlled environment was tality and decomposition are affected by individual variables, possibly due to abrasion against debris, the net mesh, or the such as water temperature, flow rate, and debris load within cod end. The faster decomposition of fins in live-deployed lar- the net. vae relative to dead-deployed larvae could have been caused by The findings presented here are important for reducing the swimming of the larvae in the cod end prior to death. This result impact of sampling strategies and for better understanding the suggests that the decomposition rates of fin structures may not effects of sampling design and environmental stress on sur- be as predictable as the decomposition rates of other characters; vival in larval Shortnose Sturgeon. Our conservative estimate of hence, estimating the time of death based on fin decay alone is environment-related mortality was 4–25% of the overall num- not recommended. ber of collected larvae. Shorter net deployments and immediate The rates of Shortnose Sturgeon larval decomposition in the sorting of larvae may reduce sampling-related mortality to 10– controlled environment and in the D-frame nets suggest that 20% or even less. However, it will be important to determine pristine larvae can be encountered in the nets only for a short the causes of environment-related mortality and to establish time after their death (substantially less time than the common whether these risks can be reduced. Like many other sturgeon overnight deployments). On the other hand, a high level of decay species, Shortnose Sturgeon spawn immediately below dams; would suggest that death occurred at least 24 h before the nets therefore, it is necessary to quantify the impact of fluctuating were checked. Therefore, in short to medium deployments (up river discharge on the survival and growth of early life stages. to overnight), the deaths of pristine larvae can be attributed to Previous work on other species, such as salmonids (Bell et al. sampling, whereas the deaths of highly decayed larvae could be 2008; Korman and Campana 2009; Warren et al. 2009) and related to environmental sources. catostomids (Weyers et al. 2003; Peterson and Jennings 2007), We propose a conservative approach wherein all Shortnose has clearly shown an adverse influence of discharge regulation Sturgeon larvae that are classified as pristine or partly de- on the abundance, growth, and survival of age-0 fish. Since the composed should be considered sampling-related mortalities, survival of early life stages is the most likely bottleneck to popu- whereas only highly decayed larvae should be assigned to lation persistence across sturgeon species (Gross et al. 2002), an environment-related mortality. Naturally, if flows are high, es- understanding of the sources of larval mortality is exceedingly pecially when coupled with a high debris load, the captured important for the successful management and conservation of larvae would be expected to decompose faster due to abrasion these protected species. and crushing within the net, resulting in higher-than-normal decomposition scores. The interpretation of partly decayed larvae would depend on the period of net deployment. In nets that ACKNOWLEDGMENTS are deployed overnight, partly decayed larvae are likely to be We thank Brent M. Wilson, Laura Qi, Faith M. Penny, gear-related mortalities, while in short deployments (e.g., 1– Andrew Hazen Brown, Jennifer R. Adams, Andrew Taylor, Christine Adams, Lillian P. Fanjoy, and Joel R. Chase for their Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 2 h) such larvae would probably be classified as environmental mortalities. Decomposition-based rates of environment-related help with fieldwork collections of larvae and image analysis in mortality exclude mortality due to predation. In addition, larvae the laboratory. We are grateful to the Hartt Island RV Resort for that die due to desiccation, starvation, or any other cause may providing accommodations during the sampling season and to also be consumed by predators after their death. Consequently, Fisheries and Oceans Canada for use of laboratory space at the the rates of environment-related mortality developed here should Mactaquac Biodiversity Facility. The editors and anonymous be interpreted as additional to the known or suspected rates of reviewers provided excellent comments and suggestions predation on larvae. that improved the manuscript. This study was supported by Mathematics of Information Technology and Complex Systems Future Research and Concluding Remarks grants to J.W. and M.K.L. and by New Brunswick Wildlife This study represents a first step in distinguishing between Trust Fund and Natural Sciences and Engineering Strategic and sampling- and environment-related mortality in Shortnose Stur- Discovery Research grants to M.K.L. geon larvae. Further research is required to address the impact of net deployments shorter than 6 h. In addition, the potential for REFERENCES delayed mortality in live-released larvae must be determined. Abramoff,` M. D., P. J. Magalhaes,˜ and S. J. Ram. 2004. Image processing with Wild-captured sturgeon larvae have been used for conservation ImageJ. Biophotonics International 11(7):36–42. MORTALITY IN LARVAL SHORTNOSE STURGEON 605

Auer, N. A. 1996. Importance of habitat and migration to sturgeons with em- Killgore, K. J., S. T. Maynord, M. D. Chan, and R. P.Morgan II. 2001. Evaluation phasis on Lake Sturgeon. Canadian Journal of Fisheries and Aquatic Sciences of propeller-induced mortality on early life stages of selected fish species. 53(Supplement 1):152–160. North American Journal of Fisheries Management 21:947–955. Auer, N. A., and E. A. Baker. 2002. Duration and drift of larval Lake Stur- Killgore, K. J., A. C. Miller, and K. C. Conley. 1987. Effects of turbulence on geon in the Sturgeon River, Michigan. Journal of Applied Ichthyology 18: yolk-sac larvae of Paddlefish. Transactions of the American Fisheries Society 557–564. 116:670–673. Beamesderfer, R. C. P., and R. A. Farr. 1997. Alternatives for the protection and Korman, J., and S. E. Campana. 2009. Effects of hydropeaking on nearshore restoration of sturgeons and their habitat. Environmental Biology of Fishes habitat use and growth of age-0 Rainbow Trout in a large regulated river. 48:407–417. Transactions of the American Fisheries Society 138:76–87. Bell, E., S. Kramer, D. Zajanc, and J. Aspittle. 2008. Salmonid fry stranding mor- Kynard, B. 1997. Life history, latitudinal patterns, and status of the Shortnose tality associated with daily water level fluctuations in Trail Bridge Reservoir, Sturgeon, Acipenser brevirostrum. Environmental Biology of Fishes 48:319– Oregon. North American Journal of Fisheries Management 28:1515–1528. 334. Benke, A. C., and C. E. Cushing, editors. 2005. Rivers of North America. Kynard, B., and M. Horgan. 2002. Ontogenetic behavior and migration of Elsevier Academic Press, Burlington, Massachusetts. Atlantic Sturgeon, Acipenser oxyrinchus oxyrinchus, and Shortnose Stur- Benson, A. C., T. M. Sutton, R. F. Elliott, and T. G. Meronek. 2006. Biological geon, A. brevirostrum, with notes on social behavior. Environmental Biology attributes of age-0 Lake Sturgeon in the lower Peshtigo River, Wisconsin. of Fishes 63:137–150. Journal of Applied Ichthyology 22:103–108. Legendre, P., and L. Legendre. 1998. Numerical ecology, 2nd English edition. Birstein, V. J., W. E. Bemis, and J. R. Waldman. 1997. The threatened status Elsevier, Amsterdam. of acipenseriform species: a summary. Environmental Biology of Fishes Manly, B. F. J. 2007. Randomization, bootstrap and Monte Carlo methods in 48:427–435. biology, 3rd edition. Chapman and Hall/CRC, London. Bruch, R. M., and F. P. Binkowski. 2002. Spawning behavior of Lake Sturgeon Mohler, J. W. 1997. Immersion of larval Atlantic Salmon in calcein solutions to (Acipenser fulvescens). Journal of Applied Ichthyology 18:570–579. induce a non-lethally detectable mark. North American Journal of Fisheries Caroffino, D. C., T. M. Sutton, and D. J. Daugherty. 2009. Assessment Management 17:751–756. of the vertical distribution of larval Lake Sturgeon drift in the Peshtigo Moser, M. L., M. Bain, M. R. Collins, N. Haley, B. Kynard, J. C. O’Herron River, Wisconsin, USA. Journal of Applied Ichthyology 25(Supplement 2): II, G. Rogers, and T. S. Squiers. 2000. A protocol for use of Shortnose and 14–17. Atlantic sturgeons. NOAA Technical Memorandum NMFS-OPR-18. Clarke, K. D., T. C. Pratt, R. G. Randall, D. A. Scruton, and K. E. Smokorowski. NMFS (National Marine Fisheries Service). 1998. Recovery plan for the Short- 2008. Validation of the flow management pathway: effects of altered flow nose Sturgeon (Acipenser brevirostrum). NMFS, Shortnose Sturgeon Recov- on fish habitat and fishes downstream from a hydropower dam. Canadian ery Team, Silver Spring, Maryland. Technical Report of Fisheries and Aquatic Sciences 2784. Oksanen, J., R. Kindt, P. Legendre, B. O’Hara, G. L. Simpson, P. Solymos, M. Conte, F. S., S. I. Doroshov, P. B. Lutes, and E. M. Strange. 1988. Hatchery H. H. Stevens, and H. Wagner. 2009. vegan: community ecology package, manual for the White Sturgeon Acipenser transmontanus Richardson with R package version 1.15-3. R Foundation for Statistical Computing, Vienna. application to other North American acipenseridae. Cooperative Extension, Available: vegan.r-forge.r-project.org/. (August 2011). University of California, Division of Agriculture and Natural Resources, Peterson, R. C., and C. A. Jennings. 2007. Effects of river discharge on abun- Publication 3322, Oakland. dance and instantaneous growth of age-0 Carpsuckers in the Oconee River, COSEWIC (Committee on the Status of Endangered Wildlife in Canada). 2005. Georgia, USA. River Research and Applications 23:1016–1025. COSEWIC assessment and update status report on the Shortnose Sturgeon R Development Core Team. 2010. R: a language and environment for statistical Acipenser brevirostrum in Canada. COSEWIC, Ottawa. computing, version 2.8. R Foundation for Statistical Computing, Vienna. Counihan, T. D., A. I. Miller, M. G. Mesa, and M. J. Parsley. 1998. The effects Sansom, R. S., S. E. Gabbott, and M. A. Purnell. 2010. Non-random decay of of dissolved gas supersaturation on White Sturgeon larvae. Transactions of chordate characters causes bias in fossil interpretation. Nature 463:797–800. the American Fisheries Society 127:316–322. Sykes, G. 2009. Nechako White Sturgeon monitoring. Report to British Crook, D. A., D. J. O’Mahony, A. C. Sanger, A. R. Munro, B. M. Gillan- Columbia Ministry of Environment, Project 4170, Triton Environmental Con- ders, and S. Thurstan. 2009. Development and evaluation of methods for sultants, Kamloops, British Columbia. osmotic induction marking of Golden Perch Macquaria ambigua with cal- Usvyatsov, S., J. Picka, R. S. Hardy, T. D. Shepherd, J. Watmough, and M. K. cein and alizarin red S. North American Journal of Fisheries Management 29: Litvak. 2012. Modeling the timing of spawning and hatching of Shortnose 279–287. Sturgeon, Acipenser brevirostrum, in the Saint John River, New Brunswick, Downloaded by [Department Of Fisheries] at 20:10 28 May 2013 Dadswell, M. J., B. D. Taubert, T. S. Squiers, D. Marchette, and J. Buckley. 1984. Canada. Canadian Journal of Fisheries and Aquatic Sciences 69:1316–1328. Synopsis of biological data on Shortnose Sturgeon, Acipenser brevirostrum Venables, W. N., and B. D. Ripley. 2002. Modern applied statistics with S, 4th LeSueur 1818. NOAA Technical Report NMFS 14. edition. Springer, New York. Gross, M. R., J. Repka, C. T. Robertson, D. H. Secor, and W. Van Winkle. Warren, D. R., A. G. Ernst, and B. P. Baldigo. 2009. Influence of spring floods 2002. Sturgeon conservation: insights from elasticity analysis. Pages 13–30 in on year-class strength of fall- and spring-spawning salmonids in Catskill W. Van Winkle, P. J. Anders, D. H. Secor, and D. A. Dixon, editors. Biology, Mountain streams. Transactions of the American Fisheries Society 138:200– management, and protection of North American sturgeon. American Fisheries 210. Society, Symposium 28, Bethesda, Maryland. Wei, Q. W., B. Kynard, D. G. Yang, X. H. Chen, H. Du, L. Shen, and H. Zhang. Jessop, B. M., and C. J. Harvie. 2003. A CUSUM analysis of discharge patterns 2009. Using drift nets to capture early life stages and monitor spawning of by a hydroelectric dam and discussion of potential effects on the upstream the Yangtze River Chinese Sturgeon (Acipenser sinensis). Journal of Applied migration of American Eel elvers. Canadian Technical Report of Fisheries Ichthyology 25(Supplement 2):100–106. and Aquatic Sciences 2454. Weyers, R. S., C. A. Jennings, and M. C. Freeman. 2003. Effects of pulsed, Kahn, J., and M. Mohead. 2010. A protocol for use of Shortnose, Atlantic, Gulf, high-velocity water flow on larval Robust Redhorse and V-Lip Redhorse. and Green sturgeons. NOAA Technical Memorandum NMFS-OPR-45. Transactions of the American Fisheries Society 132:84–91. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:11 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Catch-and-Release Rates of Sport Fishes in Northern Wisconsin from an Angler Diary Survey Jereme W. Gaeta a , Ben Beardmore a , Alexander W. Latzka a , Bill Provencher b & Stephen R. Carpenter a a Center for Limnology , University of Wisconsin–Madison , 680 North Park Street, Madison , Wisconsin , 53706 , USA b Agriculture and Applied Economics , University of Wisconsin–Madison , 427 Lorch Street, Madison , Wisconsin , 53706 , USA Published online: 24 May 2013.

To cite this article: Jereme W. Gaeta , Ben Beardmore , Alexander W. Latzka , Bill Provencher & Stephen R. Carpenter (2013): Catch-and-Release Rates of Sport Fishes in Northern Wisconsin from an Angler Diary Survey, North American Journal of Fisheries Management, 33:3, 606-614 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785997

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:606–614, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785997

MANAGEMENT BRIEF

Catch-and-Release Rates of Sport Fishes in Northern Wisconsin from an Angler Diary Survey

Jereme W. Gaeta,* Ben Beardmore, and Alexander W. Latzka Center for Limnology, University of Wisconsin–Madison, 680 North Park Street, Madison, Wisconsin 53706, USA Bill Provencher Agriculture and Applied Economics, University of Wisconsin–Madison, 427 Lorch Street, Madison, Wisconsin 53706, USA Stephen R. Carpenter Center for Limnology, University of Wisconsin–Madison, 680 North Park Street, Madison, Wisconsin 53706, USA

ical component of many regional economies, such as those of Abstract northern Wisconsin (Ditton et al. 2002; Peterson et al. 2003; Recreational freshwater fisheries are key components of local USFWS and USCB 2008b). The biotic and abiotic factors affect- economies in many regions. The quality of these fisheries can be ing the status and quality—where quality is defined as a complex affected not only by harvest but also by catch-and-release practices. Documenting catch and release among sport fish taxa is, measure of human satisfaction associated with catch rates, fish therefore, important to fisheries researchers studying sport fishes size, and diversity (Cox and Walters 2002)—of inland recre- and managers regulating these fisheries. We used an angler di- ational fisheries are numerous and complex (Lewin et al. 2006; ary survey to assess taxon-specific effort, catch, harvest, release, Arlinghaus et al. 2007). Although angler behavior is sometimes and reason for release during the 2011 open-water season. Our overlooked in management plans (Johnson and Carpenter 1994; study included information on 5,007 fishing trips taken by 652 anglers. These anglers visited 279 lakes spanning 11,761.5 km2 of Beard et al. 2003), angling itself is a critical driver influenc- northern Wisconsin. Muskellunge Esox masquinongy, black bass ing sport fisheries (Lewin et al. 2006; Arlinghaus et al. 2007). (i.e., Smallmouth Bass Micropterus dolomieu and Largemouth Bass Angling can affect fish and fish populations indirectly through M. salmoides), Northern Pike E. lucius, Walleye Sander vitreus,and boating traffic and noise, nutrient input, habitat degradation, panfish were released at rates of 99, 97, 86, 67, and 67%, respec- habitat loss, and the introduction of exotic species, and can have tively, when targeted by anglers. This study is the first to document Downloaded by [Department Of Fisheries] at 20:11 28 May 2013 black bass catch-and-release rates in the region and corroborates direct effects through harvest or catch and release (Lewin et al. previous findings of Muskellunge and Walleye catch-and-release 2006). Harvest and catch-and-release practices are known to rates based on creel surveys. Voluntary catch and release was the drive many aspects of fish ecology and, as a result, have been a most common reason for release. Our findings suggest that regula- major focus of fisheries researchers and managers (Lewin et al. tions may be much more generous than the harvest rates practiced 2006; Arlinghaus et al. 2007). by anglers and that catch-and-release angling practices may be an important factor affecting these sport fish populations. Historically, harvest effects have been the primary focus of angling research and regulations and catch-and-release effects rarely have been considered, until recently (Arlinghaus et al. Inland fisheries provide important provisioning (food and 2007). Harvest can affect fish communities in a variety of ways economic activity) and cultural (recreational) ecosystem ser- ranging from extreme cases of collapsed recreational fisheries vices (Postel and Carpenter 1997; Holmlund and Hammer (Post et al. 2002) to, more commonly, altered population size 1999). With U.S. angling expenditures at US$26.3 billion an- structures and densities (Olson and Cunningham 1989; Lewin nually (USFWS and USCB 2008a), angling can serve as a crit- et al. 2006). Catch-and-release angling practices can have lethal

*Corresponding author: [email protected] Received July 18, 2012; accepted March 12, 2013 606 MANAGEMENT BRIEF 607

or sublethal effects on released individuals through a variety of Micropterus dolomieu, Largemouth Bass M. salmoides, and mechanisms (as reviewed by Arlinghaus et al. 2007). However, panfish, a general category for the smaller sport fish: Bluegill catch and release is commonly mandated by fisheries regulations Lepomis macrochirus, Yellow Perch Perca flavescens, and Black (e.g., bag limits, length-based limits, and seasons) even though Crappie Pomoxis nigromaculatus. the population level effects of lethal and sublethal catch and release can be similar to harvest (Muoneke and Childress 1994; Parkinson et al. 2004; Cooke and Schramm 2007). METHODS Despite recent research highlighting the importance of catch Northern Wisconsin, the study area, is one of the densest lake and release to understanding sport fish population dynamics regions in the world (Magnuson et al. 2006). The region contains (e.g., Kerns et al. 2012), taxon-specific effort, catch, harvest, thousands of lakes that range in size from small lakes of less than release, and reason for release are not fully documented in the 1 ha to large lakes spanning several thousand hectares (Peterson literature for many regions (Deroba et al. 2007; Myers et al. et al. 2003) and cover 13% of the landscape (Buffam et al. 2011), 2008). Therefore, identifying taxon-specific angling patterns which is vegetated by upland conifer–hardwood forests (Stearns across regions where economies depend on recreational fisheries 1951; Brown and Curtis 1952). The ecosystem services provided may be useful to researchers and lake managers alike (Fayram by this lake-rich, forested landscape have attracted a 4.5-fold 2003; Deroba et al. 2007). The purpose of this investigation increase in population density over the last century (Peterson was to document taxon-specific effort, catch, harvest, release, et al. 2003; Carpenter et al. 2007), and fishing has become the and the reason for release of popular sport fishes in northern dominant tourist attraction and a pillar of the regional economy Wisconsin and to compare these findings with previous creel (Peterson et al. 2003). studies. Specifically, we used an angler diary survey to docu- Data were acquired from the 2011 Northern Highlands Lake ment effort and catch of Muskellunge Esox masquinongy,North- District Boater Survey (Figure 1), which employed a diary sur- ern Pike E. lucius, Walleye Sander vitreus, Smallmouth Bass vey to document individual fishing trips. Anglers were recruited Downloaded by [Department Of Fisheries] at 20:11 28 May 2013

FIGURE 1. The 2011 Northern Highlands Lake District Boater Survey diary form. 608 GAETA ET AL.

at boat landings of 136 lakes in Vilas and Oneida counties in recruitment effort with no contact). At these lakes, a census of 2011 between Memorial Day weekend (May 29–30) and Labor vehicles in the parking lot was taken and a recruitment brochure Day (September 5). Time constraints limited us to 136 lakes. was left on each windshield. Survey packages were then mailed Because our survey was a part of a larger survey focusing on to any boaters that returned the attached request. Participants invasive species, we attempted to recruit anglers at nine pop- were asked to complete a short initial survey that summarized ular lakes in the region with known populations of Eurasian past trips and then to keep track of their boating activities for water-milfoil Myriophyllum spicatum. To ensure a representa- the remainder of the season in a series of monthly log books, tive sample, the other 127 lakes were chosen to span the regional and a follow-up survey was mailed out after ice formed on the range of size and conductivity among publicly accessible lakes. lakes. The survey protocol followed the tailored-design method Of accessible lakes, our random sample was stratified first us- (Dillman et al. 2009). Participants received a $5 incentive at ing the third quartile threshold of conductivity (86 µS/cm) and the onset of the study, and regular contact through postcards second using the third quartile threshold of lake area (254 ha and replacement surveys was maintained throughout the study. for high conductivity lakes and 171.3 ha for low conductivity Each booklet that was returned was also entered into a draw that lakes). Multiple visits to each lake occurred at different times of had a 1 in 25 chance of receiving a check for $50. Study lakes day and were evenly split between weekdays and weekends to were limited to Forest, Iron, Oneida, Price, and Vilas counties account for temporal variation in angling activities. All boaters spanning nearly 14,000 km2 of northern Wisconsin (Figure 2). encountered at the boat landing were approached and invited to Questions regarding angling activity were included in the participate in the study. Many of the lakes in the sample were too boater diary, which included relevant information such as boat- remote and received too little boater traffic for in-person recruiting hours per trip (i.e., effort; Figure 1). If respondents partici- ment to be efficient (i.e., more than one person-hour of survey pated in recreational or guided angling they were asked to report Downloaded by [Department Of Fisheries] at 20:11 28 May 2013

FIGURE 2. Study lakes in northern Wisconsin. The map shows county lines, a locator map of the state of Wisconsin with study counties in gray, and a locator map of the Laurentian Great Lakes region with Wisconsin in gray. Study lakes are shown as light gray and overlaid with black points. MANAGEMENT BRIEF 609

(1) which taxa they targeted, (2) how many fish from each taxon back the initial survey. Of these, 860 boaters (65%) indicated were caught (whether or not they were targeted), (3) how many that they used their boat for recreational fishing. A total of 822 were released, and (4) whether release was “required by law” boaters completed the trip diary portion of the survey, recording (i.e., required by the lake season, length, or bag regulations), 6,694 trips. These boaters participated in recreational angling on “voluntary release,” or both “required by law” and “voluntary 75% of boating trips: pleasure cruising on 25%, swimming on release.” From these data we were able to determine several as- 12%, water skiing on 8%, with guided angling and work-related pects of taxon-specific angling activities in the region including uses comprising less than 2% of all boating trips. Of the 822 taxon-specific effort (as indicated by boat-hours per trip), CPUE diary respondents, 652 were anglers (i.e., they participated in (the number of fish caught per boat-hour), harvest rate (the ratio recreational or guided angling on at least one trip) and accounted of fish harvested to fish caught), the catch-and-release ratio (the for a total of 5,007 fishing trips within the study counties. These complement of harvest rate; that is, the ratio of fish released to anglers visited 279 lakes spread across 11,761.5 km2 of northern fish caught: Fayram 2003), and reason for release. Wisconsin (Figure 2). Several nonparametric statistical approaches were used to Angler effort in northern Wisconsin was approximately twice analyze the data at the P ≤ 0.05 significance level; all statisti- as great for Muskellunge, Walleye, and panfish than for North- cal tests were performed in R cran statistical package (version ern pike, Smallmouth Bass, and Largemouth Bass (Figure 3). 2.15.1; R Development Core Team 2012). Single-proportion Although angler effort for Muskellunge was high, the overall 95% CIs were estimated using the Wilson score method in- Muskellunge CPUE was the lowest among targeted taxa with corporating continuity correction (Newcombe 1998). We tested one Muskellunge caught for ∼14 boat-hours of effort with a whether angler-specific proportion of fish released differed sig- trip median of zero (Figure 4). Overall CPUEs for Walleye nificantly among taxa using a nonparametric multiple com- and Northern Pike were approximately two boat-hours per fish. parison test following a Kruskal–Wallis test (R cran package One Smallmouth Bass or Largemouth Bass and over four pan- ‘pgirmess’, version 1.5.4: Siegel and Castellan 1988). A two- fish were caught per boat-hour. Muskellunge, Smallmouth Bass, dimensional Kolmogorov–Smirnov test was used to determine and Largemouth Bass had the highest angler-specific catch-and- whether a relationship existed between angler-specific propor- release rate (Figure 5A); nearly every individual captured in tion of fish released and avidity (defined here as the proportion these taxa were released (99, 97, and 97%, respectively; Fig- of reported trips that a boater participated in angling) for all ure 5B). Anglers, however, harvested approximately 14% of taxa (Garvey et al. 1998). Nonparametric multiple comparison tests performed after a Kruskal–Wallis test were also used to test whether angler-specific proportion of fish released differed significantly among black bass taxa (i.e., Smallmouth Bass and Largemouth Bass) in the mandatory catch-and-release season (May 7–June 17) relative to the regular season (after June 18) as well as whether angler-specific catch-and-release rates differed between anglers recruited via the intercept and anglers recruited via windshield brochures. Walleyes were targeted at lakes with several management regimes. However, effort (boat-hours), the number of lakes, and the number of Walleyes caught were less than 5% for all but two management regimes: the general Wis-

Downloaded by [Department Of Fisheries] at 20:11 28 May 2013 consin inland waters 15-in length minimum with a bag limit of five (henceforth referred to as the “general” regulation) and a management regime of no minimum length limit and a total bag limit of five with only one over 14 in (henceforth referred to as the “no minimum” regulation). We tested whether the angler-specific proportion of Walleyes released differed between these two management regimes using a Mann–Whitney U-test (Hollander and Wolfe 1999).

RESULTS In all, 1,528 boaters intercepted at boat launches were invited to participate and 1,497 (96%) gave their consent and received FIGURE 3. Percent of angler effort (boat-hours) spent targeting various sport a trip log. Of 4,888 windshield ﬂiers, 233 (5%) postcards re- ﬁsh taxa in northern Wisconsin determined from angler diary surveys in the questing a survey package were returned. Of the 1,730 total summer of 2011. The values do not sum to 100% as anglers could target more individuals who received a survey package, 1,325 (77%) mailed than one taxon per trip. Bars are shown with 95% CIs. 610 GAETA ET AL.

captured Northern Pike and 33% of both targeted Walleyes and panfish. Walleyes and panfish were released at a significantly lower rate than any other taxa. No significant relationship was observed between angler avidity and angler-specific release rate. Angler release rates of nontargeted fish were only different than release rates for targeted fish at the 95% confidence level for panfish, which were released more often when targeted. No significant difference was observed among angler-specific proportion of black bass released in the early catch-and-release-only season and the regular open-water season (Figure 6). No significant difference in angler-specific catch-and-release rates was observed between anglers recruited via the intercept method (n = 415, once item nonresponse was accounted for) and anglers recruited via windshield brochures (n = 136, once item nonresponse was accounted for) for any taxon. Angler-specific proportions of Walleyes released did not significantly differ between the general and no-minimum regulations (Figure 7); release rates did, however, tend to be lower under the “no minimum” regulation (P = 0.06). Voluntary release was observed in ∼50% or more of individuals in all taxa except Walleye, for which “required by law” was the dominant reason for release (Figure 8).

FIGURE 4. Catch per unit effort (number of fish caught per boat-hour) for sport fish taxa when targeted by anglers in northern Wisconsin during the DISCUSSION summer of 2011. Overall CPUE for all boat-hours in all lakes are shown as gray The angler diary surveys in northern Wisconsin during the points. Boxplots represent trip-specific CPUE for which outliers are removed; shown with medians, first and third quantiles, and a range of 1.5 times the summer of 2011 indicated that fisheries in northern Wiscon- interquartile range. Outliers are represented in overall CPUE estimates (gray sin are largely catch-and-release fisheries for all taxa. Even for points). consumptive fisheries, i.e., Walleye and panfish, approximately Downloaded by [Department Of Fisheries] at 20:11 28 May 2013

FIGURE 5. Taxon-specific proportion of fish released in northern Wisconsin during the 2011 open-water season. (A) Proportion of targeted fish released as grouped by angler. Release proportions for taxa with matching letters were not significantly different (P > 0.05). Boxplots are shown with medians, first and third quantiles, range (1.5 times the interquartile range), and outliers beyond the range. Angler sample sizes are shown below boxplots. (B) The overall observed taxon-specific proportion of fish released when the taxon was targeted (gray) and nontargeted (white). Bars are shown with 95% CIs. MANAGEMENT BRIEF 611

FIGURE 6. Angler-specific proportion of Largemouth Bass and Smallmouth FIGURE 7. Angler-specific proportion of Walleyes released in northern Wis- Bass released in northern Wisconsin during the 2011 catch-and-release and consin during the 2011 catch-and-release and regular seasons. The “general regular seasons. Release proportions with matching letters were not significantly regulation” is the default Wisconsin inland waters Walleye regulation of a 15-in different (P > 0.05). Boxplots are shown with medians, first and third quantiles, length minimum with a bag limit of five. The “no minimum regulation” is a and range (1.5 times the interquartile range) all at 1.0. Outliers beyond the range. special Walleye regulation with a no-minimum length limit and a total bag limit Angler sample sizes are shown below boxplots. of five with only one over 14 in. Release proportions with matching letters were not significantly different (P > 0.05). Boxplots are shown with medians, first 66% of captured fish were released. However, “required by law” and third quantiles, range (1.5 times the interquartile range), and outliers beyond the range. Angler and lake sample sizes are shown below boxplots. was the dominant reported reason for release of Walleyes. Like- wise, we observed a trend of lower Walleye catch-and-release rates under the “no minimum” regulation relative to the “general” regulation. Catch-and-release rates have not previously been documented in the primary literature for Smallmouth Bass or Largemouth Bass in our study region. Our findings suggest that release rates of these taxa are not influenced by regulation. The angling survey also corroborates previous creel surveys of northern Wisconsin Muskellunge and Walleye fisheries (Fayram 2003; Kerr 2007) and provides further insight into previous work

Downloaded by [Department Of Fisheries] at 20:11 28 May 2013 on Northern Pike and panfish fisheries (Beard and Kampa 1999; Margenau et al. 2003). According to a survey of lake managers across the nation, voluntary catch-and-release of black bass has been increasing across the United States and has been suspected of reducing the effectiveness of harvest regulations in northern latitudes (Noble 2002). However, estimates of catch-and-release rates in black bass fisheries are rare in the literature (Myers et al. 2008). We are the first to quantify catch-and-release rates of black bass in northern Wisconsin and our results mirror the observed national trend of high release rates for black bass in recent years. For example, a creel survey study of black bass catch-and-release rates from the 1970s to 2000s in four Texas reservoirs and two Florida lakes found that catch-and-release rates increased over time in all six study systems (Myers et al. 2008). Catch-and- FIGURE 8. Taxon-specific angler reason for release of targeted fish in northern release rates by the 2000s ranged from 58% to 99%. Smallmouth Wisconsin during the 2011 open-water season. 612 GAETA ET AL.

Bass and Largemouth Bass caught by anglers in our survey were under the “no minimum” regulation relative to the “general” released at rates of 96.6% and 96.9%, respectively. 15-in length-limit regulation (P = 0.06). This trend may Harvest can be beneficial to Largemouth Bass condition indicate that minimum size limits play a role in constraining (Perry et al. 1995) and even improve black bass fishery qual- Walleye harvest in the region and warrants further exploration. ity (i.e., maintain large fish in the population) in areas where Release rates of both Northern Pike and panfish in north- growth rates and population productivity are low, such as northern Wisconsin have been previously documented; however, ern Wisconsin (Beamesderfer and North 1995). Therefore, some previous studies used methodologies that are not directly degree of harvest may be necessary to maintain quality black comparable with our study (Beard and Kampa 1999; Margenau bass fisheries. To explore whether regulations were inhibiting et al. 2003). Margenau et al. (2003) used a creel survey of anglers from harvesting black bass or whether high release rates 55 northern Wisconsin lakes from 1990 to 1999 to assess the were due to voluntary release, we tested for a reduction in angler- regional catch-and-release rate of Northern Pike. Although specific catch-and-release rates in the regular open-water season they used a more conservative metric that included anglers that relative to the mandatory catch-and-release season. No signifi- targeted only Northern Pike, they found that 88.1% of Northern cant difference in catch-and-release rates was observed between Pike were released, very similar to our observed release rate seasons for either Largemouth Bass or Smallmouth Bass. Such of 86.5%. Using creel surveys of various panfish species from high catch-and-release rates highlight the need for further re- 1980 to 1991, Beard and Kampa (1999) found catch-and-release search to explore whether length-based and bag-limit harvest rates (based on fish caught and harvested per hour) in northern regulations may be ineffective for managing the quality of black Wisconsin ranged from 17.8% to 76.6% depending on species. bass fisheries, which, for these fisheries, is the most prevalent Our observed catch-and-release rate of 67.1% for panfish fell management goal (Paukert et al. 2007). Likewise, fisheries re- within their observed catch-and-release range. However, our searchers and managers should continue investigating the mech- results are not directly comparable, as Beard and Kampa (1999) anisms through which black bass fisheries may be affected by included ice-fishing creels in their study. catch-and-release. Angling diary surveys are often used to assess recreational We found that angler diary surveys and traditional creel sur- fisheries (Hunt et al. 2007; Kerr 2007; Beardmore et al. 2011). veys yielded very similar estimates of harvest and catch-and- While the findings of these studies are relevant to fisheries re- release rates. Our findings serve as independent confirmation of searchers and managers, several inherent limitations must be previous regional creel surveys of Muskellunge, Walleye, North- acknowledged. Common limitations include low participation ern Pike, and panfish. For instance, both Muskellunge (Marge- rates and associated response biases (e.g., avidity), item non- nau and Petchenik 2004; Kerr 2007; Landsman et al. 2011) response, recording errors, and angler bias (Fisher 1996; Kerr and Walleye (Hewett and Simonson 1998) are highly sought- 2007). Other factors known to influence angler survey results after sport fishes. Both species are economically and culturally were unaccounted for in our analysis, and these included angler important to northern Wisconsin, albeit for different reasons. experience and specialization, whether an angler is transient Muskellunge provide predominantly a catch-and-release trophy or local, and whether an angler is familiar with the lake (Kerr fishery dominated by highly specialized anglers (Margenau and 2007). Petchenik 2004; Landsman et al. 2011). Our findings corrobo- We recognize that the brevity of the angling-specific portion rate the only other Muskellunge angler survey in the region, a of our survey also contributed to the limitations of our study. As creel survey that found catch-and-release rates of 99.9% in 2000 a result, our CPUE findings are not directly comparable with (Fayram 2003), matching our 2011 observed catch-and-release other studies such as creel surveys. Additional information such

Downloaded by [Department Of Fisheries] at 20:11 28 May 2013 rates of 99.3% when targeted and 100% when not targeted. as asking for the number of anglers on the boat and the hours Walleye is also a highly sought taxon partially for sport, angling for each selected target species would have greatly im- but largely because of their value as a “table” fish (Hewett and proved the study. Including a third release reason of “required Simonson 1998). As a consequence of their culinary value, Wall- by law, but would have released voluntarily if not required by eye catch-and-release rates are among the lowest of Wisconsin law” and asking the angler to report the number of fish per taxa sport fishes (Fayram 2003). An analysis of creel surveys from released per release reason would have increased our ability 215 lakes during 1990–2000 indicated that the catch-and-release to draw conclusions from these data. Gathering information on rate among lakes with 15-in minimum length limits was 66.7% taxon-specific angler motivation, satisfaction, and catch orienta- and that harvest rates had not significantly changed during the tion (Arlinghaus 2006) would also have been a valuable addition 1990s (Fayram 2003). When not accounting for lake regulations to our study. Even though we acknowledge the uncertainty in (although 60% of the lakes had a 15-in minimum), we found that some aspects of the survey, these uncertainties had no effect on the regional catch-and-release rate of Walleyes in 2011 was also our findings regarding catch and release. 66.7%. Our findings suggest that Walleye remains a popular In conclusion, we corroborated results from previous regional food fish and catch-and-release rates have been constant over creel surveys, even ones using slightly differing methodologies, the last three decades. However, our analysis of angler-specific suggesting that diary survey studies are as effective as creel sur- catch-and-release rates suggests that harvest tends to be higher veys for documenting regional angler behavior. We also found MANAGEMENT BRIEF 613

that anglers in northern Wisconsin predominantly practice catch Buffam, I., M. G. Turner, A. R. Desai, P. C. Hanson, J. A. Rusak, N. R. Lottig, and release even in consumptive fisheries, such as for Walleye E. H. Stanley, and S. R. Carpenter. 2011. Integrating aquatic and terrestrial and panfish. Reasons for release vary among taxa. In the case components to construct a complete carbon budget for a north temperate lake district. Global Change Biology 17:1193–1211. of Smallmouth Bass and Largemouth Bass, our findings sug- Carpenter,S.R.,B.J.Benson,R.Biggs,J.W.Chipman,J.A.Foley,S.A. gest that length-based and bag-limit regulations may be much Golding, R. B. Hammer, P. C. Hanson, P. T. J. Johnson, A. M. Kamarainen, more generous than the harvest rates actually practiced by most T. K. Kratz, R. C. Lathrop, K. D. McMahon, B. Provencher, J. A. Rusak, anglers. The effects of catch and release on black bass behav- C. T. Solomon, E. H. Stanley, M. G. Turner, M. J. Vander Zanden, C. H. Wu, ior and population dynamics may, therefore, be more important and H. L. Yuan. 2007. Understanding regional change: a comparison of two lake districts. BioScience 57:323–335. than the effects of harvest in determining responses of these fish- Cooke, S. J., and H. L. Schramm. 2007. Catch-and-release science and its ap- eries to angling and should be investigated. Indeed, our findings plication to conservation and management of recreational fisheries. Fisheries raise questions as to whether alternative management strategies, Management and Ecology 14:73–79. such as mandatory harvest of Largemouth Bass or outreach Cox, S., and C. Walters. 2002. Maintaining quality in recreational fisheries: to alter angler perception, may be necessary to reach certain how success breeds failure in management of open-access sport fisheries. Pages 107–119 in T. J. Pitcher and C. E. Hollingworth, editors. Recreational management goals. Future research by fisheries researchers and fisheries: ecological, economic and social evaluation. Blackwell Scientific managers should continue to investigate mechanisms through Publications, Oxford, UK. which catch and release may be affecting these north temperate Deroba, J. J., M. J. Hansen, N. A. Nate, and J. M. Hennessy. 2007. Tempo- fisheries. ral profiles of Walleye angling effort, harvest rate, and harvest in northern Wisconsin lakes. North American Journal of Fisheries Management 27:717– 727. ACKNOWLEDGMENTS Dillman, D. A., J. D. Smyth, and L. M. Christian. 2009. Internet, mail, and mixed-mode surveys: the tailored design method, 3rd edition. Wiley, Hobo- This material is based upon work supported by the Na- ken, New Jersey. tional Science Foundation under Grant Number (Award Number Ditton, R. B., S. M. Holland, and D. K. Anderson. 2002. Recreational fishing CNH-0909281) with additional support from the North Tem- as tourism. Fisheries 27(3):17–24. perate Lakes–Long Term Ecological Research program and a Fayram, A. H. 2003. A comparison of regulatory and voluntary release of Graduate Engineering Research Scholars fellowship to J. Gaeta. Muskellunge and Walleyes in northern Wisconsin. North American Journal of Fisheries Management 23:619–624. We thank K. Anderson for her work on earlier drafts of the boater Fisher, M. R. 1996. Estimating the effect of nonresponse bias on angler surveys. survey as well as G. Jackson and our crew of undergraduate re- Transactions of the American Fisheries Society 125:118–126. search assistants for the fieldwork. We also thank J. Hansen, J. Garvey, J. E., E. A. Marschall, and R. A. Wright. 1998. From star charts Kitchell, C. Kucharik, and J. Vander Zanden as well as R. Eades to stoneflies: detecting relationships in continuous bivariate data. Ecology and four anonymous reviewers for helpful comments on earlier 79:442–447. Hewett, S., and T. Simonson. 1998. Wisconsin’s Walleye management plan: versions of this manuscript. moving management into the 21st century. Wisconsin Department of Natural Resources, Administrative Report 43, Madison. Hollander, M., and D. A. Wolfe. 1999. Nonparametric statistical methods, 2nd REFERENCES edition. Wiley, New York. Arlinghaus, R. 2006. On the apparently striking disconnect between motivation Holmlund, C. M., and M. Hammer. 1999. Ecosystem services generated by fish and satisfaction in recreational fishing: the case of catch orientation of German populations. Ecological Economics 29:253–268. anglers. North American Journal of Fisheries Management 26:592–605. Hunt, L. M., B. N. Boots, and P. C. Boxall. 2007. Predicting fishing participation Arlinghaus, R., S. J. Cooke, J. Lyman, D. Policansky, A. Schwab, C. Suski, and site choice while accounting for spatial substitution, trip timing, and trip S. G. Sutton, and E. B. Thorstad. 2007. Understanding the complexity of context. North American Journal of Fisheries Management 27:832–847. catch-and-release in recreational fishing: an integrative synthesis of global Johnson, B. M., and S. R. Carpenter. 1994. Functional and numerical re- Downloaded by [Department Of Fisheries] at 20:11 28 May 2013 knowledge from historical, ethical, social, and biological perspectives. Re- sponses: a framework for fish-angler interactions? Ecological Applications 4: views in Fisheries Science 15:75–167. 808–821. Beamesderfer, R. C. P., and J. A. North. 1995. Growth, natural mortality, and Kerns, J. A., M. S. Allen, and J. E. Harris. 2012. Importance of assessing predicted response to fishing for Largemouth Bass and Smallmouth Bass pop- population-level impact of catch-and-release mortality. Fisheries 37:502– ulations in North America. North American Journal of Fisheries Management 503. 15:688–704. Kerr, S. J. 2007. Characteristics of Ontario Muskellunge (Esox masquinongy) Beard, T. D., Jr., S. P. Cox, and S. R. Carpenter. 2003. Impacts of daily bag fisheries based on volunteer angler diary information. Environmental Biology limit reductions on angler effort in Wisconsin Walleye lakes. North American of Fishes 79:61–69. Journal of Fisheries Management 23:1283–1293. Landsman, S. J., H. J. Wachelka, C. D. Suski, and S. J. Cooke. 2011. Evalua- Beard, T. D., Jr., and J. M. Kampa. 1999. Changes in Bluegill, Black Crappie, and tion of the physiology, behaviour, and survival of adult Muskellunge (Esox Yellow Perch populations in Wisconsin during 1967–1991. North American masquinongy) captured and released by specialized anglers. Fisheries Re- Journal of Fisheries Management 19:1037–1043. search 110:377–386. Beardmore, B., M. Dorow, W. Haider, and R. Arlinghaus. 2011. The elasticity Lewin, W. C., R. Arlinghaus, and T. Mehner. 2006. Documented and poten- of fishing effort response and harvest outcomes to altered regulatory policies tial biological impacts of recreational fishing: insights for management and in eel (Anguilla anguilla) recreational angling. Fisheries Research 110:136– conservation. Reviews in Fisheries Science 14:305–367. 148. Magnuson, J. J., T. K. Kratz, and B. J. Benson, editors. 2006. Long-term dy- Brown, R. T., and J. T. Curtis. 1952. The upland conifer-hardwood forests of namics of lakes in the landscape: long-term ecological research on north northern Wisconsin. Ecological Monographs 22:217–234. temperate lakes. Oxford University Press, New York. 614 GAETA ET AL.

Margenau, T. L., S. J. Gilbert, and G. R. Hatzenbeler. 2003. Angler catch Perry, W. B., W. A. Janowsky, and F. J. Margraf. 1995. A bioenergetics sim- and harvest of Northern Pike in northern Wisconsin lakes. North American ulation of the potential effects of angler harvest on growth of Largemouth Journal of Fisheries Management 23:307–312. Bass in a catch-and-release fishery. North American Journal of Fisheries Margenau, T. L., and J. B. Petchenik. 2004. Social aspects of Muskellunge Management 15:705–712. management in Wisconsin. North American Journal of Fisheries Management Peterson, G. D., T. D. Beard Jr., B. E. Beisner, E. M. Bennett, S. R. Carpenter, 24:82–93. G. S. Cumming, C. L. Dent, and T. D. Havlicek. 2003. Assessing future Muoneke, M. I., and W. M. Childress. 1994. Hooking mortality: a review for ecosystem services: a case study of the Northern Highlands Lake District, recreational fisheries. Reviews in Fisheries Science 2:123–156. Wisconsin. Conservation Ecology [online serial] 7(3): article 1. Myers, R., J. Taylor, M. Allen, and T. F. Bonvechio. 2008. Temporal trends in Post, J. R., M. Sullivan, S. Cox, N. P. Lester, C. J. Walters, E. A. Parkinson, A. voluntary release of Largemouth Bass. North American Journal of Fisheries J. Paul, L. Jackson, and B. J. Shuter. 2002. Canada’s recreational fisheries: Management 28:428–433. the invisible collapse? Fisheries 27(1):6–17. Newcombe, R. G. 1998. Two-sided confidence intervals for the single propor- Postel, S., and S. Carpenter. 1997. Freshwater ecosystem services. Pages 195– tion: comparison of seven methods. Statistics in Medicine 17:857–872. 214 in G. C. Daily, editor. Nature’s services: societal dependence on natural Noble, R. L. 2002. Reflections on 25 years of progress in Black Bass man- ecosystems. Island Press, Washington, D.C. agement. Pages 419–431 in D. P. Philipp and M. S. Ridgway, editors. Black R Development Core Team. 2012. R: a language and environment for statistical Bass: ecology, conservation, and management. American Fisheries Society, computing. R Foundation for Statistical Computing, Vienna. Symposium 31, Bethesda, Maryland. Siegel, S., and N. J. Castellan Jr. 1988. Nonparametric statistics for the behav- Olson, D. E., and P. K. Cunningham. 1989. Sport-fisheries trends shown by ioral sciences, 2nd edition. McGraw-Hill, New York. an annual Minnesota fishing contest over a 58-year period. North American Stearns, F. 1951. The composition of the sugar maple–hemlock–yellow birch Journal of Fisheries Management 9:287–297. association in northern Wisconsin. Ecology 32:245–265. Parkinson, E. A., J. R. Post, and S. P. Cox. 2004. Linking the dynamics of USFWS (U.S. Fish and Wildlife Service), and USCB (U.S. Census Bureau). harvest effort to recruitment dynamics in a multistock, spatially structured 2008a. 2006 national survey of fishing, hunting, and wildlife-associated recre- fishery. Canadian Journal of Fisheries and Aquatic Sciences 61:1658–1670. ation. USFWS/USCB, FHW/06-NAT, Washington, D.C. Paukert, C., M. McInerny, and R. Schultz. 2007. Historical trends in creel limits, USFWS (U.S. Fish and Wildlife Service), and USCB (U.S. Census Bureau). length-based limits, and season restrictions for Black Basses in the United 2008b. 2006 national survey of fishing, hunting, and wildlife-associated recre- States and Canada. Fisheries 32:62–72. ation: Wisconsin. USFWS/USCB, FHW/06-WI, Washington, D.C. Downloaded by [Department Of Fisheries] at 20:11 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:12 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Survival and Traits of Reconditioned Kelt Steelhead Oncorhynchus mykiss in the Yakima River, Washington Douglas R. Hatch a , David E. Fast b , William J. Bosch b , Joseph W. Blodgett b , John M. Whiteaker a , Ryan Branstetter a & Andrew L. Pierce a a Columbia River Inter-Tribal Fish Commission , 729 Northeast Oregon Street, Suite 200, Portland , Oregon , 97232 , USA b Yakama Nation Fisheries , Post Office Box 151, Toppenish , Washington , 98948 , USA Published online: 24 May 2013.

To cite this article: Douglas R. Hatch , David E. Fast , William J. Bosch , Joseph W. Blodgett , John M. Whiteaker , Ryan Branstetter & Andrew L. Pierce (2013): Survival and Traits of Reconditioned Kelt Steelhead Oncorhynchus mykiss in the Yakima River, Washington, North American Journal of Fisheries Management, 33:3, 615-625 To link to this article: http://dx.doi.org/10.1080/02755947.2013.788586

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:615–625, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.788586

ARTICLE

Survival and Traits of Reconditioned Kelt Steelhead Oncorhynchus mykiss in the Yakima River, Washington

Douglas R. Hatch* Columbia River Inter-Tribal Fish Commission, 729 Northeast Oregon Street, Suite 200, Portland, Oregon 97232, USA David E. Fast, William J. Bosch, and Joseph W. Blodgett Yakama Nation Fisheries, Post Ofﬁce Box 151, Toppenish, Washington 98948, USA John M. Whiteaker, Ryan Branstetter, and Andrew L. Pierce Columbia River Inter-Tribal Fish Commission, 729 Northeast Oregon Street, Suite 200, Portland, Oregon 97232, USA

Abstract We evaluated the traits and survival to release of reconditioned kelt steelhead Oncorhynchus mykiss in the Yakima River (Washington State, USA). From 2001 to 2011, we captured a total of 9,738 downstream-migrating kelts at an irrigation diversion facility, an average about 27% of each annual wild steelhead return. Captured kelts were reared for 4.5–10 months in an artificial environment, treated for diseases and parasites, and fed both krill and pellets. Surviving reconditioned fish were released into the Yakima River during the peak of the upstream migration of prespawn steelhead. Reconditioned steelhead kelts were predominantly (>92%) female. Annual survival to release ranged from 20% to 62% and averaged 38% over the course of the study, the surviving reconditioned kelts showing increases in FL, weight, and Fulton’s K condition factor compared with their preconditioning status. Kelts in good condition and those with bright coloration at the time of collection were more likely to survive than those of poorer status at collection. Postrelease timing of upstream migration by reconditioned kelts was spread over several months and correlated well with the run timing of prespawn migrants upstream. The empirical results we observed demonstrate the potential for kelt reconditioning to provide recovery benefits for repeat spawning imperiled wild populations in highly developed river systems. Downloaded by [Department Of Fisheries] at 20:12 28 May 2013

Populations of wild steelhead Oncorhynchus mykiss in the increasing abundance of imperiled species in the wild include Columbia River Basin (CRB) have declined dramatically from nature reserves (Margules and Pressey 2000), restoration ecol- historical levels (Nehlsen et al. 1991; NRC 1996; Williams ogy (Dobson et al. 1997), artificial propagation and release of et al. 1999) and are now listed as endangered under the Endan- progeny from wild parents (Cuenco et al. 1993), captive rearing gered Species Act. The average abundance of wild steelhead (Berejikian et al. 2001; Carmona-Catot et al. 2012), and use (anadromous summer run/in-river maturing; there are no win- of cryopreserved wild gametes to produce offspring artificially ter run/ocean-maturing steelhead) in the Yakima River subbasin (Cloud et al. 1990). Historically, there have been no artificial over the last two decades is only 2% of pre-1890 abundance breeding programs for Yakima River steelhead, and annual adult levels (Howell et al. 1985). Causes for these declines include returns to the Yakima River average 98% wild, the rest being a host of environmental and human-induced factors (Raymond some hatchery strays from other CRB tributaries (Yakama Na- 1988; NRC 1996; Williams et al. 1999). Common tools for tion unpublished data). Some measures to restore habitats have

*Corresponding author: [email protected] Received July 17, 2012; accepted March 15, 2013 615 616 HATCH ET AL.

recently been implemented to assist population rebuilding ef- 2007). Improving survival of downstream-migrating kelt could forts (YSFWPB 2004). Due to the inherently long-term nature be a valuable restoration strategy for increasing abundance and of habitat-related recovery efforts, more immediate strategies to productivity in CRB steelhead populations, especially in up- increase abundance are being evaluated that capitalize on the per watersheds such as the Yakima River. Transportation of iteroparous life history steelhead exhibit. downstream-migrating kelts around hydropower dams is one Iteroparity among CRB steelhead has been documented as potential method to improve kelt survival that is currently being far back as 1895 (Evermann 1895), with early accounts of evaluated in the CRB (Evans et al. 2008). repeat spawners entering natal tributaries (Long and Griffin Another method for increasing the iteroparity rate of CRB 1937; Whitt 1954) extending as far inland as the Snake River steelhead is by artificially reconditioning kelts. Reconditioning Basin (Evermann 1895). To some degree, successful iteropar- is the practice of capturing, holding, and feeding postspawned ity appears to have been a component of all steelhead inhab- salmon or steelhead in an artificial rearing environment for the iting the CRB. Despite mortality during secondary migrations, purpose of improving survival and regeneration of gonads for re- iteroparous fish are thought to contribute substantially to the ge- peat spawning. This concept has been applied to Atlantic Salmon netic and demographic structure of salmonid populations (Ward Salmo salar (Gray et al. 1987; Crim et al. 1992; Johnston et al. and Slaney 1988; Fleming and Reynolds 2004; Keefer et al. 1992), Brown Trout Salmo trutta L. (Poole et al. 1994), and 2008). Given the multiple migratory and reproductive pheno- Arctic Char Salvelinus alpinus L. (Boyer and Van Toever 1993). types that contribute to reproductive success and population Null et al. (2013) reconditioned postspawned hatchery-origin structure in steelhead, preserving as many of these reproduc- steelhead at Coleman National Fish Hatchery, California, for a tive phenotypes as possible is critical for this complex species short period, released these fish into the Sacramento River, and (Nielsen et al. 2011). used acoustic telemetry to study postspawn migration patterns. Iteroparity rates vary among populations and probably have The study we describe in this paper is the first to evaluate sur- decreased in recent years, especially in upper watersheds, due to vival and traits of wild steelhead reconditioned in an artificial anthropogenic factors such as salmonid harvest, habitat degra- environment for several months. dation, and dam construction. Whitt (1954) estimated that ap- Our objectives were to describe the abundance of adult steel- proximately 2% of adult Clearwater River steelhead were repeat head migrating upstream and the abundance of kelts migrating spawners. Unfortunately, Whitt’s estimates were conducted af- downstream in the Yakima River; describe the traits of kelt ter the construction of two hydropower facilities on main-stem steelhead in the Yakima River, including sex ratio, condition, river sites; the estimates also relied on scale analysis, which and timing; briefly describe the “reconditioning” process we de- may have resulted in an underestimation of kelt (postspawned, veloped and employed; report on the survival of reconditioned downstream migrant) abundance due to scale reabsorption kelt steelhead to release; and assess survival relative to several (Seamons et al. 2009). More recently, iteroparity rates for independent variables. Klickitat River steelhead were reported to be 3.3% from 1979 to 1981 (Howell et al. 1985). Repeat spawners composed 1.6% of the Yakima River wild run (from data in Hockersmith et al. METHODS 1995). Keefer et al. (2008) sampled steelhead kelts at main- Study area and monitoring facilities. The Yakima River stem hydropower dams from 2001 to 2004 and estimated that Basin is located in south-central Washington State. From its between 0.5% and 1.2% of those fish from the Snake River headwaters near the crest of the Cascade Range, the Yakima and tributaries and 2.9–9.0% of fish from the Lower Columbia River flows 344 km southeastward to its confluence with the

Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 exhibited an iteroparous life history. This is in contrast to the Columbia River (river kilometer [rkm] 540; Figure 1). Steel- Kalama River, an un-impounded tributary of the lower CRB, head populations primarily spawn upstream from Prosser Dam where iteroparity rates in excess of 17% have been documented (rkm 76) in Satus Creek, Toppenish Creek, the Naches River, for the winter run steelhead (Busby et al. 1996). and other tributaries (Conley et al. 2009). The Prosser Diversion While successful iteroparity rates are low for many reaches of Dam (rkm 76; Figure 1) has three fish ladders each equipped the CRB relative to steelhead throughout their range, kelts tend with passive integrated transponder (PIT) detectors and fish to be quite abundant. Emigrating steelhead kelts averaged 46.1% count windows. Video camera and recording systems were de- of annual prespawn upstream runs in the Clackamas River from ployed at each of the three fishway count windows where they 1960 to 1964 (Gunsolus and Eicher 1970). Evans et al. (2004) recorded migrating fish 24 h per day year round. The video was estimated that 17% of the Snake River steelhead population reviewed daily to produce real time counts of fish migrating was observed as kelts in the Lower Granite Dam juvenile by- past Prosser Dam. We used these counts to represent estimated pass facility during a 10-week monitoring period in the spring annual steelhead spawner abundance. of 2000. Though kelt abundance appears to be quite high, poor We evaluated river flow with respect to steelhead using emigration survival of steelhead from upper Columbia River wa- three different time periods. We queried the U.S. Bureau of tersheds to the ocean appears to be the underlying limiting factor Reclamation hydromet database to obtain average daily stream inhibiting iteroparity (Wertheimer and Evans 2005; Wertheimer flow data (m3/s) for the Yakima River near Prosser Dam (see YAKIMA RIVER STEELHEAD 617

FIGURE 1. The Yakima River Basin, steelhead kelt reconditioning facilities, and monitoring locations.

http://www.usbr.gov/pn/hydromet/yakima/). We then calculated transferred to a temporary holding tank containing oxygenated mean monthly flows and standardized each mean by divid- well water (13.8◦C). Each fish was anesthetized in a buffered ing by that month’s 10-year average. The first time period solution of tricaine methanesulfonate (MS-222) at 600 µL/L, was simply the standardized monthly flow during a calendar weighed, measured for FL, and judged by experienced fish cul- year. The second time period was the standardized monthly ture staff as to maturation status, sex, condition (good = 1; fair = Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 flow data during the kelt migration (March–May), and the third 2; poor = 3) and coloration (bright = 1; intermediate = 2; dark = time period was standardized to the steelhead run (September– 3). Methods similar to those employed by Keefer et al. (2008) April). were used to assess fish condition and coloration. Condition was After spawning in tributaries, a portion of the steelhead kelts based on the degree of visible external damage (e.g., abrasions, migrating downstream are inadvertently diverted into an irriga- lesions, fungal infections; see Evans 2003), and coloration (an tion canal that has a maximum flow of 42.5 m3/s near Prosser indicator of physiological state) was based on the degree of the Dam (USBOR 2006). The canal is equipped with a fish bypass fish’s silvery, ocean-like external appearance. A uniquely coded system that guides fish through the Chandler Juvenile Moni- PIT tag was injected into the pelvic girdle of each kelt (Prentice toring Facility (CJMF; Figure 1), where a separator is used to et al. 1990) for individual fish identification during recondition- separate large and small fish from the bypass. This separator is ing and postrelease tracking. Our evaluation was limited to kelts monitored daily from mid-March through late June annually. At retained for reconditioning as described here. Prosser Dam, the 2001–2011 mean Yakima River flow during Due to the success of Ivermectin in treating the parasitic the kelt migration (March–May) was 88.0 m3/s. freshwater Copepod Salmincola californiensis in this project’s Kelt collection, holding, and release. All kelt steelhead pilot studies conducted in 2000 (Evans and Beaty 2000), we were removed from the separator at the CJMF by dipnetting and diluted it with saline (1:30) and injected 1 to 3 mL into the 618 HATCH ET AL.

posterior end of the fish’s esophagus, using a plastic syringe and color. Additionally, we calculated correlation coefficients (Johnson and Heindel 2001). Fish were also given an initial among these fish variables along with annual abundance of kelt injection of oxytetracycline and fed the dietary supplement hw- and prespawn steelhead, the proportion that kelt steelhead made wiegandt multi vit (Wiegandt GmbH - Aquaristics Products, up of the previous run, length and weight change, and standard- Germany). Kelts were kept in one of four circular tanks, 6.1 m ized flow periods (Sokal and Rohlf 2000). diameter × 1.2 m high. Water flow to the four tanks ranged from 570 to 950 L/min of 13.8◦C well water, thus subjecting kelts to continuous current. The carrying capacity of the individual tanks RESULTS was set at a maximum of 200 fish based on standard aquacul- Prespawn Steelhead ture guidelines (Piper et al. 1982) and goals established for this From 2001 through 2011 the mean annual return of steelhead program. Formalin (37% concentration) was administered five to the Yakima River was 3,577 fish (Table 1). Because Fish times weekly for 1 h diluted 1:6000 in all reconditioning tanks migrated over Prosser Dam during all months of the year, the to prevent fungal outbreaks. On the basis of information gained run was defined as from July 1 of one year through June 30 of from a literature review (Evans et al. 2001) and our 2001 feed the following year. The majority of fish passed upstream from study (Hatch et al. 2002), reconditioned kelts were fed a com- September through April, peaking in October (Figure 2). The bination of frozen krill and 6.0-mm pellet feed manufactured median date of passage at Prosser Dam was highly variable, by Bio-Oregon. Krill was used initially to enhance the feeding occurring as early as October 18 and as late as December 26 response, and after 4–6 weeks the pellets were introduced. Feed (Table 1). Spawning in upstream tributaries generally occurred was administered 3–5 times daily at a rate of 1–2% body weight between February and June, peaking between early March and or until fish seemed satiated. The tanks were covered to pro- early May depending on stream elevation. vide shelter from sun and relieve stress from outside movement. Any kelt mortalities were removed daily and the tanks were Kelt Steelhead swept and flushed every 10–14 d as needed. The tank walls The mean annual Yakima River kelt emigration through the were painted white and the centers dark to discourage the kelts CJMF near Prosser Dam was 885 fish (Table 1). This collection from rubbing the walls. We found eye damage became prevalent represented an unknown portion of the total Yakima River kelt when fish excessively rubbed the walls. Aerators were placed abundance, as some fish passed over the dam instead of through in the tanks to break up the surface and introduce oxygen. The the irrigation canal where they could be collected. Assuming aerators appeared to reduce stress by providing added security that all upstream prespawn migrants survived to emigrate down- for the fish. stream as kelts, the average annual collection would represent Before their release from the reconditioning facility, surviv- about 27% of the total kelt population (Table 1); if only half of ing steelhead were again weighed, measured (FL), sexed, and upstream prespawn steelhead survived as kelts, the average col- scanned for PIT tags. From 2006 to 2011, kelt tanks were tran- lection would represent about 54% of the total population. Kelt sitioned from well water to river over a 3–5 d period prior to emigration occurred from March through July, peaking in April. release. Prior to 2006, kelts were tempered with river water in The median date of passage at Prosser Dam for kelt steelhead the tanker truck prior to release. Fish were released from mid- occurred as early as April 13, and as late as April 30, with a October to early December, concurrent with the peak return mean passage date of April 24 (Table 1). The abundance of emi- of the natural spawning run. Releases were made upstream of grating kelt steelhead correlated strongly with the abundance of Prosser Dam from 2001 to 2007 and a few kilometers down-

Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 stream of Prosser Dam from 2008 to 2011. The release location and timing allowed reconditioned kelts to naturally select their 80 30 70 migration timing, spawning location, spawn timing, and mates. 25 Upstream migration timing of reconditioned kelts in 2008–2011 60 20 was determined by detecting PIT tags postrelease at Prosser 50 Dam. 40 15 30 Statistical methods. To evaluate the change in fish form 10 as a result of reconditioning, we calculated Fulton’s condition 20 5 factor (K) at collection and at release (Ricker 1975). We used 10 the calculation K = w/l, where w = fish weight (kg) and l = 0 0 fish length (cm). 1-Jul 1-Aug 1-Sep 1-Oct 1-Nov 1-Dec 1-Jan 1-Feb 1-Mar 1-Apr 1-May 1-Jun One-way analysis of variance (ANOVA) was used to deter- Pre-spawn (L axis) Recondioned (R axis) mine differences, at collection, in kelt length, weight, condition, and color across years (Sokal and Rohlf 2000). We used FIGURE 2. Average timing of migration upstream by prespawn steelhead (2000–01 through 2010–11 steelhead run years) and reconditioned kelt steelhead ANOVA tests to evaluate the association of survival (0, 1) with (PIT detection, 2008–2011) at Prosser Dam. Y-axes denote number of fish several fish trait variables, including FL, weight, K, condition, counted per day; the right axis shows results for reconditioned fish. YAKIMA RIVER STEELHEAD 619

TABLE 1. Abundance and kelt trait data for upstream-migrating (prespawn) steelhead at Prosser Dam and downstream-migrating (kelt) steelhead at the Chandler Juvenile Monitoring Facility (CJMF) for 2000–01 through 2010–11 steelhead run years (July 1–June 30). Kelt FL, weight, condition, and color are mean values for annual collections.

Prespawn steelhead at Prosser Dam Kelt steelhead at collection, CJMF Median Median Proportion date of date of of prespawn FL Weight Percent Year Abundance passage Abundance passage run (cm) (kg) Condition Color female 2000–01 3,089 26 Dec 727 19 Apr 0.24 64.8 2.02 1.636 1.786 97.2 2001–02 4,525 19 Nov 1,157 24 Apr 0.26 63.4 1.96 1.561 1.603 92.6 2002–03 2,235 14 Dec 826 13 Apr 0.37 68.0 2.43 1.548 1.703 96.4 2003–04 2,755 26 Oct 998 25 Apr 0.36 60.3 1.67 1.603 1.732 93.4 2004–05 3,451 21 Oct 803 21 Apr 0.23 64.0 1.94 1.657 1.761 96.8 2005–06 2,005 18 Oct 520 13 Apr 0.26 66.5 2.10 1.648 1.660 95.2 2006–07 1,537 12 Nov 587 29 Apr 0.38 64.3 2.09 1.592 1.500 91.9 2007–08 3,310 10 Nov 847 30 Apr 0.26 62.0 1.84 1.597 1.587 92.4 2008–09 3,450 18 Nov 622 28 Apr 0.18 64.2 2.01 1.617 1.543 93.2 2009–10 6,796 3 Nov 1,659 21 Apr 0.24 62.0 1.80 1.621 1.509 89.7 2010–11 6,196 8 Nov 992 28 Apr 0.16 64.8 2.15 1.672 1.523 89.7 Average 3,577 9 Nov 885 24 Apr 0.27 63.7 1.97 1.615 1.620 92.9

the immigrating prespawn steelhead run of the same spawning years were dominated by females (Table 1; mean = 92.9% population (R2 = 0.66; Figure 3). female), whereas the mean proportion of females in the annual We found signiﬁcant differences across years (Table 1) for upstream prespawn migration was about 70% (Yakama Nation kelt FL (mean 63.7 cm; ANOVA: F = 65.96, df = 10, P < unpublished data). 0.001), weight (mean 1.97 kg; ANOVA: F = 68.33, df = 10; P < 0.001), condition at the time of collection (ANOVA: F = = = 2.77, df 10; P 0.002), and color at the time of collection Kelt Reconditioning F = = P < (ANOVA: 20.88, df 10; 0.001). Kelts from 2003 Mean survival of kelt steelhead that were reconditioned for were in the best condition and those from 2011 were in the 4.5–10 months was 38.0% (range 20.1–62.4%; Table 2). Fish poorest overall condition. Kelts from 2007 were the brightest that survived the reconditioning treatment showed increases in and those from 2001 were the darkest. Collections from all FL in 10 of the 11 years studied and increases in weight in all years; the mean increases were 0.61 cm and 0.50 kg, respectively

1800 (Table 2). Survival was positively correlated with the portion of the run that is seen to be kelts (r = 0.67; Table 3). Kelts in good 1600

Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 and fair condition on arrival survived reconditioning at higher 1400 rates than those in poor condition (χ2 = 52.59, P < 0.001); 1200 survival was 45%, 36%, and 0% for fish in good, fair, and 1000 poor condition, respectively (Figure 4). Bright and intermediate 800 colored kelts survived reconditioning better than darker fish 600 (χ2 = 30.98, P < 0.001), survival being 44%, 36%, and 32% 400 for bright, intermediate, and dark fish, respectively (Figure 4),

Downstream Kelt Collecon 200 although only a few dark ﬁsh were collected (Table 4). Migration 0 timing of PIT-tagged, reconditioned kelts was well correlated 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 (r = 0.87) with run timing of upstream prespawn migrants at Upstream 'Pre-spawn' Migraon Prosser Dam (Figure 2). Kelt survival was not associated with Fulton’s K factor at intake (ANOVA: F = 1.05, df = 1,061, FIGURE 3. Kelt abundance at Chandler Juvenile Monitoring Facility as a P = 0.166). Kelt survival decreased with length at collection function of upstream prespawn steelhead count at Prosser Dam for 2000–01 = = < through 2010–11 steelhead run years. The least squares linear regression line is (ANOVA: F 2.77, df 42, P 0.001; Figure 5). Kelt survival as follows: Downstream kelt collection = 0.057 × upstream prespawn abun- was not associated with ﬁsh weight at collection (ANOVA: F = dance + 323.66 (R2 = 0.663, P < 0.001). [Figure available in color online.] 1.14, df = 217, P = 0.078). 620 HATCH ET AL.

TABLE 2. Abundance and kelt trait data for reconditioned kelt steelhead that survived to release, 2001–2011.

Means at Mean FL (cm) Mean weight (kg) Mean Fulton’s K collection No. No. Survival Year Reconditioned Released (%) Collected Released Changed Collected Released Changed Collected Released Condition Color

2001 508 108 21.3 64.18 66.82 2.65 2.00 3.22 1.22 0.031 0.048 1.56 1.81 2002 420 142 33.8 62.68 63.39 0.71 1.92 2.41 0.49 0.030 0.037 1.48 1.50 2003 482 301 62.4 67.35 68.03 0.68 2.37 3.04 0.67 0.035 0.044 1.51 1.72 2004 694 288 41.5 59.29 60.05 0.77 1.59 2.04 0.45 0.027 0.034 1.56 1.65 2005 427 86 20.1 62.15 62.28 0.13 1.82 2.26 0.44 0.029 0.036 1.62 1.68 2006 279 85 30.5 65.78 65.86 0.09 2.08 2.59 0.51 0.031 0.039 1.63 1.63 2007 422 221 52.4 63.14 63.28 0.15 1.98 2.31 0.33 0.031 0.036 1.56 1.44 2008 472 266 56.4 61.49 62.93 1.44 1.83 2.34 0.52 0.029 0.037 1.53 1.50 2009 510 141 27.6 63.22 64.82 1.61 1.96 2.61 0.65 0.030 0.040 1.54 1.55 2010 1,100 426 38.7 60.91 60.45 −0.46 1.70 2.00 0.30 0.028 0.033 1.58 1.51 2011 680 223 32.8 63.30 64.13 0.83 2.04 2.50 0.47 0.032 0.038 1.61 1.49 Average 545 208 38.0 63.05 63.35 0.61 1.92 2.41 0.50 0.030 0.037 1.56 1.57

TABLE 3. Pearson correlation matrix r values (shaded) and associated unadjusted P-values (unshaded) for mean annual variables from Tables 1 and 2. Additionally, three standardized (Std.) flow variables measured near Prosser Dam are presented. The standardized monthly flow is the mean monthly flow for a year divided by the 10-year average. The standardized spring monthly flow includes only March through May; the standardized steelhead run flow includes September through April. Significant correlations and their associated P-values are in bold.

K Factor Prespawn Std. Std. spring Std. Steelhead steelhead Kelt Proportion Kelt At At Condition at Color at monthly monthly run monthly abundance abundance of kelts survival FL Wt. collection release collection collection flow flow flow

Prespawn steelhead 1.000 0.814 −0.637 −0.265 −0.176 −0.231 −0.256 −0.330 0.313 −0.416 0.194 0.021 −0.127 abundance Kelt abundance 0.002 1.000 −0.130 0.053 −0.367 −0.350 −0.388 −0.474 −0.106 −0.262 −0.151 −0.249 −0.324 Proportion of kelts 0.035 0.704 1.000 0.677 −0.240 −0.155 0.125 −0.065 −0.664 0.204 −0.237 −0.044 0.128 Kelt survival 0.430 0.876 0.022 1.000 −0.192 −0.304 0.297 −0.120 −0.697 −0.285 0.185 0.307 0.324 FL 0.605 0.267 0.478 0.571 1.000 0.862 0.142 0.723 −0.001 0.335 −0.198 −0.159 −0.222 Weight 0.493 0.292 0.650 0.363 0.001 1.000 0.335 0.919 0.044 0.568 −0.371 −0.373 −0.387 K Factor At collection 0.447 0.238 0.713 0.374 0.676 0.314 1.000 0.671 −0.224 0.027 0.375 0.355 0.376 At release 0.322 0.141 0.849 0.725 0.012 0.001 0.024 1.000 −0.058 0.462 −0.154 −0.163 −0.160 Condition at 0.348 0.756 0.026 0.017 0.997 0.897 0.507 0.865 1.000 0.045 0.213 0.085 0.015 collection Color at collection 0.203 0.437 0.547 0.396 0.314 0.068 0.937 0.152 0.895 1.000 −0.709 −0.710 −0.659 Std. monthly flow 0.568 0.658 0.483 0.585 0.559 0.261 0.256 0.650 0.530 0.015 1.000 0.965 0.859 Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 Std. spring monthly 0.952 0.461 0.898 0.358 0.641 0.259 0.284 0.632 0.803 0.014 0.001 1.00 0.945 flow Std. steelhead run 0.709 0.331 0.707 0.330 0.512 0.240 0.255 0.639 0.964 0.028 0.001 0.001 1.00 monthly flow

TABLE 4. Composition (% [n]) of surviving reconditioned kelt steelhead according to condition and color categories at the time of collection.

Color at collection Condition at collection Bright Intermediate Dark Total Good 25.9 (525) 17.9 (362) 0.6 (13) 44.5 (900) Fair 19.1 (387) 35.1 (711) 1.3 (26) 55.5 (1124) Poor 0.0 (0) 0.0 (0) 0.0 (0) 0.0 (0) Total 45.1 (912) 53.0 (1073) 1.9 (39) 100 (2024) YAKIMA RIVER STEELHEAD 621

FIGURE 5. Mean survival for reconditioned kelts by FL at collection, replicated by years 2001–2011, in fish captured at CJMF, Yakima River. The least squares linear regression line is as follows: survival = 0.979 – 0.009 × FL (R2 = 0.014, P < 0.001). Error bars = 95% confidence interval. Length-group 50 cm includes all fish smaller than 50 cm, and length-group 85 cm includes all fish greater than 85 cm.

low survival of individuals collected over the last few weeks of the season (Figure 6). To further investigate the week effect, we divided weekly collections into two groups—weeks 10–23 and weeks 24 onward—and compared the association of survival and collection group. Survival rates were 39.2% for the early arriving group and 29.3% for the later arriving group (χ2 = 3.72, P = 0.054).

Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 DISCUSSION We found prespawn steelhead abundance to be positively correlated with kelt abundance. Although this was expected, it differs somewhat from what Narum et al. (2008) found in the Snake River, where kelt steelhead proportions were not signifi- FIGURE 4. Mean survival for reconditioned kelts by (a) condition rating (1 = cantly correlated with escapement proportions of the reporting good, 2 = fair, 3 = poor) and (b) color rating (1 = bright, 2 = intermediate, 3 = groups. The strong correlation we observed may be a result of dark) at collection, replicated by years 2001–2011, in fish captured at CJMF, the geographic proximity of Prosser Dam to steelhead spawn- Yakima River. Error bars = 95% confidence interval. ing aggregates, relative to locations in the Snake River study. In the Yakima River Basin, the majority of returning steelhead We found significant differences in survival for both the year spawn in Toppenish and Satus creeks (Yakama Nation unpub- and the statistical week in which kelts were collected (year lished data), the upper reaches of which are at most 80–100 km association: χ2 = 395.13, P < 0.001; week effect, ANOVA: upstream from Prosser Dam (Figure 1). The furthest upstream F = 5.09, df = 18, P < 0.001). The year association was mainly spawning activity occurs in the upper reaches of the Yakima attributable to low survival of reconditioned fish in 2001 and and Naches rivers, approximately 200–240 km upstream from 2005 (Table 2), and the week association was primarily due to Prosser Dam. In contrast, the upper reaches of Snake River 622 HATCH ET AL.

River is a considerably smaller system and may demonstrate the effect. Alternatively, steelhead returning to the Snake River are generally larger, driven by populations that spend an additional year in the ocean relative to other CRB populations (Busby et al. 1996), and these older steelhead may be less likely to be successfully iteroparous (Keefer et al. 2008). We did find that reconditioning survival decreases with increasing fish length (Figure 5), which supports this notion; however, the mean FL of kelts in our collection was 63 cm; few fish in the collection were longer than 78 cm, a size-class more commonly found in the Snake River, and the decrease in survival rate with increasing length was slight. We did find that an increase in spawner density negatively affected the proportion of kelts collected in the Yakima River,which could also explain why the largest populations in Narum et al. (2008) had the lowest kelt proportions collected at Lower Granite Dam. While the overall relationship of spawner abundance and kelt abundance we observed was positive, two key pieces of data we collected support the idea that spawner abundance may have a negative effect on kelt proportion and their survival. First, we found a negative relationship between the portion of the run seen as kelts and prespawn fish abundance (r = –0.64; Table 3). This is probably not a function of river environment and collection efficiency since flow was not correlated with the proportion of kelts collected (a measure of collection efficiency), even though a greater proportion of the overall flow is diverted into the irrigation canal in low-flow years than in high flow years (Table 3). The second piece of evidence is the positive correlation we observed between the proportion of kelts collected and the kelts that survive reconditioning (r = 0.68; Table 3). A possible explanation for these findings is density-dependent competition on the spawning grounds. High spawner abundance results in more time on the spawning grounds for both male and female fish, with greater competition for redd sites among females and greater competition for access to females among males (Quinn 2005). The end result would be greater energy expenditure, more injuries, and ultimately fewer kelts (and those in poorer condition), emigrating through the system. Evidence support-

Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 ing this is that the greater proportion of kelts collected was correlated with fish collected in better condition (r = –0.66; Table 3) and the kelts in better condition survived the reconditioning process at a higher rate (r = –0.70). Stearns (1992) documented an evolutionary life history trade-off between energy investment in current reproduction and survival to breed FIGURE 6. Mean survival for reconditioned kelts by (a) collection year and again. The physical competition among males consumes energy (b) statistical week, replicated by years, in fish captured at CJMF, Yakima River, (Jonsson et al. 1997) and often results in physical injury, possi- 2001–2011. Error bars = 95% confidence interval. Statistical week 10 includes bly accounting for the sex ratio of kelt steelhead being skewed fish arriving before week 10 in some years and statistical week 26 includes fish in favor of females. The high proportion of females among the arriving later than week 26 in some years. emigrating kelts may be indicative of an evolutionary advantage of female iteroparity. A trend toward higher postspawn female steelhead spawning habitats are more than 700 km upstream of survival, relative to males, is consistent with data from other Lower Granite Dam on the Snake River (ICBTRT 2003). Other iteroparous salmonid populations (Keefer et al. 2008; Seamons than one population, Narum et al. (2008) did not find that pop- and Quinn 2010). Kelts collected earlier in the season sur- ulation distance influenced kelt composition, but the Yakima vived better than those collected late in the season (χ2 = 3.72, YAKIMA RIVER STEELHEAD 623

P < 0.054), also most probably a result of greater energy the notion (Helle 1989; Garant et al. 2001) and some does not expenditure. (Holtby and Healey 1986). Survival rates through the reconditioning process were as In summary, we demonstrate that environmentally threatened expected for the various condition factors we examined. While wild steelhead kelts can be collected and reconditioned to at- the proportion of kelts in good condition entering our recon- tain survival rates considerably higher than if no action were ditioning program (>40%) was comparable to the proportion taken. Fish condition, collection date, and prespawn abundance of good-condition emigrating kelts reported by Keefer et al. influenced reconditioning survival, suggesting that selection of (2008), we observed better survival-to-release rates for fair con- fish at intake and the number of fish collected for reconditioning dition and darker colored kelts than the survival-to-return rates can be tailored to achieve program goals. Achieving reasonable in the wild observed by Keefer et al. (2008); they found that fair survival rates by reconditioning wild kelt steelhead is a first step fish were 2.5–5.7 times less likely to return relative to kelts in toward the development and implementation of this new stock good condition. Similarly, Evans et al. (2008) reported return recovery tool. However, to provide demographic and genetic rates for kelts in fair condition as 0.8% across all treatments, benefits to the population, reconditioned kelts must migrate and sites, and years. This suggests that the reconditioning program spawn successfully after release. Surviving fish increased in may provide a relatively greater benefit for fair condition fish weight and length during reconditioning, and most resumed up- versus good condition fish. stream migration upon release, suggesting that these fish prob- The survival rates (20–62%) we observed for captive, recon- ably did spawn. However, additional studies of the reproductive ditioned steelhead kelts were lower than reported survival rates success of reconditioned kelts are required to quantify the ben- of approximately 80–95% for Atlantic Salmon kelts reared in efit of the reconditioning program. Studies of the reproductive freshwater or seawater (Gray et al. 1987; Johnston et al. 1987, success of reconditioned kelts using genetic parentage analysis 1990; Dumas et al. 1991). However, our results were consistent and studies using physiological indicators of reconditioned kelts with the kelt survival rates of 28–55% reported by Moffett et al. at release are underway. Additionally, studies of kelt migration (1996) for Atlantic Salmon artificially reconditioned in fresh- success, spawn timing and location, gamete quality, and progeny water over two successive years and the reconditioned steelhead viability are in progress and will help quantify the potential of kelt return rates of 26% reported by Null et al. (2013). The an- this management tool. Potential risks such as residualism of nual variation we found in survival to release was correlated reconditioned kelts (Null et al. 2013), leading to negative eco- with condition, suggesting that the environmental conditions logical effects to fish in the river, should be studied. These risks (i.e., temperature and flow) steelhead experience throughout the should be weighed against benefits such as increased abundance, winter and spring prior to collection (during holding, spawning, maintenance of the iteroparous phenotype, maintenance of the and repeat out-migration) along with prespawn steelhead den- genetic diversity of the population (Crespi and Teo 2002), in- sities may play a substantial role in whether these fish survive creased population productivity (Fleming and Reynolds 2004; as captive kelts. Similar condition-dependent mortality in kelt Seamons and Quinn 2010), and protection against cohort fail- steelhead returns was reported by Keefer et al. (2008) for fish ure (Fleming and Reynolds 2004; Wilbur and Rudolf 2006). The that remained in the river. empirical results we observed demonstrate the potential of kelt Seamons and Quinn (2010) reported that repeat spawning reconditioning to provide recovery benefits for imperiled repeat adults have life-time reproductive success more than twice that spawning wild populations in highly developed river systems. of one-time spawners, and the average number of offspring produced by both male and female repeat spawners is much higher

Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 (1.9 times higher for females and 2.7 times higher for males). ACKNOWLEDGMENTS As the cumulative number of progeny gained by surviving to Monitoring and evaluation efforts for the steelhead kelt re- spawn in multiple subsequent years outweighs the number of conditioning project are the result of a cooperative effort by progeny lost by not spawning in a given single year, an occa- many individuals from a variety of agencies, including the sional skipped spawning may constitute an adaptive trait in long- Yakama Nation Fisheries, the Columbia River Inter-Tribal Fish lived iteroparous fish (Rideout et al. 2005). Seamons and Quinn Commission, the U.S. Bureau of Reclamation, the University (2010) further reported that repeat spawners grew substantially of Idaho, the Pacific States Marine Fisheries Commission, and between their first and second breeding seasons (female mean the National Marine Fisheries Service. We would especially like growth of 41 mm; males 71 mm) and estimated this additional to thank the many managers, biologists, technicians, and staff female growth would result in an average increase in fecundity who have contributed to the protection of Yakima Basin steel- of about 400 eggs, or about 10%. Surviving captive kelts grew a head and their habitats through the years. Reviews by Matthew comparable amount during reconditioning (Table 2), suggesting Keefer and three anonymous reviewers improved this paper. that an increase in fecundity could be expected. The hypothe- This work was funded by the Bonneville Power Administration sis that larger fish are more productive has been tested at the through the Northwest Power and Conservation Council’s Fish population level with equivocal results: some evidence supports and Wildlife Program. 624 HATCH ET AL.

REFERENCES Gray, R. W., J. D. Cameron, and A. D. McLennan. 1987. Artificial recondi- Berejikian, B. A., E. P. Tezak, and S. L. Schroder. 2001. Reproductive behavior tioning, spawning and survival of Atlantic Salmon, Salmo salar L., kelts and breeding success of captively reared Chinook Salmon. North American in salt water and survival of their F1 progeny. Aquaculture and Fisheries Journal of Fisheries Management 21:255–260. Management 18:309–326. Boyer, J. N., and W. Van Toever. 1993. Reconditioning of Arctic Char (Salveli- Gunsolus, R. T., and G. J. Eicher. 1970. Evaluation of fish-passage facilities at nus alpinus) after spawning. Aquaculture 110:279–284. the North Fork project on the Clackamas River in Oregon. Fish Commission Busby, P. J., T. C. Wainwright, G. J. Bryant, L. J. Lierheimer, R. S. Waples, of Oregon, Portland. F. W. Waknitz, and I. V. Lagomarsino. 1996. Status review of West Coast Hatch, D., P. Anders, A. Evans, J. Blodgett, B. Bosch, D. Fast, and T. New- steelhead from Washington, Idaho, Oregon, and California. NOAA Technical some. 2002. Kelt reconditioning: a research project to enhance iteroparity in Memorandum NMFS-NWFSC-27. Columbia Basin steelhead (Oncorhynchus mykiss). 2001 Annual Report to Carmona-Catot, G., P. B. Moyle, and R. E. Simmons. 2012. Long-term captive the Bonneville Power Administration, Project 2000-01700, Portland, Oregon. breeding does not necessarily prevent reestablishment: lessons learned from Helle, J. H. 1989. Relation between size-at-maturity and survival of progeny Eagle Lake Rainbow Trout. Reviews in Fish Biology and Fisheries 22:325– in Chum Salmon, Oncorhynchus keta (Walbaum). Journal of Fish Biology 342. 35(Supplement A):99–107. Cloud, J. G., W. H. Miller, and M. J. Levanduski. 1990. Cryopreservation of Hockersmith, E., J. Vella, L. Stuehrenberg, R. N. Iwamoto, and G. Swan. 1995. sperm as a means to store salmonid germ plasm and to transfer genes from Yakima River radio-telemetry study: steelhead, 1989–93. Report to the Bon- wild fish to hatchery populations. Progressive Fish-Culturist 52:51–53. neville Power Administration, Project 89-089, Portland, Oregon. Conley, A., J. Freudenthal, D. Lind, P. Mees, and R. Visser. 2009. 2009 Yakima Holtby, L. B., and M. C. Healey. 1986. Selection for adult size in female Coho steelhead recovery plan, extracted from the 2005 Yakima subbasin salmon Salmon (Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic recovery plan with updates, September 30, 2009. Yakima Basin Fish and Sciences 43:1946–1959. Wildlife Recovery Board, Yakima, Washington. Howell, P., K. Jones, D. Scarnecchia, L. LaVoy, W. Rendra, and D. Ortmann. Crespi, B. J., and R. Teo. 2002. Comparative phylogenetic analysis of the evo- 1985. Stock assessment of Columbia River anadromous salmonids—volume lution of semelparity and life history in salmonid fishes. Evolution 56:1008– II: steelhead stock summaries stock transfer guidelines, information needs. 1020. Report to the Bonneville Power Administration, Project 83-335, Portland, Crim, L. W., C. E. Wilson, Y. P. So, D. R. Idler, and C. E. Johnston. 1992. Feed- Oregon. ing, reconditioning, and rematuration responses of captive Atlantic Salmon ICBTRT (Interior Columbia Basin Technical Recovery Team). 2003. Indepen- (Salmo salar) kelt. Canadian Journal of Fisheries and Aquatic Sciences dent populations of Chinook, steelhead and Sockeye for listed evolutionarily 49:1835–1842. significant units within the interior Columbia River domain. National Ma- Cuenco, M. L., T. W. H. Backman, and P. R. Mundy. 1993. The use of supple- rine Fisheries Service, Northwest Fisheries Science Center, ICBTRT, Tech- mentation to aid in natural stock restoration. Pages 269–293 in J. G. Cloud and nical Review Draft, Seattle. Available: www.nwfsc.noaa.gov/trt/col docs/ G. H. Thorgaard, editors. Genetic conservation of salmonid fishes. Plenum, independentpopchinsteelsock.pdf. (July 2012). New York. Johnson, K. A., and J. A. Heindel. 2001. Efficacy of manual removal and iver- Dobson, A. P., A. D. Bradshaw, and A. J. M. Baker. 1997. Hopes for the future: mectin gavage for control of Salmincola californiensis (Wilson) infestation restoration ecology and conservation biology. Science 277:515–522. of Chinook Salmon, Oncorhynchus tshawytscha (Walbaum), captive brood- Dumas, J., L. Barriere,` D. Blanc, J. Godard, and S. J. Kaushik. 1991. Recondi- stocks. Journal of Fish Diseases 24:197–203. tioning of Atlantic Salmon (Salmo salar) kelts with silage-based diets: growth Johnston, C. E., S. R. Farmer, R. W. Gray, and M. Hambrook. 1990. Recon- and reproductive performance. Aquaculture 96:43–56. ditioning and reproductive responses of Atlantic Salmon kelts (Salmo salar) Evans, A. F. 2003. Development and application of steelhead (Oncorhynchus to photoperiod and temperature manipulation. Canadian Journal of Fisheries mykiss) kelt identification techniques. Master’s thesis. Oregon State Univer- and Aquatic Sciences 47:701–710. sity, Corvallis. Johnston, C. E., R. W. Gray, A. McLennan, and A. Paterson. 1987. Effects Evans, A. F., and R. E. Beaty. 2000. Identification and enumeration of steelhead of photoperiod, temperature, and diet on the reconditioning response, blood (Oncorhynchus mykiss) kelts at Little Goose Dam juvenile bypass separa- chemistry, and gonad maturation of Atlantic Salmon kelts (Salmo salar) held tor, 1999. Annual Report to U.S. Army Corps of Engineers, Walla Walla in freshwater. Canadian Journal of Fisheries and Aquatic Sciences 44:702– District, Contract DACW68-99-M-3102, Columbia River Inter-Tribal Fish 711. Commission, Portland, Oregon. Johnston, C. E., M. J. Hambrook, R. W. Gray, and K. G. Davidson. 1992. Evans, A. F., R. E. Beaty, M. S. Fitzpatrick, and K. Collis. 2004. Identifica- Manipulation of reproductive function in Atlantic Salmon (Salmo salar) kelts Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 tion and enumeration of steelhead kelts at a Snake River hydroelectric dam. with controlled photoperiod and temperature. Canadian Journal of Fisheries Transactions of the American Fisheries Society 133:1089–1099. and Aquatic Sciences 49:2055–2061. Evans, A. F., R. E. Beaty, D. R. Hatch, J. Blodgett, and D. Fast. 2001. Kelt Jonsson, N., B. Jonsson, and L. P. Hansen. 1997. Changes in proximate com- reconditioning: a research project to enhance iteroparity in Columbia Basin position and estimates of energetic costs during upstream migration and steelhead (Oncorhynchus mykiss). 2000 Annual Report to the Bonneville spawning in Atlantic Salmon Salmo salar. Journal of Animal Ecology 66: Power Administration, Project 2000-017, Portland, Oregon. 425–436. Evans, A. F., R. H. Wertheimer, M. L. Keefer, C. T. Boggs, C. A. Peery, and K. Keefer, M. L., R. H. Wertheimer, A. F. Evans, C. T. Boggs, and C. A. Peery. Collis. 2008. Transportation of steelhead kelts to increase iteroparity in the 2008. Iteroparity in Columbia River summer-run steelhead (Oncorhynchus Columbia and Snake rivers. North American Journal of Fisheries Manage- mykiss): implications for conservation. Canadian Journal of Fisheries and ment 28:1818–1827. Aquatic Sciences 65:2592–2605. Evermann, B. W. 1895. A preliminary report upon salmon investigations in Long, J. B., and L. E. Griffin. 1937. Spawning and migratory habits of the Idaho in 1894. U.S. Fish Commission Bulletin 15:253–284. Columbia River steelhead trout as determined by scale studies. Copeia Fleming, I. A., and J. D. Reynolds. 2004. Salmon breeding systems. Pages 264– 1937:62. 294 in A. P. Hendry and S. C. Stearns, editors. Evolution illuminated: salmon Margules, C. R., and R. L. Pressey. 2000. Systematic conservation planning. and their relatives. Oxford University Press, Oxford, UK. Nature 405:243–253. Garant, D., J. J. Dodson, and L. Bernatchez. 2001. A genetic evaluation of Moffett, I. J. J., G. J. A. Kennedy, and W. W. Crozier. 1996. Freshwater recon- mating system and determinants of individual reproductive success in Atlantic ditioning and ranching of Atlantic Salmon, Salmo salar L., kelts: growth and Salmon (Salmo salar L.). Journal of Heredity 92:137–145. reproductive performance. Fisheries Management and Ecology 3:35–44. YAKIMA RIVER STEELHEAD 625

Narum, S. R., D. Hatch, A. J. Talbot, P. Moran, and M. S. Powell. 2008. Seamons, T. R., M. B. Dauer, J. Sneva, and T. P. Quinn. 2009. Use of parentage Iteroparity in complex mating systems of steelhead Oncorhynchus mykiss assignment and DNS genotyping to validate scale analysis for estimating (Walbaum). Journal of Fish Biology 72:45–60. steelhead age and spawning history. North American Journal of Fisheries Nehlsen, W., J. E. Williams, and J. A. Lichatowich. 1991. Pacific salmon at the Management 29:396–403. crossroads: stocks at risk from California, Oregon, Idaho, and Washington. Seamons, T. R., and T. P. Quinn. 2010. Sex-specific patterns of lifetime repro- Fisheries 16(2):4–21. ductive success in single and repeat breeding steelhead trout (Oncorhynchus Nielsen, J. L., S. M. Turner, and C. E. Zimmerman. 2011. Electronic tags mykiss). Behavioral Ecology and Sociobiology 64:505–513. and genetics explore variation in migrating steelhead kelts (Oncorhynchus Sokal, R. R., and F. J. Rohlf. 2000. Biometry: the principles and practices of mykiss), Ninilchik River, Alaska. Canadian Journal of Fisheries and Aquatic statistics in biological research. Freeman, New York. Sciences 68:1–16. Stearns, S. C. 1992. The evolution of life histories. Oxford University Press, NRC (National Research Council). 1996. Upstream: salmon and society in the New York. Pacific Northwest. National Academy Press, Washington, D.C. USBOR (U.S. Bureau of Reclamation). 2006. Fish protection at water diver- Null, R. E., K. S. Niemela, and S. F. Hamelberg. 2013. Post-spawn migrations of sions: a guide for planning and designing fish exclusion facilities. USBOR, hatchery-origin Oncorhynchus mykiss kelts in the Central Valley of California. Water Resources Technical Publication, Denver. Available: www.usbr. Environmental Biology of Fishes 96:341–353. gov/pmts/hydraulics lab/pubs/manuals/fishprotection/Fish%20Protection% Piper, R. G., I. B. McElwain, L. E. Orme, J. P. McCraren, L. G. Fowler, and J. R. 20at%20Water%20Diversions.pdf. (July 2012). Leonard. 1982. Fish hatchery management. U.S. Fish and Wildlife Service, Ward, B. R., and P. A. Slaney. 1988. Life history and smolt-to-adult survival Washington, D.C. of Keogh River steelhead trout (Salmo gairdneri) and the relationship to Poole, W. R., M. G. Dillane, and K. F. Whelan. 1994. Artificial reconditioning smolt size. Canadian Journal of Fisheries and Aquatic Sciences 45:1110– of wild Sea Trout, Salmo trutta L., as an enhancement option: initial results 1122. on growth and spawning success. Fisheries Management and Ecology 1:179– Wertheimer, R. H. 2007. Evaluation of a surface flow bypass system for steelhead 192. kelt passage at Bonneville Dam, Washington. North American Journal of Prentice, E. F., T. A. Flagg, C. S. McCutcheon, D. F. Brastow, and D. C. Cross. Fisheries Management 27:21–29. 1990. Equipment, methods, and an automated data-entry station for PIT Wertheimer, R. H., and A. F. Evans. 2005. Downstream passage of steelhead tagging. Pages 335–340 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. kelts through hydroelectric dams on the lower Snake and Columbia rivers. Jester Jr., E. D. Prince, and G. A. Winans, editors. Fish-marking techniques. Transactions of the American Fisheries Society 134:853–865. American Fisheries Society, Symposium 7, Bethesda, Maryland. Whitt, C. R. 1954. The age, growth, and migration of steelhead trout in the Quinn, T. P. 2005. The behavior and ecology of Pacific salmon and trout. Clearwater River, Idaho. Master’s thesis. University of Idaho, Moscow. American Fisheries Society, Bethesda, Maryland. Wilbur, H. M., and V. H. W. Rudolf. 2006. Life-history evolution in uncertain Raymond, H. L. 1988. Effects of hydroelectric development and fisheries en- environments: bet hedging in time. American Naturalist 168:398–411. hancement on spring and summer Chinook Salmon and steelhead in the Williams, R. N., P. A. Bisson, D. L. Bottom, L. D. Calvin, C. C. Coutant, M. W. Columbia River basin. North American Journal of Fisheries Management Erho Jr., C. A. Frissell, J. A. Lichatowich, W. J. Liss, W. E. McConnaha, 8:1–24. P. R. Mundy, J. A. Stanford, and R. R. Whitney. 1999. Return to the river: Ricker, W. E. 1975. Computation and interpretation of biological statistics of scientific issues in the restoration of salmonid fishes in the Columbia River. fish populations. Fisheries Research Board of Canada Bulletin 191. Fisheries 24(3):10–19. Rideout, R. M., G. A. Rose, and M. P.M. Burton. 2005. Skipped spawning in YSFWPB (Yakima Subbasin Fish and Wildlife Planning Board). 2004. Yakima female iteroparous fishes. Fish and Fisheries 6:50–72. subbasin plan. Northwest Power and Conservation Council, Portland, Oregon. Downloaded by [Department Of Fisheries] at 20:12 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:13 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 An Individual-Based Model for Population Viability Analysis of Humpback Chub in Grand Canyon William E. Pine III a , Brian Healy b , Emily Omana Smith b , Melissa Trammell c , Dave Speas d , Rich Valdez e , Mike Yard f , Carl Walters g , Rob Ahrens h , Randy Vanhaverbeke i , Dennis Stone i & Wade Wilson j a Department of Wildlife Ecology and Conservation , University of Florida , 110 Newins- Ziegler Hall, Gainesville , Florida , 32611 , USA b U.S. National Park Service , 1824 South Thompson Street, Flagstaff , Arizona , 86001 , USA c U.S. National Park Service , 324 South State Street, Suite 200, Salt Lake City , Utah , 84111 , USA d U.S. Bureau of Reclamation , 125 South State Street, Room 6107, Salt Lake City , Utah , 84138 , USA e SWCA Environmental Consultants , 172 West 1275 South, Logan , Utah , 84321 , USA f U.S. Geological Survey, Grand Canyon Monitoring and Research Center , 2255 North Gemini Drive, Flagstaff , Arizona , 86001 , USA g Fisheries Centre , University of British Columbia , 2202 Main Mall, Vancouver , British Columbia , V6T 1Z4 , Canada h School of Forest Resources and Conservation, Fisheries Program , University of Florida , 7922 Northwest 71st Street, Gainesville , Florida , 32653 , USA i U.S. Fish and Wildlife Service , 323 North Leroux Street, Flagstaff , Arizona , 86001 , USA j U.S. Fish and Wildlife Service, Conservation Genetics Laboratory, Dexter Fish Technology Center , 7127 Hatchery Road, Dexter , New Mexico , 88230 , USA Published online: 24 May 2013.

To cite this article: William E. Pine III , Brian Healy , Emily Omana Smith , Melissa Trammell , Dave Speas , Rich Valdez , Mike Yard , Carl Walters , Rob Ahrens , Randy Vanhaverbeke , Dennis Stone & Wade Wilson (2013): An Individual-Based Model for Population Viability Analysis of Humpback Chub in Grand Canyon, North American Journal of Fisheries Management, 33:3, 626-641 To link to this article: http://dx.doi.org/10.1080/02755947.2013.788587

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 North American Journal of Fisheries Management 33:626–641, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.788587

ARTICLE

An Individual-Based Model for Population Viability Analysis of Humpback Chub in Grand Canyon

William E. Pine III* Department of Wildlife Ecology and Conservation, University of Florida, 110 Newins-Ziegler Hall, Gainesville, Florida 32611, USA Brian Healy and Emily Omana Smith U.S. National Park Service, 1824 South Thompson Street, Flagstaff, Arizona 86001, USA Melissa Trammell U.S. National Park Service, 324 South State Street, Suite 200, Salt Lake City, Utah 84111, USA Dave Speas U.S. Bureau of Reclamation, 125 South State Street, Room 6107, Salt Lake City, Utah 84138, USA Rich Valdez SWCA Environmental Consultants, 172 West 1275 South, Logan, Utah 84321, USA Mike Yard U.S. Geological Survey, Grand Canyon Monitoring and Research Center, 2255 North Gemini Drive, Flagstaff, Arizona 86001, USA Carl Walters Fisheries Centre, University of British Columbia, 2202 Main Mall, Vancouver, British Columbia V6T 1Z4, Canada Rob Ahrens School of Forest Resources and Conservation, Fisheries Program, University of Florida, 7922 Northwest 71st Street, Gainesville, Florida 32653, USA Randy Vanhaverbeke and Dennis Stone U.S. Fish and Wildlife Service, 323 North Leroux Street, Flagstaff, Arizona 86001, USA Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 Wade Wilson U.S. Fish and Wildlife Service, Conservation Genetics Laboratory, Dexter Fish Technology Center, 7127 Hatchery Road, Dexter, New Mexico 88230, USA

Abstract We developed an individual-based population viability analysis model (females only) for evaluating risk to populations from catastrophic events or conservation and research actions. This model tracks attributes (size, weight, viability, etc.) for individual ﬁsh through time and then compiles this information to assess the extinction risk of the

*Corresponding author: billpine@uﬂ.edu Received August 28, 2012; accepted March 13, 2013

626 POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 627

population across large numbers of simulation trials. Using a case history for the Little Colorado River population of Humpback Chub Gila cypha in Grand Canyon, Arizona, we assessed extinction risk and resiliency to a catastrophic event for this population and then assessed a series of conservation actions related to removing speciﬁc numbers of Humpback Chub at different sizes for conservation purposes, such as translocating individuals to establish other spawning populations or hatchery refuge development. Our results suggested that the Little Colorado River population is generally resilient to a single catastrophic event and also to removals of larvae and juveniles for conservation purposes, including translocations to establish new populations. Our results also suggested that translocation success is dependent on similar survival rates in receiving and donor streams and low emigration rates from recipient streams. In addition, translocating either large numbers of larvae or small numbers of large juveniles has generally an equal likelihood of successful population establishment at similar extinction risk levels to the Little Colorado River donor population. Our model created a transparent platform to consider extinction risk to populations from catastrophe or conservation actions and should prove useful to managers assessing these risks for endangered species such as Humpback Chub.

Historically, Humpback Chub Gila cypha in Grand Canyon junction with efforts to control exotic predators (mainly trout were thought to be widely distributed with spawning popula- [family Salmonidae]) and with increased water temperatures tions in both the main-stem Colorado River and several tribu- associated with low reservoir levels in Lake Powell (Coggins tary systems (Ryel and Valdez 1995, observations summarized et al. 2011). Climate conditions driving reservoir water levels, in Webb et al. 2002). Since the completion of the Glen Canyon hydrology, and water deliveries that control riverine water Dam (GCD) in 1963 and the filling of Lake Powell upstream temperatures (Pulwarty and Melis 2001) are increasingly of Grand Canyon, water temperatures in this river reach have uncertain as are management options to control nonnative fish. generally been considered too low for spawning and success- Coupled with increasing uncertainty over future water resource ful recruitment of Humpback Chub (Hamman 1982a; Kaeding development projects in key tributary systems such as the Little and Zimmerman 1983). Currently only one tributary, the Little Colorado River from mining and other water use projects, Colorado River, is known to support Humpback Chub spawn- these (and other) risks to Humpback Chub in Grand Canyon ing and rearing (Gorman and Stone 1999), and efforts are on- persist. going in three other tributaries (Shinumo, Havasu, and Bright Our knowledge of juvenile Humpback Chub population ecol- Angel creeks) to evaluate their potential as translocation sites ogy is much sparser than for adult life stages. Based on our col- (Trammell et al. 2012). Other aggregations of adult and juve- lective experiences, the current paradigm for juveniles suggests nile fish are found at several locations in the main stem, with that (1) there is some evidence of strong density dependence in most associated with tributary inflows (Ryel and Valdez 1995; survival rates of juveniles rearing in the Little Colorado River, Paukert et al. 2006). There is little evidence of successful recruit- (2) there is little apparent relationship between age-2 recruitment outside of the Little Colorado River (Valdez and Masslich ment reconstructed from PIT tag data and estimated adult abun- 1999; Andersen et al. 2010), and tagging data suggest that most dance, and (3) strong age-1 juvenile cohorts are seen in hoop net fish originate as dispersers from the Little Colorado River (i.e., sampling roughly every other year, but by age 2 these cohorts they are “sink” populations [Paukert et al. 2006]). Juveniles and are typically no more abundant than fish from less abundant adults from the Little Colorado River population continue to use age-1 cohorts. As an example, Ryel and Valdez (1995) tracked

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 the Little Colorado River and main stem seasonally. one particularly strong cohort (1993) and suggested it suffered Humpback Chub populations have declined from historical much higher dispersal and mortality rates, along with relatively levels, and the risk of extinction is sufficient to warrant listing poor growth, than surrounding cohorts. These observations are under the U.S. Endangered Species Act (ESA; Coggins et al. key to developing an effective translocation strategy because if 2006). Three primary reasons for the decline of Humpback Chub juvenile mortality rates are indeed strongly density dependent, populations in Grand Canyon are widely considered, including with the Little Colorado River being “filled to rearing capacity” (1) negative interactions with nonnative fish (Coggins et al. every year, then it may be possible to remove substantial propor- 2011; Yard et al. 2011), (2) loss of essential habitats following tions of the Little Colorado River juveniles for transplants and flow modifications post-GCD (Converse et al. 1998; Stone and scientific study without endangering the Little Colorado River Gorman 2006), and (3) temperature changes in the main-stem population. If there is similar density dependence in juvenile river due to cold, hypolimnetic releases from GCD (Hamman survival rates of fish transplanted to and spawned in other trib- 1982a; Clarkson and Childs 2000). After a period of decline utaries, extinction risks for these tributaries will be much lower through the 1980s and 1990s, the Little Colorado River popu- than would be predicted from the small adult population sizes lation has shown increased recruitment since at least 2002 and likely to develop in them. now appears to have stabilized at a relatively large population To minimize extinction risks, management agencies have of 6,000–9,000 + adults (Coggins et al. 2006; Coggins and adopted a series of goals and actions to conserve this species, Walters 2009). However, these increases have occurred in con- including translocation of Humpback Chub to tributaries to 628 PINE ET AL.

the Colorado River in Grand Canyon (USFWS 2008, 2011) capacities and potential adult population sizes than the Little and hatchery-based refuge populations (USFWS 2008, 2011). Colorado River, how likely are they to go extinct due to demo- Specifically, the 1995 Biological Opinion on the operation of graphic stochasticity, so as to require continuing translocations GCD included the establishment of a second aggregation of from the Little Colorado River for permanent establishment? Humpback Chub downstream of GCD as a Reasonable and Finally, given 1–3 above, what translocation strategies are most Prudent Alternative, and the 2011 Biological Opinion calls for appropriate to achieve a successful second population? coordinated efforts to “expand the role of tributaries and their ability to contribute to the growth and expansion of main-stem METHODS aggregations” (USBOR 1995; USFWS 2011). Additionally, to address key research needs on early life history of Humpback Individual-based models (IBMs) are widely used in ecology Chub, there is increased interest in allowing permitted take of (DeAngelis and Mooij 2005), often for small populations. Their a very small number of fish for research purposes (i.e., otolith benefits include the following: (1) population attributes (i.e., analyses) as well as better quantifying the risks to the popula- recruitment, mortality) are described with distributions rather tion due to incidental mortalities from sampling (USFWS 2011; than mean values, (2) individual animals are explicitly tracked, Hunt et al. 2012). The Little Colorado River population is seen which is the scale that drives population dynamics, and (3) re- as a potential source of juvenile fish for translocations or per- sults are expressed as the sum of characteristics of the individual mitted take for scientific study. Conservation measures from the survivors (DeAngelis and Rose 1992; Letcher et al. 1998). The 2011 Biological Opinion (USFWS 2011) will require juvenile model allows entry and saving of population dynamics param- translocation of Humpback Chub from the Little Colorado River eters, and “running the model” consists of simulating multiple on an annual basis (Valdez et al. 2000) to augment tributary replicate population histories each with random survival events populations in Grand Canyon and maintain refuge populations. over time for each individual fish. The model also assumes This strategy could have impacts to the source population in random variation in juvenile survival rates associated with en- the Little Colorado River as well as an unknown potential for vironmental factors likely to cause variation in such rates. An success in tributaries targeted for translocations (USFWS 2010, IBM can also represent density-dependent survival patterns of 2011). Thus, using a model to screen risks to the Little Colorado juveniles over multiple juvenile “stanzas” representing distinct River population and inform translocation strategies would ben- ontogenetic habitat-use patterns and juvenile size–age ranges, efit managers in balancing conservation measures with risks to which allows for testing of sensitivity to stanza complexity. Fi- the sustainability of source populations and the likelihood of nally, the IBM can run large numbers of stochastic simulations success of target populations. quickly to allow rapid screening of policy options and sensitiv- For a wide range of management scenarios, quantitative pop- ity analysis of the results to uncertainties about key population ulation viability models can be used to evaluate the likelihood parameters. of populations approaching size or vital rate thresholds deemed critical to continued species persistence (National Research The Individual-Based Population Model Council 1995; McGowan and Ryan 2009, 2010). Quantitative In the model we developed, each individual-based simulation models are beneficial because their founding assumptions are trial begins by constructing a list of No individual female fish, transparent and, owing to their probabilistic origins, the uncer- i = 1, 2, . . . ,No. Each fish i is assigned an initial age ai based tainty of predictions are readily quantified and can be incor- on an assumed initial stable age distribution calculated from porated into the decision-making process (Hilborn and Mangel age-specific survival rates S(a) (with a standard deviation of Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 1997). Despite the applicability associated with this approach, 0.5 on age-0 survival). For each simulated year t = 1,2,.,t, each few examples of ESA-engendered questions or risk assessments individual either becomes a year older or dies between year t and are available in the literature. t + 1 and is removed from the list with probability 1-S(ai). The We developed an individual-based population model for population in year t is just the sum of the surviving population Nt. evaluating extinction risk to the Little Colorado River Hump- These random individual choices result in simple “demographic back Chub population through the collection of individuals for stochasticity” in age-1 and older abundance. ESA-mandated conservation purposes (refuge establishment, Total larvae production Et for each year t is calculated from translocations, permitted take) and also to assess the likelihood the surviving individuals as a sum over individuals of age- that populations can be established in tributaries outside the specific fecundities, i.e., Little Colorado River. Specifically, we use the model to evaluate

the following: (1) What is the current extinction risk for the Lit- Nt tle Colorado River population and is it resilient to a catastrophic Et = F(ai ). (1) event? (2) What proportion of the Little Colorado River juvenile i=1 population, and at what sizes or ages, can be removed without increasing extinction risk to the Little Colorado River population? The age-speciﬁc fecundity relationship F(a) is assumed to be (3) Given that other tributaries may have lower juvenile rearing ﬁxed over time, and is calculated by assuming that fecundity POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 629

is linearly proportional to body weight at age minus weight at data to allow direct estimates of stanza-specific As, Bs,sowe maturity wmat, where weight at age w(a) is assumed to follow specify alternative hypotheses about the distribution of survival the von Bertalanffy relationship and density effects over each life history stanza (Table 1). This ∗ is done by specifying mortality rates M s and density effect −Ka 3 ∗ w (a) = [1 − e ] . (2) B s for each life history stanza relative to each other (Walters and Martell 2004). The As and Bs are calculated from these Here, K is the von Bertalanffy metabolic coefficient. For ages hypothesized relative rates using such that w(a) >wmat, F(a) is given by ln (A) M∗ A = e s (9) s M∗ F (a) = f [w (a) − wmat, (3) s B = ∗ . Bs B s ∗ s−1 (10) where f is an estimate of larvae per unit relative weight. s B s s=0 As The real heart of our simulation is in determining the fates of the Et larvae for each year. Larvae that survive their first These scales and conversion formulas are used to assure that the year are added with age ai = 1 to the list of older fish for by-stanza parameters As, Bs are scaled so that their overall mean the next simulated year. Larvae survival to age-1 is predicted prediction across stanzas follows the overall A, B relationship using a multistanza Beverton–Holt stock–recruitment relation- of equation (4). ship that allows for density dependent survival rates during each “Environmental” variation in maximum survival rates As is stanza and for random “environmental effects” on survival. The simulated by assuming lognormal variation in these rates, so deterministic form of the average larvae to age-1 recruitment that the stanza-specific rate Ast for each year is given by: relationship is assumed to be Wst Ast = As e . (11) AEt N1 = . (4) 1 + BEt Here Wst is a year- and stanza-specific normal random deviate, with mean 0 and standard deviation σs. The assumption of inde- Here, A is maximum survival rate at low larvae densities, and pendent values of these deviates across stanzas ignores possible B represents the strength of density effects that lead to max- correlation across stanzas in impact of factors such as flooding imum possible recruitment A/B at very high Et. As noted by and flow-related changes in habitat area. The population via- Beverton and Holt (1957) in their original derivation, and later bility impacts of such correlations can be simulated simply by by Moussalli and Hilborn (1986), this overall form of relation- increasing the assumed standard deviation of the Wst. ship is maintained when fish in fact pass through multiple juve- Recruitment for each year is simulated in our model by start- = nile stanzas s 1,... ,S, where survival through each stanza ing with the Et individual larvae computed for that year, then follows the same form as equation (4) with maximum survival simulating survival of these individuals through each of the stan- rate and density dependence parameters As,Bs, The overall pa- zas by assuming each live individual at the start of each stanza rameters A,B are then given by has stanza-specific survival rate

S = Ast . = Sst (12) A As (5) 1 + Bs Nst

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 s=1 S s Here, Nst = Et for stanza 1, and for s > 1 is the number of − B = B s As (A0 = 1). (6) individual survivors from a previous stanza (s 1). Each of final s=1 s=0 number of recruiting individuals (i.e., individuals surviving the final stanza S), NSt, is added to the list of live individuals Ni The “compensation ratio” (CR) parameterization (Walters and and assigned age ai = 1, in preparation for beginning the next Martell 2004) is used where simulation year. To simulate removal of juveniles for sampling and transplant CR A = (7) policies, the Ast values were modified as “harvest rates” hst for EPRo each stanza, in which case Ast in equation (12) is multiplied by CR − 1 (1−h ) before applying the modified S to the individuals enter- B = . (8) st st (Ro)(EPRo) ing stanza s. Likewise, to simulate transplants into a population, the stanza entry numbers Nst can be modified upward by year- The CR is the Goodyear compensation ratio (of maximum sur- and stanza-specific stocking rates TPst (absolute number of fish vival rate to survival rate at average size Ro) and EPRo is average transplanted or stocked into the population) before survival rates larvae production per age-1 and older individual. There is rarely for the stanza are applied. 630 PINE ET AL.

TABLE 1. Input parameter estimates and sources for Humpback Chub in Grand Canyon. Input parameter estimates with high uncertainty are marked in bold italics.

Model parameter Estimate Source Initial age-1 and older female 3,000 for Little Colorado River, Population estimates for Little Colorado River population size (No) zero for other tributaries and Colorado River (Van Haverbeke et al. 2011; Coggins and Walters 2009) Age-1 and older female Humpback 3,000 for Little Colorado River Mark–recapture in Little Colorado River (Van Chub carrying capacity estimates 104–156 age 1 and older (small Haverbeke et al. 2011); guessed for smaller (No) capacity) tributaries based on relative habitat area 291–437 age 1 and older (medium capacity) 2,000–2,500 age 1 and older (high capacity) Average long term age-1 recruitment 1,500–2,500 Mark–recapture in Little Colorado River (Van (Ro) Haverbeke et al. 2011); guessed for smaller tributaries based on relative habitat area Recruitment compensation ratio (CR) 2.0 or 4.0 Average from stock–recruitment meta-analyses (Myers et al. 1999; Goodwin et al. 2006) Annual survival rates of age-1 and 0.4 for age 1 Ratios of fish age 1–2 in closed mark–recapture older fish 0.65 for age 2 (Van Haverbeke et al. 2011); ASMR fitting to 0.86 for age 3 and older PIT tag recovery data (Coggins et al. 2006; Coggins and Walters 2009) for age 2 and older VonBertalanffy growth (K) 0.14 Estimated from PIT tag growth rates (Coggins and Walters 2009) Length at maturity–maximum length, 0.5 Gives maturity at 4–5 years (Ryel and Valdez, implies weight at maturity (wmat) 1995) Larvae produced per body weight of 600 Based on single fecundity estimate of 2,523 spawners ( f ) (Hamman 1982a) × 0.1 egg to larvae survival Relative mortality rates of juvenile Larvae <30 mm: 3.0 Assumes M decreases linearly with length ∗ stanzas (M s ) Larvae 30–60 mm: 2.0 (Lorenzen 2000) Juveniles 60–80 mm: 1.0 Large juv. 80–130 mm: 0.8 Relative density dependence in Larvae <30 mm: 5.0 Assumes density effect decreases with increasing ∗ mortality by stanza (B s ) Larvae 30–60 mm: 2.0 size, no direct data Juveniles 60–100 mm: 1.0 Large juveniles 80–130 mm: 1.25

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 Standard deviation of environmental 0.50 per stanza for all of age Standard deviation of log age-1 abundance effect on survival (σe) 0–1.5 estimates from mark–recapture

In summary, the input data consists of the following: and B (EPRo is calculated from the age-fecundity relationship and average survivorship from the S(a) rates), noting (1) an assumed initial population size No and a population size that Ro determines average long-term population size, which estimate at carrying capacity; can differ considerably from the initial population size (2) age-specific but time-independent survival probabilities No; S(a) for recruited (age-1 and older) individuals, assumed (5) stanza-specific relative juvenile survival and density effect ∗ ∗ for parameter entry convenience to be the same for ages parameters M s ,B s ; a = 3 and older; (6) the standard deviation(s) for lognormal environmental ef- σ (3) growth-fecundity parameters K, wmat, and f ; fects on stanza survivals, e; and (4) the recruitment scale and compensation parameters Ro, and (7) management policy parameters, namely the juvenile stanza- CR needed to calculate overall Beverton–Holt parameters A specific harvest–transplant removal rates hst and stocking POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 631

rates TPst (harvest rates of older fish can also be represented annually in the Little Colorado River and adjacent Colorado by reducing the annual survival rates S(a). River main stem since the early 1990s, such that most Hump- back Chub in that tributary are now PIT-tagged by the time Note that changes in the survival rate and management policy they reach 5–7 years of age and they are repeatedly sampled removal parameters can also be used to simulate other changes over their long life of 30 + years with annual survival of about that might affect juvenile survival rates, such as invasions by ex- 85% (Coggins et al. 2006). This sampling program has provided otic warmwater or coldwater predators, emigration of fish from key demographic information on survival rates and recruitment, translocation sites, or harvesting of older fish. Also, No can be along with relatively precise estimates of population size and set very low compared to the expected average long-term popu- trend (Coggins et al. 2006; Coggins and Walters 2009) over this lation size given approximately by Ro/(1–Sadult), to explore the time period. risk of extinction following some catastrophic mortality event that leaves only No survivors. Modeled Scenarios Scenario 1: Extinction risk and recovery of the Little Col- Estimation of Recruitment and Survival Parameters orado River population following a catastrophic mortality Mark–recapture data provide direct estimates of many of the event.—The Little Colorado River population of Humpback model inputs for the Little Colorado River Humpback Chub Chub is the largest known (Meretsky et al. 2000), and likely population, and other inputs used have been approximated or about 73% of all remaining Humpback Chub rangewide (3,300 inferred from published meta-analyses (Table 1). In assem- adults in the upper Colorado River basin versus 6,000–9,000+ bling input parameter estimates, we were surprised at how adults in Grand Canyon; Coggins et al. 2006; Coggins and Wal- low the published estimate of fecundity for Humpback Chub ters 2009) spawn successfully in this one tributary system. Given was (mean = 2,523 eggs/female, range = 330–5,445, N = 8; the critical conservation importance of this one tributary system Hamman 1982a) relative to the closely related Bonytail G. ele- there is concern among management agencies (USBOR 1995; gans (mean = 25,090 eggs, range = 5,850–37,700 eggs, N = USFWS 2008) about some sort of catastrophic event from spills, 5; Hamman 1982b) and Roundtail Chub G. robusta (mean = runoff from mining, or forest fire within the basin (e.g., haz- 18,699 eggs, range = 13,816–26,903 eggs, N = 5; Brouder et al. ardous materials spill from the Highway 89 bridge located 55– 2006). This fecundity estimate is critical, since it determines 65 km upstream). This type of event could cause a singular large the initial number of larvae entering the multistanza density- mortality similar to the sodium hydroxide spill on the Cheaka- dependent calculation and hence the overall juvenile survival mus River in British Columbia (BCME 2006). We estimated rates As needed to result in observed numbers of age-1 recruits. extinction probability by assuming that 95% of the Humpback Increasing assumed fecundity results in reduced As survivals Chub stock was eliminated (reduction in the number of females and a dampened impact of early stanza harvest rates hst. from 3,000 to 150) and then simulated the population recovery There is no direct way to estimate the critical recruitment over a 200-year period 1,000 times using a base CR value of 4 compensation ratio CR from Grand Canyon Humpback Chub and again using the more conservative CR value of 2. population data; the historical data span too narrow a range Scenario 2: Assessing risk to the Little Colorado River donor of adult abundances to allow direct assessment of relative re- population.—Removal of Humpback Chub from the Little Col- cruitment success at very low population densities compared orado River population for any reason must be carefully consid- to juvenile production at higher population size. After re- ered so as to not increase the extinction risk of the population viewing estimates for stream- and lake-rearing Pacific salmon (USFWS 2008). For both research and translocation purposes,

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 (Coho Salmon Oncorhynchus kisutch and Sockeye Salmon On- larvae and juvenile size-classes of Humpback Chub are those corhynchus nerka [Myers et al. 1999]) that undergo similar most often considered for permitted take from the Little Col- ontogenetic habitat shifts and likely changes in predation risk orado River. But at what removal levels would removal increase we used a CR value of either 2 or 4. In our simulations a value the extinction risk to the Little Colorado River population? of 2 would be considered very conservative and a value of 4 Humpback Chub spawn in the Little Colorado River primar- a less conservative approach to compare policy options. The ily in March and April and juveniles first recruit to standard ∗ maximum by-stanza survival rates As implied by the M s values sampling gear in early summer and remain present through- in Table 1 are quite reasonable given the high likelihood that out the summer, fall, and overwinter (Kaeding and Zimmerman mortality rates Ms are strongly size dependent as in other fish 1983; Gorman and Stone 1999). Catch of these juveniles is species (Lorenzen 2000). However, closed (Petersen) mark– highly dependent on river discharge and turbidity (Stone 2010) recapture estimates for age-1 and age-2 juveniles in the Lit- with annual catch ranging from 10s to 1,000s of fish in these tle Colorado River are apparently not strongly correlated (R. size-classes. The Little Colorado River Humpback Chub popu- VanHaverbeke, unpublished), suggesting the possibility of a lation requires recruitment levels of about 2,700–2,900 recruits higher compensation ratio and stronger compensation for older per year to have a basically stable population size. We assessed ∗ juveniles than assumed in the base B s values in Table 1. In- the risk to the Little Colorado River population by modeling tensive mark–recapture programs have been conducted almost removals of 10–50% of the Humpback Chub recruits in four 632 PINE ET AL.

size categories ranging from larvae (<30 mm TL) to three size to determine a carrying capacity scaler/m2 of shoreline habi- categories of juveniles (30–60 mm TL, 60–80 mm TL, and tat. This scaler was then applied to the approximate shoreline 80–130 mm TL). For each of these scenarios we assessed a area of critical habitat in each tributary system to estimate the small removal of 10% of each size-class and a large removal of potential carrying capacity. We used three carrying capacities 50% of each size-class for a 5-year period, assessed the yield (low, medium, and high carrying capacity; Table 1) in tributary of fish in the last year of the removal (number cropped), the streams ranging from about 100–2,500 age-1 and older female extinction risk to the population following removal, and then fish (tributary characteristics summarized in Valdez et al. 2000). monitored the population recovery over an additional 5-year All other assumptions were the same as those calculated for the period. For each cropping level, this 10-year time period was Little Colorado River and identified in Table 1. repeated 1,000 times. We compared the extinction risk in each removal scenario to the base case of no removals (with extinc- RESULTS tion risk = 0) as well as the resulting age-1 and older abundance We found that Humpback Chub populations are resilient with at the end of the 10-year simulation with the base case. We also low extinction risk from catastrophe or permitted take of larvae assessed how sustained removals of 10% of the larvae or 50% or juveniles for research or translocation purposes at proposed of the large juveniles for a 200-year period would impact the levels (e.g., 10s to 100s of fish for stomach content or otolith Little Colorado River population of Humpback Chub by com- analyses). Our results suggest that the Little Colorado River paring these sustained removals to baseline simulations with no population is resilient to some sort of catastrophic event but removals. that recovery following a major mortality event could take 50– Scenario 3: Informing translocation strategies.—Scenario 3 100 years. We also found that the potential benefits in terms of focuses on informing translocation policies by assessing trade- expanding knowledge of juvenile Humpback Chub population offs in size-class and number of fish stocked into a tributary ecology or establishing spawning populations in other tributaries stream from the donor population (USFWS 2008). The goal exceeds the risks to the Little Colorado River population of neg- was to identify stocking policies that have a high probability of ative effects following removals for translocation. Finally, we establishment (>90%, equivalent to a risk of extinction <10%) found that if tributary streams are suitable recipients for translo- over a 50-year time period. This was done by iteratively stocking cated juvenile Humpback Chub from the Little Colorado River varying numbers of fish (females) into the donor streams for donor population, then these populations have a high probably five consecutive years until the probability of establishment of establishment and persistence. over a 50-year time period met the success criteria above. We identified numbers and size-classes of Humpback Chub stocked Scenario 1: Recovery from Catastrophe < that would be necessary to result in an extinction risk of 10% At both the base conservative CR of 2 and a CR of 4 we predict over 50 years with translocations initially occurring for five recovery of Humpback Chub populations following episodic re- consecutive years. We assessed stocking using the same four duction by 95% within approximately 100 years of impact under size-classes identified above. We also evaluated how a higher both scenarios (Figure 1). Risk of extinction within 200 years emigration rate or predation rate of 40% of age-1 translocated was zero for both CR scenarios evaluated and recovery was more fish would affect the establishment rate. rapid at the higher CR = 4 scenarios. As an extreme example, if The recipient streams are other tributaries outside of the Lit- only 5 female fish (5 of 3,000) survived the catastrophic event tle Colorado River (i.e., Havasu, Shinumo, and Bright Angel then extinction risk increases to between 5% and 8% for a CR creeks) that are considerably smaller in discharge than the Little of 4 and about 22–25% for a CR of 2.

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 Colorado River but currently do not support Humpback Chub populations either because of geographic barriers to migration Scenario 2: Extinction Risk to Little Colorado or high abundances of predatory nonnative species (e.g., Rain- River from Removals bow Trout Oncorhynchus mykiss or Brown Trout Salmo trutta We estimated a baseline recruitment level of age-1 fish in Bright Angel Creek, Omana-Smith et al. 2012). Based on our through simulations with no removals. These baseline levels current knowledge of Humpback Chub ecology and knowledge represent the population of recruits without removal for CR val- of the flow, temperature, substrate and invertebrate production ues of 2 and 4, and correspond to about 2,900–3,000 age-1 and (Oberlin et al. 1999) in these streams, they are likely suitable older female fish each year, respectively (Figure 2A). for hosting reproducing Humpback Chub populations. Larvae <30 mm TL.—Our results for a CR value of 2 and a In this scenario, we first had to calculate the potential fe- 10% removal of larvae annually for 5 years results in removal male Humpback Chub carrying capacity of each recipient stream levels of approximately 25,000–26,000 larvae each year and which would provide reasonable upper bounds for population ultimately less than a 1% reduction from the baseline (no regrowth (Table 1). Since these limiting factors are unknown, as moval) level of age-1 and older Humpback Chub after 5 years an approximation we calculated the potential carrying capac- of recovery following removal (Table 2; Figure 2B). As the ity of the Little Colorado River using abundance estimates from removal percentage increased to 50%, this resulted in about mark–recapture data and expert opinion on habitat requirements 127,000–128,000 larvae a year available for harvest during the POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 633

FIGURE 1. Recovery of the Humpback Chub population (age 1 and older) following a 95% reduction in abundance due to a catastrophic event with a recruitment Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 compensation ratio of (A) CR = 2and(B) CR = 4. Red colors indicate higher observation densities and dark blue colors indiciate lower observation densities (indicated in the legend) from the 1,000 replicate simulation trials.

5-year removal period leading to a reduction in age 1 + Hump- (Table 2) ﬁve years postremoval. A higher removal rate of 50% back Chub of 7.6% 5 years postremoval (Figures 2A, 2C). In increased the number of small juveniles removed (7,500–8,000) comparison, a less conservative CR value of 4 would result in and also reduced age-1 and older Humpback Chub population a similar yield but smaller population reduction in age-1 and size by 11.9% 5 years postremoval (Table 2). Using a less con- older Humpback Chub of 4.2% (Table 2; Figure 2E). This result servative CR of 4 removals of 50% led to lower yields (5,000– of a lower predicted impact to age-1 and older Humpback Chub 5,500) and not as high a reduction in age-1 and older abundance populations following removals when a higher CR was assumed postremoval period (11%; Table 2) was consistent across all removal levels assessed (Table 2). Medium juveniles 60–80 mm TL.—Removals of medium- Small juveniles 30–60 mm TL.—Assuming a CR value of 2, sized juveniles annually for 5 years using a CR = 2or4were removals of 10% of the small juveniles annually for 5 years generally similar to the smaller juveniles. Removals of 10% yielded about 1,400–1,500 each year and resulted in an esti- each year for 5 years resulted in 300–350 juveniles for harvest mated reduction in age-1 and older population size of <1% and reduced age-1 and older abundance by about 2% 5 years 634 PINE ET AL.

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 FIGURE 2. Change in Humpback Chub age-1 and older abundance following different cropping scenarios. Panels (A) and (D) represent a “base” case with no cropping and a compensation ratio (A) CR = 2and(D)CR = 4. Panels (B) and (E) represent 10% annual removal of larvae (<30 mm TL) each year for 5 years and a (B) CR = 2and(E)CR = 4 scenario. Panels (C) and (F) represent 50% removals of large juveniles (80–130 mm TL) each year for 5 years with (C) CR = 2and(F)CR = 4. Red colors indicate higher observation densities and dark blue colors indiciate lower observation densities (indicated in the legend) from the 1,000 replicate simulation trials.

postremoval. Increasing the removal rate to 50% per year re- to yield about 150–200 fish each year during the 5-year removal duced populations of age-1 and older Humpback Chub by about period and result in <1% reduction in age-1 and older popu- 14% 5 years postremoval and yield increased to 1,650–1,750 lation size 5 years after removal. Increasing the removal rate juveniles per year during the removal period. Again, a less con- to 50% per year yielded about 800–850 fish for removal and servative CR = 4 reduced the age-1 and older population impacts reduced the population of age-1 and older fish by about 13% of removing medium-sized juveniles (about 12%) and yield of 5 years following the end of removals (Figure 2C). Results were juveniles available to harvest (1,400–1,450 medium-sized juve- similar with a CR = 4 yielding about 750–800 large juvenile niles; Table 2). females available for removal and a reduction in age 1 + abun- Large juveniles 80–130 mm TL.—For a CR = 2 removal of dance by 11% 5 years after the removal period ended (Table 2; 10% of the largest juvenile female Humpback Chub is estimated Figure 2). POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 635

TABLE 2. Scenario 2 results assessing the change in number of age-1 and older Humpback Chub in the Little Colorado River from cropping either 10% or 50% of the available female fish from different size-classes for translocation or research purposes at different recruitment compensation (CR) levels for a 5-year period and then monitoring population recovery for a 5-year period. One thousand simulation trials were run for the 10-year simulation. The yield is the average number of fish available for cropping in the last year of the 5-year removal period. The approximate N is the number of age-1 and older fish at the end of the 5-year recovery period following the 5 years of cropping.

Approximate percent change in age-1+ Approximate and older abundance Percent N age-1 over base value Approximate Extinction Case CR cropped and older range of no cropping yield probability Base 2 0 2,900–3,000 0 0 0 Base 4 0 2,900–3,000 0 0 0 Larvae <30 mm TL 2 10 2,900–2,950 <1% 25,000–26,000 0 Larvae <30 mm TL 2 50 2,700–2,750 7.6% 127,000–128,000 0 Larvae <30 mm TL 4 50 2,800–2,050 4.2% 128,000–129,000 0 Small juveniles 30–60 mm TL 2 10 2,900–2,950 <1% 1,400–1,500 0 Small juveniles 30–60 mm TL 2 50 2,550–2,650 11.9% 7,500–8,000 0 Small juveniles 30–60 mm TL 4 50 2,600–2,650 11.0% 5,000–5,500 0 Medium juveniles 60–80 mm TL 2 10 2,850–2,950 1.7% 300–350 0 Medium juveniles 60–80 mm TL 2 50 2,500–2,600 13.6% 1,650–1,750 0 Medium juveniles 60–80 mm TL 4 50 2,550–2,650 11.9% 1,400–1,450 0 Large juveniles 80–130 mm TL 2 10 2,900–2,950 <1% 150–200 0 Large juveniles 80–130 mm TL 2 50 2,550–2,600 12.7% 800–850 0 Large juveniles 80–130 mm TL 4 50 2,600–2,650 11.0% 750–800 0

Scenario 3: Informing Translocation Strategies provides strict regulation on any purposeful or incidental take We found consistent patterns in our translocation strategy of the species or population of concern that may “cause jeop- evaluation where translocation success is equally likely from a ardy” and alter the likelihood of a species survival or recovery. policy of stocking larger numbers of small female fish (gener- These types of actions are often covered as part of Section 7 ally 1,000–1,200 <30-mm-TL female fish per stream per year consultations under the ESA and included in Biological Opin- for 5 years) or stocking fewer numbers of larger-sized female ions required by the U.S. Fish and Wildlife Service for all listed Humpback Chub (15–30 females >60 mm TL per stream per and threatened species (USFWS 2011). Our model and case year for 5 years) had similar probability of extinction rates history example builds on recommendations from National Re- (<10%) over a 50-year time period and a CR of 2 (Table 3). search Council (NRC 1995) and McGowan and Ryan (2009, At the end of the 50-year simulation period, each stream would 2010) to use population models to evaluate take (incidental be expected to support about 30–90 age-1 and older females, or permitted) effects on listed species because it (1) is based on mathematical and probabilistic predictions, (2) is explicit, Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 which would represent as much as 28% of a small stream with low carrying capacity or as little as 4% of a stream with larger transparent, and readily available for review or reassessment, carrying capacity. This is because larger streams take longer and (3) accounts for uncertainty in predictions and assump- time periods to reach carrying capacity than smaller streams. tions and allows that uncertainty to be incorporated into the decision making framework. This type of framework removes the subjectivity that is often present when making management decisions related to take requests for research or conservation DISCUSSION purposes and ultimately helps screen policies and management Our model and case history analysis with Humpback Chub actions to find the best choices to deliver intended conservation populations in Grand Canyon provides clear guidance on sev- benefits. eral important regulatory and policy commitments. We provide Population viability analysis (PVA) models are very common a transparent framework in which to assess risks to populations in assessing the risk of extinction for many plant and animal pop- from permitted take of an endangered species (McGowan and ulations (Reed et al. 1998, 2002), but these models are not with- Ryan 2010). Endangered and threatened species are also pro- out criticism (Coulson et al. 2001; Ellner et al. 2002). A common vided legal protection under the ESA (as well as similar laws criticism of PVA models is that their ability to accurately assess in other countries, i.e., Species at Risk Act in Canada), which extinction rates is low because available data for most threatened 636 PINE ET AL.

TABLE 3. Minimum numbers of female Humpback Chub by size to be translocated annually for the ﬁrst 5 years only of a 50-year simulation from the Little Colorado River (donor) to a small, medium, and large stream (recipients) before extinction probability is <10% for 1,000 trials. Low and high carrying capacity (as number of age-1 and older females) for each stream was estimated as follows: Little Colorado River = 4,000–9,000; small stream = 104–156; medium stream = 291–437; large stream = 2,000–2,500. Each run is 1,000 simulations for 50 years and a compensation ratio CR of 2 (low) or 4 (high). The number of female ﬁsh stocked, the extinction probability (%), and the estimated number of subadult (age 1 to age 3) and adult (age 4 and older) are presented.

Number female fish stocked (extinction probability) (Estimated final mean number of subadult and adult fish after 50 years) Compensation Size (mm) ratio Small stream Medium stream Large stream <30 Low 1,200 (∼8%) 1,300 (∼9%) 1,000 (∼9%) (subadults: 61) (subadults: 38) (subadults:74) (adults: 21) (adults: 14) (adults: 25) High 1,100 (∼9%) 1,200 (∼9%) 1,100 (∼9%) (subadults: 62) (subadults: 46) (subadults: 82) (adults: 21) (adults: 17) (adults: 27) 30–60 Low 125 (∼9%) 125 (∼9%) 125 (∼9%) (subadults:58) (subadults: 39) (subadults: 79) (adults: 21) (adults: 14) (adults: 26) High 115 (9%) 125 (∼9%) 125 (∼8%) (subadults:60) (subadults:46) (subadults: 82) (adults: 21) (adults: 17) (adults: 27) 60–80 Low 30 (∼9%) 30 (∼9%) 30 (∼9%) (subadults:57) (subadults: 39) (subadults: 81) (adults: 20) (adults: 15) (adults: 27) High 30 (∼8%) 30 (∼9%) 30 (∼9%) (subadults:66) (subadults: 47) (subadults: 84) (adults: 23) (adults: 17) (adults: 27) 80–130 Low 15 (∼9%) 15 (∼8%) 15 (∼8%) (subadults:62) (subadults: 40) (subadults: 82) (adults: 22) (adults: 15) (adults: 27) High 15 (∼7%) 15 (∼7%) 15 (∼8%) (subadults:67) (subadults: 47) (subadults: 89) (adults: 23) (adults: 17) (adults: 29)

species is limited (Ellner et al. 2002). Humpback Chub have to easily assess the full range of possible outcomes in terms of been studied extensively for 20 + years and many key popula- extinction risk (Scenarios 1 and 2) or population establishment tion parameters are directly estimable for this population (Ta- (Scenario 3). Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 ble 1). Our model directly incorporates uncertainty in many of Like all models, our model relied on a variety of key param- these population parameters such as variation in survival, which eters and assumptions that we calculated for Humpback Chub, follows recommendations from Coulson et al. (2001) and others borrowed from similar species, or estimated based on our collec- that PVAmodel input parameters capture the distribution of pos- tive knowledge. We found that even with a compensation ratio sible values. We also present our results in terms of population lower (and thus more conservative) than any measured value for trajectories based on 1,000s of PVA model simulations, each an ﬁsh populations (Myers et al. 1999; Goodwin et al. 2006; Ricard individual population trajectory based on 100s or 1,000s of in- et al. 2012), the Little Colorado River population of Humpback dividual ﬁsh (the IBM) using a set of parameter-starting values Chub is (1) resilient to catastrophic mortality events (Scenario drawn from Table 1. Individual-based models are very widely 1) and (2) unlikely to be put at any higher risk of extinction used in ecology and by design capture variation among individ- under any of the permitted take requests recently proposed ual animals (i.e., growth, fecundity) and the environments they for research purposes (e.g., <100 larvae or small juveniles for encounter (i.e., mortality rates) (DeAngelis and Mooij 2005). otolith collection) or for translocation or refuge development By presenting the distribution of large numbers of individual purposes (600–800 individuals between 30 and 80 mm TL; population trajectories, based on IBM simulations from a large Scenarios 2). Additionally we found that under translocation number of individual animals (Figures 1–3), managers are able scenarios evaluated to three candidate streams, there are large POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 637

differences in the numbers of fish required for low extinction In our model, this compensatory improvement is partitioned probability (<10%) dependent on fish size (Scenario 3). into each of the life stages specified along the Beverton–Holt A key assumption in our model and associated policy curve. Compensation in survival is likely related to the ability recommendations is the role of the compensation ratio CR of juvenile organisms to select habitat that reduced competitive in our population predictions. The CR describes the relative interactions at small spatial and temporal scales. The diversity improvement in juvenile survival at low stock abundance of these “foraging-arenas” (Ahrens et al. 2012) available in na- compared to the stock at natural size. This ratio represents tal habitats is likely to determine the strength of compensation the recruitment compensation potential of the population when abundance declines. At lower juvenile densities, remain- (Goodyear 1980). Higher CR values imply populations that ing individuals, such as Humpback Chub translocated to a re- have higher compensatory juvenile survival at low population cipient stream, have more food resources available, spend less sizes relative to unexploited population sizes. This implies that time searching for food, and as a result have reduced exposure high CR populations are more resilient than populations with to predation. All of these factors may combine to lead to higher low CR values because they have a stronger compensatory survival at lower stock sizes and lower risk of extinction, which response (Walters and Martell 2004). Populations with higher influences the rate of establishment in translocated populations. CR values would thus recover from catastrophe (Scenario 1), be One interesting aspect of our results is informing the size and resilient to permitted take (Scenario 2), or expand to carrying number of Humpback Chub to be translocated to the different capacity quickly when translocated to other streams (Scenario tributary populations. In comparing number of individuals avail- 3). Our use of very conservative CR values of 2 and 4 in our able to be cropped from the Little Colorado River population, modeling scenarios follows recommendations from McGowan from the largest female juveniles (80–130 mm TL) to larvae and Ryan (2010) of considering the role of compensation in (<30 mm TL) we found that stocking densities were nearly assessing the potential impacts of removals on an endangered 30 × different (Table 2). The extinction risk to the Little Col- species. orado River population from removals of larvae or juveniles at Because of high fecundity in most fish populations, the po- the levels calculated in Scenario 3 is negligible. As an example, tential to produce large numbers of eggs and potentially large Table 2 demonstrates that removal of 10% of female larvae from numbers of small juveniles exists even when populations are the Little Colorado River yields at least 25,000–26,000 larvae low. Our simulations used the only published estimates of fe- per year over the 5-year removal period. Results from Scenario cundity for Humpback Chub (Hamman 1982a) of about 2,500 3 in Table 3 however suggest that only about 1,000–1,300 fe- eggs per female. These estimates were based on induced ovu- male larvae need to be stocked per year for a 5-year period to lation and manual stripping of female Humpback Chub in a establish a population with an extinction probability of <10% hatchery setting. There is some evidence to suggest that Hump- in each tributary (assuming similar growth and survival rates back Chub or similar species use a batch spawning strategy as the Little Colorado River and zero emigration from recipient (Johnston and Page 1992) such that the annual fecundity esti- stream, see below). From a practical perspective it is impossi- mates are likely much higher than those reported by Hamman ble to determine the sex on larval or juvenile Humpback Chub (1982a). For example, estimates for the congeneric Bonytail so in application twice the number of larvae listed in Table 3 averaged about 25,000 eggs/fish for slightly larger fish (487– would need to be removed. Even at these higher removal levels, 564 mm TL; Hamman 1982b) and Roundtail Chub averaged the results in Table 2 suggest that the extinction risk to the Lit- about 18,700 eggs/fish for mostly smaller fish (Brouder et al. tle Colorado River donor population would be negligible. This 2006) than the Humpback Chub (355–406 mm) assessed by demonstrates that the recommended stocking level is likely only

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 Hamman (1982a). The length differences alone did not fully about 1% of the female larvae produced each year in the Little account for the intraspecific (and likely the interspecific) differ- Colorado River and removals of this level are very unlikely to ences in eggs per female reported. As such, using the published increase extinction risk or lead to any change in the age-1 and estimates of Hamman (1982a) for Humpback Chub makes our older population size (Table 2). results assessing impacts to the donor population conservative. Our results also suggest that managers should be patient in Beverton and Holt (1957) examined the relation between their expectations for translocated Humpback Chub populations parental spawning stock and subsequent recruitment and mor- to reach carrying capacity. We found that if stocking occurs for tality rates of juvenile fish. These authors noticed that the rela- 5 years and then the population is monitored for an additional tionship between spawning stock (number of adults) was gener- 45 years (total of 50 years), then the expected number of age- ally a poor predictor of recruitment except at low parental stock 1 individuals in the population will still be low, <100 at all sizes and they suggested that this was because juvenile mortal- stocking sizes and levels. This is not unexpected when compared ity rates must decline when fewer eggs are produced. If juvenile to Scenario 1 results (Figure 1) that indicate it would take more mortality rates were not strongly density dependent, then re- than 100 years for populations to recover to carrying capacity cruitment would increase across a range of stock size, which (CR = 2) following our simulated catastrophic mortality event. is not what data from more than 300 different fish stocks show One reason for this long period of time required to reach (Myers et al. 1999; Goodwin et al. 2006; Ricard et al. 2012). carrying capacity is that the population age structure in the 638 PINE ET AL.

FIGURE 3. Translocation scenario where 30 medium (60–80 mm TL) and 30 large (80–130 mm TL) juveniles are stocked into a recipient stream for a 5-year Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 period, and then the stockings are stopped and the population monitored. Note the rapid build-up in population abundance during the 5-year stocking period, a decline in abundance as the population transitions to becoming self-sustaining, and then a slow build-up in population size. Red colors indicate higher observation densities and dark blue colors indiciate lower observation densities (indicated in the legend) from the 1,000 replicate simulation trials.

recipient stream has to mature into a reproducing, self-sustaining the population growth towards carrying capacity. For this newly population. As an example, a manager might choose to translo- founded population to be self-sustaining, enough individuals cate and stock 30 each of two sizes of juveniles (60–80 mm need to survive and spawn to counteract the effects of genetic TL and 80–130 mm TL) for 5 years, in an attempt to accelerate drift and inbreeding. In general, current conservation guidelines establishment of a population in a tributary system. This initial suggest an effective population size (Ne; the number of breeding stocking would lead to a rapid build-up in population size for individuals contributing genetic diversity to future generations) the first 5 years as the translocated fish were released (Figure 3), should be a minimum of Ne = 50 to prevent genetic drift and but natural mortality would slowly remove these translocated inbreeding and an Ne = 500 is needed to allow the population fish leading to declines in the population. At the same time the to adapt to changing environmental conditions and retain its “survivors” from the initial stocking would continue to grow evolutionary potential (Franklin 1980; Soule´ 1987). Ultimately to reach sexual maturity and their progeny would contribute to if the translocated population is not able to recruit to adulthood POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 639

and reproduce then the recipient streams simply become sink natal homing found in many fish stocks (Dittman and Quinn populations only maintained by translocations from the Little 1996; Thorrold et al. 2001) even at very small spatial scales Colorado River. (Stewart et al. 2003). The imprinting process is not understood in Humpback Chub (or many other species) and this likely com- Management Implications plex process may occur at very early ages such that translocated Using a PVA model that explicitly accounts for compensa- fish at an early age may be more likely to have already imprinted tion in survival at different life stages, we demonstrated that the on the donor stream before reaching the recipient stream. Ad- Little Colorado River aggregation of Humpback Chub is robust ditionally, some emigration may occur rapidly following the to removals of up to 50% of the larvae or small juveniles over initial release of translocated individuals (e.g., Shinumo Creek; a 5-year period for translocation or research purposes. At these Spurgeon 2012) and acclimation of translocated individuals to removal levels, the risk of extinction did not increase and gen- the recipient stream may reduce the emigration response or erally within 5 years the population had returned to preremoval potentially predator encounters (Brennan et al. 2006). As an ex- levels. We found that removals of the the smallest, youngest ample, if we translocated 30 60–80-mm-TL juveniles and thirty size-classes (larvae and small juveniles) had the least potential 80–130-mm-TL juveniles into the stream with a small carrying impact on the population of age-1 and older individuals, which capacity of 300 individuals for 5 years, we can expect at the is not surprising given that the highest rates of natural mortal- end of 50 years to have about 120 age-1 and older Humpback ity are in the earliest life stages. As such, the removal of even Chub in this stream at an extinction risk of <1%. However, if 1000s of individuals at these early life stages had basically no we suspect that about 40% of the age-1 fish will either emigrate predicted impact on the age-1 and older population. We also or die at a higher rate than the Little Colorado River (perhaps found that the stocking levels required to have a relatively high from a novel nonnative predator), then we can expect that at the probability of establishing additional spawning aggregations in end of that 50-year time period only 25 age-1 and older Hump- other recipient tributaries was much lower than the removal lev- back Chub to be present. If all translocated fish are susceptible els that would put the Little Colorado River donor population at to some sort of mass emigration or mortality event due to the risk of impairment. donor stream being unsuitable, then the extinction risk in the These results are of interest to resource managers at several donor stream will be very high. This highlights the clear need to levels. First, incidental take assessments for listed species are carefully assess the recipient stream for risks to the transloca- often challenging to assess because of an absence of a frame- tion success from factors such as nonnative predators and also to work to evaluate the risks to the species from the allowed take think about the biology of the fish being translocated from their (McGowan and Ryan 2010). Our results suggest that for Hump- donor stream, especially as it relates to factors with as strong an back Chub, and likely other similar species, the risks to these evolutionary significance as natal imprinting. populations from permitted removals of larvae and small juveniles in the numbers required for translocations or for research ACKNOWLEDGMENTS purposes, such as otolith analyses, are very small. Second, we Funding support for this project was provided by the Bureau identified the trade-offs in terms of fish size and number to be of Reclamation and the National Park Service. We recognize the translocated from the donor Little Colorado River population University of Florida and the U.S. Geological Survey Florida to other tributary systems in an attempt to establish additional Cooperative Fish and Wildlife Research Unit for fiscal and ad- spawning populations. These guidelines should help to design ministrative assistance. future translocation efforts, increasing the likelihood of success

Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 and reducing costs. We found that the success of translocations is likely dependent on the survival and growth rates of fish in REFERENCES recipient streams being similar to the donor Little Colorado Ahrens, R. N. M., C. J. Walters, and V. Christensen. 2012. Foraging arena theory. Fish and Fisheries 13:41–59. River, and that emigration from the recipient stream is low. On- Andersen, M. E., M. W. Ackerman, K. D. Hilwig, A. E. Fuller, and P. D. Alley. going research led by National Park Service fisheries staff as 2010. Evidence of young Humpback Chub overwintering in the mainstem part of pilot translocation studies have documented that prey Colorado River, Marble Canyon, Arizona, USA. Open Fish Science Journal resources, flows, temperature, and substrate at least in Shinumo 3:42–50. and Havasu creeks are likely suitable to support translocated BCME (British Columbia Ministry of Environment). 2006. Incident summary report, CN Rail: August 5, 2005 spill of 40,000 litres of 73% so- Humpback Chub populations (B. Healy, unpublished). A key lution of sodium hydroxide into the Cheakamus River north of Squamish concern however is the presence of nonnative Rainbow Trout British Columbia. BCME, Environmental Protection, Lower Mainland (Shinumo, Bright Angel, and Havasu creeks) and Brown Trout Region, DGI 561257, Surrey. Available: www.env.gov.bc.ca/epd/regions/ (Bright Angel Creek). Removal efforts for Rainbow Trout are lower mainland/special studies/cheakamus/cheakamus.pdf. (January 2013). ongoing in Shinumo and Havasu creeks to reduce predation Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Springer-Science, Dordrecht, The Netherlands. risk for translocated Humpback Chub and their progeny. An Brennan, N. P., M. C. Darcy, and K. M. Leber. 2006. Predator-free enclosures additional concern is the potentially high rate of emigration improve post-release survival of stocked Common Snook. Journal of Exper- from the donor stream by translocated fish because of strong imental Marine Biology and Ecology 335:302–311. 640 PINE ET AL.

Brouder, M. J., D. D. Rogers, and L. D. Avenetti. 2006. Observations on the Grand Canyon. Transactions of the American Fisheries Society 112:577– reproductive biology of Roundtail Chub, Gila robusta, in the upper Verde 594. River, Arizona. Western North American Naturalist 66:260–262. Letcher, B. H., J. A. Priddy, J. R. Walters, and L. B. Crowder. 1998. An Clarkson, R. W., and M. R. Childs. 2000. Temperature effects on hypolimnial- individual-based, spatially explicit simulation model of the population dy- release dams on early life stages of Colorado River basin big-river fishes. namics of the endangered red-cockaded woodpecker, Picoides borealis.Bio- Copeia 2000:402–412. logical Conservation 86:1–14. Coggins, L. G., Jr., W. E. Pine III, C. J. Walters, and S. J. D. Martell. 2006. Lorenzen, K. 2000. Allometry of natural mortality as a basis for assessing opti- Age-structured mark–recapture analysis: a virtual-population-analysis-based mal release size in fish-stocking programmes. Canadian Journal of Fisheries model for analyzing age-structured capture–recapture data. North American and Aquatic Sciences 57:2374–2381. Journal of Fisheries Management 26:201–205. McGowan, C. P., and M. R. Ryan. 2009. A quantitative framework to evalu- Coggins, L. G., Jr., and C. J. Walters. 2009. Abundance trends and status of the ate incidental take and endangered species population viability. Biological Little Colorado River population of Humpback Chub: an update considering Conservation 142:3128–3136. data from 1989–2008. U.S. Geological Survey, Open-File Report 2009–1075, McGowan, C. P., and M. R. Ryan. 2010. Arguments for using population models Reston, Virginia. in incidental take assessments for endangered species. Journal of Fish and Coggins, L. G., Jr., M. D. Yard, and W. E. Pine III. 2011. Nonnative fish Wildlife Management 1:183–188. control in the Colorado River in Grand Canyon, Arizona: an effective program Meretsky, V. J., R. A. Valdez, M. E. Douglas, M. J. Brouder, O. T. Gor- or serendipitous timing? Transactions of the American Fisheries Society man, and P. C. Marsh. 2000. Spatiotemporal variation in length–weight 140:456–470. relationships of endangered Humpback Chub: implications for conserva- Converse, Y. K., C. P. Hawkins, and R. A. Valdez. 1998. Habitat relationships tion and management. Transactions of the American Fisheries Society 129: of subadult Humpback Chub in the Colorado River through Grand Canyon: 419–428. spatial variability and implications of flow regulation. Regulated Rivers: Re- Moussalli, E., and R. Hilborn. 1986. Optimal stock size and harvest rate in search and Management 14:267–284. multistage life history models. Canadian Journal of Fisheries and Aquatic Coulson, T., G. M. Mace, E. Hudson, and H. Possingham. 2001. The use Sciences 43:135–141. and abuse of population viability analysis. Trends in Ecology and Evolu- Myers, R. A., K. G. Bowen, and N. J. Barrowman. 1999. Maximum reproductive tion 16:219–221. rate of fish at low population sizes. Canadian Journal of Fisheries and Aquatic DeAngelis, D. L., and W. M. Mooij. 2005. Individual-based modeling of eco- Sciences 56:2404–2419. logical and evolutionary processes. Annual Review of Ecology, Evolution, NRC (National Research Council). 1995. Science and the endangered species and Systematics 36:147–168. act. National Academy Press, Washington, D.C. DeAngelis, D. L., and K. A. Rose. 1992. Which individual-based approach Oberlin, G. E., J. P. Shannon, and D. W. Blinn. 1999. Watershed influence on is most appropriate for a given problem? Pages 67–87 in D. L. DeAngelis the macroinvertebrate fauna of ten major tributaries of the Colorado River and L. J. Gross, editors. Individual-based models and approaches in ecology: through Grand Canyon, Arizona. Southwestern Naturalist 44:17–30. populations, communities, and ecosystems. Chapman and Hall, New York. Omana-Smith, E. C., B. D. Healy, W. C. Leibfried, and D. P. Whiting. 2012. Dittman, A. H., and T. P. Quinn. 1996. Homing in Pacific salmon: mechanisms Bright Angel Creek trout reduction project: winter 2010–2011 report. Na- and ecological basis. Journal of Experimental Biology 199:83–91. tional Park Service, Natural Resource Technical Report NPS/GRCA/NRTR– Ellner, S. P.,J. Fieberg, D. Ludwig, and C. Wilcox. 2002. Precision of population 2012/646, Fort Collins, Colorado. Available: www.nature.nps.gov/ viability analysis. Conservation Biology 16:258–261. publications/nrpm/nrtr.cfm#2012. (February 2013). Franklin, I. R. 1980. Evolutionary change in small populations. Pages 135– Paukert, C. P., L. G. Coggins Jr., and C. E. Flaccus. 2006. Distribution and 150 in M. E. Soule´ and B. A. Wilcox, editors. Conservation biology: an movement of Humpback Chub in the Colorado River, Grand Canyon, based evolutionary-ecological perspective. Sinauer, Sunderland, Massachusetts. on recaptures. Transactions of the American Fisheries Society 135:539–544. Goodwin, N. B., A. Grant, A. L. Perry, N. K. Dulvy, and J. D. Reynolds. Pulwarty, R. S., and T. S. Melis. 2001. Climate extremes and adaptive man- 2006. Life history correlates of density-dependent recruitment in ma- agement on the Colorado River: lessons from the 1997–1998 ENSO event. rine fishes. Canadian Journal of Fisheries and Aquatic Sciences 63: Journal of Environmental Management 63:307–324. 494–509. Reed, J. M., L. S. Mills, J. B. Dunning Jr., E. S. Menges, K. S. McKelvey, Goodyear, C. P. 1980. Compensation in fish populations. Pages 253–280 in R. Frye, S. R. Beissinger, M. C. Anstett, and P. Miller. 2002. Emerging issues C. H. Hocutt and J. R. Stauffer, editors. Biological monitoring of fish. DC in population viability analysis. Conservation Biology 16:7–19. Heath and Company, Lexington, Massachusetts. Reed, J. M., D. Murphy, and P. F. Brussard. 1998. The efficacy of population Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 Gorman, O. T., and D. M. Stone. 1999. Ecology of spawning Humpback Chub, viability analysis. Wildlife Society Bulletin 26:244–251. Gila cypha, in the Little Colorado River near Grand Canyon, Arizona. Envi- Ricard, D., C. Minto, O. P. Jensen, and J. K. Baum. 2012. Examining ronmental Biology of Fishes 55:115–133. the knowledge base and status of commercially exploited marine species Hamman, R. L. 1982a. Spawning and culture of Humpback Chub. Progressive with the RAM legacy stock assessment database. Fish and Fisheries 13: Fish-Culturist 44:213–216. 380–398. Hamman, R. L. 1982b. Induced spawning and culture of Bonytail Chub. Pro- Ryel, R. J., and R. A. Valdez. 1995. Development of a population model for gressive Fish-Culturist 44:201–203. Humpback Chub (Gila cypha) in Grand Canyon: supplement 5—life history Hilborn, R., and M. Mangel. 1997. The ecological detective: confronting models and ecology of the Humpback Chub (Gila cypha) in the Colorado River, with data. Princeton University Press, Princeton, New Jersey. Grand Canyon, Arizona. U.S. Bureau of Reclamation, Upper Colorado River Hunt, T. A., D. L. Ward, C. R. Propper, and A. C. Gibb. 2012. Effects of capture Region, Salt Lake City, Utah. by trammel net on Colorado River native fishes. Journal of Fish and Wildlife Soule,´ M. E. 1987. Introduction. Pages 1–10 in M. E. Soule,´ editor. Viable Management 3:133–141. populations for conservation. Cambridge University Press, Cambridge, UK. Johnston, C. E., and L. M. Page. 1992. The evolution of complex reproduc- Stewart, I. J., T. P. Quinn, and P. Bentzen. 2003. Evidence for fine-scale native strategies in North American minnows (Cyprinidae). Pages 600–621 in tal homing among island beach spawning Sockeye Salmon, Oncorhynchus R. L. Mayden, editor. Systematics, historical ecology, and North American nerka. Environmental Biology of Fishes 67:77–85. freshwater fishes. Stanford University Press, Stanford, California. Stone, D. M. 2010. Overriding effects of species-specific turbidity thresholds Kaeding, L. R., and M. A. Zimmerman. 1983. Life history and ecology of on hoop-net catch rates of native fishes in the Little Colorado River, Arizona. the Humpback Chub in the Little Colorado and Colorado rivers of the Transactions of the American Fisheries Society 139:1150–1170. POPULATION VIABILITY ANALYSIS OF HUMPBACK CHUB 641

Stone, D. M., and O. T. Gorman. 2006. Ontogenesis of endangered Humpback vices, Phoenix. Available: www.fws.gov/southwest/es/arizona/Documents/ Chub (Gila cypha) in the Little Colorado River, Arizona. American Midland Biol Opin/110112 HFE NNR.pdf. (January 2013). Naturalist 155:123–135. Valdez, R. A., S. W. Carothers, M. E. Douglas, M. Douglas, R. J. Ryel, K. Thorrold, S. R., C. Latkoczy, P. K. Swart, and C. M. Jones. 2001. Natal homing R. Bestgen, and D. L. Wegner. 2000. Research and implementation plan for in a marine fish metapopulation. Science 291:297–299. establishing a second population of Humpback Chub in Grand Canyon. U.S. Trammell, M., B. Healy, E. Omana Smith, and P. Sponholtz. 2012. Humpback Department of the Interior, Grand Canyon Monitoring and Research Center, Chub translocation to Havasu Creek, Grand Canyon National Park: imple- Final Report, Flagstaff, Arizona. mentation and monitoring plan. National Park Service, Natural Resource Valdez,R. A., and W. J. Masslich. 1999. Evidence of reproduction by Humpback Report NPS/GRCA/NRR – 2012/586, Fort Collins, Colorado. Available: Chub in a warm spring of the Colorado River in Grand Canyon, Arizona. www.nature.nps.gov/publications/nrpm/nrr.cfm#2012. (February 2013). Southwestern Naturalist 44:384–387. USBOR (U.S. Bureau of Reclamation). 1995. Operation of Glen Canyon Dam: Van Haverbeke, D. R., D. M. Stone, and M. J. Pillow. 2011. Mark–recapture and final environmental impact statement. USBOR, Upper Colorado River Re- fish monitoring activities in the Little Colorado River in Grand Canyon during gion, Salt Lake City, Utah. 2010. Annual Report to U.S. Geological Survey, Grand Canyon Monitoring USFWS (U.S. Fish and Wildlife Service). 2008. Final biological opinion for and Research Center, Flagstaff, Arizona. the operation of Glen Canyon Dam (February 27, 2008). USFWS, Arizona Walters, C. J., and S. J. D. Martell. 2004. Fisheries ecology and management. Ecological Services, Phoenix. Princeton University Press, Princeton, New Jersey. USFWS (U.S. Fish and Wildlife Service). 2010. A genetic management plan for Webb, R. H., T. S. Melis, and R. A. Valdez. 2002. Observations of environmental captive and translocated endangered Humpback Chub in the lower Colorado change in Grand Canyon, Arizona. U.S. Geological Survey, Water-Resources River basin. USFWS, Dexter National Fish Hatchery and Technology Center, Investigations Report 02-4080, Tucson, Arizona. Dexter, New Mexico. Yard, M. D., L. G. Coggins Jr., C. V.Baxter, G. E. Bennett, and J. Korman. 2011. USFWS (U.S. Fish and Wildlife Service). 2011. Final biological opinion on the Trout piscivory in the Colorado River, Grand Canyon: effects of turbidity, operation of Glen Canyon Dam including high flow experiments and non- temperature, and fish prey availability. Transactions of the American Fisheries native fish control (December 23, 2011). USFWS, Arizona Ecological Ser- Society 140:471–486. Downloaded by [Department Of Fisheries] at 20:13 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:14 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Estimating and Evaluating Mechanisms Related to Walleye Escapement from Rathbun Lake, Iowa Michael J. Weber a , Mark Flammang b & Randall Schultz c a Department of Natural Resource Ecology and Management , Iowa State University , 339 Science Hall II, Ames , Iowa , 50011 , USA b Iowa Department of Natural Resources , 15053 Hatchery Place, Moravia , Iowa , 52571 , USA c Iowa Department of Natural Resources , Chariton Research Station , 24570 U.S. Highway 34, Chariton , Iowa , 50049 , USA Published online: 28 May 2013.

To cite this article: Michael J. Weber , Mark Flammang & Randall Schultz (2013): Estimating and Evaluating Mechanisms Related to Walleye Escapement from Rathbun Lake, Iowa, North American Journal of Fisheries Management, 33:3, 642-651 To link to this article: http://dx.doi.org/10.1080/02755947.2013.788588

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:642–651, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.788588

ARTICLE

Estimating and Evaluating Mechanisms Related to Walleye Escapement from Rathbun Lake, Iowa

Michael J. Weber* Department of Natural Resource Ecology and Management, Iowa State University, 339 Science Hall II, Ames, Iowa 50011, USA Mark Flammang Iowa Department of Natural Resources, 15053 Hatchery Place, Moravia, Iowa 52571, USA Randall Schultz Iowa Department of Natural Resources, Chariton Research Station, 24570 U.S. Highway 34, Chariton, Iowa 50049, USA

Abstract Reservoir fisheries present managers a unique set of challenges. One of the most obvious, yet overlooked, challenges to maintaining sustainable reservoir fisheries is escapement. Yet, little is known about the escapement of reservoir fishes or factors influencing escapement. From June 2009 through March 2011, 4,137 Walleyes Sander vitreus >300 mm were collected in the tailrace of Rathbun Lake, Iowa, individually tagged with visual implant tags, and returned to the reservoir to estimate escapement rates and identify influential factors. During this time, escapement was estimated as (1) the percentage of tagged Walleyes at large that were recaptured in the tailrace and (2) the proportion of tagged to untagged Walleyes present in Rathbun Lake. Tailrace recapture rates during individual sampling events ranged between 0.0% and 2.1% and minimum annual tailrace escapement was estimated as 13.6% (SE, 3%). In spring 2010, 8% (SE, 1%) of the Rathbun Lake Walleye population had been tagged and by spring 2011, 26% (SE, 2%) of the Walleyes were tagged, indicating that they had previously escaped. To understand factors related to escapement, individual fish capture–recapture data were analyzed in Program MARK using a multistate model to estimate apparent survival, detection, and escapement probability. Probability of escapement increased with increasing mean daily discharge and decreased with increasing fish length and release distance from the dam. Variableweights indicated that discharge was the primary factor related to escapement. Escapement probability increased exponentially with daily discharge and doubled as discharge increased from 40 to 60 m3/s. Our results suggest a substantial proportion of the Rathbun Lake Walleye population has been lost recently, due in part to record water releases, making management Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 of this reservoir fishery challenging.

Reservoirs are a prominent feature across much of North Allen et al. 2008). One of the most obvious challenges to America and are constructed for multiple purposes, including, managing reservoir fisheries is escapement, which may result in but not limited to, water regulation, irrigation, and recreation. substantial loss of populations (Lewis et al. 1968; Navarro and Management of these fisheries is confounded by numerous Cauley 1993; Paller et al. 2006). However, escapement is not factors, including diverse stakeholder interests, socioeconomic simply a within-reservoir concern. Reservoir fish communities conflicts, watershed land use, flood control, novel food web are composed largely of nonnative piscivores (Rahel 2000; Eby interactions, habitat alterations, and degraded water quality et al. 2006) and those that escape may threaten downstream (Willis 1986; Stein et al. 1995; Miranda and DeVries 1996; communities through competition, predation, and hybridization

*Corresponding author: [email protected] Received September 6, 2012; accepted March 12, 2013

642 WALLEYE ESCAPEMENT 643

(Hughes 1986; Philipp 1991). Thus, escapement may have important Walleye Sander vitreus fishery, and serves as one of profound population, community, and food web effects in both four broodstock sources for Iowa hatcheries (Flammang 2009; reservoir and downstream ecosystems, shifting paradigms of Mason 2010). Walleyes are native to the drainage, but the Rath- fisheries management (Graeb et al. 2008). bun Lake population has been maintained entirely by stocking Dramatic water level fluctuations are a prominent feature of in recent years (Mitzner 2002). Historically, survival of stocked reservoirs that can produce both positive and negative effects on Walleye (mean = 27,500,000 fry/year; 80,000 ∼200-mm fisheries. Managers are interested in relationships between hy- Walleye/year from 1988 to 2011) has been considered to be drological variables and fish populations in regulated systems the primary determinant of year-class strength (Mitzner 2002) because flow rates may be manipulated to improve fisheries and fish escapement has not been considered to have a sub- (Groen and Schroeder 1978; Sammons and Bettoli 2000). Dis- stantial effect on the population. However, discharge rates from charge rates may not affect escapement of fishes during periods spring through summer 2001 (mean = 33 m3/s) were associated of low flow (Navarro and McCauley 1993) but appear to play an with dramatic declines in Walleye catch per unit effort in fall influential role when a critical discharge threshold is surpassed gill-net samples (80%), spring broodstock catch rates (50%), (Rischbieter 1996, 1998). In some instances, escapement may and angler catch rates (16% of the mean annual catch rate; also be seasonal, with higher rates occurring during the spring Flammang 2009). Additionally, anecdotal information suggests when fish gather on rocky habitat on the dam face for spawn- that recent sustained high release rates from 2009 through 2011 ing (Lewis et al. 1968; Groen and Schroeder 1978; Powell and (Figure 2) has resulted in high escapement and corresponding Spencer 1979). Yet, little formal evaluation exists identifying declines in the population (Flammang 2010). Thus, quantifi- the relationship between reservoir discharge, or other variables, cation of escapement and an understanding of the influential and escapement rates. factors at Rathbun Lake are needed. Rathbun Lake is a 4,450 ha flood control reservoir located Quantifying and understanding factors influencing fish es- on the Chariton River in south-central Iowa (Figure 1). Rath- capement can help reduce escapement and its effects on pop- bun Lake is one of the state’s largest reservoirs, supports an ulations and ecosystems. Walleyes are an important sport fish Downloaded by [Department Of Fisheries] at 20:14 28 May 2013

FIGURE 1. Rathbun Lake, Iowa, depicting the tailrace where Walleyes were collected with electroﬁshing and eight release locations in the reservoir (stars). [Figure available in color online.] 644 WEBER ET AL.

80 600 bounded on both sides with rock rip rap, and is approximately 3 Discharge (m /s) 230 m long, with a rock rip rap border on its lower end 210 m Walleye CPUE (#/hr) 500 Walleye CPUE (#/hr) downstream of the outlet preventing fish movement outside of 60 this area during periods of low flow (<10 m3/s) and limiting fish /s) 3 400 movement during periods of higher flow. Walleye abundance and recruitment in the Chariton River is extremely low (M. 40 300 Flammang, unpublished data). Thus, all Walleyes captured in 200 the tailrace were presumed to have originated from Rathbun Discharge (m 20 Lake. 100 Tailrace sampling.—Most Sander spp. that escape reservoirs remain in the immediate tailwaters below the outlet structure 0 0 (Jernejcic 1986; Spoelstra et al. 2008), suggesting the majority Apr 01 Aug 01 Dec 01 Apr 01 Aug 01 Dec 01 Apr 01 of fish that escaped from the reservoir likely remained in the 2009 2010 2011 immediate downstream area. We conducted 41 sampling events in the tailrace between June 11, 2009, and March 21, 2011, to FIGURE 2. Mean monthly discharge (m3/s) from Rathbun Lake, Iowa, during 2009–2011 (solid line) and catch per unit effort (CPUE) of Walleyes collected collect Walleyes that had escaped the reservoir. The tailrace was in the tailrace (solid bars). Note: sampling did not occur when discharge rates electrofished using an ETS Electrofishing ABP-3 electrofishing exceeded 43 m3/s. unit powered by a 6,500-W generator. Electrofishing (pulsed DC, 15 A, 500 V) was conducted with the current along each side throughout the Midwest and are one of the most widely intro- of the tailrace on consecutive downstream transects. Additional duced fishes to reservoirs (Rahel 2000). Our approach was to transects moved progressively toward the middle of the tailrace. quantify Walleye escapement rates and examine potential influ- Due to safety concerns and declines in electrofishing efficiency, 3 ential factors at Rathbun Lake, Iowa. Walleyes were collected sampling only occurred when discharge was <43 m /s. Effort in the tailrace following escapement, individually tagged, and was the total time required to make a direct path from the outlet returned to the reservoir. Escapement was then quantified using structure to the lower boundary of the study area for all transects three approaches. First, tailrace escapement was estimated as per day. All Walleyes ≥300 mm captured were measured to the the percentage of tagged fish at large that were recaptured in nearest 1 mm total length (TL), weighed to the nearest 1 g, given the tailrace during each sampling event and over a 1-year pe- a pectoral fin clip, and were individually tagged by inserting a riod. Second, in-lake estimates of escapement were calculated visual implant (VI) tag with a specific alphanumeric code just as the proportion of individuals captured during spring and fall below the skin on the lower jaw. Walleyes <300 mm were not reservoir sampling that had been previously tagged. Third, daily VI tagged due to constraints associated with tagging smaller escapement was estimated with individual fish capture history individuals. Thus, all estimates of escapement are in reference data using a multistate model in Program MARK and individual to Walleye >300 mm. All Walleyes collected on a given date (Walleye length, release distance from the dam, number of es- were collectively released into Rathbun Lake at one of eight capements) and environmental (spring spawning aggregations, locations (Figure 1). On March 21, 2011, water levels were too discharge rate) covariates were evaluated to examine factors re- low for electrofishing but permitted the use of a large bag seine lated to escapement. We hypothesized that reservoir discharge (30.5 m long × 1.8 m tall with 10-mm mesh) that was presumed rates would be the primary factor influencing Walleye escape- to have removed all individuals from the tailrace with an eight- Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 ment and result in substantial loss of the population. pass removal effort (i.e., no fish were captured following three consecutive pulls). Based on recaptures of Walleye >300 mm with a fin clip but not a VI tag from October 2, 2009, through METHODS November 15, 2010, annual tag loss was estimated at 11.3%. Study area.—Rathbun Lake dam consists of an earth-filled Tailrace escapement was estimated using two approaches. embankment lined with rip rap, and water discharge is regulated First, tailrace escapement was estimated as the proportion of through a bottom-release outlet structure. The outlet structure tagged Walleyes at large that were recaptured during each of is 24 m from shore, surrounded by rip rap, and consists of two the 40 resampling events. Second, we used tagged individuals gates (3.5 × 2 m) that are 15 m deep with a wing wall on collected during 6 days of sampling (October 2, 5, 6, 12, 15, each side. Discharge rates are managed by the U.S. Army Corps 28, 2009) which served as a finite population from which to es- of Engineers and vary temporally. The current release regime timate annual tailrace escapement. Annual tailrace escapement allows for minimal flows from 0.3 m3/s to a maximum release of was then estimated as the number of tagged fish from Octo- 23 m3/s in the months of April, May, June, September, October, ber 2–28, 2009, (216 fish tagged and released >300 mm) that and November. A maximum release of 34 m3/s is permitted in were recaptured in the tailrace over a 1-year period (November July and August and a maximum release of 43 m3/s is permitted 2, 2009, through November 15, 2010). Both tailrace escape- from December through March. The tailrace is 30 m wide, ment estimates were corrected for tag loss by multiplying the WALLEYE ESCAPEMENT 645

percentage of tagged Walleye recaptured by 11.3% and adding this value back to the original tag recovery rate (Miranda et al. 2002). Both tailrace escapement estimates provided direct information on the number and proportion of fish that had escaped but were considered conservative because they did not account for mortality, emigration, or detection probability of tagged individuals. Rathbun Lake sampling.—Because all Walleyes collected in the tailrace were believed to originate from Rathbun Lake, escapement was also estimated in-lake as the proportion of fish collected in the lake that had been previously tagged. Because one of the assumptions of capture–recapture analysis is equal detection, mortality, and emigration of tagged and nontagged individuals (Burnham et al. 1987), in-lake estimates were considered less biased by these factors than tailrace escapement estimates. Walleye were sampled in the lake each year during spring broodstock collections and fall annual lake surveys. First, Walleyes were sampled during spring broodstock egg col- FIGURE 3. Multistate study design illustrating two locations: the lake (L) and escapement to the tailrace (E). The arrow that loops back to the lake is lections from April 1–10, 2010, and April 5–12, 2011. Adult defined by the probability of an individual being recaptured in the lake on the Walleye were targeted near the dam face during 4-h sets using occasion after release (Si = probability of true survival over the interval i; Fi = multifilament nylon gill nets that were 30.5 m long by 1.8 m probability of remaining in the lake; pk = probability of recapture in the lake deep, composed of 64-mm bar mesh. A total of 360 sets were for each occasion k). The arrow connecting location L and E is the probability completed in 2010 and 288 sets were completed in 2011. In that a fish released in the lake on occasion k–1 is detected in the tailrace at some time between k–1 and k. Modified from Horton et al. (2011). the fall, Walleyes were sampled from October 20–22, 2010, and October 18–20, 2011, during annual lake surveys. Walleyes were sampled with experimental monofilament gill nets con- within brief time periods relative to the time between tagging, sisting of five panels each 7.6 m in length with bar mesh sizes recapture does not affect subsequent survival or recapture, fates of 19, 25, 38, 51, and 64 mm. A total of 15 24-h sets occurred of individuals within and among cohorts are independent, and during each year. Sample sites were randomly selected through- individuals in a cohort have the same survival and recapture out the reservoir and were fixed across years. The percent of probability for each time interval (Burnham et al. 1987). tagged Walleyes during each sampling event was calculated as Two possible locations existed where Walleyes could be de- the number of tagged fish collected divided by the total number tected, the tailrace and the lake (Figure 3). Tagged and released of fish >300 mm collected times 100. In-lake escapement esti- individuals survive with probability Si, are faithful to the lake mates were corrected for tag loss by multiplying the percentage with probability Fi, and are recaptured with the probability pk of tagged fish by 11.3% (annual tag loss) and adding this value during discrete sampling occasions. The probability that an in- back to the original tag recovery rate (Miranda et al. 2002). dividual escaped the lake (1–Fi) and was detected following In-lake escapement estimates were averaged within each of the escapement (fi*) is (1–Fi)fi* (Cormack 1964; Jolly 1965; Seber four primary periods (spring 2011, fall 2011, spring 2012, and 1965). Because Walleyes could not move from the tailrace to

Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 fall 2012). Rathbun Lake, the only escapement probability possible dur- Capture–recapture analysis.—Our third approach to esti- ing i is 1–Fi (the probability of escaping from the lake to the mate escapement utilized live encounter histories consisting tailrace) and Fi (the probability of remaining in the lake): tran- of 41 sampling events and 4,137 individually tagged Walleye sition probability of individuals from the tailrace to the lake >300 mm. Individual encounter histories were analyzed in Pro- was ﬁxed at zero (Horton et al. 2011). When a tagged Walleye gram MARK (White and Burnham 1999) using the live capture was captured in the tailrace, it was censored from the dataset multistate model to generate maximum-likelihood estimates of and entered as a new individual after being released in the lake apparent survival (S), detection probability (p), and transition to allow for the continued use of this individual in the dataset probability (psi; White et al. 2006). The multistate model is without biasing estimates of survival, detection, or escapement an extension of the Cormack-Jolly–Seber live recapture model (W. Kendall, Colorado State University, personal communica- extended for multiple areas. Thus, the use of multistate models tion). A more detailed explanation of the encounter histories and allows simultaneous inference and separation of survival and how they are used in the multistate model to estimate survival, location estimates. Model assumptions include the following: detection probability, and transition probability can be found in tagged individuals are representative of the population to which White and Burnham (1999) and White et al. (2006). inference is made, number of individuals tagged is known, tag- We developed a set of 20 a priori hypotheses to evalu- ging does not affect survival, releases and recaptures are made ate the effects of sampling date, season (spring versus the 646 WEBER ET AL.

remainder of the year), Walleye total length (length), Wall- escaped and were captured three times, and one escaped and eye release distance from the dam (release distance), number was captured four times (Figure 4). of times escaped (trips), and mean daily discharge (discharge) on Walleye escapement. Due to a limited number of recaptures Tailrace and In-Lake Estimates available and large number of parameters that needed to be es- During an individual tailrace sampling event, between 0.0% timated with the full group and time dependent model, survival and 29.2% of the individuals captured >300 mm had been previ- was held constant in all models [S(.)]. Additionally, due to sub- ously tagged (6.2 ± 0.1% [mean ± SE]; Table 1) and 0.0–2.1% stantial differences in sampling effort between Rathbun Lake of the tagged individuals at large were recaptured (0.4 ± 0.0%). and the tailrace, detection probability was modeled as group (g; Cumulatively, the 41 tailrace sampling events recaptured 18% reservoir and tailrace) and time (t) dependent for all models. This of the tagged individuals at large a second time over a 648-d pe- approach provided improved estimates of escapement probabil- riod. Finally, 29 of the 216-tagged-individuals part of the finite ity (psi), the parameter of primary interest. Release distance was population were recaptured in the tailrace over a 1-year period. calculated as a linear distance from the release location down Corrected for tag loss, minimal annual tailrace escapement was the center of the reservoir to the control structure and mean daily estimated at 13.6% (SE, 3%). discharge was calculated as the mean daily discharge rate be- During spring 2010, 1,387 Walleyes >300 mm were cap- tween electrofishing sampling periods in the tailrace. Hypothe- tured in Rathbun Lake of which 120 had previously escaped, ses were stated in model form in Program MARK and compared resulting in an in-lake escapement estimate of 8.2% (SE, 1.3%) using Akaike’s Information Criterion corrected for small sample over 280 d. In the fall of 2010, 9 of the 34 Walleyes captured size (AICc; Burnham and Anderson 1998). Because a reliable in the lake were tagged, resulting in an in-lake escapement esti- goodness-of-fit statistic does not exist for multistate models, mate of 22.5% (SE, 6.1%) over 491 d. During spring 2011, 974 model selection was determined with AICc rather than QAICc of 2,858 Walleyes collected in the lake were tagged, resulting (Conn et al. 2004). Models with lower AICc values were con- in an in-lake escapement estimate of 26.3% (SE, 2.3%) over sidered more parsimonious and closer to the unknown “truth”. 649 d. Finally, 11 of 37 Walleyes captured in the lake during fall A i value of 0–2 provides “substantial”support in explaining 2011 had previously escaped, resulting in an in-lake escapement variation in the given data whereas a value of 4–7 provides “con- estimate of 23.2% (SE, 5.1%) over 855 d. siderably less”support (Burnham and Anderson 1998). Akaike weights (Wi) were also calculated to address potential uncer- Capture–Recapture Data tainty concerning the selection of the top model (Burnham and From our a priori model set, three models received moderate Anderson 1998). Variableweights (Vi) were calculated to exam- to substantial support in explaining variation in Walleye escape- ine the relative importance of each external variable by summing ment (Table 2). The best supported model (i = 0) explaining Wi over all models in which a given variable appeared. After run- escapement was influenced by Walleye length, release distance ning all candidate models, model averaging was used to obtain from the dam, and mean daily discharge. Additionally, the model robust estimates of daily apparent survival, detection probabil- where escapement was related only to mean daily discharge ity, and escapement probability (Burnham and Anderson 1998). received substantial support (i < 2) and the model where es- Model averaging allows parameters that appear in all candidate capement was related to release distance from the dam and mean models to be estimated based on Akaike weights, therefore al- daily discharge was moderately supported (i = 3.04). Param- lowing models with greater weight to provide more information eter weights indicated that mean daily discharge was likely the in predicting parameter values (Burnham and Anderson 1998). most important factor influencing escapement (Vi = 0.99) fol-

Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 Escapement estimates obtained from Program MARK were not lowed by release distance from the dam (Vi = 0.72) and Walleye affected by detection probability, emigration, or mortality but length (Vi = 0.59). The number of previous escapements had only provided a daily estimate of escapement that could not be little influence on escapement probability (Vi = 0.00) with most extrapolated to an annual escapement rate. Walleyes escaping only one time (Figure 4). Model averaging indicated that daily escapement probability was approximately 0.01% and increased with increasing mean daily discharge and decreased with release distance from the dam and Walleye total RESULTS length (Figure 5). Escapement increased exponentially as mean A total of 7,450 Walleyes were captured in the tailrace and daily discharge increased from 8 to 61 m3/s. In addition to dis- returned to the reservoir between June 15, 2009, and March 21, charge, Walleyes released at the upper end of the reservoir had 2011 (Table 1). Of these fish, 4,137 were ≥300 mm and were a 70% lower probability of escaping than those released closest tagged with a VI tag, given a left pectoral fin clip, and returned to to the dam. Finally, a 760-mm Walleye was 32% less likely the lake at one of eight locations. An additional 3,008 Walleyes to escape than a 300-mm Walleye. Models also indicated that <300 mm were returned without an individual tag. Of the 4,137 Walleye daily survival was high during the study (99.9% [SE, individuals implanted with a VI tag and returned to the lake, 0.0004]) and that detection probability was low in both the lake 284 escaped again and were captured in the tailrace twice, nine (0.002–0.044) and the tailrace (<0.001–0.63). WALLEYE ESCAPEMENT 647

TABLE 1. Electroﬁshing effort, total number of Walleyes captured, catch per hour (CPUE), number of Walleyes <300 mm captured, number of new Walleyes >300 mm tagged that day, number of Walleyes >300 mm recaptured that day, percent of Walleyes >300 mm captured that had been tagged corrected for tag loss, and percent of tagged Walleyes at large that were recaptured corrected for tag loss.

Catch Number Fish Tagged ﬁsh Electroﬁshing Number per hour Number >300 mm Number tagged recaptured Date effort (hours) caught (CPUE) <300 mm tagged recaptured (%) (%) Jun 11, 2009 0.60 58 96.7 18 40 0 0.0 0.0 Aug 12, 2009 0.66 38 57.6 0 38 0 0.0 0.0 Oct 2, 2009 0.43 14 32.4 2 12 0 0.0 0.0 Oct 5, 2009 0.56 33 59.2 1 32 0 0.0 0.0 Oct 6, 2009 0.42 16 38.1 3 13 0 0.0 0.0 Oct 12, 2009 0.49 61 125.4 27 34 0 0.0 0.0 Oct 15, 2009 0.64 97 151.4 54 43 0 0.0 0.0 Oct 28, 2009 0.80 160 199.2 76 82 2 2.7 1.1 Nov 2, 2009 0.73 88 120.4 34 54 0 0.0 0.0 Nov 5, 2009 0.62 65 104.3 23 41 1 2.7 0.3 Nov 10, 2009 0.69 71 102.7 21 48 1 2.3 0.3 Nov 19, 2009 0.60 84 140.0 36 44 2 5.1 0.5 Nov 23, 2009 0.77 64 83.5 9 51 1 2.2 0.2 Nov 30, 2009 0.62 39 62.7 12 27 0 0.0 0.0 Dec 14, 2009 0.61 91 149.2 11 78 2 2.9 0.4 Dec 15, 2009 0.90 100 111.1 15 83 1 1.3 0.2 Jan 6, 2010 0.85 126 147.6 22 99 5 5.6 0.8 Jan 12, 2010 0.69 61 88.3 10 50 1 2.2 0.1 Jan 20, 2010 0.62 281 452.0 41 231 9 4.3 1.2 Jan 25, 2010 0.58 99 170.5 38 58 0 0.0 0.0 Feb 5, 2010 0.47 165 353.6 17 140 8 6.4 0.8 Feb 8, 2010 0.77 122 158.3 18 103 1 1.1 0.1 Feb 10, 2010 0.50 52 104.0 12 40 0 0.0 0.0 Mar 9, 2010 0.99 281 306.1 97 175 8 5.1 0.6 Mar 11, 2010 0.45 116 62.2 18 85 13 17.0 0.9 Mar 15, 2010 0.33 23 75.8 8 15 0 0.0 0.0 Mar 31, 2010 0.83 135 109.2 10 99 26 29.2 1.7 Apr 6, 2010 0.83 134 172.8 37 81 15 20.6 0.9 Apr 9, 2010 0.43 96 320.9 62 29 5 19.2 0.3 Apr 10, 2010 0.47 58 110.6 5 46 7 16.9 0.4

Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 Apr 20, 2010 0.67 341 149.3 221 104 16 17.1 0.9 May 14, 2010 1.37 25 244.5 1 20 4 22.3 0.2 Sep 20, 2010 0.58 36 44.8 0 35 1 3.2 0.1 Sep 21, 2010 0.58 127 62.1 13 110 4 4.0 0.2 Nov 15, 2010 0.89 126 137.1 14 105 7 7.4 0.3 Nov 22, 2010 0.50 198 238.0 8 176 14 8.9 0.7 Dec 3, 2010 0.72 144 343.1 21 119 4 3.7 0.2 Dec 13, 2010 1.42 437 307.7 30 400 26 7.2 1.1 Dec 16, 2010 1.40 400 285.7 25 331 25 8.4 0.9 Dec 14, 2011 0.88 209 237.5 15 176 18 11.4 0.6 Mar 21, 2011a NAb 2,579 NA 1923 590 67 12.6 2.1 Mean 0.72 133 156.5 27 101 7 6.2 0.4 Total 27.96 7450 NA 3008 4137 294 NA 18.0

aSeine was used in place of electroﬁshing due to low water levels. bNA = Not available. 648 WEBER ET AL.

TABLE 2. Number of parameters (K), Akaike’s Information Criterion (AICc), i (the difference in AICc values between the best ﬁt model and other models), AICc weight (Wi), and model likelihood for multistate models developed to estimate apparent survival (S), detection probability (p), and escapement probability (psi) for Walleye in Rathbun Lake, Iowa, from 2009 to 2011. Covariates evaluated included group (g; reservoir and tailrace), time (t), Walleye length (length), Walleye release distance from the dam (release distance), average daily discharge (discharge), and number of times previously escaped (trips).

Model Model K AICc i Wi likelihood Deviance S(.) p(g*t) psi(length + release distance + discharge) 26 12814.76 0.00 0.59 1.00 12762.40 S(.) p(g*t) psi(discharge) 27 12816.53 1.77 0.24 0.41 12762.14 S(.) p(g*t) psi(release distance + discharge) 32 12817.80 3.04 0.13 0.22 12753.26 S(Trips) p(g*t) psi(discharge) 28 12820.54 5.78 0.03 0.06 12764.12 S(.) p(g*t) psi(trips + length + release distance) 26 12832.19 17.43 0.00 0.00 12779.83 S(.) p(g*t) psi(release distance) 27 12833.90 19.14 0.00 0.00 12779.51 S(.) p(g*t) psi(length + release distance) 23 12834.22 19.46 0.00 0.00 12787.94 S(.) p(g*t) psi(t) 82 12837.23 22.47 0.00 0.00 12669.67 S(.) p(g*t) psi(trips + discharge) 25 12839.93 25.16 0.00 0.00 12789.59 S(.) p(g*t) psi(trips + length + discharge) 24 12865.97 51.20 0.00 0.00 12817.66 S(.) p(g*t) psi(trips) 23 12866.34 51.58 0.00 0.00 12820.06 S(.) p(g*t) psi(spring) 23 12896.37 81.61 0.00 0.00 12850.09 S(.) p(g*t) psi(trips + length + release distance + discharge) 22 12899.97 85.21 0.00 0.00 12855.71 S(.) p(g*t) psi(length + discharge) 20 12903.56 88.80 0.00 0.00 12863.35 S(.) p(g*t) psi(.) 23 12903.70 88.94 0.00 0.00 12857.42 S(.) p(g*t) psi(trips + release distance) 25 12912.96 98.20 0.00 0.00 12862.63 S(.) p(g*t) psi(trips + length) 19 12913.40 98.64 0.00 0.00 12875.21 S(.) p(g*t) psi(length) 15 12929.18 114.41 0.00 0.00 12899.05 S(.) p(g*t) psi(trips + release distance + discharge) 33 13063.88 249.12 0.00 0.00 12997.31 S(.) p(g) psi(t) 16 13407.10 592.33 0.00 0.00 13374.96

DISCUSSION collected in the tailrace during a 2-year period. Tailrace escape- Our results indicate a substantial proportion of the Wall- ment was estimated at >13% after 1 year and in-lake estimated eye population escaped from Rathbun Lake from June 2009 escapement up to 26% after 21 months. It should be noted that through March 2011. A total of 4,137 Walleyes >300 mm that escapement estimates from ﬁsh recaptured in the tailrace are were presumed to have originated from Rathbun Lake were conservative as it does not account for mortality and low detection probability, indicating escapement is potentially higher than our estimates. Reservoir Walleye population densities are 5000 inversely related to reservoir discharge rates (Willis and Stephen 1987) and escapement of Walleyes often creates important tail-

Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 3,843 race fisheries (Cross 1964; Groen and Schroeder 1978; Jernejcic 4000 1986). However, quantification of escapement rates of fishes is rarely documented. Substantial loss of Walleyes and Saugers 3000 Sander canadensis from reservoirs in South Dakota (∼700,000 age-0 Sander spp. over 24 h; Walburg 1971), Ohio (3–30%; Armbruster 1962; Cross 1964; Erickson and Stevenson 1972; 2000 Spoelstra et al. 2008), and Kansas (60%; Groen and Schroeder

Number of walleye Number 1978) have also been noted and are comparable to escapement 1000 estimates documented here. Escapement of other adult pisci- 284 vores can also be quite high. For instance, >30% (>2,600 fish) 9 1 0 of a population of Largemouth Bass Micropterus salmoides es- 1234caped from a reservoir in southern Illinois (Lewis et al. 1968), whereas escapement of Muskellunge Esox masquinongy from a Number of times escaped Minnesota reservoir was nearly 80% (8 of 10 fish implanted with FIGURE 4. Number of times each of the 4,137 tagged Walleyes emigrated radio transmitters) over a 5-month period (Weiss 2009). Due to from Rathbun Lake to the tailrace during 2009–2011. the negative effects of fish escapement (often apex predators) on WALLEYE ESCAPEMENT 649

0.06 1978; Powell and Spencer 1979). Because Walleyes congre- gate on the rocky dam face at Rathbun Lake during springtime 0.05 spawning activities (Flammang, unpublished data), we hypothesized that Walleye escapement during the spring would also be 0.04 higher than other seasons. However, models indicated that there was little support of this hypothesis. Instead, mean daily dis- 0.03 charge rate was the most inﬂuential parameter affecting Walleye escapement. Similarly, Rainbow Trout Oncorhynchus mykiss 0.02 (McLellan et al. 2008) and age-0 Sander spp. entrainment (Walburg 1971) were also related to several variables associ-

Emigration probability Emigration 0.01 ated with water discharge rates. Previously, discharge rates have 0.00 been suggested to have an influential role in regulating fish es- 0 10203040506070capement once a critical threshold was surpassed (Groen and Mean discharge (cms) Schroeder 1978; Rischbieter 1996, 1998), but discharge may 0.014 have little effect on escapement during periods of lower flow (Navarro and McCauley 1993). Our results support these hy- 0.012 potheses and suggest that escapement increased exponentially as flow rates increased from 8 to 61 m3/s. Maximum allowable 0.010 release rates for Rathbun Lake are 34 m3/s for most of the year with the exception of December through March when discharge 0.008 rates are allowed to increase to 43 m3/s (USACE 1980). How- ever, due to flooding during this study, daily discharge rates 0.006 increased to 85 m3/s from the end of July through the end of Oc- tober 2010, dramatically increasing the probability of Walleye

Emigration probability Emigration 0.004 escapement. 0.002 Fisheries managers often work with water regulatory agencies to manipulate reservoir water levels and discharge rates 200 300 400 500 600 700 800 to enhance fisheries (Willis 1986; Sammons and Bettoli 2000). Walleye length (mm) For example, some reservoirs are managed to have high spring 0.016 Estimate elevations with the goal of increasing spawning and nursery 0.014 95% confidence intervals habitat for fishes and to enhance growth, survival, and recruitment (Ploskey et al. 1996; Sammons et al. 1999; Sammons and 0.012 Bettoli 2000). Similarly, discharge rates may be reduced when 0.010 possible to reduce fish escapement from reservoirs. In several instances, state and federal fisheries management agencies have 0.008 worked collectively with water managers to minimize impacts of discharge on aquatic resources (Groen and Schroeder 1978; 0.006 Wahus 1996). Currently, fisheries managers at Rathbun Lake Emigration probability Emigration Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 0.004 are working with the U.S. Army Corps of Engineers to more effectively control discharge to minimize escapement from the 0.002 reservoir when possible, while maintaining a flow regime that 024681012141618does not negatively impact the native downstream stream fauna. Release distance from dam (km) In addition to release rates, release distance from the dam and Walleye length were also important factors related to es- FIGURE 5. Relationships between average daily discharge (m3/s; top panel), Walleye length (mm; middle panel), and release distance (km; bottom panel) capement probability. Previous work has also shown that older, and escapement probability (mean [dashed line] ± 95% CI [solid lines]) in and presumably larger, fish have lower escapement rates (Lewis Rathbun Lake from 2009 to 2011. et al. 1968; Jernejcic 1986). For instance, in the Yellowstone River juvenile Sauger were more likely to be entrained than reservoir populations and downstream communities, a detailed adults (Hiebert et al. 2000). The Walleye population in Rath- understanding of factors influencing escapement is needed for bun Lake is sustained primary through stocking Walleye fry improved management of these ecosystems. and fingerlings (Mitzner 2002). We did not evaluate escape- Previously, escapement was documented to occur primarily ment of fry or fingerlings and the extrapolation of these results during the spring during periods of spawning aggregations near to Walleyes <300 mm should be done cautiously. However our the discharge area (Lewis et al. 1968; Groen and Schroeder results, in addition to others (Walburg 1971), suggest that small 650 WEBER ET AL.

individuals may experience high rates of escapement. More than ACKNOWLEDGMENTS 3,000 Walleyes <300 mm were captured in the tailrace and re- We thank the staff of the Iowa Department of Natural Re- turned to the reservoir indicating that escapement of smaller sources who assisted with data collection and G. White and B. individuals may be a concern. Thus, managers should stock Kendall for assistance with data analysis. J. Scholly provided Walleyes at the largest possible size and stockings should be GIS support and M. Brown and four anonymous reviewers pro- done as far away from the dam as possible to maximize retention vided comments on a previous version of this manuscript. Fund- in the reservoir. This hypothesis is supported by examination of ing for this project was provided by Iowa fishing license sales. stocking success that found Walleyes stocked at larger sizes had lower mortality, which was composed of both mortality and escapement (Mitzner 2002). Future work should evaluate REFERENCES < Allen, M. S., S. Sammons, and M. J. Maceina, editors. 2008. Balancing fisheries escapement of Walleyes 300 mm to support this hypothesis. management and water uses for impounded river systems. American Fisheries Additionally, recovery efforts that “rescue” fish from tailraces Society, Symposium 62, Bethesda, Maryland. should return them to the reservoir as far away from the dam as Armbruster, D. C. 1962. Observations on the loss of Walleyes over and through possible to maximize retention within the reservoir. Berlin Dam. Ohio Department of Natural Resources, Division of Wildlife, River fish populations have been theorized to be composed Technical Report W-62, Columbus. Burnham, K. P., and D. A. Anderson. 1998. Model selection and inference. of two subpopulations made up of either sedentary or mobile Springer-Verlag, New York. individuals (Funk 1957; Vokoun and Rabeni 2005). Thus, it is Burnham, K. P., D. R. Anderson, G. C. White, C. Brownie, and K. H. Pol- possible that two subpopulations of Walleyes exist in Rathbun lock. 1987. Design and analysis methods for fish survival experiments based Lake: one that is predisposed to disperse and has a higher rate of on release-recapture. American Fisheries Society, Monograph 5, Bethesda, escapement than the other subpopulation that is sedentary and Maryland. Conn, P. B., W. L. Kendall, and M. D. Samuel. 2004. A general model for the remains in the reservoir. We tested the hypothesis that the num- analysis of mark-resight, mark-recapture, and band-recovery data under tag ber of times a Walleye escaped would influence escapement loss. Biometrics 60:900–909. probability but found no evidence to support this hypothesis. Cormack, R. M. 1964. Estimates of survival from the sighting of marked animals. However, only Walleyes that had already escaped the reservoir Biometrika 51:429–438. into the tailrace were tagged during this study. Thus, these in- Cross, J. E. 1964. Walleye distribution and movements in Berlin Reservoir, Ohio. Ohio Department of Natural Resources, Division of Wildlife, Publication dividuals that had already escaped may have had higher rates W-337, Columbus. of escapement than the population in Rathbun Lake, potentially Eby, L. A., W. J. Roach, L. B. Crowder, and J. A. Stanford. 2006. Effects of inflating our escapement estimates. stocking-up freshwater food webs. Trends in Ecology and Evolution 21:576– Minimizing escapement of Walleyes from reservoirs would 584. assist managers in stabilizing and maintaining these impor- Erickson, J., and F. Stevenson. 1972. Evaluation of environmental factors of Ohio reservoirs in relation to the success of Walleye stocking. Ohio Depart- tant fisheries. Even during low to moderate discharge rates ment of Natural Resources, Federal Aid in Sport Fish Restoration, Project 3 (<40 m /s), daily Walleye escapement probability was approx- F-29-R-11, Final Report, Columbus. imately 0.01% resulting in 8,352 Walleyes escaping annually Flammang, M. K. 2009. Creel survey of Lake Rathbun. Iowa Department of at a replacement cost of more than US$82,000 (Southwick Natural Resources, Job Completion Report, Des Moines. and Loftus 2003). In addition, the average Walleye angler har- Flammang, M. K. 2010. Causes and impacts of Walleye out-migration at Rath- bun Lake, Iowa. Iowa Department of Natural Resources, Job Progress Report, vest rate on Rathbun Lake is 0.1 fish/h with the average trip Des Moines. lasting 5 h (Flammang 2009). Therefore, these lost Walleyes Funk, J. L. 1957. Movement of stream fishes in Missouri. Transactions of the could support an additional 16,704 angling trips at a cost of American Fisheries Society 85:39–57.

Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 $23.00 per trip for a total of more than $384,000 (USFW and Graeb, B. D. S., S. R. Chipps, D. W. Willis, J. P. Lott, R. P. Hanten, W. Nelson- USCB 2008). Thus, fish escapement from reservoirs not only Stastny, and J. W. Erickson. 2008. Walleye response to Rainbow Smelt population decline and liberalized angling regulations in a Missouri River reservoir. has ecological consequences but economic consequences as Pages 275–292 in M. S. Allen, S. Sammons, and M. J. Maceina, editors. Bal- well. ancing fisheries management and water uses for impounded river systems. The use of fish barriers may be justified due to the extensive American Fisheries Society, Symposium 62, Bethesda, Maryland. economic cost, potential ecological ramifications, and reduced Groen, C. L., and T. A. Schroeder. 1978. Effects of water level management on angler satisfaction associated with Walleye escapement. The Walleye and other coolwater fishes in Kansas reservoirs. Pages 278–283 in R. L. Kendall, editor. Selected coolwater fishes of North America. American use of behavioral, nonphysical barriers appears to be success- Fisheries Society, Special Publication 11, Bethesda, Maryland. ful at diverting fishes in some instances (McCauley et al. 1996; Hamel, M. J., M. L. Brown, and S. R. Chipps. 2008. Behavioral responses of Hamel et al. 2008). Currently, we are unaware of any existing Rainbow Smelt to in situ strobe lights. North American Journal of Fisheries literature evaluating the response of Walleyes to nonphysical Management 28:394–401. barriers, but such barriers may be useful tools for managers Hiebert, S., R. Wydoski, and T. Parks. 2000. Fish entrainment at the lower Yellowstone diversion dam, Intake Canal, Montana, 1996–1998. U.S. De- seeking to reduce escapement from reservoirs, thereby main- partment of Interior, Bureau of Reclamation, Denver. taining important sport fisheries while protecting downstream Horton, G. E., B. H. Letcher, and W. L. Kendall. 2011. A multistate capture– ecosystems. recapture modeling strategy to separate true survival from permanent WALLEYE ESCAPEMENT 651

emigration for a passive integrated transponder tagged population of stream Rischbieter, D. 1998. Contribution of Frenchman Lake spill to the fishery of Lit- fish. Transactions of the American Fisheries Society 140:320–333. tle Last Chance Creek. California Department of Water Resources, Northern Hughes, N. F. 1986. Changes in the feeding biology of the Nile Perch, Lates District, Red Bluff. niloticus (L.) (Pisces: Centropomidae), in Lake Victoria, East Africa since Sammons, S. M., and P. W. Bettoli. 2000. Population dynamics of a reservoir its introduction in 1960, and its impact on the native fish community of the sport fish community in response to hydrology. North American Journal of Nyanza Gulf. Journal of Fish Biology 29:541–548. Fisheries Management 20:791–800. Jernejcic, F. 1986. Walleye migration through Tygart Dam and angler utiliza- Sammons, S. M., L. G. Dorsey, P. W. Bettoli, and F. C. Fiss. 1999. Effects of tion of the resulting tailwater and lake fisheries. Pages 294–300 in G. E. Hall reservoir hydrology on reproduction by Largemouth Bass and Spotted Bass and M. J. Van Den Avyle, editors. Reservoir fisheries management: strate- in Normandy Reservoir, Tennessee. North American Journal of Fisheries gies for the 80’s. American Fisheries Society, Southern Division, Reservoir Management 19:78–88. Committee, Bethesda, Maryland. Seber, G. A. F. 1965. A note on the multiple-recapture census. Biometrika Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both 52:249–259. death and immigration-stochastic model. Biometrika 52:225–247. Southwick, R. I., and A. J. Loftus, editors. 2003. Investigation and monetary Lewis, W. M., R. Heidinger, and M. Konikoff. 1968. Loss of fishes over the values of fish and freshwater mussel kills. American Fisheries Society, Special drop box spillway of a lake. Transactions of the American Fisheries Society Publication 30, Bethesda, Maryland. 97:492–494. Spoelstra, J. A., R. A. Stein, J. A. Royle, and E. A. Marschall. 2008. Movement Mason, M. 2010. 2010 Fish culture section report. Iowa Department of Natural of reservoir-stocked riverine fish between tailwaters and rivers. Transactions Resources, Fisheries Bureau, Des Moines. of the American Fisheries Society 137:1530–1542. McCauley, D. J., L. Montuori, J. E. Navarro, and A. R. Blystra. 1996. Us- Stein, R. A., D. R. DeVries, and J. M. Dettmers. 1995. Food-web regulation ing strobe lights, air bubble curtains for cost-effective fish diversion. Hydro by a planktivore: exploring the generality of the trophic cascade hypothesis. Review 15:42–51. Canadian Journal of Fisheries and Aquatic Sciences 52:2518–2526. McLellan, H. J., S. G. Hayes, and A. T. Scholz. 2008. Effects of reser- USACE (U.S. Army Corps of Engineers). 1980. Chariton River basin water reg- voir operations on hatchery coastal Rainbow Trout in Lake Roosevelt, ulation manual, Rathbun Lake. USACE, Water Management Section, Kansas Washington. North American Journal of Fisheries Management 28:1201– City, Missouri. 1213. USFWS (U.S. Fish and Wildlife Service), and USCB (U.S. Census Bureau). Miranda, L. E., R. E. Brock, and B. S. Dorr. 2002. Uncertainty of exploita- 2008. 2006 national survey of fishing, hunting, and wildlife-associated recre- tion estimates made from tag returns. North American Journal of Fisheries ation. USFWS/USCB, FHW/06-NAT, Washington, D.C. Management 22:1358–1363. Vokoun, J. C., and C. F. Rabeni. 2005. Home range and space use patterns Miranda, L. E., and D. R. DeVries, editors. 1996. Multidimensional approaches of Flathead Catfish during the summer–fall period in two Missouri streams. to reservoir fisheries management. American Fisheries Society, Symposium Transactions of the American Fisheries Society 134:509–517. 16, Bethesda, Maryland. Wahus, D. J. 1996. U.S. Army Corps of Engineers natural resource manage- Mitzner, L. 2002. Effectiveness of Walleye fry and fingerling stocking in Rath- ment program. Pages 186–189 in L. E. Miranda and D. R. DeVries, editors. bun Lake, Iowa, 1990–1999. North American Journal of Fisheries Manage- Multidimensional approaches to reservoir fisheries management. American ment 22:1007–1013. Fisheries Society, Symposium 16, Bethesda, Maryland. Navarro, J. E., and D. J. McCauley. 1993. Fish escapement from two storage Walburg, C. H. 1971. Loss of young fish in reservoir discharge and year-class reservoirs in Michigan. Rivers 4:36–47. survival, Lewis and Clark Lake, Missouri River. Pages 441–448 in G. E. Paller, M. H., D. E. Fletcher, M. M. Standora, T. B. Grabowski, T. A. Jones, S. A. Hall, editor. Reservoir fisheries and limnology. American Fisheries Society, Dyer, and J. J. Isely. 2006. Emigration of fish from two South Carolina cool- Special Publication 8, Bethesda, Maryland. ing reservoirs. North American Journal of Fisheries Management 26:976– Weiss, J. L. 2009. An evaluation of Muskellunge stocking in a southern Min- 982. nesota impoundment, Lake Zumbo. Minnesota Department of Natural Re- Philipp, D. P. 1991. Genetic implications of introducing Florida Largemouth sources, Division of Fish and Wildlife, Federal Aid in Sport Fish Restoration, Bass, Micropterus salmoides floridanus. Canadian Journal of Fisheries and Completion Report F-29-(R)P-28, St. Paul. Aquatic Sciences 48(Supplement 1):58–65. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation Ploskey, G. R., J. M. Nestler, and W. M. Bivin. 1996. Predicting Black Bass from populations of marked animals. Bird Study 46(Supplement 1):120–139. reproductive success from Bulls Shoals Reservoir hydrology. Pages 422–441 White, G. C., W. L. Kendall, and R. J. Barker. 2006. Multistate survival models in L. E. Miranda and D. R. DeVries, editors. Multidimensional approaches and their extensions in program MARK. Journal of Wildlife Management Downloaded by [Department Of Fisheries] at 20:14 28 May 2013 to reservoir fisheries management. American Fisheries Society, Symposium 70:1521–1529. 16, Bethesda, Maryland. Willis, D. W. 1986. Review of water level management on Kansas reservoirs. Powell, D. H., and S. L. Spencer. 1979. Parallel-bar barrier prevents fish loss Pages 110–114 in G. E. Hall and M. J. Van Den Avyle, editors. Reservoir over spillways. Progressive Fish-Culturist 41:174–175. fisheries management: strategies for the 80’s. American Fisheries Society, Rahel, F. J. 2000. Homogenization of fish faunas across the United States. Southern Division, Reservoir Committee, Bethesda, Maryland. Science 288:854–856. Willis, D. W., and J. L. Stephen. 1987. Relationships between storage ratio and Rischbieter, D. 1996. Emigration of fish from Antelope Reservoir during periods population density, natural recruitment, and stocking success of Walleye in of spill. California Department of Water Resources, Northern District, Red Kansas reservoirs. North American Journal of Fisheries Management 7:279– Bluff. 282. This article was downloaded by: [Department Of Fisheries] On: 28 May 2013, At: 20:16 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Fish Passage through Culverts in Central Michigan Warmwater Streams Andrew S. Briggs a & Tracy L. Galarowicz a a Department of Biology , Central Michigan University , Brooks Hall 217, Mount Pleasant , Michigan , 48859 , USA Published online: 28 May 2013.

To cite this article: Andrew S. Briggs & Tracy L. Galarowicz (2013): Fish Passage through Culverts in Central Michigan Warmwater Streams, North American Journal of Fisheries Management, 33:3, 652-664 To link to this article: http://dx.doi.org/10.1080/02755947.2013.788589

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:652–664, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.788589

ARTICLE

Fish Passage through Culverts in Central Michigan Warmwater Streams

Andrew S. Briggs* and Tracy L. Galarowicz Department of Biology, Central Michigan University, Brooks Hall 217, Mount Pleasant, Michigan 48859, USA

Abstract Road culverts can alter stream flow, remove fish habitat, and limit fish movement. Little is known about the effects of culverts on fish movement in low-order, low-gradient streams in an agricultural setting. Eleven sites on first- and second-order streams in central Michigan were examined for the effects of culvert type (box, bottomless box, pipe arch, and bottomless pipe arch) on fish passage in an agricultural setting. The selected culverts were not obvious barriers to fish passage as none were perched above the stream. Four reaches (two upstream and two downstream of a culvert) were sampled three times in spring, summer, and fall 2011 at each site. Mark–recapture sampling was used to observe fish movement, and the probability of fewer fish being found upstream of a culvert relative to downstream was modeled with logistic regression. Based on recapture movements, limited passage was assumed for at least one fish species during each season at a pipe arch culvert and for one species during summer at a box culvert. The best models from logistic regression revealed that culvert length had the largest effect on the proportion of Creek Chub Semotilus atromaculatus being found upstream out of the variables used. FishXing, a computer program used to predict fish passage through culverts, predicted that five culverts had passage-limiting flows to either Western Blacknose Dace Rhinichthys obtusus or Creek Chub during at least one season. However, all culverts were predicted to have some passable flows during each season (except one culvert during one season). Results suggest that even though culverts may not be obvious barriers, fish passage can still be limited.

Human modifications such as dams, channelization, and road movement (e.g., Warren and Pardew 1998; Gagen and Landrum crossings can have adverse effects on stream ecosystems. Mod- 2000; Burford et al. 2009; Bouska and Paukert 2010). ifications to natural stream hydrology can disrupt or alter the Movements are important for fish as stream conditions

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 transport of sediments in streams by increasing or decreasing vary spatially and temporally. Fish locate preferred habitats or stream velocity and altering the flow pattern (Kesel and Yodis avoid unfavorable conditions (Schlosser and Angermeier 1995; 1992; Ligon et al. 1995; Poff et al. 1997). Dams often decrease Fausch et al. 2002; Jackson 2003), improving survival and peak flows in streams, leading to decreased stream complex- growth (Gowan and Fausch 2002). Movements allow fish to ity and changes in channel pattern, sediment size, and channel reach spawning habitat (Pess et al. 2003), capture prey (Clapp width (Ligon et al. 1995; Friedman et al. 1998). Modifications et al. 1990), or avoid predators (Harvey 1991) and can vary can have large effects on fish by altering or removing their seasonally (Snedden et al. 1999; Bettinger and Bettoli 2004; habitats (Trautman and Gartman 1974; Ligon et al. 1995; Poff Mellina et al. 2005). Furthermore, movements allow fish species et al. 1997), shifting population abundances towards tolerant or to persist, colonize, or recolonize and promote gene flow (Brown generalist fish species (Trautman and Gartman 1974). In many and Kodric-Brown 1977; Meffe and Sheldon 1990; Peterson cases, such as the installation of road crossings, human modifi- and Bayley 1993). If longitudinal connectivity through a stream cations can limit the longitudinal connectivity of streams for fish network is reduced, the risk of extinction by fragmentation is

*Corresponding author: [email protected] Received November 15, 2012; accepted March 12, 2013

652 FISH PASSAGE THROUGH CULVERTS 653

increased (Winston et al. 1991; Park et al. 2008) or the recovery was restricted (Cahoon et al. 2007). In small streams in the of threatened fish assemblages is delayed (Detenbeck et al. Ouachita Mountains of Arkansas, families of fish were affected 1992). Barriers to longitudinal movement can reduce or even differently by different road crossing types (Warren and Pardew extirpate resident and anadromous fish populations (Gibson 1998). et al. 2005). The majority of culvert passage studies in peer-reviewed lit- Road crossings, such as culverts, can reduce longitudinal erature have focused on streams in the western (e.g., Belford connectivity for fish in streams (Warren and Pardew 1998). and Gould 1989; Beechie et al. 1994; Rosenthal 2007; Burford Bridges generally allow streams to keep their natural mean- et al. 2009; Price et al. 2010) or southern (e.g., Warren and der pattern, bank stability, and streambed as they typically are Pardew 1998; Benton et al. 2008; Vander Pluym et al. 2008; not built in the stream channel, thus allowing natural fish move- Norman et al. 2009) United States. However, few studies to date ment (Harper and Quigley 2000). Culverts are road crossings have focused on modified streams in an agricultural setting. that pass under a roadway to allow continuous flow of water. Agricultural low-order streams are important as they provide Culvert designs are diverse, using various shapes and materials crucial habitat to diverse fish assemblages including resident determined by stream width, peak flows, stream gradient, cost, and migrating species along with species unique to the river and whether fish passage is desired (Baker and Votapka 1990; system (Wang et al. 2000; Meyer et al. 2007). However, agri- Ead et al. 2002; Gibson et al. 2005). Culvert design can influ- cultural streams often have reduced riparian zones, higher water ence the rate of fish movement through the culvert (Warren and temperatures, increased nutrients, increased sediment, and de- Pardew 1998; Bouska and Paukert 2010) or reduce fish habitat, graded native fish assemblages (Waite and Carpenter 2000). including loss of stream habitat (e.g., refuge habitat and change Streams without riparian zones typically have lower index-of- in streambed substrate composition), riparian habitat, and pos- biotic-integrity scores and species richness (Stauffer et al. 2000). sible upstream habitat (Harper and Quigley 2000). Previous Nevertheless, agricultural streams have higher index-of-biotic- studies have recommended bottomless culverts over culverts integrity scores and species richness than streams in urbanized with bottoms for both salmonid and invertebrate passage (Carey areas (Wang et al. 2000). and Wagner 1996; Gibson et al. 2005; Resh 2005; Price et al. Our objective was to determine if culvert type influences the 2010) and box culverts over pipe arch culverts for fish passage passage of warmwater fishes in small (first- or second-order) (Warren and Pardew 1998; Bouska and Paukert 2010). central Michigan streams. The movement of marked fish through Culverts typically become barriers to fish passage due to culverts was compared among culvert designs, fish species, and constrictive size and poor installation for fish movement. Con- seasons. Additionally, differences in fish distribution, richness, strictive culverts increase water velocity through the culvert, and total lengths above and below the culverts were analyzed. limiting the upstream passage of fishes that cannot maintain The fish passage results were compared to predictions made by adequate swim speeds (Warren and Pardew 1998; Gagen and FishXing (a computer program that is often used by managers Landrum 2000; Gibson et al. 2005). Some culverts perch or to predict fish passage) to determine if it is a viable option for hang above the stream, limiting passage to fish with adequate managers to predict fish passage in central Michigan streams. leaping abilities (Taylor 2000; Burford et al. 2009; Bouska and Paukert 2010). Shallow water inside culverts, occurring when culverts are set too high or are too wide, can make culverts METHODS impassable to fish that require certain water depths to swim ef- Study sites.—Sampling was conducted in first- and second- fectively (Gibson et al. 2005). Culverts with high slopes often order streams in Isabella and Bay counties, Michigan (Figure 1).

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 limit the passage of fish species as they increase stream veloc- Groundwater and tiled agricultural runoff fed the streams, which ity through the culvert (Carey and Wagner 1996; Bouska and fed larger-order streams that eventually reached Saginaw Bay, Paukert 2010). Fish also can have difficulty passing culverts that Lake Huron. The streams varied in substrate, cover, and amount become clogged with debris (Beechie et al. 1994). of instream and riparian vegetation, but all were small (stream Culvert passage studies have encompassed both salmonid widths ranged from 0.6 to 7.6 m during sampling periods) and (e.g., Belford and Gould 1989; Taylor 2000; Gibson 2005; Sheer highly modified (e.g., channelization, riparian zone conversion and Steel 2006) and warmwater (e.g., Warren and Pardew 1998; to row crops, and dredging). Bouska and Paukert 2010) fish species. Most passage studies Sample sites were selected based on culvert type and acces- have found at least a portion of culverts to be partial or to- sibility to the site. The eleven sites include two bottomless box tal barriers to fish movement (e.g., Warren and Pardew 1998; (BB) culverts, three box (B) culverts, five pipe arch (P) culverts, Taylor 2000; Gibson 2005; Sheer and Steel 2006; Bouska and and one bottomless pipe arch (PB) culvert (Table 1). Although Paukert 2010). However, fish passage can differ by species. For few bottomless box culverts exist in the state of Michigan, two example, the passage of White Sucker Catostomus commer- are located in Isabella and Bay counties. The box and bottom- sonii, Creek Chub Semotilus atromaculatus, and Flathead Chub less box culverts were constructed of concrete while the pipe Platygobio gracilis was not affected by culverts in Montana; arch culverts were constructed of corrugated steel. Reconnais- however, the passage of Longnose Dace Rhinichthys cataractae sance in 2010 during low flows revealed that none of the selected 654 BRIGGS AND GALAROWICZ Downloaded by [Department Of Fisheries] at 20:16 28 May 2013

FIGURE 1. Location of sites sampled in (A) Isabella and (B) Bay counties, Michigan. Names of crossroads are provided. Culvert shape codes are as follows: B = box, BB = bottomless box, P = pipe arch, and PB = bottomless pipe arch. See Table 1 for site descriptions.

sites had hanging or perched culverts or otherwise were obvious distance equal to the length of the culvert was between reach 1 barriers to ﬁsh movement. and reach 2 and between reach 3 and reach 4 at each site. This Field sampling.—Four 25-m stream reaches were sampled was done so movement across natural stream reaches (from at each of the 11 sample sites (Figure 2). Two reaches were lo- reach 1 to reach 2 and from reach 3 to reach 4) could be com- cated downstream of the culvert (reach 1 and reach 2); two were pared to movement across the culvert (reach 2 to reach 3). The upstream of the culvert (reach 3 and reach 4). An unsampled motivation for ﬁsh to move across a culvert and across a natural FISH PASSAGE THROUGH CULVERTS 655

TABLE 1. Name, location (county), stream, and description of the culverts (shape, length, width, and slope) at sites sampled in Michigan. Stream orderand mean stream velocity, width, depth, and temperature (◦C) are given for spring (March 18–May 26), summer (June 14–July 28), and fall (September 11–November 22), 2011. Culvert shape codes are as follows: B = box, BB = bottomless box, P = pipe arch, and PB = bottomless pipe arch. See Figure 1 for map.

Stream velocity in Stream width in Stream depth in Stream temperature Culvert Culvert Culvert ◦ culvert (m/sec) culvert (m) culvert (cm) ( C) Culvert length width slope Stream Site County Stream shape (m) (m) (%) order Spring Summer Fall Spring Summer Fall Spring Summer Fall Spring Summer Fall

B1 Isabella Jordan Creek B 22.2 3.65 0.33 1st 0.11 0.09 0.06 3.65 3.65 3.65 46.53 23.80 16.47 10.7 21.3 9.6 B2 Bay Culver Creek B 12.4 4.35 0.02 2nd 0.11 0.02 0.02 4.35 4.35 4.35 70.05 51.72 51.93 12.3 20.1 8.3 B3 Bay Bradford Creek B 12.3 2.41 0.12 1st 0.07 0.02 0.02 2.54 2.41 2.41 54.52 33.60 34.53 9.7 20.2 11.8 BB1 Isabella John Neff Drain BB 13.9 11.00 0.07 1st 0.37 0.21 0.16 2.54 2.26 1.67 27.32 21.83 16.53 16.8 23.9 9.9 BB2 Bay Culver Creek BB 16.5 9.50 0.04 2nd 0.08 0.03 0.04 9.50 9.50 9.50 44.95 19.92 19.65 12.9 23 10.1 P1 Isabella John Neff Drain P 12.4 3.50 1.18 1st 0.47 0.31 0.24 2.88 2.30 2.10 13.72 9.84 7.47 10.7 23.7 10.4 P2 Isabella John Neff Drain P 15 2.70 0.14 1st 0.14 0.06 0.04 2.42 2.38 2.45 18.67 15.23 15.47 13.8 24.5 10.3 P3 Isabella Lewis Drain P 17.4 3.60 0.81 1st 0.25 0.17 0.16 3.39 3.07 3.10 18.00 9.92 4.60 14.3 18.6 9.4 P4 Isabella Lewis Drain P 20.9 3.50 0.47 1st 0.06 0.02 0.03 3.13 3.14 3.13 31.96 27.40 21.20 10.9 20.4 13.8 P5 Isabella Killenbeck Drain P 21.3 4.00 0.01 2nd 0.17 0.10 0.08 3.78 3.62 3.60 16.36 11.76 9.40 13.6 19.2 13.3 PB1 Isabella Jordan Creek PB 7.1 3.60 0.47 1st 0.12 0.05 0.04 3.04 2.78 2.90 38.60 20.77 24.60 13.5 18.4 9.6

stream reach was assumed to be equal. Sample reaches of 25 m seasons were analyzed as VIE tag retention decreases over time were determined to be appropriate during exploratory sampling (Northwest Marine Technology 2008). in October 2010 since movement directly upstream and down- Site characteristics.—Stream and culvert characteristics stream of culverts was the focus and longer reaches increased above, below, and in the culverts were recorded. Measurements the probability of nearby culverts (closest culvert downstream included the depth of the culvert embedded into the substrate of a study culvert was approximately 560 m away and others (cm), culvert length (m) and width (m), the height and distance were at least 930 m away) or habitats (scouring often effects between corrugations (cm), stream velocity (m/s) inside the cul- the habitat near culverts) influencing fish near the sample sites. vert and in each reach, water depth (cm) in the culvert and other Also, fish in small headwater streams typically do not move far reaches, stream width (cm) inside the culvert and each reach, (Storck and Momot 1981; Mundahl and Ingersoll 1983; Hill the pool depth at the outlet of the culvert (cm), and culvert slope and Grossman 1987). Having multiple culverts on one stream (%). Culvert entrance type (headwall, projecting, mitered, or was determined to be acceptable since this is common in many wing walls), the substrate type, and any obvious obstructions fish passage studies (e.g., Warren and Pardew 1998; Schaefer to fish passage such as excessive debris in the culvert were et al. 2003; Nislow et al. 2011; Pepino´ et al. 2012) and likely noted. Culvert dimensions were measured once, but all other unavoidable in agricultural streams since roads are required for measurements were recorded during each visit. Stage height agriculture. (m) and flow (m/s; 60% of depth in five cross-sectional subsec- Each site was sampled three times per season with at least tions) were recorded within each reach during each visit with a 14 d between visits in spring (March 18–May 26), summer (June Marsh-McBirney Flo-Mate Model 2000. Temperature (◦C) was 14–July 28), and fall (September 11–November 22) 2011 with recorded at each site during each visit using a YSI 650MDS. a Smith-Root LR-24 backpack electrofisher. The spring season FishXing simulation.—Fish and culvert data were used for started shortly after ice was off the streams. The summer and fall FishXing simulations. Prolonged swimming speed (cm/s), burst

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 seasons were started near the summer solstice (June 21) and au- speed (cm/s), and time to exhaustion under prolonged (min) tumnal equinox (September 23), respectively. Two passes were and burst speeds (s) for specific fish species were provided by made at each stream reach; block nets contained fish within the FishXing when available. Western Blacknose Dace Rhinichthys reach during sampling. Captured fish were identified to species, obtusus and Creek Chub were chosen for FishXing simulations weighed (g), and measured (TL; mm). All fish ≥60 mm (except because they were commonly captured. Since limited species Brook Stickleback Culaea inconstans) were tagged using visual data were available in FishXing, a burst speed of 1.48 m/s was implant elastomer (VIE) tags (similar to Gagen and Landrum used for Western Blacknose Dace (Nelson et al. 2008; FishX- 2000 [>40 mm], Cahoon et al. 2007 [>60 mm], and Rosenthal ing did have Blacknose Dace prolonged speed, 0.38 m/s); a 2007 [>60 mm]). Tagged fish received a uniquely colored VIE prolonged speed of 0.44 m/s (Leavy and Bonner 2009) and tag representing the season, date of capture, and stream reach as Goldfish Carassius auratus burst speed (1.37 m/s) from FishX- well as a partial fin clip to aid in detection of recaptures. After ing (similar to Cahoon et al. 2007) were used for Creek Chub. the fish sufficiently recovered from the procedure, they were Culvert data, including culvert shape, height, width, material, released in the center of the stream reach from which they were entrance type, slope, low and high flows through the culvert, captured. Fish were released in slack water or a pool if pos- and tailwater conditions, were recorded in the field and used as sible so they were not immediately swept downstream. Upon simulation inputs. Outputs of FishXing included a fish passage recapture, previous tags were recorded. Only recaptures within summary with a range of passable stream flows, the percent of 656 BRIGGS AND GALAROWICZ

was calculated for each site during each season. Lower species richness upstream of culverts potentially indicates that some species are negatively affected by culverts. The proportion of species richness at each site upstream of the culverts (i.e., upstream species richness/total species richness) was compared among culvert types using an analysis of variance (ANOVA; α = 0.05) for each season. Tukey’s test was used to determine which culvert types differed if the ANOVA found a difference among culvert types (α = 0.05). Analyses were performed in SAS (version 9.1). Proportional movement (Pr) was calculated for the experimental reaches (movement between reaches 2 and 3) and the control reaches (movement between reaches 1 and 2 and between reaches 3 and 4) as follows:

Pr = M/R, (1)

where M is the number of fish that moved and R equals the amount of recaptures in both stream reaches (similar to Bouska and Paukert 2010). Directional movement upstream was calculated by letting M equal the number of fish that moved upstream and R equal the amount of fish initially captured in the downstream reach (Warren and Pardew 1998; Bouska and Paukert 2010). Fish were assumed to have limited passage if the proportion of recaptured fish that moved through the culvert was less than movement through both of the reference reaches (or one of the reference reaches if that was all that was available). Metrics were determined for each species at each site during each season. The proportion of taggable-size fish captured upstream of the culverts (i.e., number ≥60 mm upstream/number ≥60 mm total) was calculated; fewer fish upstream indicates fish have difficulty passing the culvert (Nislow et al. 2011; Pepino´ et al. 2012). Mean total lengths of each species upstream and downstream of the culverts were calculated for each site during each season. Differences between total lengths upstream and downstream of the culverts were tested using a two-sample t-test for each species during each season at each site (α = 0.05). An ANOVAwas used to test for differences in total lengths between

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 the upstream and downstream reaches among culvert types for each species (α = 0.05). Larger fish upstream of a culvert suggests larger fish with better swimming capabilities are passing the culvert. Conversely, smaller fish upstream suggests larger fish cannot pass the culvert due to low water levels. However, fish size upstream and downstream of the culverts also could be influenced by the habitat present. An ANOVA also was used to FIGURE 2. Diagram of the four stream reach locations (R1 = reach 1, R2 = test for differences in recapture rates and species richness among = = reach 2, R3 reach 3, and R4 reach 4) sampled at each sample site. Each seasons with a post hoc Tukey’s test (α = 0.05). Analyses were reach was 25 m in length. The unsampled (US) length of stream was equal to the length of the culvert. Diagram is not drawn to scale. performed in SAS (version 9.1). The probability that an equal or greater proportion of fish would be found upstream of a culvert relative to downstream of passable stream flows, and barrier type (depth, leap, velocity, the culvert was modeled with logistic regression as the propor- and pool depth) present at the flows recorded in the field. tion of fish found upstream related to culvert length, stream Data analysis.—Mean species richness (total number of depth within the culvert, outlet pool depth, stream velocity species captured) upstream and downstream of the culverts within the culvert, proportion of stream area upstream relative FISH PASSAGE THROUGH CULVERTS 657

to stream area downstream, proportion of culvert width relative 14 to bank-full stream width, culvert type, and season. Since the proportion of fish found upstream of culverts was calculated 12 Z for each season, the model could potentially use up to 33 sites (11 sites multiplied by three seasons). A site was assumed to 10 have a lower proportion of fish upstream if there were a higher YZ proportion of fish downstream of the culvert and the standard 8 errors did not overlap. Dummy variables were used to represent 6 culvert type and season since the variables are categorical. A Y Kolmogorov–Smirnov univariate test for normality was used to 4

test for normality of the data. Culvert length, outlet pool depth, Mean Recapture Rate (%) proportion of culvert width relative to bank-full stream width, 2 and stream velocity within the culvert were log transformed to meet assumptions of normality. A Spearman correlation was 0 used to check for correlation among variables and none was Spring Summer Fall found. The best Kullbeck-Liebler model from logistic regres- FIGURE 3. Mean + SE recapture rates from all sites during spring, summer, sion was selected using Akaike information criterion with cor- and fall 2011. Bars containing the same letters (Y or Z) are not statistically rection for small sample sizes (AICC; Burnham and Anderson different (ANOVA: F = 5.53; df = 2, 30; P = 0.009). 2002). All analyses were performed in SAS (version 9.1). Ini- passage was only assumed for Western Blacknose Dace at site tial categories for the culvert type dummy variables were box P5, which had limited upstream movement (proportion of fish culverts, bottomless box culverts, pipe arch culverts, and bot- that moved from reach 2 to reach 3 (Pr → ) = 0, number of fish tomless pipe arch culverts. However, SAS found quasicomplete 2 3 analyzed (n) = 2; Pr → = 0.67, n = 3) and limited bidirectional separation causing a lack of confidence in the models using cul- 1 2 movement through the culvert (Pr ↔ = 0, n = 1; Pr ↔ = 0.6, vert type (Allison 2008). Bottomless box and bottomless pipe 2 3 1 2 n = 5). Differences in species richness upstream and down- arch categories were then combined to remove the quasicom- stream of culverts did not differ among culvert types (Figure 4; plete separation. ANOVA: F = 1.73; df = 3, 29; P = 0.18), and mean species FishXing results were compared to the fish passage results by richness did not differ from summer or fall (ANOVA: F = 0.43; calculating the percentage of times FishXing correctly predicted df = 2, 32; P = 0.65). passage. If FishXing predicted that no passage should occur but fish tagged downstream of the culvert were subsequently cap- 1.2 Spring tured upstream of the culvert, FishXing results were recorded 1.0 Y Y as incorrect for that culvert during that sampling period. Simu- 0.8 Y Y lations were performed for Western Blacknose Dace and Creek 0.6 Chub at each site during each season. 0.4 0.2 0.0 Summer RESULTS 1.0 Y Y Y Spring 0.8 ss:Site species richness ss:Site Z 0.6

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 Sites were sampled March 18–May 26 with an average of = 0.4 28.5 d (range 21–38 d) between the first and second visits 0.2 = and 22.0 d (range 14–30 d) between the second and third vis- 0.0 Fall its. There were no obvious barriers to fish passage at any sites 1.0 Y Y Y during this period. A total of 4,511 fish were captured among all 0.8 Z sites; 1,627 fish were tagged during the first two visits. Western 0.6

Blacknose Dace, Central Mudminnow Umbra limi, Common richne species Upstream 0.4 Shiner Luxilus cornutus, Creek Chub, and White Sucker were 0.2 commonly found. Only 4.2% of all tagged ﬁsh were recaptured 0.0 BBBPPB combining the second and third visits. Spring recapture rates were lower than fall recapture rates at all sites among seasons FIGURE 4. Mean + SE proportion of upstream species richness relative to (Figure 3; ANOVA: F = 5.53; df = 2, 30; P = 0.009). Fish recap- total species richness at each site for each culvert type during spring, summer, tured included Western Blacknose Dace, Central Mudminnow, and fall 2011. Bars containing the same letters within seasons (Y or Z) are not F = = P = Campostoma anomalum statistically different (ANOVA; spring: 1.73; df 3, 29; 0.18; summer: Central Stoneroller , Common Shiner, F = 5.11; df = 3, 29; P = 0.006; fall: F = 4.34; df = 3, 29; P = 0.012). Culvert Creek Chub, Johnny Darter Etheostoma nigrum, Pumpkinseed shapes are designated as follows: B = box culvert, BB = bottomless box culvert, Lepomis gibbosus, and Yellow Perch Perca ﬂavescens. Limited P = pipe arch culvert, and PB = bottomless pipe arch culvert. 658 BRIGGS AND GALAROWICZ

Summer TABLE 2. Top 15 out of 37 Kullback-Liebler models produced from logistic Sites were sampled June 14–July 28 with an average of 16.2 d regression to predict the probability that an equal or greater proportion of Creek Chub will be found upstream of a culvert with their corresponding AIC , i (range = 14–20 d) between the first and second visits and 17.7 d C (delta AIC), and wi (Akaike weight). Data for the models were collected in (range = 15–22 d) between the second and third visits. There spring, summer, and fall from sites in central Michigan. were no obvious barriers to fish passage at any sites during this period. A total of 4,502 fish were captured among all sites; 1,058 fish were tagged during the first two visits. Western Blacknose Variables in model AICc iwi Dace, Central Mudminnow, Common Shiner, Creek Chub, Fat- Culvert length 27.16 0.00 0.16 head Minnow Pimephales promelas, Green Sunfish Lepomis Culvert length and culvert 27.44 0.28 0.14 cyanellus, and Johnny Darter were commonly captured. Sim- width:bank-full stream width ilar to the spring, there were few recaptures (5.3%). Summer Culvert length and velocity in culvert 28.10 0.94 0.10 recapture rates were not different from spring or fall (Figure 3; Culvert length and stream depth in 28.23 1.07 0.09 = = = ANOVA: F 5.53; df 2, 30; P 0.009). Fish recaptured culvert were Western Blacknose Dace, Brown Bullhead Ameiurus neb- Culvert length and culvert type 28.26 1.09 0.09 ulosus, Central Mudminnow, Creek Chub, Fathead Minnow, Culvert length and outlet pool depth 28.55 1.38 0.08 Green Sunfish, Johnny Darter, Pearl Dace Margariscus mar- Culvert length and 28.59 1.43 0.08 garita, Northern Pike Esox lucius, Pirate Perch Aphredoderus upstream:downstream area sayanus, Pumpkinseed, and White Sucker. Limited upstream Culvert length and season 29.58 2.41 0.05 movement through the culvert was assumed for Central Mud- Outlet pool depth and culvert 30.39 3.23 0.03 = = = = minnow at site B1 (Pr2→3 0, n 1; Pr1→2 0.5, n 2) width:bank-full stream width = = = and Creek Chub at site P5 (Pr2→3 0, n 4; Pr1→2 0.17, Culvert width:bank-full stream width 31.45 4.29 0.02 = n 6; no site had limited bidirectional movement). The species Upstream:downstream area 31.57 4.41 0.02 richness upstream of culverts relative to the total species rich- Upstream:downstream area and culvert 31.61 4.44 0.02 ness at the site was lower for bottomless box culverts (Figure 4; width:bank-full stream width = = = ANOVA: F 5.11; df 3, 29; P 0.006), and mean species Upstream:downstream area and season 32.10 4.94 0.01 = richness did not differ from spring or fall (ANOVA: F 0.43; Outlet pool depth 32.12 4.96 0.01 = = df 2, 32; P 0.65). Stream depth in culvert and stream 32.68 5.52 0.01 width:bank-full stream width Fall Sites were sampled September 11–November 22 with an av- Logistic Regression erage of 31.2 d (range = 22–42 d) between the first and second visits and 21.2 d (range = 14–35 d) between the second and third Due to sample size restrictions (number of sites and consis- visits. There were no obvious barriers to fish passage at any sites tency of fish species being captured at sites), models were only during this period. A total of 6,582 fish were captured among all produced for Creek Chub and a maximum of two variables were sites; 1,976 fish were tagged during the first two visits. Western included in the models. Logistic regression produced 37 models Blacknose Dace, Bluegill Lepomis macrochirus, Central Mud- with either one or two variables included for Creek Chub. The minnow, Creek Chub, Green Sunfish, and White Sucker were top model produced included only the culvert length variable (Table 2). The equation for probability of an equal or greater Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 commonly captured. Fall recapture rates (10.3%) were higher than spring but not different from summer (Figure 3; ANOVA: proportion of Creek Chubs found upstream of a culvert relative F = 5.53; df = 2, 30; P = 0.009). Western Blacknose Dace, to downstream was as follows: Bluegill, Brown Bullhead, Central Mudminnow, Creek Chub, ≥ Fathead Minnow, Green Sunfish, Johnny Darter, Largemouth P( proportion of Creek Chubs upstream) . − . . − . Bass Micropterus salmoides, Lepomis spp. hybrids, Northern = e12 18 9 92(log(len))/1 + e12 18 9 92(log(len)), (2) Pike, Pirate Perch, Pumpkinseed, and White Sucker were recaptured. Central Mudminnow had limited bidirectional movement where len equals culvert length (m; Figure 5a). Culvert length at site BB2 (Pr2↔3 = 0.33, n = 6; Pr1↔2 = 0.45, n = 11; no was included in eight models, which were the eight best models limited upstream movement), and Western Blacknose Dace had produced (Table 2). Culvert length had the highest combined limited upstream movement at site P5 (Pr2→3 = 0.33, n = 3; weight of all the variables (Table 3). Pr3→4 = 1, n = 1). Species richness upstream of culverts relative Since the effect of culvert type was the main objective, the to the total species richness at the site was lower for bottomless best model including culvert type was examined. This model box culverts (Figure 4; ANOVA: F = 4.34; df = 3, 29; P = was the fifth best model and had a i less than 2, indicating a 0.012), and mean species richness did not differ from spring or substantial amount of support (Table 2; Burnham and Anderson summer (ANOVA: F = 0.43; df = 2, 32; P = 0.65). 2002). According to this model, box culverts had the highest FISH PASSAGE THROUGH CULVERTS 659

TABLE 3. Combined wi (Akaike weight) of each variable from the 37 FishXing Kullback-Liebler models produced from logistic regression to predict proba- FishXing predicted that five culverts had some passage lim- bility that an equal or greater proportion of Creek Chub will be found upstream of a culvert in central Michigan. iting flows to Westerm Blacknose Dace and Creek Chub during at least one season (Table 4). One pipe arch culvert (P5) was Variable Combined wi predicted to have passage limiting flows for both species during each season and another pipe arch (P3) had passage limiting Culvert length 0.79 flows for Western Blacknose Dace during each season. How- Culvert width:bank-full stream width 0.24 ever, all the culverts except P3 had measured flows that were Upstream:downsteam area 0.15 passable flows at some point during each season and thus were Outlet pool depth 0.15 not total barriers. None of the flows measured in the pipe arch Velocity in culvert 0.13 culvert P3 during spring were predicted to allow fish passage. Stream depth in culvert 0.13 The sites that were assumed to have limited upstream passage Culvert type 0.11 based on recapture movements also had some flows that were Season 0.08 barriers to fish passage as predicted by FishXing. Upstream passage through the culvert by Western Blacknose Dace at site P3 during spring and by Creek Chub at site P5 during spring probability of allowing equal or greater proportions of Creek was observed even though FishXing predicted that each culvert Chub upstream. Pipe arch and all bottomless type culverts was a barrier at some or all flows. However, FishXing was ranked second and third, respectively, in probability of allowing only assumed to be incorrect at site P3. This was because the equal or greater proportion of Creek Chub upstream (Figure 5b). exact flow when the Creek Chub passed the culvert at site P5 was unknown. For all the other sites and seasons that FishXing 1.2 a. predicted flows that were barriers to fish passage, there were either no recaptures or no detected movement through reference 1.0 reaches or the culverts. Therefore, FishXing was never classified as incorrect at these sites. 0.8

0.6 DISCUSSION Culverts differentially affected ﬁsh species with some be- 0.4 ing partial or temporary barriers without visually being obvious barriers to movement. Culverts in central Michigan streams

0.2 affected upstream passage of Western Blacknose Dace, Cen- tral Mudminnow, and Creek Chub. Species richness upstream of the culverts was lower than downstream, suggesting some 0.0 b. ortion of creek chub upstream Box culverts species are affected more than others. However, it is possible Pipe arch culverts that differences in species richness could be due to habitat differ- 1.0 Bottomless culverts ences upstream and downstream of culverts (Osborne and Wiley 1992; Poff and Allan 1995; Lammert and Allan 1999). Sites in 0.8

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 the Ouachita River, Arkansas (Gagen and Landrum 2000), and the Monongahela National Forest, West Virginia (Nislow et al. 0.6 2011), also had lower species richness upstream of culverts rel-

Probability of = or > prop Probability ative to downstream. Culverts also differentially affected ﬁsh 0.4 species in Montana (Cahoon et al. 2007) and families in the Ouachita Mountains, Arkansas (Warren and Pardew 1998). 0.2 Although culverts may affect longitudinal movement without being obvious barriers, culvert type did not have a major

0.0 role inﬂuencing this longitudinal movement. However, general 0 5 10 15 20 25 30 trends of the best model including culvert type indicated that box Culvert length (m) culverts were most effective and bottomless culverts were least effective for allowing an equal or greater proportion of Creek FIGURE 5. Probability of an equal or greater proportion of Creek Chub up- Chub upstream. In previous studies, box culverts were twice as stream of culverts in central Michigan, based on models created by logistic likely to allow passage of cyprinids as low water crossings in regression. The (a) best model based on AICC criteria that includes only the culvert length variable and the (b) best model based on AICC criteria that Great Plains streams (Bouska and Paukert 2010) and allowed includes the culvert type variable (culvert length also is included in the model). similar or higher movement rates for ﬁsh as natural reaches in 660 BRIGGS AND GALAROWICZ

TABLE 4. Discharges (m3/s) measured and passable discharges (m3/s) and corresponding barriers found by the FishXing computer program for the 11 central Michigan culverts during spring, summer, and fall 2011. Culvert shapes are as follows: B = box culvert, BB = bottomless box culvert, P = pipe arch culvert, and PB = bottomless pipe arch culvert. See Table 1 for site descriptions.

Western Blacknose Dace Creek Chub Discharge Passable discharge Passable discharge Site Season range (m3/s) range (m3/s) Barrier type range (m3/s) Barrier type Spring 0.11–0.25 0.11–0.12 Velocity 0.11–0.14 Velocity B1 Summer 0.02–0.51 0.02–0.15 Velocity 0.02–0.18 Velocity Fall 0.03–0.08 All None All None Spring 0.07–0.68 All None All None B2 Summer 0.02–0.18 All None All None Fall 0.02–0.08 All None All None Spring 0.08–0.20 All None All None B3 Summer 0.01–0.04 All None All None Fall 0.02–0.03 All None All None Spring 0.10–0.62 All None All None BB1 Summer 0.03–0.37 All None All None Fall 0.04–0.06 All None All None Spring 0.09–0.84 All None All None BB2 Summer 0.02–0.22 All None All None Fall 0.01–0.14 All None All None Spring 0.10–0.34 0.10–0.16 Velocity 0.01–0.17 Velocity P1 Summer 0.03–0.36 0.03–0.11 Velocity 0.03–0.11 Velocity Fall 0.02–0.05 All None All None Spring 0.03–0.14 0.03–0.07 Velocity 0.03–0.10 Velocity P2 Summer 0.01–0.10 0.01–0.09 Velocity All None Fall 0.01–0.02 All None All None Spring 0.03–0.54 None Velocity 0.03–0.04 Velocity P3 Summer 0.01–0.31 0.01–0.02 Velocity 0.01–0.04 Velocity Fall 0.01–0.04 0.01–0.02 Velocity 0.01–0.04 Velocity Spring 0.04–0.09 All None All None P4 Summer 0.01–0.03 All None All None Fall 0.01–0.03 All None All None Spring 0.04–0.28 0.04–0.08 Velocity 0.04–0.11 Velocity P5 Summer 0.01–0.30 0.01–0.08 Velocity 0.01–0.12 Velocity Fall 0.00–0.16 0.00–0.08 Velocity 0.00–0.11 Velocity Spring 0.07–0.27 All None All None Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 PB1 Summer 0.02–0.07 All None All None Fall 0.02–0.05 All None All None

Arkansas streams (Warren and Pardew 1998). Bottomless cul- The bottomless box culvert at site BB1 had riprap (concrete verts allowing a lower proportion of Creek Chub upstream was and boulders along the shoreline) to prevent bank erosion and unexpected since they could potentially contain favorable habi- a deep plunge pool downstream of the culvert. These features tat due to natural substrate. However, there were comparatively were preferred habitat for Creek Chub (Fraser and Sise 1980) fewer bottomless culverts (n = 6) used to create the model than and may have played a role in the higher proportion of Creek culverts with bottoms (n = 16). Including a few culverts of each Chub downstream. Riprap has lead to increased diversity or design is common for mark–recapture studies examining ﬁsh species richness in both lake and stream ecosystems (Jennings passage (e.g., Warren and Pardew 1998 [1, 2, or 4 of each de- et al. 1999; Eros˜ et al. 2008). However, the deep plunge pool sign]; Benton 2008 [2 of each design]; Vander Pluym et al. 2008 could be indicative of high stream velocity through the culvert. [3 of each design]; Norman et al. 2009 [1 or 2 of each design]) Culvert length was an important factor in allowing upstream due to the intensity and time required for the ﬁeld sampling. passage of Creek Chub in an agricultural setting. This is unique FISH PASSAGE THROUGH CULVERTS 661

since previous studies have not focused on culvert length as a fewer culverts had passage-limiting flows in the summer than major factor (Warren and Pardew 1998; Benton 2008; Norman in the spring and in the fall than in the spring and summer. This 2009). However, in northeastern Kansas streams culvert dimen- was likely due to the lower stream velocities during summer sions (a combination of culvert slope, length, and width) did and fall, as barriers predicted by FishXing were always velocity impact fish movement (Bouska and Paukert 2010). The recap- barriers. ture data from this study support culvert length as an important FishXing can be a useful tool in predicting fish passage as it factor since the two culverts that were assumed to limit upstream allows users to analyze many culverts with little data collection fish passage (B1 and P5) were the two longest culverts in the (Cahoon et al. 2005). However, FishXing predictions do not study. The importance of culvert length could be a behavioral always support actual fish passage results and often over predict response by Creek Chub. However, factors such as turbulence culverts as barriers (Cahoon et al. 2005; Cahoon et al. 2007). In could play a role in Creek Chub swimming ability and culvert this study, FishXing was only classified as incorrect at one site passage (Webb 2004; Tritico and Cotel 2010). during one season, due in part to low recapture and movement Since none of the culverts were perched or set at high slopes rates in many instances. In addition, for sites with recaptured (>2%; Coffman 2005), Creek Chub behavior could influence fish and FishXing-predicted passage-limiting flows (but not a movement through culverts. Adult Creek Chub prefer to re- total barrier), the stream velocity at the time the fish passed side in pools with large cover including boulders, branches, and the culvert was unknown. FishXing may be a suitable option overhanging stream banks (Fraser and Sise 1980). Since the for predicting fish passage in low-gradient agricultural streams, sampled culverts did not have pools and lacked cover, Creek where biological data are not often collected. Although swim Chub may have perceived the longer culverts as too far of a speeds for warmwater fishes, especially burst swim speeds, are distance to travel without the presence of pools and cover for lacking, FishXing is free to download and predictions made by refuge or their perceptual range is not large enough to real- it are useful as long as users consider that the predictions may ize the presence of potential habitat across culverts. Similarly, be conservative. Longfin Damselfish Stegastes diencaeus choose homing paths This study has implications for larger-scale fish movement that detour around sand barriers instead of swimming across the in streams, specifically highly modified agricultural streams. sand barrier to potential habitat in coral reefs (Turgeon et al. Recapture rates were low over the course of this study; however, 2010). Since most studies examining the perception of barri- low recapture rates and high turnover rates do not imply fish ers focus on terrestrial ecosystems (e.g., Bakker and Van Vuren are moving far (Rodr´ıguez 2002). Highly modified, agricultural 2004; McDonald and St. Clair 2004; Gillies and St. Clair 2008), streams may not have been considered with the idea of high the perception of barriers by fish is relatively unknown. Other turnover rates not implying long movements or the proposal species may have different perceptual ranges due to size, swim- of restricted movement by Gerking (1959). These streams are ming abilities, or habitat preferences (Lima and Zollner 1996; channelized, often dredged, and have highly fluctuating flows Bakker and Van Vuren 2004; McDonald and St. Clair 2004; due to the tiling of crop fields (especially during spring after Turgeon et al. 2010). The perceptual range could lead to dif- snow melt). Channelization and dredging reduce the presence ferent factors being important to other species or culverts in of pools, woody debris, and other refuge for fish when stream agricultural streams not being a barrier entirely. For example, a velocity increases after rain events or snow melt (Gorman and culvert lacking cover such as boulders or woody debris may be Karr 1978). With such low recapture rates, it is unlikely that all a barrier for Creek Chub whereas Johnny Darter may not find tagged fish not recaptured were just outside the sample area, in this culvert to be a barrier as they prefer slow moving streams the reference reaches, or inside the culverts. The fish may have

Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 with sandy substrate (Hubbs et al. 2004). increased their home ranges in search of suitable habitat during Season had little influence on the proportion of Creek Chub periods of high flows (Hill and Grossman 1987) or, conversely, upstream of culverts but did impact other aspects of the fish during dry periods when suitable habitat decreases as pools dry assemblages. Although species richness did not differ among up (Paloumpis 1958; Capone and Kushlan 1991). Unlike other seasons, different species were found during each season and reported movement rates (Storck and Momot 1981; Mundahl densities of species varied among seasons. Recapture rates also and Ingersoll 1983; Skalski and Gilliam 2000; Walker et al. varied, with higher recapture rates in the fall than in the spring. 2013), the potentially high movement rates in this study could be This higher recapture rate could be indicative of less movement indicative of movement rates in agricultural streams. Although made by fishes during the fall. The fall had lower stream ve- recapture rates were low in comparison to most stream studies locities, which could result in less movement since fish were that use VIE tags (e.g., Hill and Grossman 1987 [26%]; Olsen less likely to be swept downstream. Also, maintaining position and Vøllestad 2001 [24.2%]; Bruyndoncx et al. 2002 [40–54%]; in a stream at higher velocities has a higher metabolic cost and Petty and Grossman 2004 [60–80%]), the recapture rates in this may lead to fish movement downstream until refuge is found study were within the range reported elsewhere (e.g., Smithson (Facey and Grossman 1990). Likewise, some fish species ex- and Johnston 1999 [8.5–40%]; Brennan et al. 2005 [2.8%]; hibit increased home range sizes after storms increase stream Vander Pluym et al. 2008 [1.91–9.96%]). Mortality or lost tags flows (Hill and Grossman 1987). FishXing also predicted that were not the cause of low recapture rates since fish with only 662 BRIGGS AND GALAROWICZ

secondary tags (i.e., partial fin clips) were not recovered and Bettinger, J. M., and P. W. Bettoli. 2004. Seasonal movement of Brown Trout in tagging procedures from other studies (e.g., Catalano et al. 2001; the Clinch River, Tennessee. North American Journal of Fisheries Manage- Bruyndoncx et al. 2002; Roberts and Angermeier 2004; Brennan ment 24:1480–1485. Bouska, W. W., and C. P. Paukert. 2009. Road crossing designs and their impact et al. 2005) were followed. The higher recapture rates during fall on fish assemblages of Great Plains streams. Transactions of the American could imply less movement during that season than in spring. Fisheries Society 139:214–222. Brennan, N. P., K. M. Leber, H. L. Blankenship, J. M. Ransier, and R. DeBruler Management Implications Jr. 2005. An evaluation of coded wire and elastomer tag performance in juve- Culvert length is an important factor for fish seeking stream nile Common Snook under field and laboratory conditions. North American reaches upstream of culverts in agricultural streams (a setting Journal of Fisheries Management 25:437–445. with few previous studies) when culverts are not perched or set Brown, J. H., and A. Kodric-Brown. 1977. Turnover rates in insular biogeogra- phy: effect of immigration on extinction. Ecology 58:445–449. at high slopes and stream velocity is not a large factor. Managers Bruyndoncx, L., G. Knaepkens, W. Meeus, L. Bervoets, and M. Eens. 2002. The of agricultural streams should limit the length of culverts or put evaluation of passive integrated transponder (PIT) tags and visible implant in bridges that do not interfere with the stream channel. Based elastomer (VIE) marks as new marking techniques for the bullhead. Journal on the best model, culverts should be less than 17 m in length to of Fish Biology 60:260–262. allow at least a 50% probability of an equal proportion of Creek Burford, D. D., T. E. McMahon, J. E. Cahoon, and M. Blank. 2009. Assessment of trout passage through culverts in a large Montana drainage during summer Chub upstream and less than 13 m for a 75% probability. With low flow. North American Journal of Fisheries Management 29:739–752. the results of this and previous studies (Warren and Pardew 1998; Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel Bouska and Paukert 2010), box culverts should be installed inference: a practical information-theoretic approach, 2nd edition. Springer- over corrugated pipe arch culverts. Although bottomless culverts Verlag, New York. did not enhance fish passage, future studies should examine Cahoon, J. E., T. McMahon, L. Rosenthal, M. Blank, and O. Stein. 2007. Warm water species fish passage in eastern Montana culverts. Montana Department the effect of bottomless culverts further. Perception of possible of Transportation, Final Report FHWA/MT-07-009/8182, Helena. barriers by fish is relatively unknown and could be an important Cahoon, J. E., O. Stein, M. Blank, T. McMahon, and D. Burford. 2005. Fish consideration when examining fish passage through culverts and passage at road crossings in a Montana watershed. Montana Department of other road crossings. Transportation, Report FHWA/MT-05-002/8160, Helena. Capone, T. A., and J. A. Kushlan. 1991. Fish community structure in dry-season stream pools. Ecology 72:983–992. ACKNOWLEDGMENTS Carey, M., and P. Wagner. 1996. Salmon passages and other wildlife activi- We thank K. Broadway, J. Frey, J. Garavaglia, J. Gladish, ties in Washington State. Washington State Department of Transportation, C. Harris, E. Martin, T. McVay, and all other volunteers who Olympia. helped with fieldwork. We thank D. Woolnough and T. Gehring Catalano, M. J., S. R. Chipps, M. A. Bouchard, and D. H. Wahl. 2001. Evaluation for providing crucial input. Thank you to C. Freiburger for of injectable fluorescent tags for marking centrarchid fishes: retention rate and effects on vulnerability to predation. North American Journal of Fisheries allowing us to borrow surveying equipment and reviewing this Management 21:911–917. manuscript. This manuscript was improved by the additional Clapp, D. F., R. D. Clark Jr., and J. S. Diana. 1990. Range, activity, and habitat comments from anonymous reviewers. Funding was provided of large, free-ranging Brown Trout in a Michigan stream. Transactions of the by Central Michigan University, the Saginaw Bay Walleye Club, American Fisheries Society 119:1022–1034. and the Conservation Fund. Coffman, J. S. 2005. Evaluation of a predictive model for upstream fish passage through culverts. Master’s thesis. James Madison University, Harrisonburg, Virginia. REFERENCES Detenbeck, N. E., P. W. DeVore, G. J. Niemi, and A. Lima. 1992. Recovery of Allison, P. D. 2008. Convergence failures in logistic regression. SAS (Statistical temperate-stream fish communities from disturbance: a review of case studies Analysis Systems) Global Forum 2008, Paper 360-2008, SAS Institute, Cary, and synthesis of theory. Environmental Management 16:33–53. Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 North Carolina. Available: www2.sas.com/proceedings/forum2008/360- Ead, S. A., N. Rajaratnam, and C. Katopodis. 2002. Generalized study of hy- 2008.pdf. (April 2012). draulics of culvert fishways. Journal of Hydraulic Engineering 128:1018– Baker, C. O., and F. E. Votapka. 1990. Fish passage through culverts. U.S. De- 1022. partment of Transportation, Federal Highway Administration, General Tech- Eros,˜ T., B. Toth,´ A. Sevcsik, and D. Schmera. 2008. Comparison of fish assem- nical Report FHWA-FL-90-006, Portland, Oregon. blage diversity in natural and artificial rip-rap habitats in the littoral zone of a Bakker, V. J., and D. H. Van Vuren. 2004. Gap-crossing decisions by the red large river (River Danube, Hungary). International Review of Hydrobiology squirrel, a forest-dependent small mammal. Conservation Biology 18:689– 93:88–105. 697. Facey, D. E., and G. D. Grossman. 1990. The metabolic cost of maintaining Beechie, T., E. Beamer, and L. Wasserman. 1994. Estimating Coho Salmon rear- position for four North American stream fishes: effects of season and velocity. ing habitat and smolt production losses in a large river basin, and implications Physiological Zoology 63:757–776. for habitat restoration. North American Journal of Fisheries Management Fausch, K. D., C. E. Torgersen, C. V. Baxter, and H. W. Li. 2002. Landscapes 14:797–811. to riverscapes: bridging the gap between research and conservation of stream Belford, D. A., and W. R. Gould. 1989. An evaluation of trout passage through fishes. BioScience 52:483–498. six highway culverts in Montana. North American Journal of Fisheries Man- Fraser, D. F., and T. E. Sise. 1980. Observations on stream minnows in a patchy agement 9:437–445. environment: a test of a theory of habitat distribution. Ecology 61:790–797. Benton, P. D., W. E. Ensign, and B. J. Freeman. 2008. The effect of road Friedman, J. M., W. R. Osterkamp, M. L. Scott, and G. T. Auble. 1998. Down- crossings on fish movements in small Etowah Basin streams. Southeastern stream effects of dams on channel geometry and bottomland vegetation: Naturalist 7:301–310. regional patterns in the Great Plains. Wetlands 18:619–633. FISH PASSAGE THROUGH CULVERTS 663

Gagen, C. J., and C. Landrum. 2000. Response of warm-water fish to road cross- among Blacknose Dace (Rhinichthys atratulus) populations across a gradient ings on Ouachita Mountain streams. Report to U.S. Department of Agriculture of urbanization. Physiological and Biochemical Zoology 81:25–42. Forest Service, Ouachita National Forest, Hot Springs, Arkansas. Nislow, K. H., M. Hudy, B. H. Letcher, and E. P. Smith. 2011. Variation in Gerking, S. D. 1959. The restricted movement of fish populations. Biological local abundance and species richness of stream fishes in relation to dispersal Reviews 34:221–242. barriers: implications for management and conservation. Freshwater Biology Gibson, R. J., R. L. Haedrich, and C. M. Wernerheim. 2005. Loss of fish habitat 56:2135–2144. as a consequence of inappropriately constructed stream crossings. Fisheries Norman, J. R., M. M. Hagler, M. C. Freeman, and B. J. Freeman. 2009. Appli- 30(1):10–17. cation of a multistate model to estimate culvert effects on movement of small Gillies, C. S., and C. C. St. Clair. 2008. Riparian corridors enhance movement fishes. Transactions of the American Fisheries Society 138:826–838. of a forest specialist bird in fragmented tropical forest. Proceedings of the Northwest Marine Technology. 2008. Visible implant elastomer tag project National Academy of Sciences of the USA 105:19774–19779. manual: guidelines on planning and conducting projects using VIE and asso- Gorman, O. T., and J. R. Karr. 1978. Habitat structure and stream fish commu- ciated equipment. Northwest Marine Technology, Shaw Island, Washington. nities. Ecology 59:507–515. Available: www.nmt.us/support/appnotes/ape06.pdf. (January 2013). Gowan, C., and K. D. Fausch. 2002. Why do foraging stream salmonids move Olsen, E. M., and L. A. Vøllestad. 2001. An evaluation of visible implant elas- during summer? Environmental Biology of Fishes 64:139–153. tomer for marking age-0 Brown Trout. North American Journal of Fisheries Harper, D. J., and J. T. Quigley. 2000. No net loss of fish habitat: an audit of Management 21:967–970. forest road crossings of fish-bearing streams in British Columbia, 1996–1999. Osborne, L. L., and M. J. Wiley. 1992. Influence of tributary spatial position on Canadian Technical Report of Fisheries and Aquatic Sciences 2319. the structure of warmwater fish communities. Canadian Journal of Fisheries Harvey, B. C. 1991. Interactions among stream fishes: predator-induced habitat and Aquatic Sciences 49:671–681. shifts and larval survival. Oecologia 87:29–36. Paloumpis, A. A. 1958. Response of some minnows to flood and drought condi- Hill, J., and G. D. Grossman. 1987. Home range estimates for three North tions in an intermittent stream. Iowa State College Journal of Science 32:547– American stream fishes. Copeia 1987:376–380. 561. Hubbs, C. L., K. F. Lagler, and G. R. Smith. 2004. Fishes of the Great Lakes Park, D., M. Sullivan, E. Bayne, and G. Scrimgeour. 2008. Landscape-level region, revised edition. University of Michigan Press, Ann Arbor. stream fragmentation caused by hanging culverts along roads in Alberta’s Jackson, S. D. 2003. Design and construction of aquatic organism passage boreal forest. Canadian Journal of Forest Research 38:566–575. at road-stream crossings: ecological considerations in the design of river Pepino,M.,M.A.Rodr´ ´ıguez, and P. Magnan. 2012. Impacts of highway cross- and stream crossings. Pages 20–29 in C. L. Irwin, P. Garrett, and K. P. ings on density of Brook Charr in streams. Journal of Applied Ecology McDermott, editors. 2003 proceedings of the international conference on 49:395–403. ecology and transportation. Center for Transportation and the Environment, Pess, G. R., T. J. Beechie, J. E. Williams, D. R. Whitall, J. I. Lange, and J. R. North Carolina State University, Raleigh. Klochak. 2003. Watershed assessment techniques and the success of aquatic Jennings, M. J., M. A. Bozek, G. R. Hatzenbeler, E. E. Emmons, and M. D. restoration activities. Pages 185–201 in R. C. Wissmar and P. A. Bisson, Staggs. 1999. Cumulative effects of incremental shoreline habitat modifica- editors. Strategies for restoring river ecosystems: sources of variability and tion on fish assemblages in north temperate lakes. North American Journal uncertainty in natural and managed systems. American Fisheries Society, of Fisheries Management 19:18–27. Bethesda, Maryland. Kesel, R. H., and E. G. Yodis. 1992. Some effects of human modifications Peterson, J. T., and P. B. Bayley. 1993. Colonization rates of fishes in experimen- on sand-bed channels in southwestern Mississippi, U.S.A. Environmental tally defaunated warmwater streams. Transactions of the American Fisheries Geology and Water Sciences 20:93–104. Society 122:199–207. Lammert, M., and J. D. Allan. 1999. Assessing biotic integrity of streams: effects Petty, J. T., and G. D. Grossman. 2004. Restricted movement by Mottled Sculpin of scale in measuring the influence of land use/cover and habitat structure on (Pisces: Cottidae) in a southern Appalachian stream. Freshwater Biology fish and macroinvertebrates. Environmental Management 23:257–270. 49:631–645. Leavy, T. R., and T. H. Bonner. 2009. Relationships among swimming ability, Poff, N. L., and J. D. Allan. 1995. Functional organization of stream fish assem- current velocity association, and morphology for freshwater lotic fishes. North blages in relation to hydrological variability. Ecology 76:606–627. American Journal of Fisheries Management 29:72–83. Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, Ligon, F. K., W. E. Dietrich, and W. J. Trush. 1995. Downstream ecological R. E. Sparks, and J. C. Stromberg. 1997. The natural flow regime: a paradigm effects of dams: a geomorphic perspective. BioScience 45:183–192. for river conservation and restoration. BioScience 47:769–784. Lima, S. L., and P. A. Zollner. 1996. Towards a behavioral ecology of ecological Price, D. M., T. Quinn, and R. J. Barnard. 2010. Fish passage effectiveness of re- Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 landscapes. Trends in Ecology and Evolution 11:131–135. cently constructed road crossing culverts in the Puget Sound region of Wash- McDonald, W. R., and C. C. St. Clair. 2004. The effects of artificial and natural ington State. North American Journal of Fisheries Management 30:1110– barriers on the movement of small mammals in Banff National Park, Canada. 1125. Oikos 105:397–407. Resh, V. H. 2005. Stream crossings and the conservation of diadromous inver- Meffe, G. K., and A. L. Sheldon. 1990. Post-defaunation recovery of fish as- tebrates in South Pacific island streams. Aquatic Conservation: Marine and semblages in southeastern blackwater streams. Ecology 71:657–667. Freshwater Ecosystems 15:313–317. Mellina, E., S. G. Hinch, K. D. MacKenzie, and G. Pearson. 2005. Sea- Roberts, J. H., and P. L. Angermeier. 2004. A comparison of injectable fluores- sonal movement patterns of stream-dwelling Rainbow Trout in north-central cent marks in two genera of darters: effects on survival and retention rates. British Columbia, Canada. Transactions of the American Fisheries Society North American Journal of Fisheries Management 24:1017–1024. 134:1021–1037. Rodr´ıguez, M. A. 2002. Restricted movement in stream fish: the paradigm is Meyer, J. L., D. L. Strayer, J. B. Wallace, S. L. Eggert, G. S. Helfman, and N. E. incomplete, not lost. Ecology 83:1–13. Leonard. 2007. The contribution of headwater streams to biodiversity in river Rosenthal, L. R. 2007. Evaluation of distribution and fish passage in relation networks. Journal of the American Water Resources Association 43:86–103. to road culverts in two eastern Montana prairie streams. Master’s thesis. Mundahl, N. D., and C. G. Ingersoll. 1983. Early autumn movements and Montana State University, Bozeman. densities of Johnny (Etheostoma nigrum) and Fantail (E. Flabellare) darters Schaefer, J. F., E. Marsh-Matthews, D. E. Spooner, K. B. Gido, and W. J. in a southwestern Ohio stream. Ohio Journal of Science 83:103–108. Matthews. 2003. Effects of barriers and thermal refugia on local movement Nelson, J. A., P.S. Gotwalt, C. A. Simonetti, and J. W. Snodgrass. 2008. Environ- of the threatened Leopard Darter, Percina pantherina. Environmental Biology mental correlates, plasticity, and repeatability of differences in performance of Fishes 66:391–400. 664 BRIGGS AND GALAROWICZ

Schlosser, I. J., and P. L. Angermeier. 1995. Spatial variation in demographic Trautman, M. B., and D. K. Gartman. 1974. Re-evaluation of the effects of man- processes of lotic fishes: conceptual models, empirical evidence, and impli- made modifications on Gordon Creek between 1887 and 1973 and especially cations for conservation. Pages 392–401 in J. L. Nielsen and D. A. Powers, as regards its fish fauna. Ohio Journal of Science 74:162–173. editors. Evolution and the aquatic ecosystem: defining unique units in pop- Tritico, H. M., and A. J. Cotel. 2010. The effects of turbulent eddies on the sta- ulation conservation. American Fisheries Society, Symposium 17, Bethesda, bility and critical swimming speed of Creek Chub (Semotilus atromaculatus). Maryland. Journal of Experimental Biology 213:2284–2293. Sheer, M. B., and E. A. Steel. 2006. Lost watersheds: barriers, aquatic Turgeon, K., A. Robillard, J. Gregoire,´ V. Duclos, and D. L. Kramer. 2010. habitat connectivity, and salmon persistence in the Willamette and lower Functional connectivity from a reef fish perspective: behavioral tactics for Columbia River basins. Transactions of the American Fisheries Society 135: moving in a fragmented landscape. Ecology 91:3332–3342. 1654–1669. Vander Pluym, J. L., D. B. Eggleston, and J. F. Levine. 2008. Impacts of road Skalski, G. T., and J. F. Gilliam. 2000. Modeling diffusive spread in a hetero- crossings on fish movement and community structure. Journal of Freshwater geneous population: a movement study with stream fish. Ecology 81:1685– Ecology 23:565–574. 1700. Waite, I. R., and K. D. Carpenter. 2000. Associations among fish assemblage Smithson, E. B., and C. E. Johnston. 1999. Movement patterns of stream structure and environmental variables in Willamette Basin streams, Oregon. fishes in a Ouachita Highlands stream: an examination of the restricted Transactions of the American Fisheries Society 129:754–770. movement paradigm. Transactions of the American Fisheries Society 128: Walker, R. H., G. L. Adams, and S. R. Adams. 2013. Movement patterns of 847–853. Southern Redbelly Dace, Chrosomus erythrogaster, in a headwater reach of Snedden, G. A., W. E. Kelso, and D. A. Rutherford. 1999. Diel and seasonal an Ozark stream. Ecology of Freshwater Fish 22:216–227. patterns of Spotted Gar movement and habitat use in the lower Atchafalaya Wang, L., J. Lyons, P. Kanehi, R. Bannerman, and E. Emmons. 2000. Watershed River basin, Louisiana. Transactions of the American Fisheries Society 128: urbanization and changes in fish communities in southeastern Wisconsin 144–154. streams. Journal of the American Water Resources Association 36:1173– Stauffer, J. C., R. M. Goldstein, and R. M. Newman. 2000. Relationship of 1189. wooded riparian zones and runoff potential to fish community composition Warren, M. L., Jr., and M. G. Pardew. 1998. Road crossings as barriers to in agricultural streams. Canadian Journal of Fisheries and Aquatic Sciences small-stream fish movement. Transactions of the American Fisheries Society 57:307–316. 127:637–644. Storck, T., and W. T. Momot. 1981. Movements of the Creek Chub in a small Webb, P. W. 2004. Response latencies to postural disturbances in three species Ohio stream. Ohio Journal of Science 81:9–13. of teleostean fishes. Journal of Experimental Biology 207:955–961. Taylor, R. N. 2000. Humboldt County culvert inventory and fish passage eval- Winston, M. R., C. M. Taylor, and J. Pigg. 1991. Upstream extirpation of four uation. Final Report to California Department of Fish and Game, Contract minnow species due to damming of a prairie stream. Transactions of the FG-7068 IF, Ross Taylor and Associates, McKinleyville, California. American Fisheries Society 120:98–105. Downloaded by [Department Of Fisheries] at 20:16 28 May 2013 This article was downloaded by: [Department Of Fisheries] On: 20 June 2013, At: 00:52 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

North American Journal of Fisheries Management Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ujfm20 Rapid Visual Assessment to Determine Sex in Brook Trout David C. Kazyak a , Robert H. Hilderbrand a & Amanda E. Holloway b a Appalachian Laboratory , University of Maryland Center for Environmental Science , 301 Braddock Road, Frostburg , Maryland , 21532 , USA b Krieger School of Arts and Sciences , Johns Hopkins University , 3400 North Charles Street, Baltimore , Maryland , 21218 , USA Published online: 29 May 2013.

To cite this article: David C. Kazyak , Robert H. Hilderbrand & Amanda E. Holloway (2013): Rapid Visual Assessment to Determine Sex in Brook Trout, North American Journal of Fisheries Management, 33:3, 665-668 To link to this article: http://dx.doi.org/10.1080/02755947.2013.785998

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. North American Journal of Fisheries Management 33:665–668, 2013 C American Fisheries Society 2013 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2013.785998

MANAGEMENT BRIEF

Rapid Visual Assessment to Determine Sex in Brook Trout

David C. Kazyak* and Robert H. Hilderbrand Appalachian Laboratory, University of Maryland Center for Environmental Science, 301 Braddock Road, Frostburg, Maryland 21532, USA Amanda E. Holloway Krieger School of Arts and Sciences, Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218, USA

enhanced understanding of sex-specific life history variability Abstract may improve our ability to manage fishes, but candidate ap- Although sex-specific processes play a considerable role in the proaches must be evaluated before they can be used. ecology of many fishes, nonlethal tools to determine sex in most Male and female Brook Trout are known to be genetically species outside of the spawning season are lacking. We identified a suite of sexual characteristics in Brook Trout Salvelinus fontinalis (Phillips et al. 2002) and morphologically distinct (Power 1980), by surveying the available literature, consulting biologists, and re- but there are no established, practical methods to determine the viewing images of known-sex individuals. Using pretraining and sex of Brook Trout during periods outside of the spawning sea- posttraining testing of fisheries professionals, we assessed the util- son. A useful approach would be to use known-sex individuals ity of color and morphology for sexing fish across a length gradient obtained during the spawning season to identify characters that (110–251 mm TL) by presenting images of mature fish collected during the spawning season. Rapid visual assessment proved to be should be retained throughout the year. Our objectives were to an effective approach for determining the sex of Brook Trout. Aver- (1) evaluate the effectiveness of rapid visual assessment (RVA) age accuracy significantly improved from 71.5% before training to for sexing Brook Trout; (2) determine whether the effectiveness 92% after training, and the proportion of fish scored as unknown of RVA can be improved through a short training session; and was reduced (from 9.5% before to 2.0% after training). We found (3) examine factors that contribute to the effectiveness of RVA a small yet significant positive relationship between TL and the proportion of fish that were correctly sexed. With the exception of for determining the sex of Brook Trout. a single 113-mm male, all individuals were correctly sexed by at least 79% of respondents after training. No significant differences were found among respondents based on education, experience, or METHODS confidence level. The effectiveness of rapid visual assessment also We used backpack electrofishing to collect 111 Brook Trout did not significantly differ between male and female Brook Trout. from Big Run, a tributary to the Savage River in western Rapid visual assessment is a viable technique for the determination Maryland. Fish were collected on 6–8 October 2011, approx- Downloaded by [Department Of Fisheries] at 00:52 20 June 2013 of sex in Brook Trout. imately 1 week prior to the observed peak of spawning activity. This sampling period was chosen to maximize the proportion of mature, known-sex individuals observed within the population. Practical methods of sex determination in fishes are rare but Each fish was mildly anesthetized using tricaine methanesul- need to be made available to researchers and managers alike. fonate (80 mg/L; buffered with 0.2-mM NaHCO3,pH= 7). We Ideal techniques are nonlethal, quick, and affordable while also measured the TL (mm) of each fish and attempted to manually allowing sex determination throughout the year and at all stages express gametes from all captured individuals. If gametes were of ontogeny. Such methods open new doors in fisheries research expressed, the fish was identified as male (n = 36; 113–243 mm to identify sex-specific vital rates and their implications for TL) or female (n = 29; 110–251 mm TL) accordingly. The population-level processes; for example, Brook Trout Salvelinus fish were collected in a no-kill management zone and could fontinalis exhibit considerable life history variation, and tactics not be lethally sampled. Consequently, the sex of individuals may vary by sex (Hutchings 1996; Hutchings et al. 1999). An that did not express gametes was considered to be unknown

*Corresponding author: [email protected] Received September 18, 2012; accepted March 6, 2013 Published online May 29, 2013 665 666 KAZYAK ET AL.

(n = 46; 107–216 mm TL), and these individuals were excluded TABLE 1. Characters that were used to distinguish between male and female from the study. Standardized photographs of the left side of Brook Trout. each individual were collected by using consumer-grade digital Male Female photography equipment mounted on a tripod at a fixed distance (20 cm) above a white background. We analyzed these pho- Slab-sided bodya Fusiform body tographs, reviewed pertinent literature (Power 1980; Holloway Deeper head Head not as tall 2012), and conducted informal conversations with biologists to More colorfula Less colorful identify secondary sexual characteristics in Brook Trout. Longer, pointed snout Shorter, rounded snout At the East Coast Trout Management and Culture Workshop Kype present (on larger fish)a Kype absent V in 2012, we conducted a survey of 57 voluntary participants Black smudges on sides of Black smudges absent or (University of Maryland Institutional Review Board Project ventral surfacea less bold 329909-1). The survey was broken into four components. First, aFirst reported by Power (1980). we collected demographic data from each participant (education level, experience, and confidence in sex identification). Next, participants were shown a series of slides (hereafter, pretraining were once again asked to identify each fish as male, female, or test) developed from a randomly selected subset (n = 25) of unknown (see supplemental material in the online version of this our images of known-sex fish. Each slide was shown for 10 s article). and clearly stated the TL of the fish. For each slide, partici- A paired t-test was used to compare the pretraining and post- pants were asked to identify the sex of the fish as male, female, training percentages of fish that were correctly identified or or unknown. Based on the results of a morphological analysis that were identified as unknown. We used a two-sample t-test of sexual dimorphism in Brook Trout (Holloway 2012) and a to compare the accuracy of RVA for males and females, and literature review, we presented a 5-min training session high- ANCOVA was used to evaluate changes in the accuracy of RVA lighting a suite of secondary sexual characteristics that are useful before and after training and with respect to fish TL. Analysis of for distinguishing male and female Brook Trout (Table 1; Fig- variance was used to determine whether significant differences ure 1). Also included in the training were pictures of additional in accuracy existed among demographic attributes of the survey known-sex fish, and distinguishing characters were highlighted. participants. Because the response data were percentages, we Finally, the pretraining images were shown to participants in used an arcsine transformation for all analyses to better meet a random order (hereafter, posttraining test), and participants the assumptions required for t-tests, ANOVA, and ANCOVA. Downloaded by [Department Of Fisheries] at 00:52 20 June 2013

FIGURE 1. Comparison of external features used in rapid visual assessment of sex in Brook Trout. Note the pointed snout, slab-sided body, and dark black abdominal smudge on the male (top) compared with the rounded snout, more fusiform body, and relative lack of a black abdominal smudge on the female (bottom). These individuals were collected from Big Run, Maryland, during the October spawning season. [Figure available in color online.] MANAGEMENT BRIEF 667

Finally, we used linear regression on untransformed data to evaluate whether pretraining performance could predict posttraining improvement.

RESULTS A suite of secondary sexual characteristics was identified in Brook Trout (Table 1; Figure 1). In general, males were laterally compressed, with a deep head and a long, pointed snout. In contrast, females were torpedo-shaped, with a smaller head and a short, rounded snout. On large males, a kype was sometimes present. Males tended to be more brightly colored than females, although vibrant females were also observed. Many males had dark smudges along the sides of the abdomen. These smudges were less common and typically less pronounced on female Brook Trout. While none of these features was strictly dichoto- mous, the suite of characters taken together appeared to offer considerable insight into the sex of an individual. Anecdotally, the black smudges are retained throughout the year but are less pronounced at times other than the spawning season. Survey participants were able to correctly identify the sex of Brook Trout with much greater accuracy than simple guessing, FIGURE 2. Percentage of Brook Trout that were correctly sexed via rapid both before the training session (mean = 71.5%, SD = 16.0%; visual assessment, exhibiting a significant positive relationship with training (pretraining versus posttraining) and fish TL. Untransformed data are shown for t56 = 9.8, P < 0.001) and after training (mean = 92.0%, SD = = < visual clarity. Note the substantially reduced variation in the posttraining scores 5.7%; t56 32.2, P 0.001). However, the percentage of fish relative to the pretraining scores. that were correctly sexed was significantly greater after the train- Oncorhynchus ing session (t56 =−11.3, P < 0.001). In addition, the percentage spp. (Beacham and Murray 1986; Brykov et al. of fish that were identified as being of unknown sex was sig- 2010). Our study demonstrates the effectiveness of RVA for use nificantly lower after the training session (mean = 2.0%, SD = with mature, riverine Brook Trout during the spawning season. = 3.5%) than before training (mean = 9.5%, SD = 10.1%; t56 = Accuracy rates were high (mean 92.0%) and uncertainty 7.7, P < 0.001). There was no evidence of an interaction between fish TL and training with regards to the effectiveness of RVA. Length had a weak positive relationship with the percentage of fish that were correctly sexed for both pretraining and posttraining tests (R2 = 0.14, P < 0.05; Figure 2). However, with the exception of a single 113-mm male, all individuals were correctly sexed by at least 79% of respondents after training (Figure 2). When that individual was omitted from our analysis, Downloaded by [Department Of Fisheries] at 00:52 20 June 2013 the average posttraining accuracy was 94.4%. There was a strong negative relationship between the pretraining scores and the improvement that occurred after training (n = 57, adjusted R2 = 0.87, P < 0.001; Figure 3): respondents that scored poorly before training exhibited significantly greater posttraining improvement than respondents that had high scores on the pretraining test. Accuracy during the pretraining and posttraining tests was not affected by the respondents’ experience, position, education, or perceived confidence in sexing fish (P > 0.05). There was no difference in the accuracy of sexing male fish versus female fish (t22, 28 =−1.7, P = 0.10).

DISCUSSION FIGURE 3. Linear regression of the relationship between pretraining scores and posttraining improvement in the accuracy of Brook Trout sex determi- Rapid visual assessment based on morphology has been nation. Pretraining scores demonstrated a strong inverse relationship with the successfully used to determine the sex of Paciﬁc salmon improvement that was observed after a short training session. 668 KAZYAK ET AL.

was low (mean = 2.0%) after training had been provided. Overall, RVAappears to be a viable approach for determining After the smallest individual fish in the survey was excluded the sex of Brook Trout in some applications. This approach is from analysis, overall accuracy rates exceeded 94%. Sexual fast, affordable, and minimally invasive, but it lacks the level maturity in small Brook Trout (<120 mm) is uncommon but of certainty that may be required for some applications and well documented in stream-dwelling populations (McFadden it may not be as effective for use with immature individuals. 1961; Wydoski and Cooper 1966). In other locations, where Identification of additional sex-specific traits that are effective larger Brook Trout are the focus of field studies, we expect at all stages of ontogeny and that have been cross validated that secondary sexual characteristics will be more pronounced with genetic markers would substantially enhance our ability to and that RVA will be even more effective. Although some determine sex-specific life history attributes and variables for participants had very high scores (up to 96% accuracy) on the the purpose of better management of fishes. pretraining test, marked overall improvement (mean = 20.5%) during the posttraining test suggests that knowledge of a few key morphological characteristics can significantly improve a ACKNOWLEDGMENTS person’s ability to accurately sex Brook Trout. There were no Field discussions with Joe Zydlewski (U.S. Geological Sur- significant differences in performance based on demographic vey, Maine Cooperative Fish and Wildlife Research Unit) in- variables (e.g., experience, education level, or confidence) spired this research project. Mark Colaw and Wes Wagner aided before or after training, further suggesting that this approach with the field component of this study. Alan Heft and Matt can be rapidly learned by fisheries personnel. Sell (Maryland Department of Natural Resources) facilitated The RVA approach appears to perform well across sizes and the implementation of the survey. This paper is scientific se- sexes within the Big Run population of Brook Trout. Accuracy ries contribution number 4757 from the University of Maryland rates for male and female Brook Trout were not different, and Center for Environmental Science, Appalachian Laboratory. TL differences, although statistically significant, did not have a strong impact on the accuracy of RVA. Larger fish were correctly sexed more often, probably due to their more pronounced REFERENCES secondary sexual characteristics. However, even the smaller fish Beacham, T. D., and C. B. Murray. 1986. Sexual dimorphism in length of were correctly sexed at least 79% of the time, which approaches upper jaw and adipose fin of immature and maturing Pacific salmon (On- the accuracy rates reported by Beacham and Murray (1986) for corhynchus). Aquaculture 58:269–276. Brykov, V. A., A. D. Kukhlevsky, and A. V. Podlesnykh. 2010. Incomplete spawning Pacific salmon. Brook Trout are thought to exhibit congruence between morphobiological characters and sex-specific molecular morphological variation among populations, and transferabil- markers in Pacific salmons: 1. analysis of discrepancy in five species of the ity remains untested. Although we cannot directly evaluate the genus Oncorhynchus. Russian Journal of Genetics 46:867–872. transferability of RVA,we are encouraged that in the pretraining Casselman, S. J., and A. I. Schulte-Hostedde. 2004. Reproductive roles predict test, some biologists from other regions effectively sexed Brook sexual dimorphism in internal and external morphology of Lake Whitefish, Coregonus clupeaformis. Ecology of Freshwater Fish 13:217–222. Trout collected at Big Run, thus indicating that secondary sexual Fleming, I. A., and M. R. Gross. 1994. Breeding competition in a Pacific salmon characteristics are conserved between populations. (Coho: Oncorhynchus kisutch): measures of natural and sexual selection. The Brook Trout used in this study were collected during Evolution 48:637–657. the spawning season in order to allow definitive determination Holloway, A. E. 2012. Morphometric analysis for nonlethal sex determi- of sex. However, almost every morphological character that we nation in Brook Trout: a new tool for research and management. Mas- ter’s thesis. Johns Hopkins University, Baltimore, Maryland. Available: used should be visible throughout the year. Coloration may vary jscholarship.library.jhu.edu/handle/1774.2/36076. (September 2012). Downloaded by [Department Of Fisheries] at 00:52 20 June 2013 with season, with males in particular showing bright oranges Hutchings, J. A. 1996. Adaptive phenotypic plasticity in Brook Trout, Salvelinus and reds. Another key, sexually dimorphic character is the shape fontinalis, life histories. Ecoscience 3:25–32. of the snout. Males tend to have more angular snouts, whereas Hutchings, J. A., A. Pickle, C. R. McGregor-Shaw, and L. Poirier. 1999. In- the snouts of females are rounded, and this difference was no- fluence of sex, body size, and reproduction on overwinter lipid depletion in Brook Trout. Journal of Fish Biology 55:1020–1028. ticeable even in the smallest fish we were able to accurately sex Janhunen, M., N. Peuhkuri, and J. Piironen. 2009. Morphological variability (Holloway 2012). Exaggerated head features have been among three geographically distinct Arctic Charr (Salvelinus alpinus L.) pop- observed in males of other salmonids (Fleming and Gross ulations reared in a common hatchery environment. Ecology of Freshwater 1994; Casselman and Schulte-Hostedde 2004), and these Fish 18:106–116. sexual characteristics appear to be retained after an individual McFadden, J. T. 1961. A population study of the Brook Trout, Salvelinus fontinalis. Wildlife Monographs 7. has reached reproductive age (Janhunen et al. 2009). Survey Phillips, R. B., M. P. Matsuoka, and K. M. Reed. 2002. Characterization of participants also noted that around the spawning season, charr chromosomes using fluorescence in situ hybridization. Environmental females often exhibit a swollen vent. This additional character Biology of Fishes 64:223–228. may help biologists when using RVA in the field, but it was Power, G. 1980. The Brook Charr, Salvelinus fontinalis. Pages 141–203 in not feasible for RVA based on the photographs presented in E. K. Balon, editor. Charrs, salmonid fishes of the genus Salvelinus.DrW. Junk, The Hague, The Netherlands. our survey. Although the accuracy of RVA may vary to some Wydoski, R. S., and E. L. Cooper. 1966. Maturation and fecundity of Brook degree, the primary characters used to determine sex in Brook Trout from infertile streams. Journal of the Fisheries Research Board of Trout should be effective throughout the year. Canada 23:623–649.