An Evaluation of Redd Counts As a Measure of Bull Trout Population Size and Trend Philip J

This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:34 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 Evaluation of Redd Counts as a Measure of Bull Trout Population Size and Trend Philip J. Howell a & Paul M. Sankovich b a U.S. Forest Service, Forestry and Range Sciences Laboratory, 1401 Gekeler Lane, La Grande, Oregon, 97850, USA b U.S. Fish and Wildlife Service, Columbia River Fisheries Program Office, La Grande Field Office, 3502 Highway 30, La Grande, Oregon, 97850, USA

Available online: 07 Feb 2012

To cite this article: Philip J. Howell & Paul M. Sankovich (2012): An Evaluation of Redd Counts as a Measure of Bull Trout Population Size and Trend, North American Journal of Fisheries Management, 32:1, 1-13 To link to this article: http://dx.doi.org/10.1080/02755947.2011.649192

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 32:1–13, 2012 American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2011.649192

ARTICLE

An Evaluation of Redd Counts as a Measure of Bull Trout Population Size and Trend

Philip J. Howell* U.S. Forest Service, Forestry and Range Sciences Laboratory, 1401 Gekeler Lane, La Grande, Oregon 97850, USA Paul M. Sankovich U.S. Fish and Wildlife Service, Columbia River Fisheries Program Ofﬁce, La Grande Field Ofﬁce, 3502 Highway 30, La Grande, Oregon 97850, USA

Abstract The use of redd counts to monitor abundance and trend of bull trout Salvelinus confluentus has been clouded by uncertainties concerning measurement error, life history variation, and correspondence of redd counts to adult population size. We compared census redd counts with population estimates of mature females for a migratory fluvial population of bull trout (primarily ≥ 300 mm fork length) and for a population of small (<200 mm), likely resident, bull trout. We also compared the measurement error of the experienced surveyors who conducted the redd counts to that of a group of inexperienced surveyors. Although the regression of redd counts on adult females for the migratory population was statistically significant, a large proportion of the variation in the relationship was unexplained (r2 = 0.47). Despite that variation, redd counts accurately reflected a greater than 50% decline in the population over 10 years; however, 5-year trends in redd counts could be misleading. Power analysis parameterized by using the variation in the number of females per redd and measurement error of experienced surveyors indicated that minimum declines of 44–56% or increases of 78–118% over 10–15 years would be necessary for detection using traditional statistical criteria. Geometric mean abundance of migratory adults derived from redd counts and adults-per-redd values from the present study and published averages were similar to measured adult numbers in most cases. For both migratory and resident populations, redd counts by experienced surveyors were substantially more accurate and precise than those by inexperienced surveyors. Counts of migratory bull trout redds were more accurate and precise than counts of resident bull trout redds, which were significantly smaller and consistently underestimated. Thus, bull trout redd counts can be used to estimate abundance levels and to detect substantial longer-term changes in abundance, particularly for migratory populations. However, the reliability of the counts depends on the skill of the surveyors. Downloaded by [Department Of Fisheries] at 01:34 01 March 2012

Estimates of abundance and trend (i.e., change in abun- groups of populations (metapopulations) and objectives for dance over time) are critical in determining the status of species, maintaining stable or increasing trends in adult abundance. particularly those that are suspected to be rare or declining. Consequently, monitoring abundance and trend is important for Bull trout Salvelinus conﬂuentus are listed as threatened under assessing recovery of the species. the Endangered Species Act in the coterminous United States Redd counts have been and continue to be the most widely (USFWS 1999) and are classiﬁed as sensitive in Alberta, British used method for monitoring adult bull trout abundance (Rieman Columbia, and the Yukon Territory, Canada (ASRD/ACA 2009). and Myers 1997; Dunham et al. 2001) and are used extensively The draft bull trout recovery plan (USFWS 2002) contains for other salmonids, particularly anadromous forms (Gallagher numeric objectives for adult abundance in core areas or et al. 2007). However, among salmonid species there are

*Corresponding author: [email protected] Received April 4, 2011; accepted August 8, 2011

1 2 HOWELL AND SANKOVICH

considerable differences and similarities in the timing and 2008). Different redd densities in nonindex areas could result distribution of spawning, the size of adults and redds, and life in inaccurate expansions of index counts for estimates of total history forms, as well as in the physical characteristics and redd or population abundance (Jacobs and Nickelson 1998). conditions of the spawning areas (e.g., substrate and flow). To generate estimates of population abundance, perhaps more These factors can influence error in the redd counts and, critical than the relationship of redd counts to true redd numbers consequently, design of the surveys (Gallagher et al. 2007). is the relationship of redd counts to numbers of adults. Com- Redd counts offer several advantages for monitoring bull bined redd counts and corresponding independent abundance trout abundance. Since redds are constructed by reproductive estimates of spawning bull trout for the Wigwam River and fish homing to natal streams, they directly relate to the metric tributaries of Lake Pend Oreille and the Metolius River were of interest (adult bull trout) and generally reflect the spatial highly correlated but only on a log scale (Dunham et al. 2001). distribution of breeding populations. Consequently, they do not The number of adults per redd varied substantially from 1.03 rely on additional assumptions or data regarding survival or to 3.33, which Dunham et al. (2001) suggested could be due to distribution, which are needed when using estimates of other error in the estimates or to varying life history patterns. Addi- life stages. Bull trout spawn during early fall when flows and tional work has expanded the upper end of the range of adults turbidity are low, conditions that are conducive to detecting per redd to 4.3 (Taylor and Reasoner 2000). Thus, there is a need redds (Rieman and Myers 1997). Redd surveys can also be to develop population-specific relationships, or at least regional less expensive and more easily conducted than other inventory and life history-related relationships, to derive more accurate methods, such as trapping or mark–recapture (Dunham et al. abundance estimates. 2001). Redd counts can be used alone as an index of adult Life history variation could affect both redd counts and es- abundance to determine population trends (Rieman and Myers timates of adult abundance. Bull trout have been typified as 1997; Maxell 1999) or can be combined with information on having two primary life history forms: migratory and resident the ratio of redds to adults in order to estimate population size (Rieman and McIntyre 1993). Migratory juvenile bull trout rear (Al-Chokhachy et al. 2005; Murdoch et al. 2009). in tributaries, where spawning also occurs, for several years be- Despite these advantages, the reliability of applying bull trout fore moving downstream to larger rivers (fluvial forms) or lakes redd counts to estimate adult abundance depends on several (adfluvial) to mature. Later, as adults, they return to spawn in factors influencing the precision and accuracy of these counts. natal streams, after which they move back down to the rivers or Such factors include measurement error in counting redds and lakes (Pratt 1992). Resident forms, on the other hand, remain the relationship of redd counts to the number of adults. High in upper reaches and smaller tributaries during all life stages. measurement error or a low correlation between redd counts Although there may be substantial size differences between resi- and the number of adults could, at best, obscure patterns in dent and migratory adults (e.g., 150–300 and >300 mm, respec- abundance and reduce power to detect trends (Maxell 1999); at tively; Goetz 1989; Mullan et al. 1992; Pratt 1992), there is little worst, they could lead to erroneous conclusions. size differential between resident and migratory juveniles or be- In several previous analyses of bull trout redd counts, the tween older migratory juveniles and resident adults (Rieman measurement error and adults-per-redd relationship were un- and McIntyre 1993). Consequently, it is difficult to identify and known (Rieman and Myers 1997; Maxell 1999; Al-Chokhachy enumerate resident adults based on size or other external char- et al. 2005). Dunham et al. (2001) found substantial measure- acteristics. Although other authors have speculated about the ment error from high variation in counts among surveyors. Sur- potential influence of differences in life history on redd counts veyor counts were 28–254% of the best estimate of the “true” (Al-Chokhachy et al. 2005), factors related to classifying adults number of redds. In a study limited to experienced surveyors, and redds as resident or migratory (e.g., maturity, size at ma- Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 Muhlfeld et al. (2006) found less measurement error (78–130% turity, and redd size) have not been evaluated and the influence of the best estimates) and applied detection efficiencies to cor- of life histories on redd counts has not been explicitly assessed. rect redd counts for observer error and provide confidence in- Bull trout may also spawn either annually or in alternate years tervals (CIs). (Pratt 1992). Failure to account for nonspawning adults could Besides variability among surveyors, the timing, frequency, result in underestimates of adult numbers based on redd counts. and spatial coverage of redd surveys can profoundly influence Thus, while there is some evidence suggesting correlation the ability to estimate total redd numbers (Dunham et al. 2001; between redd counts and adults, there is also evidence of sub- Courbois et al. 2008). For example, redd surveys are frequently stantial variability in redd counts from measurement error and limited to “index” areas (e.g., Al-Chokhachy et al. 2005) rather differences in life history that could affect the utility of redd than randomly or probabilistically selected sites within the total counts for estimating abundance. Previous evaluations of bull spawning distribution or complete surveys of the spawning dis- trout redd counts were lacking in (1) estimates of measurement tribution (Stevens et al. 2007; Jacobs et al. 2009). Variationin the error in the redd counts, (2) estimates of the numbers of adults spatial distribution of spawning between index and unsurveyed, for comparison, (3) estimates of the effect of life history dif- nonindex areas may influence the detection of population trends ferences on the redd counts, or (4) some combination of these. (Dunham et al. 2001; Isaak and Thurow 2006; Courbois et al. The primary purpose of this study was to robustly evaluate the BULL TROUT POPULATION SIZE AND TREND 3

N WA

OR Mill Creek

North Fork Mill Creek

Diversion dam Walla Walla River and trap

0 510 Kilometers Low Creek

Spawning

FIGURE 1. Map of Mill Creek, including the study area above the diversion dam and the bull trout spawning distribution.

relationship of redd counts to adult abundance for migratory Prosopium williamsoni also occur in upper Mill Creek. Spring and resident forms of bull trout by addressing possible sources Chinook salmon O. tshawytscha were historically present, and of error in both estimates. Our objectives were to (1) compare adults have been occasionally reintroduced to spawn. estimates of adult population size generated from redd counts (redds per mature female or redds per adult) with those derived METHODS from direct estimates of adults, (2) compare trends in abun- Female abundance.—Since redds are constructed by spawn- dance based on adult estimates versus redd counts, (3) estimate ing females (Esteve 2005), we estimated numbers of mature bull the error in the redd counts due to variability among experi- trout females for the most direct comparison with the number enced versus inexperienced observers, and (4) determine the of redds. To estimate the number of fluvial adult females, we influence of the variability in redd counts and adult estimates on used a combination of upstream trap counts and mark–recapture the detection of trends. estimates. During 1998 through 2007, we operated a trap at the upstream end of the ladder at the diversion dam from mid-May STUDY AREA through mid-October, which spanned the period when migratory This study was conducted in the upper Mill Creek wa- fish moved above the dam (Hemmingsen et al. 2000). Bull trout tershed (Walla Walla River subbasin) in northeastern Oregon trapped at the ladder were anesthetized with tricaine methane- and southeastern Washington (Figure 1). Previous redd surveys sulfonate (MS-222; 60 mg/L), measured for fork length (FL), Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 (Hemmingsen et al. 2001b, 2001c) indicated that 99% of the mi- weighed, and interrogated for a passive integrated transponder gratory bull trout population spawned upstream of a 2.5-m-high (PIT) tag; if no PIT tag was present, the fish was injected with diversion dam, which was equipped with a fish ladder, at river one. In 2002–2007, we inspected each fish for maturity by using kilometer 41. The Mill Creek watershed above the diversion ultrasound to identify mature females. Ovaries and eggs can be dam is 8,798 ha (NPPC 2004). It was designated as the mu- readily identified with this technique (e.g., Evans et al. 2004). nicipal water supply for Walla Walla, Washington, in 1918 and We used PIT tag recaptures to determine annual frequencies of has been closed to public entry since 1974. The area is largely upstream migration and gonadal maturity for individual fish. To roadless and is closed to grazing and timber harvest. The tribu- estimate the number of females trapped during 1998–2001 prior taries to Mill Creek above the diversion dam are short (<6km to our use of ultrasound, we multiplied the total trap count of perennial flow) and contribute small amounts of flow (≤0.1982 300-mm or larger bull trout (see below) in those years by the m3/s [≤7ft3/s] each) during summer base flow conditions (U.S. mean fraction of females trapped in 2002–2007. Forest Service, unpublished data). Besides bull trout, rainbow A telemetry study indicated that most fluvial adult bull trout trout Oncorhynchus mykiss, steelhead (anadromous rainbow in the Mill Creek drainage overwintered downstream from trout), sculpins Cottus spp., western brook lampreys Lampetra the dam; however, a few overwintered upstream of the dam richardsoni, suckers Catostomus spp., and mountain whitefish (Hemmingsen et al. 2001b, 2001c). In 2003–2007, bull trout 4 HOWELL AND SANKOVICH

captured in the trap were marked with a caudal punch to pling of resident-sized females smaller than 300 mm was limited facilitate a mark–recapture (resight) estimate to account for any to Low Creek (Figure 1), where only resident-sized fish smaller fluvial adults that had overwintered upstream from the dam than 300 mm had been observed to spawn. and thus were not included in the counts at the ladder trap. The In early August 2003 and 2005, we blocknetted and elec- location of the caudal punch was changed each year so that trofished every third pool and riffle to obtain depletion esti- marked fish from various release years could be distinguished. mates of mature resident females in Low Creek. We completed To obtain mark–resight estimates, a single diver snorkeled at least two passes until we achieved a 67% reduction from the all the pools and other habitats of sufficient depth to hold fluvial previous pass. Fish were examined endoscopically for maturity adult-sized fish on two to three consecutive days during mid- to by following the methods of Swenson et al. (2007). Densities late August prior to the onset of spawning. The diver recorded were determined by dividing removal estimates of female adults the number of marked and unmarked bull trout (300 mm or by the surface area of the units sampled. These densities were larger) that were observed. Trapping data for the previous 5 years expanded across the available habitat area in the reach where indicated almost all of the migratory fish returning to spawn the density samples were collected to obtain a total abundance in the population were 300 mm or larger (Hemmingsen et al. estimate of mature resident females using methods described 2001b, 2001c). We estimated the total number of unmarked, by Hankin and Reeves (1988). Prior to the depletion estimates, 300-mm or larger bull trout by incorporating the number of habitat units throughout Low Creek were classified as pools or marked and unmarked bull trout observed while snorkeling riffles and the unit areas were measured. Because of possible and the number of marked bull trout released upstream of the bias in electrofishing depletion estimates, in 2005 we evalu- trap at the time of snorkeling into Bailey’s modification of ated bias in our estimates by following the methods of Peterson the Peterson mark–recapture formula (Ricker 1975), which et al. (2004). In a subsample of eight pools and eight riffles accounts for replacement of marked fish into the population, longitudinally spaced throughout the distribution of bull trout in Low Creek, we applied a caudal fin clip to 130-mm or larger M × C + N = ( 1), bull trout that were collected during the removal passes and then (R + 1) returned the marked fish to the blocknetted units. The next day, we completed electrofishing removal passes in these blocknet- where N is the population size, M is the number of marked ted units, and the marked and unmarked bull trout in the catch bull trout, C is the total number of 300-mm or larger bull trout were measured and counted. that were observed (recaptured) by the snorkeler, and R is the Redd counts.—We conducted three to four redd counts at number of marked bull trout observed by the snorkeler. We then 2–3-week intervals during September and October in Mill Creek multiplied the estimate of unmarked adults (N − M)bythe and its tributaries upstream of the diversion dam to the upstream female fraction of the bull trout trapped that year to determine limits of fish distribution. Data for Mill Creek indicated that the the number of unmarked females above the trap. Since the spawning duration was from the last week of August through percentage of marked fish in the snorkel count exceeded 10% of October (Hemmingsen et al. 2001b) and that 94% of the redds the total snorkel count, CIs were based on a binomial distribu- remained visible during that survey interval (Hemmingsen et al. tion (Seber 1982). To estimate the number of 300-mm or larger 2001a). Thus, the spatial extent, frequency, and timing of our females that remained above the trap in 1998–2002 prior to our redd counts should have provided essentially a complete count mark–recapture estimates, we multiplied the number of females or “census” of the number of detectable redds (after Gallagher trapped in each of those years by the mean proportion that the et al. 2007; Jacobs et al. 2009). unmarked females from the mark–recapture estimates respre- On each survey, each new redd observed was flagged and Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 sented in the number of females trapped during 2003–2007. uniquely numbered, and the combined area of the pit and tailspill To identify mature resident females (<300 mm), we used was calculated using measurements of the maximum length endoscopy to examine bull trout (130–299 mm FL) collected and width (Zimmerman and Reeves 2000). Flagged redds from by electrofishing a systematic sample of habitat units (23% of previous surveys that year were remeasured. We used the largest the pools and 14% of the riffles) of Mill Creek and its tribu- redd area measured during all of the surveys to ensure that the taries above the diversion dam in 2002. Endoscopy has been area represented a fully completed redd. Lengths of bull trout shown to be a highly accurate technique for determining matu- observed on or within 1 m of the redds were visually estimated rity and sex in small salmonids (Swenson et al. 2007). Similar to the nearest 50-mm length-class. For comparison of redd sizes, work we previously conducted in Silver Creek (Powder River we also measured redds using the same methods in the Little subbasin, Oregon) indicated that mature resident females were Minam River (adjacent Grande Ronde River subbasin), where, 140–200 mm (Hemmingsen et al. 2001b). Therefore, to be con- like Low Creek, all bull trout observed during previous spawning servative, we inspected every captured bull trout that was 130 surveys and other sampling (e.g., electrofishing) were smaller mm or larger. Since no mature, resident-sized females were than 300 mm. found that year in Mill Creek and tributaries where fluvial-sized Surveys were conducted by three to six individuals with prior fish had previously been observed to spawn, subsequent sam- experience in counting bull trout redds, and over the course of BULL TROUT POPULATION SIZE AND TREND 5

the study most of the surveys were done by the same individ- ratios for Mill Creek and from observer error in redd counts uals. Each of the three of the surveyors who conducted redd from the test reaches. We compared minimum detectable dif- counts throughout the study had about 14 years of experience ferences in trends in 5-year increments from 5 to 25 years. We in conducting bull trout redd surveys. A surveyor added later also tested varying assumptions about significance level (α) and had 9 years of experience in counting bull trout redds. Most of one- or two-tailed tests for significance. Following the methods the surveyors had additional experience surveying redds of Chi- of Maxell (1999), an α level of 0.05 and a two-tailed test were nook salmon, coho salmon O. kisutch, and summer steelhead. compared with an α level of 0.20 and a one-tailed test to de- A group of inexperienced surveyors, who were conducting bull tect declines. The latter permits increased detection of declines trout redd counts in other adjacent portions of the Walla Walla (i.e., lower minimum decreases or earlier detection) but does not and Umatilla River subbasins, were also evaluated for measure- distinguish increasing trends from stable trends. However, this ment error for purposes of comparison with the experienced would meet monitoring needs for trends in abundance under the surveyors counting redds in Mill Creek and its tributaries. At draft recovery plan. Power was held constant at 0.80, and the the time of evaluation, the inexperienced surveyors had 2 weeks CV was assumed to be constant with respect to abundance. of training that covered redd characteristics and observation of We also compared estimates of adult population size based bull trout redds and 6 weeks of experience in counting bull trout on direct methods to estimates from expansions of redd counts. redds during regularly scheduled surveys. To directly determine the total number of fluvial adults (fe- To estimate potential observer error in our redd surveys, in males and males), we combined the numbers of all 300-mm 2004 we established three 1-km test reaches in a section of or larger bull trout in our upstream trap counts and all un- main-stem Mill Creek that was used by larger, fluvial adults and marked, 300-mm or larger bull trout upstream of the trap from three 1-km reaches in Low Creek, which was used by smaller, our mark–resight estimates (see section on female abundance resident-sized adults. Redds were identified and flagged dur- in Methods). For estimates of population size based on redd ing the first survey on 21–22 September. To determine the best counts, we compared estimates from the 10 years of Mill Creek estimate of the “true” number of visible new redds in the test redd counts by using the mean adults-per-redd for Mill Creek reaches, we followed methods similar to those used by Dunham and mean adults-per-redd values from Dunham et al. (2001) et al. (2001) and Muhlfeld et al. (2006). During the 3-week pe- and Al-Chokhachy et al. (2005) as examples where population- riod after the first survey (a typical interval between surveys), specific data might not be available. We used geometric mean redds were identified and flagged at weekly intervals by an ex- abundance for 5- and 10-year periods during 1998–2007 and re- perienced surveyor that was not responsible for routine surveys lated these to six abundance classes (1–50; 50–250; 250–1,000; of those reaches. At the end of the 3 weeks, flags were removed 1,000–2,500; 2,500–10,000; and 10,000–100,000 adults) used from all redds identified after the first survey. Each of the survey- by the International Union for Conservation of Nature (IUCN ors responsible for the routine surveys of fluvial redds (i.e., all 2001) for its Red List of Threatened Animals and by the U.S. survey sections except Low Creek) and the inexperienced sur- Fish and Wildlife Service (USFWS 2008) to assess status of veyors used for comparison independently counted unflagged core areas during its 5-year status review of bull trout. Similar redds in each of the test reaches on main-stem Mill Creek. The to the trend analysis, geometric means are used when changes experienced surveyor responsible for routine surveys of redds tend to be exponential, as in population growth. of suspected resident fish in Low Creek and the same crew of inexperienced surveyors independently counted redds in the three RESULTS test reaches in Low Creek. On the day after the evaluation, four experienced surveyors together resurveyed the test reaches to Female Estimates and Redd Counts Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 verify the best estimate of the number of new redds since the In our systematic sample of Mill Creek and tributaries first survey. Only the experienced surveyors evaluated for ob- above the trap, we collected 54 bull trout (113–255 mm FL) server error conducted the redd surveys in 2004–2007 and most to identify potentially resident mature females smaller than of the surveys in prior years. 300 mm that would not be accounted for in our trap counts and Analysis.—Trends in redd counts and adult females were ini- mark–recapture estimates. Ninety-six percent of the sampled tially analyzed using linear regression. Counts were loge trans- fish were 125–255 mm; 81% of the sampled fish were from formed based on an assumption of exponential changes in abun- Mill Creek, where most of the redds in previous spawning dance (Maxell 1999) and to normalize the distributions. Because surveys were counted. We did not collect any bull trout from the of apparent nonlinear patterns and autocorrelation in the linear 256–299-mm size-class. We found 13 mature males (159–255 regressions, we also used polynomial regression. To evaluate mm) but no mature females, suggesting that smaller, potentially effects of error in the redd counts in detecting trends, we used resident females did not occur or were present at low densities. TRENDS software (Gerrodette 1993), similar to the work by We also examined redd size for evidence of smaller, resi- Maxell (1999). This allowed us to determine the minimum de- dent adults. Pairwise comparisons of redd area indicated that tectable proportional increase and decrease in abundance using redds in main-stem Mill Creek and the North Fork Mill Creek coefficients of variation (CV = SD/mean) from female : redd were significantly larger than redds in Low Creek and other 6 HOWELL AND SANKOVICH

TABLE 1. Quartile area (m2) of bull trout redds in Mill Creek and its trib- 200 utaries and in the Little Minam River (25% = 25th percentile; 75% = 75th percentile). 175 Females 2 Redds Redd area (m ) 150 Number Stream of redds Median 25% 75% 125

Mill Creek 533 1.09 0.69 1.80 Number 100 North Fork Mill Creek 40 0.99 0.66 1.46 Low Creek 270 0.23 0.15 0.34 75 Bull Creek 14 0.45 0.14 0.84 Paradise Creek 16 0.39 0.25 0.63 50 Little Minam River 320 0.23 0.15 0.35 25 1998 2000 2002 2004 2006 tributaries (Tables 1, 2; Kruskal–Wallis one-way analysis of FIGURE 2. Number of fluvial adult female bull trout and number of bull trout variance, Dunn’s pairwise comparison). This is also consistent redds in Mill Creek, 1998–2007. with the size of fish observed during spawning surveys and other sampling of those streams. For example, no 300-mm or larger Low Creek and the Little Minam River had nearly identically bull trout were observed in Low Creek during spawning surveys, sized smaller redds and smaller spawners (<300 mm; Table 1). whereas fish occupying or adjacent to redds in Mill Creek were During 2002–2007, the number of mature migratory adult commonly that size. Consequently, we used redd counts from females trapped in Mill Creek ranged from 32 to 90, and the Mill Creek and North Fork Mill Creek to compare with our esti- mean proportion of mature females was 0.51 (SD = 0.02). The mates of larger mature females from trap counts of migratory fe- FL of mature females was 430 mm (N = 408, SD = 38). Only males and unmarked females (≥300 mm) in our mark–recapture five mature females were smaller than 300 mm, and they were estimates above the trap. This was also corroborated by previous all 289 mm or larger. From our PIT tag capture histories of telemetry data indicating that large, radio-tagged adults spawned Mill Creek adults, only 8% of the recaptures at the upstream in main-stem Mill Creek and North Fork Mill Creek but not in trap (N = 576) were not in successive years, and all mature Low Creek (Hemmingsen et al. 2001b, 2001c). In any case, females initially captured had mature gonads when recaptured redds in the excluded tributaries (except Low Creek) accounted in following years. Unmarked, 300-mm or larger females in for less than 3% of the total redds each year. Differences in redd the mark–recapture estimate above the trap accounted for an sizes and the corresponding size of spawners were also consis- average of 21% (SD = 5) of the total number of females. The tent when we included the Little Minam River in the comparison. average lower CI for the mark–recapture estimate of unmarked = = females was 17% (SD 5) of the estimate, and the upper CI TABLE 2. Results of Dunn’s test (Q test statistic) for pairwise comparisons = of bull trout redd area at the sites listed in Table 1. was 29% (SD 5) of the estimate. Redd counts in Mill Creek and North Fork Mill Creek during 1998–2007 ranged from 56 Difference to 180 (Figure 2). The mean number of females per redd during Comparison of ranks QP< 0.01 that period was 0.9 (SD = 0.2). In 2003 and 2005, we estimated that there were 48 (95% Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 Mill Creek versus Low Creek 397.89 21.12 Yes CI = 27–69) and 97 (95% CI = 69–125) mature females, Mill Creek versus Bull Creek 267.42 3.92 Yes respectively, in Low Creek. Mature females ranged from 142 Mill Creek versus Paradise 246.02 3.85 Yes to 198 mm, and mature males ranged from 135 to 193 mm. Sex Creek ratios were slightly skewed toward males (44–47% of the fish Mill Creek versus North Fork 35.70 0.86 No were female). The numbers of redds counted in Low Creek in Mill Creek those same years were 28 and 43, respectively, yielding 1.7 and North Fork Mill Creek versus 362.18 8.48 Yes 2.3 females/redd—about twice the ratio for Mill Creek. Low Creek We recaptured 56% of the marked bull trout (61% in pools, Paradise Creek versus Low 151.86 2.34 No 46% in riffles) during the second electrofishing depletion esti- Creek mate from the subsampled units where we used mark–recapture Mill Creek versus Low Creek 503.41 20.78 Yes to evaluate potential bias in the depletion estimates on which Mill Creek versus Little 503.40 21.95 Yes our population sizes for Low Creek were based. However, we Minam River captured only one unmarked bull trout, which was smaller than Little Minam River versus 0.01 0.00 No 130 mm. The absence of any unmarked fish in the size range Low Creek of mature females during the recapture and the generally higher BULL TROUT POPULATION SIZE AND TREND 7 Downloaded by [Department Of Fisheries] at 01:34 01 March 2012

FIGURE 3. Mean redd counts, expressed as a percentage of the best estimate of the true number of redds, as determined by experienced and inexperienced surveyors in test reaches containing redds of large, ﬂuvial adult bull trout (Mill Creek) and redds of small, resident bull trout (Low Creek).

electrofishing efficiencies for larger-sized fish (Peterson et al. estimate counted by all experienced surveyors, which represents 2004) suggest that our depletion estimates of mature females how redds counts would be totaled in a typical survey, was 99%. were not substantially biased. In Low Creek, where only smaller, suspected-resident-sized fish spawned, redd counts of the experienced surveyor, who Redd Count Measurement Error conducted all of the annual redd counts in that stream during Redd counts of individual experienced surveyors used to the study, underestimated the total redds in the test reaches by conduct spawning surveys in Mill Creek were 67–122% of the an average of 45% (Figure 3). Adjusting the redd counts by that best estimate of the number of fluvial bull trout redds in the three amount produced females-per-redd ratios (0.9, 1.2) similar to test reaches (Figure 3). The grand mean proportion of the best those in Mill Creek. Redd counts of inexperienced surveyors for 8 HOWELL AND SANKOVICH

200 5.0 5.2

175 4.8 5.0

150 4.6 4.8

125 4.4 4.6

Redds 100 Ln Redds

Ln Females 4.2 4.4

75 + Females 4.0 O Redds 4.2 50

3.8 4.0 25 25 50 75 100 125 150 Females 3.6 3.8 1998 2000 2002 2004 2006 FIGURE 4. Linear regression (dashed lines = 95% confidence interval) of the Year number of redds against the number of adult female bull trout in Mill Creek, 1998–2007 (redds = 12.7 + [1.04 × females]; r2 = 0.47; P = 0.03). FIGURE 5. Linear and polynomial regressions of loge(redd count) and loge(adult females) for bull trout in Mill Creek over time (1998–2007). The 2 linear equations are loge(females) = 167.52 – (0.08 × year) (r = 0.49, P = = − × 2 = = the test reaches in Mill Creek, which are used here only for com- 0.024) and loge(redds) 184.80 (0.09 year) (r 0.48, P 0.027). The polynomial equations are loge(females) = 4.42 + (0.19 × year) − (0.02 × parison purposes, deviated substantially more from the best esti- 2 2 = = = + × − = year )(r 0.78, P 0.005) and loge(redds) 4.479 (0.247 year) mate, were consistently positively biased (mean 161%, range (0.03 × year2)(r2 = 0.83, P = 0.002). = 117–211%), and were more variable (experienced: CV = 0.25; inexperienced: CV = 0.49; Figure 3). In Low Creek, inexperienced surveyors consistently underestimated the from the linear regression for female counts. If the significance number of redds to a greater extent (68%) than the experienced criteria are relaxed by using an α of 0.20 with lower certainty surveyor, and counts were more variable (CV = 0.56) than and a one-tailed test to be able to distinguish between declines those for the redds of fluvial bull trout. versus stable or increasing trends, detection of declines of 28% over 15 years are possible using the variation in females per Population Size and Trend redd observed in Mill Creek. Use of inexperienced surveyors Linear regression indicated the relationship between redds would require a minimum decline of 70% or an increase of and females in Mill Creek was significant but not highly corre- 224% over 15 years. Minimum detectable changes over 5 years lated (Figure 4). However, separate linear regressions for redds are quite large, particularly using counts of novice surveyors, and females over time revealed highly similar, significant declin- which require almost a total collapse of the population or an ing trends in the population (Figure 5); Durbin–Watson statistics 11-fold increase. for females (1.32, P = 0.049) and redds (0.96, P = 0.008) sug- Redd counts and abundance estimates for Mill Creek sup- gested that the data were autocorrelated. Quadratic polynomial port the difficulty of accurately assessing trends over a period regressions reduced apparent autocorrelation (Durbin–Watson of 5 years. If we examine the data for 1998–2005, some of Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 statistic; females: 2.11, P = 0.233; redds: 2.29, P = 0.344), the potential short-term discrepancies between redd and female significantly improved the fit of the data (F-statistic for the in- counts become apparent. Females estimates during 1998–2005 crease in r2; females: P = 0.010; redds: P = 0.003), and likewise were stable based on log-linear regression (r2 = 0.05, P = 0.59, showed very similar trends (analysis of covariance: P = 0.623). coefficient =−0.01). However, the trend in the redd counts dur- The variation in the annual number of females per redd in ing the first 5 years of this period and the trend during the last Mill Creek and among test counts of the surveyors was similar 5 years suggest different conclusions. The trend in redd counts (CV in Table 3). The minimum detectable changes in abundance for 1998–2002 indicates a significant, steeply increasing trend over 15 years of monitoring based on that variation would be a (r2 = 0.84, P = 0.03, coefficient = 0.14), whereas the trend in decrease of 44–47% or an increase of 78–87%, assuming typical redd counts for 2001–2005 conversely indicates a significant, statistical criteria (α = 0.05, power = 0.8, and the two-tailed steeply declining trend with similar statistics (r2 = 0.85, P = test) and no change in CV with abundance (Table 3). The trend 0.03, coefficient =−0.19). Thus, neither of these 5-year trends in Mill Creek redd counts from the linear regression (Figure 5) in redd counts is consistent with the population trend. was equivalent to a 55% decrease in abundance over 10 years, In terms of abundance, geometric means of the Mill Creek which is consistent with the minimum detectable decline of 52% population size predicted from redd counts based on mean adults over 10 years predicted in Table 3 and similar to the 52% decline per redd observed during 1998–2007 were similar to direct BULL TROUT POPULATION SIZE AND TREND 9

TABLE 3. Minimum detectable changes (%) in adult bull trout abundance for a given number of years of monitoring, based on ﬂuvial redd counts in Mill Creek. Parameters in bold italics highlight the differences in the scenarios (CV = coefﬁcient of variation). Monitoring period (years) Scenario Parameter or condition Type of change 5 10 15 20 25 Variation in females/redd CV = 0.23 α = 0.05 Two-tailed test Increase 239 105 78 64 56 Decrease −68 −52 −44 −40 −37

Variation in redd counts by experienced CV = 0.25 observer in test reaches α = 0.05 Two-tailed test Increase 276 118 87 71 62 Decrease −74 −56 −47 −42 −39

Variation in redd counts by novice CV = 0.49 observer in test reaches α = 0.05 Two-tailed test Increase 1,115 335 224 175 147 Decrease −92 −78 −70 −64 −60

Variation in females/redd and a one-tailed CV = 0.23 test with reduced signiﬁcance level α = 0.20 One-tailed test Decrease −41 −32 −28 −25 −22

estimates of adults (within 3–12% of the geometric mean) when mean adult estimates for Mill Creek would not be statistically averaged over a 5- or 10-year period of record (Figure 6). If av- different from the direct adult estimate, except for 1998–2002 erage values of adults per redd from the literature were used in when using the highest value of adults per redd (Holm–Sidak the absence of estimates for the population, predicted geometric multiple-comparison test). Likewise, the population estimate relative to the abundance classes used to evaluate status would 600 generally be similar except in some cases when using the highest

Adults adults-per-redd value from the literature (Figure 6). 500 Redd-predicted adults @ 1.76 adults/redd Redd-predicted adults @ 2.16 adults/redd Redd-predicted adults @ 2.68 adults/redd 400 DISCUSSION There are a number of possible factors inﬂuencing variability Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 300 in the relationship of redd counts to female or adult counts and

Number the reliability of using redd counts to estimate adult abundance

200 and trend. First, there is measurement error in both counts. We attempted to minimize error in the adult female count to the

100 extent possible. There was likely little error in our trap counts of migratory adults moving upstream prior to spawning, which accounted for an average of 79% of the female adults, or in 0 98 - 02 03 - 07 98 - 07 our identification of adult females by use of ultrasound (Evans et al. 2004). Our snorkel counts used to estimate the number Years of nonmigratory adult females (≥300 mm FL) in the spawning FIGURE 6. Geometric mean (±95% confidence interval) adult bull trout population were largely a complete survey of potential holding abundance in Mill Creek during three periods (1998–2002, 2003–2007, and habitat above the dam, so there was no error associated with 1998–2007) from direct estimates and from predictions using redd counts for sample reach selection. There is error associated with estimating Mill Creek (1.76 adults/redd) and averages from Dunham et al. (2001; 2.16 the size of fish while snorkeling (O’Neal 2007). However, we adults/redd) and Al-Chokhachy et al. (2005; 2.68 adults/redd). Dashed lines < ≥ delineate abundance classes (0–50, 50–250, and 250–1,000 adults; categories used only two size-classes ( 300 and 300 mm FL). Data from IUCN 2001 and USFWS 2008). from trapping, electrofishing, and other snorkeling indicated 10 HOWELL AND SANKOVICH

that there were few 200–299-mm bull trout in the area sampled, Other studies have suggested higher correlation between so there should have been little uncertainty in classifying the redds and spawning adults than we observed (Dunham et al. fish observed into the two size-groups since most fish would 2001; Johnston et al. 2007). This may be at least partially due to have been at least 300 mm or would have been substantially differences in the range of redd counts and spawner abundance smaller. Our maturity data also indicated that very few mature, used in the analyses. In Mill Creek, there was a threefold range nonmigratory females smaller than 300 mm were excluded from in both redd and female counts, whereas there was a 10-fold our mark–recapture estimates. range in the counts used by Dunham et al. (2001; <100 to Measurement error in redd counts can be substantial >1,000), although the variation in spawners per redd in that (Dunham et al. 2001). As suggested by Muhlfeld et al. (2006), analysis (CV = 0.46; J. Dunham, U.S. Geological Survey, we attempted to minimize this source of error by primarily unpublished data) was much greater than for Mill Creek (CV = using a consistent crew of experienced surveyors. The range in 0.23). Spawners and redds in Smith-Dorrien Creek, Alberta, average differences in individual surveyor counts from the best where there was a 1:1 relationship between spawning females estimate of redds (67–122%) in test reaches in Mill Creek was and redd counts (Johnston et al. 2007), increased more than similar to that of experienced surveyors in Montana (78–130%; 20-fold from 1992 to 1998 (Johnston 2005). As indicated by Muhlfeld et al. 2006). The deviation of the combined counts the power analysis, the detectability of changes in abundance of the surveyors from the best estimate (–1%) suggested little based on redd counts depends on the magnitude of change. bias in the total redd counts, assuming the errors were similarly The size of the spawning fish and corresponding redd size and distributed among years. Differences among surveyors in errors detectability may also account for some of the variability in redd from undercounting (missed redds) and overcounting (false : adult relationships among populations. For example, adfluvial identifications) tended to cancel each other out, as occurred for bull trout spawners in Smith-Dorrien Creek (Johnston et al. Dunham et al. (2001). In the Swan River, equal compensation in 2007) average about 600 mm when they first spawn (Stelfox the two types of error occurred at densities of about 5 redds/km 1997) and create redds that are approximately 2 m2 (Johnston (Muhlfeld et al. 2006). Average redd density in the Mill Creek 2005). Likewise, for the primarily adfluvial bull trout spawning test reaches was 6 redds/km. in portions of the Metolius River and tributaries (included in Our power analysis assumes lognormal measurement error the Dunham et al. 2001 analysis), redd area averaged 3–5 m2 in the redd counts. That may not be the case (Muhlfeld et al. (Higgins et al. 2005) and female spawners averaged 578 mm 2006), which could lead to overestimating the power of log- (Riehle et al. 1997). For adfluvial bull trout in the Flathead River, linear regression to detect trends. Since we did not have separate Montana, redd area averaged 2.3–3.7 m2 and spawners averaged estimates of missed redds versus falsely identified redds, we 611 mm (Shepard et al. 1984). In Mill Creek, redds and spawners could not determine the error structure in a manner similar to were smaller (mean redd area = 1.4 m2; mean FL of mature that described by Muhlfeld et al. (2006). females = 430 mm). As was evident in Low Creek, where redds One factor related to surveyor error that was not evaluated averaged 0.2 m2 and where adults were smaller than 200 mm, in this study or in previous studies is the year-to-year variation smaller redds probably increase the error associated with redd in surveyor error. For example, estimates of observation error counts. We previously examined the relationship between small in redd counts for this study and the study by Muhlfeld et redds and adults (132–214 mm) in another apparent resident al. (2006) are based on a single evaluation and assume that population (Hemmingsen et al. 2001b). A population estimate there were no differences in error among years. However, the and redd counts based on methods similar to those used for Low variation in the females per redd over 10 years (CV = 0.23) Creek resulted in 42 adults/redd, suggesting very high negative in Mill Creek, which includes all sources of measurement bias in the redd counts. This was attributed to a small redd size Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 error plus any true annual variation in the number of redds coupled with a fine, granitic spawning substrate that made the constructed by females, was similar to the variation among the redds very difficult to detect. experienced surveyors in the 1 year of test counts (CV = 0.25), For this study, we were fortunate to have discrete resident suggesting little year-to-year variation in surveyor error. versus migratory populations to evaluate separately. This would There was substantially more measurement error in the be more complicated for populations consisting of a mix of counts by inexperienced surveyors of both fluvial and resident migratory and resident females. Small, presumably “sneaker” bull trout redds. These were not totally novice surveyors, how- males (∼150–250 mm) were observed near redds occupied by ever. In fact, the 8-week period of training and experience of this larger (>300 mm) females and males in Mill Creek. Similar group at the time of evaluation was considerably greater than breeding behavior involving small males has been observed the training period provided to many surveyors that are hired in other migratory bull trout populations (McPhail and Baxter to conduct redd surveys of other bull trout populations in the 1996) and is often observed in other large, migratory salmonids region (Moore et al. 2006). This suggests that the bias and im- (Esteve 2005). Small, mature males were also well distributed in precision of the inexperienced surveyors in our study were not our sample from Mill Creek and its tributaries. However, their higher, and could have even been lower, than would be expected presence would not in itself constitute a resident population, for other inexperienced surveyors. which would also have to include resident adult females. BULL TROUT POPULATION SIZE AND TREND 11

Understanding not only the relationship between redds have similar life history characteristics (e.g., migratory versus and spawners but also the relationship between redds and the resident, alternate-year spawning) should be used. total number of adults can be useful in estimating effective Error in redd counts can come from a variety of sources, population size and determining abundance criteria for recovery including aspects of survey design (e.g., spatial and temporal plans (Rieman and Allendorf 2001). For example, bull trout distribution) and observer variability. Consequently, estimates may spawn in successive years or in alternate years (Pratt of that uncertainty should be included with redds counts similar 1992). Consequently, any alternate-year spawners that did not to other methods for estimating abundance, such as depletion migrate above our trap in years when they did not spawn would and mark–recapture, to help with their interpretation. not be accounted for in our adult estimates. However, our data Given the limitations of redd counts, a precautionary indicated that in Mill Creek, almost all of the adults that over- approach in using them to assess status, particularly for an wintered below the trap migrated upstream to the spawning area imperiled species like the bull trout, appears warranted. For and were mature annually. In Flathead Lake, on the other hand, example, using lower levels of statistical significance and where alternate-year spawning is more prevalent, about 40% of a one-tailed test for trend analysis, as suggested by Maxell the adult population remained in the lake and did not spawn in a (1999), would permit detecting lower rates of decline or reduce given year (Fraley and Shepard 1989). Consequently, adult esti- the time needed to detect those declines, while still being mates based solely on spawner : redd relationships or trap counts consistent with the recovery objective of distinguishing between might exclude the nonspawning and nonmigratory adults that declines and stable-to-increasing populations. Likewise, use are present below or above the trap, as occurred in Mill Creek. of conservative adult : redd conversion factors for estimating abundance where population-specific data are lacking will Conclusions reduce the likelihood of overestimating the abundance of the Variation in the relationship between redd counts and adult population. This would be particularly important in the case of abundance can limit the use of redd counts in detecting trends small, depressed populations since they are more vulnerable to in abundance or estimating adult abundance. Detection of short- extirpation and since redd counts at low redd densities tend to term trends based on redd counts (e.g., 5 years) can be mis- be inflated (Muhlfeld et al. 2006). leading and likely requires extreme changes in abundance to be Two important factors influencing the variability of redd accurate. Guidelines for evaluating trends for status assessment counts are redd size, which is related to adult size and life (Morris et al. 1999; IUCN 2001; USFWS 2008) recommend a history form, and measurement error in redd surveys. Counts minimum of 10 years of data and preferably data for three gen- of redds produced by small, presumably resident adults greatly erations, which for bull trout has been interpreted as a 15-year underestimated the abundance of redds and adults and can be period based on 5 years/generation (Maxell 1999). From our re- more variable. Consequently, redd counts are likely to be a more sults, detection of trends over 10–15 years by using traditional reliable and sensitive indicator of abundance for larger, migra- statistical criteria would be possible for only the higher-risk tory forms than for small, resident forms. Redd surveys that status categories defined by the IUCN (2001) and the USFWS include estimates of redd area and spawner lengths could be (2008). However, our study also suggests greater potential power used to sort migratory versus resident redds and to describe the of redd counts to detect population trends than previously re- size and distribution of life history forms, which are also useful ported. For example, the CVs in the redd counts for the Flathead attributes in assessing the status of populations. and Swan rivers were considerably higher (0.42–0.67; Maxell Previous studies (Dunham et al. 2001; Muhlfeld et al. 2006) 1999) than we observed and would require correspondingly and the data from this study point out the importance of us- larger changes or a longer time to detect. However, this differing skilled redd surveyors to reduce measurement error. Redd Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 ence was not because redds counts for Mill Creek were “better” surveys are not a purely mechanical, objective method that is but because the CVs used for Mill Creek were directly deter- immune to operator error; and skill is not solely determined mined from variation in the female : redd ratios and measure- by experience. Although experienced surveyors generally pro- ment error in the redd counts, whereas the CVs for the Flathead duced redd counts that were more accurate and precise in the and Swan rivers were based on annual changes in redd counts, test reaches in this study, a mix of experienced and inexperi- which included large declines. enced surveyors had the fewest counting errors in the study by Abundance estimates derived from redd counts and the num- Dunham et al. (2001). The use of skilled surveyors across the ber of adults per redd can provide a confidence band that in- range of the species may be very difficult. Hiring, training, and cludes the actual size of the population. Like trend analysis, retaining skilled surveyors will clearly be a challenge. longer-term data sets were more informative. Ten-year geo- Regardless of the skill of the surveyors or the approach that metric means have been similarly used to assess abundance of is used to analyze redd counts, the reliability of redd counts in salmon and steelhead populations listed under the Endangered assessing status is dependent on a well-designed and executed Species Act in the interior Columbia River basin (ICTRT 2007). sampling plan. Bias in selecting areas to survey (e.g., index Where direct estimates of adults per redd are not available for reaches) or in the timing and frequency of surveys (e.g., peak the population of interest, values from other populations that counts) may increase the error in the counts and the variability 12 HOWELL AND SANKOVICH

in their relationship to abundance. Other abundance monitor- Goetz, F. A. 1989. Biology of the bull trout Salvelinus confluentus: a literature ing techniques may be helpful in evaluating assumptions made review. U.S. Forest Service, Willamette National Forest, Eugene, Oregon. using redd counts (e.g., adults per redd) and may be more suit- Hankin, D. G., and G. H. Reeves. 1988. Estimating total fish abundance and total habitat area in small streams based on visual estimation methods. Canadian able for some bull trout populations (e.g., those occurring sym- Journal of Fisheries and Aquatic Sciences 45:834–844. patrically with brook trout Salvelinus fontinalis) and for certain Hemmingsen, A. R., B. L. Bellerud, S. L. Gunckel, and P. J. Howell. 2001a. spawning conditions (e.g., substrate type). Likewise, monitoring Bull trout life history, genetics, habitat needs, and limiting factors in central approaches that assess the risk of decline rather than measure and northeast Oregon. 1998 Annual Report to the Bonneville Power Admin- abundance or trend can be useful in situations where the error istration, Project 199405400, Portland, Oregon. Hemmingsen, A. R., S. L. Gunckel, and P. J. Howell. 2001b. Bull trout life in abundance estimates is high and the power to detect trends is history, genetics, habitat needs, and limiting factors in central and northeast low (Staples et al. 2005). Oregon. 1999 Annual Report to the Bonneville Power Administration, Project 199405400, Portland, Oregon. ACKNOWLEDGMENTS Hemmingsen, A. R., S. L. Gunckel, and P. M. Sankovich. 2001c. Bull trout life history, genetics, habitat needs, and limiting factors in central and northeast Funding was provided by the Bonneville Power Administra- Oregon. 2000 Annual Report to the Bonneville Power Administration, Project tion, U.S. Forest Service, and U.S. Fish and Wildlife Service. 199405400, Portland, Oregon. Larry Boe, Steve Starcevich, Alan Hemmingen, Jason Shappart, Hemmingsen, A. R., S. L. Gunckel, J. K. Shappart, B. L. Bellerud, D. V. Stephanie Gunckel, John Brunzell, Ian Tattam, David Gaudette, Buchanan, and P. J. Howell. 2000. Bull trout life history, genetics, habi- Lisa Gaudette, Ari Martinez, Blane Bellerud, David Crabtree, tat needs, and limiting factors in central and northeast Oregon. 1997 An- nual Report to the Bonneville Power Administration, Project 199405400, Mae Noble, Devin Best, Dana McCoskey, and David Kwas- Portland, Oregon. niewski assisted with collection of field data. David Buchanan, Higgins, A., T. G. Wise, and S. Jacobs. 2005. Timing and location of spawn- Alan Hemmingsen, and Steve Jacobs provided administrative ing by bull trout and kokanee in the Odell Lake watershed and Heising support. Thanks to Jason Dunham, Russ Thurow, Paul Wilson, Spring of the Metolius River watershed, 2003–04. Oregon Department of Fish and Wildlife, Information Reports 2005–04, Corvallis. Available: nrimp.dfw. Howard Schaller, Clint Muhlfeld, and an anonymous reviewer state.or.us/crl/Reports/Info/TrapperInfoReport2005.pdf. (April 2011). for helpful suggestions on previous drafts. Reference to trade ICTRT (Interior Columbia Technical Recovery Team). 2007. Viability crite- names does not imply endorsement by the U.S. Government. ria for application to interior Columbia basin salmonid ESUs. NOAA Na- tional Marine Fisheries Service, Northwest Regional Office, Seattle. Avail- able: www.nwfsc.noaa.gov/trt/trt viability.cfm. (April 2011). REFERENCES IUCN (International Union for Conservation of Nature). 2001. IUCN red Al-Chokhachy, R., P. Budy, and H. Schaller. 2005. Understanding the signifi- list categories and criteria, version 3.1. IUCN, Cambridge, UK. Available: cance of redd counts: a comparison between two methods for estimating the www.iucn.org/themes/ssc/red-lists.htm. (April 2011). abundance of and monitoring bull trout populations. North American Journal Isaak, D. J., and R. F. Thurow. 2006. Network-scale patterns of spatial and of Fisheries Management 25:1505–1512. temporal variation in Chinook salmon redd distributions: patterns inferred ASRD/ACA (Alberta Sustainable Resource Development and Alberta Conser- from spatially continuous replicate surveys. Canadian Journal of Fisheries vation Association). 2009. Status of the bull trout (Salvelinus confluentus) and Aquatic Sciences 63:285–296. in Alberta: update 2009. ASRD, Wildlife Status Report 39 (Update 2009), Jacobs, S. E., and T. E. Nickelson. 1998. Use of stratified random sampling Edmonton. to estimate the abundance of Oregon coastal coho salmon. Oregon Depart- Courbois, J. Y., S. L. Katz, D. J. Isaak, E. A. Steel, R. F. Thurow, A. M. ment of Fish and Wildlife, Fish Research Project F-145-R-09, Final Report, Wargo Rub, T. Olsen, and C. E. Jordan. 2008. Evaluating probability sam- Portland. pling strategies for estimating redd counts: an example with Chinook salmon Jacobs, S. E., W. Gaeuman, M. A. Weeber, S. L. Gunckel, and S. J. Starcevich. (Oncorhynchus tshawytscha). Canadian Journal of Fisheries and Aquatic Sci- 2009. Utility of a probabilistic sampling design to determine bull trout popu- ences 65:1814–1830. lation status using redd counts in basins of the Columbia River plateau. North Dunham, J., B. Rieman, and K. Davis. 2001. Sources and magnitude of sampling American Journal of Fisheries Management 29:1590–1604.

Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 error in redd counts for bull trout. North American Journal of Fisheries Johnston, F. D. 2005. Demographic and life-history responses of an over- Management 21:343–352. exploited bull trout (Salvelinus confluentus) population to zero harvest regu- Esteve, M. 2005. Observations of spawning behaviour in Salmoninae: Salmo, lations. Master’s thesis. University of Calgary, Alberta. Oncorhynchus,andSalvelinus. Reviews in Fish Biology and Fisheries Johnston, F. D., J. R. Post, C. J. Mushens, J. D. Stelfox, A. J. Paul, B. Lajeunesse. 15:1–21. 2007. The demography of an over-exploited bull trout (Salvelinus confluentus) Evans, A. F., M. S. Fitzpatrick, and L. K. Siddens. 2004. Use of ultrasound population. Canadian Journal of Fisheries and Aquatic Sciences 64:113–126. imaging and steroid concentrations to identify maturational status in adult Maxell, B. A. 1999. A prospective power analysis on the monitoring of bull trout steelhead. North American Journal of Fisheries Management 24:967–978. stocks using redd counts. North American Journal of Fisheries Management Fraley, J. J., and B. B. Shepard. 1989. Life history, ecology and population status 19:860–866. of bull trout (Salvelinus confluentus) in the Flathead lake and river system, McPhail, J. D., and J. S. Baxter. 1996. A review of bull trout (Salvelinus confluen- Montana. Northwest Science 63:133–143. tus) life-history and habitat use in relation to compensation and improvement Gallagher, S., P.K. Hahn, and D. H. Johnson. 2007. Redd counts. Pages 197–234 opportunities. University of British Columbia, Fisheries Management Report in D. H. Johnson, B. M. Shrier, J. S. O’Neal, J. A. Knutzen, X. Augerot, T. 104, Vancouver. A. O’Neil, and T. N. Pearsons, editors. Salmonid field protocols handbook: Moore, T. L., S. J. Starcevich, S. E. Jacobs, and P. J. Howell. 2006. Migratory techniques for assessing status and trends in salmon and trout populations. patterns, structure, abundance, and status of bull trout populations from sub- American Fisheries Society, Bethesda, Maryland. basins in the Columbia Plateau and Blue Mountain provinces. 2005 Annual Gerrodette, T. 1993. Program TRENDS: user’s guide. Southwest Fisheries Sci- Report to the Bonneville Power Administration, Project 199405400, Portland, ence Center, La Jolla, California. Oregon. BULL TROUT POPULATION SIZE AND TREND 13

Morris,W.,D.Doak,M.Groom,P.Kareiva,J.Fieberg,L.Gerber,P.Mur- Rieman, B. E., and D. L. Myers. 1997. Use of redd counts to detect trends phy, and D. Thomson. 1999. A practical handbook for population viability in bull trout (Salvelinus confluentus) populations. Conservation Biology analysis. The Nature Conservancy, Arlington, Virginia. 11:1015–1018. Muhlfeld, C. C., M. L. Taper, D. F. Staples, and B. B. Shepard. 2006. Observer Seber, G. A. F. 1982. Estimation of animal abundance and related parameters. error structure in bull trout redd counts in Montana streams: implications Oxford University Press, New York. for inference on true redd numbers. Transactions of the American Fisheries Shepard, B., K. Pratt, and J. Graham. 1984. Life histories of westslope cut- Society 135:643–654. throat and bull trout in the upper Flathead River basin, Montana. Montana Mullan, J. W., K. Williams, G. Rhodus, T. Hillman, and J. McIntyre. 1992. Department of Fish, Wildlife, and Parks, Kalispell. Production and habitat of salmonids in mid-Columbia River tributary streams. Staples, D. F., M. L. Taper, and B. B. Shepard. 2005. Viable population moni- U.S. Fish and Wildlife Service, Monograph 1, Washington, D.C. toring. Conservation Biology 19:1908–1916. Murdoch, A. R., T. N. Pearsons, and T. W. Maitland. 2009. The number of redds Stelfox, J. D. 1997. Seasonal movements, growth, survival and population status constructed per female spring Chinook salmon in the Wenatchee River basin. of the adfluvial bull trout population in lower Kananaskis Lake, Alberta. Pages North American Journal of Fisheries Management 29:441–446. 309–316 in W. C. Mackay, M. K. Brewin, and M. Monita, editors. Friends NPCC (Northwest Power and Conservation Council). 2004. Walla Walla sub- of the bull trout conference proceedings. Trout Unlimited Canada, Bull Trout basin plan. NPCC, Portland, Oregon. Available: www. nwcouncil.org/fw/ Task Force (Alberta), Calgary. subbasinplanning/wallawalla/plan/. (April 2011). Stevens, D. L., D. P. Larsen, and A. R. Olsen. 2007. The role of sample sur- O’Neal, J. S. 2007. Snorkeling. Pages 325–339 in D. H. Johnson, B. M. Shrier, veys: why should practitioners consider using a statistical sampling design? J. S. O’Neal, J. A. Knutzen, X. Augerot, T. A. O’Neil, and T. N. Pearsons, Pages 11–23 in D. H. Johnson, B. M. Shrier, J. S. O’Neal, J. A. Knutzen, editors. Salmonid field protocols handbook: techniques for assessing status X. Augerot, T. A. O’Neil, and T. N. Pearsons. Salmonid field protocols and trends in salmon and trout populations. American Fisheries Society, handbook: techniques for assessing status and trends in salmon and trout Bethesda, Maryland. populations. American Fisheries Society, Bethesda, Maryland. Peterson, J. T., R. F. Thurow, and J. W. Guzevich. 2004. An evaluation of Swenson, E. A., A. E. Rosenberger, and P. J. Howell. 2007. Validation of multipass electrofishing for estimating the abundance of stream-dwelling endoscopy for determination of maturity in small salmonids and sex of salmonids. Transactions of the American Fisheries Society 133:462–475. mature individuals. Transactions of the American Fisheries Society 136: Pratt, K. L. 1992. A review of bull trout life history. Pages 5–9 in P. J. Howell 994–998. and D. V. Buchanan, editors. Proceedings of the Gearhart Mountain bull trout Taylor, G., and A. Reasoner. 2000. Bull trout Salvelinus confluentus population workshop. American Fisheries Society, Oregon Chapter, Corvallis. and habitat surveys in the McKenzie and Middle Fork Willamette basins. Ricker, W. E. 1975. Computation and interpretation of biological statistics of fish 1999. Bonneville Power Administration, Project 1995-05300, Portland, populations. Bulletins of the Fisheries Research Board of Canada 191:1–382. Oregon. Riehle, M., W. Weber, A. M. Stuart, S. L. Thiesfeld, and D. E. Ratliff. 1997. USFWS (U.S. Fish and Wildlife Service). 1999. Determination of threatened Progress report of the multiagency study of bull trout in the Metolius River status for bull trout in the coterminous United States. Federal Register 64:(1 system, Oregon. Pages 137–144 in W. C. Mackay, M. K. Brewin, and M. November 1999):58910–58933. Monita, editors. Friends of the bull trout conference proceedings. Trout Un- USFWS (U.S. Fish and Wildlife Service). 2002. Bull trout (Salvelinus conflu- limited Canada, Bull Trout Task Force (Alberta), Calgary. entus) draft recovery plan. USFWS, Portland, Oregon. Rieman, B. E., and F. W. Allendorf. 2001. Effective population size and ge- USFWS (U.S. Fish and Wildlife Service). 2008. Bull trout (Salvelinus conflu- netic conservation criteria for bull trout. North American Journal of Fisheries entus) 5-year review: summary and evaluation. USFWS, Portland, Oregon. Management 21:756–764. Zimmerman, C. E., and G. H. Reeves. 2000. Population structure of sym- Rieman, B. E., and J. D. McIntyre. 1993. Demographic and habitat requirements patric anadromous and nonanadromous Oncorhynchus mykiss: evidence from for conservation of bull trout. Intermountain Research Station, General Tech- spawning surveys and otolith microchemistry. Canadian Journal of Fisheries nical Report INT-302, Ogden, Utah. and Aquatic Sciences 57:2152–2162. Downloaded by [Department Of Fisheries] at 01:34 01 March 2012 This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:36 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 Dam Removal and Implications for Fish Health: Ceratomyxa shasta in the Williamson River, Oregon, USA Charlene N. Hurst a , Richard A. Holt a & Jerri L. Bartholomew a a Department of Microbiology, Oregon State University, 220 Nash Hall, Corvallis, Oregon, 97331, USA Available online: 08 Feb 2012

To cite this article: Charlene N. Hurst, Richard A. Holt & Jerri L. Bartholomew (2012): Dam Removal and Implications for Fish Health: Ceratomyxa shasta in the Williamson River, Oregon, USA, North American Journal of Fisheries Management, 32:1, 14-23 To link to this article: http://dx.doi.org/10.1080/02755947.2012.655843

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 32:14–23, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.655843

ARTICLE

Dam Removal and Implications for Fish Health: Ceratomyxa shasta in the Williamson River, Oregon, USA

Charlene N. Hurst, Richard A. Holt, and Jerri L. Bartholomew* Department of Microbiology, Oregon State University, 220 Nash Hall, Corvallis, Oregon 97331, USA

Abstract The removal of dams on a river is one potential tool for the ecological restoration of native salmonid fishes. However, the removal of barriers also introduces risks, such as the introduction of fish pathogens into previously isolated populations. The proposed removal of four dams on the Klamath River, Oregon–California, provides an opportunity for examining the disease risks associated with dam removal. A salmonid pathogen endemic to the region, Ceratomyxa shasta, is responsible for high mortality in juvenile Chinook salmon Oncorhynchus tshawytscha and coho salmon O. kisutch below the dams. Above the dams, parasite densities are lower and not implicated in salmonid mortality, except in the Williamson River tributary, where high parasite densities raise concerns over the restoration of anadromous fish that are likely to take advantage of spawning habitat in that river. In the current study, baseline information on parasite density, distribution, and genotype composition in the Williamson River was gathered to determine how salmonid reintroduction might be affected by parasite dynamics. Assay of water samples highlighted two areas of high parasite density: between the mouth of the Williamson River and the confluence of the Sprague River tributary, and above the Spring Creek confluence. Despite these high parasite densities, mortality did not occur in sentinel coho or Chinook salmon. Genetic analyses of parasites from water samples and infected fish demonstrated that C. shasta genotype II was dominant and was associated with stocked nonnative rainbow trout O. mykiss. The absence of pathogenicity of this parasite genotype for Chinook and coho salmon suggests that reintroduction plans will not initially be adversely affected by the high parasite densities in the Williamson River. However, following dam removal, returning adult salmon will transport parasite genotypes present below the dams upstream. These genotypes are likely to become established and may reach densities that could affect juvenile Chinook and coho salmon.

Anadromous salmonids require both freshwater and marine Positive effects of dam removal on ﬁsh populations have

Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 environments for the completion of their life history with fresh- been demonstrated in several rivers. On the Milwaukee River, water environments providing critical spawning and early life Wisconsin, removal of the Woolen Mills Dam resulted in im- stage rearing habitats. Over the last century, a variety of factors proved habitat and increased numbers of the native smallmouth have led to the decline of anadromous salmonids in the Pacific bass while nonnative common carp Cyprinus carpio declined Northwest, including habitat degradation brought about by agri- (Kanehl et al. 1997). Similarly, colonization of habitat post dam culture, the raising of livestock, forest usage, urbanization, and removal by anadromous salmonids occurred in the Mad River, dams. These effects have been further compounded by overhar- California; Rogue River, Oregon; and Clearwater River, Idaho vest, disease, and ocean conditions (Fresh 1997; Mundy 1997; (Winter 1990). However, there are also risks associated with Pearcy 1997). One solution proposed for mitigation of these dam removal. The release of large amounts of sediment from effects is dam removal, which would return the river to a more reservoirs could negatively affect fish populations, especially if natural hydrograph and provide access to historically utilized toxins have accumulated in the sediment (Stanley and Doyle habitats (Bednarek 2001). 2003), and introduction or amplification of pathogens may

*Corresponding author: [email protected] Received December 21, 2010; accepted August 17, 2011

14 DAM REMOVAL AND IMPLICATIONS FOR FISH HEALTH 15

occur as fish populations separated by the dams come into con- the actinospore stage (Bartholomew et al. 1997). Parasite dis- tact (Brenkman et al. 2008; Zielinski et al. 2010). tribution is seasonal, with infection occurring in sentinel fish Few case studies have compared pathogen dynamics and as- exposed between April and mid-December (Ratliff 1981; Ching sociated disease risks prior to and following dam removal or fish and Munday 1984; Hendrickson et al. 1989), coinciding with the passage. One study on the Deschutes River, Oregon, examined presence of adult spring Chinook salmon, the fish most likely to the risk of introducing Myxobolus cerebralis as a result of fish utilize upper basin habitat (Hamilton et al. 2005). passage above the dam (Zielinski et al. 2010, 2011). Prepassage In the upper Klamath basin, high densities of both C. shasta monitoring showed that M. cerebralis was established in at least and the polychaete host occur in the Williamson River, an area one tributary below the dams, thus increasing the risk of the of critical spawning habitat. From the mouth of the Williamson parasite being introduced above the dams. To improve the long- River to the Sprague River confluence (river kilometer [rkm] 19), term success of fish passage, the authors recommended transfer 90% parasite-induced mortality has been observed in nonnative of eggs or fry raised in pathogen-free water rather than passing (na¨ıve) rainbow trout O. mykiss (RBT; freshwater form) exposed naturally reared juveniles (Zielinski et al. 2010) and contin- in cages in the river (Hemmingsen et al. 1988; Buchanan et al. ued monitoring for M. cerebralis in areas where the pathogen 1989; Hallett and Bartholomew 2006; Stocking et al. 2006). As- was detected (Zielinski et al. 2011). On the Cowlitz River, say of water samples collected during sentinel studies in 2004 Washington, passage of adult spring Chinook salmon On- demonstrated that parasite densities in the Williamson River corhynchus tshawytscha above the Mayfield Dam began in were similar to those in the lower Klamath River where se- 1996. Although no studies were conducted to assess changes in vere disease occurs (Hallett and Bartholomew 2006). In addi- pathogen effects before and after fish passage, observational data tion, a survey for the polychaete host identified a large popu- suggests that prevalence of Ceratomyxa shasta has increased. lation at the Williamson River confluence with Klamath Lake Sentinel studies conducted in the late 1990s (shortly after rein- (Stocking and Bartholomew 2007). These studies demonstrated troduction began) demonstrated a less than 10% prevalence of that appropriate conditions exist for C. shasta to persist in the C. shasta in juvenile spring Chinook salmon upstream of the Williamson River. dam, while 70–84% infection prevalence occurred in juvenile Complicating predictions of how C. shasta will affect rein- spring Chinook collected from traps in 2010. In addition, mor- troduction plans is our changing understanding of the parasite. tality of spring Chinook salmon in Cowlitz Hatchery, located Four distinct genotypes of C. shasta (0, I, II, and III) occur in below the dam, has increased since fish passage (E. Ray, Wash- sympatry with their salmonid hosts (Atkinson and Bartholomew ington Department of Fish and Wildlife, personal communica- 2010a, 2010b) and are host-specific (Hurst 2010). Thus, the tion). A third study in the Elwha River, Washington, collected ability of the parasite to establish and reproduce is influenced baseline information on multiple fish pathogens prior to dam by the species and strain of salmonid hosts available. In the removal and will also assess changes in pathogen dynamics Williamson River, anadromous salmon have been extirpated post dam removal (Brenkman et al. 2008). The researchers hy- since 1917 (Hamilton et al. 2005), leaving redband trout O. pothesize that the prevalence of endemic pathogens (such as mykiss newberrii as the only native host for C. shasta. However, Renibacterium salmoninarum) may increase above the dams since 1925, nonnative RBT have been stocked into the Spring and that nonendemic pathogens (such as infectious hematopoi- Creek tributary of the Williamson River to supplement recre- etic necrosis virus) could be introduced. In addition, pathogens ational fishing. Fish stocking occurs during the peak months of that co-evolved with resident fish above the dams may be more parasite production—from May through August—and because virulent in na¨ıve, reintroduced stocks (Brenkman et al. 2008). of their high susceptibility to C. shasta, these fish are assumed Thus, pathogens may be influenced by interactions between res- to die if not caught by anglers (W. Tinniswood, Oregon Depart- Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 ident and reintroduced populations and could hinder salmonid ment of Fish and Wildlife [ODFW], personal communication). recovery in spite of dam removal. This strategy ensures that these fish will succumb to ceratomyx- In the Klamath River, Oregon–California, dam removal and osis before interbreeding with native redband trout but may also fish passage have been proposed for increasing salmonid habitat amplify certain parasite genotypes in the Williamson River. in the basin. However, the success of salmonid reintroduction in Understanding how biotic factors such as fish host species the Klamath River basin is contingent, in part, on their ability to and polychaete host densities, and abiotic factors such as tem- cope with the endemic parasite C. shasta. This parasite causes perature and management practices influence parasite dynamics ceratomyxosis, a disease of salmon and trout characterized by is important for predicting the success of reintroduced anadro- severe inflammation, hemorrhage, and necrosis of the intes- mous salmonids in the upper Klamath basin. The first objective tine (Bartholomew et al. 1989). Infection often results in death, of this study evaluates the risk of disease resulting from infection although fish native to rivers where the parasite is endemic are with C. shasta in the Williamson River. The approach was to less likely to develop the disease (Zinn et al. 1977; Bartholomew determine distribution, density, and genotype composition of C. 1998). Ceratomyxa shasta has a complex life cycle that requires shasta in fish and polychaete hosts and water samples collected two hosts, the salmonid host releasing the myxospore stage of throughout the Williamson River. The second objective exam- the parasite and the polychaete Manayunkia speciosa releasing ines how stocking practices may alter parasite dynamics. Here 16 HURST ET AL.

the approach was to compare parasite genotype composition, ments are met these fish are expected to repopulate their historic parasite density, and mortality of native and nonnative sentinel ranges, the Williamson River providing critical habitat. fish in the Williamson River and at a site downstream in the The Williamson River is a headwater of the Klamath River main-stem Klamath River where stocking does not occur. basin and flows into upper Klamath Lake. This river is approximately 120 rkm and is predominately a spring-fed system with inputs from two major tributaries, Sprague River and METHODS Spring Creek, with confluences at rkm 19 and 27.5, respectively. Water sources for the Sprague River include a combination of Density, Distribution, and Genotype Composition of snowmelt and cool springs, while Spring Creek is largely in- Ceratomyxa shasta in the Williamson River fluenced by cool springs (Gannett et al. 2007). For this study, Study location.—The Klamath River is 480 rkm and extends the main-stem Williamson River was divided into three reaches from the Williamson River in Oregon to the Pacific Ocean in based on geomorphology. Reach 1 comprised the area from the California. The upper and lower portions of the river are di- mouth of the Williamson River (rkm 0) to the Sprague River vided by a series of five dams. Copco I was completed in 1917, confluence (rkm 19), reach 2 included the area between the blocking fish passage above rkm 314. With installation of Iron Sprague River and Spring Creek (rkm 27.5), and reach 3 con- Gate dam in 1962, upstream fish passage was further restricted sisted of the area above Spring Creek to rkm 42. Twenty sites to the lower 307 rkm (Hamilton et al. 2005). The Klamath Basin were selected within the three reaches: six sites in reach 1, three Restoration Agreement (Klamath Restoration 2010a) and Kla- sites in reach 2, and seven sites in reach 3. In addition, two sites math Hydrologic Settlement Agreement (Klamath Restoration were sampled in each tributary. Sites were selected based on 2010b) propose increasing fish habitat in the upper basin by accessibility and spatial representation of the reaches. the removal of the four lower dams (Iron Gate, Copco I and Water sampling.—Four 1-L samples of water were collected II, and J. C. Boyle) and providing fish passage facilities at the from each site (Figure 1) in September 2008, June 2009, and fifth dam (Keno) beginning in 2020. Historically, the ranges of May and June 2010. These months were selected because spring Chinook salmon and coho salmon O. kisutch and steel- parasite density is highest from late spring to late summer head (anadromous rainbow trout) extended into the upper basin (Hendrickson et al. 1989). Thermometers were placed about (Hamilton et al. 2005), and after conditions in the above agree- 6 in below the surface, and Global Positioning System latitudes Downloaded by [Department Of Fisheries] at 01:36 01 March 2012

FIGURE 1. Klamath basin in relation to the United States on the left; Williamson River and tributaries on the right, where oval boundaries indicate river reaches, ﬁlled circles depict water sample sites, triangles represent sentinel ﬁsh exposure sites, and hollow boxes indicate collection sites for polychaete(Manayunkia speciosa). [Figure available in color online.] DAM REMOVAL AND IMPLICATIONS FOR FISH HEALTH 17

and longitudes were recorded for each site. Sites were the same Polychaete density and infection prevalence.—Areas sur- at all time points with the exception of rkm 36 in June 2009 to veyed for polychaetes were selected based on water sample replace sites at rkm 36.3 and 36.6 as high water levels prevented data (parasite densities of > 1 parasite/L within the area and no access. All samples were kept on ice during transport to the John parasite detected above the area indicate infected host popula- L. Fryer Salmon Disease Laboratory, Oregon State University, tions) and suitable polychaete habitat (fine benthic organic mat- Corvallis, Oregon (SDL). ter, boulders, or Cladaphora spp.; Stocking and Bartholomew Water was filtered 24–48 h after collection, and the filter 2007). Two areas fit these criteria: rkm 18.2–19.2, which strad- and its retentate were frozen at –20◦C until processed (Hallett dles the Sprague confluence (reaches 1 and 2), and rkm 33–34, and Bartholomew 2006). The processing procedure was mod- above Spring Creek (reach 3). To focus our search efforts within ified in 2010 to use acetone to dissolve the filter membrane each area, two 1-L water samples were collected ∼every 0.1 rkm instead of cutting the filters, leading to an increase in the sensi- and processed using qPCR as described above. Water samples tivity of the molecular assay described below. We extracted DNA from one 0.1-rkm segment within rkm 33–34 and two segments from three of the four replicate samples from each site during within rkm 18.2–19.2 contained parasite densities greater than each collection time. Each of the three samples was assayed 1 parasite/L. Ten substrate samples were collected from the in duplicate by quantitative polymerase chain reaction (qPCR; segment within rkm 33–34 and seven samples from the two seg- Hallett and Bartholomew 2006), along with negative (molecular ments within rkm 18.2–19.2 on July 20–21, 2010. Within rkm grade water) and positive controls (C. shasta-infected tissue), 18.2–19.2, where water was too deep to wade, two divers with using an ABI7300 qPCR sequence detection system. Interassay scuba collected benthic invertebrates. For sample collection, a variability was assessed from the standard deviations of the pos- scrub pad was used to dislodge benthic invertebrates and as- itive control samples. Quantitative cycle threshold (Cq) values sociated substrate contained within a Hess Sampler (256 cm2). (a measurement of parasite DNA) for the three site replicates Material was then collected in the sampler’s 80-µm aquatic were averaged, and tests for inhibition were conducted on one mesh. Samples were transferred to Whirl-Pak plastic bags, im- sample from each site during each collection time (Hallett and mediately preserved with a 95% solution of ethanol, and placed Bartholomew 2009). Samples were considered inhibited if the on ice for transport to the SDL. sample fluoresced later than the negative control in the presence Preserved samples were emptied into a 30 × 20-cm2 180- of a known amount of synthetic DNA. Inhibited samples were D10 tray (Wildco, Yulee, Florida) subdivided into 15 sections. reanalyzed with a 1:10 dilution of parasite DNA to molecular Three 5-cm2 subsamples were randomly selected, and material grade water if a difference of greater than one cycle occurred within each section was collected using a disposable pipette between the sample and the negative control. We determined and placed in 5-mL glass vials with a 1:3 ratio of Rose Ben- parasites per liter based on a one-spore standard reference sam- gal dye (20 mg/L 95% solution of ethanol) to a 95% solu- ple with a Cq value of 32.5. Sites with less than 1 parasite/L tion of ethanol. After 24 h, the sample was filtered through an were considered negative. Parasite density threshold values of 80-µm sieve to remove dye and then stored in a 95% solution 10 parasites/L and 100 parasites/L were extrapolated from a of ethanol. Polychaetes were sorted using a dissecting micro- one-spore standard and on previous correlations (Hallett and scope at 100 × magnification. This procedure was repeated for Bartholomew 2006). all three subsamples. Individual polychaetes were placed into Spore stage identification.—To determine the parasite stage 1.5-mL microcentrifuge tubes with ∼100 µL of a 95% solution (myxospore or actinospore) at locations where water samples of ethanol and refrigerated until processing. Polychaete den- tested positive (the qPCR assay does not differentiate between sities were calculated by averaging the three subsamples and spore stages), 5 L of water were collected at rkm 18.2 and 33 extrapolating to polychaetes per square meter. Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 in June 2009, in conjunction with the water sampling. Sites We extracted DNA from individual polychaetes according to were selected based on results from water samples collected in Stocking and Bartholomew (2007) with the following modifica- September 2008. Samples were transported on ice to the SDL, tions: DNA extraction buffer was reduced to 100 µL, proteinase and water was poured into separate 25-L fiberglass tanks with air K was reduced to 5 µL, and following the addition of RNase, stones. Ten C. shasta-susceptible rainbow trout (RBT; Roaring sample incubation at 37◦C was extended to 1 h. After boiling, River Hatchery, Scio, Oregon) were placed in each tank. After 72 pools of five polychaetes were created by adding 2 µL of each h of exposure, flow into each tank was resumed with pathogen- sample into a tube (10 µL total), then adding 2 µL of the pooled free water and fish were monitored daily for signs of disease. DNA into 198 µL of distilled water. Pooled DNA was assayed Moribund fish were euthanatized using tricaine methanesul- for C. shasta by qPCR (Hallett and Bartholomew 2006). Poly- fonate (MS-222; Argent Laboratories, Redmond, Washington), chaetes from positive pools were then assayed individually by and a swab of the intestinal tissue was examined under a com- qPCR using a 1:100 dilution of the original sample to determine pound microscope at 200 × for the presence of myxospores. infection prevalence. Intestinal tissues were collected for sequencing of the parasite Sequencing.—We extracted C. shasta DNA from water sam- DNA (see below). Surviving fish at 60 d postexposure were ples, polychaetes, and fish intestinal tissues as above. Aliquots euthanatized and processed as above. of 1 µL of the final eluate for water samples, 1 µL of a 1:100 18 HURST ET AL.

dilution for fish samples, and 2 µL of a 1:100 dilution for poly- Sequencing.—To determine if parasite genotype differed be- chaete samples was used in the genotyping polymerase chain tween the two upper basin sites, parasite DNA was sequenced reaction (PCR) assay (Atkinson and Bartholomew 2010a). Sam- from 10 C. shasta-positive RBT exposed at each location in ples were purified using ExoSAP-IT (USB, Cleveland, Ohio) June 2006–2010. When possible, fish were selected from both and submitted to the Center for Genomics and Bioinformat- the mortality and survivor groups for parasite genotype compar- ics at Oregon State University for sequencing using an ABI ison. Sequencing was performed as detailed above. Prism 3100 genetic analyzer. Parasite genotype was determined Statistical analyses.—Differences in Cq values between sen- according to the number of trinucleotide repeats in the inter- tinel sites (Williamson River and Keno eddy), months, and years nal transcribed spacer region I, and genotype proportions in were analyzed using a one-way ANOVA with statistical signif- mixed samples were identified by comparing chromatogram icance determined with an α = 0.05. The data met the assump- peak heights (Atkinson and Bartholomew 2010a). tion of normality, and significant effects were tested using a Statistical analyses.—Statistical significance was deter- Tukey’s honestly significant difference (HSD) test. Sentinel fish mined with an α = 0.05 for all tests. Temperature differences exposures were one component of a comprehensive monitoring among river reaches for June 2009 and 2010 were assessed us- study, and as a result we did not have an appropriate experi- ing a one-way analysis of variance (ANOVA). Significant effects mental design for statistical analysis of fish mortality between were examined using Tukey’s tests. The relationship between sentinel sites. We used S-plus version 8.1 (Tibco) to perform temperature and mean parasite density was examined using a statistical analyses. polynomial regression. Both the one-way ANOVA and polynomial regression were performed using S-plus version 8.1 statistical software (Tibco, Palo Alto, California). Differences in RESULTS parasite density (Cq values) among river reaches (1, 2, and 3) Density, Distribution, and Genotype Composition of and sampling times (month–year combination) were examined Ceratomyxa shasta in the Williamson River using a nested ANOVA (PROC GLM, SAS version 9.2, Cary, Parasite density differed between reaches and sites, and also North Carolina), where site was nested within reach, and month by month–year combination (Table 1). Parasite density was and year were tested as one variable because we were not able highest in reach 1, there being greater than 10 parasites/L at to sample all months over all years. Significant results were all sites. Densities in reach 2 were less than 1 parasite/L, except followed by Tukey’s tests. at rkm 19.2 in June 2009 where parasite densities exceeded 1 parasite/L. In reach 3, parasite densities at two sites were 1–10 parasites/L: just above Spring Creek at rkm 28 and 33 (Tukey’s Sentinel Fish Studies in the Upper Basin HSD: P < 0.05). At the remaining five sites, parasite densities Fish exposures.—Sentinel fish exposed as part of a com- were less than 1 parasite/L (Figure 2). The month–year vari- prehensive parasite monitoring study from 2006 to 2010 were able also affected parasite density, densities in May 2010 being used to assess differences in RBT mortality at rkm 7.6 in the ◦ ◦ lower than in September 2008, June 2009, and 2010 (Tukey’s Williamson River (42 30.820 N, 121 55.021 W), a system in HSD: P < 0.05). In both the Spring Creek and Sprague River which stocking of RBT occurs and at Keno eddy (42◦8.9835N, ◦ tributaries, parasite densities were less than 1 parasite/L during 122 0.9332 W), a site 47 rkm downstream of the Williamson all sampling times. Variability between qPCR assays was low, River where stocking does not occur (Figure 1). In 2009, the as standard deviations for both reference samples were 1.1 and Williamson River site was moved 6 rkm downstream for acces- 0.9 cycles. sibility reasons (42◦29.661N, 121◦56.095W). Nonnative RBT Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 Water temperatures were significantly different (Figure 2) (Roaring River Hatchery) were held at both sites in May and between river reaches but did not differ between June 2009 and June 2006–2010. Chinook and coho salmon (Iron Gate Hatchery June 2010 (one-way ANOVA: F2, 24 = 21.0, P < 0.001; F 1, 24 = strains, Hornbrook, California) were exposed to determine the 2.8, P = 0.11). June temperatures in reach 2 ranged from 5◦C effects of upper basin parasite dynamics on native anadromous fish strains. Fall Chinook salmon were exposed at both sites in TABLE 1. Results of ANOVAfor the effects of site nested within reach, reach, May and June 2006–2010. Coho salmon exposures were only and month–year combination on parasite density in the Williamson River. conducted in the Williamson River and took place in May and June 2009. Fish exposures and subsequent monitoring followed Type III sum methods in Stocking et al. (2006). Source of squares df FP-value Water sampling.—To determine parasite density during sen- Site (reach) 716.68 13 8.99 <0.0001 tinel fish exposures, three 1-L samples of water were taken at the Month–year 65.57 3 3.56 0.015 beginning and end of fish exposure at each site, except in June Error (subplot) 1,024.49 167 2008 when water samples were only collected at the beginning Reach 2,216.83 2 20.16 0.0001 of the exposures. Water samples were processed and assayed Error (whole plot) 715.32 13 using qPCR, as above. DAM REMOVAL AND IMPLICATIONS FOR FISH HEALTH 19

FIGURE 2. Cycle threshold (Cq) values and temperatures of C. shasta for Williamson River water samples collected in 2008, 2009, and 2010 by reach. Spring Creek and Sprague River conﬂuences of the Williamson River occur at rkm 27.5 and 19, respectively, as indicated by the vertical lines. Note: two sites occur between rkm 26 and 28, the lower Cq and temperature values correlating with the site in reach 2. [Figure available in color online.]

to 14◦C and were lower than temperatures in reaches 1 and Infection occurred in RBT held in water collected from either 3, which ranged from 15–21◦C and 12–22◦C, respectively. A reach 1 (100%) or reach 3 (20%), demonstrating presence of the wider temperature range was observed across the three reaches actinospore stage of the parasite. All fish were infected with in June (5–22◦C) than in September (8.5–16◦C) and May (8– genotype II; genotype 0 was not detected. 14◦C). Temperatures in the Sprague River were warm: 23◦Cin One polychaete assemblage was identified at rkm 18.3 in June and 16◦C in May and September. Spring Creek maintained reach 1 and one assemblage at rkm 33.2 in reach 3. Polychaete a relatively constant cool temperature of 5–8◦C regardless of density at the site below the Sprague River was calculated as month. In the Williamson River main stem, parasite DNA was 4.08 × 103 polychaetes/m2, and infection prevalence was 1.1% detected between 10◦C and 21◦C. A polynomial regression (y = (n = 185). Only genotype II was detected in the polychaetes. 0.0641 × 2 – 2.1108x + 50.144, R2 = 0.1976, P < 0.001) In the site within reach 3, polychaete density was estimated showed that approximately 20% of the variability in parasite at 78 polychaetes/m2 and infection prevalence could not Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 density was explained by temperature (Figure 3). be determined. Two of the four known genotypes of C. shasta were detected in the Williamson River. The proportions of parasite genotypes Sentinel Fish Studies in the Upper Basin II (pathogenic for nonnative RBT) and 0 (associated with native Mortality of sentinel nonnative RBT differed between the redband trout) in water samples changed spatially and tempo- Williamson River and Keno eddy, while Chinook salmon mor- rally, larger proportions of genotype 0 being detected in reach 1 tality was consistently below 5% at both sites. No mortality oc- than in reach 3. Genotype 0 generally accounted for 10–30% of curred in coho salmon exposed in the Williamson River. From the parasites present in water samples from all sampling times 2006 to 2010, mortality in RBT held at the Williamson River in reach 1, although in May 2010 water collected from rkm 18.2 sentinel site during May and June ranged from 97% to 100%, contained 82% genotype 0. This was the only time when the pro- and infection was from genotype II. Mortality in RBT exposed portion of genotype 0 was greater than genotype II. Genotype 0 at Keno eddy ranged from 0% to 75% (Figure 4) and fish were was not detected at any sites in reach 3, except in June 2009 when infected with either genotype 0 (50%) or genotype II (50%), it comprised approximately 10%. Parasite densities in reach 2 while exposure survivors were infected with genotype 0. Chi- were below the sequencing threshold, except in June 2009 when nook salmon were infected with genotypes II and III (Atkinson genotype II accounted for 100% of the parasites detected. and Bartholomew 2010a). 20 HURST ET AL.

FIGURE 3. Polynomial regression plot of temperature versus C. shasta cycle threshold (Cq) values for all months and reaches. Main-stem Williamson River samples are indicated using ﬁlled squares, and tributary data are depicted with hollow squares. The dashed horizontal line indicates parasite production (1 parasite/L). [Figure available in color online.] Downloaded by [Department Of Fisheries] at 01:36 01 March 2012

FIGURE 4. Mortality of nonnative RBT (bars) and average C. shasta cycle threshold (Cq) values (triangles) for the Williamson River and Keno eddy sentinel sites for years 2006–2010. The dashed line designates the Cq value that corresponds with 1 parasite/L. [Figure available in color online.] DAM REMOVAL AND IMPLICATIONS FOR FISH HEALTH 21

Parasite density was affected by year and site, but we did basin potentially allows for an expansion of the distribution of not detect effects of month. Densities in 2006 and 2007 were C. shasta. Chinook salmon historically utilized spawning habi- higher than those in 2008–2010 (one-way ANOVA: F4, 99 = tats in both the Williamson and Sprague rivers (Hamilton et al. 20.2, P < 0.0001). Variation among reference samples on assay 2005), and the recent removal of the Chiloquin Dam on the plates was up to three cycles or a 10-fold difference in parasite Sprague River in 2008 is expected to increase salmonid habitat. density and likely affected yearly comparisons in 2006–2007, Although C. shasta was not detected in the Sprague River (this whereas variation in 2008–2010 was generally less than 1 cycle study), this may have been a result of warm temperatures that and did not equate to a ten-fold change in density. Parasite may limit both host utilization of the habitat (Cherry et al. 1977) density at Keno eddy was typically less than 1 parasite/L, which and the parasite’s reproduction. However, Chinook salmon re- is 10–100-fold less than in the Williamson River (where density turning in the spring are likely to utilize the Sprague River when was usually > 10 parasites/L; one-way ANOVA: F1, 99 = 232.3, water temperatures are lower and are more favorable for both P < 0.0001). the parasite and the host. The short-term effects of C. shasta on coho salmon reintroduction success are likely to be minimal. The historical distri- DISCUSSION bution of coho salmon likely extended to Spencer Creek, lo- The success of salmon restoration in the Klamath River is cated less than 1 rkm upstream of Keno eddy (Hamilton et al. dependent on the availability of spawning and rearing habitat in 2005), suggesting it is unlikely these fish will migrate into the the upper basin and on the level of exposure to pathogens that Williamson River. Even if migration into the Williamson River cause high mortality. In the Williamson River high densities of were to occur, the survival of coho salmon after exposure to the myxozoan C. shasta reflect the influence of water tempera- high doses of genotype II in this study suggests that this geno- ture variability, location of polychaete populations, and move- type may be adapted to different host species in the upper and ment of the fish hosts on parasite dynamics. Only two of the three lower Klamath River basin. This differential pathogenicity has parasite genotypes present in the upper basin were detected: also been supported in laboratory challenges (Hurst 2010). The genotype II (associated with nonnative RBT) was dominant in risk of disease from C. shasta for coho salmon migrating into the Williamson River, and at Keno eddy (located 47 rkm down- the vicinity of Keno eddy is currently low based on low para- stream in the main-stem Klamath River) genotype 0, associated site densities and predominance of genotype 0 (Atkinson and with native redband trout, was dominant. This genotype distri- Bartholomew 2010b). However, long-term effects of coho rein- bution likely reflects the presence of stocked nonnative RBT in troduction would likely include the introduction and establish- the system, which has provided a new host for the parasite. ment of a lower-basin genotype II specific for coho salmon, With dam removal, modification, or both, the migration of potentially increasing disease risk over time. anadromous salmonids into the upper Klamath basin is likely Because of the specificity of C. shasta genotypes for their to alter parasite dynamics by providing new hosts for existing fish host species, the fish host(s) responsible for parasite ampli- parasite genotypes and introducing new genotypes. Chinook fication can be inferred from the genotype ratio. Native redband salmon that historically inhabited the Williamson and Sprague trout become infected with C. shasta genotype 0 but do not of- rivers (Hamilton et al. 2005) are the focus of reintroduction ef- ten develop ceratomyxosis (Buchanan et al. 1989; Atkinson and forts (Hooten and Smith 2008). The short-term effects of the Bartholomew 2010b). In contrast, nonnative RBT are highly sus- high parasite densities in the Williamson River are likely to be ceptible to infection and disease with genotype II (Atkinson and negligible for juvenile Chinook salmon as disease is not associ- Bartholomew 2010a, 2010b). The low proportion of genotype 0 ated with exposure to the genotypes present (0 and II; Atkinson in the Williamson River suggests that native redband trout are Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 and Bartholomew 2010a; Hurst 2010). However, adult Chinook not responsible for the high parasite densities. Thus, the stock- salmon will introduce genotype I as they migrate into new habi- ing of nonnative RBT to supplement recreational fishing is the tats (Atkinson and Bartholomew 2010a). As conditions in the most likely explanation for the amplification of parasites in the Williamson River below the Sprague River confluence are al- Williamson River. This explanation is supported by higher mor- ready conducive to parasite propagation, this introduced parasite tality of RBT and higher density of genotype II in the Williamson genotype is likely to become established. Thus in the long- River as compared with Keno eddy, where stocking does not oc- term, we predict juvenile Chinook salmon that migrate through cur. This result has prompted the ODFW to reevaluate the RBT this area of the river after late May could encounter increased stocking plan (Tinniswood, personal communication). disease risk from C. shasta. Introduction of adult salmon in- Patterns of fish movement can also be inferred from the fected with genotype I does not present a risk for the native genotype ratio. The presence of C. shasta genotypes 0 and II redband trout as this species does not become diseased after throughout the main-stem Williamson River suggests that both exposure to this parasite genotype (Atkinson and Bartholomew nonnative RBT and native redband trout are found throughout 2010b). the main-stem river. Conversely, the inability to detect genotype In addition to risks from establishment of new parasite geno- II from water samples at Keno eddy indicates that the non- types, reintroduction of anadromous fish into the upper Klamath native RBT are unlikely to migrate this distance downstream. 22 HURST ET AL.

Following dam removal, lower basin genotypes I and II (associ- tions, such as the Williamson River, following reintroduction. ated with coho salmon) will likely be detected in the Williamson Results of this study also suggest that current management prac- River and at Keno eddy as both Chinook and coho salmon mi- tices, such as stocking nonnative fish, should be reevaluated as grate into this area. restoration plans progress. Trends in parasite density and distribution are affected by temperature throughout the main-stem Williamson River and its ACKNOWLEDGMENTS tributaries. Water temperatures in the lower Williamson River We thank those who provided assistance in the field: Bill Tin- (reach 1) are influenced by input from the Sprague River, where niswood and Roger Smith (ODFW), the Nature Conservancy, land use practices have decreased natural riparian shading, re- and private landowners (Lonesome Duck Resort, Waterwheel sulting in warmer temperatures (Boyd et al. 2002). Thus, in Campground, Williamson River Resort, and Sportsman’s Park the Williamson River below the Sprague River confluence, in Keno), as well as Julie Alexander, Michelle Jordan, Gerri temperatures rise above 10◦C in April, peak in July, then ◦ Buckles, and Shawn Harris from the Bartholomew Laboratory continue to decline, falling below 10 C in October (USGS; at Oregon State University (OSU). We thank Steve Christy, Matt http://waterdata.usgs.gov/or/nwis/rt; this study). Consequently, Stinson, and Jill Pridgeon for aid in processing samples, and parasite densities below the Sprague River confluence were high Julie Alexander, Adam Ray, Xuan Che, and Sophie Dong who during all sample times. Water temperatures above the Sprague provided much needed statistical guidance. Advice and com- River confluence were largely influenced by Spring Creek, a ◦ ments on this manuscript from Kym Jacobson at the Hatfield constant source of cold water (5–10 C) even during the sum- Marine Science Center and Kate Field, Sascha Hallett, Stephen mer months. Although both hosts may be present, temperatures Atkinson, and Sarah Bjork (OSU) were greatly appreciated. are below the threshold of both actinospore production in the We also thank the Roaring River Hatchery (ODFW) and Iron polychaete host (Bjork 2010) and parasite proliferation in the Gate Fish Hatchery (California Department of Fish and Game) fish host (Schafer 1968; Johnson 1975; Udey et al. 1975; Ratliff for fish. Funding was provided by the ODFW Restoration and 1981; Hendrickson et al. 1989). Therefore, the cool-water input Enhancement Program. from Spring Creek likely limits parasite production in reach 2. Parasite density above Spring Creek (reach 3) may be limited by water flow, as temperatures were similar to those recorded be- REFERENCES low the Sprague River confluence. The upper portion of reach 3, Atkinson, S. D., and J. L. Bartholomew. 2010a. Disparate infection patterns of Ceratomyxa shasta (Myxozoa) in rainbow trout Oncorhynchus mykiss and at rkm 42, is ephemeral, with no flow from late summer through Chinook salmon Oncorhynchus tshawytscha correlate with ITS-1 sequence winter (USGS; http://waterdata.usgs.gov/or/nwis/rt) and re- variation in the parasite. International Journal of Parasitology 40:599–604. duced flows occur further downstream where polychaetes were Atkinson, S. D., and J. L. Bartholomew. 2010b. Spatial, temporal and host identified. This flow variability may naturally constrain poly- factors structure the Ceratomyxa shasta (Myxozoa) population in the Klamath chaete populations, which require a certain amount of flow River basin. Infection, Genetics and Evolution 10:1019–1026. Bartholomew, J. L. 1998. Host resistance to infection by the myxosporean for filter feeding and to prevent desiccation (Stocking and parasite Ceratomyxa shasta: a review. Journal of Aquatic Animal Health Bartholomew 2007; Bjork and Bartholomew 2009). Thus, water 10:112–120. flow may restrict the range of the parasite in the upper reaches Bartholomew, J. L., J. S. Rohovec, and J. L. Fryer. 1989. Ceratomyxa shasta, of the Williamson River. a myxosporean parasite of salmonids. U.S. Fish and Wildlife Service Fish With dam removal proposed as a tool in many restoration Disease Leaflet 80. Bartholomew, J. L., M. J. Whipple, D. G. Stephens, and J. L. Fryer. 1997. efforts, it is important to consider how pathogen dynamics in a The life cycle of Ceratomyxa shasta, a myxosporean parasite of salmonids, system may change once dams are removed (Brenkman et al. requires a freshwater polychaete as an alternate host. Journal of Parasitology Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 2008). In cases where fish passage above the dams has never oc- 83:859–868. curred, resident fish have been isolated from pathogens present Bednarek, A. T. 2001. Undamming rivers: a review of the ecological impacts of below the projects and may have decreased resistance. Con- dam removal. Environmental Management 27:803–814. Bjork, S. J. 2010. Factors affecting the Ceratomyxa shasta infectious cycle versely, isolated resident fish could transmit novel pathogens and transmission between polychaete and salmonid hosts. Doctoral disser- to reintroduced fish stocks. Thus, when considering the disease tation. Oregon State University, Corvallis. Available: ir.library.oregonstate. risk from an endemic parasite such as C. shasta, some level of edu/xmlui/handle/1957/15435. (December 2010). parasite-induced effects should be factored into reintroduction Bjork, S. J., and J. L. Bartholomew. 2009. Effects of Ceratomyxa shasta dose efforts. Dam removal may be expected to reduce some pathogen on a susceptible strain of rainbow trout and comparatively resistant Chinook and coho salmon. Diseases of Aquatic Organisms 86:29–37. risks below dams as a result of improved flows and temperature Boyd, M., S. Kirk, M. Wiltsey, and B. Kasper. 2002. Upper Klamath Lake as well as increased movement of fish. However, in areas up- drainage total maximum daily load (TMDL) and water quality management river of the effects of dam removal, environmental conditions plan (WQMP). Oregon Department of Environmental Quality, Portland. may not change significantly and reintroduction may have un- Brenkman, S. J., S. L. Mumford, M. House, C. Patterson. 2008. Establishing intended consequences. Thus, disease risks should be evaluated baseline information on the geographic distribution of fish pathogens endemic in Pacific salmonids prior to dam removal and subsequent recolonization by establishing baseline information on pathogens above and by anadromous fish in the Elwha River, Washington. Northwest Science below the current barriers and by monitoring high-risk loca- 82(Special Issue):42–52. DAM REMOVAL AND IMPLICATIONS FOR FISH HEALTH 23

Buchanan, D. V., A. R. Hemmingsen, D. L. Bottom, R. A. French, K. P. Cur- Klamath Restoration. 2010a. Klamath Basin restoration agreement for the rens, and F. C. Shrier. 1989. Native trout project. Oregon Department of sustainability of public and trust resources and affected communities. Fish and Wildlife, Fish Research Project, Annual Progress Report F-136-R, Klamath Restoration, Yreka, California. Available: klamathrestoration.gov/ Portland. sites/klamathrestoration.gov/files/Klamath-Agreements/Klamath-Basin- Cherry, D. S., K. L. Dickson, J. Cairns Jr., and J. R. Stauffer. 1977. Preferred, Restoration-Agreement-2-18-10signed.pdf. (November 2010). avoided, and lethal temperatures of fish during rising temperature conditions. Klamath Restoration. 2010b. Klamath hydrologic settlement agree- Journal of the Fisheries Research Board of Canada 34:239–246. ment. Klamath Restoration, Yreka, California. Available: http://www. Ching, H. L., and D. R. Munday. 1984. Geographic and seasonal distribution of kbrt.org/files/Klamath Hydroelectric Settlement Agreement.pdf. (January the infectious stage of Ceratomyxa shasta Noble, 1950, a myxozoan salmonid 2012). pathogen in the Fraser River system. Canadian Journal of Zoology 62: Mundy, P. R. 1997. The role of harvest management in the future of Pacific 1075–1080. salmon populations: shaping human behavior to enable the persistence of Fresh, K. L. 1997. The role of competition and predation in the decline of Pacific salmon. Pages 315–330 in D. J. Stouder, P. A. Bisson, and R. J. Naiman, edi- salmon and steelhead. Pages 245–276 in D. J. Stouder, P. A. Bisson, and R. tors. Pacific salmon and their ecosystems: status and future options. Chapman J. Naiman, editors. Pacific salmon and their ecosystems: status and future and Hall, New York. options. Chapman and Hall, New York. Pearcy, W. G. 1997. Salmon production in changing ocean domains. Pages Gannett, M. W., K. E. Lite, Jr., J. L. LaMarche, B. J. Fisher, and D. J. Polette. 331–351 in D. J. Stouder, P. A. Bisson, and R. J. Naiman, editors. Pacific 2007. Ground-water hydrology of the upper Klamath basin, Oregon and Cal- salmon and their ecosystems: status and future options. Chapman and Hall, ifornia. U.S. Geological Survey, Scientific Investigations Report 2007-5050, New York. version 1.1, Washington, D.C. Available: pubs.usgs.gov/sir/2007/5050/. Ratliff, D. E. 1981. Ceratomyxa shasta: epizootiology in Chinook salmon of (October 2010). central Oregon. Transactions of the American Fisheries Society 110:507– Hallett, S. L., and J. L. Bartholomew. 2006. Application of a real-time PCR 513. assay to detect and quantify the myxozoan parasite Ceratomyxa shasta in Schafer, W. E. 1968. Studies on the epizootiology of the myxosporidian Cer- water samples. Diseases of Aquatic Organisms 71:109–118. atomyxa shasta (Noble). California Fish and Game 54:90–99. Hallett, S. L., and J. L. Bartholomew. 2009. Development and application of Stanley, E. H., and M. W. Doyle. 2003. Trading off: the ecological effects of a duplex QPCR for river water samples to monitor the myxozoan parasite dam removal. Frontiers in Ecology and the Environment 1:15–22. Parvicapsula minibicorni. Diseases of Aquatic Organisms 86:39–50. Stocking, R. W., and J. L. Bartholomew. 2007. Distribution and habitat charac- Hamilton, J. B., G. L. Curtis, S. M. Snedaker, and D. K. White. 2005. Dis- teristics of Manayunkia speciosa and infection prevalence with the parasite, tribution of anadromous fishes in the upper Klamath River watershed prior Ceratomyxa shasta, in the Klamath River, Oregon/California, USA. Journal to hydropower dams: a synthesis of the historical evidence. Fisheries 30(4): of Parasitology 93:78–88. 10–20. Stocking, R. W., R. A. Holt, J. S. Foott, and J. L. Bartholomew. 2006. Spatial Hemmingsen, A. R., D. V. Buchanan, D. L. Bottom, R. A. French, K. P. Cur- and temporal occurrence of the salmonid parasite Ceratomyxa shasta in the rens, and F. C. Shrier. 1988. Native trout project. Oregon Department of Oregon–California Klamath River basin. Journal of Aquatic Animal Health Fish and Wildlife, Fish Research Project Annual Progress Report F-136-R, 18:194–202. Portland. Udey, L. R., J. L. Fryer, and S. K. Pilcher. 1975. Relation of water tempera- Hendrickson, G. L., A. Carleton, and D. Manzer. 1989. Geographic and sea- ture to ceratomyxosis in rainbow trout (Salmo gairdneri) and coho salmon sonal distribution of the infective stage of Ceratomyxa shasta (Myxozoa) in (Oncorhynchus kisutch). Journal of the Fisheries Research Board of Canada Northern California. Diseases of Aquatic Organisms 7:165–169. 32:1545–1551. Hooten, B., and R. Smith. 2008. A plan for the reintroduction of anadro- USGS (United States Geological Survey). 2010. USGS real-time water data mous fish in the upper Klamath basin. Oregon Department of Fish for Oregon. USGS Water Resources, Washington, D.C. Available: water- and Wildlife, draft report, Klamath Falls. Available: www.dfw.state.or.us/ data.usgs.gov/or/nwis/rt. (December 2010). agency/commission/minutes/08/05 may/C 2 Draft%20Plan%20for%20the% Winter, B. D. 1990. A brief overview of dam removal effects in Washing- 20Reintroduction%20of%20Anadromous%20Fish%20in%20the%20Upper% ton, Oregon, Idaho, and California. NOAA Technical Memorandum NMFS 20Klamath%20Basin.pdf. (September 2010). F/NWR-28. Hurst, C. N. 2010. Ceratomyxa shasta-related concerns for reintroduced Zielinski, C. M., H. V.Lorz, and J. L. Bartholomew. 2010. Detection of Myxobo- anadromous salmonids into the upper Klamath basin California/Oregon, lus cerebralis in the lower Deschutes River basin, Oregon. North American USA. Master’s thesis. Oregon State University, Corvallis. Available: Journal of Fisheries Management 30:1032–1040.

Downloaded by [Department Of Fisheries] at 01:36 01 March 2012 ir.library.oregonstate.edu/xmlui/handle/1957/19652. (December 2010). Zielinski, C. M., H. V. Lorz, S. L. Hallett, L. Xue, and J. L. Bartholomew. Johnson, K. A. 1975. Host susceptibility, histopathologic and transmission stud- 2011. Comparative susceptibility of Deschutes River, Oregon, Tubifex tubifex ies on Ceratomyxa shasta, a myxosporidian parasite of salmonid ﬁsh. Doc- populations to Myxobolus cerebralis. Journal of Aquatic Animal Health 23: toral dissertation. Oregon State University, Corvallis. 1–8. Kanehl, P. D., J. Lyons, and J. E. Nelson. 1997. Changes in the habitat and Zinn, J. L., K. A. Johnson, J. E. Sanders, and J. L. Fryer. 1977. Susceptibility ﬁsh community of the Milwaukee River, Wisconsin, following removal of of salmonid species and hatchery strains of Chinook salmon (Onchorhyncus the Woolen Mills Dam. North American Journal of Fisheries Management tshawytscha) to infections by Ceratomyxa shasta. Journal of the Fisheries 17:387–400. Research Board of Canada 34:933–936. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:38 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 Habitat for Age-0 Shovelnose Sturgeon and Pallid Sturgeon in a Large River: Interactions among Abiotic Factors, Food, and Energy Intake Dawn R. Sechler a , Quinton E. Phelps a , Sara J. Tripp a , James E. Garvey a , David P. Herzog b , David E. Ostendorf b , Joseph W. Ridings b , Jason W. Crites b & Robert A. Hrabik b a Fisheries and Illinois Aquaculture Center, Department of Zoology, Southern Illinois University, 1125 Lincoln Drive, Carbondale, Illinois, 62901, USA b Missouri Department of Conservation, Open Rivers and Wetlands Field Station, 3815 East Jackson Boulevard, Jackson, Missouri, 63755, USA Available online: 08 Feb 2012

To cite this article: Dawn R. Sechler, Quinton E. Phelps, Sara J. Tripp, James E. Garvey, David P. Herzog, David E. Ostendorf, Joseph W. Ridings, Jason W. Crites & Robert A. Hrabik (2012): Habitat for Age-0 Shovelnose Sturgeon and Pallid Sturgeon in a Large River: Interactions among Abiotic Factors, Food, and Energy Intake, North American Journal of Fisheries Management, 32:1, 24-31 To link to this article: http://dx.doi.org/10.1080/02755947.2012.655848

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 32:24–31, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.655848

MANAGEMENT BRIEF

Habitat for Age-0 Shovelnose Sturgeon and Pallid Sturgeon in a Large River: Interactions among Abiotic Factors, Food, and Energy Intake

Dawn R. Sechler,* Quinton E. Phelps, Sara J. Tripp, and James E. Garvey Fisheries and Illinois Aquaculture Center, Department of Zoology, Southern Illinois University, 1125 Lincoln Drive, Carbondale, Illinois 62901, USA David P. Herzog, David E. Ostendorf, Joseph W. Ridings, Jason W. Crites, and Robert A. Hrabik Missouri Department of Conservation, Open Rivers and Wetlands Field Station, 3815 East Jackson Boulevard, Jackson, Missouri 63755, USA

these organisms to complete key life history stages (Galat and Abstract Zweimuller 2001; Garvey et al. 2010). The habitats within the The main stems of large rivers throughout the world have been main channel and its adjacent areas are dynamic, creating het- highly modified with little consideration for effects on fishes that erogeneity that is probably critical for early growth and survival rely on these areas to complete their life histories. Particularly important is the ability of riverine habitats to provide foraging op- of fish, particularly for species that recruit in these areas, such portunities for young fish. We explored how temperature, flow, and as freshwater drum Aplodinotus grunniens, blue catfish Ictalu- food availability influenced diet content, prey selection (Strauss’s rus furcatus, paddlefish Polyodon spathula, and North Ameri- linear selectivity index), and energy condition of age-0 shovelnose can sturgeons (Acipenseridae). Although extrachannel habitats sturgeon Scaphirhynchus platorynchus and pallid sturgeon S. al- (e.g., backwaters and floodplains) are often the focus of river bus in major habitat areas (e.g., islands, channel borders, wing dikes, and side channels) of the middle Mississippi River during conservation and restoration (Gutreuter 2004; Prato and Hey spring (March–May) and summer (June–August) 2008. Standard- 2006), main-channel areas also have been altered in many ways ized diet mass (dry mass standardized for fish body mass) of the for purposes such as flood control and navigation and require ◦ age-0 sturgeon peaked at about 19 C and at a flow velocity of 0.5 restoration attention (see Garvey et al. 2010). The impact of m/s. Although potential prey taxa were diverse, the diets for age-0 these changes on fishes that depend on these areas is not well sturgeon of all sizes were dominated by mayflies (Ephemeroptera) and midge larvae (Chironomidae) across all habitats. As age-0 understood (Dettmers et al. 2001). Identifying and quantifying

Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 sturgeon grew, the relative energy return per habitat appeared to key habitat types in these rivers are critical for understand- diverge; island tips upstream of the main channel and channel bor- ing the effects of degradation and for determining restoration der areas behind wing dikes provided higher gains in standardized goals. diet mass than other habitats. No differences in energy condition The middle Mississippi River (extending 320 river kilometers (kJ/g) occurred among habitats, although large (51–200 mm total length [TL]) age-0 sturgeon had higher energy densities than their [rkm] between the confluences of the Ohio and Missouri rivers) small (≤50 mm TL) counterparts. Enhancement of areas with flow has been highly channelized, leaving few remaining side chan- and substrates that facilitate the production and availability of nels, sandbars, and instream islands (Hurley et al. 2004; Koch midges and mayflies (e.g., instream island complexes) is critical for et al. 2012). Loss of these habitat types may have negatively af- the recruitment of age-0 Scaphirhynchus sturgeon in large rivers. fected the resident sturgeons of the genus Scaphirhynchus.The more common species, the shovelnose sturgeon S. platorynchus, The main channels of large rivers contain many species of and its federally endangered congener, the pallid sturgeon S. fish (Dettmers et al. 2001) and may be important for allowing albus, have high mortality rates and low recruitment due to a

*Corresponding author: [email protected] Received August 2, 2011; accepted October 6, 2011

24 MANAGEMENT BRIEF 25

multitude of anthropogenic impacts, such as overharvesting and At the sizes collected, it was impossible to morphologically dis- habitat loss (Colombo et al. 2007; Tripp et al. 2009). tinguish shovelnose sturgeon from pallid sturgeon. Thus, this Recruitment in these sturgeon species is probably driven by research focuses on patterns at the genus level, assuming that the availability, quantity, and quality of habitat that is present the shovelnose sturgeon predominated in the samples (Hrabik in or lies adjacent to the main river channel after their lar- et al. 2007; Schrey et al. 2007). At each site, a Marsh–McBirney vae settle from the drift and become benthic-foraging juveniles flowmeter was used to measure surface water velocity (m/s) and (Galat and Zweimuller 2001; Braaten et al. 2008). The density a Quanta or Hydrolab handheld meter or sonde was used to of Scaphirhynchus larvae and juveniles differs among habitat collect water temperature and dissolved oxygen data. types in the middle Mississippi River; small fish occupy areas At each macrohabitat, 500 mL of trawl contents (consist- of moderate flow that are sheltered from the main river channel, ing of a mixture of substrate, invertebrates, and organic matter) whereas larger, older individuals become more dispersed in the were randomly subsampled from a near-bank trawl tow and river (Phelps et al. 2010a). We evaluated the relative energetic an off-bank trawl tow. The contents were then preserved in value of major main-stem habitat types in the middle Mississippi 95% ethanol, and macroinvertebrates were later extracted by River for small (≤50 mm total length [TL]) and large (51–200 hand for identification and quantification. The 500-mL subsam- mm TL) age-0 shovelnose sturgeon and pallid sturgeon. We ple represented macroinvertebrate families in the entire trawl hypothesized that habitats would differ in food availability and (D. Sechler, unpublished data). Prey densities were quantified thus would differ in potential energy return, depending on flow as the number of individuals of each taxon per 3-min trawl and prey associated with each habitat area. We examined macro- tow. habitats and other abiotic and biotic factors to evaluate whether The structure of Scaphirhynchus sturgeon stomachs may they are more conducive for successful foraging and thus better cause food to back up into the esophagus while other food is growth and survival to advanced life stages. being processed (Held 1969). Thus, the esophagus was removed along with the stomach for studying the diet composition of each age-0 sturgeon. Stomach and esophagus contents were carefully METHODS removed and rinsed onto a gridded dish. Macroinvertebrates Once per week during May–September 2008, age-0 shov- were enumerated and identified to the family level by using elnose sturgeon and pallid sturgeon were collected between dichotomous keys (Merritt and Cummins 1996). Where head 1000 and 1530 hours via 3-min tows with a mini-Missouri capsules were present, head capsule widths (mm) were mea- bottom trawl (see Herzog et al. 2005 for full gear de- sured for later calculation of biomass; such measurements were scription) at six island complexes in the middle Missis- performed for Ephemeroptera (mayflies), Diptera (true flies) pu- sippi River: Mosenthein (rkm 298–303, where rkm is the pae, Chironomidae (midges; Diptera), Plecoptera (stoneflies), distance upstream of the Ohio River confluence), Osborne Trichoptera (caddisflies), and Odonata (damselflies and dragon- Chute (rkm 233–236), Rockwood (rkm 161–166), Grand flies). If other macroinvertebrates were encountered, they were Tower–Cottonwood (rkm 126–129), Marquette (rkm 77– measured according to the methods of Benke et al. (1999). In 80), and Angelo (rkm 8). At each island complex, seven addition, if a macroinvertebrate was identified by a body part macrohabitats were sampled (Figure 1): channel border– that was not used for measurement, it was counted to improve dike (CBD), channel border–open (CBO), island–downstream the accuracy of prey frequency of occurrence data (Chipps and tip (IDT), island–main channel (IMC), island–upstream Garvey 2007). Scion imaging software was used to measure tip (IUT), side channel (SC), and main channel. Sampling all organisms at 500× magnification by using a digital camera in each macrohabitat type included near-bank trawling (water (Cohu, Inc., San Diego, California) mounted on a Wild Model Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 depth <3) and off-bank trawling (water depth > 3m).Given M5A dissecting scope. that the highest densities and greater variability in lengths of Macroinvertebrate measurements were converted to dry mass young sturgeon occur near islands and SCs (Phelps et al. 2010a), (µg) values by using length–dry mass regressions (e.g., Benke about 70% of the sampling was concentrated in these areas. The et al. 1999; Chipps and Garvey 2007). For each age-0 sturgeon, remaining 30% of sampling effort was expended in the CBO diet mass was calculated in terms of micrograms of dry mass and CBD strata. In addition, the three channel macrohabitats per gram of fish wet mass (i.e., µg/g) to standardize the amount are homogeneous, whereas heterogeneity exists around island across sizes of sturgeon sampled. After the diet analysis was complexes; therefore, more sampling effort was needed around completed (i.e., the esophagus and stomach were removed and islands to capture the unique differences in abiotic and biotic the contents were identified and measured), the energy density factors. Trawl sites were excluded when river conditions (e.g., (kJ/g) of age-0 sturgeon was determined to further investigate low water or the presence of obstructions) prevented sampling. whether standardized diet mass was related to energy density. The TL (mm) and wet mass (g) of each age-0 sturgeon were Each sturgeon was dried in an oven at 60◦C to a constant mass recorded; each fish was placed into a vial that contained a 95% and was then weighed. The dried sturgeon were pulverized, ethanol solution accompanied by specific trawl information (is- pressed into a pellet, and burned in a Parr 1425 semi-micro land, macrohabitat, date, water velocity, and water temperature). bomb calorimeter. 26 SECHLER ET AL.

FIGURE 1. Distribution of macrohabitat types and trawling locations sampled for age-0 Scaphirhynchus sturgeon (shovelnose sturgeon and pallid sturgeon) and macroinvertebrates in the middle Mississippi River during 2008.

For each macroinvertebrate taxon, the frequency of occur- pared across macrohabitats with a one-way ANOVA combined rence and percent by number in diets and trawl samples were with Tukey’s pairwise comparisons. Mean standardized diet calculated. These data were used to calculate Strauss’s linear mass (µg/g) was compared with water velocity and tempera- index of selectivity (Strauss’s L; Strauss 1979) for the four ture by using nonlinear regression models; we assumed that 2 Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 dominant prey categories: mayﬂy nymphs, chironomid larvae, a parabolic function (i.e., Y = x + x ) would be appropriate dipteran pupae (including chironomid pupae), and other (i.e., for both comparisons. Previous research suggests that growth all other prey consumed). Strauss’s L was calculated as L = and diet intake should peak at some intermediate depth and ri − pi, where ri is the relative abundance of prey item i in the temperature (Phelps et al. 2010b). Indices of macroinvertebrate gut and pi is the relative abundance of prey item i in the habitat. abundance from trawls (mean number per 3-min trawl tow) also This index was used to determine prey size and macrohabitat were compared across macrohabitats during each season via diet selectivity by age-0 sturgeon. Strauss’s L ranges from −1 one-way ANOVA. Preliminary analysis of medium-sized age-0 to 1; positive values indicate selection, negative values indicate sturgeon (51–100 mm TL) and larger age-0 sturgeon showed no avoidance or inaccessibility, and a value of 0 indicates that the difference in their diets, so these two groups were pooled (here- prey type is consumed in proportion to its availability in the after, “large size-class”) to increase sample size (D. Sechler, environment (Strauss 1979; Chipps and Garvey 2007). unpublished data). All diet analyses were therefore conducted Mean daily water temperatures (◦C) were compared among by separating small (≤50 mm TL) and large (51–200 mm TL) macrohabitat types by using a one-way analysis of variance sturgeon because age-0 sturgeon probably behave differently (ANOVA) coupled with Tukey’s pairwise comparisons. Site over ontogeny. To detect differences in the standardized diet temperature and surface water velocity (m/s) were also com- mass and age-0 sturgeon energy content across macrohabitats, MANAGEMENT BRIEF 27

we used ANOVA with Tukey’s pairwise comparisons for both size-classes. To further investigate whether size-related selectivity varied across macroinvertebrate taxa and macrohabitats, we conducted a multivariate ANOVA (MANOVA) on both sturgeon size-classes with each of the four prey groups as dependent variables and macrohabitat as the independent variable. Trends found with the MANOVAs were further evaluated by use of univariate ANOVAs. All analyses were conducted with the general linear models procedure in the Statistical Analysis System (α = 0.05).

RESULTS In total, 404 age-0 Scaphirhynchus sturgeon were captured during 2008; 266 of these fish were categorized as small age- 0 sturgeon (mean TL = 26.5 mm, SD = 0.03), and 138 were categorized as large age-0 sturgeon (mean TL = 103.1 mm, SD = 0.24). No age-0 sturgeon were caught in the main-channel macrohabitat; therefore, the main-channel habitat was excluded from the analyses. One-way ANOVA (F5, 135 = 4.78, P = 0.0005) and Tukey’s pairwise comparisons indicated that water temperatures in the IMC and IUT macrohabitats were cooler than those in all other macrohabitats (Figure 2). Water velocity differed significantly across macrohabitats (F5, 117 = 5.75, P < 0.0001; Figure 2). Tukey’s pairwise comparisons indicated differences in water velocity across macrohabitats; CBD and IUT macrohabitats had the lowest water velocities, and IDT habitat had the highest water velocities. The parabolic regression models of standardized diet mass in relation to water temperature (diet mass = [2.99 × temperature] − [0.078 × temperature2]; r2 = 0.79) and velocity (diet mass = [83.2 × velocity] − [64 × velocity2]; r2 = 0.73) were both significant (P < 0.0001), suggesting that an intermediate peak diet mass existed for both FIGURE 2. Mean (±SE) water temperature (top panel) and water velocity relationships. Macroinvertebrate abundance per trawl differed (bottom panel) at each macrohabitat type (see Figure 1) in the middle Mississippi significantly among macrohabitats (MANOVA: Wilks’ lambda River during May–September 2008. Sample sizes are given at the bottom of each bar. Bars without a letter in common are significantly different (P < 0.05). = 0.94, F20, 1,705 = 1.61, P = 0.0400). However, the taxonomic categories did not significantly differ in abundance among habitat types (univariate ANOVAs: all P > 0.05; Figure 3); the different among macrohabitats (F = 47.13, P < 0.0001); Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 5, 129 exception was Ephemeroptera (univariate ANOVA: F5, 517 = large sturgeon sampled at CBD habitats had the highest mean 3.41, P = 0.005). For three prey categories (dipteran pupae, diet mass, and those sampled at SC habitats had the lowest mean chironomid larvae, and other prey), relative abundances were diet mass (Figure 4). Diet selectivity by both small age-0 stur- generally higher in areas associated with the channel border and geon (Wilks’ lambda: F20, 508 = 3.83, P < 0.0001) and large IUTs. age-0 sturgeon (Wilks’ lambda: F20, 233 = 2.78, P = 0.0001) Empty stomachs were rare in all macrohabitats during 2008: differed significantly among macrohabitats (Figure 5). Selectiv- the percentage of age-0 sturgeon with empty stomachs was 2% ity by small sturgeon differed significantly among habitats for in IUT habitat (total n = 37 sturgeon); 1% in CBD habitat all four invertebrate prey categories (all P < 0.05): Strauss’s L (n = 68); and 0% in the CBO (n = 33), IDT (n = 94), IMC varied from −1 to 1 for mayfly nymphs, from −0.02 to 1.00 (n = 129), and SC (n = 38) habitats. Roughly 2% of small for dipteran pupae, from −0.05 to 1.00 for chironomid larvae, age-0 sturgeon had empty stomachs, and approximately 1% of and from −1.00 to 0.01 for the “other” category. The patterns large age-0 sturgeon had empty stomachs. The standardized were generally similar for the large age-0 sturgeon, although diet mass (µg/g) for small age-0 sturgeon did not vary across differences in mean Strauss’s L only occurred for mayflies and macrohabitats (F5, 258 = 1.95, P = 0.0859; Figure 4). However, chironomid larvae. Energy density (kJ/g) of age-0 sturgeon the standardized diet mass for large sturgeon was significantly did not significantly vary among macrohabitats for the small 28 SECHLER ET AL.

FIGURE 4. Mean (±SE) standardized diet mass (µg dry weight of stomach and esophagus contents per g of sturgeon wet mass) for small (≤50 mm total length) and large (51–200 mm) size-classes of age-0 Scaphirhynchus sturgeon across six macrohabitat types in the middle Mississippi River, May–September 2008. Sample sizes are given at the bottom of each bar. Bars without a letter in common are significantly different (P < 0.05). Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 FIGURE 3. Mean (±SE) number of macroinvertebrates in trawl samples from 2008 growing season for age-0 shovelnose sturgeon and pallid six macrohabitats (abbreviations defined in Figure 1) of the middle Mississippi sturgeon, and these differences may affect diet selection, energy River during May–September 2008. Invertebrates were aggregated into four primary prey groups: ephemeropterans (EPH; mayflies); dipteran pupae (DP; density, and ultimately growth and survival during early life includes chironomid pupae); chironomid larvae (CH); and other (all other prey (J. E. Garvey and S. R. Chipps, in press). The effect of these consumed). Sample sizes are the same as shown in Figure 2. Bars without a habitats on age-0 sturgeon may depend on fish size. Although letter in common are significantly different (P < 0.05). the small size-class of age-0 sturgeon appeared to forage in a consistent manner across macrohabitats, our results suggest that

size-class (F5, 99 = 1.34, P = 0.2533) or the large size-class the larger age-0 sturgeon garnered different energy returns de- (F5, 31 = 1.14, P = 0.36; Figure 6). Mean energy density pending on the habitat they occupied. Our analysis indicated that was higher in large age-0 sturgeon than in small individuals habitats adjacent to the main stem near the channel border and (Figure 6). behind wing dikes provided more food for large age-0 sturgeon. Standardized diet mass for large sturgeon in SC areas was con- DISCUSSION siderably lower, suggesting that foraging opportunities are lim- We found that the main-stem areas of the middle Mississippi ited in these areas, particularly during low-flow periods late in River varied in physical and biological characteristics during the the age-0 sturgeon growing season (e.g., fall), when thes habitats MANAGEMENT BRIEF 29 Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 FIGURE 5. Mean (±SE) values of Strauss’s linear selectivity index for small (≤50 mm total length) and large (51–200 mm) size-classes of age-0 Scaphirhynchus sturgeon across six macrohabitats (abbreviations defined in Figure 1) in the middle Mississippi River, May–September 2008. Selection is presented for four prey categories: ephemeropterans (EPH); dipteran pupae (DP; includes chironomid pupae); chironomid larvae (CH); and other (all other prey consumed). Sample sizes are given at the bottom of each bar. Bars without a letter in common are significantly different (P < 0.05).

can become disconnected from the main channel. Additionally, Although we did ﬁnd some habitat-speciﬁc differences in stage height (about 8 m) remained high throughout spring and standardized diet mass, the energy density of the small and summer during 2008, thereby potentially increasing available large size-classes of age-0 sturgeon did not depend on habi- spawning habitats, nursery habitats, and catch rates of age-0 tat. Small sturgeon had lower energy densities, probably be- sturgeon (Phelps et al. 2010b) in comparison with the previous cause more body mass was allocated to energy-neutral mate- 4 years. A better understanding of the factors that made 2008 a rials, such as skeleton and bony scutes. In contrast, a greater more favorable year for growth, survival, and recruitment of age- proportion of energy was probably allocated to storage and 0 Scaphirhynchus sturgeon will provide knowledge that is vital energy-rich body structures in age-0 sturgeon of the larger size- to enhancing the management of these declining populations. class, leading to higher energy density. Energy density is often 30 SECHLER ET AL.

organisms. Mayfly nymphs, which probably embed in or cling to firm sand or gravel substrate, occurred similarly in all macrohabitats (i.e., based on trawl samples); this prey group often appeared in diets of both the small and large age-0 sturgeon, and selection often was positive. In summary, these results suggest that there is no apparent size-specific or age-specific dietary shift exhibited by age-0 Scaphirhynchus sturgeon among habitats. Although mayfly nymphs are larger and thus are the most energetically beneficial prey common in the diets, the clear preference of age-0 sturgeon for midge larvae suggests that they are an abundant and energetically efficient food source for young sturgeon in the middle Mississippi River. Our results are similar to those reported by Hoover et al. (2007) for older Scaphirhynchus sturgeon in the middle Mississippi River; in their study, mayflies and caddisflies were important diet components for adult Scaphirhynchus sturgeon. Pallid sturgeon have also been observed to consume fish as they age (Gerrity et al. 2006; Hoover et al. 2007). However, it is not known whether pallid sturgeon smaller than 200 mm TL will engage in pis- civory; we did not document any fish in age-0 sturgeon diets during the present study. The middle Mississippi River and other large rivers that harbor young Scaphirhynchus sturgeon and other river-obligate fish are continually modified for multiple uses; therefore, the quantity and quality of habitat must be defined accurately. Phelps et al. (2010a) found that the macrohabitat scheme used herein provided insight into the relative density of age-0 Scaphirhynchus sturgeon. Similarly, we found that food availability and energy density also varied across these habitat types in biologically meaningful ways. Areas of contrasting flow probably create substrate that is amenable to the production and perhaps suspension of midges and mayflies and appear to be impor- FIGURE 6. Mean (±SE) energy density (kJ/g) of small (≤50 mm total length) tant, particularly for older sturgeon life stages. Not surprisingly, and large (51–200 mm) size-classes of age-0 Scaphirhynchus sturgeon across six the natural habitat types where these conditions occur include macrohabitats in the middle Mississippi River, May–September 2008. Sample sizes are given at the bottom of each bar. Bars without a letter in common are the IUT areas and channel borders. An artificial counterpart, the significantly different (P < 0.05). wing dike, creates alternating areas of scour and deposition that probably facilitate both invertebrate production and the foraging success of age-0 Scaphirhynchus sturgeon. At a finer scale, the an indicator of body condition and should be related to dietary parabolic functions we fit to our data suggest that standardized Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 intake. diet mass in age-0 sturgeon peaked in areas characterized by a Despite the wide variety and apparently high density of po- temperature of about 19◦C and a flow velocity of 0.5 m/s. These tential prey items available to age-0 sturgeon in the middle data indicate that management efforts to create heterogeneity in Mississippi River, only a few prey categories were consumed flow conditions will facilitate substrates favored by midges and in different quantities than available in the habitats. These re- mayflies, thereby improving Scaphirhynchus sturgeon foraging sults are nearly identical to those found for age-0 shovelnose success, growth, and ultimately recruitment. Ideally, this could sturgeon in the upper Missouri River (Braaten et al. 2007). In be accomplished by creating a series of island complexes in particular, we found that both the small and large size-classes of the main channel of the middle Mississippi River (Koch et al. age-0 sturgeon across habitats consistently consumed chirono- 2012). mids in proportions greater than their abundance in the trawls. Future efforts should investigate shallow and deeper (>3m) Chironomids are often associated with fine substrates and may transects and the effects of depth, water temperature, water ve- have been easily suspended in the river, thus entering the drift locity, and sturgeon movements on diet composition and energy and becoming more accessible to foraging sturgeon. Dipteran density of age-0 Scaphirhynchus sturgeon. Sturgeon movement pupae were fairly rare in trawl samples but still appeared in di- is also of importance because large age-0 sturgeon are probably ets, again suggesting some preference for these benthic-oriented mobile in the river (Phelps et al. 2010a); therefore, diet contents MANAGEMENT BRIEF 31

from a specific habitat sampled at one point in time may not eries techniques, 3rd edition. American Fisheries Society, Bethesda, reflect an individual’s actual foraging habitat because the stur- Maryland. geon may have foraged elsewhere before moving into the habitat Garvey, J. E., B. Ickes, and S. Zigler. 2010. Challenges in merging fisheries research and management: the upper Mississippi River experience. Hydrobi- of capture. In addition, examination of both main-channel and ologia 640:125–144. SC habitats associated with island point-bar macrohabitats may Gerrity, P. C., C. S. Guy, and W. M. Gardner. 2006. Juvenile pallid sturgeon provide further insight into how the aforementioned variables are piscivorous: a call for conserving native cyprinids. Transactions of the may differ in these additional macrohabitats. This resolution on American Fisheries Society 135:604–609. a finer scale of macrohabitat sampling, when combined with Gutreuter, S. 2004. Challenging the assumption of habitat limitation: an example from centrarchid fishes over an intermediate spatial scale. River Research and trawling depth data, may provide more-specific information on Applications 20:413–425. which habitats are necessary for age-0 Scaphirhynchus stur- Held, J. W. 1969. Some early summer foods of the shovelnose sturgeon in the geon to successfully recruit to adult life stages. Lastly, this Missouri River. Transactions of the American Fisheries Society 98:514–517. study provides baseline information that can be used to further Herzog, D., V. A. Barko, J. S. Scheibe, R. A. Hrabik, and D. E. Ostendorf. 2005. explore crucial early life history stages of Scaphirhynchus spp. Efficacy of a benthic trawl for sampling small-bodied fishes in large river systems. North American Journal of Fisheries Management 25:594–603. sturgeons. Hoover, J. J., S. G. George, and K. J. Killgore. 2007. Diet of shovelnose sturgeon and pallid sturgeon in the free-flowing Mississippi River. Journal of Applied ACKNOWLEDGMENTS Ichthyology 23:494–499. Hrabik, R. A., D. P. Herzog, D. E. Ostendorf, and M. D. Petersen. 2007. Lar- We are grateful to Mike Hill and Justin Seibert for assistance vae provide first evidence of successful reproduction by pallid sturgeon, in field collection of age-0 sturgeon. We also thank Nicholas Scaphirhynchus albus, in the Mississippi River. Journal of Applied Ichthyol- Wahl, Lionel Grant, and Michael O’Keefe for assisting with ogy 23:436–443. the identification of aquatic insects. Funding for this research Hurley, K. L., R. J. Sheehan, R. C. Heidinger, P. S. Wills, and B. Clevenstine. 2004. Habitat use by middle Mississippi River pallid sturgeon. Transactions was provided by the U.S. Army Corps of Engineers St. Louis of the American Fisheries Society 133:1033–1041. District. Koch, B., R. C. Brooks, A. Oliver, D. Herzog, J. E. Garvey, R. Hrabik, R. Colombo, Q. Phelps, and T. Spier. 2012. Habitat selection and movement of naturally occurring pallid sturgeon in the Mississippi River. Transactions of REFERENCES the American Fisheries Society 141:112–120. Benke, A. C., A. D. Huryn, L. A. Smock, and J. B. Wallace. 1999. Length- Merritt, R. W., and K. W. Cummins. 1996. An introduction to the aquatic insects mass relationships for freshwater macroinvertebrates in North America, with of North America, 3rd edition. Kendall/Hunt, Dubuque, Iowa. particular reference to the southeastern United States. Journal of the North Phelps, Q. E., S. J. Tripp, J. E. Garvey, D. P. Herzog, D. E. Ostendorf, J. W. American Benthological Society 18:308–343. Ridings, J. W. Crites, and R. A. Hrabik. 2010a. Habitat use during early life Braaten, P. J., D. B. Fuller, L. D. Holte, R. D. Lott, W. Viste, T. F. Brandt, and history infers recovery needs for shovelnose sturgeon and pallid sturgeon in R. G. Legare. 2008. Drift dynamics of larval pallid sturgeon and shovelnose the middle Mississippi River. Transactions of the American Fisheries Society sturgeon in a natural side channel of the upper Missouri River, Montana. 139:1060–1068. North American Journal of Fisheries Management 28:808–826. Phelps, Q. E., S. J. Tripp, W. D. Hintz, J. E. Garvey, D. P. Herzog, D. E. Osten- Braaten, P. J., D. B. Fuller, and N. D. McClenning. 2007. Diet composition of dorf, J. W. Ridings, J. W. Crites, and R. A. Hrabik. 2010b. Water temperature larval and young-of-year shovelnose sturgeon in the upper Missouri River. and river stage influence mortality and abundance of naturally occurring Mis- Journal of Applied Ichthyology 23:516–520. sissippi River Scaphirhynchus sturgeon. North American Journal of Fisheries Chipps, S. R., and J. E. Garvey. 2007. Assessment of diets and feeding patterns. Management 30:767–775. Pages 473–513 in C. Guy and M. Brown, editors. Analysis and interpretation Prato, T., and D. Hey. 2006. Economic analysis of wetland restoration along of freshwater fisheries data. American Fisheries Society, Bethesda, Maryland. the Illinois River. Journal of the American Water Resources Association Colombo, R. E., J. E. Garvey, N. D. Jackson, R. Brooks, D. P. Herzog, 42:125–131. R. A. Hrabik, and T. W. Spier. 2007. Harvest of Mississippi River sturgeon Schrey, A. W., B. L. Sloss, R. J. Sheehan, R. C. Heidinger, and E. J. Heist. 2007.

Downloaded by [Department Of Fisheries] at 01:38 01 March 2012 drives abundance and reproductive success: a harbinger of collapse? Journal Genetic discrimination of middle Mississippi River Scaphirhynchus sturgeon of Applied Ichthyology 23:441–451. into pallid, shovelnose, and putative hybrids with multiple microsatellite loci. Dettmers, J. M., D. H. Wahl, D. A. Soluk, and S. Gutreuter. 2001. Life in the fast Conservation Genetics 8:683–693. lane: fish and foodweb structure in the main channel of large rivers. Journal Strauss, R. E. 1979. Reliability estimates for Ivlev’s electivity index, the forage of the North American Benthological Society 20:255–265. ratio, and a proposed linear index of food selection. Transactions of the Galat, D. L., and I. Zweimuller. 2001. Conserving large-river fishes: is the American Fisheries Society 108:344–352. highway analogy an appropriate paradigm? Journal of the North American Tripp, S. J., R. E. Colombo, and J. E. Garvey. 2009. Declining recruitment Benthological Society 20:266–279. and growth of shovelnose sturgeon in the middle Mississippi River: implica- Garvey, J. E., and S. R. Chipps. In press. Quantifying diets and en- tions for conservation. Transactions of the American Fisheries Society 138: ergy flow. In A. Zale, D. Parrish, and T. Sutton, editors. Fish- 416–422. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:39 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 Assessing Red Drum Juvenile Stocking in a South Carolina Estuary Using Genetic Identification Michael R. Denson a , Karl Brenkert IV a , Wallace E. Jenkins a & Tanya L. Darden a a South Carolina Department of Natural Resources, Post Office Box 12559, Charleston, South Carolina, 29422-2559, USA Available online: 15 Feb 2012

To cite this article: Michael R. Denson, Karl Brenkert IV, Wallace E. Jenkins & Tanya L. Darden (2012): Assessing Red Drum Juvenile Stocking in a South Carolina Estuary Using Genetic Identification, North American Journal of Fisheries Management, 32:1, 32-43 To link to this article: http://dx.doi.org/10.1080/02755947.2011.649577

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 32:32–43, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2011.649577

ARTICLE

Assessing Red Drum Juvenile Stocking in a South Carolina Estuary Using Genetic Identiﬁcation

Michael R. Denson,* Karl Brenkert IV, Wallace E. Jenkins, and Tanya L. Darden South Carolina Department of Natural Resources, Post Ofﬁce Box 12559, Charleston, South Carolina 29422-2559, USA

Abstract The South Carolina Department of Natural Resources has been stocking red drum Sciaenops ocellatus since 1988 to evaluate parameters critical to their successful survival and recruitment in South Carolina estuaries. From 1999 to 2002, between 600,000 and 1,000,000 juvenile red drum were stocked each year in two tributaries of Charleston Harbor.The harbor and each tributary were partitioned into three independent strata and randomly sampled monthly for two decades, allowing population trends before, during, and after stocking to be evaluated. Using microsatellite- based parentage analysis, we examined the contribution of stocked age-0 juvenile red drum (15–60 mm total length) to the local population 1 year after release by using fishery-independent sampling. Analysis of these data showed that the highest contributions (88.9%) were close to the stocking site in years with low natural recruitment, whereas in years with high natural recruitment, contributions were lower and stocking was less effective in increasing catch per unit effort. The results of stocking 600,000 small juveniles/year from 1999 to 2001 in one of the study tributaries (Ashley River) indicated that stocked fish did not displace wild fish but had an additive effect on their abundance, supporting the hypothesis that trophic resources are not limiting for postlarval age-0 red drum within Charleston Harbor. The high observed variability in contribution among years of stocking similar-sized red drum suggests that (1) interpretation on a year-class-specific basis is necessary to fully understand the effects of stocking and (2) marine stock enhancement programs would benefit substantially from evaluation in the context of wild annual recruitment patterns.

The red drum Sciaenops ocellatus is an important recre- Historically, marine stock enhancement efforts have been ational fish species that is distributed along the South Atlantic perceived as unsuccessful due to their inability to demonstrate and Gulf coasts of the USA; populations of red drum have a quantitative contribution to the fisheries (Shelbourne 1964; Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 substantially declined, due in part to increased fishing pressure Blaxter 2000; Chan et al. 2003). In many cases, however, (e.g., Vaughan and Carmichael 2000). As a result, many states these enhancement programs were targeted at commercially have enacted successively more-restrictive regulations on recre- harvested species (Atlantic cod Gadus morhua and plaice Pleu- ational anglers to allow a portion of the estuarine-dependent ronectes platessa), making it unlikely that small-scale stocking subadult population to reach sexual maturity and join the programs could impact catches with low numbers of stocked spatially separate offshore spawning aggregations (Peters and fish (Shelbourne 1975; Danielssen and Gjosaeter 1994). Typ- McMichael 1987; Ross et al. 1995; Rooker et al. 1998). In ically, research-scale stocking programs release too few fish addition, Texas, Florida, and South Carolina have developed to be detected in open systems when evaluating changes in stock enhancement research programs to address specific fisheries-dependent or fisheries-independent catch per unit ef- questions about life history, stocking strategies, and tagging fort (CPUE) as a measure of success (Tveite 1971; Blaxter 2000; methods (McEachron et al. 1995; Arnold et al. 1996; Jenkins Scharf 2000). In addition, historic egg, larvae, and small juve- et al. 1997, 2004, 2005; Serafy et al. 1999). nile stocking programs implemented in both Europe and the

*Corresponding author: [email protected] Received June 3, 2010; accepted August 18, 2011

32 GENETIC IDENTIFICATION TO ASSESS RED DRUM JUVENILE STOCKING 33

USA released fish that were not tagged or marked in a manner to utilize genetic marks as the primary tracking tool. As reported permitting the qualitative assessment of their impacts (Blaxter by Jenkins et al. (2004) and Smith et al. (2004), the results of 1976; Solemdal et al. 1984). stocking OTC-marked fish suggest that stocked age-0 juveniles During the last 30 years, more sophisticated marking could make a substantial contribution to wild populations. The techniques (e.g., chemical or thermal marking of otoliths) longevity of OTC marks and inherent variability in reader iden- have allowed researchers to mark larval fish prior to their tification of the marks, however, make the use of genetic iden- release (Tsukamoto 1988; Volk et al. 1990; Brooks et al. 1994; tification with microsatellites preferable to other methods for Beckman and Schulz 1996). Since its inception in 1988, the ascertaining contribution levels (Denson and Smith 2008). red drum stocking program administered by the South Carolina This paper summarizes the evaluation of stocking red drum in Department of Natural Resources (SCDNR) has marked all two rivers within the Charleston Harbor estuary. The estuary was fish prior to release to facilitate quantitative evaluation of the randomly sampled monthly over two decades, allowing popula- efficacy of stocking. During this time, SCDNR has used a tion trends before, during, and after stocking to be evaluated. The variety of marking methods, including external anchor tags research examined the contribution of stocked red drum juve- embedded in the dorsal musculature or through the abdominal niles (15–60 mm total length [TL]) on the localized population wall, coded wire tags injected into the snout, chemical marking 1 year after release, when they recruit to fishery-independent of otoliths through immersion in oxytetracycline hydrochloride sampling. Microsatellite analysis of individuals collected from (OTC), and most recently genetic identification (Jenkins et al. fisheries-independent sampling was used to assess contribution. 1997, 2002, 2004, 2005; Renshaw 2003; Robbins et al. 2008). Hatchery genotypes were determined from captive broodstock, In their review of genetic techniques and application to fish- subsamples of juveniles retained from each release group, and eries research, Jones et al. (2010) indicated that microsatellite samples collected in the wild. The efficacy of stocking age-0 markers were the primary choice for parentage assignment. Sev- red drum juveniles for enhancement of local populations and eral researchers have reported using these nuclear DNA markers for further elucidating the factors that may be controlling red to discriminate released hatchery fish from their wild cohorts by drum recruitment was also examined. matching hatchery offspring to broodstock with a low probability of mismatch (Ferguson and Danzmann 1998; Perez-Enriquez and Taniguchi 1999; MacDonald et al. 2004). Microsatellites METHODS provide sufficient resolution for identifying hatchery-produced Broodstock collection and spawning.—Red drum broodstock red drum collected from an estuary and for conducting parent- were acquired from the wild and produced hatchery year-classes age analyses to determine whether individual fish are progeny during four consecutive years (1999–2002). In spring and sum- of broodstock used in the stocking program (Liu and Cordes mer 1999, 11 broodstock were collected from the local offshore 2004; Tringali 2006; Karlsson et al. 2008; Tringali et al. 2008). adult population by SCDNR staff via bottom longline. Passive The use of nonlethal genetic tools from small fin clips is an integrated transponder tags were implanted into the fish; the attractive alternative to sacrificing fish to extract otoliths. fish were sexed, measured, weighed, and returned to shore in The use of genetic tools for stocking evaluation was demon- live tanks. Upon arrival at the hatchery, fish were transferred to strated by red drum stock enhancement work conducted in South 3.8-m-diameter tanks. Each tank was equipped with a biological Carolina by Jenkins et al. (2004), during which approximately bead filter (Armant Aquaculture, Vacherie, Louisiana) and ul- 600,000 age-0 juveniles were released annually between 1995 traviolet sterilizer (Aqua Ultraviolet, Temecula, California). Six and 1998 in the Colleton River estuary. All fish released during fish (three males and three females) were placed in one tank, and the experiment were marked prior to release by immersion in five fish (two males and three females) were placed in another Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 OTC (Jenkins et al. 2002). Fish collected from the wild were tank. A fin clip was removed from each fish for genetic analy- evaluated for both OTC and genetic marks. Renshaw (2003) sis and stored in a solution of 1% sarcosyl, 6-M urea, 20-mM demonstrated that genetic analyses could be used as a substi- sodium phosphate, and 1-mM EDTA buffered at pH 6.8. tute for OTC marking; in blind tests, fish identified as being of In 2000, an additional 14 red drum adults were collected and hatchery origin by OTC marks were also correctly assigned by divided into tanks at sex ratios similar to that of the 1999 col- analyses of microsatellites. While both methods indicated that lection. In 2001, two tanks of fish captured during the previous the fish of hatchery origin made up a substantial component years were recombined to provide a different genetic makeup; (∼19%) of each year-class in the target estuary, due to errors in the fish were reconditioned and spawned to produce juveniles reading OTC marks there was some disagreement as to which for the fall 2001 stocking efforts. These fish were subsequently fish were stocked. In addition, the lack of baseline CPUE data released, and nine new fish were obtained from the wild to for the stocked area meant that it was not possible to demonstrate produce the 2002 hatchery year-class. whether stocking supplemented the existing population or dis- Red drum were conditioned and spawned by using com- placed wild fish. For a stocking program to be effectively evalu- pressed photoperiod and temperature cycles to simulate natural ated, such a metric must be documented. Thus, a new multiyear seasonal fluctuations. Control of spawning allowed us to prepare study was initiated in 1999 within the Charleston Harbor estuary the nursery ponds at the Waddell Mariculture Center (WMC) 34 DENSON ET AL.

FIGURE 1. Map of the Charleston Harbor estuary, South Carolina, showing the three sampling strata (Charleston Harbor, Ashley River, and Wando River) monitored by monthly stratiﬁed random sampling for red drum.

several weeks prior to stocking with larvae to optimize nursery sampling gear in early summer; therefore, age-1 data included rearing conditions. In addition to larvae produced from brood- only red drum less than 400 mm TL that were collected between stock held at SCDNR, a private hatchery (Southland Fisheries, July and December of each year. From 1992 to 1998, mean red Hollywood, South Carolina) provided larvae from broodstock drum CPUE for the Ashley River ranged from 0.3 to 1.4 fish/net collected in South Carolina waters during 1999 for the first three set, while the Charleston Harbor stratum ranged from 1.2 to 5.3 stocking events. fish/net set and the Wando River ranged from 1.2 to 8.4 fish/net During each year, 2–3-d-old red drum larvae were stocked at set. These data suggested that the Ashley River usually had a a density of 1.4 million larvae/ha in fertilized saltwater ponds smaller population of age-1 red drum than the Wando River. (see Jenkins et al. 2004) and were harvested approximately The specific stocking locations within the study tributaries 30 d after stocking. During harvest, fish were group-weighed to were previously identified by agency researchers as “good” estimate the total biomass harvested from each pond. In addition, nursery habitat, characterized by small creeks adjacent to high- at least three subsamples from each pond were group-weighed marsh, mud flats, and oyster reefs where red drum had been and individually counted to estimate the number of fish per unit previously captured (Figure 1). The fish were transported to weight. In addition, 25 fish/pond were individually weighed and boat landings adjacent to stocking sites in a 2,275-L, oxygenated measured to document the size-frequency distribution. hauling trailer and were acclimated for approximately 1 h before Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 Control groups of red drum were retained from every release being removed from the trailer and placed in a boat-mounted group to document posthandling survival. Genetic samples were tank equipped with compressed oxygen and high-pressure air also collected from a subsample of 50 fish from each production stones. Fish were then transported to the stocking locations, unit. Fish were maintained in recirculating seawater systems where they were dispersed throughout the marsh during flood for 30 d. tide at or close to the time of high water. Stocking sites.—The two locations chosen for stocking were Each river was stocked during three sequential years to allow the Ashley and Wando rivers, which converge in Charleston for replication of treatments. In 1999, approximately 617,000 Harbor. Sample sites in the two rivers and the harbor were red drum juveniles (19–104 mm TL) from seven ponds were re- partitioned into three separate strata in a statewide random leased into the Ashley River on six separate occasions between fishery sampling survey (Figure 1). Beginning in 1990, each October 13 and November 30 (Table 1). Significant differences stratum was routinely sampled for numerous species by using (Kruskal–Wallis analysis of variance: H = 319.741, df = 6, standardized gear (trammel nets) throughout a 2-m tidal ampli- P < 0.001) in the size of fish originating from the different tude. Red drum were only available to the gear when the water ponds dictated that the reported mean was weighted. Four of the was less than 1 m in depth and receded from the smooth cord- release groups (574,191 fish) were made up of red drum that grass Spartina alterniflora marsh. Red drum first recruit to the measured 19–47 mm TL (mean = 28.2 mm TL). The remaining GENETIC IDENTIFICATION TO ASSESS RED DRUM JUVENILE STOCKING 35

TABLE 1. Mean total length (TL) and number (N) of red drum released into Genetic methods.—Genomic DNA was extracted from red the Ashley and Wando rivers from 1999 to 2002. Numbers of fish stocked were drum fin clips by using the SprintPrep Plasmid Purification sys- estimated by using gravimetric techniques; asterisk indicates that the mean TL of the 1999 year-class was weighted by the number and TL of fish from two tem (Agencourt Bioscience) according to the manufacturer’s stocking batches. directions and was stored in 10-mM tris and 0.1-mM EDTA. All tissue samples were genotyped at eight microsatellite loci in Ashley River Wando River three multiplexed panels (Table 2). Amplification of DNA was µ Mean TL Mean TL performed in 20- L reactions containing approximately 10 ng of µ TL range TL range DNA, 0.2 mM of each deoxynucleotide triphosphate, 0.3 Mof × Year N (mm) (mm) N (mm) (mm) forward and reverse primers, 1 HotMaster Taq buffer, 3.5 mM of MgCl2, and 0.1 units of HotMaster Taq DNA polymerase (Ep- 1999 617,190 31.0∗ 19–104 pendorf). Thermal cycling conditions included 2 min at 94◦C; 18 2000 604,884 23.5 16–46 513,920 23.4 16–37 cycles of 30 s at 94◦C, 30 s at annealing temperature (decreasing 2001 721,417 21.1 17–28 344,949 20.4 17–28 from 55◦Cto43◦C with a 1.5◦C decrease every two cycles), and 2002 621,962 22.3 13–36 40 s at 65◦C; 17 cycles of 30 s at 94◦C, 30 s at 43◦C, and 40 s at 65◦C; and one cycle of 60 min at 65◦C. The amplified products were separated on a CEQ 8000 automated sequencer (Beckman Coulter) along with a fluorescently labeled, 400-base-pair size = 37,117 fish were significantly larger (mean 63.0 mm TL). The standard. Genotypes were scored using the Beckman Coulter mean size of stocked fish for the entire release group in 1999 CEQ 8000 fragment analysis software. Two readers indepen- was 31.0 mm TL. In 2000, the Ashley River was stocked with dently scored all data, and differences were reconciled in con- 604,884 juveniles (16–46 mm TL) on four stocking trips be- ference. All broodstock red drum were genotyped at least twice. tween September 29 and November 8, and the Wando River was Expected allele and genotype frequencies were estimated stocked on five occasions with 513,920 juveniles (16–37 mm by using Levene’s (1949) correction. The effective number TL) between October 18 and November 28. In 2001, the of alleles (AE) was calculated according to Kimura and Crow Ashley River was stocked with 721,417 juveniles (17–28 mm (1964) as TL) during five trips between September 24 and November 9, while the Wando River was stocked with 344,949 juveniles A = / − H , (17–28 mm TL) during two trips on November 9 and 14. In 2002, E 1 (1 E) the Wando River was stocked with 621,962 fish (13–36 mm TL) during three trips between September 11 and 26 and was the only where HE is the expected heterozygosity. The significance river stocked in the Charleston Harbor estuary during that year. of any deviation from Hardy–Weinberg equilibrium was

TABLE 2. Characteristics of microsatellite loci used to identify stocked red drum. The number of alleles and the effective number of alleles (AE) were calculated based on red drum collected from 2000 to 2003 within the Charleston Harbor estuary. GenBank accession Repeat Multiplex Alleles Locus number Forward (F) and reverse (R) primer sequence motif group (n) AE

Soc014 AF183146 (F) GTATGTATTAAGGGCACAAGGTG (GT)21 1165.2 (R) GATTGCTGCTGGACAGACTG Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 Soc017 AF183148 (F) CGCCCGTCTACGTGACAGTATG (GT)14 1166.5 (R) ATAGCTGCGCATCATTCGGTTG Soc243 AF073283 (F) GACGGGGATGCCATCTGC (CCT)9 273.5 (R) AATGCGAAAAAGACGAAACAGT Soc083 AF073269 (F) TGCTGTAATTGAAAAGCAGTGTAC (GT)19 2177.4 (R) AGCGGAACTAGAATTGGTTTTATA a Cne612 — (F) CAAGTGCACGGTATGTGAG (GT)20 2102.5 (R) AGGAACCTGACCAATCAAA Soc029 AF183148 (F) GGCAAATAGTACAGAAAATTACATGG (GT)10 352.7 (R) GAT TTCTCTCAGTGACTGGCAGTT Soc129 AF073275 (F) GCGGCTGCAACACAAGAATT (ATCT)4 32011.4 (R) TGCACGGGGAAACAGAACG Soc060 AF073267 (F) TCTATTGAAGCCTGTAAGTTAGTT (GAG)9 352.6 (R) CAAGGAAGGAGTGGGGAATGACAA

aSource: R. Chapman, South Carolina Department of Natural Resources, personal communication. 36 DENSON ET AL.

evaluated with sequential Bonferroni corrections (Rice 1989) Rousset 1995a) implemented in GENEPOP version 3.3, and ge- as determined by an unbiased approximation of Fisher’s exact netically similar releases were combined. The reference groups test using a Markov-chain randomization method implemented then consisted of six genetically dissimilar populations: wild in GENEPOP version 3.3 (Guo and Thompson 1992; Raymond fish and the release groups that were held in tanks D1, D2–D3, and Rousset 1995a, 1995b). Larvae were evaluated for adher- D4–D5, C3, and C4. Individuals that did not match WMC brood- ence to expected Mendelian proportions by using a chi-square stock by exclusion analysis were designated as “stocked” or test. Linkage disequilibrium was tested between all pairs of loci “wild” if they were assigned by maximum likelihood analysis to in all wild year-class samples, again by using a Markov-chain- any combination of Southland Fisheries reference populations or based randomization method as implemented in GENEPOP. to the wild reference population, respectively, with a probability In 1999, red drum broodstock and hatchery juveniles were of greater than 90%. Any individual that did not meet these crite- sampled at WMC, while only hatchery juveniles were sampled at ria was designated unknown. The validation test was conducted the Southland Fisheries hatchery. For 2000–2002, progeny were on a subset (WMC-released and wild) of age-1 fish belonging to produced only at the WMC and all broodstock genotypes and the 1999 year-class. Out of 105 samples, 98 assignments were hatchery juvenile genotypes were available. For the progeny of in agreement among the two methods (93.3% concordance). In the 1999–2002 year-classes, exclusion analysis implemented in all cases, the GENECLASS2 method was more conservative PROBMAX version 1.3 (Danzmann 1997) was used for parent- (i.e., identified fish as wild when identified by PROBMAX as age assignment. Broodstock genotypic data were compared with cultured). Therefore, although the assumptions associated with genotypic data from wild and hatchery samples. Probabilities mixed-stock analysis (GENECLASS2) algorithms are not usu- of compatibility by chance (PCCs) were calculated using the ally met in stocking scenarios, our confidence level in this case postbroodstock case described by Bravington and Ward (2004) appears sound as we find high concordance among exclusion based on population allele frequencies from wild fish caught (PROBMAX) and mixed-stock analysis methods as well as zero during 1999–2002. The results from the PCC analysis deter- false-positive identifications. In a worst-case scenario, we would mined the criteria that were needed to classify red drum as underestimate the contribution of Southland Fisheries-released stocked, wild, or unknown. The probability of correctly identi- fish by 6.7%. fying a stocked fish with all eight loci had a 99.99% confidence Sampling.—The SCDNR’s stratified random sampling ef- level (PCC = 4.4 × 10−5) based on an average of 15 potential forts began in 1990, and each stratum was sampled monthly by parental pairs annually within the hatchery broodstock. Occa- using trammel nets deployed from the stern of a specially de- sionally, loci failed to amplify and could not be scored; as the signed Florida net boat (Tremblay). Locations were randomly number of scored loci decreased, the probability of correctly selected from 30 sampling locations in each stratum that could identifying a stocked fish decreased to an average confidence be effectively sampled with the gear type. These sites were typ- level of 99.96% for seven loci (PCC = 4.2 × 10−4) and 99.83% ically along marsh banks with gradual slopes, near oyster reefs, for six loci (PCC = 1.6 × 10−3). Based on the PCCs, assign- mud flats, on either side of creek mouths, and near structure. ment criteria for the project were (1) scoring of zero to three Locations with drop-offs, fast-moving water, and submerged loci was insufficient to make a genetic determination of cultured stumps or structure were avoided. The nets were 137 m long, versus wild (i.e., the fish was classified as unknown); (2) if four consisting of 2.4-m-deep monofilament in three panels (two to eight loci were scored and any single locus was inconsistent outside panels: 10-cm stretch; one internal panel: 5-cm stretch). with broodstock profiles, the fish was classified as wild; and The nets were deployed in an arc next to the bank while the boat (3) if seven or eight loci were scored and all were consistent was moving at approximately 28 km/h to trap the fish between with broodstock profiles, the fish was classified as stocked. the bank and net. After the net was set, the boat was driven Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 For the 1999 year-class, additional analysis was required through the sampling area and the fish were driven into the net. in order to identify the progeny of Southland Fisheries brood- Red drum recruit to this sampling gear at 10 months of age, stock. Fish that did not match WMC broodstock were examined when they are approximately 250 mm TL. The CPUE data col- with analysis implemented in GENECLASS2 (Piry et al. 2004), lected for each month were used to determine year-class strength which assigned individuals to reference populations based on and regional variability within each year-class. Red drum CPUE Bayesian methods (Rannala and Mountain 1997). The probabil- within a stratum was compared across years to evaluate changes ity of assignment was determined using a Markov-chain Monte between the years of stocking and the years prior to stocking. Carlo resampling algorithm (Paetkau et al. 2004). Reference Additionally, within-year red drum CPUE was compared across populations were created from presumed wild fish (based on strata to evaluate potential changes in relationships of CPUE otolith results) and all hatchery releases (from Southland Fish- among strata during the stocking experiment. Analyses of CPUE eries and WMC). Reference populations for the WMC releases were conducted for each stratum and year (1999–2002) by using were included to provide a validation test among the two parent- a general linear model. The CPUE data were loge(x + 1) trans- age assignment methods. The probability that the separate re- formed prior to analysis, and Tukey’s multiple comparison tests lease groups were genetically similar was determined by using were used to identify individual differences between strata. Due an unbiased estimate of the Fisher’s exact test (Raymond and to the schooling behavior of subadult red drum, the variation GENETIC IDENTIFICATION TO ASSESS RED DRUM JUVENILE STOCKING 37

TABLE 3. Summary of the number of hatchery red drum released, the fishery-independent sampling catch of red drum in trammel nets (1 year after release), and the percentage contribution of hatchery fish to the number of red drum sampled for years during and immediately after the stocking experiment (Ashley = Ashley River; Wando = Wando River; Harbor = Charleston Harbor). For the 1999 year-class, sample sizes and contribution from only the smaller-sized release group (see Methods) are provided in parentheses. Stocking Effort Number Hatchery fish Total annual hatchery Year-class Site density (fish/ha) (net sets) CPUE ± SD sampled percentage contribution (%) 1999 Ashley 630 41 1.73 ± 3.73 54 (34) 88.9(82.4) Wando 0 32 1.00 ± 2.42 41 (38) 14.6(7.9) Harbor 0 38 0.74 ± 1.03 26 (20) 30.8(10.0) 51.2(35.9) 2000 Ashley 618 47 2.72 ± 7.97 59 35.6 Wando 184 33 3.97 ± 7.43 89 13.5 Harbor 0 50 3.68 ± 4.33 45 6.718.7 2001 Ashley 737 51 1.04 ± 1.64 94 28.4 Wando 124 35 5.20 ± 8.46 49 2.0 Harbor 0 46 4.02 ± 6.87 64 1.614.0 2002 Ashley 0 50 1.10 ± 2.44 17 0.0 Wando 232 30 5.47 ± 10.36 47 0.0 Harbor 0 47 6.91 ± 14.88 37 0.00.0

in CPUE estimates increases as CPUE increases (independent those captured in the Charleston Harbor stratum were stocked of sampling effort). A more biologically meaningful α level of fish. Finally, after stocking about 620,000 small juveniles in the 0.10 was used for the CPUE statistical evaluations. Wando River during 2002, no stocked red drum were detected in any samples from the three strata the following year. Arnott et al. (2010) evaluated wild red drum CPUE from the RESULTS SCDNR trammel-net survey from 1991 to 2007 and provided a Genotypes were evaluated for departure from Hardy– template-standardized analysis across multiple strata, including Weinberg equilibrium at each locus and for linkage disequi- the Ashley River, Charleston Harbor, and Wando River. Their librium at each pair of loci. No significant differences were results illustrated the synchrony of age-1 red drum abundance detected between observed and expected allele frequencies af- between strata in each wild year-class (regional variability be- ter Bonferroni correction. A χ2 test of Mendelian inheritance tween year-classes along the South Carolina coast occurred in for all loci by using offspring (n = 39) from two known 2001 all estuaries simultaneously) and support the supposition that parental crosses confirmed that no loci significantly departed strong and weak year-classes of wild fish occur (Figure 2; see from expected Mendelian inheritance (P = 0.06–0.88, χ2 = Arnott et al. 2010 for complete description). It is among these 1.16–13.50, df = 2–7). Post hoc power estimations ranged from identified annual year-class strengths that we overlay results 0.66 to 0.81 for these analyses (G*Power version 3.1; Faul et al. from our stocking experiments for evaluation of their relation- 2009). ship to recruitment patterns. The genetic analyses determined that 88.9% of the age-1 The 1999 year-class contributed 51.2% of stocked fish Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 red drum captured in the Ashley River during 2000 were fish to the population in the combined Charleston Harbor estu- that had been stocked in 1999 (Table 3). In addition, 30.8% ary (Table 3) during a very weak natural recruitment year of the fish in the adjacent Charleston Harbor stratum had been (Figure 2). The 2000–2002 year-classes of wild red drum ap- stocked in the Ashley River and 14.6% of the fish captured in peared to increase in strength, and 2002 was one of the strongest the Wando River were also stocked. Fish were also released in year-classes reported from SCDNR sampling. Contribution of the Wando River during 2000, yet of fish sampled in the Wando stocked red drum showed an inverse relationship to wild red River during 2001, only 13.5% were identified as those stocked drum year-class strength (Figure 3). The high variability in in the Ashley River or Wando River. The samples collected in contribution between stocking years of similar-sized red drum the Charleston Harbor stratum showed that even with almost suggests that interpretation on a year-class-specific basis might 1 million fish stocked into two of its tributaries, only 6.7% of be necessary to fully understand the effects of stocking. sampled age-1 red drum were stocked fish. In the Ashley River, In 1999, seven different groups of red drum were stocked fish that had been stocked in 2000 comprised 35.6% of the in the Ashley River. Significantly larger-sized red drum were collected samples. The fish stocked in 2001 contributed 28.4% stocked earlier in the season, and smaller-sized red drum were of the sampled fish in the Ashley River during 2002, while stocked later in the season. Releases were pooled into two groups only 2.0% of those captured in the Wando River and 1.6% of (release batches 1–3 and 4–7) as the fish produced from batches 38 DENSON ET AL.

Ace Basin Ashley River 4 Charleston Harbor Wando River Muddy Banks Romain Harbor Winyah Bay 2

0 Standardized CPUEStandardized

-2 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 Year class

FIGURE 2. Long-term trends in standardized age-1 red drum catch per unit effort (CPUE) from 1988 to 2007 based on trammel-net surveys in seven independent strata along the South Carolina coast. Standardization protocols were developed based on the 1990–2007 year-classes; the mean CPUE of each stratum was standardized to a mean of 0 (shown here as a dashed line) and an SD of 1 (reprinted in part from Arnott et al. 2010, with permission).

1–3 had a larger mean size (61–64 mm TL) and were released termine the effects of release timing in relation to available food earlier in the season (mid-October–early November) than the resources or predator concentrations. Although we endeavored fish produced in batches 4–7, which had a mean of 22–27 mm to produce similar-sized red drum across stocking events, vari- TL and were released throughout November. The first release ations occurred due to weather conditions and optimum tide for group comprised 6.1% of the fish released in 1999, yet 46.8% stocking. All fish were, however, reared on natural zooplankton of those subsequently identified as stocked red drum from the blooms in ambient outdoor ponds during the normal spawning 1999 year-class were genetically identified as being from this season and were stocked synchronously with wild fish of the group (Table 4). Their relative survival (stocking success) was same size at age. Even if we had excluded the larger-sized fish almost 10 times higher than that of the smaller-sized red drum from our analyses, the patterns of contribution and relation- stocked later in the season. These data support the concept of ships to natural recruitment among years would have remained higher contribution from larger stocked fish. It is, however, im- the same (i.e., contribution in the Ashley River changes from possible to determine which fish in a broad size distribution 88.9% to 82.4%; Table 3). Additionally, we detected no dif- survived. In addition, for these stocking events we did not de- ferences in distribution patterns within the estuary among the Downloaded by [Department Of Fisheries] at 01:39 01 March 2012

TABLE 4. Summarized stocking impact of the 1999 year-class of red drum, comparing larger (batches 1–3) and smaller (batches 4–7) sizes at release (TL = total length). Percentage of Number Percentage of total Release Number Mean (range) TL Release total number captured at hatchery ﬁsh Stocking success batch stocked at release (mm) date released age 1 captured (× 10−5) 116,064 61.4 (52–84) Oct 13 213,703 64.1 (40–104) Oct 27 37,350 64.2 (31–86) Nov 3 6.12946.878.1 497,821 38.9 (29–47) Nov 3 5 156,112 26.8 (24–32) Nov 15 6 141,946 22.0 (19–26) Nov 19 7 178,312 27.5 (20–40) Nov 30 93.93353.25.7 GENETIC IDENTIFICATION TO ASSESS RED DRUM JUVENILE STOCKING 39

8 100 8 A. Charleston Harbor 7 7 Ashley River Charleston Harbor Wando River 80 Contribution Charleston Harbor 6 6 5 60 5 4 4 3 40

Contribution (%) Contribution 3

CPUE (fish per net set) per net (fish CPUE 2 CPUE per net set) (fish 20 2 1 1 0 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 0 Year class 1998 1999 2000 2001 2002 2003 8 100 Year class B. Lower Wando 7 Lower Wando FIGURE 4. Relationship of year-class catch per unit effort (CPUE) of age-1 80 red drum among the three strata of the Charleston Harbor estuary during the 6 Contribution stocking experiment (1999–2002). Significant differences between strata within α = 5 a year-class ( 0.10) are identified by black shaded and open symbols; gray 60 symbols represent values that are not significantly different from those depicted 4 with black shaded or open symbols.

3 40 Contribution (%) Contribution results, and given that all red drum were produced and released

CPUE (fish per net set) per net (fish CPUE 2 20 in phase with the natural spawning season of wild red drum, we 1 refer to the combined 1999 year-class data for the remaining discussion. 0 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 The Ashley River, Charleston Harbor, and Wando River strata Year class were sampled monthly during 2000–2003. Effort in each stratum 8 100 differed by the number of available sites, distance between sites, C. Ashley River 7 and speed at which the tide changed. Each year, 41–51 net sets 80 were made in the Ashley River stratum, 38–50 net sets were 6 made in the Charleston Harbor stratum, and 30–35 net sets were )

% made in the Wando River stratum (Table 3). Red drum from the 5 60 ( Ashley River 1999 year-class began to recruit to the sampling gear in July 4 Contribution 2000. The CPUE of age-1 red drum in the Ashley River was 3 40 1.73 ﬁsh/net set, which was not signiﬁcantly lower than that for

Contribution Contribution the two adjacent strata (P = 0.256, F = 1.38, df = 2; Table 3; CPUE (fish per net set) per net (fish CPUE 2 Figure 4), representing the first time since sampling began that Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 20 1 the CPUE in the Ashley River trended as high as the Wando River and Charleston Harbor strata. In contrast, during 2000 0 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 the red drum CPUE was significantly different among the strata Year class (P = 0.011, F = 4.65, df = 2); the Ashley River CPUE was significantly lower than the CPUE in the Charleston Harbor FIGURE 3. Relationship of year-class catch per unit effort (CPUE) and stock- stratum (P = 0.010, t = 2.96, df = 2). For the 2001 year- ing contribution of age-1 red drum sampled in the three strata of the Charleston Harbor estuary (Wando = Wando River). Age-0 red drum were stocked in the class, the CPUE in the Ashley River was significantly lower estuary from 1999 to 2002. (P = 0.009, F = 4.95, df = 2) than the CPUE in Charleston Harbor and the Wando River. The trend continued for the 2002 year-class, in which significantly fewer (P = 0.000, F = 8.67, size-groups (28 smaller red drum versus 20 larger red drum were df = 2) red drum were captured in the Ashley River than in recaptured in the Ashley River; three fish of both size-groups the other two strata. Comparisons of relative year-class strength were recaptured in the Wando River; two smaller fish versus six within strata for 1999–2002 showed that in the Ashley River, larger fish were recaptured in Charleston Harbor). Therefore, year-class strength for 1999 was not significantly lower (P = without replication of the size-at-release and timing-of-release 0.759, F = 0.39, df = 3) than that for 2000–2002 due to the 40 DENSON ET AL.

100 study represents a stocking of approximately 1.5 times as many similarly sized juvenile red drum as were stocked in our study Stocked Fish Ashley River during 1999, and the estuary they used for stocking is approx- 80 Wando River imately seven times larger (141,676 ha) than the Charleston Charleston Harbor Harbor estuary. It is conceivable that had ﬁsh been stocked into Galveston Bay at higher densities, contributions similar to those 60 reported here would have been observed. Karlsson et al. (2008) also reported that dispersion of red drum was nonrandom and

40 that hatchery ﬁsh were collected in greater numbers in close

Fish Sampled (n) Sampled Fish proximity to release sites despite stocking over a much larger area than was used in either South Carolina study. 20 In Florida, Tringali et al. (2008) documented a 2.8% contribution of stocked red drum from collections made throughout Tampa Bay and its neighboring waters (though sample collec- 0 tion was not random) after a release of 2,386,879 red drum ju- 1999 2000 2001 2002 Year class veniles of three size-classes. Tringali et al. (2008) also reported higher contributions of red drum captured closer to release sites, FIGURE 5. Summary of the number of wild and stocked red drum that were perhaps due to optimum habitat available at these locations for sampled from each of the three strata of the Charleston Harbor estuary for the specific life stages. In 1999, we found that red drum stocked 1999–2002 year-classes (shaded area of each bar = wild fish; hatched area = stocked fish). throughout the Ashley River were more densely aggregated in this stratum but had also moved into the adjacent strata and made substantial contributions within months of being stocked. high contribution (88.9%) of stocked red drum during a weak If adequate habitat and resources are available close to release year-class. The CPUE for the 1999 year-class was, however, sites, juvenile red drum appear to have no need to disperse. Fur- significantly lower than that for 2000–2002 in both Charleston thermore, their strong schooling behavior provides a successful = = = Harbor (P 0.002, F 5.35, df 3) and the Wando River strategy for survival, increasing the likelihood that higher con- = = = (P 0.002, F 5.35, df 3). These trends suggest that stocking centrations of stocked fish will be observed closest to stocking red drum during years of low natural recruitment can have a sites, as was shown in each of the studies. substantial impact on the population, while stocking similar The studies by Jenkins et al. (2004), Karlsson et al. (2008), numbers during years of high natural recruitment can have a and Tringali et al. (2008) for South Carolina, Texas, and Florida, much smaller effect (Figure 5). respectively, lacked a means to contextually evaluate the contribution of stocked red drum to a wild population. In these DISCUSSION studies, red drum contribution or stocking success was linked to size of fish at release, season of release, proximity to release Contribution and Natural Recruitment sites, habitat suitability, sample design, and a suite of other fac- Stock enhancement of red drum has been occurring since tors. We propose that in order to truly understand treatment the early 1980s in the southeastern United States. The three effects, contribution must also be considered relative to natural states that currently stock red drum (i.e., South Carolina, Texas, annual population variability. and Florida) typically release fish into diverse receiving waters, Fisheries-independent sampling in the Ashley River yielded Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 at varying densities, and using different methodologies, all of relatively high CPUE values for the 1999 year-class. During which affect their contribution to the population. Jenkins et al. poor recruitment years (1999) as well as during strong (2004) reported high stocking contributions of red drum in a year-classes, the addition of small juveniles at higher densities South Carolina estuary, although this study was limited by its appears to strongly influence fishery-independent indices of low number of sampling sites and their proximity to stocking abundance in systems with historically small populations of red sites, resulting in higher estimates of contribution in these ar- drum, such as the Ashley River. In 1999, the CPUE in the Ashley eas. In the present study, monthly sampling sites were chosen River was not significantly different from that in the other strata, at random from 90 sites throughout the 21,872-ha Charleston in contrast to 2000–2002. The CPUE increased in the Ashley Harbor estuary. Based on the random sampling design, overall River to a level higher than previously measured during a poor contribution to the wild population in South Carolina from the wild year-class (1999), suggesting that stocking had an additive 1999 year-class was 51.2%. In contrast, Karlsson et al. (2008) effect on the overall population rather than displacing wild fish reported that in Texas during 2004, 1.1 million red drum juve- through competition for limited resources. Additionally, these niles (29–40 mm TL) were stocked into Galveston Bay and that data suggest that resources are available for juvenile fish sur- subsequent genetic evaluation indicated a 9.35% contribution vival in the estuary even in year-classes that may be otherwise of stocked red drum to the wild population. Therefore, their identified as below average. During 2000, CPUE increased in GENETIC IDENTIFICATION TO ASSESS RED DRUM JUVENILE STOCKING 41

the Ashley River although the contribution of stocked fish to late July and August, a time that coincides with the peak red the total catch decreased, suggesting that the population had drum spawning season. Matlock et al. (1987), Scharf (2000), increased beyond historic averages and that the stocked fish con- and Arnott et al. (2010) found intermittent occurrence of strong stituted a much smaller proportion of the total population. In the red drum year-classes across estuaries and suggested that fac- Wando River, a system with typically high CPUE, during strong tors determining abundance and overall distribution might vary year-classes (2000–2001) the stocked fish constituted only 13% over a large spatial scale that may affect life stages differentially. of sampled fish and this proportion decreased to 2.0% in the year Those authors suggested that this could be due to egg or larval after stocking. One potential explanation for these observations dispersal from spawning adults, the subsequent movement of is that in 1999, the estuary may have lacked the environmental juveniles, or environmental conditions that affect estuaries sim- conditions, food availability, or both (Munro and Bell 1997) that ilarly along the coast. It is therefore plausible that the annual were necessary to sustain wild red drum larvae, yet the trophic variation in wild year-class strength reported by others and in resources necessary for survival of juveniles (>25 mm TL) the present study (i.e., for 1999–2002) is a major factor affecting were sufficient. In the Ashley River during 1999, the unusually interpretation of stocking efficacy and must be incorporated into high CPUE, which included a large proportion (88.9%) of stock enhancement programs. stocked red drum, demonstrated that the system had the capacity to support a larger population of juvenile red drum and that stocking would be a desirable management option if the goal is Genetic Identification to increase local red drum availability for recreational anglers. Criticisms of marine stock enhancement efforts have for Similar to these findings, previous stocking with small red drum many years largely centered on a lack of adequate technol- juveniles as reported by Jenkins et al. (2004) indicated that this ogy to identify the effects of stocking in largely open systems. small juvenile life stage bypassed the limiting factors that es- These limitations have been partly addressed through (1) the tablish year-class strength and recruitment into the recreational development of mariculture techniques that can produce fish for fishery; however, without long-term fishery-independent sam- stocking that match the life history stage present in the wild pling, it was not possible for the authors to determine the effects and (2) the incorporation of long-term tags to enable quan- of stocking in the context of relative intensity of wild year-class titative monitoring (Shelbourne 1964; Blankenship and Leber strength (Smith et al. 2003; Jenkins et al. 2004, 2005). From the 1995; Blaxter 2000; Chan et al. 2003). Our study utilizes mi- present study, it is clear that as wild year-class strength increased crosatellite genotyping as a means of identifying stocked fish, the effect of stocking diminished, likely due to competition for an approach that has recently become a cornerstone standard trophic resources or habitat. If stocked fish had a competitive tool for red drum stock enhancement research (Renshaw 2003; advantage over wild cohorts, however, we would expect that Jenkins et al. 2004, 2005; Gold et al. 2008; Karlsson et al. 2008; a large proportion of individuals collected in 2002 would also Tringali et al. 2008). have been identified as stocked fish. In contrast, during 2002 no Smith et al. (2003) evaluated the same samples collected in stocked fish were identified from fishery-independent sampling this study, but they used OTC marks as indicators of stocking; efforts, suggesting that any displacement of wild fish by stocked the OTC method showed only a 78% contribution of stocked fish was not occurring. One hypothesis for the lack of stocked fish to the Ashley River in 1999, a 12% difference from our fish from the 2002 year-class was the presence of a strong wild findings. In addition, their study indicated that the contribution 2002 year-class that followed two strong year-classes (2000 and from 1999 red drum stocking was 15% for Charleston Harbor 2001). It is therefore plausible that density-dependent factors (present study: 30%) and 12% for the Wando River (present (i.e., out-competition by the more numerous wild recruits or study: 15%). These differences point to the underestimation Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 age-1–3 red drum) contributed to failure of the 2002 stocking. of stocked fish contributions by using a qualitative technique Red drum exhibit a protracted spawning season that lasts rather than a quantitative technique. Denson and Smith (2008) from August to September, and different recruitment pulses of demonstrated that in laboratory trials, mark readability degrades larvae are produced during this period. The timing of these over time and is compromised by the cutting process and otolith pulses during the 2–3-month spawning window may have sig- morphological changes. Robbins et al. (2008) evaluated a sam- nificant effects on ultimate year-class strength due to potential ple of red drum from the 2000 and 2001 year-classes and also peaks in food availability as reflected in the size and health of found discrepancies for OTC-marked fish that were processed the juvenile red drum entering the winter period, which sub- by using both techniques. Additional stocked fish were identi- sequently impacts winter survival and year-class abundance fied with the genetic marks, and hatchery fish that were sampled (Scharf 2000; Stewart and Scharf 2008). Houser and Allen from the wild were identified at a probability of 99.99%. In (1996) monitored zooplankton, which are primary foods for addition to utilizing a nonlethal sample technique, the use of larval red drum (Daniel 1987), from May through October 1991 these markers can allow the determination of which broodstock in North Inlet, South Carolina. Peaks of copepods, fiddler crab animals yielded which offspring, which parental crosses were Uca spp. zoeae, larval grass shrimp Palaemonetes spp., and most prolific for each spawning event, and which release groups larval goby Gobiosoma spp. were documented to occur during provided the greatest contribution to the overall population. 42 DENSON ET AL.

In this study, multiple broodstock groups were used; there- REFERENCES fore, we were able to determine the effective contribution from Arnold, P. I., J. E. Serafy, M. E. Clarke, and D. R. Schultz. 1996. An immuno- two different spawning batches of red drum stocked during two logical study of predation on hatchery-reared, juvenile red drum (Sciaenops different months at two different sizes, similar to what has been ocellatus, Linnaeus): description of an ELISA and predator-prey studies in nature. Journal of Experimental Marine Biology and Ecology 199:29–44. reported in evaluating the effectiveness of two hatcheries releas- Arnott, S. A., W. A. Roumillat, J. A. Archambault, C. A. Wenner, J. I. Gerhard, ing fish at different times of the year in Texas (Karlsson et al. T. L. Darden, and M. R. Denson. 2010. Spatial synchrony and temporal 2008). The current stock enhancement studies at the SCDNR dynamics of juvenile red drum (Sciaenops ocellatus) populations in South rely on multiple broodstock groups to increase hatchery effec- Carolina, USA. Marine Ecology Progress Series 415:221–236. tive population size and to provide unique genetic combinations Beckman, D. W., and R. G. Schulz. 1996. A simple method for marking fish otoliths with alizarin compounds. Transactions of the American Fisheries of broodstock that permit progeny to be used in different stock- Society 125:146–149. ing treatments. By using multiple broodstock groups, more- Blankenship, H. L., and K. M. Leber. 1995. A responsible approach to marine complex stocking studies can be designed to evaluate critical stock enhancement. Pages 167–175 in H. L. Schramm Jr. and R. G. Piper, stocking size at release across a wider range of life history editors. Uses and effects of cultured fishes in aquatic ecosystems. American stages, including larval fish. Fisheries Society, Symposium 15, Bethesda, Maryland. Blaxter, J. H. S. 1976. Reared and wild fish: how do they compare? Pages 11–26 in G. Persoone and E. Jaspers, editors. Proceedings of the 10th European sym- Conclusions and Management Implications posium on marine biology, volume 1: research in mariculture at laboratory- The results of stocking 600,000 juveniles/year (∼25 mm and pilot scale. Universa Press, Wetteren, Belgium. TL) in the Ashley River during 1999–2001 indicated that these Blaxter, J. H. S. 2000. The enhancement of marine fish stocks. Advances in stocked fish did not displace wild fish but rather had an additive Marine Biology 38:2–53. Bravington, M. V., and R. D. Ward. 2004. Microsatellite DNA markers: evaluat- effect on overall red drum abundance, supporting the hypoth- ing their potential for estimating the proportion of hatchery-reared offspring esis that trophic resources are not necessarily limiting for the in a stock enhancement programme. Molecular Ecology 13:1287–1297. postlarval age-0 life stage. Variable year-class strength of wild Brooks, R. C., R. C. Heidinger, and C. C. Kohler. 1994. Mass-marking otoliths red drum could be driven by the prevailing environmental con- of larval and juvenile walleyes by immersion in oxytetracycline, calcein, or ditions during early life stages or by unsuccessful larval ingress. calcein blue. North American Journal of Fisheries Management 14:143–150. Chan, K. S., N. C. Stenseth, M. O. Kittilson, J. Gjosaeter, K. Lekve, T. Smith, S. If a lack of egg production or a lack of larval transport is the Tveite, and D. Danielssen. 2003. Assessing the effectiveness of releasing cod key limiting factor in a given year, then direct stocking of lar- larvae for stock improvement with monitoring data. Ecological Applications vae into the estuary appears to be capable of bypassing these 13:3–22. recruitment bottlenecks. Future research should replicate these Daniel, L. B. 1987. Aspects of the biology of juvenile red drum, Sciaenops efforts to evaluate the effect of year-class strength on stocking ocellatus, and spotted sea trout, Cynoscion nebulosus (Pisces: Sciaenidae) in South Carolina. Master’s thesis. University of Charleston, Charleston, contribution and should focus specifically on determining the South Carolina. factors limiting natural recruitment and linking these factors Danielssen, D. S., and J. Gjosaeter. 1994. Release of 0-group cod Gadus morhua with measurable environmental conditions. L. on the southern coast of Norway in the years 1986–1989. Aquaculture and As managers consider the use of red drum juveniles to aug- Fisheries Management 25:129–142. ment populations, the life stage that needs to be increased should Danzmann, R. G. 1997. PROBMAX: a computer program for assigning un- + known parentage in pedigree analysis from known genotypic pools of parents be considered. Red drum are long-lived fish (40 years), and and progeny. Journal of Heredity 88:333. the use of microsatellite genotypes as permanent markers will Denson, M. R., and T. I. J. Smith. 2008. Use of oxytetracycline to mark larval allow an evaluation of the contribution of stocked fish that re- red drum: survival and mark persistence. North American Journal of Fisheries cruit to the spatially separate adult population. Although our Management 28:1779–1789. research demonstrates a contribution to the year-classes at age Faul, F., E. Erdfelder, A. Buchner, and A. G. Lang. 2009. Statistical power Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 analyses using G*Power 3.1: tests for correlation and regression analyses. 1, determining the contribution of stocked red drum to the adult Behavior Research Methods 41:1149–1160. spawning population would be an appropriate next step. Ferguson, M. M., and R. G. Danzmann. 1998. Role of genetic markers in fisheries and aquaculture: useful tools or stamp collecting? Canadian Journal of Aquatic Sciences 55:1553–1563. ACKNOWLEDGMENTS Gold, J. R., L. Ma, E. Saillant, P. S. Silva, and R. R. Vega. 2008. Ge- We thank Theodore I. J. Smith, Stacey Robbins, and the staff netic effective size in populations of hatchery-raised red drum released for stock enhancement. Transactions of the American Fisheries Society 137: of the SCDNR’s WMC and Inshore Fisheries Research Sec- 1327–1334. tion for their assistance with this project. We also thank Peter Guo, S. W., and E. A. Thompson. 1992. Performing the exact test of Hardy- Kingsley-Smith, Matt Walker, Beth Cushman, and Justin Yost Weinberg proportion for multiple alleles. Biometrics 48:361–372. for providing comments on an early draft. This research was Houser, D. S., and D. M. Allen. 1996. Zooplankton dynamics in an intertidal supported by the U.S. Fish and Wildlife Service, Federal Aid salt-marsh basin. Estuaries 19:659–673. Jenkins, W. E., M. R. Denson, C. B. Bridgham, M. R. Collins, and T. I. J. Smith. in Sport Fish Restoration Program (Grant Number F-85-R4) 2002. Retention of oxytetracycline-induced marks on sagittae of red drum. and by the South Carolina Sea Grant Consortium (Grant Num- North American Journal of Fisheries Management 22:590–594. ber R-189). This is Publication 683 from the SCDNR Marine Jenkins, W. E., M. R. Denson, C. B. Bridgham, M. R. Collins, and T. I. J. Smith. Resources Division. 2004. Year class component, growth and movement of juvenile red drum GENETIC IDENTIFICATION TO ASSESS RED DRUM JUVENILE STOCKING 43

(Sciaenops ocellatus) stocked seasonally in a South Carolina estuary. North Robbins, S. N., T. L. Darden, A. O. Ball, and M. R. Denson. 2008. Genetic American Journal of Fisheries Management 24:636–647. parentage analysis of red drum in South Carolina and the use of genotypic Jenkins, W. E., M. R. Denson, M. R. Collins, and T. I. J. Smith. 1997. Differen- data from archived otoliths. North American Journal of Fisheries Management tiating between hatchery and wild red drum by position of the first annulus on 28:1796–1801. the sagitta. North American Journal of Fisheries Management 17:1001–1004. Rooker, J. R., G. J. Holt, and S. A. Holt. 1998. Vulnerability of newly settled red Jenkins, W. E., T. I. J. Smith, and M. R. Denson. 2005. Stocking red drum: drum (Sciaenops ocellatus) to predatory fish: is early-life survival enhanced lessons learned. Pages 77–86 in M. Nickum, P. Mazik, J. Nickum, and D. by seagrass meadows? Marine Biology 131:145–151. MacKinlay, editors. Propagated fish in resource management. American Fish- Ross, J. L., T. M. Stevens, and D. S. Vaughan. 1995. Age, growth, mortality, and eries Society, Symposium 44, Bethesda, Maryland. reproductive biology of red drums in North Carolina waters. Transactions of Jones, A. G., C. M. Small, K. A. Paczolt, and N. L. Ratterman. 2010. A prac- the American Fisheries Society 124:37–54. tical guide to methods of parentage analysis. Molecular Ecology Resources Scharf, F. S. 2000. Patterns of abundance, growth, and mortality of juvenile 10:6–30. red drum across estuaries on the Texas coast with implications for recruit- Karlsson, S., E. Saillant, B. W. Bumguardner, R. R. Vega, and J. R. Gold. ment and stock enhancement. Transactions of the American Fisheries Society 2008. Genetic identification of hatchery-released red drum in Texas bays and 129:1207–1222. estuaries. North American Journal of Fisheries Management 28:1294–1304. Serafy, J. E., J. S. Ault, T. R. Capo, and D. R. Schultz. 1999. Red drum, Sciaenops Kimura, M., and J. F. Crow. 1964. The number of alleles that can be maintained ocellatus L., stock enhancement in Biscayne Bay, Florida, USA: assessment in a finite population. Genetics 49:725–738. of releasing unmarked early juveniles. Aquaculture Research 30:737–750. Levene, H. 1949. On a matching problem arising from genetics. Annals of Shelbourne, J. E. 1964. The artificial propagation of marine fish. Advances in Mathematical Statistics 20:91–94. Marine Biology 2:1–83. Liu, Z. J., and J. F. Cordes. 2004. DNA marker technologies and their applica- Shelbourne, J. E. 1975. Marine fish cultivation: pioneering studies on the culture tions in aquaculture genetics. Aquaculture 238:1–37. of the larvae of plaice (Pleuronectes platessa L.) and the sole (Solea solea MacDonald, G. J., R. G. Danzmann, and M. M. Ferguson. 2004. Relatedness L.). HMSO, London. determination in the absence of pedigree information in three cultured strains Smith, T. I. J., W. E. Jenkins, M. R. Denson, and M. R. Collins. 2003. Stock of rainbow trout (Oncorhynchus mykiss). Aquaculture 233:65–78. enhancement research with anadromous and marine fish in South Carolina. Matlock, G. C., R. L. Colura, A. F. Maciorowski, and L. W. McEachron. 1987. Pages 27–39 in Y.Nakamura,J.P.McVey,K.Leber,C.Neidig,S.Fox,and Use of on-going tagging programs to validate scale readings. Pages 279–285 K. Churchill, editors. Ecology of aquaculture and enhancement of stocks: in R. C. Summerfelt and G. E. Hall, editors. The age and growth of fish. Iowa thirtieth U.S.-Japan joint meeting on aquaculture. Mote Marine Laboratory, State University Press, Ames. Sarasota, Florida. McEachron, L. W., C. E. McCarty, and R. R. Vega. 1995. Beneficial uses of Smith, T. I. J., W. E. Jenkins, M. R. Denson, and M. R. Collins. 2004. Increasing marine fish hatcheries: enhancement of red drum in Texas coastal waters. red drum abundance in South Carolina: monitor, regulate, and stock hatchery Pages 161–167 in H. L. Schramm Jr. and R. G. Piper, editors. Uses and fish. Pages 77–86 in M. Nickum, P. Mazik, J. Nickum, and D. MacKinlay, effects of cultured fishes in aquatic ecosystems. American Fisheries Society, editors. Propagated fish in resource management. American Fisheries Society, Symposium 15, Bethesda, Maryland. Symposium 44, Bethesda, Maryland. Munro, J. L., and J. D. Bell. 1997. Enhancement of marine fisheries resources. Solemdal, P., E. Dahl, D. S. Danielssen, and E. Moksness. 1984. The cod Reviews in Fisheries Science 5:185–222. hatchery of Flødevigen: background and realities. Pages 17–45 in E. Dahl, Paetkau, D., R. Slade, M. Burden, and A. Estoup. 2004. Genetic assign- D. S. Danielssen, E. Moksness, P. Solemdal, editors. The propagation of cod ment methods for the direct, real-time estimation of migration rate: a Gadus morhua L: part 1. Institute of Marine Research, Arendal, Norway. simulation-based exploration of accuracy and power. Molecular Ecology 13: Stewart, C. B., and F. S. Scharf. 2008. Estuarine recruitment, growth, and first- 55–65. year survival of juvenile red drum in North Carolina. Transactions of the Peters, K. M., and R. H. McMichael Jr. 1987. Early life history of the red drum, American Fisheries Society 137:1089–1103. Sciaenops ocellatus (Pisces: Sciaenidae), in Tampa Bay, Florida. Estuaries Tringali, M. D. 2006. A Bayesian approach for the genetic tracking of cultured 10:92–107. and released individuals. Fisheries Research 77:159–172. Perez-Enriquez, R., and N. Taniguchi. 1999. Use of microsatellite DNA as Tringali, M. D., K. M. Leber, W. G. Halstead, R. McMichael, J. O’Hop, B. genetic tags for the assessment of a stock enhancement program for red sea Winner, R. Cody, C. Young, C. Neidig, H. Wolfe, A. Forstchen, and L. bream. Fisheries Science 65:374–379. Barbieri. 2008. Marine stock enhancement in Florida: a multi-disciplinary, Piry, S., A. Alapetite, J. M. Cornuet, D. Paetkau, L. Baudouin, and A. Estoup. stake-holder-supported, accountability-based approach. Reviews in Fisheries

Downloaded by [Department Of Fisheries] at 01:39 01 March 2012 2004. GeneClass2: a software for genetic assignment and first generation Science 16:51–57. migrants detection. Journal of Heredity 95:536–539. Tsukamoto, K. 1988. Otolith tagging of Ayu embryo with fluorescent Rannala, B., and J. L. Mountain. 1997. Detecting immigration by using mul- substances. Bulletin of the Japanese Society of Scientific Fisheries tilocus genotypes. Proceedings of the National Academy of Sciences of the 54:1289–1295. USA 94:9197–9201. Tveite, S. 1971. Fluctuations in year-class strength of cod and pollack in south- Raymond, M., and F. Rousset. 1995a. An exact test for population differentia- eastern Norwegian coastal waters during 1920–1969. Fiskeridirektorates tion. Evolution 49:1280–1283. Skrifter, Serie Havundersøkelser 16:65–76. Raymond, M., and F. Rousset. 1995b. GENEPOP (version 1.2): population Vaughan, D. S., and J. T. Carmichael. 2000. Assessment of Atlantic red drum genetics software for exact tests and ecumenicism. Journal of Heredity for 1999: northern and southern regions. NOAA Technical Memorandum 86:248–249. NMFS-SEFSC-447. Renshaw, M. A. 2003. Use of microsatellites for genetic marking of hatchery- Volk, E. C., S. L. Schroder, and K. L. Fresh. 1990. Inducement of unique otolith stock individuals in the wild as part of red drum (Sciaenops ocellatus) banding patterns as a practical means to mass-mark juvenile Pacific salmon. stock enhancement research in South Carolina. Master’s thesis. University Pages 203–215 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. Jester Jr., of Charleston, Charleston, South Carolina. E. D. Prince, and G. A. Winans, editors. Fish-marking techniques. American Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223–225. Fisheries Society, Symposium 7, Bethesda, Maryland. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:40 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 Assessment of a New Gear to Sample Walleye Eggs Jordan D. Katt a c , Keith D. Koupal b , Casey W. Schoenebeck a & W. Wyatt Hoback a a Department of Biology, University of Nebraska at Kearney, Bruner Hall of Science, 2401 11th Street, Kearney, Nebraska, 68849, USA b Nebraska Game and Parks Commission, Fisheries Division, 1617 First Avenue, Kearney, Nebraska, 68847, USA c Nebraska Game and Parks Commission, 2200 North 33rd Street, Lincoln, Nebraska, 68503, USA Available online: 14 Feb 2012

To cite this article: Jordan D. Katt, Keith D. Koupal, Casey W. Schoenebeck & W. Wyatt Hoback (2012): Assessment of a New Gear to Sample Walleye Eggs, North American Journal of Fisheries Management, 32:1, 44-48 To link to this article: http://dx.doi.org/10.1080/02755947.2012.655842

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 32:44–48, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.655842

MANAGEMENT BRIEF

Assessment of a New Gear to Sample Walleye Eggs

Jordan D. Katt*1 Department of Biology, University of Nebraska at Kearney, Bruner Hall of Science, 2401 11th Street, Kearney, Nebraska 68849, USA Keith D. Koupal Nebraska Game and Parks Commission, Fisheries Division, 1617 First Avenue, Kearney, Nebraska 68847, USA Casey W. Schoenebeck and W. Wyatt Hoback Department of Biology, University of Nebraska at Kearney, Bruner Hall of Science, 2401 11th Street, Kearney, Nebraska 68849, USA

Existing gears for sampling walleye eggs (as well as gears Abstract for sampling other demersally spawning fish) are limited to A new gear for sampling walleye Sander vitreus eggs, egg sam- assessing egg deposition in specific environments or on a single pling disks, is described, and the ability of these disks to esti- substrate. In lotic waters, kick nets, anchored plankton nets, and mate walleye egg density on mud, cobble, and rip-rap substrates is demonstrated. Each egg sampling disk was a concave steel disk surber samplers (Corbett and Powles 1986; Pitlo 1989; Dustin covered with outdoor carpet. The ability of the disks to estimate and Jacobson 2003; Foust and Haynes 2007; Chalupnicki et al. walleye egg density and total egg deposition on a defined area was 2010) can be used to sample eggs because the water current tested at Sherman Reservoir, Nebraska. The average cost of each carries the suspended eggs into the net; however, in lentic $ egg sampling disk was US 8.00. An array of 10 disks could be waters, the lack of current limits the usefulness of such gears. processed (retrieved, complete the egg recovery procedure, and re- deployed) in approximately 1 h. The egg sampling disks were an Dip nets (Grinstead 1971; Foust and Haynes 2007) can be used appropriate gear for estimating egg deposition on mud, cobble, and in both lotic and lentic waters for presence/absence sampling, rip-rap substrate and could also be useful for estimating the total but cannot be used to derive an egg density. Passive egg traps egg deposition of walleyes and other demersally spawning fish. (Kelder and Farrell 2009) and samplers made of window screen (Michaletz 1984) are vulnerable to damage during severe Effective walleye Sander vitreus management often includes weather and would fill with sediment when used on mud substrate. Cement blocks covered with furnace filter (Manny et al. Downloaded by [Department Of Fisheries] at 01:40 01 March 2012 knowledge of important spawning areas. Walleyes prefer to spawn on cobble/gravel substrate (Johnson 1961; Pitlo 1989; 2007; Thompson 2009) are less prone to severe weather, but the Lowie et al. 2001; Foust and Haynes 2007), which is often ab- accumulation of sediment when used on mud substrate could reduce the egg capture efficiency and make finding the eggs in sent in reservoirs (Katt 2009). When cobble is absent, walleyes will spawn on other substrates (Johnson 1961), including sand the filter material difficult. Newburg (1975) used egg baskets, (Niemuth et al. 1959; Dustin and Jacobson 2003), silt or soft in which a hole was dug into the substrate where the basket was placed and the fill material was placed into the basket. The muck (Roseman et al. 2002; Dustin and Jacobson 2003; Foust and Haynes 2007), flooded terrestrial and aquatic vegetation egg baskets are time consuming to deploy/retrieve (Perkins and (Priegel 1970; Holzer and Von Ruden 1984; Roseman et al. Krueger 1994), require diver assistance, and would not work on rip-rap due to the large size of the boulders. Several studies 2002; Dustin and Jacobson 2003), and flooded rip-rap (Grin- stead 1971; Weber and Imler 1974; Michaletz 1984). have used a pump or suction device (Manz 1964; Grinstead 1971; Pitlo 1989; Fielder 2002; Roseman et al. 2002) to sample

*Corresponding author: [email protected] 1Present address: Nebraska Game and Parks Commission, 2200 North 33rd Street, Lincoln, Nebraska 68503, USA. Received December 21, 2010; accepted August 30, 2011

44 MANAGEMENT BRIEF 45

eggs from the substrate. Egg deposition estimates using these METHODS devices may be inaccurate on some substrates (such as rip-rap) Each egg sampling disk (Figure 1) was an individual steel because substrate irregularities make the area difficult to sample disk blade from an agricultural field disk. The disks were con- (Perkins and Krueger 1994). It is also unlikely these devices cave and ranged in diameter from 32 to 51 cm. After a 4.5- would work on rip-rap, because eggs would fall into the large cm-diameter steel washer was welded upright in the center of interstitial spaces between boulders, or on mud, because of the concave side of the disk, outdoor carpet was glued to the the large volume of sediment suctioned into the sample (Manz disk with a construction grade adhesive. Lead-core line ranging 1964). from 1 to 4 m in length was tied to the washer with a carabineer The limitations of these existing gears prompted the on the other end. An identification number was painted on the development of a new gear, called egg sampling disks. The convex side of each disk. The area of each egg sampling disk objective of this study was to describe the egg sampling disks was calculated based on its radius. and demonstrate their ability to estimate egg deposition on Laboratory tests were conducted to evaluate whether the substrates found in reservoirs (e.g., mud, cobble, and rip-rap, as percent of eggs recovered differed when two different people defined by the modified Wentworth scale in Cummins [1962]; washed the disks or varied with disk size. The cumulative Katt et al. 2011). The egg sampling disks are compared with percent egg recovery following multiple egg recovery proce- regard to the criteria of an ideal egg sampler (inexpensive, easy dures was also estimated. Nine individual disks (34–42 cm to process, not avoided by spawning fish, protective of eggs after diameter) were placed in individual 68-L tubs that were partially capture, durable, and unimpaired by severe weather; Marsden filled with water. One hundred viable walleye eggs were placed et al. 1991) and some assumptions and potential biases are on each disk. The eggs were obtained from a hatchery, had been discussed. treated with bentonite clay to remove adhesiveness (mucked), Downloaded by [Department Of Fisheries] at 01:40 01 March 2012

FIGURE 1. An individual egg sampling disk showing the position of the washer, outdoor carpet, and lead-core line. 46 KATT ET AL.

FIGURE 2. An array of egg sampling disks deployed on (A) mud substrate and (B) cobble and rip-rap substrate. Panel (A) depicts a full array of disks deployed on mud substrate while panel (B) depicts one-third of an array of disks deployed on the cobble and rip-rap substrates.

and were water-hardened. The following day, each disk was (mud, cobble, and rip-rap). Differences in site characteristics re- removed from the tub and subjected to the established egg quired use of two deployment methods. On the mud substrate recovery procedure (described below) three separate times, with (Figure 2a), two 2.4-m-tall t-posts were pounded into the reser- each recovery procedure being performed by a single person. voir substrate 3–5 m apart. Two u-bolts were attached to each t- The egg recovery procedure consisted of partially filling a tub post near the surface of the water, and five disks were attached to (38–53 L) with water and placing an individual egg sampling each post. On the cobble and rip-rap substrate (Figure 2b), three disk in the tub. The entire disk surface was scrubbed by hand cement anchor weights were deployed 3–5 m apart. Foamcore twice. The water was poured through a 500-µm pore-size sieve line (1–3 m) was used to attach each anchor weight to a float. A after the second scrubbing. The number of eggs recovered after 4.5–cm-diameter steel ring was placed around the foamcore line each procedure from each disk was recorded. This test was and three to four disks were attached to each ring. The egg sam- conducted again with a different person completing the egg pling disks were retrieved weekly and each disk underwent the recovery procedures the following day. egg recovery procedure (as described above) to check for eggs. The percent recovery of each person was calculated as the To calculate the walleye egg density, the number of eggs number of eggs recovered divided by the number of eggs still recovered from an array of disks was adjusted based on the available for recovery after the first recovery procedure (n = combined mean percent recovery from the laboratory test. The 9 for each washer). Analysis of covariance (ANCOVA) with adjusted number of walleye eggs sampled in an array of disks

Downloaded by [Department Of Fisheries] at 01:40 01 March 2012 interactions was used to evaluate potential differences in percent was then divided by the total area of the egg sampling disks egg recovery between each person and variation with disk size in the array (eggs/m2). This density was then divided by the (α = 0.05). Cumulative percent egg recovery was estimated after number of spawn nights since the last disk check (eggs/m2 per the second and third recovery procedure to evaluate potential spawn night) to standardize sampling effort. increases in percent egg recovery with additional effort. The total number of walleye eggs spawned ( ± SE) on a Evaluation of the egg sampling disks ability to estimate wall- known area was derived by extrapolating the walleye egg density eye egg density and total walleye egg deposition took place at to the entire area of that substrate. For this study, walleye egg Sherman Reservoir (41.3033◦N, 98.8824◦W), Nebraska. Sher- densities were extrapolated only on the cobble substrate because man Reservoir is an off-stream irrigation reservoir of the Middle the area of cobble was known (1,295 m2; Katt et al. 2011) and Loup River. At conservation pool, the reservoir covers 1,151 ha remained the same with rising water levels. The areas of mud and has a maximum depth of 11 m. Disks were deployed from and rip-rap were not calculated by extrapolation because of March 18 through April 15, 2008, and from March 25 through complications with the rising water levels. April 22, 2009, to estimate egg density throughout the early Since the egg sampling disks varied in diameter (32–51 cm), spawn, peak spawn, and late spawn periods each season. linear regression (α = 0.05) was used to determine whether Egg sampling disks were deployed in arrays of 10 disks, with disk size inﬂuenced the number of eggs sampled. To test this three arrays being randomly deployed on each of three substrates relationship, weekly egg catch data from individual disks at MANAGEMENT BRIEF 47

sampling sites where all the disks sampled eggs during a week However, because multiple disks were deployed in an array, egg were used. density could still be calculated when some of the disks in the array were lost or compromised. While the egg sampling disks meet the criteria of an ideal RESULTS egg sampler, the use of this gear does involve some underlying The mean percent egg recovery after the first egg recov- assumptions and potential bias. Egg loss from the disks during ery procedure did not differ between the two people (F = retrieval likely occurred, but the quantity of egg loss is unknown, 1.18, df = 1, 15, P = 0.28) or vary with disk size (F = which affects the accuracy of the density estimations. A similar 1.18, df = 1, 15, P = 0.30). Therefore, percent egg re- proportion of eggs was assumed to be lost from each disk be- covery data were pooled within each recovery procedure to cause all the disks were retrieved using the same method. This estimate a cumulative mean percent recovery. The cumula- assumption was also made by Manny et al. (2007), using egg tive mean number of eggs recovered ( ± SE) after the first, mats (cinder blocks covered with furnace filter). By making this second, and third egg recovery procedures was 66 ± 2, assumption, egg densities from each substrate can be compared 85 ± 1, and 91 ± 1 eggs, respectively. and should be unbiased. For the purpose of estimating total egg The adjusted walleye egg densities ranged from 0 to 37 deposition on a defined area or substrate, efforts to estimate eggs/m2 per spawn night on mud substrate, from 18 to 202 the percent egg loss during disk retrieval would be needed or eggs/m2 per spawn night on cobble substrate, and from 10 to acknowledgment should be noted that the estimates made are 3,682 eggs/m2 per spawn night on rip-rap (comparison of mean conservative, such as those made in this study. egg densities from different substrates can be found in Katt et al. A concern with the egg sampling disks was that the use of 2011). The total number of walleye eggs spawned on the cob- different sizes of disks would affect the accuracy of the density ble substrate ( ± SE) was estimated to be 2,389,275 ± 505,050 estimations. However, in this study the size of the disk did not in 2008 and 1,414,140 ± 145,040 in 2009. Disk size did not affect the number of eggs sampled in the field or the number influence the number of eggs sampled during either year of the of eggs recovered in the laboratory tests. The size difference study: 2008 (F = 0.07, df = 73, P = 0.80) and 2009 (F = 1.04, of disks used was small, which made finding a relationship df = 73, P = 0.31). between egg recovery and disk size unlikely. Ideally, the same size of disks would be used, but the difficulty of finding enough DISCUSSION disks of the same size was prohibitive. According to Marsden et al. (1991), the ideal egg sampler An additional possible bias is incomplete egg recovery, which would be inexpensive, easy to process, not avoided by spawning may influence the accuracy of the density estimates. Percent egg fish, protective of eggs after capture, durable, and unimpaired recovery did not differ between two people but laboratory tests to by severe weather. The egg sampling disks described here estimate egg recovery efficiency indicated that more eggs would satisfy many of these criteria. The average cost of each disk was be recovered in multiple egg recovery procedures. However, be- US$8.00, including the material and labor for constructing the cause egg recovery was incomplete even after three recovery disks. A four-person crew was able to process an array of disks procedures, the total number of eggs on each disk would still (retrieve, conduct the egg recovery procedure, and redeploy) in need to be extrapolated. Therefore, additional recovery proce- approximately 1 h. Spawning walleyes did not appear to avoid dures were determined to not be necessary. the egg sampling disks because walleye eggs were sampled The walleye eggs used in the laboratory tests came from a from each substrate type, but whether some walleyes avoided hatchery, where they were treated with bentonite clay (mucked) the disks because of the t-posts and foamcore line used in the to remove the adhesiveness and water-hardened. Treating the Downloaded by [Department Of Fisheries] at 01:40 01 March 2012 deployment is unknown. The egg sampling disks seemed to be eggs to remove the adhesiveness likely would not have influ- protective of the eggs after capture as is evident from the need enced the results of the laboratory tests because walleye eggs to use multiple recovery procedures to collect all the eggs from naturally lose their adhesiveness once they are water-hardened the disks. The outdoor carpet covering the disks also provided a (Pitlo 1989). Since walleye spawning typically occurs at night surface for eggs to adhere to after fertilization and the concave (Eschmeyer 1950) and processing of the disks did not begin shape of the disk may help retain eggs better than a flat disk. until mid-morning, the eggs on the disks in the field were likely The egg sampling disks were durable as most of the disks water-hardened and nonadhesive. Further, crews in the field did were used during both spawning seasons. Some disks needed not observe clumps of eggs when processing the disks, an indi- only minor repairs, such as replacing the outdoor carpet and re- cation that recovered eggs were no longer adhesive. Therefore, painting the identification numbers. Some disks were lost (6.3% the use of mucked (nonadhesive) eggs in the laboratory tests of all disks used throughout the study) as a result of wave action was an accurate representation of what was encountered in the caused by strong winds, which caused a disk to be displaced field. or rip-rap boulders to fall on a disk (3.7% of disks lost); rising The estimated egg densities from each substrate and the to- water levels in the reservoir, which washed out a t-post (1.2%); tal number of walleye eggs spawned on the cobble substrate angler disturbance (1.3%); and miscellaneous causes (0.1%). are conservative estimates. Several challenges limit the preci- 48 KATT ET AL.

sion of these estimates, including the unknown quantity of eggs Fielder, D. G. 2002. Sources of walleye recruitment in Saginaw Bay, Lake lost during disk retrieval, incomplete egg recovery during the Huron. North American Journal of Fisheries Management 22:1032–1040. washing procedures, and high variability in the egg density data. Foust, J. C., and J. M. Haynes. 2007. Failure of walleye recruitment in a lake with little suitable spawning habitat is probably exacerbated by restricted Egg loss during disk retrieval and egg recovery percentages for home ranges. Journal of Freshwater Ecology 22:297–309. individual people can be estimated and the densities adjusted Grinstead, B. G. 1971. Reproduction and some aspects of the early life history based on these estimations. The egg densities presented here of walleye, Stizostedion vitreum (Mitchill) in Canton Reservoir, Oklahoma. were adjusted based on the mean recovery percentage of two Pages 41–51 in G. E. Hall, editor. Reservoir ﬁsheries and limnology. American different people performing the laboratory test, but estimates of Fisheries Society, Special Publication 8, Bethesda, Maryland. Holzer, J. A., and K. L. Von Ruden. 1984. Determine walleye spawning move- egg loss during retrieval were not made. The variability in the ments in Pool 8 of the Mississippi River. Pages 1–78 in W. Fernholz, editor. egg densities is somewhat problematic, but may be reduced by Mississippi River work unit annual report, 1982–1983. Wisconsin Depart- sampling only during the peak spawn or by using different disk ment of Natural Resources, LaCrosse. arrays. Since densities of eggs were obtained throughout the en- Johnson, F. H. 1961. Walleye egg survival during incubation on several types of tire spawning period (early spawn, peak spawn, and late spawn) bottom in Lake Winnibigoshish, Minnesota, and connecting waters. Transac- tions of the American Fisheries Society 90:312–322. were used, a wide range of egg densities was observed, which Katt, J. D. 2009. Response of walleye to the addition of spawning substrate increased the variability in this study. This study used arrays of in a Nebraska irrigation reservoir. Master’s thesis. University of Nebraska, 10 disks as the sampling unit, but more arrays with fewer disks Kearney. per array could be used to increase the number of sampling units. Katt, J. D., B. C. Peterson, K. D. Koupal, C. W. Schoenebeck, and W. W. The egg sampling disks were an appropriate gear to estimate Hoback. 2011. Changes in relative abundance of adult walleye and egg density following the addition of walleye spawning habitat in a Midwest irrigation walleye egg densities and compare densities among mud, cob- reservoir. Journal of Freshwater Ecology 26:51–58. ble, and rip-rap substrate types. The egg sampling disks could Kelder, B. F., and J. M. Farrell. 2009. A spatially explicit model to predict also be useful for estimating the egg density of other demer- walleye spawning in an eastern Lake Ontario tributary. North American sally spawning species such as white bass Morone chrysops, Journal of Fisheries Management 29:1686–1697. lake sturgeon Acipenser fulvescenus, and lake trout Salvelinus Lowie, C. E., J. M. Haynes, and R. P. Walter. 2001. Comparison of walleye habitat suitability index (HSI) information with habitat features of a walleye namaycush, among others. The disks have already been used for spawning stream. Journal of Freshwater Ecology 16:621–631. presence/absence sampling of walleye and white bass eggs at Manny, B. A., G. W. Kennedy, J. D. Allen, and J. R. P. French III. 2007. First potential spawning sites in other Nebraska irrigation reservoirs evidence of egg deposition by walleye in the Detroit River. Journal of Great (Martin 2008). Lakes Research 33:512–516. Manz, J. V. 1964. A pumping device used to collect walleye eggs from offshore spawning areas in western Lake Erie. Transactions of the American Fisheries ACKNOWLEDGMENTS Society 93:204–206. Funding for this project was provided by Federal Sport Fish Marsden, J. E., C. C. Krueger, and H. M. Hawkins. 1991. An improved trap for Restoration Project No. F-166-R. F. Koupal designed and built passive capture of demersal eggs during spawning: an efﬁciency comparison with egg nets. North American Journal of Fisheries Management 11:364–368. the egg sampling disks. B. Peterson, C. Sullivan, S. Warner, Martin, D. R. 2008. Habitat selection by spawning walleye and white bass in M. Smith, D. Bauer, P. Spirk, L. Pape, M. Porath, N. Fryda, irrigation reservoirs of the Republican River basin, Nebraska. Master’s thesis. R. Lueckenhoff, B. Olds, P. Engler, D. Martin, and M. Feeney University of Nebraska, Lincoln. aided throughout this project. R. Eades provided useful com- Michaletz, P. H. 1984. Utilization of rip-rap by spawning walleyes in Lake ments. The Nebraska Game and Parks Commission provided Francis Case near Chamberlain, South Dakota. South Dakota Department of Game, Fish and Parks, Report 84–6, Chamberlain. equipment, Tradewinds Marina allowed the use of their facili- Newburg, H. J. 1975. Evaluation of an improved walleye spawning shoal with ties, and the University of Nebraska–Kearney provided technical criteria for design and placement. Minnesota Department of Natural Re-

Downloaded by [Department Of Fisheries] at 01:40 01 March 2012 support during this project. sources Section of Fisheries Investigational Report 340. Niemuth, W., W. Churchill, and T. Wirth. 1959. The walleye: its life history, ecology and management. Wisconsin Conservation Department, Publication REFERENCES 227, Madison. Chalupnicki, M. A., J. H. Johnson, J. E. McKenna Jr., and D. E. Dittman. Perkins, D. L., and C. C. Krueger. 1994. Design and use of mesh bags to estimate 2010. Habitat selection and spawning success of walleyes in a tributary to deposition and survival of ﬁsh eggs in cobble substrate. North American Owasco Lake, New York. North American Journal of Fisheries Management Journal of Fisheries Management 14:866–869. 30:170–178. Pitlo, J. Jr. 1989. Walleye spawning habitat in Pool 13 of the upper Mississippi Corbett, B. W., and P. M. Powles. 1986. Spawning and larva drift of sym- River. North American Journal of Fisheries Management 9:303–308. patric walleyes and white suckers in an Ontario stream. Transactions of the Priegel, G. R. 1970. Reproduction and early life history of walleye in the Lake American Fisheries Society 115:41–46. Winnebago region. Wisconsin Department of Natural Resources Technical Cummins, K. W. 1962. An evaluation of some techniques for the collection and Bulletin 45. analysis of benthic samples with special emphasis on lotic waters. American Roseman, E. F., W. W. Taylor, D. B. Hayes, and R. L. Knight. 2002. Evidence Midland Naturalist 67:477–504. of walleye spawning in Maumee Bay, Lake Erie. Ohio Journal of Science Dustin, D. L., and P. C. Jacobson. 2003. Evaluation of walleye spawning habi- 102:51–55. tat improvement projects in streams. Minnesota Department of Natural Re- Thompson, A. L. 2009. Walleye habitat use, spawning behavior, and egg de- sources Section of Fisheries Investigational Report 502. position in Sandusky Bay, Lake Erie. Master’s thesis. Ohio State University, Eschmeyer, P. H. 1950. The life history of the walleye Stizostedion vitreum vit- Columbus. reum (Mitchill), in Michigan. Michigan Department of Conservation, Institute Weber, D. T., and R. L. Imler. 1974. An evaluation of artiﬁcial spawning beds of Fisheries Research Bulletin 3, Ann Arbor. for walleye. Colorado Division of Wildlife Special Report 34. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:41 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 Comment: Rotenone Toxicity to Rainbow Trout and Several Mountain Stream Insects Don C. Erman a b a Department of Wildlife, Fish, and Conservation Biology, University of California–Davis, One Shields Avenue, Davis, California, 95616, USA b Retired Available online: 21 Feb 2012

To cite this article: Don C. Erman (2012): Comment: Rotenone Toxicity to Rainbow Trout and Several Mountain Stream Insects, North American Journal of Fisheries Management, 32:1, 53-59 To link to this article: http://dx.doi.org/10.1080/02755947.2012.655844

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 32:53–59, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.655844

COMMENT

Comment: Rotenone Toxicity to Rainbow Trout and Several Mountain Stream Insects

The use of rotenone (and other piscicides) in fish manage- METHODS ment is of concern to many, particularly because of its impacts Finlayson et al. (2010) exposed hatchery rainbow trout and on nontarget aquatic species and associated communities. De- six species of larval aquatic insects to a range of rotenone con- spite its long use in fisheries, rotenone is poorly understood centrations (based on dilutions of the two formulations) for 4 with regard to those impacts (Vinson et al. 2010). At present, or 8 h. The active ingredients of Nusyn-Noxfish include 2.5% the U.S. Environmental Protection Agency (USEPA 2008) has rotenone by weight, 2.5% piperonyl butoxide (a synergist), and withdrawn its approval of rotenone for terrestrial, estuarine, an additional 2.5% other rotenoids derived from the plant used and marine uses. The major reasons for the withdrawal are the in rotenone extraction; CFT Legumine contains 5% rotenone weakness of the scientific data on the environmental effects of and 5% other rotenoids. rotenone and suspected human health effects. It has also been Organism survival was to be judged 48 h after exposure withdrawn for terrestrial use by Canada (Health Canada 2008) in the replications of various rotenone concentrations and and the European Union (2008). Rotenone is still allowed for formulations. All tests were conducted in glass beakers under use in freshwater habitats as a piscicide. static test conditions with 5–10 replications of one individual In a recent paper, Finlayson et al. (2010) drew several per vessel per test concentration. The authors did not explicitly conclusions about the relative toxicities of two rotenone formu- address whether the test vessels were aerated, only that they lations under consideration for the latest project on Silver King changed water to maintain “sufficient oxygen concentration,” Creek on the eastern slope of the Sierra Nevada (Carson-Iceberg among other factors. Some of the key test conditions are Wilderness, Humboldt-Toiyabe National Forest), California. summarized in Table 1. Current planning is controversial and contested. On September The general test conditions for judging the outcomes of tox- 6, 2011, the Federal Court for the Eastern District of California icity testing can be assessed by ratios (loading factors) that use issued a permanent injunction on the project for violation organism mass, test volume, and time. The authors reported of the Wilderness Act because of its negative effects on loading factors defined in Rand and Petrocelli (1985). I assume aquatic macroinvertebrates (Case 2:10-cv-01477-FCD-CMK, they were referring to the chapter by Parrish (1985), who sug- document 65). State and federal agencies have been removing gested that test conditions were adequate if loading factors were nonindigenous fish there by poisoning for nearly half a century; less than 0.5–0.8 g/L in static tests. Finlayson et al. (2010) and more projects are planned (USFWS 2010). The purpose of reported the maximum loading they used. By my calculations,

Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 these projects is to remove all fish except the federally listed however, only one test condition had clearly satisfactory loading Paiute cutthroat trout Oncorhynchus clarkii seleniris from the (that for mayflies [0.024 g/L]), two factors were questionable, basin. The authors evaluated the lethal effect (dosage lethal to and one was clearly outside the recommended level (Table 1). 50% of the test animals [LC50]) of Nusyn-Noxfish relative to (No values could be calculated for caddisflies because average that of CFT Legumine on rainbow trout Oncorhynchus mykiss body mass was not reported, but the loading factor was some- and six species of stream insects (Ephemeroptera [mayflies]: where between the others.) All tests were designed to last 48 h, so Baetis tricaudatus, Rhithrogena morrisoni, Plecoptera [stone- I also calculated the loading factor suggested by Sprague (1990) flies]: Classenia subulosa, Oroperla barbara; Trichoptera [cad- from his chapter on aquatic toxicology in the American Fish- disflies]: Arctopsyche grandis, Hydropsyche tana/amblis). They eries Society publication Methods for Fish Biology (Table 1). extrapolated the findings to field data from a previous Nusyn- Sprague (1990) recommended incorporating time in the loading Noxfish poisoning of Silver King Creek. I question the findings factor: test volume (L) divided by organism mass (g) times days and conclusions from this study based on the methods used, the (2 d, in this case, for the judgment of mortality). Based on his statistical analysis, and the selection of field data for analysis. experience and review, Sprague (1990) recommended a loading My comments follow the organization of Finlayson et al. factor greater than 2 if there were no aeration of test vessels (2010). or greater than 1 under other circumstances. Only the average

53 54 ERMAN

TABLE 1. Organisms and conditions used in toxicity tests of two rotenone formulations (after Finlayson et al. 2010). See text for invertebrate species names.

Test group Mean mass (g) Liquid volume (mL) Loading factora (L·g−1·d−1) Loading factorb (g/L) Rainbow trout (Nusyn-Noxﬁsh) 1.25 1,200 0.48 1.04 Rainbow trout (CFT Legumine) 0.22 300 0.68 0.73 Ephemeropteraa 0.0072 300 20.8 0.02 Plecopterab 0.206 300 0.73 0.69 Tricopteracc300 cc

aFrom Sprague (1990), who recommended values greater than 1–2 L·g−1·d−1. bFrom Parrish (1985), who recommended values less than 0.5–0.8 g/L. cData not available.

case for mayflies (20.8 L·g−1·d−1) satisfied this recommended attempt to compensate for the lack of flow, such species often loading factor for adequate water conditions. Caddisflies would move their bodies or body parts for short periods. Many ob- not have met adequate loading if the individuals had averaged servers of stream insects are familiar with the behavior of the more than 0.15 g (a loading of 1) or 0.075 g (a loading of 2). stoneflies and mayflies tested. When removed from flow, stone- The study methods also incorporated bias in the sizes used in flies (e.g., Claassenia) soon begin to perform “push-ups” in an the rainbow trout assays (Table 1). Fish averaged 5.7 times larger attempt to create a current, whereas many mayflies, particularly (1.25 g) in the Nusyn-Noxfish tests than in the CFT Legumine the Rhithrogena (family Heptageniidae), vibrate their abdominal tests (0.22 g). Differences in size, age, and other factors affect gills (Hynes 1970; Eriksen et al. 1996). The net-spinning cad- the level of toxicity for various chemicals (Sprague 1990). I am disflies (Hydropsychidae) used in Finlayson et al.(2010) require unaware of published studies that compare the effects of rainbow a current for retreat construction and net deployment, and they trout size on toxicity. It is well known that most other physio- are similarly affected in their oxygen consumption when current logical responses are sensitive to fish size. Thus, the standard is restricted (Hynes 1970). As a consequence, the organisms in research design should take size into account by either random- the test conditions (static tests lacking water current) used by izing sizes among tests, keeping size similar among tests, or Finlayson et al. were under continuous stress and would have using covariance analysis in some way. For one thing, using impaired uptake of any chemical that requires passage through different fish sizes for each toxicant caused one group to have their external gills (e.g., rotenone). Subsequent estimates of mor- an even worse loading factor than the other—quite apart from tality from poisons are probably biased and of uncertain rele- whether size itself would change their sensitivity to the poisons. vance. Whether such unnatural conditions in a laboratory mean The effect of such a difference in rainbow trout sizes confounds that toxicity is lower or higher in the field is speculative. One the test and raises doubt about the conclusions from the results could just as well make the argument that because of the poor comparing the two formulations. laboratory conditions and reduced oxygen uptake the organisms In addition to information on fish size, the authors should in the laboratory took up less toxicant (gills are the location have provided information on larval insect stage or instar. The for uptake) and hence survived better (longer) than they would toxicity of pesticides to larval insects is also influenced by age have under field conditions. Thus, any conclusions about field (larval instar). As shown for midge larvae (Diptera: Chironomi- survival are no more than speculation and have no place in the dae), the toxicity of a common pesticide (chlorpyrifos) nearly conclusions from the results. As concluded by Sprague (1985), Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 doubled over one difference in larval instar stage (Buchwalter “[i]t is therefore important that the investigator be familiar with et al. 2002). This finding means that comparison of the sensitiv- the life history and general requirements of organisms used ity of a toxin among or within species may depend on the instar in testing.” in which exposure occurs. Without such information, the gen- The three factors combined in the invertebrate tests— eralizations or extrapolations to field sensitivity by Finlayson inadequate container loading in some tests, the lack of current, et al. are misleading. and unknown instar—are severe constraints. Finlayson et al. The overall test conditions for the invertebrates were more (2010) acknowledge (page 106) that the failure of some tests to problematic than those for the rainbow trout. All of the insect last 48 h and the mortality of controls “may suggest test con- species chosen for testing are flow-dependent organisms. That ditions less than optimal for these species.” I suggest that the is, they require flow over their bodies to survive or function nor- basic conditions were inadequate for any conclusions. mally. In his classic work, Hynes (1970) reviewed this concept at some length. Without a current, these organisms shut down oxygen consumption even when the oxygen concentration in the RESULTS water is maintained at high levels. All six of the insect species Even if the test conditions had been adequate, the study failed tested possess external gills. But they depend on water move- to use statistical inference to make valid conclusions about the ment rather than static diffusion through external surfaces. In an results. The authors reported LC50 means (geometric) and 95% COMMENT 55

confidence intervals (CIs) in statistical tests but apparently relied tions. Thus, six cells or three pairs of comparisons of relative on inspection of the means for judging their significance. The rotenone formulation toxicity remained. Two of these compar- tests on rainbow trout are instructive in this regard. isons were for the same stonefly species (O. barbara)atthe For this test, the lack of water current is not critical although two exposures; the last was for the mayfly R. morrisoni at 8 h. loading problems remain. Nevertheless, the authors claimed What may appear as a fairly robust study of species, exposures, that the toxicity of CFT Legumine (LC50 = 7.4 µg/Lin4h, and formulations, was in fact two independent tests of relative 5.3 µg/L in 8 h) was greater than that for synergized Nusyn- toxicity under highly questionable laboratory conditions. Nev- Noxfish (LC50 = 7.7 µg/L in 4 h, 6.2 µg/L in 8 h) and there- ertheless, whatever the results of subsequent statistical tests or fore that the synergist in piperonyl butoxide had little effect on comparisons, generalizations about toxicity are suspect because the toxicity of rotenone for rainbow trout. But the 95% CIs of of poor methods. the means of the two formulations (their Table 2) overlapped for Based on their findings, Finlayson et al. (2010) recom- both the 4-h and 8-h tests. Use of confidence intervals acknowl- mended that fisheries managers “use unsynergized formulations edges that the means have statistical properties to be consid- because the synergized formulation is less toxic to fish and ered when judging “difference.” Therefore, the true mean value more toxic to aquatic insects” (page 109). But the findings were could be anywhere in the CI range (with a probability used untrue for rainbow trout (there was no difference) and unproven in the confidence interval). From these data alone, the authors for aquatic insects. should recognize that the actual means may not really be differ- There are additional questions concerning extrapolations ent. Again, Sprague (1990) recommended how to judge toxicity from this toxicity study that pertain to the nature of rotenone for- results, particularly when confidence intervals overlap. He pro- mulations themselves. As noted above, the make-up and amount vided a simple test statistic using average LC50 and confidence of active ingredients (all chemicals with pesticidal action) in the intervals called the f 1,2 test: two formulations are different. As the authors pointed out, to use equal concentrations of rotenone twice as much Nusyn-Noxfish 2 2 f1,2 = antilog{square root [(logef1) + (logef2) ]}, was used as CFT Legumine. In terms of all active ingredients, therefore, a test concentration of 10 µg rotenone/L made up where f 1 is the ratio of the upper 95% confidence limit to the from Nusyn-Noxfish would have 30 µg total active ingredi- average LC50 for substance 1 and f 2 is the analogous ratio for ents/L, compared with 20 µg/L for CFT Legumine. Other fish substance 2. A difference is significant if the test statistic is less toxicants with different formulations (e.g., Synpren-Fish Tox- than the ratio of the greater average LC50 divided by the lesser icant, which contains 2.5% rotenone, 2.5% piperonyl butoxide, LC50. and 5.0% other cube resins) have total active ingredients that Application of this statistical test shows that there was no when doubled to equal the 10 µg rotenone/L concentration of significant difference (95% level) in the toxicity to rainbow CFT Legumine would contain 40 µg total active ingredients/L. trout of the two formulations at either 4 or 8 h of exposure; Further, other synergized fish toxicants use synergists other thus, toxicity was not greater for CFT Legumine than for than piperonyl butoxide (e.g., Pro-Noxfish uses sulfoxide as a Nusyn-Noxfish. The inference that the synergist in Nusyn- synergist) that may have other properties. (The active ingredi- Noxfish had little effect on mortality remains; however, the ents in several common fish toxicants can be found in Finlayson bias caused by using larger fish (and poorer loading conditions) et al. 2000.) Generalizing about toxicity for the variety of with Nusyn-Noxfish than with CFT Legumine renders even formulations and active ingredients is risky for fish and even this conclusion risky. Marking (1977) reported much earlier more so for the enormous diversity of the invertebrate world. that piperonyl butoxide caused more than additive toxicity with Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 rotenone when used on rainbow trout. Field Application of Invertebrate Toxicity Finlayson et al. (2010) also concluded that the results showed for Silver King Creek that the toxicity of Nusyn-Noxfish to aquatic invertebrates was The authors used their laboratory data to compute toxic units generally greater because of the synergist (“may be twice as (TU) of field exposure to rotenone (Finlayson et al. 2010; their toxic” [page 107]) than the CFT Legumine formulation. Putting Figure 2), where TU (from Rand and Petrocelli 1985) is the field aside the issues of the test conditions, I examined the basis for rotenone concentration divided by the LC50 for a species. If the this generalization. The study used six species of insects, two TU for the average LC50 (or confidence interval) of a species exposure periods, and two rotenone formulations, potentially exceeded the field rotenone concentration, at least 50% of the yielding 24 data cells of average LC50 and 95% confidence population would be killed. Finlayson et al. used LC50 for the intervals. Nine cells lacked data either entirely or for the con- nine data cells at 8 h exposure that had any data, regardless of the fidence interval. Four other cells did not represent results for differences in 24 or 48 h to observe mortality. They used a field 48 h but were truncated at 24 h because of problems or early rotenone exposure from 1991 to 1993 in Silver King Creek of mortality. Of the remaining 11 cells, 5 had data only for one 11 µg/L for 6–18 h, reportedly based on Trumbo et al. (2000a). formulation for a given species and exposure time and could Trumbo et al. however, indicated that rotenone exposure not be used to compare the relative toxicities of the formula- in Silver King Creek (from Table 1 in Trumbo et al. 2000a) 56 ERMAN

averaged 22 h (range, 18.5–24 h) based on the only reported 2010). Such a protocol is also standard procedure in Arizona data from a downstream monitoring station. Furthermore, in (USBR 2007). each of the three years of rotenone application (1991–1993), In the area of the project conducted in the 1990s, Silver King two applications of Nusyn-Noxfish were made, separated by Creek had also been poisoned previously (USFWS 2010), first 1–2 d. In 1991, a third application was made 1 month later to with Pro-Noxfish (50 µg rotenone/L plus the synergist sulfox- the upper half of the basin when fish were found upstream from ide) in 1964 and again in 1976 (first with antimycin and then the earlier two applications, and the stream at its lower loca- twice more with Pro-Noxfish at 50 µg rotenone/L). Some stream tions (but above the downstream boundary at Llewellyn Falls) sections have received as many as eight separate rotenone ex- “was heavily doused with rotenone from backpack sprayers” posures over 29 years (USFWS 2010). Fish, and presumably when caged sentinel fish escaped to the treatment area (Flint aquatic invertebrates, develop resistance to rotenone when it is et al. 1998:22). It is apparently standard protocol, at least in used repeatedly (Orciari 1979), as is well known for insects’ California, that managers make two closely spaced, separate responses to other pesticides. Hence, one might expect increas- rotenone applications in a year. Two applications a year for 3 ing tolerance in those species that survive previous rotenone years (1994–1998) were also made in Silver Creek, California projects, which would make the response of the invertebrate (Trumbo et al. 2000b), and two applications (separated by 2 community in Silver King Creek difficult to extrapolate to other weeks) in 2007 were made to the three major tributaries of Lake locations. Davis, California (McMillin and Finlayson 2008). Plans for the latest poisoning of Silver King Creek called for application of Field Samples of Invertebrates from Silver King Creek a rotenone formulation twice a year for 2 or 3 years (USFWS Finlayson et al. (2010) also analyzed some invertebrate 2010). (A recent environmental assessment of a similar project bottom-sample data from the rotenone project on Silver King in Arizona states that a similar frequency of application is stan- Creek in the 1990s. They compared data (their Table 3) using 14 dard in that state; USBR 2007). The relevance of the toxic unit invertebrate metrics for five poisoned stations and one unpoi- concept or toxicity data in general (e.g., Table 1 in Finlayson et soned (control) station from 1991 to 1996. Personnel collected al. 2010) to two or more closely spaced rotenone applications is samples before and after rotenone applications in each year unknown and unstudied. from 1991 to 1993 and yearly thereafter. Thus, in this data set The implication from the toxic unit analysis (Figure 2 in Fin- the samples from 1991 before poisoning represented the only layson et al. 2010) was that rainbow trout would certainly be “before” data (complete sample data from 1990 for all the sta- killed (i.e., suffer more than 50% mortality) by either formulations have been lost, and the 1991 data are 15 and 27 years after tion despite the obviously limited sizes and ages of the test fish), earlier poisoning). From analysis of variance (ANOVA) the au- whereas all of the aquatic insects (with the possible exception thors found significantly lower density (critical α = 0.05) in the of the mayfly R. morrisoni, whose upper confidence interval poisoned stations than in the unpoisoned control for total inver- exceeded the toxic unit threshold) would suffer less than 50% tebrate abundance and Coleoptera abundance. But none of the mortality. The authors concluded that some mortality would oc- other 12 metrics of invertebrate assemblages were significantly cur for both species of mayfly tested and that there would be different (Table 3 in Finlayson et al. 2010). On this basis, they little, if any, mortality for the caddisfly A. grandis or the stonefly argued that there was limited impact on the macroinvertebrate O. barbara. They drew no conclusions about the caddisflies in community from the poisoning. the genus Hydropsyche spp. or the stonefly C. subulosa, possibly The use of ANOVA in this way may be statistically valid in because they lacked data on Nusyn-Noxfish toxicity for those a general sense, but it is uninformative. Such an analysis maxi- taxa. Nevertheless, they concluded that if the study were rep- mizes the importance of variability and insensitivity: sequential Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 resentative of toxicities and other mountain stream invertebrate collections during 6 years, and collections before poisoning and species, during poisoning years are not separated by effect. As noted by the authors, five poisoned stations and one control (with atten- a rotenone treatment with a mean rotenone concentration of 11 µg/L dant differences in the total number of bottom samples) were with an exposure time of 6–18 h would result in complete mortal- compared. In addition to the variability from the treatments ity to trout and some mortality of invertebrate species, but many invertebrate species would survive (page 109). and time, there was confounding of variables in both space and time (see the discussion in Waters and Erman 1990), along with In Silver King Creek (and elsewhere), however, multiple ap- station variability from the wide range of stream sizes, eleva- plications of rotenone were made each year, and the average tions, past livestock grazing, grazing exclusion at some stations, exposure time was 22 h. None of the tests conducted by Fin- stations without fish predators part of the time, habitat restora- layson et al. (2010) exceeded one-time exposures beyond 8 h. tion efforts, and so forth. Without any attempt to account for Recent planning for another poisoning of Silver King Creek this variability (say, by measuring the factors and using them as downstream from the areas exposed in the 1990s also called for variables), the conventional method of increasing sample size by the application of CFT Legumine or Noxfish twice a year for adding more stations or times may only add variability (Downes 2–3 years at rotenone concentrations of 25–50 µg/L (USFWS 2009). COMMENT 57

10000

1000

Abundance based

100

10 Taxa based Standard Deviation of Metric

0.1 1 10 100 1000 10000 100000 Mean of Metric

FIGURE 1. Means and standard deviations of 14 metrics used by Finlayson et al. (2010) to compare macroinvertebrates collected from rotenone-poisoned (solid squares) and control (open squares) stations in Silver King Creek. The underlying data (yearly estimates) for poisoned stations (solid triangles) are shown for comparison; yearly estimates for a single control station lack a standard deviation.

The ANOVA analysis used by Finlayson et al. (2010) is dent variables would make little difference in the other sources also unlikely to detect differences because of the high degree of variability. As presented, the analysis by Finlayson et al. of dependence of the standard deviation on the means. A loga- (2010) maximizes the likelihood of finding few differences in in- rithmic plot of the 14 metrics by mean versus standard deviation vertebrate communities from rotenone poisoning within the wel- for control and poisoned stations is given in Figure 1. The under- ter of unaccounted-for variation and weak statistical methods. lying values from yearly estimates of total density and the total The data set could be analyzed to avoid some of the statisti- number of taxa for poisoned stations are also shown for com- cal problems. A hypothesis of the Finlayson et al. (2010) study parison and have the same relationship. Yearly estimates of the is that for most of the 14 metrics there was no difference be- single control station lack a standard deviation. This dependence tween control and poisoned stations. If these 14 metrics reflect violates a major assumption of the ANOVA model. Although their importance as measures of the effects of rotenone, then a the 14 metrics were each considered as different factors in the reasonable hypothesis is that there is no difference in the ranks analysis by Finlayson et al., 5 of the metrics are subsets of of the 14 metrics based on poisoned or unpoisoned stations. I Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 the total abundance metric (i.e., the number of individuals in the tested such a ranking using the nonparametric Wilcoxon signed various orders) and the remaining 8 metrics (taxa counts) are rank test. This null hypothesis was rejected (P = 0.0088), that is, themselves subsets of the total number of taxa. there was probably a significant negative effect of the rotenone Analysis of variance is said to be fairly forgiving as to the poisoning on the invertebrate community in Silver King Creek. violation of assumptions, and a significant difference from a test There was contention over the latest project to poison Silver is probably valid even when the assumptions of the model are King Creek (Finlayson et al. 2010:103), in large part because violated (Chittenden 2002). But strong departures from model of the lack of species identification of the invertebrate samples. assumptions mean that the lack of significant differences is a As Finlayson et al. (2010) and others (Sprague 1990) point out, likely outcome that may not be true. For some time, invertebrate organisms, particularly invertebrates, vary in their sensitivity to ecologists have recognized that abundance data from inverte- toxins (and possibly the other active ingredients and interactions brate samples are prone to this dependence of the standard devi- of the chemicals in the formulations). Thus, one might expect ation on the mean (Elliott 1977). Logarithmic transformation of attention to focus both on determining the species composition the abundance data (Elliott 1977), other transformations (Allan of the community and on whether or not the species composition 1984), or nonparametric methods are possible improvements to is altered after poisoning. However, for the invertebrate collec- the analysis. In this case, however, transformation of the depen- tions made during the 1990s and earlier in Silver King Creek, 58 ERMAN

more than one-half of the individuals (53.7%) were identiﬁed 900 only to family or a higher classiﬁcation based primarily on larval A Poisoned 800 Control specimens. From 1984 to 1996, collections from the stations in Silver 700 King Creek made by various state and federal agency personnel 600 were sent to the former U.S. Forest Service National Aquatic ± 1 SE)

2 500 Ecosystem Monitoring Center in Provo, Utah, for processing and identiﬁcation. Annual reports of data analysis from the cen- 400

ter noted the loss of many taxa that were present before poison- 300

ing began in 1991. From the original data sheets, I used the same Density (N/m list of taxa (family, genus, and some species level identification) 200 for 1991 at each station as the base before poisoning. I compared 100 this list with that of subsequent samples through 1996 for the 0 five poisoned stations and one control (Figure 2). The percentage 1990 1991b 1991p 1992b 1992p 1993b 1993p 1994 1995 1996 similarity of taxa composition, even at these levels of identifi- 160 cation, was significantly lower in the poisoned stations (mean, B 53.4%) than in the control (68.0%) (Mann–Whitney U-test, P 140 = 0.0004). A large difference in similarity between the control (about 70% similarity to prepoisoned) and unpoisoned stations 120 (about 58%) remained 2 and 3 years after the last poisoning in 1993 (Figure 2). 100 ± 1 SE) Further effects on specific taxa were apparent in the origi- 2 80 nal data. In particular, the abundance of the cockroach-shaped stonefly Yoroperla (sometimes identified to Y. brevis), the most 60 abundant (and easily identified larva) stonefly in Silver King Density (N/m Creek before poisoning began, sharply declined with each poi- 40

0 1994b 1994p 1995b 1995p 1996b 1996p 1997 1998 Sample Period

FIGURE 3. Mean density of Yoroperla (Peltoperlidae) stonefly larvae from (A) five poisoned and one control station in Silver King Creek and (B) four poisoned stations in Silver Creek, California. Sample years followed by letters refer to periods immediately before (b) or after (p) rotenone (Nusyn-Noxfish) application.

soning event until in the last 2 years of the study (2–3 years after Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 the last application of Nusyn-Noxfish), scarcely any individuals were found in the poisoned stations (Figure 3). A nearly identical decline occurred in abundance of Yoroperla (Figure 3) after a similar poisoning project on nearby Silver Creek (Trumbo et al. 2000b). These results for Silver Creek (1994–1998) further strengthen the inference from Silver King Creek (1991–1996) because the two studies had little overlap in time, thus reducing possible climatic or hydrologic variation as a confounding factor. As might be expected, other taxa expanded in abundance, FIGURE 2. Similarity in macroinvertebrate composition based on taxa col- although the overall composition of the community and the rel- lected after rotenone (Nusyn-Noxfish) poisoning of Silver King Creek and taxa ative abundances of various taxa differed from the preproject collected (and identified to family or finer classification) in 1991 before poi- condition. The conclusion by Finlayson et al. (and, for that mat- soning. The mean similarity before poisoning in 1991 for five poisoned stations ter, Trumbo et al. 2000a, 2000b) that past poisoning caused few was significantly less (P = 0.0004) than for the control station. Sample years followed by letters refer to periods immediately before (b) or after (p) rotenone significant effects on the macroinvertebrate community of Silver application. King Creek is incorrect. COMMENT 59

Conclusions Finlayson, B. J., W. L. Somer, and M. R. Vinson. 2010. Rotenone toxicity to The study by Finlayson et al. (2010) had serious methodolog- rainbow trout and several mountain stream insects. North American Journal ical problems in toxicity testing and analysis that render their of Fisheries Management 30:102–111. Flint, R. A., W. L. Somer, and J. Trumbo. 1998. Silver King Creek Paiute conclusions suspect or incorrect. Similarly, their analysis of the cutthroat trout restoration, 1991 through 1993. California Department of Fish invertebrate data collected from Silver King Creek failed to ac- and Game, Inland Fisheries Administrative Report 98-7, Sacramento. count for variability and violated test model assumptions and Health Canada. 2008. Reevaluation note: rotenone. Available: www.hc- thereby clouded the impacts from poisoning. My analyses of sc.gc.ca/cps-spc/pubs/pest/ decisions/rev2008–01/index-eng.php. (March their data show significant impacts from Nusyn-Noxfish poison 2011). Hynes, H. B. N. 1970. The ecology of running waters. University of Toronto on a variety of metrics and on certain taxa found in the benthic Press, Toronto. invertebrate community of Silver King Creek. Marking, L. L. 1977. Methods for assessing additive toxicity of chemical mix- tures. Pages 99–108 in F. L. Mayer and J. L. Hamelink, editors. Aquatic toxicology and hazard evaluation. ASTM (American Society for Testing and ACKNOWLEDGMENTS Materials), Special Technical Publication 634, Philadelphia. I thank John G. Williams and Nancy A. Erman for their McMillin, S., and B. J. Finlayson. 2008. Chemical residues in water and sed- helpful review of drafts of this paper. iment following rotenone application to Lake Davis, California, 2007. Cal- ifornia Department of Fish and Game, Pesticide Investigations Unit, Office ∗ DON C. ERMAN 1 of Spill Prevention and Response Administrative Report 08-01, Rancho Cor- dova. Department of Wildlife, Fish, and Conservation Biology Orciari, R. D. 1979. Rotenone resistance of golden shiners from a periodically University of California–Davis reclaimed pond. Transactions of the American Fisheries Society 108:641– One Shields Avenue 645. Parrish, P. R. 1985. Acute toxicity tests. Pages 31–57 in G. Rand and S. Petro- Davis, California 95616, USA celli, editors. Fundamentals of aquatic toxicology: methods and applications. Hemisphere, New York. ∗ E-mail: [email protected] Rand, G., and S. Petrocelli, editors. 1985. Fundamentals of aquatic toxicology: 1Retired. methods and applications. Hemisphere, New York. Sprague, J. B. 1985. Factors that modify toxicity. Pages 124–163 in G. Rand and S. Petrocelli, editors. Fundamentals of aquatic toxicology: methods and REFERENCES applications. Hemisphere, New York. Allan, J. D. 1984. Hypothesis testing in ecological studies of aquatic insects. Sprague, J. B. 1990. Aquatic toxicology. Pages 491–528 in C. B. Schreck and Pages 484–507 in V. H. Resh and D. M. Rosenberg, editors. The ecology of P. B. Moyle, editors. Methods for fish biology. American Fisheries Society, aquatic insects. Praeger, New York. Bethesda, Maryland. Buchwalter, D. B., J. J. Jenkins, and L. R. Curtis. 2002. Respiratory strategy is Trumbo, J. S., S. Siepmann, and B. Finlayson. 2000a. Impacts of rotenone use a major determinant of [3H] water and [14C] chlorpyrifos uptake in aquatic on benthic macroinvertebrate populations in Silver King Creek, 1990 through insects. Canadian Journal of Fisheries and Aquatic Sciences 59:1315–1322. 1996. California Department of Fish and Game, Office of Spill Prevention Chittenden, M. E., Jr. 2002. Given a significance test, how large a sample size and Response, Administrative Report 00-5, Sacramento. is large enough? Fisheries 27(8):25–29. Trumbo, J. S., S. Siepmann, and B. Finlayson. 2000b. Impacts of rotenone Downes, B. J. 2009. Back to the future: little-used tools and principles of scien- use on benthic macroinvertebrate populations in Silver Creek, 1994 through tific inference can help disentangle effects of multiple stressors on freshwater 1998. California Department of Fish and Game, Office of Spill Prevention ecosystems. Freshwater Biology 55(Supplement 1):60–79. and Response, Administrative Report 00-7, Sacramento. Elliott, J. M. 1977. Some methods for the statistical analysis of samples of ben- USBR (U.S. Bureau of Reclamation). 2007. Draft environmental assessment thic invertebrates, 2nd edition. Freshwater Biological Association Scientific (EA): native fish restoration in Bonita Creek, Gila Box Riparian National Publication 25. Conservation Area, Graham County, Arizona. Available: www.usbr.gov/ Eriksen, C. H., G. A. Lamberti, and V.H. Resh. 1996. Aquatic insect respiration. lc/phoenix/reports/bonitacreek/bcdea.html. (October 2011).

Downloaded by [Department Of Fisheries] at 01:41 01 March 2012 Pages 29–40 in R. W. Merritt and K. W. Cummins, editors. An introduction USEPA (U.S. Environmental Protection Agency). 2008. Risks of rotenone use to the aquatic insects of North America, 3rd edition. Kendall/Hunt, Dubuque, to federally threatened California red-legged frog (Rana aurora draytonii). Iowa. USEPA, Office of Pesticide Programs, Environmental Fate and Effects Divi- EU (European Union). 2008. Commission decision of 10 April 2008 sion, Washington, D.C. concerning the noninclusion of rotenone, extract from equisetum, and USFWS (U.S. Fish and Wildlife Service). 2010. Paiute cutthroat trout chinin-hydrochlorid in annex I to Council Directive 91/414/EEC and restoration project: Silver King Creek, Humboldt-Toiyabe National For- the withdrawal of authorisations for plant protection products contain- est, Alpine County, California. Available: www.fws.gov/Nevada/protected ing these substances. Official Journal of the European Union 51(L108): species/fish/species/pct.html.(February 2011). 30–32. Available: http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri = Vinson, M. R., E. C. Dinger, and D. K. Vinson. 2010. Piscicides and OJ:L:2008:108:0030:0032:EN:PDF. (March 2011). invertebrates: after 70 years, does anyone really know? Fisheries 35: Finlayson, B. J., R. A. Schnick, R. L. Cailteux, L. DeMong, W. D. Horton, W. 61–71. McClay, C. W. Thompson, and G. J. Tichacek. 2000. Rotenone use in fish- Waters, W. E., and D. C. Erman. 1990. Research methods: concept and design. eries management: administrative and technical guidelines manual. American Pages 1–34 in C. B. Schreck and P. B. Moyle, editors. Methods for fish Fisheries Society, Bethesda, Maryland. biology. American Fisheries Society, Bethesda, Maryland. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:42 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 A Method to Train Groups of Predator-Naive Fish to Recognize and Respond to Predators When Released into the Natural Environment Justin A. Olson a , Jenae M. Olson a , Rachel E. Walsh a & Brian D. Wisenden a a Biosciences Department, Minnesota State University Moorhead, 1104 7th Avenue South, Moorhead, Minnesota, 56563, USA Available online: 29 Feb 2012

To cite this article: Justin A. Olson, Jenae M. Olson, Rachel E. Walsh & Brian D. Wisenden (2012): A Method to Train Groups of Predator-Naive Fish to Recognize and Respond to Predators When Released into the Natural Environment, North American Journal of Fisheries Management, 32:1, 77-81 To link to this article: http://dx.doi.org/10.1080/02755947.2012.661390

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 32:77–81, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.661390

MANAGEMENT BRIEF

A Method to Train Groups of Predator-Naive Fish to Recognize and Respond to Predators When Released into the Natural Environment

Justin A. Olson, Jenae M. Olson, Rachel E. Walsh, and Brian D. Wisenden* Biosciences Department, Minnesota State University Moorhead, 1104 7th Avenue South, Moorhead, Minnesota 56563, USA

cues released as signature body odors of predators and chemi- Abstract cal alarm cues passively released by injured prey (Brown 2003; Hatcheries are effective at producing large numbers of fish Ferrari et al. 2010). for augmenting fisheries or conserving endangered populations, A mechanism for acquired predator recognition that has been but the fish they produce are often predator-naive, resulting in high levels of predation mortality when the fish are first released demonstrated many times in the literature is the coupling of into natural water bodies. Fish normally acquire recognition of chemical alarm cues from injured conspecifics with the sight novel stimuli as indicators of danger when injury-released chemi- or smell of a predator (Ferrari et al. 2010). In the parlance of cal cues from conspecifics (a known indicator of an actively foraging comparative psychologists, chemical alarm cue is an uncondi- predator) are presented simultaneously with a novel stimulus (e.g., tioned stimulus that elicits an innate response (the unconditioned predator odor, image). Thus, fish in wild populations quickly learn the sight and smell of their predators. Past research has demon- response) without requiring prior experience (Suboski 1990). strated that predator-naive, hatchery-reared fish can be trained Once chemical alarm cues of conspecifics have been detected, to recognize predators and that fish trained by this method have the receiver is primed for releaser-induced recognition learning a significantly greater probability of surviving an encounter with (Suboski 1990). In this type of learning, any novel stimulus (the a predator. To implement predator training in fishery manage- conditioned stimulus) presented simultaneously with alarm cues ment, predator recognition training must be feasible on a large scale in a way that does not place an undue financial or logis- becomes associated with predation risk. A remarkable aspect of tical burden on fisheries managers. Here, we demonstrate that this type of learning is that a single pairing is sufficient to confer groups of fish can be quickly and easily conditioned to recognize near-permanent association of risk with the novel stimulus. In the odor of a novel predator and react to it with antipredator be- laboratory aquaria, releaser-induced recognition learning can be havior. This simple method could improve the cost effectiveness used to manipulate fish to associate predation risk with the sight of any stocking program, result in greater densities of managed stocks, and enhance the direct and indirect economic benefits of or odor of predators (Chivers and Smith 1994), nonpredatory

Downloaded by [Department Of Fisheries] at 01:42 01 March 2012 a fishery. stimuli such as a goldfish Carassius auratus (Chivers and Smith 1994), or even a flashing red light (Yunker et al. 1999) or elec- tronic tones (Wisenden et al. 2008). Behavioral response to indi- Effective antipredator behavior is a major determinant of cators of predation risk commonly involves reduction in activity cohort survival. The first step in any antipredator response is because predators detect prey by detecting motion (Lawrence recognition of predation risk (Lima and Dill 1990; Smith 1992). and Smith 1989; Lima and Dill 1990). Reduced activity is often Antipredator response to chemical cues released by injured con- associated with reduction in foraging activity, movement to the specifics is a widespread phenomenon among aquatic animals, bottom, an increase in shelter use and increase in shoal cohesion ranging from ciliates to amphibians (see Wisenden 2003 for a (Chivers and Smith 1998; Ferrari et al. 2010). review) because these cues are reliably released in the context of Suboski and Templeton (1989) and Olla and Davis (1989) predation. Recognition of predators and correlates of predation were among the first to anticipate that this form of learning could risk by fish is often mediated directly or indirectly by chemical be used as a potential management tool to prepare predator-naive

*Corresponding author: [email protected] Received April 21, 2011; accepted October 11, 2011 77 78 OLSON ET AL.

animals reared in captivity for release into the wild. Fisheries predator recognition learning occurs broadly across all fish taxa; management practices have long used hatchery populations to therefore, results gained from fathead minnows apply broadly augment fisheries, but a major limitation of stocking programs to economically important game species. is high mortality of stocked fish, particularly in the first days and weeks of introduction into natural water bodies (for recent examples see Berejikian et al. 1999; Karam et al. 2008; Kekal¨ ainen¨ METHODS et al. 2008; Sudo et al. 2008; Tomiyama et al. 2009; Christensen Laboratory-reared fathead minnows were obtained at 30 d and Moore 2010; Ochiwada-Doyle et al. 2010; Thorstad of age from the Environmental Protection Agency laboratories et al. 2011). in Duluth, Minnesota, and maintained in the aquatic research Several studies have applied releaser-induced recognition facility at Minnesota State University Moorhead (MSUM) on a learning to hatchery populations of economically significant diet of commercial flake food and a photoperiod of 12 h light : fish species and have shown that game species do indeed ac- 12 h dark. quire predator recognition this way (Brown and Smith 1998; Preparation of test cues.—Chemical alarm cues were pre- Berejikian et al. 1999, 2003; Mirza and Chivers 2000; Wisenden pared from fathead minnows obtained from a local bait dealer. et al. 2004; Hawkins et al. 2008), and that trained fish have a Four fathead minnows were killed by cervical dislocation significantly higher probability of surviving encounters with (MSUM Institutional Animal Care and Use Committee proto- predators than untrained fish (Olla and Davis 1989; Berejikian col 10-R/T-BIOL-010-N-Y-C) and measured (mean ± SE total et al. 1999; Mirza and Chivers 2000). length [TL] = 74.25 ± 3.03 mm). The skin was carefully fil- Hatchery-reared Chinook salmon Oncorhynchus leted from each side of the body, laid flat on a piece of wet glass tshawytscha trained to recognize predacious cutthroat to measure skin area, then transferred to a beaker of deionized trout O. clarkii reduced activity and foraging behavior in water resting on a bed of crushed ice. A total of 29.02 cm2 of skin response to cutthroat trout odor and subsequently showed a was harvested, homogenized with a hand blender for 30 s, di- 9% increase in survival when released into the wild (Berejikian luted to a final volume of 100 mL with deionized water, aliquoted et al. 1999). Hatchery-reared brook trout Salvelinus fontinalis into two 50-mL tubes, and frozen at –20◦C until needed. fingerlings trained to recognize the odor of chain pickerel Predator odor was tank water from a 185-L aquarium con- Esox niger showed a decrease in activity, decrease in foraging taining two West African saddled bichirs Polypterus endlicherii activity, and increase in shelter use when presented with (TLs = 23.4 and 24.9 cm) maintained on a diet of commercial pickerel odor (Mirza and Chivers 2000). These behavioral pellets. Selection of the predator species was made arbitrarily changes in the trout translated into a 20% increase in survival because predator recognition training is open to any stimulus. in direct encounters with pickerel in laboratory tanks and a An exotic species such as a bichir is certain to be novel to the 5% increase in survival in staged encounters in large outdoor test minnows used in this experiment. enclosures (Mirza and Chivers 2000). Conditioning protocol.—The experimental group was To date, predator recognition training has not been adopted housed in a 265-L tub filled with about 200 L of dechlorinated by fisheries managers because, in part, the logistics of imple- tap water filtered with a large sponge filter. The tub contained 30 mentation were not the focus of earlier research. Past studies fathead minnows that were all 110-d-old (TL = 33.9 ± 0.8 mm trained fish one to three at a time, which is not practical on a [mean ± SE]). Fish in the experimental tub were conditioned commercial scale. For predator recognition training to be feasi- with 100 mL of thawed minnow skin extract and 18.95 L of ble to implement, a method for training groups of hatchery fish thawed predator cue and left for 2 h. The control group of 36 needs to be developed. fathead minnows (TL = 32.4 ± 0.7 mm [mean ± SE]) were in Downloaded by [Department Of Fisheries] at 01:42 01 March 2012 In this study, we tested the efficacy of a predator recogni- an identical tub, and all were 103-d-old. The control tub received tion training paradigm specifically adapted to existing practices nothing. The size of fish did not differ between the groups (t18 = of fisheries workers. We aimed to develop a training method 1.35, P = 0.194). that does not add cost (equipment, personnel, time) and yet Experimental set up.—After the 2-h conditioning period, 15 confers the benefits of predator recognition by stocked fish. fish from each tub were arbitrarily selected and assigned to 30 We emulated the process of transporting fish in a holding tank separate test aquaria containing fresh dechlorinated tap water. (200 L) from the point of rearing (hatchery or rearing pond) to One minnow was placed in each tank. Treatment assignments the destination natural water body. Training took place in the alternated from one tank to the next to control for any effect holding tank over 2 h, estimated to be the typical transportation of location in the testing room. Test tanks were all-glass, 38-L time. The fish were then “released” into test aquaria where they aquaria with a sponge filter and a thin layer of naturally colored were tested for recognition and response to predator odor. We gravel. A length of airline tubing was wedged into the lift tube used fathead minnows Pimephales promelas as our test organ- of the sponge filter to serve as a stimulus injection tube through ism because they are well suited for laboratory study and have which test stimuli could be introduced surreptitiously. Grid been the study organism for many previous studies on predator lines (5-cm × 5-cm cells) were drawn on the small pane of recognition learning (Ferrari et al. 2010). The phenomenon of each test tank that was used for scoring fish behavior. Pieces of MANAGEMENT BRIEF 79

FIGURE 1. Activity before and after presentation of predator odor for minnows that had received no conditioning (indicated by open circles and dashed line) and for minnows that had been conditioned with predator odor and skin extract (indicated by solid triangles and solid line).

paper were placed between tanks to preclude visually mediated of squares, alpha set at 0.05) using poststimulus behavior as the social transmission of behavioral responses from one tank to response variable, conditioning treatment (present or absent) as the next during testing (Mathis et al. 1996). a factorial predictor, and prestimulus behavior as a continuous Experimental protocol.—After a 24-h acclimation period, predictor (covariate). fish were tested for a behavioral response to predator odor. A 60-mL syringe was used to withdraw 60 mL of tank water RESULTS through the stimulus injection tube. This water was discarded The activity of fish before presentation of the predator because it served only to rinse the tube of any residue. A sec- odor was a good predictor of activity after presentation of ond 60 mL was withdrawn into the syringe and retained for predator odor for control fish that were not conditioned with later use to flush test stimuli from the injection tube. A 5-min predator odor and skin conspecific skin extract (Figure 1). prestimulus observation period recorded activity and vertical However, poststimulus activity of conditioned fish was not distribution. Activity was scored as the total number of grid related to prestimulus activity levels because conditioned fish lines crossed over the 5-min observation period. Vertical distri- significantly reduced activity in the poststimulus observation bution was recorded as the grid row occupied by the fish at 10-s period (Figure 1). The treatment differences resulted in a intervals over the 5-min observation period. The row nearest the significant interaction term between prestimulus activity and surface was row 1, the row at the tank bottom was row 5. The

Downloaded by [Department Of Fisheries] at 01:42 01 March 2012 the conditioning treatment (ANCOVA: conditioning F scores from 30 observations over the 5-min period were aver- 1, 26 = 0.786, P = 0.384; preactivity F = 8.571, P = 0.007; aged into one mean score of vertical distribution per trial. After 1, 26 conditioning × preactivity F = 8.454, P = 0.007). the prestimulus observation period, 120 mL of predator odor 1, 26 In response to predator odor, most conditioned fish reduced followed by the 60 mL of previously retained tank water was vertical distribution but a few sought refuge at the surface, reintroduced through the injection line into the turbulent upflow sulting in no overall net effect of the conditioning treatment of air and water of the sponge filter. The air and water currents (ANCOVA: conditioning F = 0.447, P = 0.510; preactivity served to quickly disperse the predator odor throughout the test 1, 26 F = 10.850, P = 0.003; conditioning × preactivity F = tank. A second 5-min observation period (the poststimulus pe- 1, 26 1, 26 0.457, P = 0.505; Figure 2). riod) began immediately after cue injection. Activity and vertical distribution were recorded as before. Common antipredator responses in fishes are a reduction in activity and movement DISCUSSION toward the bottom (Lawrence and Smith 1989; Ferrari et al. We demonstrated that a group of predator-naive fish can be 2010). trained to recognize and fear a novel predator with the addi- Data were analyzed using a mixed-model analysis of covari- tion of chemical cues of injured conspecifics and odor cues of ance (ANCOVA;using PASW release 18.0 [2009], type III sums a novel predator. The training method used is cost effective and 80 OLSON ET AL.

FIGURE 2. Mean vertical distribution before and after the addition of predator odor for minnows that had received no conditioning (indicated by open circles and dashed line) and for minnows that had been conditioned with predator odor and skin extract (indicated by solid triangles and solid line). The row nearest the surface was designated as “1” and the row near the tank bottom was designated as “5.”

easily implemented with existing management practices. Chem- (1999) noted that conditioned Chinook salmon fingerlings re- ical cues can be used in place of the direct experience with duced activity in response to the odor of cutthroat trout but did predators that fish in natural water bodies use to acquire preda- not exhibit any other behavioral responses (foraging and vertical tor recognition (Olla and Davis 1989; Chivers and Smith 1995). distribution remained unchanged). Nevertheless, when released Naive fish from a hatchery or rearing pond loaded into a holding conditioned cutthroat trout reared in complex hatchery habitat tank on a delivery vehicle can be trained by adding a small vol- show a 20% increase in survival over unconditioned trout from ume of conspecific skin extract and predator odor to the holding similar hatchery habitat (Berejikian et al. 1999). tank. By the time the truck reaches the destination lake, the fish In this study, we conditioned fish to recognize and respond will be trained to recognize the novel odor as an indicator of to the odor of one novel species. We used laboratory-reared danger. Several studies show that behavioral responses such as minnows as test subjects and the odor of an exotic predator reduced activity allow trained fish to be more likely than un- as the test odorant, thus ensuring that the test subjects had no trained fish to survive encounters with predators (Mathis and previous experience or evolved response to the odor of this Smith 1993; Berejikian et al. 1999; Mirza and Chivers 2000). predator. In cases where prey species have preexisting innate Now that we have demonstrated the effectiveness of group train- recognition of predator odor, this technique of predator recogni- ing, the next step is to begin large-scale experimental implemen- tion imparts strengthened response to predator odor (Berejikian

Downloaded by [Department Of Fisheries] at 01:42 01 March 2012 tation of this management tool. et al. 2003). In complex ecosystems with multiple predators, a The aim of this experiment was to test the efficacy of group complex bouquet of predator odors may be necessary to confer training using a convenient model species, the fathead min- adequate life skills training to naive fish (Darwish et al. 2005). now. The phenomenon of predator recognition is not specific In some instances, one predator odor may be generalizable to a to minnows, to fish generally (Wisenden 2003; Ferrari et al. multiple-predator species if predators are phylogenetically re- 2010), or even to aquatic ecosystems (e.g., Griffin et al. 2000). lated (Ferrari et al. 2007) or specific predator training timed This training method is the final piece of the process required to with predators active at a particular time or season of release clear the way for large-scale implementation of this “head-start” (Pearsons 2010). management strategy on fish of commercial and conservation importance (Crane and Mathis 2011). Fish in this study responded to predator odor with reduced activity but not with any detectable change in vertical distri- ACKNOWLEDGMENTS bution. This is not an unusual occurrence in that activity is a This work was funded by research grants to B.D.W. from the more-sensitive metric of antipredator response than others such MSUM College of Social and Natural Sciences and the U.S. as vertical distribution, shelter seeking, etc. Berejikian et al. National Science Foundation. MANAGEMENT BRIEF 81

REFERENCES Lima, S. L., and L. M. Dill. 1990. Behavioral decisions made under the risk Berejikian, B. A., R. J. F. Smith, E. P. Tezak, S. L. Schroder, and C. M. Knud- of predation: a review and prospectus. Canadian Journal of Zoology 68: sen. 1999. Chemical alarm signals and complex hatchery rearing habitats 619–640. affect antipredator behavior and survival of Chinook salmon (Oncorhynchus Mathis, A., D. P. Chivers, and R. J. F. Smith. 1996. Cultural transmission of tshawytscha) juveniles. Canadian Journal of Fisheries and Aquatic Sciences predator recognition in fishes: intraspecific and interspecific learning. Animal 56:830–838. Behaviour 51:185–201. Berejikian, B. A., E. P. Tezak, and A. L. LaRae. 2003. Innate and enhanced Mathis, A., and R. J. F. Smith. 1993. Chemical alarm signals increase the predator recognition in hatchery-reared Chinook salmon. Environmental Bi- survival time of fathead minnows (Pimephales promelas) during encounters ology of Fishes 67:241–251. with northern pike (Esox lucius). Behavioral Ecology 4:260–265. Brown, G. E. 2003. Learning about danger: chemical alarm cues and local risk Mirza, R. S., and D. P. Chivers. 2000. Predator-recognition training enhances assessment in prey fishes. Fish and Fisheries 4:227–234. survival of brook trout: evidence from laboratory and field-enclosure studies. Brown, G. E., and R. J. F. Smith. 1998. Acquired predator recognition in juvenile Canadian Journal of Zoology 78:2198–2208. rainbow trout (Oncorhynchus mykiss): conditioning hatchery-reared fish to Ochiwada-Doyle, F., C. A. Gray, N. R. Loneragan, and M. D. Taylor. 2010. recognize chemical cues of a predator. Canadian Journal of Fisheries and Using experimental ecology to understand stock enhancement: comparisons Aquatic Sciences 55:611–617. of habitat-related predation on wild and hatchery-reared Penaeus plebejus Chivers, D. P., and R. J. F. Smith. 1994. Fathead minnows, Pimephales promelas, Hess. Journal of Experimental Marine Biology and Ecology 390:65–72. acquire predator recognition when alarm substance is associated with the sight Olla, B. L., and M. W. Davis. 1989. The role of learning and stress in predator of unfamiliar fish. Animal Behaviour 48:597–605. avoidance of hatchery-reared coho salmon (Oncorhynchus kisutch) juveniles. Chivers, D. P., and R. J. F. Smith. 1995. Free-living fathead minnows Aquaculture 76:209–214. rapidly learn to recognize pike as predators. Journal of Fish Biology 46: Pearsons, T. N. 2010. Operating hatcheries within an ecosystem context using 949–954. the adaptive stocking concept. Fisheries 35:23–31. Chivers, D. P., and R. J. F. Smith. 1998. Chemical alarm signalling in Smith, R. J. F. 1992. Alarm signals in fishes. Reviews in Fish Biology and aquatic predator-prey systems: a review and prospectus. Ecoscience´ 5: Fisheries 2:33–63. 338–352. Suboski, M. D. 1990. Releaser-induced recognition learning. Psychological Christensen, D. R., and B. C. Moore. 2010. Largemouth bass consumption de- Review 97:271–284. mand on hatchery rainbow trout in two Washington lakes. Lake and Reservoir Suboski, M. D., and J. J. Templeton. 1989. Life skills training for hatchery fish: Management 26:200–211. social learning and survival. Fisheries Research 7:343–352. Crane, A. L., and A. Mathis. 2011. Predator-recognition training: a conserva- Sudo, H., N. Kajihara, and T. Fujii. 2008. Predation by the swimming crab tion strategy to increase postrelease survival of hellbenders in head-starting Charybdis japonica and piscivorous fishes: a major mortality factor in programs. Zoo Biology 30:611–622. hatchery-reared juvenile Japanese flounder Paralichthys olivaceus released Darwish, T. L., R. S. Mirza, A. O. H. C. Leduc, and G. E. Brown. 2005. in Mano Bay, Sado Island, Japan. Fisheries Research 89:49–56. Acquired recognition of novel predator odour cocktails by juvenile glowlight Thorstad, E. B., I. Uglem, P.Arechavala-Lopez, F. Økland, and B. Finstad. 2011. tetras. Animal Behaviour 70:83–89. Low survival of hatchery-released Atlantic salmon smolts during initial river Ferrari, M. C. O., A. Gonzalo, F. Messier, and D. P. Chivers. 2007. Generaliza- and fjord migration. Boreal Environment Research 16:115–120. tion of learned predator recognition: an experimental test and framework for Tomiyama, T., K. Ebe, G. Kawata, and T. Fujii. 2009. Post-release predation future studies. Proceedings of the Royal Society of London B 274:1853–1859. on hatchery-reared Japanese flounder Paralichthys olivaceus in the coast of Ferrari, M. C. O., B. D. Wisenden, and D. P. Chivers. 2010. Chemical ecology Fukushima, Japan. Journal of Fish Biology 75:2629–2641. of predator-prey interactions in aquatic ecosystems: a review and prospectus. Wisenden, B. D. 2003. Chemically-mediated strategies to counter predation. Canadian Journal of Zoology 88:698–724. Pages 236–251 in S. P. Collin and N. J. Marshall, editors. Sensory processing Griffin, A. S., D. T. Blumstein, and C. S. Evans. 2000. Training captive- in the aquatic environment. Springer-Verlag, New York. bred or translocated animals to avoid predators. Conservation Biology 14: Wisenden, B. D., J. Klitzke, R. Nelson, D. Friedl, and P. Jacobson. 2004. 1317–1326. Predator-recognition training of hatchery-reared walleye and a field test of Hawkins, L. A., A. E. Magurran, and J. D. Armstrong. 2008. Ontogenetic a training method using yellow perch. Canadian Journal of Fisheries and learning of predator recognition in hatchery-reared Atlantic salmon, Salmo Aquatic Sciences 62:2144–2150. salar. Animal Behaviour 75:1663–1671. Wisenden, B. D., J. Pogatshnik, D. Gibson, L. Bonacci, A. Schumacher, and Karam, A. P., B. R. Kesner, and P. C. Marsh. 2008. Acoustic telemetry to assess A. Willett. 2008. Sound the alarm: learned association of predation risk

Downloaded by [Department Of Fisheries] at 01:42 01 March 2012 post-stocking dispersal and mortality of razorback sucker Xyrauchen texanus. with novel auditory stimuli by fathead minnows (Pimephales promelas)and Journal of Fish Biology 73:719–727. glowlight tetras (Hemigrammus erythrozonus) after single simultaneous pair- Kekal¨ ainen,¨ J., T. Niva, and H. Huuskonen. 2008. Pike predation on hatchery- ings with conspeciﬁc chemical alarm cues. Environmental Biology of Fishes reared Atlantic salmon smolts in a northern Baltic river. Ecology of Fresh- 81:141–147. water Fish 17:100–109. Yunker, W. K., D. E. Wein, and B. D. Wisenden. 1999. Conditioned alarm be- Lawrence, B. J., and R. J. F. Smith. 1989. Behavioural response of solitary fat- havior in fathead minnows (Pimephales promelas) resulting from association head minnows, Pimephales promelas, to alarm substance. Journal of Chemi- of chemical alarm pheromone with a nonbiological visual stimulus. Journal cal Ecology 15:209–219. of Chemical Ecology 25:2677–2686. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:43 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 First Report of Abundant Rudd Populations in North America Kevin L. Kapuscinski a , John M. Farrell a & Michael A. Wilkinson b a College of Environmental Science and Forestry, State University of New York, 1 Forestry Drive, Syracuse, New York, 13210, USA b New York State Department of Environmental Conservation, 270 Michigan Avenue, Buffalo, New York, 14203, USA

Available online: 29 Feb 2012

To cite this article: Kevin L. Kapuscinski, John M. Farrell & Michael A. Wilkinson (2012): First Report of Abundant Rudd Populations in North America, North American Journal of Fisheries Management, 32:1, 82-86 To link to this article: http://dx.doi.org/10.1080/02755947.2012.661391

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 32:82–86, 2012 C American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.661391

MANAGEMENT BRIEF

First Report of Abundant Rudd Populations in North America

Kevin L. Kapuscinski* and John M. Farrell College of Environmental Science and Forestry, State University of New York, 1 Forestry Drive, Syracuse, New York 13210, USA Michael A. Wilkinson New York State Department of Environmental Conservation, 270 Michigan Avenue, Buffalo, New York 14203, USA

tablished on five continents (Nico et al. 2010; Froese and Pauly Abstract 2011). The Rudd Scardinius erythrophthalmus was first introduced into The rudd was intentionally introduced into the USA during U.S. waters about a century ago, and the species’ popularity as a the late 19th or early 20th century (Burkhead and Williams baitfish in the 1980s has facilitated its spread to at least 21 states and the province of Ontario. Several established populations have 1991; Mills et al. 1993), but for several decades its distribution been identified, but low abundances have led to little research and remained limited to the northeastern USA and Oconomowoc management attention. Rudds comprised 48.7% of the 14,130 fish Lake, Wisconsin, where it was intentionally introduced (Cahn captured in spring trap-netting surveys of Buffalo Harbor (north- 1927). The popularity of the rudd as a baitfish in the 1980s facili- eastern Lake Erie) and the upper Niagara River during 2007–2008. tated its spread to the waters of at least 21 states, including the St. Rudd was the most abundant species sampled, being captured at 11 of 12 locations and comprising 23.6% of the total catch in Buffalo Lawrence River, New York, a boundary water between the USA Harbor and 70.3% of the catch in the upper Niagara River. Docu- and Canada (Klindt 1990, 1991; Crossman et al. 1992; Nico et al. mented presences and absences in historical reports indicate that 2010). Although populations of rudds became established in rudd became established in these waters between 1986 and 1991. several North American waters during the past century, their low Research is needed to understand the effects of rudds on native abundances resulted in little attention from fisheries biologists. aquatic resources, especially nearshore macrophyte assemblages and the fish they support. For example, the rudd is still considered rare in Oconomowoc Lake (Susan Byler, Wisconsin Department of Natural Re- sources, personal communication), despite its having been in- The distributions of numerous aquatic species have expanded troduced in 1916 (Cahn 1927). Similarly, rudds are present but

Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 well beyond their native ranges due to human actions such as considered rare in the Ohio waters of Lake Erie (Travis Hartman, intentional releases, the removal or bypassing of natural bar- Ohio Department of Natural Resources, personal communica- riers through the construction of shipping channels, and un- tion), the New York waters of Lake Erie outside of Buffalo intentional introductions via bait bucket transfers and ballast Harbor (Donald Einhouse, New York State Department of En- water exchanges (Mills et al. 1993). Introductions of nonna- vironmental Conservation, personal communication), and the tive ﬁshes such as the sea lamprey Petromyzon marinus and St. Lawrence River (J. M. Farrell, unpublished data). round goby Neogobius melanostomus have had serious and well- The rudd is omnivorous, and larger individuals (>150 mm) documented negative effects on native communities (Hansen often consume aquatic macrophytes (Lake et al. 2002; Hicks 1999; Janssen and Jude 2001). The rudd Scardinius erythroph- 2003), exploiting food resources that are not typically utilized thalmus is a European minnow that now has a nearly global by native ﬁshes in the north temperate regions of North Amer- longitudinal distribution due to human translocations; the rudd ica. Rudds typically grow to approximately 350 mm in length ranges from New Zealand to the Midwestern USA and is es- and 1,800 g in weight in their native range (Crossman et al.

*Corresponding author: [email protected] Received June 28, 2011; accepted November 1, 2011

82 MANAGEMENT BRIEF 83

FIGURE 1. Map of Buffalo Harbor (Lake Erie) and the upper Niagara River indicating trap-net locations. Locations sampled in 2007 are denoted by triangles, those sampled in 2008 by squares, and those sampled in both years by circles. The numbers of rudds caught and the total numbers of fish caught are also provided. 1992). The largest rudd on record was 617 mm long and weighed abundant rudd populations may nullify or reduce the benefits 3,623 g (Spremˇ et al. 2010). Such large, macrophyte-consuming of these macrophyte restoration efforts. fish are not common in North American freshwater ecosystems, so an abundant rudd population could negatively affect aquatic communities by altering macrophyte assemblages and METHODS accelerating internal nutrient loading (eutrophication) by re- A spring trap-net survey was conducted in Buffalo Har- mobilizing the nutrients stored in sediments and macrophytes bor and the upper Niagara River during 2007–2008 to cap- (Hansson et al. 1987; van Donk and Gulati 1995; Vanni 2002; ture spawning muskellunge Esox masquinongy for research and Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 Hicks 2003; Nurminen et al. 2003). Additional threats to aquatic quantifying fish assemblages at nearshore sites. Oneida (1.8–2.4 communities include (1) juvenile rudds competing with native m high with 6.1-m wings, 15.2–121.9-m leaders, and 2.54-cm fishes for benthic invertebrate prey (Crossman et al. 1992; Hicks bar measure mesh) and hoop style trap-nets (0.9–1.2 m high 2003), (2) maintaining parasites and pathogens and spreading with 6.1-m wings, 15.2–30.5-m leads, and 2.54-cm bar measure them to native fishes (Popovic´ et al. 2001), and (3) hybridizing mesh) were deployed at 11 sites from 14 May to 8 June 2007 with golden shiner Notemigonus crysoleucas (Burkhead and and at 7 sites from 15 May to 11 June 2008 (some sites were Williams 1991), thereby causing genomic extinction (i.e., the sampled in both years; Figure 1). The mesh size of our nets loss of local evolutionary lineages; Allendorf and Luikart 2007). was probably too large to capture small, young (e.g., age-1 and This report is the first to document the presence of abundant age-2) rudds. The sites sampled were typically 1–2.75 m deep rudd populations in North America, specifically in Buffalo at the trap end and adjacent to deeper water. They tended to be Harbor (northeastern Lake Erie) and the upper Niagara River. associated with fine sediment substrates with filamentous algae The presence of rudds in these waters is especially notable but undeveloped aquatic macrophytes during the netting period; because several habitat restoration projects are under way aquatic macrophytes were common at the trap-net sites later in or planned for these areas (USFERC 2007), some of which the season. Water temperatures ranged from 9.1◦C to 21.3◦C include the restoration of aquatic macrophytes. Herbivory by during our trap-net surveys in 2007 and from 10.6◦C to 21.1◦C 84 KAPUSCINSKI ET AL.

TABLE 1. Composition of the ﬁsh species captured in trap nets in Buffalo Harbor (BH) and the upper Niagara River (UNR) during spring 2007–2008.

2007 2008 BH UNR BH UNR Species N % N % N % N % Rudd Scardinius erythropthalmusa 819 23.2 1,247 64.8 719 24.1 4,102 72.1 Brown bullhead Ameiurus nebulosus 486 13.8 158 8.2 523 17.5 737 13.0 Rock bass Ambloplites rupestris 779 22.1 220 11.4 501 16.8 84 1.5 Smallmouth bass Micropterus dolomieu 669 19.0 21 1.1 599 20.0 28 0.5 Common carp Cyprinus carpioa 58 1.6 42 2.2 42 1.4 271 4.8 Northern pike Esox lucius 57 1.6 13 0.7 55 1.8 239 4.2 Pumpkinseed Lepomis gibbosus 131 3.7 37 1.9 170 5.7 19 0.3 Yellow perch Perca flavescens 173 4.9 30 1.6 55 1.8 22 0.4 Bluegill Lepomis macrochirus 74 2.1 30 1.6 121 4.0 21 0.4 Quillback Carpiodes cyprinus 94 2.7 28 0.9 12 0.2 Redhorse Moxostoma spp. 16 0.5 22 1.1 43 1.4 33 0.6 Largemouth bass Micropterus salmoides 31 0.9 18 0.9 50 1.7 11 0.2 White sucker Catostomus commersonii 60 1.7 10 0.5 21 0.7 14 0.2 Bowfin Amia calva 1 0.0 31 1.6 35 0.6 White perch Morone americanaa 10 0.3 9 0.5 23 0.8 21 0.4 Common shiner Luxilus cornutus 6 0.2 22 1.1 14 0.2 Goldfish Carassius auratusa 19 0.5 1 0.1 16 0.5 5 0.1 Black crappie Pomoxis nigromaculatus 11 0.3 1 0.1 9 0.3 1 0.0 Freshwater drum Aplodinotus grunniens 14 0.4 1 0.1 4 0.1 3 0.1 Longnose gar Lepisosteus osseus 5 0.1 4 0.1 1 0.0 Muskellunge Esox masquinongy 30.1 40.1 Channel catfish Ictalurus punctatus 3 0.1 1 0.0 3 0.1 Gizzard shad Dorosoma cepedianuma 40.120.1 Goldfish × common carpa 2 0.1 1 0.1 3 0.1 Rainbow trout Oncorhynchus mykissa 1 0.0 1 0.1 2 0.1 1 0.0 Round goby Neogobius melanostomusa 40.2 10.0 Walleye Sander vitreus 40.1 White bass Morone chrysops 10.1 20.0 Northern hog sucker Hypentelium nigricans 10.1 10.0 Muskellunge × northern pike 1 0.1 1 0.0 Green sunfish Lepomis cyanellus 10.1 Total 3,530 1,925 2,989 5,686 Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 aNonnative species.

in 2008 and were often below the 18◦C preferred by spawning was the most abundant species sampled, being captured at 11 rudds (Hicks 2003; Tarkan 2006). Each net was emptied daily or of 12 sites (Figure 1) and comprising 23.6% of the total catch every other day, and captured fish were identified to species and in Buffalo Harbor and 70.3% in the upper Niagara River. Other enumerated. In 2009, two hoop-style trap nets were deployed nonnative fishes (7 species) comprised 2.8% of the total catch in for 1–2 d each month during May–August to capture rudds for Buffalo Harbor and 4.7% in the upper Niagara River, whereas a diet analysis. Only data on the total length of rudds is reported native species (23 species) comprised 73.6% of the total catch in here because this effort was spatially and temporally limited and Buffalo Harbor but only 25.0% in the upper Niagara River. Cap- not all fish were not enumerated. tured rudds were not measured in 2007–2008, but we estimated that their total lengths ranged from 200 to 450 mm and noted the presence of several length modes indicative of age structure. RESULTS The rudds captured in trap nets in 2009 from the upper Niagara A total of 14,130 fish were captured in trap nets during 2007– River averaged 349 mm in total length (range, 138–445 mm; 2008, 6,887 (48.7%) of which were rudds (Table 1). The rudd n = 190). The rudds captured during 2007–2008 also appeared MANAGEMENT BRIEF 85

to be well conditioned and sexually mature (we only attempted The rudd’s relatively large body size and ability to obtain to extrude gametes from a subsample of these rudds, but both nutrients from algae, macrophytes, and detritus, which is con- eggs and milt were observed). sidered an adaptation to avoid competition within its native range (Johansson 1987), will probably facilitate its spread into the Great Lakes by buffering both top-down and bottom-up controls (as described for gizzard shad Dorosoma cepedianum DISCUSSION by Stein et al. 1995). Furthermore, the macrophyte-consuming Rudds were first collected in the system in a 1991 biological rudd will establish novel trophic links and possibly contaminant survey of the Buffalo River (a tributary to Buffalo Harbor; Mikol pathways in the ecosystems it invades (as reported for round et al. 1993) but were notably absent from several reports that goby by Kwon et al. 2006). Perhaps most importantly, however, documented the presence of nonnative fishes in Buffalo Harbor the rudd may cause shifts in aquatic vegetation communities and the Niagara River in 1928–1986 (State of New York Con- (van Donk and Gulati 1995; van Donk and Otte 1996; Hicks servation Department 1929; Goodyear et al. 1982; Makarewicz 2003) that many native fishes rely upon for spawning, nursery, et al. 1982; Spotila 1986). Therefore, it appears that the rudd and foraging habitat. The numerous threats that the rudd poses became established in Buffalo Harbor and the upper Niagara to native aquatic communities should be of concern to resource River after 1986 but before 1991. This timeline is consistent with managers, especially because the populations documented here the rudd becoming popular as a baitfish in the 1980s, thereby have direct access to critical habitats that support economi- implicating bait bucket transfers as the most likely source of cally and ecologically important fish communities in the Great introduction. Lakes basin. Three rudds were captured by commercial fish- Buffalo Harbor and the upper Niagara River appear to contain ermen in Long Point Bay, Lake Erie, in 2009 (Gislason et al. favorable environmental conditions for rudd reproduction and 2010), and although the origin of these rudds is unknown, they survival, based on the abundances and multiple length modes may have been immigrants from the Buffalo Harbor or Niagara that we observed. In addition, young-of-the-year rudds were River populations. Comprehensive surveys that employ multi- captured in other surveys of the upper Niagara River, indicat- ple gears (e.g., electrofishing, gill nets, seines, and trap nets) ing that they are reproducing in this water (Kapuscinski 2011). to sample the overall fish community are needed to evaluate Nurminen et al. (2003) captured rudds from Lake Hiidenvesi, the abundance of rudds relative to other fishes in the Buffalo Finland, that were up to age 15 but that on average were smaller Harbor, Niagara River, and other waters where rudds may be (mean total length, 150 mm at age 5 and 204 mm at age 9) than of concern. Research focused on elucidating the effects of rudd the rudds we captured. Quantitative analyses of the reproductive herbivory on nearshore macrophyte assemblages and the fishes dynamics and age and size structures of the rudd populations in they support will be especially important for guiding habitat Buffalo Harbor and the Niagara River will help resource man- restoration projects in the Great Lakes basin. agers understand the invasion histories of these systems and make predictions about future range expansions. The abundant rudd population in the upper Niagara River ACKNOWLEDGMENTS challenges the existing paradigm of optimal rudd habitat and Many New York State Department of Environmental Conser- requires us to expand our concept of the waters vulnerable to vation and State University of New York, College of Environ- invasion. Most studies have focused on lake populations, and mental Science and Forestry personnel provided assistance in the rudd is typically described as a littoral species that prefers the field and logistical and administrative support. The authors lentic habitats (Johansson 1987; Lake et al. 2002) and spawns Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 are especially grateful to Mike Clancy, Charles Curry, Brad Gru- on aquatic macrophytes in shallow, nearshore waters (Orr 1966; ber, Robin Holevinski, Paul McKeown, and Jon Sztukowski. The Kennedy and Fitzmaurice 1974). There are few references to authors also thank the Ontario Ministry of Natural Resources, rudds occupying slow-flowing lotic habitats (Cadwallader 1978; especially Alastair Mathers, for permitting surveys in Ontario Zerunian et al. 1986). Hicks (2003) did report the presence of waters. Gillian AvRuskin assisted with figure creation. Derek rudds in the tailrace of the Karapiro Dam, Waikato River, New Crane, Geofrey Eckerlin, and two anonymous reviewers pro- Zealand. Cadwallader (1978), who surveyed 25 New Zealand vided constructive comments on a draft of this paper. This work waters, captured recently hatched rudds in shallow, nearshore was funded by a grant from the Niagara River Greenway Eco- areas of 17 lentic waters and only one gently flowing reach of logical Fund Standing Committee and a Federal Aid in Sport a reservoir near the mouth of a stream. The discharge of the Fish Restoration Grant administered by the New York State 3 Niagara River averages about 5,550 m /s (Harrison and Hadley Department of Environmental Conservation. 1978), and the velocity at nearshore vegetated areas averages about 0.04 m/s (K. L. Kapuscinski, unpublished data). The rudd’s ability to thrive in a lotic habitat such as the Niagara REFERENCES River requires us to gain a better understanding of the environ- Allendorf, F. W., and G. Luikart. 2007. Conservation and the genetics of popu- mental conditions that affect its population dynamics. lations. Blackwell Scientific Publications, Malden, Massachusetts. 86 KAPUSCINSKI ET AL.

Burkhead, N. M., and J. D. Williams. 1991. An intergeneric hybrid of a native Kwon, T., S. W. Fisher, G. W. Kim, H. Hwang, and J. Kim. 2006. Trophic transfer minnow, the golden shiner, and an exotic minnow, the rudd. Transactions of and biotransformation of polychlorinated biphenyls in zebra mussel, round the American Fisheries Society 120:781–795. goby, and smallmouth bass in Lake Erie, USA. Environmental Toxicology Cadwallader, P. L. 1978. Acclimatisation of rudd Scardinius erythrophthalmus and Chemistry 25:1068–1078. (Pisces: Cyprinidae), in North Island of New Zealand. New Zealand Journal Lake, M. D., B. J. Hicks, R. D. S. Wells, and T. M. Dugdale. 2002. Consumption of Marine and Freshwater Research 12:81–82. of submerged aquatic macrophytes by rudd (Scardinius erythrophthalmus L.) Cahn, A. R. 1927. The European rudd (Scardinius) in Wisconsin. Copeia 162:5. in New Zealand. Hydrobiologia 470:13–22. Crossman, E. J., E. Holm, R. Cholmondeley, and K. Tuininga. 1992. First Makarewicz, J. C., R. C. Dilcher, J. M. Haynes, and K. Shump. 1982. Biological record for Canada of the rudd, Scardinius erythrophthalmus, and notes on the survey, Buffalo River and outer harbor of Buffalo, N.Y. U.S. Army Corps of introduced round goby, Neogobius melanostomus. Canadian Field-Naturalist Engineers, Buffalo, New York. 106:206–209. Mikol, G., K. Roblee, and M. Wilkinson. 1993. Fish and wildlife habitat inven- Froese, R., and D. Pauly, editors. 2011. FishBase: World Wide Web elec- tory and assessment of the lower Buffalo River watershed. New York State tronic publication. Available: www.fishbase.org, version (02/2011). (March Department of Environmental Conservation, Albany. 2011). Mills, E. L., J. H. Leach, J. T. Carlton, and C. L. Secor. 1993. Exotic species in Gislason, D., K. Reid, and K. Oldenburg. 2010. Assessment and mitigation of the Great Lakes: a history of biotic crises and anthropogenic introductions. the effects of commercial fishing activities on aquatic species at risk in Long Journal of Great Lakes Research 19:1–54. Point Bay. Species at Risk Research Fund for Ontario, Report SARF151, Nico, L., P. Fuller, and G. Jacobs. 2010. Scardinius erythrophthalmus. U.S. Toronto. Geological Survey Nonindigenous Aquatic Species Database, Gainesville, Goodyear, C. D., T. A. Edsall, D. M. Ormsby Dempsey, G. D. Moss, and P. E. Florida. Available: nas.er.usgs.gov/queries/FactSheet.aspx?speciesID=648. Polanski. 1982. Atlas of the spawning and nursery areas of Great Lakes fishes, (March 2010). volume 10. Niagara River. U.S. Fish and Wildlife Service FWS/OBS-82/52. Nurminen, L., J. Horppila, J. Lappalanien, and T. Malinen. 2003. Implications Hansen, M. J. 1999. Lake trout in the Great Lakes: basinwide stock collapse of rudd (Scardinius erythrophthalmus) herbivory on submerged macrophytes and binational restoration. Pages 417–453 in C. P. Ferreri and W. W. Taylor, in a shallow eutrophic lake. Hydrobiologia 506–509:511–518. editors. Great Lakes fishery policy and management: a binational perspective. Orr, T. S. C. 1966. Spawning behaviour of rudd, Scardinius erythrophthalmus, Michigan State University Press, East Lansing. infested with plerocercoids of Ligula intestinalis. Nature (London) 212:736. Hansson, L., L. Johansson, and L. Persson. 1987. Effects of fish grazing on nu- Popovic,´ N. T., M. Hacmanjek, and E. Teskeredziˇ c.´ 2001. Health status of rudd trient release and succession of primary producers. Limnology and Oceanog- (Scardinius erythrophthalmus hesperidicus H.) in Lake Vrana on the Island raphy 32:723–729. of Cres, Croatia. Journal of Applied Ichthyology 17:43–45. Harrison, E. J., and W. F. Hadley. 1978. Ecologic separation of sympatric Spotila, J. R. 1986. Assessment of the fishery in the Small Boat Harbor, Buffalo, muskellunge and northern pike. Pages 129–134 in R. L. Kendall, editor. NY, and related areas. Final Report to the Niagara Frontier Transportation Selected coolwater fishes of North America. American Fisheries Society, Authority. State University of New York College at Buffalo, Buffalo. Special Publication 11, Bethesda, Maryland. Sprem,ˇ N., D. Matulic,´ T. Treer, and I. Aniciˇ c.´ 2010. A new maximum length Hicks, B. J. 2003. Biology and potential impacts of rudd (Scardinius ery- and weight for Scardinius erythrophthalmus. Journal of Applied Ichthyology throphthalmus L.) in New Zealand. Pages 49–58 in R. Munro, editor. Man- 26:618–619. aging invasive freshwater fish in New Zealand. Proceedings of a Depart- State of New York Conservation Department. 1929. A biological survey of the ment of Conservation Workshop, New Zealand Department of Conservation, Erie–Niagara system: supplemental to the eighteenth annual report, 1928. J. Wellington. B. Lyon, Albany, New York. Janssen, J., and D. J. Jude. 2001. Recruitment failure of mottled sculpin Cottus Stein, R. A., D. R. DeVries, and J. M. Dettmers. 1995. Food web regulation bairdi in Calumet Harbor, southern Lake Michigan, induced by the newly by a planktivore: exploring the generality of the trophic cascade hypothesis. introduced round goby Neogobius melanostomus. Journal of Great Lakes Canadian Journal of Fisheries and Aquatic Sciences 52:2518–2526. Research 27:319–328. Tarkan, A. S. 2006. Reproductive ecology of two cyprinid fishes in an olig- Johansson, L. 1987. Experimental evidence for interactive habitat segregation otrophic lake near the southern limits of their distribution range. Ecology of between roach (Rutilus rutilus) and rudd (Scardinius erythrophthalmus)ina Freshwater Fish 15:131–138. shallow eutrophic lake. Oecologia 73:21–27. USFERC (U.S. Federal Energy Regulatory Commission). 2007. Order on offer Kapuscinski, K. L. 2011. Comparative ecology of muskellunge and nearshore of settlement and issuing new license. New York Power Authority’s Niagara fish assemblages in the Great Lakes. Doctoral dissertation. State Uni- Project 2216-066. USFERC, Washington, D.C.

Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 versity of New York, College of Environmental Science and Forestry, van Donk, E., and R. D. Gulati. 1995. Transition of a lake to turbid state six Syracuse. years after biomanipulation: mechanisms and pathways. Water Science and Kennedy, M., and P. Fitzmaurice. 1974. Biology of the rudd Scardinius ery- Technology 32:197–206. throphthalmus (L.) in Irish waters. Proceedings of the Royal Irish Academy van Donk, E., and A. Otte. 1996. Effects of grazing by ﬁsh and waterfowl on the 74B:245–303. biomass and species composition of submerged macrophytes. Hydrobiologia Klindt, R. 1990. Distribution of rudd in the St. Lawrence River. St. Lawrence 340:285–290. River Subcommittee, Report to the Great Lakes Fishery Commission, New Vanni, M. J. 2002. Nutrient cycling by animals in freshwater ecosystems. Annual York State Department of Environmental Conservation, Albany. Review of Ecology and Systematics 33:341–370. Klindt, R. 1991. Rudd in the St. Lawrence River: 1989–1990. St. Lawrence Zerunian, S., L. Valentini, and G. Gibertini. 1986. Growth and reproduction River Subcommittee, Report to the Great Lakes Fishery Commission, New of rudd and red-eye roach (Pisces, Cyprinidae) in Lake Bracciano. Italian York State Department of Environmental Conservation, Albany. Journal of Zoology 53:91–95. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:43 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 Contrasting Maturation and Growth of Northern Rock Sole in the Eastern Bering Sea and Gulf of Alaska for the Purpose of Stock Management James W. Stark a a National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Alaska Fisheries Science Center, Resource Assessment and Conservation Engineering Division, 7600 Sand Point Way Northeast, Seattle, Washington, 98115, USA Available online: 29 Feb 2012

To cite this article: James W. Stark (2012): Contrasting Maturation and Growth of Northern Rock Sole in the Eastern Bering Sea and Gulf of Alaska for the Purpose of Stock Management, North American Journal of Fisheries Management, 32:1, 93-99 To link to this article: http://dx.doi.org/10.1080/02755947.2012.655845

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 32:93–99, 2012 American Fisheries Society 2012 ISSN: 0275-5947 print / 1548-8675 online DOI: 10.1080/02755947.2012.655845

MANAGEMENT BRIEF

Contrasting Maturation and Growth of Northern Rock Sole in the Eastern Bering Sea and Gulf of Alaska for the Purpose of Stock Management

James W. Stark* National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Alaska Fisheries Science Center, Resource Assessment and Conservation Engineering Division, 7600 Sand Point Way Northeast, Seattle, Washington 98115, USA

sole) L. mochigarei in the western extent of its distribution off Abstract the Kuril Islands of the western North Pacific. Northern rock The primary purpose of this study was to provide commercial soles are most abundant in the southeastern Bering Sea (Fadeev fishery managers with the age- and length-at-maturity information 1965) where it is widely distributed in depths to 200 m but con- about northern rock sole Lepidopsetta polyxystra needed for them to set a sustainable overfishing limit and evaluate the precision centrated off the Pribilof Islands and in Bristol Bay (Acuna and of the two predictors of maturity. The estimated length at which Lauth 2008). In contrast, 80% of all the Gulf of Alaska northern 50% of eastern Bering Sea female northern rock sole matured rock soles is found in depths less than 100 m in the western and (L50) was 309 mm, which was significantly smaller than that for central areas (von Szalay et al. 2010). Gulf of Alaska females. We determined that the differences in Northern rock sole has high commercial value, primarily L50 between populations were probably the result of differences in the rate of female growth. Growth was significantly faster in from the roe and fillet products (Wilderbuer and Nichol 2007). the Gulf of Alaska than in the eastern Bering Sea during 1996 Sustainable management of northern rock sole is dependent and 1999. However, by 2007 the growth rates were similar between upon having realistic estimates of the female reproductive these areas. The variability in growth was correlated with seawater potential for each stock. Typically the most reliable correlates of temperature. There were also differences in the age at which 50% female maturity are age and body length. Previously, manage- of the females matured (A50) between the populations in the eastern Bering Sea (9 years) and the Gulf of Alaska (7 years). In contrast, ment of eastern Bering Sea northern rock soles relied on age at within the eastern Bering Sea, females maintained a similar A50 maturity estimates that were based on macroscopic anatomical over several years, which indicates that age is the most reliable maturity classifications collected by fishery observers during predictor of maturity for northern rock sole. the 1993 and 1994 eastern Bering Sea rock sole roe fishery. To improve the precision of the maturity estimates, this study

Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 was initiated as part of the Bering Sea northern rock sole The northern rock sole Lepidopsetta polyxystra is the third stock assessment conducted by the National Marine Fisheries most numerous groundfish species in the eastern Bering Sea, Service’s Alaska Fisheries Science Center (AFSC) Resource × 9 the current population and biomass being over 10 10 indi- Assessment and Conservation Engineering Division. The × 6 viduals and over 2 10 metric tons (mt; Acuna and Lauth maturity estimates are used with the AFSC survey estimates of 2008). Northern rock sole distribution overlaps widely with its population recruitment to estimate the spawning biomass and congener, rock sole L. bilineata, in the eastern Aleutian Islands spawner–recruit relationship, which are used to calculate the and southern portion of the eastern Bering Sea, the Gulf of population’s maximum sustainable yield (MSY), the fishery Alaska, and down the west coast of North America to Puget mortality rate that would give MSY (FMSY), and the overfishing Sound (Stark and Somerton 2002). Northern rock sole is pre- limit (Wilderbuer and Nichol 2007). This study employed his- dominant in the southern Bering Sea and Aleutian Islands, and tological maturity classification methods to determine female rock sole predominates in the Gulf of Alaska. Northern rock sole maturity, which are known to be less biased than the anatomical also overlaps with the ricecake sole (also known as the dusky maturity classification methods (Hunter et al. 1992) previously

*E-mail: [email protected] Received April 8, 2011; accepted August 11, 2011 93 94 STARK

used for eastern Bering Sea northern rock soles. Histological random stratified sampling of two to six specimens from each maturity assessments were previously used by Stark and 1-cm category of total body length distributed. These collec- Somerton (2002) to define the maturity of the Gulf of Alaska tions included northern rock sole specimens with body lengths northern rock sole population and its congener the rock sole. as small as 5 cm, as effected by a standard 0.89-cm-mesh cod end liner in the survey trawl. Maturity assessment.—In the laboratory the cassette of ovary METHODS tissues were then removed from the 10% formalin preservative, Sample collection.—Observers of the AFSC Fisheries Mon- dehydrated through graded ethanol, cleared, and embedded in itoring and Analysis Division collected 209 northern rock soles Paraplast blocks. The embedded ovaries were cut into sections taken in the eastern Bering Sea and delivered by commercial bot- 0.003–0.005 mm thick, mounted on a glass slide with a cover tom trawl vessels to seafood processing plants in Dutch Harbor, slip, stained with hemotoxin, and counterstained with eosin. Alaska, during February and March 2006. The observers were The cover slip was then sealed onto the slide. The stains were specifically trained in the taxonomic identification of rock soles. used to unambiguously delineate important structures within The species were differentiated using blind side coloration and the oocytes and ovary. Eosin was used to give a bright pink col- gill raker and supraorbital pore counts (Orr and Matarese 2000). oration to all yolk residing within all oocytes contained within All specimens were measured for total length (cm) and selected the ovary tissue. The eosin also stained fully developed outer using length-stratified samples of three to seven females from cell membranes (chorion) found in all oocytes; maturity stages each centimeter category. The specimen measurement scale was ranged from cortical alveioi to hydrated. The hemotoxin stain designed so that each 10 cm unit of measure spans the range from gave a distinct violet coloration to noneosinophylic structures of −5to + 5 mm total body length (L; note L was measured in cen- the cell to make them clearly discernable. All specimens were timeters and converted to millimeters for purposes of analysis). assessed according to a five-stage scale of ovary maturity based Ovary tissue was excised from each specimen, placed in a on cell structure criteria (Table 1). Females were classified ac- labeled histology cassette, which was submerged in a bottle cording to the most advanced stage of histological development containing a solution of 10% formalin. The maturity collec- that occurred in the ovary. Mature females included those in- tion included specimens as small as 21 cm. This length range dividuals that had any of the yolk-filled oocyte development was considered fully representative, based on a 1999 Gulf of stages (i.e., vitellogenesis, advanced yolk, and hydrated). Ma- Alaska northern rock sole maturity collection (Stark and Somer- ture females also included those with postovulatory follicles. ton 2002), which included specimens as small as 19 cm but All the mature females were those expected to spawn that year. found no mature females smaller than 30 cm in length. Spawning females were those with maturity stage classifications Northern rock sole growth determinations were made from of either hydrated or postovulatory follicle. unpublished AFSC bottom trawl survey age data, stratified by Statistical methods.—All analysis utilized S-Plus software length and area, taken from the eastern Bering Sea and Gulf of (version 2000 Professional release3, MathSoft Inc., Cambridge, Alaska bottoms trawl surveys of 1996, 1999, and 2007. These Massachusetts). Maturity was estimated as a function of L by survey collections were designed to provide the most represen- fitting a logistic function to the binary maturity data with gen- tative growth data available and allow for valid comparisons eralized linear modeling (Venables 1997). The between-month between years and areas. The AFSC surveys used standardized and between-area differences were tested by fitting the model fishing methods, including standardized station patterns to as- of maturity as a function of L, with and without terms distin- sess the condition of groundfish stocks across the eastern Bering guishing month and area, and then judging the significance of the terms via analysis of deviance (Venables 1997). Length at

Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 Sea. The full bottom depth distribution of northern rock soles was trawl-sampled, and for each North Paciﬁc Fishery Man- 50% maturity (L50) was estimated for each month and area by agement Council statistical area, specimens were selected using evaluating the ﬁtted model of maturity at 50% maturity and

TABLE 1. Ovary development stages and maturity categories assigned to individual northern rock sole females based on histological assessment of the most advanced cellular structures in the ovary.

Development stage Maturity category Cellular structures Perinucleus Immature Oocyte with multiple nucleoli bordering the nucleus; chorion absent. Cortical alveoli Immature Oocyte with cortical alveoli adjacent to developing chorion. Vitellogenesis Mature Oocyte predominantly contains yolk spheres. Advanced yolk Mature Oocyte ﬁlled with fused yolk; no individual yolk spheres. Hydrated Mature and spawning Oocyte expanded by hydration, located in lumen of ovary. Postovulatory follicles Mature and spawning Postovulatory follicles present. MANAGEMENT BRIEF 95

algebraically solving for length. The variance of L50 was esti- are considered to be the annuli, which are enumerated via a mated using bootstrapping (Efron and Tibshirani 1993) based on dissecting microscope. 1,000 resamplings, with replacement, of the maturity and length data. Differences in the L50 by month and area were tested with RESULTS a Z-test (Sokal 1969). The same procedures were used to compare maturity as a function of age and to compare length results Age and Length at Maturity from this study with northern data from a Gulf of Alaska study The collection included females ranging in length from 210 by Stark and Somerton (2002). Based on their study, February mm to 490 mm; ages ranged from 4 to 23 years. We found no was assumed to be the most representative period for sampling differences between the February and March collections for because all females that were expected to spawn that year would maturity at age (P = 0.89) and length (P = 0.73); consequently unambiguously appear to be mature and few would have com- the data were pooled to increase sample size to 162 for age and pleted spawning. The March collection provided the means to 209 for length. Fifty percent of female northern rock soles in the test whether smaller and younger females matured later in the eastern Bering Sea reached maturity at the age (A50) of approx- year than larger females. imately 9 years (Figure 1; Table 2), which was significantly (P Length at age was described by the von Bertalanffy growth < 0.001) older than for Gulf of Alaska females (A50 = 7 years; function, fitted using nonlinear least squares (Venables 1997). Stark and Somerton 2002). The youngest mature eastern Bering To evaluate differences in northern rock sole growth between Sea female collected was also the smallest mature female, at areas and years the von Bertalanffy model of growth was fitted 240 mm and age 6. Half the eastern Bering Sea females matured to the length-at-age data with terms distinguishing area and year (L50) at 309 mm in the eastern Bering Sea (Figure 2; Table 3), and retested for the combined areas and years. The goodness of which was significantly (P = 0.003) smaller than Gulf of Alaska fit to the data was tested comparing the separate-category and females (328 mm; Stark and Somerton 2002). The proportion combined-category models via the likelihood ratio test (Kimura of mature eastern Bering Sea female northern rock soles was 1980). Otoliths were aged by personnel of the AFSC Age and similar during February and March 2006, although the rate of Growth Program with standard validated methods (Fargo and spawning increased from less than 5% in February to approx- Chilton 1987; Kimura and Anderl 2005; Kimura et al. 2007), imately 30% in March (Figure 3). This result suggests that the which includes sectioning transversely through the otolith nu- spawning season of eastern Bering Sea northern rock soles began cleus and heating the otolith to increase the contrast between just before the February collection period and peaked during the the growth (transparent) and opaque layers. The growth layers spring.

1.0

0.8

0.6 Downloaded by [Department Of Fisheries] at 01:43 01 March 2012

0.4

Proportion of mature females 0.2

0.0

510152025

Years

FIGURE 1. Maturity estimated as a function of age for female northern rock soles collected in February and March 2006 (n = 162) in the eastern Bering Sea, as depicted by ﬁtting a logistic function to the maturity data. The mean age at 50% maturity (8.9 years) is indicated by solid vertical line; the 95% conﬁdence interval is represented by the dotted lines. 96 STARK

TABLE 2. Female northern rock sole age-at-maturity results based on ovary TABLE 3. Female northern rock sole length-at-maturity results based on histology samples collected from the eastern Bering Sea by the Alaska Fisheries ovary histology samples collected from eastern Bering Sea trawl samples (1996, Science Center during February and March 2006. The parameters of the logistic 1999, 2007) by the Alaska Fisheries Science Center by date of collection. See equation fit to the data are n (sample size), B (slope of the line) and its variance, Table 2 for additional details. A (y intercept) and its variance, the covariance of A and B (the product of the SDs of B and A and the coefficient of correlation between them), and A50 (the Sampling variable Sampling statistic age at which 50% of females were expected to reach sexual maturity) and its variance. n 209 B 0.08130 Sampling variable Sampling statistic A −25.09109 n 162 Variance of B 6.17796 B 0.74466 Variance of A 100.89386 A −6.63200 Covariance of B, A 0.00020 Variance of B 0.00107 L50 (mm) 308.62537 Variance of A 5.48510 Variance of L50 33.12513 Covariance of B, A −0.00385 A50 (years) 8.90603 Variance of A50 0.19481 DISCUSSION Length at Maturity and Growth Growth My study is the first to determine both the age and length The precision of rock sole age determinations was tested by at maturity of female northern rock soles in eastern Bering AFSC Age and Growth Program personnel (Kimura and An- Sea, and the histological maturity classification methods used derl 2005). The average reader and tester agreement for these improved accuracy of the estimates. The results from the years averaged 70% (CV = 3). To determine if area differences February collection were corroborated by the results from the in length at maturity were consistent with growth, I made the March collection and suggested that by February all females first growth comparison between regions for northern rock sole that would mature for the annual maturation cycle had matured. (Figure 4; Table 4). The results indicate that female growth rates I estimated that female northern rock soles of the eastern < were significantly (P 0.001) faster in the Gulf of Alaska than Bering Sea reached L50 at 309 mm, which was significantly in the eastern Bering Seas during 1996 and 1999. However, smaller than the Gulf of Alaska L50 estimate of 328 mm. The by 2007 the rate of female growth had increased in the east- smallest mature females found in the eastern Bering Sea were ern Bering Sea to a rate similar to that of the Gulf of Alaska 240 mm, compared with 300 mm for the Gulf of Alaska col- = population (P 0.69). lection. The smaller female L50 in the eastern Bering Sea was

1.0

0.8 Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 0.6

0.4

0.2 Proportion of mature females

0.0

100 200 300 400 500

Length (mm)

FIGURE 2. Maturity estimated as a function of total length for female northern rock soles collected in February and March 2006 (n = 209) in the eastern Bering Sea, as depicted by ﬁtting a logistic function to the maturity data. The mean length at 50% maturity (309 mm) is indicated by solid vertical lines; the 95% conﬁdence interval represented by dotted lines. MANAGEMENT BRIEF 97

0.8 February 0.7 March 0.6 0.5 0.4 0.3 0.2

Proportion of maturity stages 0.1 0 Hydrated follicles Perinucleus Vitellogenesis Postovulatory Advanced yolk Advanced Cortical alveoli Cortical Stage of ovary development

FIGURE 3. Proportion of eastern Bering Sea female northern rock soles classiﬁed within each ovarian maturity category from collections made during February (n = 114) and March (n = 95) of 2006. All females that had reached the 240-mm minimum length at maturity were included to investigate the progression of ovarian maturity.

probably due to significant differences in the rate of growth be- nuli growth correlated chronologically with summertime bottom tween the two populations. The female northern rock sole pop- temperatures for northern rock sole (R2 = 0.34), yellowfin sole ulation in the Gulf of Alaska had approximately twice the rate Limanda aspera (R2 = 0.81), and Alaska plaice Pleuronectes of growth as the Bering Sea population, during 1996 and 1999. quadrituburculatus (R2 = 0.61), based on fish aged as old as However, by 2007 the rate of growth of the eastern Bering Sea 23 years. female population had become similar to the growth rate of the Extreme average summer sea bottom temperatures occurred Gulf of Alaska population. In contrast, the Gulf of Alaska female in 1999 (0.83◦C) and 2003 (3.87◦C), which produced narrow population maintained a similar rate of growth during 1996, growth annuli within the otoliths of all three flatfish species. 1999, and 2007. Growth rates of eastern Bering Sea northern In contrast, broad annuli growth occurred during years that had rock sole, and two other flatfish species were directly affected by average bottom temperatures of 2.3◦C. Colder bottom temper- summer seawater temperatures on the shelf (Matta et al. 2010). atures correlated with slower growth in northern rock sole, and Summer is the feeding and growing season for these species, and this was confirmed for northern rock sole reared under lab- they have the same feeding grounds every year. The otolith an- oratory conditions (Hurst and Abookire 2006). The summer

TABLE 4. Female northern rock sole length at age as described by the von Bertalanffy growth equation for females in the Bering Sea and Gulf of Alaska during 1996, 1999, and 2007, based on summer Alaska Fisheries Science Center groundﬁsh surveys. Components include n (the sample size) and equation parameters

Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 L∞ (length at maximum age), k (the estimated growth increment), and t0 (the theoretical age when ﬁsh length is 0). Bering Sea Gulf of Alaska Sampling variable 1996 1999 2007 1996 1999 2007 n 280 298 263 142 227 261 L∞ (mm) 488.807 571.613 418.235 429.3187 434.9223 472.1492 var(L∞) 193.4323 1022.7257 45.9070 195.5226 430.6108 221.6947 k 0.1096 0.0710 0.1581 0.2360 0.1596 0.1668 var(k) 0.0001 0.0001 0.0001 0.0014 0.0011 0.0004 t0 0.2563 0.0982 0.9135 0.3869 −0.9860 −0.0352 var(t0) 0.0329 0.0873 0.0295 0.1944 0.5160 0.1188 cov(L∞, k) −0.1039 −0.2548 −0.0558 −0.4391 −0.6405 −0.2609 cov(L∞, t0) −1.8848 −7.9425 −0.7017 −3.6262 −12.1423 −3.4234 cov(k, t0) 0.0012 0.0022 0.0014 0.0147 0.0223 0.0059 98 STARK

1996 change would further affect northern rock sole growth through 500 changes in food supply and body metabolism and, consequently, the size at which females mature in the future. In contrast, fe- 400 males of the Gulf of Alaska northern rock sole population may

300 continue to have a lower interannual variability in growth and mean length at maturity than the eastern Bering Sea population 200 because of a more moderate temperature regime. Compared with the Bering Sea, the Gulf of Alaska is more insulated from arctic 100 temperatures because of its lower latitude and greater circula- 5101520 tion of seawater (Stabeno et al. 2001). Consequently, the Gulf of Alaska is not subject to an annual accumulation of sea ice, 1999 which can alter the timing and abundance of primary and sec- 500 ondary production and, thereby, northern rock sole metabolism

400 and growth.

300 Age at Maturity Length (mm) 200 Although female northern rock soles in the eastern Bering Sea had very different growth rates over years, the age at which 100 females matured was similar between years. Based on my study, 5101520 the eastern Bering Sea population reached A50 at an estimated age of 9 years. The A50 estimate was comparable to the pre- 2007 500 vious age estimate of 9–10 years used by eastern Bering Sea stock managers (Wilderbuer and Nichol 2007), that was based 400 on a 1994 collection. Similarities in the estimated female age

300 at maturity in the eastern Bering Sea between 1994 and 2006 suggests that age is a reliable predictor of maturity for northern 200 rock soles, at least for the eastern Bering Sea population. As a consequence of maturing at a significantly older age (2 years 100 older) than Gulf of Alaska females, the eastern Bering Sea pop- 5101520 ulation was more dependent upon older females to sustain the Years population. Females of both populations have the same potential life span (Figure 4). Compared with the younger maturing Gulf FIGURE 4. Total length at age of female northern rock soled from collections of Alaska females, delayed maturation in the eastern Bering made in the eastern Bering Sea (solid triangles, solid line) and Gulf of Alaska Sea means that these females would probably have fewer years (open squares, dashed line) during areawide groundfish assessment surveys available for spawning, potentially decreasing the individual conducted by the Alaska Fisheries Science Center. Sample sizes by year— 1996, 1999, and 2007, respectively—were 281, 298, and 263 for the Bering Sea reproductive production from each female. and 142, 227, and 261 for the Gulf of Alaska. The results suggest the spawning season begins before Febru- ary and peaks during the spring for the eastern Bering Sea popu- Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 lation, which is consistent with observations by Shubnikov and growing period bottom temperatures and spring plankton bloom Lisovenko (1964) in the southeastern Bering Sea and in the Gulf directly correlate with the extent of the annual winter sea ice of Alaska (Stark and Somerton 2002). The depth and location sheet formation and the timing and extent of the sea ice melts on of spawning is not well known for the eastern Bering Sea pop- the Bering Sea shelf (Stabeno et al. 2001). Sea ice that extends ulation. In the western Bering Sea northern rock sole spawned into the shelf during March or April indicates stratification of in depths less than 200 m (Pertseva-Ostroumova 1961), and in the water column, which produces an early and intense plank- the Gulf of Alaska spawning occurred in depths less than 65 m ton bloom at the ice edge. This type of plankton bloom can (Stark and Somerton 2002). The spawning grounds were located produce up to 65% of the annual primary productivity over the within bays, near shore, and on the inner portions of banks in shelf (Niebauer et al. 1990). In years that have strong mixing areas that had gravel and a hard sandy bottom. This type of sub- of the seawater column during March and April, there is no ice strate is essential for the attachment of the rock sole’s demersal formation, and the plankton bloom is delayed until May or June adhesive eggs. and is attenuated. Water temperatures are expected to undergo Results of my study provide fishery managers with the neces- more frequent fluctuations and rise an average of 2◦Cbythe sary maturity estimates that allow them to refine the overfishing year 2050 in the eastern Bering Sea (Hollowed et al. 2009). The and allowable catch limits of northern rock sole in the eastern MANAGEMENT BRIEF 99

Bering Sea. I determined that age-based maturity estimates evaluation of assumptions and precision. U.S. National Marine Fisheries would be much more reliable than length-based estimates and Service Fishery Bulletin 90:101–128. provides managers with a much more accurate estimate of Hurst, T. P., and A. A. Abookire. 2006. Temporal and spatial variation in potential and realized growth rates of age-0 year northern rock sole. Journal of Fish the age at which females mature. Consequently managers can Biology 68:905–919. now produce more precise estimates of the spawning biomass Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. and spawner–recruit relationship. As a result, managers will U.S. National Marine Fisheries Service Fishery Bulletin 77:765–776. have more precise estimates of the maximum sustainable yield, Kimura, D. K., and D. M. Anderl. 2005. Quality control of age data at the Alaska fishery mortality rate, and overfishing limit than was previously Fisheries Science Center. Marine and Freshwater Research 56:783–789. Kimura, D. K., D. M. Anderl, and B. J. Goetz. 2007. Seasonal marginal growth possible using other methods. on otoliths of seven Alaska groundfish species support the existence of annual patterns. Alaska Fisheries Research Bulletin 12:243–251. Matta, M. E., B. A. Black, and T. K. Wilderbuer. 2010. Climate-driven synchrony ACKNOWLEDGMENTS in otolith growth-increment chronologies for three Bering Sea flatfish species. Marine Ecology Progress Series 413:137–145. The following Alaska Fisheries Science Center personnel Niebauer, H. J., V. A. Alexander, and S. M. Henrichs. 1990. Physical and made this study possible: T. Wilderbuer, A. Hollowed, and R. biological oceanographic interaction in the spring bloom at the Bering Nelson. Collections were made by A. Baattista, K. Hardin, and Sea marginal ice edge zone. Journal of Geophysical Research 95:22229– L. Kerr. Age determinations were made by B. Matta and D. 22241. Anderl. Reviews were made by D. Somerton, G. Duker, J. Lee, Orr, J. W., and A. C. Matarese. 2000. Revision of the genus Lepidopsetta Gill, 1862 (Teleostei: Pleuronectidae) based on larval and adult morphol- and M. Wilkins. This manuscript was improved by suggestions ogy, with a description of a new species from the North Pacific Ocean and from two anonymous reviewers, editor C. Griswold and the Bering Sea. U.S. National Marine Fisheries Service Fishery Bulletin 98:539– associate editor. 582. Pertseva-Ostroumova, T. A. 1961. [The reproduction and development of Far Eastern flounders.] Academy Nakhimovsky, Institute of Oceanology, Union REFERENCES of Soviet Socialist Republics. (In Russian.). English translation by Translation Acuna, E., and R. R. Lauth. 2008. Results of the 2007 eastern Bering Sea con- Series, Fisheries Research Board of Canada 856 (1967), Nanaimo, British tinental shelf bottom trawl survey of groundfish and invertebrate resources. Columbia. NOAA Technical Memorandum NMFS-AFSC-181. Shubnikov, D. A., and L. A. Lisovenko. 1964. Data on the biology of rock Efron, B., and R. J. Tibshirani. 1993. An introduction to the bootstrap. Chapman sole of the southeastern Bering Sea. Pages 220–226 in P. A. Moiseev, editor. and Hall, New York. Soviet fisheries investigations in the northeast Pacific, part 2. Translated from Fadeev, N. S. 1965. Comparative outline of the biology of flatfishes in the south- Russian 1968: Israel Program for Scientific Translations, Jerusalem. eastern part of the Bering Sea and condition of their resources. Soviet fisheries Sokal, R. R. 1969. Biometry. Freeman, San Francisco. investigations in the northeast Pacific, part 4. Izvestiya Tikhookeanskogo Stabeno, P.J., N. A. Bond, N. B. Kachel, S. A. Salo, and J. D. Schumacher. 2001. Nauchno-Issledovatel’skogo Instituta Rybnogo Khozyaistva i Okeanografii On the temporal variability of the physical environment over the southeastern 58:112–120. (Translated from Russian by Isreal Program for Scientific Trans- Bering Sea. Fisheries Oceanography 10:81–98. lations, Jerusalem, 1968). Stark, J. W., and D. A. Somerton. 2002. Maturation, spawning and growth of Fargo, J., and D. E. Chilton. 1987. Age validation for rock sole in Hecate Strait, rock soles off Kodiak Island in the Gulf of Alaska. Journal of Fish Biology British Columbia. Transactions of the American Fisheries Society 116:776– 61:417–431. 778. Venables, W. N. 1997. Modern applied statistics with S-Plus. Springer-Verlag, Hollowed, A. B., N. A. Bond, T. K. Wilderbuer, W. T. Stockhausen, Z. T. A’mar, New York. R. J. Beamish, J. E. Overland, and M. K. Schirripa. 2009. A framework von Szalay, P. G., N. W. Raring, F. R. Shaw, M. E. Wilkins, and M. H. Martin. for modelling fish and shellfish responses to future climate change. ICES 2010. Data report 2009: Gulf of Alaska bottom trawl survey. NOAA Technical (International Council for the Exploration of the Sea) Journal of Marine Memorandum NMFS-AFSC-208. Science 66:1584–1594. Wilderbuer, T. K., and D. G. Nichol. 2007. Stock assessment and fishery eval-

Downloaded by [Department Of Fisheries] at 01:43 01 March 2012 Hunter, J. R., B. J. Macewicz, N. C. H. Lo, and C. A. Kimbrell. 1992. Fecundity, uation: Bering Sea and Aleutian Islands northern rock sole. North Paciﬁc spawning, and maturity of female Dover sole Microstomus paciﬁcus, with an Fishery Management Council, Anchorage, Alaska. This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 01:14 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 Evaluation of a Gastric Lavage Method for Sturgeons Laurent Brosse a , Pierre Dumont b , Mario Lepage c & Eric Rochard d a Cemagref, Unité Ressources Aquatiques Continentales, 50 avenue de Verdun, 33612 CESTAS Cedex, France b Faune et Parcs Québec, 201 Place Charles-Lemoyne, Longueuil, J4K 2T5, Québec, Canada c AGEDRA, 805 Chemin de Reden, 33240 Saint-André de Cubzac, France d Cemagref, Unité Ressources Aquatiques Continentales, 50 avenue de Verdun, 33612 CESTAS Cedex, France Available online: 09 Jan 2011

To cite this article: Laurent Brosse, Pierre Dumont, Mario Lepage & Eric Rochard (2002): Evaluation of a Gastric Lavage Method for Sturgeons, North American Journal of Fisheries Management, 22:3, 955-960 To link to this article: http://dx.doi.org/10.1577/1548-8675(2002)022<0955:EOAGLM>2.0.CO;2

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 22:955±960, 2002 ᭧ Copyright by the American Fisheries Society 2002

Evaluation of a Gastric Lavage Method for Sturgeons

LAURENT BROSSE* Cemagref, UniteÂ Ressources Aquatiques Continentales, 50 avenue de Verdun, 33612 CESTAS Cedex, France

PIERRE DUMONT Faune et Parcs QueÂbec, 201 Place Charles-Lemoyne, Longueuil J4K 2T5, QueÂbec, Canada

MARIO LEPAGE AGEDRA, 805 Chemin de Reden, 33240 Saint-AndreÂ de Cubzac, France

ERIC ROCHARD Cemagref, UniteÂ Ressources Aquatiques Continentales 50 avenue de Verdun, 33612 CESTAS Cedex, France

Abstract.ÐBecause of their threatened status, stur- lantic sturgeon A. oxyrinchus and shortnose sturgeons (Acipenseridae) can no longer be sacri®ced for geon to determine harmlessness; however, lavage stomach content analysis. We tested a nonlethal method ef®ciency by type of prey contained in the stomach of gastric lavage on Siberian sturgeon Acipenser baeri. The ef®ciency and harmlessness of the method were has not been studied (Haley 1998). In lake stur- tested with four different volumes of food (10, 20, 30, geon A. fulvescens, gastric lavage using water re- and 40 cm3), each composed of brown shrimp Crangon ¯ux has been tested for ef®ciency and shown no crangon, Chironomidae, earthworm Lumbriscus terres- in¯uence due to the type of prey contained in the tris, and sand goby Pomatoschistus minutus. The Sibe- stomach (Nilo 1996). However, Sprague et al. rian sturgeons were force-fed before the gastric lavage (1993) showed that water at too high a pressure was performed. Some prey were recovered from all the sturgeons, and the average food item recovery rate from could lead to internal injuries or even death by stomach contents was 67.5%; recovery of brown shrimp rupturing the swim bladder. Because of that, tech- and sand goby (78.2%) was greater than that of ver- niques using a syringe connected to a ¯exible tube miform items (51.4%). The volume of food had no sig- have been proposed and tested (Meehan and Miller ni®cant in¯uence on the prey recovery rate. No mortality 1978; Haley 1998). resulted from the gastric lavage. However, the method We chose a pulsed gastric lavage method (Foster is not totally benign because the ®sh that had undergone 1977) inspired by those that use a pressurized wa- gastric lavage averaged signi®cantly greater weight loss than the control ®sh during the 60-d follow-up period. ter reservoir (Light et al. 1983; Nilo 1996), which has the advantage of providing a continuous supply of water. The aim of this work was to test the Because many sturgeon (Acipenseridae) species ef®ciency of this method in recovering various Downloaded by [Department Of Fisheries] at 01:14 01 March 2012 have a threatened status, the study of their feeding types of prey from Siberian sturgeon A. baeri and habits must employ methods that are as safe and to determine any resulting damage, such as death ef®cient as possible. The combination of these cri- or weight loss. teria explains partly why data on sturgeon diet are scarce. Diet studies of acipenserids had involved Methods the analyses of stomach contents recovered from The tests were performed from January to corpses, such as with the Gulf sturgeon Acipenser March 1999 on farmed Siberian sturgeons at the oxyrinchus desotoi (Mason and Clugston 1993) and Cemagref experimental station (St-Seurin-sur- the shortnose sturgeon A. brevirostrum (Dadswell l'Isle, France). Four test groups of ®ve individuals 1979; Carlson and Simpson 1987). Since then, gas- were formed (N ϭ 20, lavage set), and one group tric lavage procedures have been tested on the At- of 15 individuals was kept as a control. All groups were of similar fork length (FL; 78±105 cm). Each * Corresponding author: laurent.brosse@bordeaux. lavaged sturgeon was force-fed a mixture of ®ve cemagref.fr different types of prey: earthworms Lumbriscus Received January 17, 2001; accepted January 3, 2002 terrestris, chironomid larvae, sand goby Pomatos-

955 956 BROSSE ET AL.

TABLE 1.ÐNumber of prey used for the different volumes force-fed to Siberian sturgeon.

Number of prey Force-fed Number of volume Chiro- Earth- Small Large Set sturgeon (cm3) nomids worms shrimp shrimp Fish Lavage 5 10 40 1 9 4 5 Lavage 5 20 50 2 14 8 6 Lavage 5 30 60 4 21 12 7 Lavage 5 40 60 5 28 15 9 Control 15 0 0 0 0 0 0

chistus minutus (1±3 cm total length), and small the tank at the beginning of the lavage); a 500- (cephalothorax Ͻ1 cm) and large (cephalothorax ␮m-mesh sieve for collecting the stomach con- Ͼ1 cm) brown shrimp Crangon crangon (Table 1). tents; a 6-mm OD ¯exible tube for injecting the Quantities of the ®ve prey types in the mixture to water; a 12-mm OD ¯exible tube for ¯ushing out be force-fed were known. Volumes of prey varied the water (plus the food bolus) into which the plas- and were based on previous measurement of stom- tic 6-mm-OD ¯exible tube was inserted (Figure 1). ach volumes of several sturgeons of the same Throughout the operations, which lasted less than length range. Force-feeding took about 1 min, dur- 2 min, a ®sh was held dorsal side down in an ing which a ¯exible tube (12 mm outside diameter improvised cradle, and its head and gills were con- [OD], 20 cm long) ®lled with an exact volume of tinuously hosed with freshwater. Both tubes were prey items was inserted about 5 cm into the di- inserted into the forward part of the digestive tract. gestive tract and purged with a piston. The large tube was inserted about 5 cm into the The materials used for gastric lavage were mod- tract, and the water injection tube was inserted as eled on those used by Nilo (1996). The system was far as the ®rst stomach loop, approximately 20 cm composed of a preset-pressure garden sprayer with (Figure 2). The injection tube was gently manip- a 7-L tank (maximum pressure of 2 ϫ 105 Pa in ulated in a to-and-fro motion to dislodge and carry Downloaded by [Department Of Fisheries] at 01:14 01 March 2012

FIGURE 1.ÐThe gastric lavage process used on Siberian sturgeons(modeled after Nilo 1996). Pressurized water (maximum pressure of 2 ϫ 105 Pa in the tank at the beginning of the lavage) was inserted with the small tube (6 mm outside diameter [OD]) and returned with prey items outside the ®sh via the large tube (12 mm OD). MANAGEMENT BRIEFS 957 Downloaded by [Department Of Fisheries] at 01:14 01 March 2012

FIGURE 2.ÐDigestive tract of Siberian sturgeon (119 cm fork length), showing placement of the small water in¯ow tube (6 mm outside diameter [O.D]) and the large out¯ow tube (12 mm O.D) used in gastric lavage (see Figure 1). 958 BROSSE ET AL.

TABLE 2.ÐMean gastric lavage recovery rate for prey types and volume of prey force-fed to Siberian sturgeon. Group A ϭ chironomids ϩ earthworms; Group B ϭ small and large shrimp ϩ ®sh.

Group B Force-fed Group A volume Small Large (cm3) Chironomids Earthworms Mean shrimp shrimp Fish Mean Total mean 10 62.0 20.0 41.0 77.8 70.0 92.0 79.9 64.4 20 54.7 80.0 67.4 85.7 70.0 63.3 73.0 70.8 30 45.4 45.0 45.2 89.5 76.7 77.1 81.1 66.7 40 28.3 76.0 52.2 78.6 69.3 88.9 78.9 68.2 Mean 47.6 55.3 51.4 82.9 71.5 80.3 78.2 67.5

the prey from the stomach. The length of the in- in¯uence on the recovery (P Ͻ 0.001), with a sig- jection tube was obtained by preliminary tests on ni®cantly lower recovery rate (P Ͻ 0.001) for ®ve sacri®ced Siberian sturgeons. group A (earthworms and chironomids; 51.4%) The control ®sh were not given a gastric lavage than for group B (large and small shrimp and ®sh; and only underwent simulated force-feeding (i.e., 78.2%). The volume of prey used had no signi®- introducing the tube and pushing the piston but cant in¯uence on the recovery rate (P ϭ 0.97), without prey). Fish were weighed immediately af- regardless of the prey type (0.07 Ͻ P Ͻ 0.90).

ter the gastric lavage (W0) and at 30 (W30) and 60 Impact on the Sturgeons (W60) d after lavage. The mean relative change in weight (Rt) during these periods was calculated as No mortality was observed during the 60 d fol- Rt ϭ⌺(W30 or 60 Ϫ W0)/⌺W0. The standard error of lowing lavage. No traces of injury were observed this ratio was estimated according to Cochran in the forward part of the digestive tract in the set (1977). Mortality was monitored over the 60-d test of ®sh sacri®ced immediately after gastric lavage period, at the end of which all the sturgeons (la- (N ϭ 5). A signi®cant loss of weight occurred in vage and control set; N ϭ 35) were sacri®ced so both the lavaged (P Ͻ 0.001) and the control sets that their digestive tracts could be examined. Five (P ϭ 0.001) after 60 d (Table 3). No signi®cant more individuals used for preliminary tests were difference in weight change was noted between sacri®ced immediately after force-feeding and la- the lavaged and control sets at either 30 d (P ϭ vaging to assess more immediate injuries. 0.395) or 60 d (P ϭ 0.188) postlavage. However, We used the Kruskall±Wallis H-test to test a signi®cant difference in weight change between whether recovery rate (ef®ciency of the method) the lavaged (an average loss of 7.97%) and control differed among prey types. We examined the sets (an average loss of 5.84%) was found when weight variation within (Wilcoxon matched-pairs the 30-d and 60-d data were combined (P ϭ 0.008). test) and between (Mann±Whitney U-test) the sets to test whether or not the method harmed the ®sh. Discussion All the statistics were performed using SYSTAT Overall, the stomach content recovery rate is software (SPSS 1998). satisfactory using this technique because we re-

Downloaded by [Department Of Fisheries] at 01:14 01 March 2012 covered items from 100% of the ®sh tested, which Results is higher than results for lake sturgeons (96%) by Ef®ciency of the Method Nilo (1996) using a similar method or for Atlantic Some prey were recovered for all Siberian stur- sturgeon (91%) by Haley (1998) using a syringe. geons tested (N ϭ 20), and the average prey re- Regardless of the volume fed, all prey types were covery rate was 67.5% over all the samples (Table recovered. However, the variation in the recovery 2). The type of prey was found to have a signi®cant rate among prey types may lead to a distorted pic-

TABLE 3.ÐAverage (standard deviation) of Siberian sturgeon weight and relative weight loss rate, Rt, during the 60- d period following gastric lavage (t0, t30, and t60 correspond to the 1st, 30th, and 60th days of observation).

Mean weights (g) Rt (%) Set N t t t R R R 0 30 60 t0Ϫ30 t30Ϫ60 t0Ϫ60 Lavage 20 6,145 (1,135) 5,670 (1,303) 5,655 (1,296) Ϫ7.73 (1.81) Ϫ0.26 (0.06) Ϫ7.97 (1.86) Control 15 6,160 (1,494) 5,827 (1,419) 5,800 (1,454) Ϫ5.41 (1.48) Ϫ0.46 (0.13) Ϫ5.84 (1.60) MANAGEMENT BRIEFS 959

ture of the actual diet of Siberian sturgeons ex- this method is not totally benign and may induce amined by this method. That is, we recovered ap- some long-term modi®cations in the ®sh's feeding proximately 50% of the vermiform prey and 75% behavior. We would consequently recommend us- of the larger prey; so variation in prey-type re- ing it only once on a particular animal, even if covery is likely when using this method on ®sh in recaptured beyond an interval of several months. their natural environment. Foster (1977) noted that The results we achieved with this method en- certain methods (suction pumping and regurgita- couraged us to use it on juvenile European stur- tion) gave signi®cantly different results according geon A. sturio captured in the Gironde estuary in to the prey type, whereas other methods (pulsed France (N ϭ 77; 56±103 cm FL); prey were re- or re¯ux gastric lavage) did not result in such var- covered in about 98% of the cases (Brosse et al. iations. Nilo (1996) did not observe such biases, 2000). The analysis of the contents shows that Eu- but Haley (1998) suggested a need to examine this ropean sturgeons feed mainly on small polychaetes possibility. (about 3±5 cm long) and, to a lesser degree, brown Using a pressurized water reservoir of several shrimp (2.5 cm maximum length) and sand goby liters enables rapid gastric lavage compared with (3 cm TL). other methods, such as that used by Haley (1998), in which the reservoir is a 60-mL syringe. Simi- Acknowledgments larly, using natural prey in tests provides better This study has been supported by the European simulation of natural conditions than does use of Union LIFE program, the French MinisteÁres de arti®cial food (i.e., Haley 1998). Our method limits l'AmeÂnagement du Territoire et de l'Environnement handling time and associated stress and allows for and de l'Agriculture et de la PeÃche, the Affaires Mar- the rapid release of wild ®sh near their capture itimes, the ReÂgions Aquitaine and Poitou-Charentes, site. The use of double tubing reduces problems the DeÂpartements de la Charente Maritime and de la linked with potential contractions of the pharynx Gironde, the Agence de l'eau Adour-Garonne, and or esophagus that could limit the recovery of the by a joint convention between Cemagref (France) food bolus. Moreover, the insertion of the larger and La SocieÂteÂ de la faune et des parcs du QueÂbec tube far enough into the digestive tract could pre- (Canada). Thanks to Daniel Mercier, Marcel Pelard, vent the introduction of the small tube (injection Thierry Rouault, and Philippe Jatteau for their valu- tube) into the swim bladder. able help and advice. Special thanks to Jeremy Spen- In preliminary tests conducted before our study cer for the English translation. herein reported, we found that at2hormore after sturgeons were force-fed food, very few of the References originally injected prey items were recovered, Brosse, L., E. Rochard, P. Dumont, and M. Lepage. which indicates that only the items recently in- 2000. First results on the diet of the young european gested are recovered with our technique. Having sturgeon, Acipenser sturio Linnaeus, 1758, in the access only to the forward part of the digestive Gironde estuary. Boletin Instituto EspanÄol de tract may be a handicap, compared with other OceanografõÂa 16:75±80. methods used on other sturgeon species (Haley Carlson, D. M., and K. W. Simpson. 1987. Gut content of juvenile shortnose sturgeon in the upper Hudson 1998). On the other hand, accessing the recently Estuary. Copeia 1987:796±802. Downloaded by [Department Of Fisheries] at 01:14 01 March 2012 ingested fraction of the food bolus provides the Cochran, W. G. 1977. Sampling techniques, 3rd edition. advantage of enabling the identi®cation of the lo- Wiley, New York. cal feeding ground, which could be useful for hab- Dadswell, M. J. 1979. Biology and population charac- itat studies. teristic of the shortnose sturgeon, Acipenser brevi- No mortality was observed over the 60 d fol- rostrum LeSueur 1818 (Osteichthyes: Acipenseri- dae), in the Saint John River Estuary, New Bruns- lowing gastric lavage, which supports ®ndings of wick. Canadian Journal of Zoology 57:2186±2210. similar studies (Meehan and Miller 1978; Gaudin Foster, J. R. 1977. Pulsed gastric lavage: an ef®cient et al. 1981; Light et al. 1983; Hartleb and Moring method of removing the stomach contents of live 1995). Occasional weight losses similar to those ®sh. Progressive Fish-Culturist 39:166±169. we noted in the lavaged set are regularly observed Gaudin, P., E. Martin, and L. CailleÁre. 1981. Le tubage in Siberian sturgeon broodstock during the winter gastrique chez les poissons: mise au point d'un eÂqui- pement et test chez le chabot Cottus gobio L. Bul- at the biological station (P. Williot, Cemagref, per- letin Francais de Pisciculture 282:8±15. sonal communication). However, the slight but sig- Haley, N. 1998. A gastric lavage technique for char- ni®cant difference we observed in weight change acterizing diet of sturgeons. North American Jour- between the lavaged and control sets indicates that nal of Fisheries Management 18:978±981. 960 BROSSE ET AL.

Hartleb, C. F., and J. R. Moring. 1995. An improved Nilo, P. 1996. Force des classes d'aÃge, habitats et ali- gastric lavage device for removing stomach content mentation des esturgeons jaunes (Acipenser fulves- from live ®sh. Fisheries Research 24:261±265. cens) juveÂniles du systeÁme Saint Laurent. MeÂmoire Light, R. W., P. H. Adler, and D. E. Arnold. 1983. Eval- de MaõÃtrise en biologie. UniversiteÂ du Quebec aÁ uation of gastric lavage for stomach analyses. North Montreal, Montreal. American Journal of Fisheries Management 3:81± Sprague, C. R., Beckman L. G., and S.D. Duke., 1993. 85. Prey selection by juvenile white sturgeon in res- Mason, W. T., Jr., and J. P. Clugston. 1993. Food of the ervoirs of the Columbia River. Pages 229±243 in Gulf sturgeon in the Suwannee River, Florida. R. C. Beamesderfer and A. A. Nigro, editors. Status Transactions of the American Fisheries Society 122: and habitat requirement of the white sturgeon pop- 378±385. ulations in the Columbia River downstream from Meehan, W. R., and R. A. Miller. 1978. Stomach ¯ush- McNary Dam, volume 2. Final Report of Oregon ing: effectiveness and in¯uence on survival and Department of Fish and Wildlife to Bonneville Pow- conditions of juvenile salmonids. Journal of Fish- er Administration, Portland. eries Research Board of Canada 35:1359±1363. SPSS. 1998. SYSTAT 8.0, statistics. SPSS, Chicago. Downloaded by [Department Of Fisheries] at 01:14 01 March 2012 This article was downloaded by: [Department Of Fisheries] On: 01 March 2012, At: 00:54 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

Standard Weight (Ws ) Equation and Length Categories for Shovelnose Sturgeon Michael C. Quist a , Christopher S. Guy a & Patrick J. Braaten a a Kansas Cooperative Fish and Wildlife Research Unit, Division of Biology, Kansas State University, 205 Leasure Hall, Manhattan, Kansas, 66506, USA Available online: 08 Jan 2011

To cite this article: Michael C. Quist, Christopher S. Guy & Patrick J. Braaten (1998): Standard Weight (Ws ) Equation and Length Categories for Shovelnose Sturgeon, North American Journal of Fisheries Management, 18:4, 992-997 To link to this article: http://dx.doi.org/10.1577/1548-8675(1998)018<0992:SWWSEA>2.0.CO;2

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 18:992±997, 1998 ᭧ Copyright by the American Fisheries Society 1998

Standard Weight (Ws) Equation and Length Categories for Shovelnose Sturgeon

MICHAEL C. QUIST,* CHRISTOPHER S. GUY, AND PATRICK J. BRAATEN Kansas Cooperative Fish and Wildlife Research Unit,1 Division of Biology Kansas State University, 205 Leasure Hall, Manhattan, Kansas 66506, USA

Abstract.ÐWeight±length data were compiled from 32 than or equal to stock length)´100] and relative populations of shovelnose sturgeon Scaphirhynchus pla- stock density [RSD ϭ (number of ®sh greater than torynchus (N ϭ 11,820) from nine states within the geo- or equal to a speci®ed length/number of ®sh great- graphic distribution of the species. We used the regression- line-percentile technique, which provides a 75th-per- er than or equal to stock length)´100] are commonly used to assess size structure. However, be- centile standard, to develop the standard weight (Ws) equation. The proposed equation in metric units is fore PSD or RSD can be calculated for a ®sh pop- log10Ws ϭϪ6.287 ϩ 3.330 log10FL; Ws is weight in ulation, species-speci®c length categories must be grams and FL is fork length in millimeters. The equiv- de®ned. See Gabelhouse (1984a) for methods to alent equation in English units is log10Ws ϭϪ4.266 ϩ determine stock, quality, preferred, memorable, 3.330 log10FL; Ws is weight in pounds and FL is fork length in inches. These equations are proposed for use and trophy lengths. with shovelnose sturgeon between 120 mm (5 in) and Despite the widespread use of Wr and stock den-

1,050 mm (41 in). Values for relative weight (Wr) cal- sity indices with warmwater sport®sh, Ws equa- culated with the Ws equation did not consistently in- tions and length categories are virtually nonexis- crease or decrease with increasing ®sh length, indicating tent for large-river, nongame ®shes. Therefore, the absence of length bias. We propose the following length objectives of this study were to (1) develop a W categories for calculation of proportional stock density s equation for shovelnose sturgeon Scaphirhynchus (PSD) and relative stock densities (RSDs): stock, 250 mm (10 in); quality, 380 mm (15 in); preferred, 510 mm platorynchus based on the regression-line-percentile (20 in); memorable, 640 mm (25 in); and trophy, 810 (RLP) technique, (2) evaluate the consistency of mm (32 in). We found signi®cant relations between size Wr across various lengths, (3) evaluate Wr distri- structure indices and mean population Wr. Additionally, butions of individuals and populations, (4) deter- we found signi®cant differences among incremental Wr mine length categories (i.e., stock, quality, pre- values. We believe the Ws equation and length category ferred, memorable, and trophy lengths), and (5) designations will be useful tools for managing shovelnose sturgeon populations. assess the relations between Wr and size structure for shovelnose sturgeon populations.

Relative weight [Wr ϭ 100´W/Ws, where W is Methods the observed weight and Ws is the length-speci®c standard weight value for the species; Wege and Weight±length data for shovelnose sturgeon Anderson (1978)] is commonly used as a man- were solicited from biologists within the geo- agement tool to assess the physiological well-be- graphical distribution of the species (Lee et al. ing of ®shes (i.e., condition). Relative weight is 1980). Populations represented by less than 30 in- Downloaded by [Department Of Fisheries] at 00:54 01 March 2012 easier to interpret than the traditional Fulton con- dividuals were omitted from all analyses. All data were evaluated as fork lengths (FL). One data set dition factor (K), because unlike K, Wr neither increases with increasing ®sh length nor varies by submitted as total lengths (TL) was converted to fork length with an equation developed from data species. Thus, Wr is useful when comparing within and among ®sh populations. sets that included both FL and TL [FL (mm) ϭ Stock density indices are also useful manage- (TL Ϫ 43.14)/1.02; r ϭ 0.99, P ϭ 0.0001]. Ad- ment tools to assess ®sh populations. Proportional ditionally, all weights and lengths were converted stock density [PSD ϭ (number of ®sh greater than to grams and millimeters. or equal to quality length/number of ®sh greater The RLP technique was used to develop the Ws equation for shovelnose sturgeon (Murphy et al. 1991). The minimum length for weight precision * Corresponding author: [email protected] was determined by plotting the variance-to-mean 1 The Unit is jointly supported by Kansas State Univer- sity, the Kansas Department of Wildlife and Parks, the ratio by 1-cm length-groups, as suggested by Mur- U.S. Geological Survey Biological Resources Division, phy et al. (1990). Minimum length was set where and the Wildlife Management Institute. the variance-to-mean ratio exceeded approximate- 992 MANAGEMENT BRIEFS 993

ly 0.01 (Neumann and Murphy 1991). Minimum used. All weight±length regressions were signi®- sample size for the application of the RLP tech- cant (P Յ 0.0001), and all but three correlation nique was determined by methods described in coef®cients were 0.93 or higher (Table 1).

Brown and Murphy (1996). Independence of Wr The following Ws equation for shovelnose stur- across length-classes for each study population geon was calculated with the 75th-percentile RLP was evaluated by assessing the number of signif- technique:

icant positive and negative slopes of Wr regressed on ®sh length. The total number of signi®cant pos- log10Ws ϭϪ6.287 ϩ 3.330 log10 FL; itive and negative slopes were compared with a where W is standard weight in grams and FL is binomial test to detect any length-related bias in s in millimeters. The equivalent in English units is the Ws equation (Mehta and Patel 1995). Length categories were determined by methods log10Ws ϭϪ4.266 ϩ 3.330 log10 FL; described by Gabelhouse (1984a). He suggested

that minimum stock, quality, preferred, memora- where Ws is standard weight in pounds and FL is ble, and trophy lengths be determined from lengths in inches. These equations are proposed for use encompassed by 20±26%, 36±41%, 45±55%, 59± with shovelnose sturgeon from 120 mm (5 in) to 64%, and 74±80% of world record length, respec- 1,050 mm (41 in).

tively. We used the largest ®sh in our data set Although Wr may vary among lengths in a given (1,052 mm FL; 8,164 g) as the world record length population, there should be no consistent pattern

because the largest shovelnose sturgeon recorded of increasing or decreasing Wr values for a series by the National Freshwater Fishing Hall of Fame of populations. When Wr was regressed on ®sh weighed 2,155 g. length for 32 populations, 20 populations had sig- Several authors have noted seasonal variations ni®cant slopes (P Յ 0.05). The number of positive in stock density indices (Carline et al. 1984; Serns slopes (11 slopes) was not signi®cantly different 1985; Willis et al. 1993). However, nearly all data than the number of negative slopes (9 slopes; P ϭ sets represented summer or early fall sampling, 0.41), indicating no length bias associated with the

which made separation by season impractical. Ws equation. Similarly, size-related biases can result from gear Forty-three percent of the populations had mean

selectivity (Hamley 1975; Reynolds and Simpson Wr values within the suggested benchmark target 1978). Relations between Wr and size structure range of 95±100 (Wege and Anderson 1978). Al- were determined for those populations known to though this is similar to those percentages reported be sampled with gill or trammel nets from riverine for white bass Morone chrysops (37%), palmetto

habitats. Mean Wr was plotted as a function of PSD bass (a hybrid of female striped bass Morone sax- and as a function of incremental Wr and RSDs [i.e., atilis ϫ male white bass; 32%; Brown and Murphy stock to quality (S±Q), quality to preferred (Q±P), 1991), and white crappie Pomoxis annualaris preferred to memorable (P±M), and memorable to (Neumann and Murphy 1991), 85% of these pop- trophy length (M±T)]. The relationships were an- ulations were from Montana. Sixty percent of the

alyzed with correlation techniques. Differences of Montana populations had mean population Wr val- Wr among length categories were tested with anal- ues between 95 and 105. Conversely, 66% of the Downloaded by [Department Of Fisheries] at 00:54 01 March 2012 ysis of variance (ANOVA; SAS 1989). A proba- shovelnose sturgeon populations from states other

bility level of 0.05 was used to reject the null than Montana had mean population Wr values be- hypothesis in all statistical tests. tween 80 and 90. These data suggest that universal target values may be inappropriate for shovelnose Results and Discussion sturgeon. Weight±length regressions were developed for Numerous researchers have described differ-

32 shovelnose sturgeon populations from nine ences in Wr values associated with lentic and lotic states (N ϭ 11,820; Table 1). We found that 30 populations (e.g., Fisher et al. 1996); however, we

populations were required to develop a Ws equa- are unaware of any studies identifying variation in tion with the RLP technique (sample variance of Wr along a longitudinal scale in lotic ecosystems. slopes ϭ 0.0018; Michael L. Brown, South Dakota The variability between lentic and lotic systems State University, personal communication). Shov- has prompted the development of differing target

elnose sturgeon 120 mm and longer were included ranges for Wr. For example, Fisher et al. (1996) in the weight±length regressions because the vari- found that Wr values for burbot Lota lota were ance-to-mean ratio was below 0.01 for all ®sh consistently lower in lotic ecosystems than in len- 994 QUIST ET AL.

TABLE 1.ÐSample size (N; ®sh Ն 120 mm), intercept (a), slope (b), and correlation coef®cient (r) for regressions of log10weight (g) on log10fork length (mm). Mean relative weight (Wr) values for shovelnose sturgeon length categories (stock to quality, S±Q; quality to preferred, Q±P; preferred to memorable, P±M; and memorable to trophy, M±T) are provided. Numbers in parentheses below the mean are sample size and SD of mean. All regressions were signi®cant at P Յ 0.0001.

Mean W values for length categories: State and r location NabrS±Q Q±P P±M M±T Illinois, Missouri Mississippi River, 30 Ϫ5.392 2.941 0.91 122 58 53 Pool 26 (15, 32.17) (6, 14.19) a Mississippi River, 72 Ϫ6.473 3.381 0.96 86 99 96 78 Cape Girardeau (15, 21.38) (23, 23.17) (25, 32.16) (8, 20.65) Iowa, Illinois Mississippi River, 149 Ϫ7.292 3.678 0.98 75 86 90 98 Pool 13 (5, 10.66) (50, 8.55) (80, 9.50) (14, 13.19) Kansas Kansas River, Lawrence 125 Ϫ5.793 3.119 0.99 91 83 84 76 (11, 10.97) (29, 5.94) (80, 6.88) a Kansas River, Ogden 233 Ϫ6.619 3.426 0.95 79 82 84 82 a (24, 6.81) (206, 6.81) a Kansas, Missouri Missouri River, 47 Ϫ5.972 3.174 0.99 78 80 78 79 St. Joseph (5, 7.29) (13, 7.42) (21, 11.93) a Missouri River, 39 Ϫ6.196 3.265 0.99 81 81 82 89 Kansas City a (8, 4.88) (23, 6.52) a Louisiana Red River 45 Ϫ4.925 2.824 0.96 151 106 91 91 a (8, 23.87) (27, 10.70) (9, 7.60) Montana Big Muddy River 357 Ϫ6.206 3.294 0.98 98 95 99 92 (14, 20.44) (47, 13.16) (212, 11.02) (74, 8.01) Marias River 54 Ϫ6.432 3.380 0.95 96 100 a (36, 12.20) Missouri River, Loma 1,160 Ϫ6.089 3.267 0.95 96 105 106 a (49, 16.00) (590, 12.33) Missouri River, 69 Ϫ6.246 3.323 0.99 126 103 98 108 Whiteside a (11, 8.61) (5, 3.56) (37, 12.82) Missouri River, Stafford 441 Ϫ6.540 3.423 0.98 102 95 97 103 Ferry a (27, 10.59) (51, 12.79) (256, 11.04) Missouri River, Robinson 1,922 Ϫ6.457 3.394 0.97 105 92 107 102 Bridge (16, 14.54) (85, 1.32) (599, 14.45) (1,004, 11.19) Missouri River, Fort Peck 732 Ϫ6.290 3.317 0.96 91 93 91 (11, 7.66) (454, 9.00) (262, 9.17) Missouri River, Milk River 168 Ϫ5.874 3.159 0.96 93 88 83 con¯uence (12, 5.81) (108, 7.52) (46, 7.95) Missouri River, Wolf Point 112 Ϫ6.084 3.229 0.97 89 84 82 (25, 9.28) (73, 8.35) (14, 8.63) Missouri River, Red Water 58 Ϫ6.041 3.229 0.98 82 89 91 Downloaded by [Department Of Fisheries] at 00:54 01 March 2012 River con¯uence a a (5, 8.11) Missouri and Yellowstone 782 Ϫ6.449 3.382 0.98 90 90 98 95 River con¯uence (15, 114.77) (61, 10.94) (484, 11.55) (188, 10.11) Powder River 70 Ϫ7.540 3.768 0.95 96 101 a (54, 8.92) Tongue River 272 Ϫ6.706 3.474 0.93 96 100 a (178, 10.59) Yellowstone River, 133 Ϫ6.736 3.478 0.98 110 90 91 95 Highway 23 bridge a (21, 11.99) (48, 11.81) (55, 10.49) Yellowstone River, Forsyth 187 Ϫ6.988 3.574 0.97 96 100 (12, 7.05) (101, 11.73) Yellowstone River, Miles 60 Ϫ8.211 4.000 0.97 93 96 City (8, 10.11) (36, 9.85) Yellowstone River, Powder 588 Ϫ7.189 3.643 0.96 91 93 91 99 River con¯uence a a (37, 9.35) (400, 11.07) Yellowstone River, Glendive 853 Ϫ6.821 3.509 0.98 89 93 90 95 (7, 7.74) (29, 14.64) (123, 7.67) (501, 11.23) MANAGEMENT BRIEFS 995

TABLE 1.ÐContinued.

Mean W values for length categories: State and r location NabrS±Q Q±P P±M M±T Yellowstone River, 565 Ϫ6.483 3.399 0.97 101 100 101 above Intake (6, 18.86) (60, 11.25) (334, 10.39) Yellowstone River, 1,158 Ϫ6.349 3.337 0.99 96 92 90 92 below Intake (29, 18.44) (95, 10.86) (337, 8.27) (582, 9.67) Nebraska Platte River 117 Ϫ6.001 3.203 0.94 88 87 90 (13, 12.09) (97, 7.99) (7, 3.83) North Dakota Lake Oahe 91 Ϫ4.579 2.669 0.88 78 69 (38, 9.41) (53, 8.56) Lake Sakakawea 45 Ϫ5.826 3.134 0.98 86 99 89 78 (15, 21.00) (23, 23.17) (12, 6.21) (21, 11.19) Wisconsin Chippewa River 1,085 Ϫ4.572 3.381 0.85 81 89 83 a (522, 8.09) (562, 7.84) a Represents less than ®ve individuals.

tic ecosystems. Consequently, they suggested tar- er than propose different target values. Similarly, get values of 95±105 for lentic populations and Willis et al. (1991) identi®ed a broad distribution 75±85 for lotic ecosystems. Kruse and Hubert in body condition of yellow perch Perca ¯avescens (1997) found similar relations with cutthroat trout and suggested that the variability in condition re- Oncorhynchus clarki but elected to develop sepa- inforces the concept that universal target values rate equations for lotic and lentic populations rath- may be inappropriate in many situations. There- fore, we suggest target values of 95±105 for Mon- tana populations and 80±90 for other populations of shovelnose sturgeon. We recommend the following length categories be used when calculating stock density indices for shovelnose sturgeon: stock, 250 mm (10 in); quality, 380 mm (15 in); preferred, 510 mm (20 in); memorable, 640 mm (25 in); and trophy, 810 mm (32 in). All populations had PSD values greater than 79, except for a Mississippi River population (Table 1). The number of populations with high PSD values is not surprising because shovelnose sturgeon are typically collected with large-mesh (250-mm-bar-measure) nets. The primary gears

Downloaded by [Department Of Fisheries] at 00:54 01 March 2012 used to sample shovelnose sturgeon for this study were either gill or trammel nets. Thus, stock-length to quality-length shovelnose sturgeon were less likely to be sampled. Stock density indices are used to quantify size structure of a ®sh population. It is debatable whether such indices re¯ect growth, mortality, and recruitment. However, Willis (1989) and Guy and Willis (1995) documented relations between PSD and growth. Gabelhouse (1984a) suggested that low PSD values may re¯ect poor habitat, over- harvest, and reduced food supply. The relation be- FIGURE 1.ÐRelationship between mean population rel- tween PSD and Wr against growth are probably ative weight of memorable length to trophy length (Wr M± T) and relative stock density of memorable length to tro- one of cause and effect (Willis 1989; Gabelhouse phy length (RSD M±T) for shovelnose sturgeon. 1991; Guy and Willis 1995). We were unable to 996 QUIST ET AL.

P±M and M±T ®sh (P Յ 0.05; Figure 2). These

data illustrate the importance of size-speci®c Wr analyses.

Although the relation between PSD and Wr probably will vary with growth, we surmise that these indices will be useful in assessing shovelnose sturgeon size structure and condition. At best, these indices will provide information on growth, mortality, and recruitment of shovelnose sturgeon in large-river ecosystems. We encourage ®sheries bi-

ologists to develop Ws equations and size structure categories for large-river species so that these tools will be available for condition and size structure assessment.

Acknowledgments We thank the following individuals for contrib- uting the data sets used in this project: Kenneth Backes, Mel Bowler, Lyle Christensen, Randy Claramunt, David Day, Bill Gardner, Doug Hen- ley, Robin Hofpar, Edward Peters, Greg Power, Bobby Reed, Mike Ruggles, Anne Tews, and Gary Tilyou. In addition, we thank Mike Brown for statistical assistance. The U.S. Army, Department of Defense and Kansas State University, Division of FIGURE 2.ÐRelations between mean relative weight Biology, provided funding for this project. (Wr) by length categoryÐstock to quality (S±Q), quality to preferred (Q±P), preferred to memorable (P±M), and References memorable to trophy (M±T)Ðfor shovelnose sturgeon. Bars represent ϮSE. Data points without a letter in com- Brown, M. L., and B. R. Murphy. 1991. Standard mon represent signi®cant differences (␣ϭ0.05) among weights (Ws) for striped bass, white bass, and hybrid length categories. striped bass. North American Journal of Fisheries Management 11:451±467. Brown, M. L., and B. R. Murphy. 1996. Selection of a obtain growth data for shovelnose sturgeon used minimum sample size for application of the regres- in the development of the Ws equation. Thus, we sion-line-percentile technique. North American used relations between Wr and PSD to evaluate the Journal of Fisheries Management 16:427±432. utility of these management tools for large-river Carline, R. F., B. L. Johnson, and T. J. Hall. 1984. Es- species. timation and interpretation of proportional stock Mean population W values were positively cor- density for ®sh populations in Ohio impoundments. r North American Journal of Fisheries Management related (r ϭ 0.47, P ϭ 0.033) with PSD with a 4:139±154. Downloaded by [Department Of Fisheries] at 00:54 01 March 2012 curvilinear regression model. Thus, populations Fisher, S. J., D. W. Willis, and K. L. Pope. 1996. An with higher PSD values had higher mean condition assessment of burbot (Lota lota) weight±length data values. Similar results have been reported for other from North American populations. Canadian Jour- species, such as northern pike Esox lucius (Willis nal of Zoology 74:570±575. and Scalet 1989), crappies Pomoxis spp. (Gabel- Gabelhouse, D. W., Jr. 1984a. A length-categorization system to assess ®sh stocks. North American Jour- house 1984b; Guy and Willis 1995), and brook nal of Fisheries Management 4:273±285. trout Salvelinus fontinalis (Johnson et al. 1992). Gabelhouse, D. W., Jr. 1984b. An assessment of crappie We also plotted relations between incremental stocks in small midwestern private impoundments. RSD and Wr. The best relationship occurred be- North American Journal of Fisheries Management 4:371±384. tween Wr of M±T and RSD of M±T (r ϭ 0.68, P ϭ 0.0001; Figure 1). In addition, we found dif- Gabelhouse, D. W., Jr. 1991. Seasonal changes in body condition of white crappie and relations to growth ferences among incremental W values. Mean W r r in Melvern Reservoir, Kansas. North American generally decreased as ®sh attained greater Journal of Fisheries Management 11:50±56. lengths. Stock-length to quality-length shovelnose Guy, C. S., and D. W. Willis. 1995. Population char- sturgeon had signi®cantly greater Wr values than acteristics of black crappies in South Dakota waters: MANAGEMENT BRIEFS 997

a case for ecosystem-speci®c management. North Pages 11±25 in G. D. Novinger and J. G. Dillard, American Journal of Fisheries Management 15: editors. New approaches to the management of 654±765. small impoundments. American Fisheries Society, Hamley, J. M. 1975. Review of gill net selectivity. Jour- North Central Division, Special Publication 5, Be- nal of the Fisheries Research Board of Canada 32: thesda, Maryland. 1943±1969. SAS Institute. 1989. SAS procedures guide for personal Johnson, S. L., F. J. Rahel, and W. A. Hubert. 1992. computers, version 6.03. SAS Institute, Cary, North Factors in¯uencing the size structure of brook trout Carolina. populations in beaver ponds in Wyoming. North Serns, S. L. 1985. Proportional stock density indexÐis American Journal of Fisheries Management 12: it a useful tool for assessing ®sh populations in 118±124. northern latitudes? Wisconsin Department of Nat- Kruse, C. G., and W. A. Hubert. 1997. Proposed stan- ural Resources, Research Report 132, Madison.

dard weight (Ws) equations for interior cutthroat Wege, G. J., and R. O. Anderson. 1978. Relative weight trout. North American Journal of Fisheries Man- (Wr): a new index of condition for largemouth bass. agement 17:784±790. Pages 79±91 in G. D. Novinger and J. G. Dillard, Lee, D. S., and ®ve coauthors. 1980. Atlas of North editors. New approaches to the management of American freshwater ®shes. North Carolina Muse- small impoundments. American Fisheries Society, um of Natural History, Raleigh. North Central Division, Special Publication 5, Be- Mehta, C., and N. Patel. 1995. StatXact 3 for Windows, thesda, Maryland. statistical software for exact nonparametric infer- Willis, D. W. 1989. A proposed standard length±weight ence, user manual. CYTEL Software, Cambridge, equation for northern pike. North American Journal Massachusetts. of Fisheries Management 9:203±208. Murphy, B. R., M. L. Brown, and T. A. Springer. 1990. Willis, D. W., C. S. Guy, and B. R. Murphy. 1991. De-

Evaluation of the relative weight index, with new velopment and evaluation of a standard weight (Ws) applications to walleye. North American Journal of equation for yellow perch. North American Journal Fisheries Management 10:85±97. of Fisheries Management 11:374±380. Murphy, B. R., D. W. Willis, and T. A. Springer. 1991. Willis, D. W., B. R. Murphy, and C. S. Guy. 1993. Stock The relative weight index in ®sheries management: density indices: development, use, and limitations. status and needs. Fisheries 16(2):30±38. Reviews in Fisheries Science 1:203±222. Neumann, R. M., and B. R. Murphy. 1991. Evaluation Willis, D. W., and C. G. Scalet. 1989. Relations between

of the relative weight (Wr) index for assessment of proportional stock density and growth and condition white crappie and black crappie populations. North of northern pike populations. North American Jour- American Journal of Fisheries Management 11: nal of Fisheries Management 9:488±492. 543±555. Reynolds, J. B., and D. E. Simpson. 1978. Evaluation Received January 19, 1998 of ®sh sampling methods and rotenone census. Accepted April 21, 1998 Downloaded by [Department Of Fisheries] at 00:54 01 March 2012