This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 19:58 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Dam Removal Increases American Eel Abundance in Distant Headwater Streams Nathaniel P. Hitt a , Sheila Eyler b & John E. B. Wofford c a U.S. Geological Survey, Leetown Science Center, Aquatic Ecology Branch, 11649 Leetown Road, Kearneysville, West Virginia, 25430, USA b U.S. Fish and Wildlife Service, Maryland Fishery Resources Office, 177 Admiral Cochrane Drive, Annapolis, Maryland, 21401, USA c National Park Service, Shenandoah National Park, 3655 Highway 211 East, Luray, Virginia, 22835, USA Version of record first published: 20 Jul 2012.
To cite this article: Nathaniel P. Hitt, Sheila Eyler & John E. B. Wofford (2012): Dam Removal Increases American Eel Abundance in Distant Headwater Streams, Transactions of the American Fisheries Society, 141:5, 1171-1179 To link to this article: http://dx.doi.org/10.1080/00028487.2012.675918
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. Transactions of the American Fisheries Society 141:1171–1179, 2012 American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.675918
ARTICLE
Dam Removal Increases American Eel Abundance in Distant Headwater Streams
Nathaniel P. Hitt* U.S. Geological Survey, Leetown Science Center, Aquatic Ecology Branch, 11649 Leetown Road, Kearneysville, West Virginia 25430, USA Sheila Eyler U.S. Fish and Wildlife Service, Maryland Fishery Resources Office, 177 Admiral Cochrane Drive, Annapolis, Maryland 21401, USA John E. B. Wofford National Park Service, Shenandoah National Park, 3655 Highway 211 East, Luray, Virginia 22835, USA
Abstract American eel Anguilla rostrata abundances have undergone significant declines over the last 50 years, and migra- tion barriers have been recognized as a contributing cause. We evaluated eel abundances in headwater streams of Shenandoah National Park, Virginia, to compare sites before and after the removal of a large downstream dam in 2004 (Embrey Dam, Rappahannock River). Eel abundances in headwater streams increased significantly after the removal of Embrey Dam. Observed eel abundances after dam removal exceeded predictions derived from autoregres- sive models parameterized with data prior to dam removal. Mann–Kendall analyses also revealed consistent increases in eel abundances from 2004 to 2010 but inconsistent temporal trends before dam removal. Increasing eel numbers could not be attributed to changes in local physical habitat (i.e., mean stream depth or substrate size) or regional population dynamics (i.e., abundances in Maryland streams or Virginia estuaries). Dam removal was associated with decreasing minimum eel lengths in headwater streams, suggesting that the dam previously impeded migration of many small-bodied individuals (<300 mm TL). We hypothesize that restoring connectivity to headwater streams could increase eel population growth rates by increasing female eel numbers and fecundity. This study demonstrated that dams may influence eel abundances in headwater streams up to 150 river kilometers distant, and that dam removal may provide benefits for eel management and conservation at the landscape scale. Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 American eels Anguilla rostrata exhibit complex life his- cluding new initiatives to improve fish passage (ASMFC 2000). tory strategies characterized by long-distance movements be- Dams were also recognized as a cause for recognizing American tween marine habitats for spawning and freshwater habitats for eel as a species of special concern in Canada (COSEWIC 2006) growth and development (Oliveira 1999). Historically, Ameri- and for the U.S. Fish and Wildlife Service’s recent decision to can eels were widespread throughout the rivers and estuaries of evaluate listing American eels as a threatened species under the North America’s Atlantic coast, but the construction of dams has Endangered Species Act (USFWS 2011). significantly reduced the amount of accessible habitat for diadro- Although American eels are capable of passing some signifi- mous fishes such as eels (Busch et al. 1998). Significant declines cant natural barriers (e.g., the Great Falls of the Potomac River, in American eel abundances (Haro et al. 2000; Fenske et al. which has several consecutive falls >6 m), dams may limit the 2011) have triggered new efforts for fishery management, in- upstream movement of eels such that eel numbers often decrease
*Corresponding author: [email protected] Received October 17, 2011; accepted February 26, 2012 1171 1172 HITT ET AL.
above dams (Goodwin et al. 1999; Machut et al. 2007) and in- of Engineers breached the dam. The dam removal was the re- crease immediately below dams (Wiley et al. 2004; Machut sult of many years of work by nonprofit organizations and city, et al. 2007). Consequently, barriers may influence stream com- state, and federal government agencies. The dam removal was munity composition and population dynamics in upstream and intended to benefit anadromous clupeids (e.g., American shad downstream directions. Upstream of dams, decreased eel den- Alosa sapidissima) and striped bass Morone saxatilis as well sities may influence stream fish communities by removing a as catadromous American eels (A. Weaver, Virginia Depart- native piscivore which could otherwise comprise over 25% of ment of Game and Inland Fisheries, personal communication). the total fish biomass in streams (Smith and Saunders 1955; In addition to Embrey Dam, a small dam on the Thornton River Ogden 1970). Freshwater mussel distributions may also be lim- (Fletcher’s Mill Dam, ∼1 m high) near the boundary of Shenan- ited through restrictions of the fish host movements that are doah National Park (SNP) was removed in 2009 to promote necessary for upstream dispersal of mussel glochidia (Williams fish passage; however, it was not considered further because et al. 1993; Watters 1996). Downstream of dams, increased eel preliminary analyses indicated no significant differences in eel densities may increase intraspecific competition and decrease abundance between the Thornton watershed and other focal wa- per capita growth rates (Machut et al. 2007). Reduced access tersheds. We evaluated fish community and physical habitat data to headwater streams may also influence eel stock–recruitment from headwater streams in SNP (Figure 1) located between 118 dynamics by decreasing the production of female eels (Krueger and 150 rkm upstream from the former location of Embrey Dam and Oliveira 1999). (Table 1). Dam removal has proven effective for restoring historical up- Eel population analysis.—National Park Service personnel stream migrations of diadromous salmonid and clupeid fishes sampled fish communities in 117 wadeable stream sites within (Hill et al. 1996; Kiffney et al. 2009), but comparatively little is SNP annually from 1996 to 2010 (rarely excluding years; see known about American eel responses to dam removal. On one Table 1). Of these sites, 32 supported American eels during at hand, American eels have been observed upstream of dams that least one sampling event. We limited our analysis to 15 sites that are known to limit other migratory fishes (Busch et al. 1998), had >7 annual collections, including samples before and after suggesting that dam removal is relatively unimportant for eel 2004 (Figure 1; Table 1). Each site was delimited within stan- distributions. On the other hand, decreased eel abundances up- dardized 100-m reaches, and fish communities were sampled stream from dams (Goodwin et al. 1999; Machut et al. 2007) using standard three-pass backpack electrofishing techniques. suggest that dams permit only a subset of the total migratory Individual eel abundances were recorded for each pass, and the population to move upstream. If true, partial barriers to migra- pooled weight of all eels and the minimum individual length tion could affect eel populations by influencing sex ratios and (TL) per site were recorded (Atkinson 2002). We estimated eel fecundity. An understanding of the effects of dams could there- fore inform conservation and restoration priorities for American eels. TABLE 1. Attributes of sample sites within Shenandoah National Park. The In this study, we used a 15-year data set to evaluate how locations of the sites are shown in Figure 1; the watersheds correspond to those American eel populations in headwater streams responded to in Figures 2–4. the removal a large downstream dam on the Rappahannock River in Virginia. Our objectives were twofold. First, we evalu- Site Fluvial Number of sample ated temporal trends in American eel abundance, biomass, and elevation distance to years before and body size before and after dam removal. Second, we evaluated Site Watershed (m) dam (km) after dam removal evidence for competing hypotheses involving changes in local 1F003 Thornton 362 118 8, 4 Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 physical habitats and population dynamics across larger spatial 1F030 Thornton 352 119 7, 4 scales. Our study provides the first analysis of American eel 1F145 Thornton 415 120 5, 3 > responses to dam removal at the stream network scale ( 100 1FVA2 Thornton 382 120 7, 4 river kilometers [rkm]). 1FVA3 Thornton 428 122 7, 4 2F015 Rose 340 125 5, 4 2F016 Rose 414 127 8, 4 METHODS 2F017 Rose 642 129 8, 4 Study area.—Embrey Dam was located near the fall line 2F038 Hughes 293 121 8, 4 on the Rappahannock River in Virginia, (Figure 1). Down- 2F039 Hughes 370 122 8, 4 stream of the dam site, the river is influenced by tidal flows 2F040 Hughes 402 123 8, 4 over the course of its 170-rkm distance to the Chesapeake Bay 2F072 Rapidan 316 146 6, 5 (Figure 1). The dam spanned a width of 235 m and a height of 2F093 Rapidan 285 145 8, 7 6.7 m and was constructed in 1910 for hydroelectric produc- 2F135 Rapidan 412 148 8, 5 tion and municipal water supply, replacing a dam built in 1855 2FVA4 Rapidan 507 150 7, 7 (Feeney 2004). On February 23, 2004, the U.S. Army Corps DAM REMOVAL BENEFITS AMERICAN EELS 1173 Downloaded by [Department Of Fisheries] at 19:58 25 September 2012
FIGURE 1. American eel distribution within Shenandoah National Park. The regional map indicates the former location of Embrey Dam on the Rappahannock River (circle), the locations of Maryland control stream sites (squares), and the Rappahannock River watershed (cross hatches). Study estuaries are indicated for the Rappahannock River (A), York River (B), and James River (C). Fluvial distances from the Embrey Dam site to Shenandoah National Park sample sites are listed in Table 1. 1174 HITT ET AL.
abundances within sampling reaches as the sum of eel counts spaced points along 11 equidistant lateral transects within the across passes and combined site-level data into four focal wa- 100 m reach. At each sample point, stream depth was recorded to tersheds for analysis (Table 1). Each of the 15 SNP focal sites in the nearest millimeter and the dominant substrate was recorded this analysis was located upstream from the Embrey Dam site. as silt, sand, gravel, cobble, boulder, or bedrock (Wentworth We used time series analysis and nonparametric and para- 1922). To assess substrate size trends, substrate types were nu- metric statistical tests to evaluate the effects of dam removal merically coded (i.e., silt = 0, sand = 1, etc.) to calculate mean on eel abundances in headwater streams. First, we used au- conditions (Bain and Stevenson 1999). American eels are typ- toregressive integrated moving average (ARIMA) techniques ically associated with pools in lotic environments (Jenkins and (Box et al. 2008) to derive a null model for expected eel abun- Burkhead 1994), and we assumed that changes in pool habi- dances in the absence of dam removal and to evaluate the signifi- tat would be reflected by changes in mean stream depth and cance of observed changes in eel abundances after dam removal. substrate size over time. The ARIMA techniques were useful because preliminary analy- Second, we evaluated eel abundances in additional streams ses revealed potential autocorrelation in eel abundances among of the Chesapeake Bay region to control for the effects of dam years and ARIMA models incorporate such temporal autocorre- removal in the SNP sites. We reasoned that if oceanic-scale lation to forecast mean and variance of estimates (Zhang 2003). processes were influencing headwater eel numbers (i.e., mass We parameterized the null model using eel abundance data from effects; sensu Shmida and Wilson 1985), eel numbers would ex- 1996 to 2003 (i.e., before dam removal) to forecast abundances hibit similar trends outside the Rappahannock River watershed. from 2004 to 2010 (i.e., after dam removal). Best-fitting ARIMA We examined eel time series data from the Maryland Biolog- model parameters (number of autoregressive terms, number of ical Stream Survey (MBSS). The MBSS fish community data nonseasonal differences, and the number of lagged forecast er- were collected by Maryland Department of Natural Resources rors) were selected from the function “auto.arima” in the R personnel annually from 2000 to 2010. Stream sites were sam- library “forecast.” We inferred the effects of dam removal based pled using two-pass backpack electrofishing techniques during on the departure of observed eel abundances after dam removal summer base-flow conditions within blocknetted 75-m sample from the 95% confidence intervals of the null model predic- reaches (MDNR 2010). We evaluated five sites that contained eel tions. Our analysis of predicted confidence intervals provided a records and were not separated from the ocean by dams. Sites method to estimate the significance of temporal changes without were located within watersheds of the lower Potomac River, bias due to the nonindependence of residuals common to linear Pocomoke River, and Patuxent River in the southwestern por- modeling techniques (Box et al. 2008). Koutroumanidis et al. tion of the Chesapeake Bay (Figure 1). Mean stream widths (2006) used similar methods for analysis of fisheries catch rates. of the selected MBSS stream sites ranged from 1.7 to 6.6 m Second, we used Mann–Kendall analysis (Mann 1945; (average = 3.8 m) and were located within 10 rkm of the tidal- Kendall 1975) to evaluate temporal trends in eel abundances influence zone (Figure 1). among sites within three time periods: before dam removal Third, we evaluated eel abundances within estuaries of the (1996–2003), after dam removal (2004–2010), and within the Rappahannock River, York River, and James River (Figure 1) to entire period of record. The Mann–Kendall statistics provided a understand whether or not the Embrey Dam removal coincided nonparametric analysis of increasing and decreasing eel abun- with unusually high or low rates of recruitment from marine dances and ranged from −1 (decreasing trends) to + 1 (increas- areas (i.e., mass effects). Estuary data were collected by the Vir- ing trends). We reported Mann–Kendall P-values as an index ginia Institute of Marine Science using a trawl survey designed of the relative strength of temporal trends but did not interpret to estimate the abundance of juvenile fish in the Virginia portion significance based on a critical α level because Mann–Kendall of the Chesapeake Bay (Tuckey and Fabrizio 2010). The trawl
Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 P-values are biased by serial autocorrelation (Yue and Wang surveys sampled eel abundances within estuaries of the Rap- 2004). Instead, we reasoned that sites would tend to exhibit a pahannock River, York River, and James River during spring random distribution of increasing and decreasing abundances months (April to June), and we evaluated annual data collected prior to dam removal but would shift to increasing abundances between 1996 and 2010. Sampling was conducted monthly at after 2004 if dam removal increased colonization rates. We also both fixed and randomly selected stations within each estuary. plotted average minimum eel lengths and pooled biomass among The index for American eels is an annual weighted geometric focal watersheds over time and estimated differences in pre- and mean catch per tow of all eels greater than 152 mm TL (Tuckey postdam mean conditions using t-tests. and Fabrizio 2010). All analyses were conducted in R version Alternative hypotheses.—We considered local physical habi- 2.13.1 (R Development Core Team 2011). tat and regional population dynamics as alternatives to dam re- moval to explain temporal changes in eel abundance, size, and biomass. First, we quantified the interannual variation in sub- RESULTS strate size and stream depth within SNP sample sites as possible Mean American eel abundances within SNP watersheds in- confounding factors from dam removal. National Park Service creased from 1.6 to 3.9 eels/100 m after the removal of Em- personnel collected physical habitat samples at three evenly brey Dam in 2004 (Table 2). Postdam eel abundances exceeded DAM REMOVAL BENEFITS AMERICAN EELS 1175
TABLE 2. Eel population attributes and environmental conditions in Shenandoah National Park watersheds pre- and postremoval of Embrey Dam. Values are means, with SDs in parentheses. Differences between pre- and postdam means are indicated by different lowercase letters (t-tests assuming unequal variance) using a Bonferroni correction for α = 0.05/5 = 0.01 (t =−2.79, P = 0.006). Sample sizes are listed in Table 1.
Eel abundance / Minimum total Pooled eel Mean stream Mean substrate 100 m length (mm) biomass (g) depth (m) size-class Watershed Pre Post Pre Post Pre Post Pre Post Pre Post Hughes 2.4 3.8 389 313 522 707 0.17 0.16 3.1 3.0 (1.3) (2.9) (96) (168) (321) (400) (0.02) (0.03) (0.4) (0.4) Rapidan 1.4 4.7 452 226 783 819 0.18 0.21 3.2 3.1 (0.7) (4.4) (52) (136) (282) (461) (0.02) (0.02) (0.3) (0.1) Rose 0.6 2.3 545 323 725 335 0.13 0.18 3.1 3.0 (0.7) (1.7) (104) (160) (222) (293) (0.02) (0.03) (0.3) (0.2) Thornton 2.0 4.2 368 234 538 520 0.11 0.10 2.9 2.5 (0.6) (3.6) (40) (78) (255) (205) (0.02) (0.01) (0.4) (0.3) All 1.6 y 3.9 z 426 269 628 620 0.15 0.17 3.0 2.9 (1.6) (5.0) (96) (135) (287) (388) (0.04) (0.05) (0.3) (0.4)
ARIMA null model predictions for all focal watersheds Headwater streams generally supported smaller eels after (Figure 2) and exhibited a time-lag response to dam removal: dam removal than before dam removal (Figure 3; Table 2). observed abundances exceeded predicted values (>95% confi- Prior to dam removal, average minimum eel lengths ranged dence intervals) within 4 years after dam removal in the Hughes, from 545 mm (Rose River watershed) to 368 mm (Thornton Rapidan, and Thornton River watersheds and within 2 years in River watershed); after dam removal, the range of average min- the Rose River watershed (Figure 2). Mann–Kendall analysis imum total lengths dropped to between 323 mm (Rose River supported the ARIMA model results, indicating nine sites (60%) watershed) and 226 mm (Rapidan River watershed) (Table 2). with decreasing abundance trends prior to dam removal (i.e., Moreover, no eels less than 300 mm TL were detected in any τ<0) but all sites with increasing abundance trends after dam SNP watershed before 2004, but eels of that length were present removal (τ>0) (Table 3). Analysis of the combined data set in each watershed after dam removal (Figure 3). Average total (1996–2010) showed 13 sites with increasing trends and 2 sites eel biomass decreased on average from 401 g to 159 g after dam with decreasing trends in eel abundance (Table 3). removal (Table 2) but exhibited substantial spatial and temporal
TABLE 3. Mann–Kendall τ-statistics for time series analysis of American eel abundances.
Before dam removal After dam removal Whole data set Site τ P τ P τ P 1F003 −0.189 0.612 0.667 0.308 0.469 0.045
Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 1F030 −0.150 0.759 0.667 0.308 0.135 0.633 1F145 −0.738 0.130 0.816 1.000 −0.433 0.195 1FVA2 1.000 1.000 0.548 0.470 0.526 0.061 1FVA3 0.265 0.525 0.183 1.000 0.060 0.871 2F015 0.316 0.613 0.913 0.149 0.509 0.085 2F016 0.504 0.148 0.548 0.470 0.627 0.011 2F017 1.000 1.000 0.707 0.371 0.408 0.148 2F038 −0.390 0.272 0.913 0.149 0.116 0.670 2F039 −0.222 0.530 0.548 0.470 −0.032 0.944 2F040 0.197 0.605 0.548 0.470 0.201 0.430 2F072 1.000 1.000 0.632 0.289 0.426 0.155 2F093 −0.591 0.070 0.781 0.023 0.217 0.308 2F135 −0.321 0.385 0.800 0.086 0.530 0.019 2FVA4 −0.233 0.610 0.476 0.204 0.198 0.403 1176 HITT ET AL.
FIGURE 2. Interannual variation in American eel abundance within Shenan- doah National Park watersheds. Solid lines show the average observed abun- dances within focal watersheds. Black dashed lines indicate the autoregressive integrated moving average (ARIMA) model predictions for 2004–2010 (param- eterized from 1996 to 2003 data; see text). Gray dashed lines indicate the upper and lower 95% confidence limits for mean predicted abundances. Sites within watersheds are listed in Table 1.
variation (Figure 3). Although individual length data were not FIGURE 3. Interannual variation in American eel minimum total length (solid available, the minimum length, total number of eels collected, line, left axis) and pooled biomass (dashed line, right axis) within Shenandoah and biomass data indicate that eel abundances increased due National Park watersheds. Sites within watersheds are listed in Table 1. primarily to the immigration of eels <300 mm TL. Mean depth and substrate size-classes were highly variable
Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 across SNP watersheds and exhibited no significant differences ical habitat changes could explain the observed increases in eel between pre- and postdam conditions (Table 2; Figure 4). Across abundance over time. SNP watersheds, mean depths ranged from 0.04 to 0.20 m and American eel abundances within Maryland streams and Vir- showed inconsistent temporal patterns (Figure 4). For instance, ginia estuaries exhibited no distinct changes coincident with 2004 yielded some of the lowest mean depths in the Hughes dam removal on the Rappahannock River (Figure 5). Mean eel and Thornton River watersheds but the highest in the Rapidan abundances in Maryland streams ranged from 3.2 to 22.0 indi- River watershed (Figure 4). Mean depths in the Rose River wa- viduals/75 m between 2000 and 2010 and exhibited no consis- tershed showed an increasing trend (Figure 4) but increased by tent increases after 2004 (Figure 5A). In contrast, estuarine eel only 0.05 m on average after dam removal (Table 2). Among abundances generally decreased over time in the Rappahannock watersheds, mean substrate size ranged from approximately 2.5 River (Figure 5B) as well as in the York River (Figure 5C) and to 3.5 across years, suggesting substrate fluctuations around James River (Figure 5D). It is therefore unlikely that oceanic- cobble-dominated systems (cobble = 3; Figure 4). Pre- ver- scale dynamics could explain the observed population increases sus postdam comparisons of mean substrate size-class within in the Rappahannock River tributaries. Instead, we observed in- watersheds indicated that substrate size has not changed in a creasing eel numbers in headwater streams despite decreasing systematic direction (Table 2). It is therefore unlikely that phys- regional trends. DAM REMOVAL BENEFITS AMERICAN EELS 1177
FIGURE 5. Interannual variation in American eel abundances within (A) Maryland nontidal wadeable streams and the estuaries of (B) the Rappahannock River, (C) the York River, and (D) the James River, Virginia. Vertical dotted lines indicate the year of dam removal on the Rappahannock River. In panel (A), the horizontal dashed line indicates 2 SDs from the mean abundances (solid line).
dant species in a New Jersey stream, comprising 20% of all observed fishes (and 37% of biomass, second only to white suckers Catostomus commersonii, at 47% of total biomass). In contrast, eel numbers in the SNP study sites never exceeded 2% FIGURE 4. Interannual variation in mean stream depth (solid line, left axis) and substrate size-class (dashed line, right axis) within watersheds used for of the total catch because fish assemblages were numerically American eel analysis in Shenandoah National Park. See text for substrate class dominated by eastern blacknose dace Rhinichthys atratulus (up definitions. to 52% of the total catch) and brook trout Salvelinus fontinalis (up to 77% of the total catch) (J. E. B. Wofford, National Park
Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 DISCUSSION Service, unpublished data). Thus, we would expect eel abun- Our study provides new inferences regarding the landscape- dances to increase over the near term in SNP streams without level effects of dam removal. Prior studies have shown localized limitations due to intraspecific competition for food or micro- effects of dam removal on fish populations, but our analysis is habitats. Moreover, measures of regional connectivity are gen- the first to our knowledge to demonstrate such influences on fish erally more powerful than local physical habitat variables for populations far upstream (i.e., 150 rkm). Our results also show modeling anguillid distributions and abundance (Smogor et al. that the immigration of small-bodied individuals (<300 mm TL) 1995; Domingos et al. 2006). was primarily responsible for the observed increases in eel num- Increasing eel abundances may influence stream fish commu- bers. Although Embrey Dam did not prevent eel passage, our re- nities by altering predation and competition pressures. Although sults indicate that it depressed eel abundances and altered eel size brook trout are currently the dominant piscivore in most SNP structure within connected headwater catchments. The benefits streams, fish typically comprise a relatively small portion of of dam removal may therefore extend far into headwater areas. lotic brook trout diets (Reed and Bear 1966). As a result, in- Although SNP sites showed increasing eel numbers over creasing eel numbers could affect the predation rates on benthic time, total eel abundances remained relatively low. For instance, fishes, which comprise the majority of American eel fish diets Ogden (1970) reported that American eels were the most abun- (Ogden 1970). Such increased predation on benthic fishes may 1178 HITT ET AL.
influence the top-down regulation of stream food webs (Power this paper. This research was supported by Shenandoah National et al. 1985), and thus migration barriers which reduced Ameri- Park, the National Park Service’s Inventory and Monitoring can eel numbers could have ecosystem-level consequences (e.g., Program, and the U.S. Geological Survey. Use of trade, product, Pringle 1997). However, such effects would take several years or firm names is for descriptive purposes only and does not imply to observe because American eels typically shift from inverte- endorsement by the U.S. Government. brates to fish and crayfish diets at approximately 400 mm TL (Ogden 1970; Lookabaugh and Angermeier 1992) and small- bodied eels were primarily responsible for the increased abun- REFERENCES dances we observed. Moreover, because American eels spend ASMFC (Atlantic States Marine Fisheries Commission). 2000. Interstate fishery several years in freshwater habitats before their spawning out- management plan for American eel (Anguilla rostrata). ASMFC, Fisheries migration (i.e., 6–21 years; Jessop 2010), additional sampling Management Report 36, Washington, D.C. Atkinson, J. B. 2002. Shenandoah National Park fisheries: monitoring protocol. will be necessary to assess fish community responses to chang- Shenandoah National Park, Luray, Virginia. ing eel abundances. Bain, M. B., and N. J. Stevenson, editors. 1999. Aquatic habitat assessment: Increasing the headwater stream abundances of American common methods. American Fisheries Society, Bethesda, Maryland. eels could affect regional population dynamics because head- Barbin, G. P., and J. D. McCleave. 1997. Fecundity of the American eel Anguilla ◦ water reaches provide vital habitats for the growth and devel- rostrata at 45 N in Maine, U.S.A. Journal of Fish Biology 51:840–847. Bednarek, A. T. 2001. Undamming rivers: a review of the ecological impacts of opment of female eels. First, access to headwater streams could dam removal. Environmental Management 27:803–814. increase per capita fecundity because American eel body sizes Box, G. E. P., G. M. Jenkins, and G. C. Reinsel. 2008. Time series analysis: typically increase with distance from the ocean (Lookabaugh forecasting and control, 4th edition. Wiley, New York. and Angermeier 1992; Smogor et al. 1995) and eel fecundity Busch, W. D. N., S. J. Lary, and C. M. Castiglione. 1998. Evaluating stream increases with body size (Barbin and McCleave 1997). Second, habitat for diadromous fish in Atlantic coast watersheds: a preliminary as- sessment. Habitat Hotline Atlantic (November):1–3. only female American eels are typically observed in headwa- Catalano, M. J., M. A. Bozek, and T. D. Pellett. 2007. Effects of dam re- ter streams (Goodwin and Angermeier 2003), and so the rela- moval on fish assemblage structure and spatial distributions in the Baraboo tive abundance of females could increase if restored headwater River, Wisconsin. North American Journal of Fisheries Management 27: connectivity reduced the downstream crowding associated with 519–530. high abundances of male fish (Krueger and Oliveira 1999). Con- COSEWIC (Committee on the Status of Endangered Wildlife in Canada). 2006. Assessment and status report on the American eel Anguilla rostrata in Canada. servation and restoration efforts for American eels could there- COSEWIC, Ottawa. fore benefit by considering headwater connectivity as a possible Domingos, I., J. L. Costa, and M. J. Costa. 2006. Factors determining length mechanism by which to increase eel numbers throughout their distribution and abundance of the European eel, Anguilla anguilla, in the range. River Mondego (Portugal). Freshwater Biology 51:2265–2281. Dam removal presents several ecological trade-offs for con- Fausch, K. D., B. E. Rieman, J. B. Dunham, M. K. Young, and D. P. Peterson. 2009. Invasion versus isolation: trade-offs in managing native salmonids with sideration in fisheries management. Over the short term, dam barriers to upstream movement. Conservation Biology 23:859–870. removal may increase downstream sedimentation and decrease Feeney, B. 2004. Embrey Dam removal opens 100s of miles of river to fish. water quality, but fish populations and communities may benefit Chesapeake Bay Journal 14(2):5. from increased abundance and resilience with restored stream Fenske, K. H., M. J. Wilberg, D. H. Secor, and M. C. Fabrizio. 2011. An age- network connectivity (Bednarek 2001; Hart et al. 2002; Stanley and sex-structured assessment model of American eels (Anguilla rostrata) in the Potomac River, Maryland. Canadian Journal of Fisheries and Aquatic and Doyle 2003). In some cases, barriers may be used as a man- Sciences 68:1024–1037. agement tool to prevent the immigration of undesirable species Goodwin, K. R., and P. L. Angermeier. 2003. Demographic characteristics of (Fausch et al. 2009). Although American eels are well known American eel in the Potomac River drainage, Virginia. Transactions of the
Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 for their long-distance catadromous migrations, barrier removal American Fisheries Society 132:524–535. could also benefit nondiadromous freshwater fishes by permit- Goodwin, K. R., P. L. Angermeier, and D. J. Orth. 1999. Assessing impacts of hydropower dams on upstream migration of American eel. Final Report to ting fish movement and recolonization within stream networks the Virginia Department of Game and Inland Fisheries, Richmond, Virginia. (Winston et al. 1991; Catalano et al. 2007; Hitt and Angermeier Haro, A., W. Richkus, K. Whalen, A. Hoar, W. D. Busch, S. Lary, T. Brush, 2008, 2011). Dams are ubiquitous in river systems worldwide and D. Dixon. 2000. Population decline of the American eel: implications for (Poff and Hart 2002), but the rate of dam removal is increas- research and management. Fisheries 25(9):7–16. ing through time (Stanley and Doyle 2003) and our analysis Hart, D. D., T. E. Johnson, K. L. Bushaw-Newton, R. J. Horowitz, A. T. Bednarek, D. F. Charles, D. A. Kreeger, and D. J. Velinsky. 2002. Dam re- suggests that dam removal confers ecological benefits for fish moval: challenges and opportunities for ecological research and river restora- conservation and management across large spatial scales. tion. BioScience 52:669–681. Hill, M. J., E. A. Long, and S. Hardin. 1996. Effects of a dam removal on Dead Lake, Chipola River, Florida. Proceedings of the Annual Con- ACKNOWLEDGMENTS ference Southeastern Association of Fish and Wildlife Agencies 48(1994): 512–523. We thank T. Tuckey, K. Whiteford, M. Kashiwagi, A. Weaver, Hitt, N. P., and P. L. Angermeier. 2008. Evidence for fish dispersal from spatial N. Dammeyer, A. Williams, C. Snyder, J. Schaberl, S. Faulkner, analysis of stream network topology. Journal of the North American Bentho- and K. Cooper for assistance with the development and review of logical Society 27:304–320. DAM REMOVAL BENEFITS AMERICAN EELS 1179
Hitt, N. P., and P. L. Angermeier. 2011. Fish community and bioassessment R Development Core Team. 2011. R: a language and environment for statisti- responses to stream network position. Journal of the North American Ben- cal computing. R Foundation for Statistical Computing, Vienna. Available: thological Society 30:296–309. www.R-project.org/. (October 2011). Jenkins, R. E., and N. M. Burkhead. 1994. Freshwater fishes of Virginia. Amer- Reed, E. B., and G. Bear. 1966. Benthic animals and foods eaten by brook trout ican Fisheries Society, Bethesda, Maryland. in Archuleta Creek, Colorado. Hydrobiologia 27:227–237. Jessop, B. M. 2010. Geographic effects on American eel (Anguilla rostrata) Shmida, A., and M. V. Wilson. 1985. Biological determinants of species diver- life history characteristics and strategies. Canadian Journal of Fisheries and sity. Journal of Biogeography 12:1–20. Aquatic Sciences 67:326–346. Smith, M. W., and J. W. Saunders. 1955. The American eel in certain fresh waters Kendall, M. G. 1975. Rank correlation methods. Griffin Publishers, London. of the maritime provinces of Canada. Journal of the Fisheries Research Board Kiffney, P. M., G. R. Pess, J. H. Anderson, P. Faulds, K. Burton, and S. C. Riley. of Canada 12:238–269. 2009. Changes in fish communities following recolonization of the Cedar Smogor, R. A., P. L. Angermeier, and C. K. Gaylord. 1995. Distribution and River, WA, USA by Pacific salmon after 103 years of local extirpation. River abundance of American eels in Virginia streams: tests of null models across Research and Applications 25:438–452. spatial scales. Transactions of the American Fisheries Society 124:789– Koutroumanidis, T., L. S. Iiadis, and G. K. Sylaios. 2006. Time-series model- 803. ing of fishery landings using ARIMA models and fuzzy expected intervals Stanley, E. H., and M. W. Doyle. 2003. Trading off: the ecological effects of software. Environmental Modelling and Software 21:1711–1721. dam removal. Frontiers in Ecology and the Environment 1:15–22. Krueger, W. H., and K. Oliveira. 1999. Evidence for environmental sex deter- Tuckey, T. D., and M. C. Fabrizio. 2010. Estimating relative juvenile abundance mination in the American eel, Anguilla rostrata. Environmental Biology of of ecologically important finfish in the Virginia portion of Chesapeake Bay. Fishes 55:381–398. Annual Report to the Virginia Marine Resources Commission, Project F-104- Lookabaugh, P. S., and P. L. Angermeier. 1992. Diet patterns of American eel, R-14, Virginia Institute of Marine Science, Gloucester Point. Anguilla rostrata, in the James River drainage, Virginia. Journal of Freshwater USFWS (U.S. Fish and Wildlife Service). 2011. Endangered and threat- Ecology 7:425–431. ened wildlife and plants; 90-day finding on a petition to list the Ameri- Machut, L. S., K. E. Limburg, R. E. Schmidt, and D. Dittman. 2007. Anthro- can eel as threatened. Federal Register 76:189(29 September 2011):60431– pogenic impacts on American eel demographics in Hudson River tributaries, 60444. New York. Transactions of the American Fisheries Society 136:1699–1713. Watters, G. T. 1996. Small dams as barriers to freshwater mussels Mann, H. B. 1945. Nonparametric tests against trend. Econometrica 13:245– (Bivalvia, Unionoida) and their hosts. Biological Conservation 75:79– 259. 85. MDNR (Maryland Department of Natural Resources). 2010. Maryland biolog- Wentworth, C. K. 1922. A scale of grade and class terms for clastic sediments. ical stream survey, sampling manual: field protocols. MDNR, Report 12- Journal of Geology 30:377–392. 2162007-19, Annapolis. Wiley, D. J., R. P. Morgan II, R. H. Hilderbrand, R. L. Raesly, and D. L. Ogden, J. C. 1970. Relative abundance, food habits, and age of the American Shumway. 2004. Relations between physical habitat and American eel abun- eel, Anguilla rostrata (LeSueur), in certain New Jersey streams. Transactions dance in five river basins in Maryland. Transactions of the American Fisheries of the American Fisheries Society 99:54–59. Society 133:515–526. Oliveira, K. 1999. Life history characteristics and strategies of the American Williams, J. D., M. L. Warren Jr., K. S. Cummings, J. L. Harris, and R. J. Neves. eel, Anguilla rostrata. Canadian Journal of Fisheries and Aquatic Sciences 1993. Conservation status of freshwater mussels of the United States and 56:795–802. Canada. Fisheries 18(9):6–22. Poff, N. L., and D. D. Hart. 2002. How dams vary and why it matters for the Winston, M. R., C. M. Taylor, and J. Pigg. 1991. Upstream extirpation of four emerging science of dam removal. BioScience 52:659–668. minnow species due to damming of a prairie stream. Transactions of the Power, M. E., W. J. Matthews, and A. J. Stewart. 1985. Grazing minnows, American Fisheries Society 120:98–105. piscivorous bass, and stream algae: dynamics of a strong interaction. Ecology Yue, S., and C. Wang. 2004. The Mann-Kendall test modified by effective 66:1448–1456. sample size to detect trend in serially correlated hydrological series. Water Pringle, C. M. 1997. Exploring how disturbance is transmitted upstream: go- Resources Management 18:201–218. ing against the flow. Journal of the North American Benthological Society Zhang, G. P. 2003. Time series forecasting using a hybrid ARIMA and neural 16:425–438. network model. Neurocomputing 50:159–175. Downloaded by [Department Of Fisheries] at 19:58 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 19:59 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Quantifying Cumulative Entrainment Effects for Chinook Salmon in a Heavily Irrigated Watershed Annika W. Walters a c , Damon M. Holzer a , James R. Faulkner a , Charles D. Warren b , Patrick D. Murphy b & Michelle M. McClure a a National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, Washington, 98112, USA b Idaho Department of Fish and Game, Anadromous Fish Screen Program, 99 Highway 93N, Salmon, Idaho, 83467, USA c U.S. Geological Survey, Wyoming Cooperative Fish and Wildlife Research Unit, Department 3166, University of Wyoming, 1000 East University Avenue, Laramie, Wyoming, 82071, USA Version of record first published: 19 Jul 2012.
To cite this article: Annika W. Walters, Damon M. Holzer, James R. Faulkner, Charles D. Warren, Patrick D. Murphy & Michelle M. McClure (2012): Quantifying Cumulative Entrainment Effects for Chinook Salmon in a Heavily Irrigated Watershed, Transactions of the American Fisheries Society, 141:5, 1180-1190 To link to this article: http://dx.doi.org/10.1080/00028487.2012.679019
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. Transactions of the American Fisheries Society 141:1180–1190, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.679019
ARTICLE
Quantifying Cumulative Entrainment Effects for Chinook Salmon in a Heavily Irrigated Watershed
Annika W. Walters,*1 Damon M. Holzer, and James R. Faulkner National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, Washington 98112, USA Charles D. Warren and Patrick D. Murphy Idaho Department of Fish and Game, Anadromous Fish Screen Program, 99 Highway 93N, Salmon, Idaho 83467, USA Michelle M. McClure National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, Washington 98112, USA
Abstract Pacific salmon Oncorhynchus spp. experience multiple small-scale disturbances throughout their freshwater habi- tat, but the cumulative effect of these disturbances is often not known or not easily quantifiable. One such disturbance is water diversions, which can entrain fish and alter streamflow regimes. Threatened Lemhi River (Idaho) Chinook salmon O. tshawytscha smolts encounter 41–71 water diversions during their out-migration. We used passive inte- grated transponder tag data to model the entrainment rate of Chinook salmon smolts as a function of the proportion of water removed by an irrigation diversion. Under median-streamflow conditions with unscreened diversions, the estimated cumulative effect of the diversions was a loss of 71.1% of out-migrating smolts due to entrainment. This is a large potential source of mortality, but screening is an effective mitigation strategy, as estimated mortality was reduced to 1.9% when all diversions were screened. If resources are limited, targeting the diversions that remove a large amount of water and diversions in locations with high fish encounter rates is most effective. Our modeling approach could be used to quantify the entrainment effects of water diversions and set screening priorities for other watersheds. Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 Pacific salmon Oncorhynchus spp. are an integral component effects of local, small-scale habitat disturbances and restoration of Pacific Northwest stream ecosystems but have experienced efforts at the population level (Bartz et al. 2006; Honea et al. extensive population declines due to habitat alteration, stocking 2009; Roni et al. 2010). Quantification of population-level ef- practices, hydropower development, and climate (Ruckelshaus fects is challenging, but one important habitat disturbance that et al. 2002; McClure et al. 2003). Most salmon recovery plans may be quantifiable is the direct effect of irrigation diversions include strong recommendations for improving habitat condi- on the out-migration success of juvenile salmonids. tions; the actions implemented based on these recommendations Water diversion for irrigation is a major threat that is typically consist of small projects at multiple locations through- currently faced by fish populations (Rosenberg et al. 2000). out the spawning and rearing habitat of a population. Salmon Water diversion can alter streamflow regimes and entrain fish biologists are working to determine methods for quantifying the in irrigation canals (Gebhards 1958; Post et al. 2006). Although
*Corresponding author: [email protected] 1Present address: U.S. Geological Survey, Wyoming Cooperative Fish and Wildlife Research Unit, Department 3166, University of Wyoming, 1000 East University Avenue, Laramie, Wyoming 82071, USA. Received October 28, 2011; accepted March 19, 2012
1180 CUMULATIVE ENTRAINMENT EFFECTS 1181
the importance of streamflow alteration is receiving greater sions reduce streamflow in the Lemhi River and its tributaries to research attention (Poff and Zimmerman 2010), the direct the extent that most tributaries are disconnected from the main effects of entrainment by water diversions are not as well stem (Tire et al. 2011). Screening of water diversions began in studied. For a fish population, the entrainment effect of any 1958 (Schill 1984), and currently most of the water diversions individual diversion may be minimal, but the cumulative effects on the Lemhi River and its large tributary (Hayden Creek) are of multiple diversions could be considerable. screened. However, there are still many unscreened diversions Entrainment is the process by which fish travel into irrigation on the other tributaries. canals at a water diversion. The fate of entrained fish depends The Lemhi River basin historically supplied productive on whether the diversion is screened. In an unscreened diver- spawning and rearing grounds for Chinook salmon, but the pop- sion, fish will enter the irrigation system and likely die; if the ulation has experienced substantial declines in the last 50 years. diversion is screened, fish are bypassed and returned to the main The population is part of the Snake River spring–summer Chi- river channel (Zydlewski and Johnson 2002). The few studies nook salmon evolutionarily significant unit, which is listed of individual diversions have reported entrainment rates ranging as threatened under the Endangered Species Act. Due to its from 1% to 79% (Carlson and Rahel 2007; Gale et al. 2008). It size and location, the Lemhi River population is considered a is challenging to quantify the number of fish that are entrained at key population for the recovery of this evolutionarily signifi- a single diversion, and estimating the population-level effect is cant unit (ICTRT 2007). Spawning currently occurs primarily even harder. As a result, entrainment studies have mainly been in the upper Lemhi River and Hayden Creek (Figure 1). We conducted for nonmigratory species at one or a few diversions use the term “upper Lemhi River” to refer to the main-stem (Schrank and Rahel 2004; Unwin et al. 2005; Post et al. 2006). Lemhi River above the confluence with Hayden Creek; the term Despite the lack of quantified effects, managers have long recog- “lower Lemhi River” refers to the main-stem Lemhi River be- nized the potential for irrigation diversion entrainment to have a low the confluence with Hayden Creek. Chinook salmon in substantial negative effect on fish populations. As a result, fish the Lemhi River basin have demonstrated three migration life screens were built as early as the 1890s, although they were of- histories within the same cohort: (1) out-migration during the ten discontinued due to high maintenance costs (Clothier 1953). first spring after emergence as age-0 early smolts; (2) down- Screens are still expensive to build and maintain, but they are stream migration during the fall as age-0 fall parr and subse- becoming a common conservation practice. Similar to the nega- quent overwintering in the lower Lemhi River and Salmon River; tive effects of diversions, the benefits of screening have not been and (3) out-migration during the next spring as age-1 smolts well quantified at the population level (Moyle and Israel 2005; (Bjornn 1978; Lutch et al. 2003). In this study, we focused on but see Gale et al. 2008). age-1 smolts, which represent one of the common migration The goal of this study was to evaluate the effect of entrain- strategies. ment due to water diversion for migrating Chinook salmon O. tshawytscha smolts and the benefit conferred by screening those Geospatial Model diversions. We used Lemhi River (Idaho) Chinook salmon as a We used a spatially explicit GIS-based simulation model case study, because the Lemhi River basin experiences extensive to assess the effects of diversion entrainment on Lemhi River withdrawals of water due to irrigation diversions. Our objectives Chinook salmon smolts. We first identified the location and size were to (1) explore how entrainment varied with diversion rate of each diversion in the watershed. We then developed a model and variation in streamflow, (2) predict fish mortality in the to estimate the probability that a fish would be entrained at a Lemhi River basin for unscreened and screened scenarios, and diversion. Finally, we routed fish through the stream network, (3) evaluate various management approaches to the prioritiza- with individuals removed at each diversion based on the modeled
Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 tion of screening efforts. probability of entrainment. Water diversion locations.—We obtained all points of diver- METHODS sion (PODs) in the Lemhi River basin and the associated wa- ter rights data from the Idaho Department of Water Resources Study Site (IDWR; Figure 1). For each POD, we summarized the total The Lemhi River is a tributary to the Salmon River in central amount of legal water withdrawals and provided this value as a Idaho (Figure 1). Its watershed encompasses about 3,300 km2 new attribute. Many of the PODs in the IDWR database have of forest, rangeland, and irrigated land. The climate is semiarid, been consolidated into a single diversion for screening purposes, with cold winters and warm summers. Precipitation generally so in the final analysis we only considered the list of irrigation increases with elevation (22–115 cm/year) and primarily falls as diversions and associated fish screens developed by the Idaho snow. Melt from mountain snowpack is the predominant source Department of Fish and Game (IDFG) Anadromous Fish Screen of streamflow, but the Lemhi River basin also has groundwater Program. Each of these diversions has a screen design discharge inputs that modify the influence of annual freshets. Diversions that corresponds to the maximum discharge that can be diverted. occurred in the Lemhi River basin as early as 1855, and today Lemhi River gage data are not available at every POD; thus, to there are over 250 gravity-fed irrigation diversions. These diver- estimate streamflow, we modeled natural streamflow (low [Q80, 1182 WALTERS ET AL. Downloaded by [Department Of Fisheries] at 19:59 25 September 2012
FIGURE 1. Map of the Lemhi River basin, Idaho. Each point of diversion is denoted by a circle; the size of the circle represents the legal rate of water diversion in cubic feet per second (CFS; 1 CFS = 0.03 m3/s) for that site. Areas where spring Chinook salmon spawn and rear are highlighted in bold. CUMULATIVE ENTRAINMENT EFFECTS 1183
i.e., the flow level that was exceeded 80% of the time], median average relationship between entrainment probability and pro- [Q50], and high [Q20] streamflows under no-diversion condi- portion of streamflow diverted, we modeled the average of the tions, estimated on the basis of watershed characteristics) and entrainment probabilities across sites within years. This reduced then subtracted the cumulative water rights upstream of the the data to six data points (one for each year; Table A.2). POD, accounting for the fact that some of the diverted water The model relating entrainment rate to the proportion of would return to the Lemhi River via return flow (percent return streamflow diverted was fitted using weighted least squares, flow estimated based on gage data; see Appendix). with the weights equal to the inverse of the estimated variances Entrainment rate.—The IDFG Anadromous Fish Screen Pro- of the mean entrainment probabilities on the logit scale. We gram has monitored fish entrainment since 2003 by installing assumed that if no streamflow was diverted at a given site, then automated passive integrated transponder (PIT) tag readers on no fish would be entrained in that irrigation diversion; likewise, fish screen bypass pipes. The PIT tag interrogation stations we assumed that if 100% of the streamflow was diverted at a (Biomark, Inc., Boise, Idaho) documented the movement of given site, then all passing fish would be entrained at that site. PIT-tagged fish (tagged as part of the routine IDFG monitoring To impose these constraints in our models, we used the logit program) through fish screen bypasses on two to four diversions transformation (logit[x] = loge[x/{1 − x}]) for both the entrain- from 2003 to 2008. ment probability estimates and the proportion of flow diverted. To estimate the probability that an individual fish would On the logit scale, the model for entrainment probability at a be entrained at a monitored irrigation diversion, we examined diversion site for a cohort of fish in a season (pi) as a func- records of smolts out-migrating during spring (March 1–June tion of the proportion of streamflow diverted at that site (pdivi) 30) 2003–2008 (PTAGIS 2011). We did not include fall-tagged was parr; although some of these individuals actively migrate down- stream and out of the Lemhi River basin, others hold over for logit(pi ) = β0 + β1logit(pdivi ) + εi . (1) the winter in the lower reaches of the Lemhi River. Separate release cohorts of fish were created for each irrigation diversion Here, εi represents random error terms that are assumed to with a PIT tag detector because the detector dates of operation be normally distributed with a mean of 0 and a variance of varied. For each release cohort, we created capture histories for σ 2 on the logit scale. We used weighted least squares to fit individual fish and used the Cormack–Jolly–Seber model (Cor- the models, with the weights being equal to the inverse of the mack 1964; Jolly 1965; Seber 1965) to estimate joint detection estimated variances of the entrainment probability estimates and survival probabilities. We assumed that the probability of on the logit scale. Weighting in this way allowed observations detection at a fish bypass was the same as the probability of with more precise entrainment probability estimates to have entrainment; however, detection efficiencies are potentially less more influence in the model fit. Due to the unknown time of than 1.0 at every site, so the resulting entrainment probabilities passage for individual undetected fish, there was a large amount are actually minimum estimates. of missing information on the proportion of streamflow diverted, We related these entrainment probabilities to the average thus precluding the use of commonly used capture–recapture daily proportion of streamflow that was diverted during the models that allow individual covariates (e.g., Lebreton et al. migration season at the monitored diversion site (Table A.1). 1992). The daily proportion of streamflow diverted was calculated Fish routing.—The final stage in the model was simulating on the basis of daily Lemhi River streamflow at each mon- fish loss to or passage by the irrigation diversions. Since many itored diversion site (estimated from the nearest streamflow of the PODs have been consolidated for screening purposes, gage) and the daily estimated discharge that was diverted by we considered the diversions that were monitored and screened
Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 the irrigation diversion. The discharge diverted was unknown by the IDFG Anadromous Fish Screen Program. For these and was predicted using a step function based on historic diversions, the most accurate estimate of discharge diverted is values from an evaluation of diversion operation plans (DHI related to fish screen design, as each fish screen has a maximum 2003). If streamflow at a diversion site was low (
as stream area. We divided all reaches in currently occupied 10,000 total survival values was used as the SE of predicted total rearing streams into 200-m segments and inserted each irriga- survival. tion diversion into its correct position within this network. The predicted proportion of smolts in each segment was assumed Simulation Model to Estimate Multiple Entrainment Events to be equivalent to the proportion of stream area. Starting from When irrigation diversions are screened, an individual fish the upstream-most position, we routed fish downstream, adding may be entrained multiple times; this could increase stress and smolts at new reach segments and removing smolts at diversions travel time. To estimate how many times a fish is entrained and based on entrainment estimates. bypassed, we ran a Monte Carlo simulation. In the simulation, We compared the simulated effects of entrainment for each time a fish encountered an irrigation diversion a random screened and unscreened scenarios under low, median, and high number between 0 and 1 was generated; if the number was May streamflow conditions. If an irrigation diversion was un- less than the probability of entrainment for that diversion, the screened, it was assumed that a smolt would not survive entrain- fish was assumed to be entrained. The probability of entrain- ment, and survival was set at 0.01. If a diversion was screened, ment was the value from the previous analysis under median survival was set at 0.99 because a previous laboratory study May conditions. The number of times each individual fish was showed that survival rates at screens were greater than 0.99 entrained and bypassed at a diversion was counted, and the sim- for juvenile Chinook salmon (Swanson et al. 2004). In addition ulation was run for 10,000 fish. The simulation used only the 41 to the current situation, in which all diversions on the main- screened irrigation diversions that all smolts encounter as they stem Lemhi River are screened, a series of potential screening pass through the lower Lemhi River; therefore, this simulation management strategies was tested, including screening based on provided a minimum estimate of multiple entrainment rates. location, diversion rights, or entrainment rates. We estimated SEs of predicted total survival using Monte RESULTS Carlo simulation. For each of 10,000 iterations, a slope and The weighted linear regression yielded convincing evidence an intercept for the entrainment rate model were drawn from that the mean estimated entrainment probability for a Chi- a multivariate normal distribution with the mean vector equal nook salmon smolt was associated with the mean proportion to the estimated model parameters from the regression model of streamflow diverted (P = 0.0037; Table 1; Figure 2). The for entrainment rates (Table 1) and with the covariance matrix proportion of fish entrained was slightly less than the propor- equal to that for the estimated model parameters, both on the tion of streamflow diverted; variability in the proportion of fish logit scale. In addition, within each iteration, each diversion site entrained increased as the proportion of streamflow diverted had a separate random error term that was drawn from a normal increased (Figure 2). Based on this relationship, the modeled distribution with a mean of 0 and a variance that was equal entrainment rate at individual irrigation diversions was rela- to the estimated residual variance from the regression model— tively low: on average, the percentage of smolts entrained at an that is, for each diversion site, within each iteration the predicted individual diversion was 4% under high-streamflow conditions, entrainment rate was
−1 pˆ h,i = logit (b0,h + b1,hpdivi + eh,i ), (2)
where h is the index for simulation iteration, i is the index for irrigation diversion site, b0 and b1 are simulated regression
Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 parameters, pdiv is the logit-transformed proportion of water diverted, and e is the random error term. We ran the entrainment calculator for each iteration of the simulation, and we recorded the resulting cumulative total survival values. The SD of those
TABLE 1. Parameter estimates and associated P-values for a weighted linear regression relating the probability of Chinook salmon smolt entrainment to the proportion of streamflow removed by an irrigation diversion in the Lemhi River basin, Idaho. The P-values are two-sided; pdiv is the proportion of streamflow diverted. FIGURE 2. Estimates of mean ( ± SE) entrainment probability for Chinook Coefficient Estimate SE tPsalmon smolts versus the mean proportion of streamflow diverted at irrigation diversions within the Lemhi River basin, Idaho, 2003–2008. Entrainment prob- Intercept −0.560 0.238 −2.356 0.0780 abilities are means of the monitored diversion sites for each year. The solid line Logit(pdiv) 0.907 0.150 6.062 0.0037 is from the weighted regression described in Table 1; dotted lines represent the 95% confidence interval. CUMULATIVE ENTRAINMENT EFFECTS 1185
100
80
60
40 % Mortality 20
0
high median low FIGURE 4. Probability of zero, one, or multiple entrainment events for a Flow conditions Chinook salmon smolt as it passes through the 41 irrigation diversions in the lower main-stem Lemhi River. Probability was estimated with a simulated run of 10,000 fish. FIGURE 3. Entrainment mortality (mean ± SE) during the Chinook salmon smolt out-migration under scenarios of unscreened (open squares) and screened (shaded circles) irrigation diversions and high (Q20), median (Q50), and low mortality, but even when the 40 diversions with the highest en- (Q80) streamflow conditions. trainment rates were screened the mortality rate was still 39.6% due to the other unscreened diversions. When approximately 6% under median-streamflow conditions, and 10% under low- 40 diversions were screened based on their location or on the streamflow conditions. The maximum estimated entrainment amount of discharge diverted, mortality dropped to 28.7% and rate for a single diversion was 45% (low-streamflow conditions), 26.0%, respectively (Table 2). and the minimum entrainment rate was less than 1%. Estimated cumulative effects under scenarios of unscreened irrigation diversions were high. Under median May streamflow DISCUSSION conditions, 71% of migrating smolts would be lost to entrain- Irrigation is the largest use of freshwater in the United States, ment in irrigation canals (Figure 3). This percentage increased to and much of this water is obtained through diversion from over 88% during low-streamflow conditions. During low stream- flow, a greater proportion of the water was diverted, resulting TABLE 2. Potential screening management strategies for irrigation diversions in higher entrainment rates (Figure 3). Under high-streamflow and the associated predictions of entrainment mortality of Chinook salmon conditions, smolt mortality decreased to 54% (Figure 3). Under smolts under median May streamflow conditions. The screening scenario based scenarios in which diversions were screened, mortality dropped on entrainment probability involves screening those diversions with the highest entrainment probabilities under median May streamflow conditions. to between 1% and 4% for all streamflow conditions. For the scenarios involving screened diversions, we assumed Screening Diversions Mortality very low mortality with entrainment (1%), such that a fish could scenario screened (N)(%) become entrained multiple times. Most fish were entrained one None screened 0 71.1 Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 or two times while passing through the 41 irrigation diversions in the lower Lemhi River (Figure 4). Approximately 12% were All screened 89 1.9 never entrained, and about 15% were entrained four or more Screening based on location times. Main-stem Lemhi River 70 7.7 Screening of all irrigation diversions resulted in a decline in Lower main-stem Lemhi River 41 28.7 smolt mortality from 71.1% to 1.9% (Table 2). Screening of Lemhi River tributaries 19 70.4 the diversions located on the main-stem Lemhi River resulted Screening based on water rights in large reductions in Chinook salmon smolt mortality (7.7% >0.57 m3/s (20 ft3/s) diverted 22 42.5 mortality), while screening only the diversions within the trib- >0.28 m3/s (10 ft3/s) diverted 42 26.0 utaries had little effect (70.4% mortality). Screening the largest >0.14 m3/s (5 ft3/s) diverted 70 9.6 irrigation diversions was more important than screening the Screening based on entrainment probability smaller diversions; however, even when all diversions greater Top-10 diversions 10 66.6 than 0.14 m3/s (5 ft3/s) were screened, there were still effects Top-20 diversions 20 59.6 (9.6% mortality). Screening of diversions based on entrainment Top-40 diversions 40 39.6 probability (i.e., proportion of water diverted) also decreased 1186 WALTERS ET AL.
natural streams and rivers (Hutson et al. 2004). The effect of tory migrants in a basin with many irrigation diversions. In the this extensive network of irrigation diversions on fish popula- Lemhi River basin, the strong effects are driven by the num- tions is not well known. In this study, we evaluated the direct ber of diversions, although several of the irrigation diversions effects of diversion structures on Chinook salmon smolts in a divert large amounts of water and have high individual entrain- heavily irrigated watershed. Although most individual irriga- ment rates (25–45% [depending on streamflow] if unscreened). tion diversions only entrained a small proportion of migrat- It is possible for one irrigation diversion to have considerable ing smolts, the predicted cumulative effect of water diversion effects; in the Yellowstone River, one diversion accounted for on smolt out-migration survival was substantial. However, the more than half of all nonfishing mortality in saugers Sander direct effects of diversions were successfully mitigated by a canadensis (Jaeger et al. 2005). comprehensive screening program. The methods developed for In this study, we focused only on smolt out-migration, but this study can provide an approach for quantifying the effects many juvenile Chinook salmon out-migrate as parr during the of irrigation diversions and setting screening priorities in other previous fall. Differing migration strategies expose cohorts to watersheds. varying risks in relation to the management of irrigation with- We found a consistent relationship between the proportion drawals and streamflow. For parr that migrate all the way to of streamflow diverted and the entrainment rate. Our relation- the Salmon River during fall, entrainment rates should be lower ship differed from the expected S-shaped curve that describes because the amount of irrigation withdrawal decreases in Octo- a scenario in which very few fish are entrained at low water ber and November. For parr that overwinter in the lower Lemhi diversion rates, entrainment increases sharply as more water River and continue their out-migration during the next spring, is diverted, and entrainment reaches nearly 100% at high wa- entrainment rates in the upper Lemhi River would be lower, and ter diversion levels (Moyle and Israel 2005). However, in our entrainment rates in the lower Lemhi River would be compara- study, the average proportion of streamflow diverted at any in- ble to those of smolts. In the six monitored irrigation diversions, dividual irrigation diversion was generally less than 50%, and parr entrainment was lower than smolt entrainment, with 1– a sharp increase in entrainment could be possible when the 12% of tagged parr entrained in a monitored irrigation diversion majority of water is flowing into an irrigation canal. As a re- during a given year (2003–2007) in comparison with 6–34% sult, our entrainment curve likely provides a conservative es- of smolts (C.D.W., unpublished data). Although the movement timate of entrainment at high levels of water diversion. Due patterns during migration are straightforward, parr and smolts to the large number of irrigation diversions, we used the same could also be moving locally and encountering irrigation diver- entrainment curve for all diversions, but the relationship will sions multiple times before migration, increasing the probability likely differ between diversion sites. Other factors that could of entrainment. Future studies should incorporate all life history affect entrainment rates include fish species and life history, types and non-migration-related movement. the configuration of the irrigation diversion, the timing of wa- In addition to providing estimates of entrainment, our ap- ter diversion, and the physical location of the irrigation di- proach allowed us to quantify the effectiveness of screening version site (Schrank and Rahel 2004; Bahn 2007; Grimaldo measures in the Lemhi River basin. The basin has undergone et al. 2009; Svendsen et al. 2010). Bahn (2007) found that an intensive screening program, and at this point all irriga- the best predictors of entrainment rate for irrigation diversions tion diversions that are likely to be encountered by Chinook were the amount of discharge diverted and the upstream slope. salmon smolts during out-migration are screened. Screening For studies that are focused on the effect of only a subset of has been shown to be an effective management tool (Gale specific water diversions, individual entrainment curves should et al. 2008), but there are still few studies that quantify the be calculated that take into account the unique characteristics potential benefits of screening, especially at the population
Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 of each irrigation diversion site. Diversion-specific streamflow level or for out-migrating anadromous fish (Moyle and Israel and water diversion rates would allow further refinement of the 2005). Our modeling approach showed that the screening of model. irrigation diversions can reduce mortality during smolt out- We found that the entrainment effect of a single irrigation migration from 50–90% to 1–4%. This result is useful for diversion was low, with only 6% of smolts entrained by an evaluating the vast amount of resources invested in screen- individual diversion on average. However, the cumulative ef- ing. Screening of an irrigation diversion in the Lemhi River fects of irrigation diversions for the population were consider- basin costs, on average, $3,600–4,600 per 0.03 m3/s (= 1ft3/s) able; 71.1% of migrating smolts were entrained under median- of diversion capacity. The average design discharge for an ir- streamflow conditions. A 71% entrainment rate is equal to the rigation diversion in this basin is about 0.42 m3/s (15 ft3/s), percentage of age-0 westslope cutthroat trout O. clarkii lewisi but the largest diversions divert almost 1.7 m3/s (60 ft3/s). In that were entrained during downstream movement in Skalkaho addition to the cost of building diversion screens, the IDFG Creek, Montana (Gale et al. 2008), but is much higher than rates employs full-time seasonal workers to clean and maintain the observed in other studies (Schrank and Rahel 2004; Post et al. screens in the Lemhi River basin. Results from this study could 2006; Carlson and Rahel 2007). The higher entrainment rate in be combined with cost estimates to conduct cost–benefit analy- our study is attributable to the fact that we considered obliga- ses for proposed screening projects. CUMULATIVE ENTRAINMENT EFFECTS 1187
By examining various screening strategies, the model pro- ment effects. Our results suggest that these effects can be suc- vides support for current methods of prioritizing screening cessfully mitigated by screening, although the effects of multiple efforts. In the Lemhi River basin, screening the irrigation di- entrainments in screened diversions on travel time and stress re- versions that divert more water and that are located on the main quire further study. While diversion screening is widely viewed stem will save more smolts than the screening of smaller diver- as important, this is one of the first studies to quantify the ben- sions along the tributaries. The results suggest that screening efits at the population level. In addition, our approach allows based on the size of the diversion was a consistently good strat- a comparison of the costs and benefits of various screening egy. Screening of the diversions with the highest entrainment approaches, which can help managers to prioritize the limited probability (based on proportion of water diverted) was not as funds dedicated to restoration efforts. This study also highlights effective as expected because many of these diversions were the importance of evaluating the cumulative effects of small- located higher in the watershed and thus encountered fewer scale disturbances and restoration efforts. fish. Due to the varying physical characteristics of individual irrigation diversions, these general guidelines should be supple- mented with a targeted approach aimed at diversions that may ACKNOWLEDGMENTS have high entrainment rates attributable to the timing of water We thank Michael Ciscell and Morgan Case (IDWR) for withdrawal or to the diversion configuration. In the Lemhi River providing the water rights GIS layers and guidance on how to basin, it was necessary to screen almost all of the diversions po- analyze them; we also thank Bryan Nordlund, Paul McElhany, tentially encountered by smolts for maximum effect. To reduce Al Zale, Tim Grabowski, and two anonymous reviewers for mortality to less than 10%, screening was required for approxi- providing helpful comments on the manuscript. A.W.W. was mately 70 of the 89 irrigation diversions encountered by smolts supported by a National Research Council research associate- during their out-migration. Our models focused on Chinook ship. The use of trade or product names does not constitute salmon, which primarily spawn and rear in the main stem, but endorsement by the U.S. Government. for species that are highly migratory tributary spawners (e.g., steelhead O. mykiss, bull trout Salvelinus confluentus, and west- slope cutthroat trout), screening of diversions on the tributaries REFERENCES may be much more important. The Lemhi River basin has over Bahn, L. 2007. An assessment of losses of native fish to irrigation diversions on selected tributaries of the Bitterroot River, Montana. Master’s thesis. Montana 100 unscreened diversions on the tributaries, possibly creating State University, Bozeman. negative effects for other species. Bartz, K. K., K. M. Lagueux, M. D. Scheuerell, T. Beechie, A. D. Haas, and The model assumed very high survival (99%) for smolts that M. H. Ruckelshaus. 2006. Translating restoration scenarios into habitat con- encountered screened irrigation diversions, but the assumption ditions: an initial step in evaluating recovery strategies for Chinook salmon was based on a laboratory study that only considered screen (Oncorhynchus tshawytscha). Canadian Journal of Fisheries and Aquatic Sci- ences 63:1578–1595. mortality. In the field, there may also be non-screen-related Bjornn, T. C. 1978. Survival, production, and yield of trout and Chinook salmon mortality: for example, the fish could experience mortality when in the Lemhi River, Idaho. Idaho Cooperative Fishery Research Unit, Bulletin exiting the bypass pipes if they exit into poor-quality habitat or 27, Moscow. into a pool with waiting predators. The estimates also do not Carlson, A. J., and F. J. Rahel. 2007. A basinwide perspective on entrainment account for the potential fitness costs of being entrained or for of fish in irrigation canals. Transactions of the American Fisheries Society 136:1335–1343. the possibility that the majority of fish will be entrained more Clothier, W. 1953. Fish loss and movement in irrigation diversions from the than once. Multiple entrainments can result in increased travel west Gallatin River, Montana. Journal of Wildlife Management 17:144–158. time, which is associated with decreased survival during out- Cormack, R. M. 1964. Estimates of survival from the sightings of marked
Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 migration (Scheuerell et al. 2009). Harnish (2007) found the animals. Biometrika 51:429–438. median entrainment time to be 7 d, but the estimate was for DHI Water and Environment, Inc. 2003. Evaluation of diversion operation plans to meet negotiated flow targets for salmon and steelhead in the Lemhi River nonmigratory fish (westslope cutthroat trout juveniles) without basin using the MIKE BASIN model. Report for the Idaho Department of the downstream-directed movement of out-migrating Chinook Water Resources, DHI Inc., Boise, Idaho. Available: http://www.idwr.idaho. salmon. In addition, the experience could be stressful or could gov/waterboard/WaterPlanning/mike basin/PDFs/Lemhi-Modeling-Final.pdf. alter behavior. Studies on delta smelt Hypomesus transpacificus (April 2011). found that plasma cortisol concentrations increased during ex- Gale, S. B., A. V. Zale, and C. G. Clancy. 2008. Effectiveness of fish screens to prevent entrainment of westslope cutthroat trout into irrigation canals. North posure to fish screens, indicating increased physiological stress American Journal of Fisheries Management 28:1541–1553. (Young et al. 2010). Thus, our estimates of mortality are proba- Gebhards, S. V.1958. Fish loss in irrigation canals of the Salmon River drainage bly minimum estimates. Given the high likelihood that a smolt as determined by electric shocker. Idaho Department of Fish and Game, will be entrained multiple times in a screened system, future Special Report, Salmon. studies should explore other possible sources of mortality and Grimaldo, L. F., T. Sommer, N. Van Ark, G. Jones, E. Holland, P. B. Moyle, B. Herbold, and P. Smith. 2009. Factors affecting fish entrainment into nonlethal effects. massive water diversions in a tidal freshwater estuary: can fish losses Irrigation diversions are potentially a significant source of be managed? North American Journal of Fisheries Management 29:1253– mortality for out-migrating smolt populations due to entrain- 1270. 1188 WALTERS ET AL.
Harnish, R. A. 2007. Fish screen efficiency and effects of screened and un- Available: https://research.idfg.idaho.gov/Fisheries Research Reports/Res- screened irrigation canals on the downstream movement of westslope cut- Schill1983. (April 2011). throat trout juveniles in Skalkaho Creek, Montana. Master’s thesis. Montana Schrank, A. J., and F. J. Rahel. 2004. Movement patterns in inland cut- State University, Bozeman. throat trout (Oncorhynchus clarki utah): management and conservation im- Honea, J. M., J. C. Jorgensen, M. M. McClure, T. D. Cooney, K. Engie, D. M. plications. Canadian Journal of Fisheries and Aquatic Sciences 61:1528– Holzer, and R. Hilborn. 2009. Evaluating habitat effects on population status: 1537. influence of habitat restoration on spring-run Chinook salmon. Freshwater Seber, G. A. F. 1965. A note on the multiple recapture census. Biometrika Biology 54:1576–1592. 52:249–259. Hutson, S. H., N. L. Barber, J. F. Kenney, K. S. Linsey, D. S. Lumia, and Svendsen, J. C., K. Aarestrup, M. G. Deacon, and R. H. B. Christensen. 2010. M. A. Maupin. 2004. Estimated use of water in the United States in Effect of a surface oriented travelling screen and water abstraction practices 2000. U.S. Geological Survey, Circular 1268, Reston, Virginia. Available: on downstream migrating Salmonidae smolts in a lowland stream. River http://pubs.usgs.gov/circ/2004/circ1268/pdf/circular1268.pdf. (April 2011). Research and Applications 26:353–361. ICTRT (Interior Columbia Technical Recovery Team) 2007. Viability criteria Swanson, C., P. S. Young, and J. J. Cech. 2004. Swimming in two-vector for application to interior Columbia River basin salmonid ESUs. National flows: performance and behavior of juvenile Chinook salmon near a simulated Marine Fisheries Service, review draft, Seattle. Available: http://www.nwfsc. screened water diversion. Transactions of the American Fisheries Society noaa.gov/trt/col/trt viability.cfm. (April 2011). 133:265–278. Jaeger, M. E., A. V. Zale, T. E. McMahon, and B. J. Schmitz. 2005. Sea- Tyre, A. J., J. T. Peterson, S. J. Converse, T. Bogich, D. Miller, M. Post van der sonal movements, habitat use, aggregation, exploitation, and entrainment of Burg, C. Thomas, R. Thompson, J. Wood, D. C. Brewer, and M. C. Runge. saugers in the lower Yellowstone River: an empirical assessment of factors 2011. Adaptive management of bull trout populations in the Lemhi Basin. affecting population recovery. North American Journal of Fisheries Manage- Journal of Fish and Wildlife Management 2:262–281. ment 25:1550–1568. Unwin, M. J., M. Webb, R. J. Barker, and W. A. Link. 2005. Quantifying Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both production of salmon fry in an unscreened irrigation system: a case study death and immigration-stochastic model. Biometrika 52:225–247. on the Rangitata River, New Zealand. North American Journal of Fisheries Lebreton, J.-D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992. Modeling Management 25:619–634. survival and testing biological hypotheses using marked animals: a unified Young, P. S., C. Swanson, and J. J. Cech. 2010. Close encounters with approach with case studies. Ecological Monograph 62:67–118. a fish screen III: behavior, performance, physiological stress responses, Lutch, J., B. Leth, A. Apperson, A. Brimmer, and N. Brindza. 2003. and recovery of adult delta smelt exposed to two-vector flows near a Idaho supplementation studies, annual progress report. Idaho Depart- fish screen. Transactions of the American Fisheries Society 139:713– ment of Fish and Game, Report 03-37, Boise. Available: https://research. 726. idfg.idaho.gov/Fisheries Research Reports/Res03-37Lutch1997-2004 Idaho Zydlewski, G. B., and J. R. Johnson. 2002. Response of bull trout fry to four types Supplementation Studies.pdf. (April 2011). of water diversion screens. North American Journal of Fisheries Management McClure, M. M., E. E. Holmes, B. L. Sanderson, and C. E. Jordan. 2003. 22:1276–1282. A large-scale, multispecies status assessment: anadromous salmonids in the Columbia River basin. Ecological Applications 13:964–989. Moyle, P. B., and J. A. Israel. 2005. Untested assumptions: effectiveness of screening diversions for conservation of fish populations. Fisheries 30(5):20– APPENDIX: ESTIMATION OF LEMHI RIVER 28. STREAMFLOW FOR EACH POINT OF DIVERSION PTAGIS (PIT Tag Information System). 2011. PIT tag information system Web site. Pacific States Marine Fisheries Commission, Portland, Oregon. To estimate Lemhi River streamflow at each point of diver- Available: www.ptagis.org. (January 2011). sion (POD), we first modeled what the natural hydrology would Poff, N. L., and J. K. H. Zimmerman. 2010. Ecological responses to altered be if there was no water diversion (Qundiverted). For each POD, flow regimes: a literature review to inform the science and management of we subtracted the cumulative legal water rights upstream of the environmental flows. Freshwater Biology 55:194–205. POD (WR), taking into account that the proportion of water Post, J. R., B. T. Van Poorten, T. Rhodes, P. Askey, and A. Paul. 2006. Fish entrainment into irrigation canals: an analytical approach and application rights utilized (i.e., the diversion fraction [DF]) might be less to the Bow River, Alberta, Canada. North American Journal of Fisheries than 1.0 and that some water would be returned via return flow Management 26:875–887. (RF): Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 Roni, P., G. Pess, T. Beechie, and S. Morley. 2010. Estimating changes in coho salmon and steelhead abundance from watershed restoration: Q = Q − × × − . how much restoration is needed to measurably increase smolt pro- diverted undiverted [ WR DF (1 RF)] (3) duction? North American Journal of Fisheries Management 30:1469– 1484. We used locations with available gage data as a check on our Rosenberg, D. M., P. McCully, and C. M. Pringle. 2000. Global-scale environ- estimates. mental effects of hydrological alterations: introduction. Bioscience 50:746– Natural hydrology (Q ) was estimated by using 751. undiverted Ruckelshaus, M. H., P. Levin, J. B. Johnson, and P. M. Kareiva. 2002. The StreamStats, a program developed by the U.S. Geological Sur- Pacific salmon wars: what science brings to the challenge of recovering vey (water.usgs.gov/osw/streamstats/idaho.html). StreamStats species. Annual Review of Ecology and Systematics 33:665–706. provides streamflow statistics based on basin characteristics Scheuerell, M. D., R. W. Zabel, and B. P. Sandford. 2009. Relating juvenile and assumes no anthropogenic hydrologic alterations. We migration timing and survival to adulthood in two species of threatened used StreamStats to calculate monthly hydrological indices Pacific salmon (Oncorhynchus spp.). Journal of Applied Ecology 46:983– 990. (exceedance probabilities Q20, Q50, and Q80) for 20 locations, 2 Schill, D. 1984. Evaluating the anadromous fish screen program on the upper and we regressed the indices against watershed area (km ). Salmon River. Idaho Department of Fish and Game, Special Report, Boise. Watershed area was calculated using the drainage area provided CUMULATIVE ENTRAINMENT EFFECTS 1189
in the U.S. Geological Survey’s National Hydrography Dataset, subtracted WR to match gage data. Under high-streamflow and a proximity function was used to assign each diversion to conditions, water rights owners have high water rights, which its nearest National Hydrography Dataset stream reach. For all were not included in the WR calculation. Therefore, under monthly hydrologic indices, the relationship with watershed high-streamflow conditions, we set the diversion rate at 150% area (ranging from 40 to 3,100 km2) was well described by a of WR. For some individual diversions, 150% of legal wa- power function: y = axb (R2 ≥ 0.94). For each diversion location, ter rights was more than the diversion ditch could handle, but we calculated watershed area and used the regression equations at a larger scale 150% of WR was comparable to the cu- to get an estimate of natural hydrology under low (Q80), median mulative maximum capacity of the diversion ditches. During (Q50), and high (Q20) streamflow conditions. Due to the large low-streamflow conditions, water rights owners will probably number of locations (>200), we did not directly use StreamStats not be able to utilize their full water rights; this is especially for all PODs. The Lemhi River has a substantial natural ground- true for junior water rights owners, who are mainly located water component that was not captured in the StreamStats along the upper Lemhi River and the tributaries. We set the di- analysis. To correct for this, we examined the difference between version rate equal to 50% of WR for the tributaries and upper the predicted median StreamStats streamflow and the median Lemhi River and to 70% of WR for the lower Lemhi River. For streamflow from gage data for winter months, when diversion high- and low-streamflow conditions, percentages were chosen was not occurring. We added 0.85 m3/s to streamflow in Lemhi so that predicted streamflow matched that calculated from gage Big Springs Creek, 1.42 m3/s to streamflow in Hayden Creek, data. 1.70 m3/s to streamflow in the upper Lemhi River, and 2.55 m3/s Not all of the diverted water is lost from the system; a propor- to streamflow in the lower Lemhi River. For the analysis, we tion will return to the stream through surface and groundwater focused on May streamflow because May is the time period flow (i.e., RF). We chose RF fractions so that predicted median during which smolt migration and water diversion are both May streamflow hydrology matched the median May stream- occurring. flow levels at the locations for which gage data were available. We calculated the WR impacting a POD by using GIS to The RF was 65% for the upper Lemhi River, 75% for the lower compute the aggregated values above each diversion. We did not Lemhi River, and 72% for all other areas. These values are within include extra-high water rights in these values. Under median- the range (50–99%) of reported RF fractions for a study of indi- streamflow conditions, we assumed that 100% of the legal di- vidual diversions (DHI Water and Environment, Inc. 2003). We version rights (DF = 1.0) were exercised; however, for high- assumed the same RF fractions for low- and high-streamflow and low-streamflow conditions, we adjusted the proportion of conditions.
TABLE A.1. Detection and covariate data for cohorts of Chinook salmon smolts released into the Lemhi River, Idaho, 2003–2008. Detection sites are the irrigation diversions that were equipped with passive integrated transponder tag readers. First day of release is given as the day of year (January 1 = day 1).
Estimated entrainment probability Detection Number of First day Streamflow Proportion of Year site fish released Mean SE of release (m3/s) streamflow diverted
2003 L03 202 0.109 0.039 113 5.72 0.172 L03A 202 0.109 0.039 113 5.88 0.105
Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 L09 119 0.051 0.035 135 7.88 0.113 2004 L03 196 0.535 0.054 105 0.91 0.647 L03A 350 0.106 0.024 95 0.72 0.279 L08A 174 0.105 0.035 106 1.31 0.459 L09 35 0.200 0.127 128 3.67 0.160 2005 L03 203 0.552 0.053 97 2.12 0.405 L03A 218 0.079 0.029 95 1.85 0.238 L08A 141 0.203 0.050 104 2.71 0.295 L09 98 0.238 0.066 111 3.63 0.200 2006 L03 175 0.156 0.037 113 6.24 0.143 L06 119 0.083 0.036 116 6.26 0.414 L30 119 0.183 0.050 116 6.99 0.075 2007 L03 128 0.064 0.036 107 6.82 0.091 L06 251 0.039 0.019 86 5.34 0.260 L30 142 0.208 0.056 103 6.57 0.075 2008 L03 65 0.027 0.027 114 10.83 0.107 L30 47 0.138 0.064 120 7.50 0.081 1190 WALTERS ET AL.
TABLE A.2. Yearly means of detection and covariate data for cohorts of Chinook salmon smolts released into the Lemhi River, 2003–2008. First day of release is given as the day of year (January 1 = day 1).
Estimated entrainment probability Number of Mean first day Mean streamflow Mean proportion of Year estimates of release Mean SE (m3/s) streamflow diverted 2003 3 120.3 0.090 0.019 6.50 0.130 2004 4 108.5 0.237 0.102 1.65 0.387 2005 4 101.8 0.268 0.101 2.58 0.284 2006 3 115.0 0.141 0.030 6.49 0.210 2007 3 98.7 0.103 0.053 6.24 0.142 2008 2 117.0 0.082 0.055 9.17 0.094 Downloaded by [Department Of Fisheries] at 19:59 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:00 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Quantifying Latent Impacts of an Introduced Piscivore: Pulsed Predatory Inertia of Lake Trout and Decline of Kokanee Erik R. Schoen a , David A. Beauchamp b & Nathanael C. Overman a c a Washington Cooperative Fish and Wildlife Research Unit, School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington, 98195-5020, USA b U.S. Geological Survey, Washington Cooperative Fish and Wildlife Research Unit, School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington, 98195-5020, USA c Puget Sound Energy, Post Office Box 97034 PSE-09S, Bellevue, Washington, 98009-9734, USA Version of record first published: 20 Jul 2012.
To cite this article: Erik R. Schoen, David A. Beauchamp & Nathanael C. Overman (2012): Quantifying Latent Impacts of an Introduced Piscivore: Pulsed Predatory Inertia of Lake Trout and Decline of Kokanee, Transactions of the American Fisheries Society, 141:5, 1191-1206 To link to this article: http://dx.doi.org/10.1080/00028487.2012.681104
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. Transactions of the American Fisheries Society 141:1191–1206, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.681104
ARTICLE
Quantifying Latent Impacts of an Introduced Piscivore: Pulsed Predatory Inertia of Lake Trout and Decline of Kokanee
Erik R. Schoen* Washington Cooperative Fish and Wildlife Research Unit, School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington 98195-5020, USA David A. Beauchamp U.S. Geological Survey, Washington Cooperative Fish and Wildlife Research Unit, School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington 98195-5020, USA Nathanael C. Overman1 Washington Cooperative Fish and Wildlife Research Unit, School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington 98195-5020, USA
Abstract Introduced long-lived predators often cause significant impacts on their prey, but these impacts can be masked from detection due to high “predatory inertia”: time lags in population growth and dietary ontogeny. We evaluated whether predation by introduced lake trout Salvelinus namaycush could explain the 88% decline in escapement of kokanee Oncorhynchus nerka during 2005–2009 in Lake Chelan, Washington. We quantified the strength and trend of predation impacts with field sampling, a hydroacoustic assessment of kokanee production, and bioenergetics and age-structured population models of lake trout. Lake trout consumption of kokanee exceeded kokanee production, indicating strong predation impacts at the start of the decline. Fully piscivorous lake trout (>550 mm fork length) were responsible for 83% of this predation. The population model predicted that a pulse of strong stocked cohorts crossed this piscivorous size threshold, causing the biomass of fully piscivorous lake trout to expand by roughly 70–300% during 2004–2009 and driving predation pressure to peak levels. Together, these results suggested that lake trout predation was a large and growing source of kokanee mortality during the decline. Counterintuitively, predation pressure was projected to increase even if the numbers of harvestable lake trout declined, as strong cohorts grew to piscivorous size while succumbing to mortality. Angler catch rates of lake trout declined by 40% during 2004–2007, Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 as was predicted by the population model; this decline in catch masked the rise in predation pressure. This analysis demonstrates the potential for introduced predators exhibiting high predatory inertia to cause strong, latent impacts on prey that would be unexpected based on harvest trends and prior dynamics alone. Forward-looking monitoring and modeling analyses are clearly advantageous for managers who seek to maintain ecosystems in long-term “balance” by detecting and reversing incipient changes in predation.
Introductions of long-lived predators often cause dramatic 2009). Although piscivores such as lake trout Salvelinus na- changes to freshwater ecosystems, but these impacts can be dif- maycush, northern pike Esox lucius, and walleye Sander vitreus ficult to predict and quantify (Eby et al. 2006; Martinez et al. can reduce or extirpate prey fish populations (He and Kitchell
*Corresponding author: [email protected] 1Present address: Puget Sound Energy, Post Office Box 97034 PSE-09S, Bellevue, Washington 98009-9734, USA. Received December 14, 2011; accepted March 24, 2012 1191 1192 SCHOEN ET AL.
1990; Bowles et al. 1991; McMahon and Bennett 1996; Patankar predation impacts may require several additional years to be- et al. 2006; Bystrom¨ et al. 2007; Ellis et al. 2011), these species come effective. also appear to interact weakly with the same prey taxa in other A recent decline in the abundance of kokanee Oncorhynchus systems (Richards et al. 1991; McMahon and Bennett 1996; nerka (lacustrine sockeye salmon) in Lake Chelan, Washington, Martinez et al. 2009; Ellis et al. 2011) and the conditions un- raised concerns about lake trout predation (Martinez et al. derlying these differences are poorly understood. Studies using 2009), but abundance trends did not suggest an obvious link before–after comparisons to evaluate impacts are often con- between lake trout introduction and the kokanee’s decline. founded by other processes, such as habitat alteration (Dextrase Lake trout were first introduced into the lake during 1980 and and Mandrak 2006), eutrophication (VanderZanden et al. 2003), were stocked heavily from 1990 to 2000 to establish a trophy or climate change (Rahel and Olden 2008; Sharma et al. 2009). fishery (Figure 1; Washington Department Fish and Wildlife, Complicating matters, many long-lived predators exhibit traits unpublished data). Stocking was terminated after this 11-year such as lengthy generation times and dietary ontogenies that pulse, and harvest limits were liberalized to reduce potential delay their maximum impacts for years after they are intro- predation impacts (Martinez et al. 2009). Natural reproduction duced. These traits have been termed “high predatory inertia” of lake trout was first documented in 2000 (Duke Engineering in the context of manipulating predator populations to control and Services 2000). Interestingly, kokanee escapement in- nonnative prey fish (Stewart et al. 1981). Here, we explore how creased roughly fourfold from the early 1990s to the mid-2000s, high predatory inertia may also inhibit efforts to evaluate the coinciding with the establishment of lake trout. From this ecological impacts of introduced piscivores by making latent peak, kokanee escapement declined by 88% during 2005–2009 strong interactions difficult to distinguish from persistent weak before partially recovering (Figure 1; Keesee and Keller 2012). interactions. Reports from the fishery indicated that lake trout harvest rates High predatory inertia complicates standard approaches to mirrored these trends, increasing substantially during the 1990s quantifying food web patterns and dynamics. Strong predatory to mid-2000s (CCPUD 2007; Martinez et al. 2009) before interactions are most convincingly identified with direct exper- declining from 2005 to 2008 (A. Jones, Darrell and Dad’s imental manipulation (Paine 1980; Schindler 1998; Carpenter Family Guide Service, personal communication). Positively et al. 2001), but controlled experiments are often impractical in correlated densities of predators and prey generally suggest large or unique systems, especially when the effects of predators that top-down control is weak (e.g., Worm and Myers 2003; develop slowly. More commonly, predation impacts are evalu- ated indirectly by using either bioenergetics models or popula- tion dynamics models, and a combination of these approaches can be useful for quantifying the impacts of long-lived preda- tors (e.g., Stewart et al. 1981; Luecke et al. 1994; Schindler et al. 1998). Bioenergetics analysis provides quantitative es- timates of predation rates, but it requires intensive sampling that is generally practical only for relatively short periods and it does not predict future dynamics in changing systems (Ney 1993; Beauchamp et al. 2007; Chipps and Wahl 2008). Pop- ulation models can incorporate dynamics on longer temporal scales, but they do not identify mechanisms of interaction and they are sensitive to confounding events and ecological time
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 lags (Stenseth et al. 1997; White et al. 2006; Peckarsky et al. 2008). Despite these challenges, strong interactions between long- lived predators and their prey are particularly important to iden- tify in advance because of the time lags involved with detecting and responding to them. For example, juvenile lake trout are difficult to sample, so an increase in recruitment may go unno- ticed for 4–7 years until fish become fully vulnerable to angling or other sampling gears (Shuter et al. 1998). Another two or more years may pass before harvestable lake trout become fully piscivorous (Ruzycki et al. 2003), and an upsurge in predation FIGURE 1. Number of lake trout stocked into Lake Chelan (top panel; Wash- may remain undetected until the affected prey cohort reaches the ington Department of Fish and Wildlife, unpublished data) and an index of fishery or spawning grounds (Ellis et al. 2011). Finally, these kokanee escapement in the Lake Chelan drainage (bottom panel; data from Keesee and Keller 2012). Escapement was estimated using the area-under-the- predators are slow to respond to changes in stocking or har- curve method (Neilson and Geen 1981) based on data from spawner surveys vest (Stewart et al. 1981), so management efforts to mitigate conducted every 7–14 d in five major tributaries of the lake. LATENT PREDATION IMPACTS OF LAKE TROUT 1193
Ware and Thomson 2005); therefore, one obvious interpretation ics was consistent with a strong predation interaction during the of these synchronized abundance trends was that lake trout decline. predation had weak impacts on kokanee. However, unlike many previously studied lake trout populations (e.g., Stewart et al. 1981; Luecke et al. 1994; Ruzycki et al. 2003), the population in STUDY AREA Lake Chelan had an irregular stocking history and stable recruit- Lake Chelan is a deep (mean depth = 144 m; maximum ment patterns could not be assumed. An unstable age structure depth = 453 m), glacially formed lake located in the Cascade could decouple the trends of harvest and prey consumption, Range of north-central Washington (48◦N, 120◦W; Figure 2). potentially masking a strong predation interaction. The lake is long and narrow (length = 81 km; maximum width Do lake trout interact weakly with kokanee in Lake Chelan, <3 km) and is composed of two basins: Lucerne Basin in or can strong latent impacts explain the recent kokanee de- the northwest and Wapato Basin in the southeast (Kendra and cline? Our objectives were to (1) quantify the rate of lake Singleton 1987). The lake is ultra-oligotrophic (total phospho- trout predation on kokanee at the beginning of the decline in rus averages 3.2 µg/L) and monomictic. Native fish species 2005; (2) determine the impact of this predation rate by com- include the bridgelip sucker Catostomus columbianus, burbot paring it with the production rate and biomass of the kokanee Lota lota, largescale sucker Catostomus macrocheilus, northern population in 2005; and (3) determine whether the timing of pikeminnow Ptychocheilus oregonensis, peamouth Mylocheilus longer-term trends in lake trout and kokanee population dynam- caurinus, slimy sculpin Cottus cognatus, threespine stickleback Downloaded by [Department Of Fisheries] at 20:00 25 September 2012
FIGURE 2. Map of Lake Chelan, north-central Washington, showing the two lake basins, principal sampling sites (asterisks), and hydroacoustic transects (lines). 1194 SCHOEN ET AL.
Gasterosteus aculeatus, and westslope cutthroat trout O. clarkii two depths in the hypolimnion during late summer. Kokanee lewisi. Native bull trout Salvelinus confluentus were extirpated were sampled opportunistically by angling and with horizon- from the lake in approximately 1950. Many nonnative fish and tal midwater gill nets and large “curtain” gill nets (Beauchamp invertebrate species have been introduced into the lake, primar- et al. 2009). ily to enhance sport fisheries; such species include lake-resident Fork length (FL; mm), wet weight (g), and sex of captured Chinook salmon O. tshawytscha, kokanee, lake trout, opossum fish were recorded in the field. Lake trout stomachs were col- shrimp Mysis diluviana, rainbow trout O. mykiss, and small- lected and frozen immediately, and gonads were weighed to mouth bass Micropterus dolomieu (Brown 1984; Wydoski and determine reproductive investment. For age and growth analy- Whitney 2003). Chinook salmon, kokanee, rainbow trout, and sis, we collected opercles and otoliths from lake trout and scales westslope cutthroat trout are currently stocked annually. from kokanee. Vertical thermal profiles were collected with a Kokanee were introduced into Lake Chelan in 1917 and have Hydrolab Datasonde (Hach Environmental) at each sampling supported a popular fishery for decades (Brown 1984; Hagen site. 1997; Duke Engineering and Services 2000). Most kokanee Hydroacoustic surveys.—To quantify the biomass of koka- in Lake Chelan exhibit a 3- or 4-year egg-to-egg life cycle nee, a hydroacoustic survey was conducted during moonless (Truscott and Peven 1988; Peven 1989, 1990). Over 90% of nights on 30–31 August 2005, coinciding with late-summer kokanee spawning takes place in the Stehekin River and its thermal stratification when schooling behavior was minimized tributaries at the north end of the lake during September and (Luecke and Wurtsbaugh 1993). The survey consisted of 22 October (Peven 1990; Keesee and Keller 2012). The kokanee transects in a zigzag pattern (Figure 2). We stratified the sur- population declined substantially during the late 1970s after vey into three ecologically distinct lake regions: the Stehekin the introductions of M. diluviana and Chinook salmon (Brown River area (2 transects), the remainder of the Lucerne Basin 1984), but it recovered during the 1980s and 1990s. Spawner (16 transects), and the Wapato Basin (4 transects). All tran- surveys have been conducted in major tributaries annually since sects were conducted in the pelagic portion of the lake (areas 1981. Escapement is estimated using the area-under-the-curve with water depths of at least 15 m), which comprised over 90% method (Neilson and Geen 1981), assuming a spawner residence of the total lake surface area. Hydroacoustic sampling was con- time of 15 d (Brown 1984). These surveys are considered to ducted from a 7-m boat with a 200-kHz echosounder (Model DE indicate the escapement trend but not the complete number of 6000; BioSonics). A 6.7◦ split-beam transducer was mounted spawners (CCPUD 2007). Harvest removes only a small fraction on a tow body facing downward at a depth of 1 m and was of kokanee escapement (Brown 1984). Kokanee are stocked towed at 8–10 km/h. Data were acquired using a minimum target directly into the lake at the Wapato Basin, but it is unclear strength threshold of −55 dB, a 0.4-ms pulse width, and a ping whether these fish contribute to the fishery or to the spawning rate of 1 ping/s. Data were analyzed using Echoview version population (Duke Engineering and Services 2000). 4.2 (Myriax). Kokanee density was estimated by echo counting single acoustic targets, and density was converted to an estimate of to- METHODS tal biomass by using data on body weight and lake bathymetry. We used field data and a bioenergetics model to quantify The density of targets was relatively low (0–5 targets/1,000 m3), ontogenetic and seasonal patterns in prey consumption by lake and no schools were observed on the echograms. We assumed trout. We compared these consumption values with the esti- that all small (<330 mm FL) pelagic targets were kokanee be- mated kokanee biomass and production rate from a hydroa- cause kokanee comprised 95% of the midwater gill-net catch and coustics survey to determine the impact of lake trout predation because modal sizes of acoustic targets corresponded with the
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 on the kokanee population. Finally, using stocking records and size distribution of kokanee. We converted target strength values a population model, we reconstructed the past abundance and to FLs using Love’s (1971) equation and a total length (TL)–FL age structure of lake trout and then projected these trends into relationship for kokanee (FL = 0.939 × TL; Hyatt and Hubert the future to characterize likely changes in predation pressure 2000). Estimates of FL (mm) were converted to weight (W;g) under different scenarios of natural reproduction. by using a relationship developed from kokanee sampled during Field sampling.—We conducted standardized sampling ev- this study (r2 = 0.99, N = 93, P < 0.0001): ery 3 months during August 2004 through May 2006 to quantify the seasonal diet, distribution, and growth pattern of lake trout. W = 0.00000402 × FL3.20. (1) Lake trout were captured with horizontal sinking gill nets that were fished overnight at five fixed sites (Figure 2). At each sam- We estimated kokanee age from scales to determine the size- pling site, four depth strata were sampled with one small-mesh at-age relationship, and we used size modes of hydroacoustic net (stretched mesh sizes = 2.5, 3.2, 3.8, 5.1, 6.4, and 7.6 cm) targets to assign targets to age-classes (age 0 = 30–100 mm and one large-mesh net (stretched mesh sizes = 8.9, 10.2, 11.4, FL; age 1 = 100–200 mm FL; ages 2–4 = 200–330 mm FL). 12.7, and 15.2 cm). The depth strata (0–15, 15–30, 30–50, and The estimated body weights of individual targets were added 50–70 m) corresponded with the epilimnion, metalimnion, and to determine the biomass of each age-class sampled in 2-m LATENT PREDATION IMPACTS OF LAKE TROUT 1195
depth intervals (from 2 to 200 m) within each transect. Kokanee biomass density (kg/1,000 m3) was calculated for each interval within each transect by dividing the detected biomass by the volume acoustically sampled. These depth-specific volumetric densities were multiplied by the total volume of the depth stra- tum within the corresponding lake region (Kendra and Singleton 1987) and were summed across all depths; the sum was divided by the surface area of the lake region to yield the areal biomass density (kg/ha) for each transect. We calculated the mean areal biomass density of each kokanee age-class in each lake region, expanded these values by the surface area of each region (Table 4 of Kendra and Singleton 1987), and summed them to estimate the total biomass. The production rate (metric tons/year) of the kokanee population was calculated using the instantaneous growth rate method (Ney 1993; Hayes et al. 2007). Diet analysis.—We analyzed lake trout diets as input for the FIGURE 3. Length frequency distribution of lake trout captured with gill nets bioenergetics model. Lake trout stomach contents were identi- in Lake Chelan during August 2004–June 2005 and August 2005–June 2006. fied to species for prey fishes and to order and life stage for invertebrates, and the blotted wet weight of each prey type was and Quinn (1979), recorded. The lengths of prey fish were measured or were esti- −(ω/L∞)t mated from the lengths of diagnostic bones when possible (see FLt = L∞[1 − e ], (2) Schoen and Beauchamp 2010). A subset of salmonid prey spec- imens (n = 21) was unidentifiable to the species level based where FLt is fork length (mm) at age t (years), L∞ is the asymp- on bone morphology; those specimens were analyzed geneti- totic maximum length (mm), and ω is the growth rate of young cally by the Molecular Genetics Facility, School of Aquatic and fish (mm/year); this model was fitted to empirical length and Fishery Sciences, University of Washington. Prey DNA samples age data from the aged fish (n = 188) by using maximum like- were extracted, amplified using polymerase chain reaction, and lihood estimation (Isely and Grabowski 2007). Fork length was sequenced via the methods of Buser et al. (2009), with mod- converted to wet weight (W; g) using a relationship developed ifications described by Schoen and Beauchamp (2010). Phy- from lake trout sampled during this study (r2 = 0.92, N = 504, logenetic relationships were assigned using MEGA4 software P < 0.0001): (Tamura et al. 2007). . We calculated diet proportions by weight (Chipps and Garvey W = 0.00000799 × FL3 07. (3) 2007) for four size-classes of lake trout subdivided by season. Within these groups, diet composition differed between the two We estimated the S of lake trout by using the catch curve method lake basins (Schoen and Beauchamp 2010), so we estimated (Miranda and Bettoli 2007). To satisfy the assumption of stable lakewide diet proportions as the mean of the diet proportions recruitment among years, only the cohorts that were stocked in each basin weighted by the relative abundance of lake trout in roughly equal numbers from 1990 to 2000 were included in in each basin. We assumed that the density of each lake trout this analysis (this assumption is addressed in the Discussion). size-class was proportional to the catch per unit effort (CPUE) The age frequency distribution of captured lake trout was cor-
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 for that size-class in sinking gill nets. We scaled these estimates rected for gill-net size selectivity (Hansen et al. 1997) and for to the area of benthic slope zone habitat available in each basin inequalities in effort among mesh sizes (Ruzycki et al. 2003). (Lucerne Basin: 1,728 ha; Wapato Basin: 1,767 ha) at the 15–70- Per-capita consumption by lake trout.—Per-capita rates of m depths typically occupied by lake trout (Hansen et al. 1995) consumption by lake trout were estimated with a bioenergetics to estimate the proportion of the lake trout population in each model developed by Stewart et al. (1983), with physiological basin. parameters modified by Luecke et al. (1999). Simulations were Lake trout size distribution, growth, and survival.—We used run for lake trout of ages 2–16 by using a daily time step (model field data to characterize the size distribution, growth rate, and day 1 = 1 May). Model inputs included annual growth (weight survival rate (S) of lake trout. Lake trout captured in gill nets at age), seasonal diet composition, the water temperature ex- ranged in size from 182 to 846 mm FL (n = 504), and the size perienced by the consumer (“thermal experience”), the energy distribution showed no obvious change between the 2 years of densities of prey organisms, and energy losses due to spawning. sampling (Figure 3). Lake trout age was estimated by use of op- Growth inputs for lake trout were generated from the age– ercles (Sharp and Bernard 1988) because sagittal otoliths did not length and length–weight relationships derived above (Table 1). exhibit clear annual marks. Growth in FL was characterized by Seasonal diet composition was determined from stomach fitting the von Bertalanffy model parameterization of Gallucci content data (Table 2). For simplicity, prey were grouped into 10 1196 SCHOEN ET AL.
TABLE 1. Growth, size-class, age structure, and proportion of physiological Lake trout population dynamics.—We simulated trends in maximum consumption rate (pCmax) used for bioenergetics simulations of lake lake trout abundance and size structure with a deterministic, trout in Lake Chelan, Washington. age-structured population model. The number of lake trout at Agea Wet Size-classb Age age j in year t (Njt) was projected forward to indicate the number c + + + + (years) weight (g) (FL, mm) structure pCmax of lake trout at age j 1 in year t 1(Nj 1,t 1) according to the instantaneous annual mortality rate Z (Hilborn and Walters 2 103 180–450 251.0 0.80 1992): 3 275 180–450 188.9 0.76 4 515 180–450 142.1 0.72 −Z Nj+ ,t+ = Njte . (4) 5 798 180–450 107.0 0.72 1 1 6 1,100 451–500 80.5 0.65 7 1,404 501–550 60.6 0.61 Age-1 recruits were added to the population through stocking 8 1,696 501–550 45.6 0.61 and natural reproduction. Natural reproductive rates were un- 9 1,966 551–850 34.3 0.45 known, so we bracketed this uncertainty by simulating three sce- 10 2,212 551–850 25.8 0.43 narios representing the range of possibilities: no reproduction, 11 2,432 551–850 19.4 0.43 “replacement” (a reproductive rate that was sufficient to offset 12 2,625 551–850 14.6 0.42 mortality losses over the long term), and a rapid reproductive 13 2,792 551–850 11.0 0.42 rate derived from literature values. For each scenario, the num- S 14 2,937 551–850 8.3 0.41 ber of age-0 fish stocked in each year was multiplied by the to 15 3,061 551–850 6.2 0.41 estimate the number surviving to age 1 in the subsequent year. 16 3,167 551–850 4.7 0.41 For the replacement and rapid reproduction scenarios, spawn- 17 3,256 551–850 ing also contributed to recruitment. The number of naturally spawned age-1 recruits in year t + 1(N1,t + 1) was estimated as aConsumption was estimated for lake trout (ages 2–16) growing in wet weight from + age t to age t 1. Day 1 of the simulations was 1 May. N = F S , bSize-classes were used to assign diet and thermal experience inputs to lake trout in 1,t+1 t 0 (5) the model. Values indicate the size-class on model day 1 at each age. cExpected number of lake trout at each age in a unit population of 1,000 fish (age ≥2) with the observed mortality rate. Numbers at age were adjusted on a daily time step in where Ft is the population fecundity (total number of eggs pro- simulations but are represented here on an annual basis for simplicity. Values indicate the duced) in the previous year t and S0 is egg-to-age-1 survival. numbers at age on model day 1. We used an estimate of individual fecundity (f ; eggs/mature female) reported for an introduced, low-density lake trout pop- ulation (Ruzycki et al. 2003): categories for analysis. The seasonal thermal experience of each 1.48 lake trout size-class was estimated using thermal profiles and f = 0.03 × W , (6) depth distribution patterns (Table 2; Beauchamp et al. 2007). Thermal experiences were calculated separately for each lake where W is female body weight (g). We calculated F in each year basin, and values were pooled via the method used for diet com- as the sum of f for all females in each reproductively mature position. We used prey energy density values from the literature cohort (age ≥7), assuming a sex ratio of 1:1. For the rapid (Table 3) and assumed that prey indigestibility was 3% for fishes reproduction scenario, we used an S0 value of 0.0043 estimated and 17% for invertebrates (Beauchamp et al. 2007). We simu- for low-density lake trout populations (Shuter et al. 1998). For
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 lated spawning losses by reducing body mass by 6.8% on 15 the replacement scenario, we iteratively adjusted S0 downward November for lake trout larger than 400 mm FL (Stewart et al. until the projected long-term annual population growth rate (λ) 1983; E.R.S., unpublished data). was equal to 1, meaning that natural reproduction was exactly To compare consumption rates among seasons and size- sufficient to balance mortality over time. classes of lake trout during the focal 2004–2006 period, we Under each reproduction scenario, we simulated the number also took the observed age structure into account. Daily esti- of lake trout at each age (ages 1–30) in each year from 1980 mates of consumption by individual lake trout were expanded to 2015. Although the population was young during our 2004– into aggregate seasonal and annual consumption estimates for 2006 sampling period (oldest estimated age = 19 years), we an age-structured population unit of 1,000 lake trout (ages 2– assumed that as the population matured, older age-classes would 16), with the proportion of individuals at each age determined by be represented in later years according to the observed mortality the observed S (Table 1). The number of juvenile lake trout con- rate. To predict how population trends would influence lake trout sumed by piscivorous lake trout was estimated as the consumed harvest rates, we estimated the number of lake trout that were biomass divided by the geometric mean weights of age-1 and vulnerable to harvest by using a length at 50% vulnerability (Lc) age-2 lake trout (i.e., weights estimated by the growth model). relationship developed for lake trout fisheries that were free of LATENT PREDATION IMPACTS OF LAKE TROUT 1197
TABLE 2. Seasonal thermal experience and diet composition of four lake trout size-classes in Lake Chelan, Washington. Diet proportions (by weight) are shown for prey taxa; proportions of 0.10 or more are indicated in bold italics (BUR = burbot; CHS = Chinook salmon; CYP = cyprinids; KOE = kokanee; LKT = lake trout; THS = threespine stickleback; UNS = unidentified salmonids; OTF = other fish; MYS = Mysis diluviana;OTI= other invertebrates).
Diet proportion (by weight) Size-class Thermal n (nonempty (FL; mm) Month experience (◦C) stomachs) BUR CHS CYP KOE LKT THS UNS OTF MYS OTI 180–450 Feb 5.3 10 0 0 0 0 0 0.032 0 0 0.968 0 May 6.8 16 0 0 0.483 0 0 0.066 0 0 0.447 0.004 Aug 11.1 9 0.065 0 0 0 0 0.037 0 0.037 0.759 0.102 Nov8.729000000.184 0 0.048 0.762 0.006 451–500 Feb 5.3 22 0 0 0.016 0.043 0 0.018 0 0 0.896 0.027 May 7.0 33 0 0 0.731 0 0.014 0 0 0 0.252 0.003 Aug11.01500000000.153 0.844 0.002 Nov 9.4 27 0 0 0 0.039 0 0.270 0 0.015 0.650 0.026 501–550 Feb 5.4 12 0 0 0 0.027 0 0.001 0 0.001 0.970 0.002 May 7.1 27 0 0 0.502 0 0 0.002 0 0.323 0.149 0.025 Aug 10.5 10 0 0.018 0 0.082 0 0 0.020 0.003 0.842 0.035 Nov 9.8 23 0 0 0.052 0.076 0 0.375 0 0.043 0.444 0.009 551–850 Feb 5.5 13 0 0 0.003 0.912 0 0.008 0 0.003 0.068 0.007 May 7.3 30 0 0 0.811 0.135 0 0 0 0.018 0.030 0.005 Aug 9.8 6 0.084 0 0 0 0.840 0 0 0 0.075 0 Nov 10.0 9 0 0 0 0.133 0.006 0.224 0 0.002 0.634 0
size restrictions on harvest (Shuter et al. 1998), Lake trout harvest trends.—We analyzed harvest records from the lake trout fishery to characterize trends in CPUE. The 0.421 0.669 Lc = 0.853ω L∞ . (7) lake trout fishery was confined to the Wapato Basin and a small adjacent portion of the Lucerne Basin. Four of the five primary For simplicity, we assumed that lake trout experienced a knife- charter guides on Lake Chelan provided harvest records for a edged transition to full vulnerability upon reaching Lc.Wepre- subset of their trips during 2004–2007. Harvest records con- sented the population model results as the total population size sisted of standardized questionnaire forms completed by guided (including fish ≥age 1), the number of lake trout that were anglers indicating the duration of the charter (full day or half vulnerable to harvest (fish ≥age 5), and the biomass of large day) and the number of each fish species harvested. Harvest piscivorous lake trout (fish ≥age 9). We intended these simula- records with usable data were collected on 445 trips, including tions to illustrate the potential range of population trajectories 53% of all trips by the participating guides during 2005–2007— given the known demographic constraints rather than to predict the years for which the total number of trips was known—and the true values exactly. records were well distributed across seasons and years. We
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 TABLE 3. Estimated energy density (J/g wet weight) of lake trout prey organisms.
Prey taxon Surrogate taxon Energy density (J/g) Reference Burbot 5,125 Johnson et al. (1999) Chinook salmona 5,863 Stewart and Ibarra (1991) Cyprinids Peamouth 7,093 Mazur (2004) Kokaneea Sockeye salmon 6,008 Beauchamp et al. (1989) Lake trouta 6,009 Stewart et al. (1983) Threespine stickleback 6,949 Mazur (2004) Unidentified salmonidsa Sockeye salmon 6,008 Beauchamp et al. (1989) Other fish Sculpin 4,178–4,514b Mazur (2004) Mysis diluviana (formerly M. relicta) 2,976–3,720b Lasenby (1971); Adare and Lasenby (1994) Other invertebrates Crayfish 3,318 Mazur (2004)
aEstimated for a prey weight of 100 g. bEnergy density varied seasonally within the specified range. 1198 SCHOEN ET AL.
TABLE 4. Biomass (kg) of kokanee in three regions of Lake Chelan as estimated with a hydroacoustics survey during August 2005.
Age-0 biomass Age-1 biomass Age-2–4 biomass Lake region Surface area (ha) n transects Mean SE Mean SE Mean SE Stehekin River area 704 2 332 234 2,002 1,215 20,086 14,837 Lucerne Basin 9,281 16 415 48 2,676 573 14,066 2,689 Wapato Basin 3,502 4 75 16 613 56 2,844 294 Total 13,486 22 823 67 5,291 562 36,996 4,043
calculated the annual mean CPUE, with effort defined as the Lake Trout Diet, Growth, and Survival number of full-day equivalent charters per year. Mean harvest The diets of lake trout shifted from M. diluviana to fish as on half-day trips was 57.7% of harvest on full-day trips, so we lake trout grew, and salmonid prey were particularly important counted half-day trips as 0.577 d of effort. for the largest lake trout (Table 2). Mysis diluviana represented Predation impacts of lake trout on kokanee.—To estimate the 73% of the annual diet (by weight) for the smallest lake trout impact of lake trout predation on kokanee in 2005 (the year of the (180–450 mm FL) but only 20% of the prey consumed by the hydroacoustics survey), we expanded the per-capita consump- largest lake trout (551–850 mm FL). All size-classes of lake tion of kokanee for each lake trout age-class by the abundance trout consumed substantial proportions of cyprinids during May of that age-class during 2005 as estimated by the population before thermal stratification forced lake trout into deep water, model. We compared this population-level consumption esti- which spatially segregated them from cyprinid prey (Table 2). mate with the estimates of kokanee biomass and production After growing larger than 450 mm FL, lake trout began con- from the hydroacoustics survey. The comparison was restricted suming kokanee and other salmonids, and kokanee comprised to the biomass and production of age-1 and older kokanee be- 30% of the annual diet for the largest lake trout. Kokanee were cause lake trout diets contained almost no age-0 kokanee. most prevalent in diets of the largest lake trout during February Finally, to illustrate the likely consequences of lake trout pop- (91% of the diet; Table 2). Over 90% of kokanee in lake trout ulation dynamics for kokanee, we simulated the trend in lake diets were age 1 or older (>100 mm FL; Figure 4). The largest trout predation pressure on kokanee over time. The annual per- lake trout also consumed juvenile lake trout (ages 1–2), mostly capita kokanee consumption by each lake trout age-class was during August (Table 2). The lengths of all ingested prey fishes expanded by the abundance of that age-class during each year were no more than 41% of lake trout lengths (n = 62; Figure 4). from 1980 to 2015 as estimated by the population model. We interpreted this projected trend as a relative index of predation pressure (i.e., lake trout demand for kokanee prey) because pre- dicting actual consumption rates would require yearly diet com- position data, which were only available for 2004–2006. Sta- tistical analyses were performed using R version 2.10.0 (Ihaka and Gentleman 1996).
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 RESULTS Kokanee Biomass and Production Lake Chelan supported a kokanee biomass of approximately 43.1 metric tons during August 2005 based on a quantitative hydroacoustic survey of the lake. Age-0 kokanee accounted for less than 2% of this biomass (mean ± SE = 823 ± 67 kg). Age-1 kokanee represented 12% of the total biomass (5.29 ± 0.56 metric tons); age-2 and older kokanee represented 86% of the total (37.0 ± 4.0 metric tons). Densities of all kokanee age- classes were greatest near the Stehekin River, and lower densi- FIGURE 4. Relationship between the fork length of piscivorous lake trout ties were distributed throughout the rest of the lake (Table 4). and the lengths of ingested prey fishes (total length for burbot and threespine Production by the kokanee population was an estimated 32.3 sticklebacks; fork length for all other species) in Lake Chelan. Symbols indicate metric tons/year, including 22.2 metric tons/year by age-1 and prey species, which also included lake trout. Prey fish length was no more than older kokanee. 41% of predator length (n = 62). LATENT PREDATION IMPACTS OF LAKE TROUT 1199
Lake Trout Population Dynamics and Harvest Trends The simulated trend in total lake trout abundance declined under the no-reproduction scenario but increased under the re- placement and rapid reproduction scenarios after the 2004–2006 sampling period (Figure 6A). Lake trout were originally in- troduced into the lake during 1980–1982, but under all three reproduction scenarios these early cohorts and their offspring were far outnumbered by the cohorts that were stocked during the 1990s. After initial stocking in 1980–1982, the simulated lake trout population (age ≥1) remained low during the 1980s. The simulated population grew substantially with heavy stock- ing during the 1990s, and population growth slowed in all sce- narios after stocking ceased. Under the no-reproduction sce- nario, abundance reached a maximum in 2001 and then declined (λ = 0.75). Under the replacement scenario, lake trout numbers declined from 2001 to 2005 and then increased and oscillated towards equilibrium. Under the rapid reproduction scenario, population growth slowed from 2001 to 2004 and returned to FIGURE 5. Seasonal prey consumption by a size-structured population unit λ> of 1,000 lake trout in Lake Chelan. Consumption was estimated by using a rapid annual growth ( 1.1) in 2005. bioenergetics model and is summarized for four lake trout size-classes (180– The simulated abundance of lake trout that were vulnera- 450, 451–500, 501–550, and 551–850 mm fork length). ble to harvest declined under the no-reproduction and replace- ment scenarios but increased under the rapid reproduction sce- nario after the 2004–2006 sampling period (Figure 6B). These ω The for lake trout was 0.186 mm/year, and the L∞ was trends followed the trends in total abundance after a develop- = 671 mm. The Z for lake trout was 0.2843 (n 123 fish of ages mental lag as successive cohorts became vulnerable to anglers. 7–12, r2 = 0.55), corresponding to an S of 75%. The predicted Lc was 412 mm FL, corresponding to an age of 5.1 years. The number of lake trout that were vulnerable Age-Structured Rates of Consumption by Lake Trout to harvest increased substantially from 1994 to 2005 under all To compare consumption rates among size-classes of lake scenarios, but the trajectories diverged after the last stocked co- trout and among seasons during the focal 2004–2006 period, hort became vulnerable: harvestable lake trout increased under we expanded per-capita consumption rates according to the ob- the rapid reproduction scenario but declined under both the re- served age structure. An age-structured population unit of 1,000 placement and no-reproduction scenarios (Figure 6B). Based on lake trout (ages 2–16) consumed an estimated 3,419 kg of prey harvest records, lake trout CPUE in the charter fishery declined annually, including 2,079 kg of M. diluviana and 1,264 kg of from 12.6 ± 0.6 fish/full-day charter (mean ± SE) in 2004 to fish (Figure 5). Less than 30% of the fish biomass consumed 8.7 ± 0.5 fish/full-day charter in 2005, 8.8 ± 0.5 fish/full-day was salmonid prey, including 177 kg of kokanee, 164 kg of charter in 2006, and 7.2 ± 0.3 fish/full-day charter in 2007 lake trout, 1.5 kg of Chinook salmon, and 1.7 kg of unidentified (Figure 6B). salmonids. The consumed biomass of juvenile lake trout prey The simulated biomass of large, highly piscivorous lake trout represented approximately 4,038 age-1 lake trout or 979 age-2 (>550 mm FL; age ≥9.2) expanded rapidly under all reproduc-
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 lake trout consumed annually per 1,000 adult lake trout (i.e., tion scenarios beginning in 1999 and continued to grow after the if predation was focused solely on ages 1 or 2). The smallest 2004–2006 sampling period as successive stocked cohorts grew lake trout size-class was responsible for 52% of total prey con- into the size range at which they were effective kokanee preda- sumption but only 33% of fish consumption. The largest lake tors (Figure 6C). From 2004 to 2009, the biomass of large lake trout size-class (551–850 mm FL) was responsible for 83% of trout increased 70% under the no-reproduction scenario, 201% predation on kokanee, despite representing only 13% of the lake under the replacement scenario, and 296% under the rapid re- trout population numerically. production scenario. After 2009, the simulated biomass of large Lake trout showed strong seasonal patterns in prey consump- lake trout began to decline under the no-reproduction and re- tion (Figure 5). Consumption rates were generally greatest dur- placement scenarios but continued to increase under the rapid ing July–December, when lake trout experienced higher water reproduction scenario. temperatures (Table 2). Seasonal changes in predation on differ- ent prey taxa mirrored seasonal diet composition patterns, with Predation Impacts of Lake Trout on Kokanee lake trout preying heavily on kokanee during January–March, Lake trout consumed a large proportion of the kokanee cyprinids during April–June, and smaller lake trout during July– standing stock biomass and population production in 2005. September. Population-level lake trout consumption of kokanee ranged from 1200 SCHOEN ET AL. Downloaded by [Department Of Fisheries] at 20:00 25 September 2012
FIGURE 6. Simulated population dynamics of lake trout under scenarios of no reproduction, replacement, and rapid reproduction: (A) number of lake trout stocked (bars) and abundance of age-1 and older lake trout (lines); (B) number of lake trout that were vulnerable to harvest (age ≥5) and lake trout catch per unit effort (CPUE) by anglers (mean ± SE); (C) biomass of the largest lake trout size-class (>550 mm fork length; age ≥9), which preyed most heavily on kokanee; and (D) an index of lake trout predation pressure on kokanee and an index of kokanee escapement (data from Keesee and Keller 2012). The shaded bar across all panels indicates the years of field sampling, 2004–2006. LATENT PREDATION IMPACTS OF LAKE TROUT 1201
an estimated 33.9 metric tons (no-reproduction scenario) to 46.8 kokanee productivity through behavioral changes (Werner and metric tons (rapid reproduction scenario), with an intermediate Peacor 2003; Hardiman et al. 2004). value (36.3 metric tons) under the replacement scenario. Con- The trends and timing of lake trout population dynamics were sumption estimates exceeded (153–211%) the estimated annual consistent with strong predation impacts on kokanee during production rate of age-1 and older kokanee and represented 53– their population decline in 2005–2009. The population model 73% of the production plus standing stock biomass of these predicted a 70–296% increase in the biomass of the largest lake kokanee in 2005. trout size-class from 2004 to 2009, driving predation pressure The index of simulated lake trout predation pressure on koka- on kokanee to all-time peak levels. The trends in large lake trout nee generally mirrored the trend of simulated large lake trout biomass and predation pressure were predicted to lag behind biomass (Figure 6C, D). Under all scenarios, simulated preda- the trend in lake trout harvest by roughly 4 years. These trends tion pressure increased roughly sixfold from 1998 to 2006. Un- were consequences of the pulsed stocking history of lake trout, der the no-reproduction scenario, simulated predation pressure observed growth and mortality rates, and three scenarios rep- reached a plateau during 2006–2008 and began to decline sub- resenting the range of potential reproduction rates. Until 2009, stantially in 2009. Under the replacement scenario, simulated the projected trends were similar for all three reproductive sce- predation pressure rose to a peak in 2009 and declined for the narios because the largest lake trout size-class was dominated next 6 years. Under the rapid reproduction scenario, simulated by stocked fish. After 2009, the trajectories of the scenarios predation pressure increased continuously. All three scenarios diverged, indicating that the biomass of large lake trout and pre- predicted that predation pressure was greater during the kokanee dation pressure could continue to increase, could stabilize, or decline (2005–2009) than in any previous year (Figure 6D). could decrease depending on natural reproductive rates. For this reason, quantifying the natural recruitment of lake trout in Lake DISCUSSION Chelan should become a critical priority. Although the population analysis depended on several impor- Impacts of Lake Trout Predation on Kokanee tant assumptions, we limited the interpretation of model results Lake trout predation likely contributed substantially to the to key conclusions that were strongly supported by data. We 88% decline in kokanee escapement between 2005 and 2009. primarily used model output to project the broad trends and Our simulations indicated that in 2005 (the year in which koka- timing of population dynamics and predation pressure, and we nee were acoustically surveyed), lake trout consumed a large only used numerical estimates of abundance in further analyses proportion of kokanee production and biomass, and predation for the model year 2005, when field data were collected (e.g., pressure reached its greatest level since lake trout were intro- diet composition and thermal experience). The mortality analy- duced into the lake. Predation pressure was projected to remain sis likely produced conservative estimates of both the predatory at or above this level through at least 2009. Despite abundance inertia of lake trout and the impacts on kokanee. Although catch and harvest trends that suggested a synchronous rise and fall curve analysis relies on the assumption of equal recruitment of predator and prey populations, this analysis suggested that among years, stocking rates actually increased slightly (mean predation caused strong and growing impacts on the declining annual increase = 4.3%) during the 1993–1999 period corre- prey population. sponding to the cohorts in the analysis. Natural reproduction also The bioenergetics analysis revealed that lake trout predation likely added a small number of recruits to the 1998 and 1999 imposed strong impacts on the kokanee population in 2005. cohorts. This inflated the relative abundance of young fish in the Annual lake trout consumption of kokanee exceeded the annual catch and led us to slightly overestimate mortality. Therefore, the production of age-1 and older kokanee and claimed more than model probably underestimated the demographic importance of
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 half the sum of annual production and standing stock biomass older, more piscivorous lake trout in the simulations, resulting in of these kokanee cohorts. This predation rate was likely unsus- conservative estimates of predation on salmonids and the time tainable (Ney 1990; Beauchamp et al. 2007), suggesting that lag before each cohort achieved its maximum predatory impact. predation contributed to the sharp declines in kokanee escape- ment beginning in 2006. Although they represented less than Unstable Predator Age Structure Produces 15% of the population, lake trout belonging to the largest size- Unexpected Dynamics class (>550 mm FL) were responsible for most (83%) of the The no-reproduction and replacement scenarios of the pop- predation on kokanee during the focal sampling period of 2004– ulation model illustrated a counterintuitive possibility: that the 2006. This was consistent with studies from other lakes, which numbers of harvestable lake trout may have declined during the have found that although the largest lake trout are relatively few mid- to late 2000s while the predation pressure on kokanee ac- in number, these large individuals consume more salmonid prey tually rose. This was possible because of high predatory inertia than do smaller lake trout (Johnson and Martinez 2000; Ruzycki combined with an unstable age structure due to the irregular et al. 2003; Beauchamp et al. 2007). The full impacts of lake stocking history of lake trout. The 11 years of heavy stocking trout in Lake Chelan were likely even greater because we did produced a “baby boom” effect, with a pulse of strong, aging not estimate nonconsumptive effects, which may have reduced cohorts dominating the demographics of the population. Each 1202 SCHOEN ET AL.
cohort became vulnerable to harvest roughly 4 years before et al. 2008). Therefore, lake trout in Lake Chelan appeared to achieving full piscivory, causing the harvestable population to reproduce at a slower rate than the fastest-growing populations peak and begin declining while predation pressure continued reported in other studies, but either moderate positive or negative to rise. An important consequence of this prediction is that population growth was possible. the latent strong predation interaction was initially concealed from researchers and managers because harvest was more eas- Disentangling Predation from Confounding Factors ily monitored than predation, and lake trout harvest varied syn- Aside from predation by lake trout, other factors may also chronously with kokanee escapement. have contributed to the kokanee decline. Floods in key spawning Consistent with this prediction of the population model, two streams during October 2003 probably limited kokanee returns lines of empirical evidence suggest that the number of har- during 2006 and 2007 (Keesee and Keller 2012). A 1-year sus- vestable lake trout did in fact decline during the mid- to late pension of kokanee stocking in 2006 may also have contributed 2000s. First, the lake trout CPUE for the charter fishery declined to low escapements in 2008 and 2009; however, hatchery koka- by over 40% from 2004 to 2007. Informal reports indicate that nee fry are heavily consumed by lake trout immediately after catch rates continued to decline until 2008 before increasing stocking in Lake Chelan (E.R.S., unpublished data), and it is un- during 2009–2011 (A. Jones, personal communication). Trends clear whether hatchery kokanee contribute to the spawning pop- in CPUE for lake trout fisheries are often hyperstable—biased ulation (Duke Engineering and Services 2000). Brown (1984) in a positive direction relative to actual abundance trends— estimated that harvest reduced kokanee escapement by less than because angler expertise generally improves over time (Shuter 5% in 1982; if harvest remained at this order of magnitude, it et al. 1998). Thus, the harvestable population may have actu- was unlikely to cause the recent decline. We did not estimate ally declined faster than suggested by the catch rate. Second, kokanee harvest in this study because many unguided anglers a comparison of gill-net catch curves suggested that lake trout participated in the kokanee fishery and our data only covered smaller than 400 mm FL were underrepresented by roughly the charter fleet. It is possible that harvest limits kokanee in 50% in the Lake Chelan population relative to an established years of low abundance, especially if compensatory growth in- self-sustaining population that was sampled with similar meth- creases body size and vulnerability to harvest (Martinez and ods in Lake Tahoe, California–Nevada (Thiede 1997; E.R.S., Wiltzius 1995; Rieman and Maiolie 1995). Previous studies in- unpublished data). This pattern provides evidence that natural dicated that during the 2005–2009 decline, kokanee were not reproduction in Lake Chelan did not produce as many recruits strongly limited by food supply, competition with M. diluviana during the early 2000s as necessary to replace the cohorts that (Schoen 2007), or predation from other piscivores such as north- were stocked during the 1990s. These weaker cohorts would be ern pikeminnow or Chinook salmon (Schoen and Beauchamp expected to have reached harvestable size during 2005–2009. 2010). The bioenergetics analysis revealed a substantial rate of can- Disentangling the impacts of high predatory inertia and con- nibalism by lake trout, which can limit recruitment in fluc- founding factors is an important step towards prioritizing lim- tuating, age-structured predator populations (Wissinger et al. ited resources for conservation. The complementary analyses 2010). presented here offer a useful approach for overcoming these Of the three scenarios bracketing the range of potential lake challenges. Population dynamics analysis predicted increasing trout reproductive rates, the replacement scenario appeared to predation pressure, but because of confounding events and the yield the best match to observed trends. The no-reproduction past volatility of kokanee escapement, this was insufficient to and replacement scenarios both predicted the observed decline determine whether predation contributed to the kokanee de- of harvestable lake trout during the mid- to late 2000s, as noted cline. Bioenergetics analysis predicted that the absolute impacts Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 above. However, lake trout were observed spawning in Lake of predation were strong during a critical snapshot in time (i.e., Chelan as early as 2000 (Duke Engineering and Services 2000), at the beginning of the decline), but the analysis did not pro- and we captured a number of lake trout that were too young vide any long-term context or predictive power outside of this to have been stocked. Thus, we considered the no-reproduction period. Conclusions based on the bioenergetics results alone scenario to be an extreme lower bound rather than a likely out- would have been particularly short-lived for this system be- come. Conversely, the constantly increasing trend in harvestable cause of the unstable age distribution of predators. Combined lake trout numbers predicted by the rapid reproduction scenario use of the population dynamics and bioenergetics analyses re- did not fit the observed decline in catch rates. This scenario used vealed strong and growing predation impacts during the kokanee the fecundity rate reported for a rapidly expanding lake trout decline. population in Yellowstone Lake (Ruzycki et al. 2003), which was near the low end of a range of lake trout fecundity estimates for other low-density, fast-growing populations (Shuter et al. Could Prey Switching Reduce Impacts on Kokanee? 1998). The rapid reproduction scenario predicted a mean λ of We interpreted the rapidly increasing biomass of large lake 1.13 after stocking ceased, well below a value recently reported trout during 1999–2009 as evidence of increasing predation for lake trout in Lake Pend Oreille, Idaho (λ = 1.63; Hansen on kokanee. However, lake trout are opportunistic predators LATENT PREDATION IMPACTS OF LAKE TROUT 1203
and may instead have switched to other prey as the kokanee risk, given the many examples of ecological impacts exerted by population declined from its peak abundance after the 2004– nonnative lake trout and the time lags associated with detecting 2006 diet sampling period. Evidence for such diet switching in and successfully reversing those impacts. Due to unanticipated other systems is limited. Lake trout in Lake Tahoe have shown surges in lake trout density and high predatory inertia, koka- great interannual variability in the proportion of fish in their diet nee were extirpated from Flathead Lake and from Priest Lake, (Richards et al. 1991), and they consumed more kokanee in years Idaho, before managers could act (Beattie and Clancey 1991; when kokanee densities were greater (Thiede 1997). However, Bowles et al. 1991; Ellis et al. 2011). Although more gradual lake trout in the Great Lakes sustained high consumption rates prey declines allowed time for predator suppression in Yel- across a 100-fold range of prey fish densities, suggesting that lowstone Lake, Wyoming, and Lake Pend Oreille, several years lake trout can severely reduce fish populations without switching elapsed before these programs were developed and implemented to alternative prey (Eby et al. 1995). Similarly, lake trout diets (Bigelow et al. 2003; Hansen et al. 2008; Martinez et al. 2009). in Flathead Lake, Montana, and bull trout and rainbow trout Less-intensive precautionary steps, such as cessation of stock- diets in Lake Pend Oreille continued to be dominated by koka- ing and removal of harvest limits, are common (Martinez et al. nee despite 80–90% declines in kokanee density in each lake 2009); however, our results demonstrate that these steps can be (Clarke et al. 2005; Beauchamp et al. 2007). Thus, while prey insufficient for reducing predation pressure on a 5–10-year time switching may have somewhat reduced lake trout per-capita pre- scale, even if reproductive rates are moderate or low. dation on kokanee, this effect was uncertain and very unlikely Understanding the processes that control the strength and to compensate for the substantial increase in biomass of large timing of impacts from long-lived predators is an important lake trout between 1999 and 2009. In the absence of long-term step toward effectively managing affected ecosystems. Past co- diet data (e.g., Eby et al. 1995; Scheuerell et al. 2005), piscivore existence or synchronized abundance trends of predators and foraging models (Breck 1993; Beauchamp et al. 1999; Mazur prey should not be interpreted as evidence of a weak preda- and Beauchamp 2006) could predict whether a prey switching tion interaction without careful consideration of predator de- threshold for lake trout is likely in this system by comparing mography and dietary ontogeny. Managers should consider the the expected energy gains associated with alternative foraging implementation of monitoring programs that allow early de- strategies. tection of changing impacts; such programs include regular assessment of predator recruitment and in-lake monitoring of Management Implications prey fishes with hydroacoustics or gill nets. Advance knowledge Aside from its predation impacts, the lake trout population of predator demographics, distribution, and spawning locations in Lake Chelan has become a valuable resource that supports may facilitate rapid and effective removal programs if these a popular fishery. If management goals include sustaining this become necessary (Ruzycki et al. 2003; Hansen et al. 2010; fishery, then unchecked population growth is undesirable be- Dux et al. 2011). Forward-looking field sampling and model- cause the body size of lake trout is likely to decline dramatically ing analyses are clearly advantageous for managers who wish to without abundant kokanee prey (Bowles et al. 1991; Stafford maintain introduced piscivores in long-term “balance” with their et al. 2002; Martinez et al. 2009). We recommend assessment prey by detecting and reversing incipient changes in predation of natural lake trout recruitment, which will strongly influence impacts. future predation rates as the pulse of stocked cohorts is replaced. If the population is growing more than a decade after stocking ceased, then lake trout reduction measures may be necessary ACKNOWLEDGMENTS to sustain kokanee, restore native westslope cutthroat trout, or This work was funded by the U.S. Geological Survey; the
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 maintain the trophy value of the lake trout fishery. Alternatively, Chelan County Public Utility District; the School of Aquatic if reproduction is slower than past stocking rates, then the lake and Fishery Sciences, University of Washington; and the Lake trout catch rate may continue to decline before stabilizing at a Chelan Sportsman’s Association. A. Buettner, C. Ekblad, M. reduced level as the strong stocked cohorts are eventually lost Grassley, B. Long, E. Lowery, C. Menard, J. Mitts, S. Piersza- from the population. lowski, C. Sergeant, and many others provided field and labo- When introduced predatory game fishes support valuable ratory assistance. T. Buser, L. Hauser, and I. Jimenez-Hidalgo fisheries, this presents a challenging management dilemma. Al- conducted the genetic analysis. Samples and invaluable local lowing these populations to expand incurs a risk of uncertain knowledge were provided by A. Jones, F. Clark, J. Heinlen, but potentially serious harm to prey and ecosystem processes, M. Lippincott, M. Polacek, and A. Viola. Comments from E. while piscivore reduction programs impose certain monetary Duffy, A. Hansen, T. Essington, E. Lowery, P.Martinez, J. McIn- and social costs on established recreational fisheries (Eby et al. tyre, and two anonymous reviewers improved the manuscript. 2006; Martinez et al. 2009). Given these incentives, indecision The Washington Cooperative Fish and Wildlife Research Unit is a tempting choice for managers (Walters and Martell 2004: is jointly sponsored by the U.S. Geological Survey; the Uni- page 10), who may choose to wait and see how severe the preda- versity of Washington; the Washington Departments of Ecol- tion impacts become. However, this approach carries substantial ogy, Fish and Wildlife, and Natural Resources; and the Wildlife 1204 SCHOEN ET AL.
Management Institute. Reference to trade names does not imply Dextrase, A., and N. E. Mandrak. 2006. Impacts of alien invasive species on endorsement by the U.S. Government. freshwater fauna at risk in Canada. Biological Invasions 8:13–24. Duke Engineering and Services. 2000. Lake Chelan fisheries investigation. Re- port to Chelan County Public Utility District 1, Lake Chelan Hydroelectric REFERENCES Project 637, Bellingham, Washington. Adare, K. I., and D. C. Lasenby. 1994. Seasonal changes in the total lipid content Dux, A. M., C. S. Guy, and W. A. Fredenberg. 2011. Spatiotemporal distribution of the opossum shrimp, Mysis relicta (Malacostraca: Mysidacea). Canadian and population characteristics of a nonnative lake trout population, with im- Journal of Fisheries and Aquatic Sciences 51:1935–1941. plications for suppression. North American Journal of Fisheries Management Beattie, W. D., and P. T. Clancey. 1991. Effects of Mysis relicta on the zoo- 31:187–196. plankton community and kokanee population of Flathead Lake, Montana. Eby, L. A., W. J. Roach, L. B. Crowder, and J. A. Stanford. 2006. Effects of Pages 39–48 in T. P. Nesler and E. P. Bergersen, editors. Mysids in fish- stocking-up freshwater food webs. Trends in Ecology and Evolution 21:576– eries: hard lessons from headlong introductions. American Fisheries Society, 584. Symposium 9, Bethesda, Maryland. Eby, L. A., L. G. Rudstam, and J. F. Kitchell. 1995. Predator responses to prey Beauchamp, D. A., C. M. Baldwin, J. L. Vogel, and C. P. Gubala. 1999. Es- population dynamics: an empirical analysis based on lake trout growth rates. timating diel, depth-specific foraging opportunities with a visual encounter Canadian Journal of Fisheries and Aquatic Sciences 52:1564–1571. rate model for pelagic piscivores. Canadian Journal of Fisheries and Aquatic Ellis, B. K., J. A. Stanford, D. Goodman, C. P. Stafford, D. L. Gustafson, D. A. Sciences 56(Supplement 1):128–139. Beauchamp, D. W. Chess, J. A. Craft, M. A. Deleray, and B. S. Hansen. 2011. Beauchamp, D. A., D. L. Parrish, and R. A. Whaley. 2009. Coldwater fishes in Long-term effects of a trophic cascade in a large lake ecosystem. Proceedings large standing waters. Pages 97–117 in S. A. Bonar, W. A. Hubert, and D. W. of the National Academy of Sciences of the USA 108:1070–1075. Willis, editors. Standard methods for sampling North American freshwater Gallucci, V. F., and T. J. Quinn II. 1979. Reparameterizing, fitting, and test- fishes. American Fisheries Society, Bethesda, Maryland. ing a simple growth model. Transactions of the American Fisheries Society Beauchamp, D. A., D. J. Stewart, and G. L. Thomas. 1989. Corroboration of 108:14–25. a bioenergetics model for sockeye salmon. Transactions of the American Hagen, J. E. 1997. An evaluation of a trout fishery enhancement program in Fisheries Society 118:597–607. Lake Chelan. Master’s thesis. University of Washington, Seattle. Beauchamp, D. A., D. H. Wahl, and B. M. Johnson. 2007. Predator–prey in- Hansen, M. J., N. J. Horner, M. Liter, M. P. Peterson, and M. A. Maiolie. 2008. teractions. Pages 765–842 in C. S. Guy and M. L. Brown, editors. Analysis Dynamics of an increasing lake trout population in Lake Pend Oreille, Idaho. and interpretation of freshwater fisheries data. American Fisheries Society, North American Journal of Fisheries Management 28:1160–1171. Bethesda, Maryland. Hansen, M. J., C. P. Madenjian, J. H. Selgeby, and T. E. Helser. 1997. Gillnet Bigelow, P., T. Koel, D. Mahony, B. Ertel, B. Rowdon, and S. Olliff. 2003. selectivity for lake trout (Salvelinus namaycush) in Lake Superior. Canadian Protection of native Yellowstone cutthroat trout in Yellowstone Lake, Yel- Journal of Fisheries and Aquatic Sciences 54:2483–2490. lowstone National Park, Wyoming. National Park Service, Water Resources Hansen, M. J., J. W. Peck, R. G. Schorfhaar, J. H. Selgeby, D. R. Schreiner, S. Division, Technical Report NPS/NRWRD/NRTR-2003/314, Fort Collins, T. Schram, B. L. Swanson, W. R. MacCallum, M. K. Burnham-Curtis, and Colorado. G. L. Curtis. 1995. Lake trout (Salvelinus namaycush) populations in Lake Bowles, E. C., B. E. Rieman, G. R. Mauser, and D. H. Bennett. 1991. Effects Superior and their restoration in 1959–1993. Journal of Great Lakes Research of introductions of Mysis relicta on fisheries in northern Idaho. Pages 65–74 21:152–175. in T. P. Nesler and E. P. Bergersen, editors. Mysids in fisheries: hard lessons Hansen, M. J., D. Schill, J. Fredericks, and A. Dux. 2010. Salmonid predator– from headlong introductions. American Fisheries Society, Symposium 9, prey dynamics in Lake Pend Oreille, Idaho, USA. Hydrobiologia 650:85–100. Bethesda, Maryland. Hardiman, J. M., B. M. Johnson, and P.J. Martinez. 2004. Do predators influence Breck, J. E. 1993. Foraging theory and piscivorous fish: are forage fish just big the distribution of age-0 kokanee in a Colorado reservoir? Transactions of the zooplankton? Transactions of the American Fisheries Society 122:902–911. American Fisheries Society 133:1366–1378. Brown, L. G. 1984. Lake Chelan fishery investigations. Chelan County Public Hayes, D. B., J. R. Bence, T. J. Kwak, and B. E. Thompson. 2007. Abundance, Utility District 1 and Washington Department of Game, Olympia. biomass, and production. Pages 327–374 in C. S. Guy and M. L. Brown, Buser,T.,N.Davis,I.Jimenez-Hidalgo,´ and L. Hauser. 2009. Genetic techniques editors. Analysis and interpretation of freshwater fisheries data. American provide evidence of Chinook salmon feeding on walleye pollock offal. North Fisheries Society, Bethesda, Maryland. Pacific Anadromous Fish Commission Bulletin 5:225–229. He, X., and J. F. Kitchell. 1990. Direct and indirect effects of predation on Bystrom,¨ P., J. Karlsson, P. Nilsson, T. Van Kooten, J. Ask, and F. Olofsson. a fish community: a whole-lake experiment. Transactions of the American 2007. Substitution of top predators: effects of pike invasion in a subarctic Fisheries Society 119:825–835. Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 lake. Freshwater Biology 52:1271–1280. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: Carpenter, S. R., J. J. Cole, J. R. Hodgson, J. F. Kitchell, M. L. Pace, D. Bade, choice, dynamics, and uncertainty. Kluwer Academic Publishers, Norwell, K. L. Cottingham, T. E. Essington, J. N. Houser, and D. E. Schindler. 2001. Massachussetts. Trophic cascades, nutrients, and lake productivity: whole-lake experiments. Hyatt, M. H., and W. A. Hubert. 2000. Proposed standard-weight (Ws) equations Ecological Monographs 71:163–186. for kokanee, golden trout and bull trout. Journal of Freshwater Ecology CCPUD (Chelan County Public Utility District). 2007. Lake Chelan fishery 15:559–563. plan. CCPUD, Wenatchee, Washington. Ihaka, R., and R. Gentleman. 1996. R: a language for data analysis and graphics. Chipps, S. R., and J. E. Garvey. 2007. Assessment of diets and feeding patterns. Journal of Computational and Graphical Statistics 5:299–314. Pages 473–514 in C. S. Guy and M. L. Brown, editors. Analysis and inter- Isely, J. J., and T. B. Grabowski. 2007. Age and growth. Pages 187–228 in C. pretation of freshwater fisheries data. American Fisheries Society, Bethesda, S. Guy and M. L. Brown, editors. Analysis and interpretation of freshwater Maryland. fisheries data. American Fisheries Society, Bethesda, Maryland. Chipps, S. R., and D. H. Wahl. 2008. Bioenergetics modeling in the 21st century: Johnson, B. M., and P. J. Martinez. 2000. Trophic economics of lake trout reviewing new insights and revisiting old constraints. Transactions of the management in reservoirs of differing productivity. North American Journal American Fisheries Society 137:298–313. of Fisheries Management 20:127–143. Clarke, L. R., D. T. Vidergar, and D. H. Bennett. 2005. Stable isotopes and gut Johnson, T. B., D. M. Mason, S. T. Schram, and J. F. Kitchell. 1999. Ontogenetic content show diet overlap among native and introduced piscivores in a large and seasonal patterns in the energy content of piscivorous fishes in Lake oligotrophic lake. Ecology of Freshwater Fish 14:267–277. Superior. Journal of Great Lakes Research 25:275–281. LATENT PREDATION IMPACTS OF LAKE TROUT 1205
Keesee, B. G., and L. M. Keller. 2012. Lake Chelan kokanee spawning ground and fishery of an alpine lake. Pages 30–38 in T. P. Nesler and E. P. Berg- surveys 2011. Chelan County Public Utility District, Final Report, Wenatchee, ersen, editors. Mysids in fisheries: hard lessons from headlong introductions. Washington. American Fisheries Society, Symposium 9, Bethesda, Maryland. Kendra, W., and L. R. Singleton. 1987. Morphometry of Lake Chelan. Wash- Rieman, B. E., and M. A. Maiolie. 1995. Kokanee population density and ington State Department of Ecology, Water Quality Investigations Section, resulting fisheries. North American Journal of Fisheries Management 15:229– Ecology Report 87-1, Olympia. 237. Lasenby, D. C. 1971. The ecology of Mysis relicta in an Arctic and a temperate Ruzycki, J. R., D. A. Beauchamp, and D. L. Yule. 2003. Effects of introduced lake. Doctoral dissertation. University of Toronto, Toronto. lake trout on native cutthroat trout in Yellowstone Lake. Ecological Applica- Love, R. H. 1971. Dorsal-aspect target strength of an individual fish. Journal of tions 13:23–37. the Acoustical Society of America 49:816–823. Scheuerell, J. M., D. E. Schindler, M. D. Scheuerell, K. L. Fresh, T. H. Sibley, Luecke, C., T. C. Edwards Jr., M. W. Wengert Jr., S. Brayton, and R. Schnei- A. H. Litt, and J. H. Shepherd. 2005. Temporal dynamics in foraging behavior dervin. 1994. Simulated changes in lake trout yield, trophies, and forage of a pelagic predator. Canadian Journal of Fisheries and Aquatic Sciences consumption under various slot limits. North American Journal of Fisheries 62:2494–2501. Management 14:14–21. Schindler, D. E., J. F. Kitchell, and R. Ogutu-Ohwayo. 1998. Ecological con- Luecke, C., M. W. Wengert Jr., and R. W. Schneidervin. 1999. Comparing sequences of alternative gill net fisheries for Nile perch in Lake Victoria. results of a spatially explicit growth model with changes in the length-weight Conservation Biology 12:56–64. relationship of lake trout (Salvelinus namaycush) in Flaming Gorge Reservoir. Schindler, D. W. 1998. Replication versus realism: the need for ecosystem-scale Canadian Journal of Fisheries and Aquatic Sciences 56:162–169. experiments. Ecosystems 1:323–334. Luecke, C., and W. A. Wurtsbaugh. 1993. Effects of moonlight and daylight Schoen, E. R. 2007. Pelagic trophic interactions in contrasting basins of Lake on hydroacoustic estimates of pelagic fish abundance. Transactions of the Chelan. Master’s thesis. University of Washington, Seattle. American Fisheries Society 122:112–120. Schoen, E. R., and D. A. Beauchamp. 2010. Predation impacts of lake trout and Martinez, P. J., P. E. Bigelow, M. A. Deleray, W. A. Fredenberg, B. S. Hansen, Chinook salmon in Lake Chelan, Washington: implications for prey species N. J. Horner, S. K. Lehr, R. W. Schneidervin, S. A. Tolentino, and A. E. and fisheries management. U.S. Geological Survey, Washington Cooperative Viola. 2009. Western lake trout woes. Fisheries 34:424–442. Fish and Wildlife Research Unit, University of Washington, Seattle. Martinez, P. J., and W. J. Wiltzius. 1995. Some factors affecting a hatchery- Sharma, S., D. A. Jackson, and C. K. Minns. 2009. Quantifying the potential sustained kokanee population in a fluctuating Colorado reservoir. North effects of climate change and the invasion of smallmouth bass on native lake American Journal of Fisheries Management 15:220–228. trout populations across Canadian lakes. Ecography 32:517–525. Mazur, M. M. 2004. Linking visual foraging with temporal prey distributions Sharp, D., and D. R. Bernard. 1988. Precision of estimated ages of lake trout from to model trophic interactions in Lake Washington. Doctoral dissertation. five calcified structures. North American Journal of Fisheries Management University of Washington, Seattle. 8:367–372. Mazur, M. M., and D. A. Beauchamp. 2006. Linking piscivory to spatial- Shuter, B. J., M. L. Jones, R. M. Korver, and N. P. Lester. 1998. A general, temporal distributions of pelagic prey fishes with a visual foraging model. life history based model for regional management of fish stocks: the inland Journal of Fish Biology 69:151–175. lake trout (Salvelinus namaycush) fisheries of Ontario. Canadian Journal of McMahon, T. E., and D. H. Bennett. 1996. Walleye and northern pike: boost or Fisheries and Aquatic Sciences 55:2161–2177. bane to Northwest fisheries? Fisheries 21(8):6–13. Stafford, C. P., J. A. Stanford, F. R. Hauer, and E. B. Brothers. 2002. Changes Miranda, L. E., and P. W. Bettoli. 2007. Mortality. Pages 229–277 in C. S. Guy in lake trout growth associated with Mysis relicta establishment: a retrospec- and M. L. Brown, editors. Analysis and interpretation of freshwater fisheries tive analysis using otoliths. Transactions of the American Fisheries Society data. American Fisheries Society, Bethesda, Maryland. 131:994–1003. Neilson, J. D., and G. H. Geen. 1981. Enumeration of spawning salmon from Stenseth, N. C., W. Falck, O. N. Bjørnstad, and C. J. Krebs. 1997. Population spawner residence time and aerial counts. Transactions of the American regulation in snowshoe hare and Canadian lynx: asymmetric food web con- Fisheries Society 110:554–556. figurations between hare and lynx. Proceedings of the National Academy of Ney, J. J. 1990. Trophic economics in fisheries: assessment of demand-supply Sciences of the USA 94:5147–5152. relationships between predators and prey. Reviews in Aquatic Sciences 2:55– Stewart, D. J., and M. Ibarra. 1991. Predation and production by salmonine 81. fishes in Lake Michigan, 1978–88. Canadian Journal of Fisheries and Aquatic Ney, J. J. 1993. Bioenergetics modeling today: growing pains on the cutting Sciences 48:909–922. edge. Transactions of the American Fisheries Society 122:736–748. Stewart, D. J., J. F. Kitchell, and L. B. Crowder. 1981. Forage fishes and their Paine, R. T. 1980. Food webs: linkage, interaction strength and community salmonid predators in Lake Michigan. Transactions of the American Fisheries
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 infrastructure: the third Tansley lecture. Journal of Animal Ecology 49:667– Society 110:751–763. 685. Stewart, D. J., D. Weininger, D. V. Rottiers, and T. A. Edsall. 1983. An ener- Patankar, R., F. A. von Hippel, and M. A. Bell. 2006. Extinction of a weakly ar- getics model for lake trout (Salvelinus namaycush): application to the Lake moured threespine stickleback (Gasterosteus aculeatus) population in Prator Michigan population. Canadian Journal of Fisheries and Aquatic Sciences Lake, Alaska. Ecology of Freshwater Fish 15:482–487. 40:681–698. Peckarsky, B. L., P. A. Abrams, D. I. Bolnick, L. M. Dill, J. H. Grabowski, Tamura, K., J. Dudley, M. Nei, and S. Kumar. 2007. MEGA4: molecular evolu- B. Luttbeg, J. L. Orrock, S. D. Peacor, E. L. Preisser, O. J. Schmitz, and tionary genetics analysis (MEGA) software version 4.0. Molecular Biology G. C. Trussell. 2008. Revisiting the classics: considering nonconsumptive and Evolution 24:1596–1599. effects in textbook examples of predator–prey interactions. Ecology 89: Thiede, G. P. 1997. Impact of lake trout predation on prey populations in Lake 2416–2425. Tahoe: a bioenergetics assessment. Master’s thesis. Utah State University, Peven, C. M. 1989. Lake Chelan spawning ground surveys 1989. Chelan County Logan. Public Utility District, Wenatchee, Washington. Truscott, K. B., and C. M. Peven. 1988. Lake Chelan spawning ground surveys, Peven, C. M. 1990. Lake Chelan spawning ground surveys 1990. Chelan County 1988. Chelan County Public Utility District, Wenatchee, Washington. Public Utility District, Wenatchee, Washington. Vander Zanden, M. J., S. Chandra, B. C. Allen, J. E. Reuter, and C. R. Goldman. Rahel, F. J., and J. D. Olden. 2008. Assessing the effects of climate change on 2003. Historical food web structure and restoration of native aquatic commu- aquatic invasive species. Conservation Biology 22:521–533. nities in the Lake Tahoe (California–Nevada) basin. Ecosystems 6:274–288. Richards, R., C. Goldman, E. Byron, and C. Levitan. 1991. The mysids and lake Walters, C., and S. Martell. 2004. Fisheries ecology and management. Princeton trout of Lake Tahoe: a 25-year history of changes in the fertility, plankton, University Press, Princeton, New Jersey. 1206 SCHOEN ET AL.
Ware, D. M., and R. E. Thomson. 2005. Bottom-up ecosystem trophic dy- Wissinger, S. A., H. H. Whiteman, M. Denoel,¨ M. L. Mumford, and namics determine fish production in the northeast Pacific. Science 308: C. B. Aubee. 2010. Consumptive and nonconsumptive effects of can- 1280–1284. nibalism in fluctuating age-structured populations. Ecology 91:549– Werner, E. E., and S. D. Peacor. 2003. A review of trait-mediated indirect 559. interactions in ecological communities. Ecology 84:1083–1100. Worm, B., and R. A. Myers. 2003. Meta-analysis of cod–shrimp interactions White, E. M., J. C. Wilson, and A. R. Clarke. 2006. Biotic indirect effects: a reveals top-down control in oceanic food webs. Ecology 84:162–173. neglected concept in invasion biology. Diversity and Distributions 12:443– Wydoski, R. S., and R. R. Whitney. 2003. Inland fishes of Washington, 2nd 455. edition. University of Washington Press, Seattle. Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:00 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Evidence for Parr Growth as a Factor Affecting Parr-to- Smolt Survival William P. Connor a & Kenneth F. Tiffan b a U.S. Fish and Wildlife Service, Idaho Fishery Resource Office, 276 Dworshak Complex Drive, Orofino, Idaho, 83544, USA b U.S. Geological Survey, Western Fisheries Research Center, 5501A Cook-Underwood Road, Cook, Washington, 98605, USA
Version of record first published: 30 Jul 2012.
To cite this article: William P. Connor & Kenneth F. Tiffan (2012): Evidence for Parr Growth as a Factor Affecting Parr-to- Smolt Survival, Transactions of the American Fisheries Society, 141:5, 1207-1218 To link to this article: http://dx.doi.org/10.1080/00028487.2012.685121
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. Transactions of the American Fisheries Society 141:1207–1218, 2012 American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.685121
ARTICLE
Evidence for Parr Growth as a Factor Affecting Parr-to-Smolt Survival
William P. Connor* U.S. Fish and Wildlife Service, Idaho Fishery Resource Office, 276 Dworshak Complex Drive, Orofino, Idaho 83544, USA Kenneth F. Tiffan U.S. Geological Survey, Western Fisheries Research Center, 5501A Cook-Underwood Road, Cook, Washington 98605, USA
Abstract Data collected on juvenile anadromous salmonids implanted with passive integrated transponder (PIT) tags are used in mark–recapture analyses to understand the factors affecting survival of fish estimated between rearing in riverine habitat and dam passage. We estimated parr-to-smolt survival of PIT-tagged naturally produced subyearling fall Chinook salmon Oncorhynchus tshawytscha to examine the previously unexplored influences of environmental and biological conditions measured prior to reservoir entry. Mean ( ± SE) parr-to-smolt survival of the early migrating cohorts was 45.4 ± 6.3% (n = 13) compared with 37.4 ± 4.7% (n = 13) for later migrating cohorts. Annual mean parr-to-smolt survival differed widely across years ranging from a low of 9.6 ± 0.5% (n = 2) in 2001 to a high of 81.7 ± 4.6% (n = 2) in 1999. Parr growth prior to reservoir entry and reservoir velocity provided the most information 2 on variability in parr-to-smolt survival (N = 26, R = 0.75, corrected Akaike’s information criterion [AICc] =−5.01). We suggest that parr growth and reservoir velocity were directly proportional to parr-to-smolt survival because fast growth and downstream movement reduces the time when fish are vulnerable to predators. The effect of reservoir velocity comports with previous published studies and supports management efforts to increase reservoir velocity. Few if any published studies explicitly relate parr growth measured on individual fish to survival estimated for their cohorts in freshwater. This study provides empirical evidence that upholds the long-held belief that any anthropogenic activity that reduces growth of juvenile salmonids during freshwater rearing has the potential to reduce their survival.
The parr-to-smolt life stage is critically important for anadro- 1964; Skalski et al. 1998) to estimate parr-to-smolt survival Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 mous salmonids because many factors can affect survival during of depressed populations of anadromous salmonids since the this time of early rearing and initiation of seaward migration. 1990s. One of the earliest and longest parr-to-smolt survival Survival varies as anadromous salmonids progress through the studies (1991–2004) conducted using PIT-tag technology was juvenile life stages. Survival during the parr and early smolt on spring–summer Chinook salmon tagged as subyearling stages can be low and vary widely (4–12% to 25–46% apparent parr in streams of the Salmon River basin in the U.S. Pacific survival for coho salmon Oncorhynchus kisutch parr, Ebersole Northwest (Figure 1; Achord et al. 2007). The fish overwintered et al. 2006; Quinn and Peterson 1996; 73–99% estimated after tagging and then migrated downstream the following survival for Chinook salmon O. tshawytscha smolts, Muir spring through Lower Granite Reservoir and past Lower Granite et al. 2001; Connor et al. 2004). Researchers have used passive Dam (Figure 1). Parr-to-smolt survival estimated between the integrated transponder (PIT) tag technology (Prentice et al. time of tagging during rearing and passage at Lower Granite 1990a, 1990b) within a mark–recapture framework (Cormack Dam ranged from 7% to 48% within a year and averaged from
*Corresponding author: william [email protected] Received June 21, 2011; accepted April 5, 2012 1207 1208 CONNOR AND TIFFAN
Lower Monumental Little Goose Ice Harbor Lower Granite MT WA * * * Dworshak Asotin Columbia The McNary * Clearwater Dalles * Salmon * * John Bonneville Day Hells Canyon OR ID WY
N Snake 0 100 200 Km
FIGURE 1. The Snake River upper reach that flows unimpounded from Hells Canyon Dam 94 km downstream to the Salmon River confluence where natural fall Chinook salmon subyearlings were seined and PIT-tagged, the Snake River lower reach that flows freely from the Salmon River confluence 65 km downstream to Asotin, Washington, the transition zone between riverine and impounded habitat that flows for 15 km downstream from Asotin to Lower Granite Reservoir, and Lower Granite Reservoir that extends from the Clearwater River confluence 51 km downstream to Lower Granite Dam. Detection data were collected on PIT-tagged fish at the dams marked with an asterisk.
8% to 25% across years (Achord et al. 2007). Though Achord became later, differed widely across years, and increased as fork et al. (2007) measured increases in fork length and variation in length at tagging and reservoir flow during migration increased, environmental conditions in some cases, they did not evaluate and it decreased as date of tagging became later and reservoir how these factors affected parr-to-smolt survival. temperature during migration increased (Connor et al. 2003a; Few studies have estimated parr-to-smolt survival for Smith et al. 2003). Fork length and date of tagging functioned cohorts distinguished by migration timing within a year over as indices of environmental conditions during rearing in both a long enough span of years to statistically explore differ- studies, whereas reservoir flow and temperature functioned as ences in survival between early and late migrants (hereafter, indices of environmental conditions during migration. between-cohort differences), differences in survival across One difference between the Connor et al. (2003a) and Smith years (hereafter, annual differences), the factors that affect et al. (2003) studies was that in addition to the migration
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 the between-cohort and annual differences in survival, and indices, Smith et al. (2003) calculated indices of environmental the relative influence the factors have on survival. This paper conditions during the period hatchery subyearling parr were builds on two relatively short-term studies on Snake River fall rearing in Lower Granite Reservoir before becoming smolts. Chinook salmon. Connor et al. (2003a) PIT-tagged natural fall Smith et al. (2003) reported that correlations between the Chinook salmon subyearling parr in natal riverine habitat of rearing indices and survival were generally stronger than the Snake River (Figure 1) during 1998–2000, divided the fish correlations between the migration indices and survival. Water into four intra-annual cohorts, and monitored their detection at is released from upstream reservoirs during the summer to Lower Granite Dam after they had become subyearling smolts. increase velocity and decrease temperature when subyearling To supplement Connor et al. (2003a), Smith et al. (2003) made parr and smolts are in Lower Granite Reservoir. Findings from weekly releases of PIT-tagged hatchery fall Chinook salmon both studies on subyearling parr suggested that the releases subyearling parr into the free-flowing river during 1995–2000 of water from upstream reservoirs increased parr-to-smolt that, in contrast to PIT-tagged natural fish, dispersed rapidly survival. However, the Smith et al. (2003) analyses of rearing into Lower Granite Reservoir where they reared as they became indices raises a question that has implications for regional man- subyearling smolts. Parr-to-smolt survival of both the natural agement of this fish population. “How much of the variability and hatchery subyearling parr decreased as migration timing in parr-to-smolt survival in the Connor et al. (2003a) study PARR GROWTH AND PARR-TO-SMOLT SURVIVAL 1209
should have been attributed to rearing conditions experienced (Yanke 2006). Growth rate during rearing also influences physi- by the natural subyearling parr before reservoir entry?” ological development and the rate of smoltification (Wedemeyer Another question relative to the Connor et al. (2003a) study et al. 1980); thus, high growth rates during rearing enhance relates to the diverse migrational behavior of juvenile Snake migrational disposition and shorten exposure time to predators. River fall Chinook salmon. Unlike yearling Chinook salmon Additionally, growth rate governs how quickly prey exceed smolts that pass downstream rapidly over compressed periods the gape limitation of size-selective predators (Krueger et al. of time in spring and early summer (Achord et al. 1996, 2007; 2011) and when prey become capable of burst speeds needed Smith et al. 2002), subyearling parr that migrate later include to elude predators (Miller et al. 1988). Though it is biologically an unknown portion of fish that pass Lower Granite Dam during intuitive to link high levels of freshwater growth to high the winter after the water supply to the PIT-tag detection system levels of natural fish survival—and a multitude studies imply is shut off during late fall (Tiffan et al. 2012). Some fish from that increases in freshwater growth translate into increases in the later-migrating cohorts studied by Connor et al. (2003a) survival (e.g., steelhead O. mykiss, Close and Anderson 1992; also passed Lower Granite Dam the year after tagging when the Atlantic salmon Salmo salar, Orciari et al. 1994; coho salmon, PIT-tag detection system was resupplied with water in spring Quinn and Peterson 1996; Chinook salmon, Beckman et al. (e.g., Connor et al. 2002). Given winter passage and passage the 1999; Ebersole et al. 2006)—we are not aware of any studies year after tagging, survival estimated with PIT-tag data under that explicitly relate parr growth measured on individual fish the mark–recapture framework is typically lower than true to estimates of parr-to-smolt survival (i.e., opposed to apparent survival. Winter passage and passage in the year after tagging survival) for their cohorts in freshwater. A finding for parr of the later migrating cohorts raise the question “Did Connor growth as an influential factor for parr-to-smolt survival would et al. (2003a) report the true difference in parr-to-smolt survival have global relevance because it would support the long-held between some early and late migrating cohorts, as well as the belief that juvenile growth influences juvenile survival. true relation between the migration indices and parr-to-smolt In this paper, we combined the data collected in the Snake survival?” This question can be partly answered by limiting River upper reach by Connor et al. (2003a) during 1998–2000 analyses to cohorts of natural subyearling parr tagged along the with data collected in that reach during 2001–2010 to (1) Snake River upper reach (Figure 1) that typically pass Lower test for between-cohort and annual differences in Chinook Granite Dam during the year of tagging before the PIT-tag salmon parr-to-smolt survival, (2) identify potential factors for detection system is dewatered (e.g., Connor et al. 2002). the between-cohort and annual differences, and (3) describe The goals of this study were to evaluate the influence of fac- the relative influence of parr growth, transition zone velocity, tors measured before and after reservoir entry on parr-to-smolt fork length (hereafter length), reservoir velocity, and reservoir survival of natural Chinook salmon subyearlings and to mini- temperature on variation in parr-to-smolt survival. mize bias in analyses associated with winter passage and passage the year after tagging. Water velocity has been hypothesized to be one underlying factor for fish movement rates (Coutant 2001) METHODS and we selected it a priori for our analyses. Tiffan et al. (2009) Data collection.—We collected Chinook salmon subyearling found large seasonal changes in velocity and that velocity parr along the Snake River upper reach by setting a beach seine declined markedly as water flowed from the free-flowing Snake parallel to the shore from a boat and hauling it straight into River into the 15-km-long transition zone immediately upstream shore. The seine was constructed with 0.48 cm mesh and was of Lower Granite Reservoir. Consequently, the mean ± SD 30.5 m long and 1.8 m deep. It was fitted with a 3.9-m3 bag and downstream movement of radio-tagged subyearlings decreased a multistranded mudline. Each end of the seine was fitted with
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 as fish moved from riverine habitat through the transition zone a weighted brail attached to 15.2-m lead ropes. Seining began (e.g., 107 ± 33 km/d down to 36 ± 18 km/d). Moreover, near the onset of fry emergence and was conducted 1 d each 46.3% of the subyearlings moved downstream faster than the week. Three to five permanent stations were sampled almost mean cross-sectional water column velocity in the free-flowing every week by setting the seine one to three times depending on river compared with only 3.6% of the fish in the transition zone. beach length. In most years, supplemental stations were sam- Tiffan et al. (2009) concluded that fish delayed their migration pled the last 2 weeks of May and first week of June to increase in the transition zone while they re-obtained velocity cues. Such the sample size of PIT-tagged fish. Sampling ended when the delay before reservoir entry might reduce survival because natural subyearlings had dispersed offshore and downstream. predation can be high in transitional habitat (Shively et al. 1996). Thousands of unmarked (no fin clips, external marks, or tags) Parr growth prior to reservoir entry is another factor that we hatchery fall Chinook salmon subyearling smolts were released selected a priori for our analysis. Parr growth has the potential upstream from four of the beach seining stations after May to influence parr-to-smolt survival because changes in growth 26, 2005, (one of seven stations) and May 27, 2008, (3 of 10 reflect changes in temperature and ration level (Geist et al. stations). After the unmarked hatchery subyearling smolts were 2010), turbidity (Sigler et al. 1984), competition for food and released upstream from these stations in 2005 and 2008, the space (Chapman 1966; Kostow 2009), and nutritional status origin (i.e., natural or hatchery) of unmarked subyearlings in 1210 CONNOR AND TIFFAN
the beach-seine catch was determined based on pupil diameter because it (1) probably occurs among cohorts in the wild and body shape, and presumed hatchery fish were processed that are truly differentiated by emergence timing and growth but not tagged as described by Tiffan and Connor (2011). (e.g., Connor et al. 2002; Connor and Burge 2003), (2) allows Natural subyearling parr were placed in a 94.6-L aerated indices of environmental and biological conditions to better livewell treated with 100 g of NaCl and 12.5 mL of Polyaqua. reflect the true differences experienced by cohorts in the wild, When seining was completed at a site, the fish were anesthetized and (3) limits the potential for tagging effects caused by stark in a 3-mL stock solution (100 g/L) of tricaine methanesulfonate differences date of tagging, length at tagging, and temperature (MS-222) per 19 L of water buffered with a sodium bicarbonate at release that are present when the data are simply divided solution and were measured to the nearest 1.0 mm in length. by arbitrarily selected discrete time intervals. To demonstrate Fish 60 mm and longer were scanned for a previously implanted (3) above, we calculated cumulative (%) tagging date distri- PIT tag (i.e., a recapture). Fish 60 mm and longer that had not butions and identified the maximum difference between the been previously implanted with a PIT tag were tagged. The distributions of cohort 1 and cohort 2 within each year. We also fish were released at the collection site after a 15-min recovery reported the maximum between-cohort and annual differences period. Water temperature at release was measured to the near- in temperatures recorded during release of the tagged fish. est 0.1◦C with a hand-held digital thermometer. We used the We established the migration period for each cohort in PIT Tag Information System (PTAGIS 2011) to upload initial three steps using PIT-tag detection data collected at Lower capture and tagging data and download recapture data. The data Granite Dam (after Connor et al. 2003a). First, we calculated included initial date of tagging, length at initial tagging, recap- a cumulative detection distribution for each cohort that only ture date, length at recapture, and temperature at release. We included detections made in the year of release. Second, we also downloaded the detection dates of PIT-tagged subyearlings calculated the date cutoff for early outliers by multiplying at the seven dams equipped with PIT-tag detection systems the interquartile range by 1.5 and subtracting the resulting (Figure 1) during the period these systems were supplied product from the 25th percentile (i.e., the lower fence date; with water (dates available at: www.ptoccentral.org/Ptoc OM). Ott 1993). Third, we calculated the date cutoff for late outliers Because PIT-tag frequencies were changed in 1999, the end by multiplying the interquartile range by 1.5 and adding the date for detection data collection in 1999 is reported online as resulting product to the 75th percentile (i.e., the upper fence September 1, but detection data were actually collected using date; Ott 1993). The first day of the migration period was the a nontypical but efficient method at Lower Granite and Little detection date observed on or after the lower fence date and Goose dams until October 31. We downloaded these detection the last day was the detection date observed on or before the data by running a recapture query in PTAGIS (recapture upper fence date. We graphed the entire PIT-tag detection date coordinators: ALS, CFM; tagging coordinator: WPC). distribution of each cohort including outliers to examine it for The flow (m3/s) data for predicting velocity in the transition detection of PIT-tagged fish immediately before the date that zone were collected by the U.S. Geological Survey at Anatone, the PIT-tag detection system was dewatered in year t (i.e., the Washington, in the lower Snake River (35 km upstream from the year of tagging) and immediately after it was resupplied with transition zone). The flow data for predicting velocity in Lower water in year t + 1 (i.e., the year after tagging). Such detections Granite Reservoir were collected by the U.S. Army Corps of for a given cohort would suggest that passage of PIT-tagged fish Engineers at Lower Granite Dam. The U.S. Army Corps of En- probably continued when the PIT-tag detection system at Lower gineers also measured daily mean temperature in the tailrace of Granite Dam was dewatered and that the parr-to-smolt survival Lower Granite Dam. We accessed the data online (DART 2011). estimate for that cohort was an underestimate of true survival. Seine catch, cohorts, and migration periods.—We divided Growth, length, velocity, temperature, and survival.—We
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 the beach-seine catch into fry (<46 mm long) and parr (>45 mm used data collected on each recaptured PIT-tagged subyearling long; after Connor et al. 2002). We divided the 1998–2010 parr to calculate its growth rate (to nearest 0.1 mm/d) as the samples of PIT-tagged (i.e., >59 mm long) natural subyearling length at recapture minus the length at initial tagging divided parr into two cohorts (e.g., 1998-1, 1998-2), each of which by the number of days that elapsed between initial tagging and contained approximately equal numbers of PIT-tagged fish as recapture. We averaged the growth rates and the lengths at tag- described by Connor et al. (2003a). In contrast to Connor et al. ging among fish in each cohort within each year to calculate (2003a), our analyses were limited to two instead of four cohorts the factors: parr growth and parr length. Velocity in the tran- because we did not analyze data on later-migrating cohorts that sition zone, reservoir velocity, and reservoir temperature were rear downstream of the Salmon River confluence that are more also included as factors as described in the following example likely to pass Lower Granite Dam during the winter or the year for cohort 1998-1. Of time spent in Lower Granite Reservoir by after tagging. The method for assigning fish to cohorts results the subyearlings studied by Tiffan et al. (2009), 13% was spent in tagging and detection date distributions for cohort 1 that in the transition zone and 87% was spent in the reservoir. We proceed on an earlier time schedule than for cohort 2, but there is multiplied the length of the May 25–July 20 migration period of considerable temporal overlap in the tagging and detection date cohort 1998-1 (57 d) by 0.13 and 0.87 to estimate the number of distributions between the two cohorts. This overlap is important days and date range, respectively, that fish from this cohort were PARR GROWTH AND PARR-TO-SMOLT SURVIVAL 1211
present in each respective reach. Given the 57-d length of the year. Second, we added the remaining factors to mixed model 3 May 25–July 20 migration period, the number of days spent and to form mixed model 4 and removed the fixed term for cohort the date ranges of presence for cohort 1998-1 were 7 d (i.e., 57 from mixed model 4 to form mixed model 5. We compared the multiplied by 0.13) and May 25–May 31 for the transition zone, AICc values of mixed models 4 and 5 to determine if the co- and 49 d (57 multiplied by 0.87) and June 1–July 20 for the hort variable could be replaced by the factors. The AICc value reservoir. We then input flow data into reach-specific velocity of mixed model 5 would be equivalent to or less than the AICc regression equations (velocity = βo + β1flow) fitted by Tiffan value of mixed model 4 if the subset of factors largely accounted et al. (2009) from data collected along transects spaced 3-km for the variability associated with the fixed term for cohort. apart in the transition zone (six transects; range of r2 values, The relative influence of the factors.—To describe the 0.98–0.99) and reservoir (16 transects; range of r2 values, 0.96– relative influence of the factors on variation in parr-to-smolt 0.99). In our example of cohort 1998-1, the mean flow for May survival, we first fitted an ordinary least-squares regression 25–31 measured at Anatone, Washington, was input into each of model from each factor to establish if the slope coefficient the six transition-zone velocity regression models. Mean transi- was positive or negative. After assessing the slope coefficients tion zone velocity was the average of the six predictions. We used from ordinary least-squares regression, we fitted the possible the same approach to calculate mean reservoir velocity with the 31 ordinary least-squares multiple regression models to predict 16 transects in the reservoir except flow was measured at Lower parr-to-smolt survival from every possible set of factors studied Granite Dam. Mean reservoir temperature was the average tem- to determine if the sign of the slope coefficients for each perature measured in the tailrace of Lower Granite Dam during factor were consistent with those estimated with ordinary May 25–July 20 (after Connor et al. 2003a). We used the single- least-squares regression. A sign change was taken as evidence release method (Cormack 1964; Skalski et al. 1998) to estimate for problematic multicollinearity (Dielman 1996). We removed parr-to-smolt survival ( ± SE) for each cohort between tagging regression models that contained problematically collinear during rearing and passage at Lower Granite Dam during year factors to form a set of candidate regression models. t. We square-root transformed all the factors and the survival We applied the information-theoretic approach for model estimates before statistical analyses to stabilize the variance. selection (Burnham and Anderson 2002) to the candidate Testing for between-cohort and annual differences in regression models. We calculated the AIC and AICc as survival.—We used a mixed model (α = 0.05; Littell et al. previously described for each candidate regression model. 1996) to test the null hypothesis H0: the difference in the The candidate regression model with the lowest AICc was least-squares mean parr-to-smolt survival estimates of the two selected as the best regression model. To rank the remaining cohorts was equal to zero, while accounting for annual variation regression models, we calculated simple differences between in parr-to-smolt survival. We used cohort as a fixed term and the best model’s AICc and the AICc of the remaining models as = − year as a random term. We then fitted the starting model and i (AICci AICcmin ). The amount of information lost by a a model without the random term for year. We calculated the regression model increased as its difference, , increased. As a Akaike’s information criterion (AIC) for each model as: AIC = general rule, regression models with -values greater than 10 −2log(£) + 2K, where £ is the model likelihood function and are not informative or plausible. Such models were removed to K is the number of estimable parameters including the intercept form the final set of regression models. (Burnham and Anderson 2002). We then recalculated each We evaluated the relative response in parr-to-smolt survival AIC using a second-order bias correction denoted as AICc as: to each factor in the final set of regression models based on AICc = AIC + [2K(K + 1) / (n − K − 1)], where n is the the total Akaike weight calculated as follows. We divided the sample size (Burnham and Anderson 2002). Of the two models, model likelihood [e(−1/2 )] of each model by the sum of all
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 we selected the model with the lowest AICc for subsequent model likelihoods to produce Akaike weights. If a factor was analyses and refer to this model as mixed model 1. in a regression model, then we assigned the factor the Akaike Factors for between-cohort and annual differences in sur- weight of the model. For example, if a factor was in eight of vival.—To identify the factors for between-cohort and annual nine final regression models it would be assigned the eight differences in parr-to-smolt survival, we used mixed modeling Akaike weights of those models. The sum of these eight Akaike and added the factors as covariates. First, we added a subset of weights was the total Akaike weight for the factor. Factors with the factors as fixed terms (selected and added into the model higher total Akaike weights were judged to have more influence together for reasons described in results) to mixed model 1 to on the variability in parr-to-smolt survival than factors with form mixed model 2 and removed the random term for year lower total Akaike weights. from mixed model 2 to form mixed model 3. We compared the Finally, we used the best regression model (i.e., lowest AICc) AICc values of mixed models 2 and 3 to determine if the random to make partial regression plots for the factors in this model. The term for year could be replaced by the factors. The AICc value plots were made as described for factor X in the simple model of mixed model 3 would be equivalent to or less than the AICc Y = β0 + β1X + β2Z. Predicting Y from Z alone (i.e., Ypred value of mixed model 2 if the subset of selected factors largely = β0 + β1Z) provides residuals (Y − Ypred). Each residual is accounted for the variability associated with the random term for referred to as an “adjusted” Y (Yadj) because the variability in Y 1212 CONNOR AND TIFFAN
TABLE 1. The range of capture dates for natural fall Chinook salmon juveniles, the number of fry (<46 mm fork length) and subyearling parr (>45 mm fork length) captured, mean ± SD fork length, total number of subyearling parr > 59 mm fork length that were PIT-tagged, the percent of the total parr catch that was large enough to PIT-tag, and the number of PIT-tagged subyearling parr that were recaptured and provided data for calculating parr growth, 1998–2010. Dates are given as month/day/year.
Numbers captured Number of Range of Mean parr Number of Percent of tagged parr capture dates Fry Parr Total fork length parr tagged parr tagged recaptured Apr 16–Jul 6, 1998 101 1,078 1,179 70 ± 14 628 58.3 110 Apr 8–Jul 2, 1999 97 1,493 1,590 69 ± 13 918 61.5 170 Apr 6–Jun 15, 2000 683 1,064 1,747 66 ± 19 380 35.7 97 Apr 6–Jun 14, 2001 552 794 1,346 56 ± 9 198 24.9 12 Apr 4–Jul 3, 2002 2,289 3,013 5,302 57 ± 11 720 23.9 168 Mar 27–Jun 26, 2003 962 4,523 5,485 60 ± 10 1,727 38.2 357 Mar 25–Jun 24, 2004 6,123 6,310 12,433 55 ± 8 1,171 18.6 150 Mar 31–Jun 23, 2005 5,462 8,119 13,581 57 ± 8 2,235 27.5 295 Mar 31–Jun 29, 2006 75 1,344 1,419 66 ± 9 945 70.3 80 Mar 29–Jul 5, 2007 4,311 7,226 11,537 54 ± 7 1,000 13.8 129 Apr 3–Jul 17, 2008 1,628 6,610 8,238 64 ± 12 2,512 38.0 595 Mar 26–Jun 18, 2009 811 3,876 4,687 59 ± 8 1,599 41.3 18 Mar 25–Jul 22, 2010 1,572 2,502 4,074 58 ± 10 682 27.3 90
attributable to Z has been removed. Next, predicting X from 1.9◦C; range, 9.1–14.8◦C) and 2001-2 (13.4 ± 2.1◦C; range, ◦ Z (i.e., Xpred = β0 + β1Z) provides a set of residuals (X − 10.3–15.0 C). The largest difference in temperature among ◦ Xpred), each of which is referred to as an adjusted X (Xadj). The years was observed between 1998 (15.1 ± 1.3 C; range, ◦ ◦ ◦ partial residual plot is constructed by plotting Yadj against Xadj 9.8–18.2 C) and 2010 (12.0 ± 1.5 C; range, 8.3–19.8 C). and fitting a predicted regression line to the data (Ypred adj = β0 In 12 of 13 annual comparisons, the migration period of sub- + β1Xadj). yearlings from cohort 1 began and ended earlier than did the mi- gration period of subyearlings from cohort 2 (see whiskers plot- RESULTS ted in Figure 2). Further examination of Figure 2 supported the parr-to-smolt survival estimates as relatively unbiased estimates Seine Catch, PIT-Tagging, Cohorts, and Migration Periods of true survival. With the exception of cohort 2001-2, only fish Fish were captured over a range of 10–17 weeks during that were considered outliers (e.g., cohort 1999-2) were detected the 13 years studied (Table 1). Natural subyearling parr were at Lower Granite Dam near the date when the PIT-tag detection more abundant in the catch than were natural fry. Mean ± SD system was dewatered (Figure 2). In the case of cohort 2001-2, parr length ranged from 54 ± 7to70± 14 mm. The number the 75th percentile of passage was relatively late and there were of natural subyearling parr that were longer than 59 mm and no outliers because only three fish were detected owing to record were PIT-tagged ranged from 198 to 2,512. The percentage low parr-to-smolt survival (Table 2). Detection the year after
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 of the natural subyearling parr that were captured and were tagging at Lower Granite Dam was only observed for cohorts large enough to PIT-tag ranged from 13.8% to 70.3%. In 2005 1998-2 (2 of 85 detections) and 2002-2 (1 of 51 detections). and 2008, totals of 26 out of 2,233 fish (1.2%) and 104 out of 2,512 (4.1%) fish, respectively, were tagged after releases of Testing for Differences in Parr-to-Smolt Survival unmarked hatchery fish began. The number of fish that were Parr-to-smolt survival of cohort 1 was higher than parr-to- recaptured and provided the data for calculating parr growth smolt survival of cohort 2 in 10 of the 13 years studied (Table 2). across all years ranged from 12 to 595. Arithmetic mean ( ± SE) parr-to-smolt survival across years The number of PIT-tagged natural subyearling parr assigned (n = 13) was 45.4 ± 6.3% for cohort 1 and 37.4 ± 4.7% to cohorts ranged from 99 to 1,256 (Table 2). When the for cohort 2. The least-squares mean ( ± SE) difference in the maximum difference was observed in the annual cumulative square-root-transformed parr-to-smolt survival estimates of co- tagging date distributions, tagging of fish from cohort 1 was horts 1 and 2 was 0.6 ± 0.2%. This difference was significantly closer to completion than tagging of fish from cohort 2, different from zero (df = 12, P = 0.03) based on results of but there was considerable overlap in the distributions. The mixed model 1 (Table 3). Arithmetic mean ( ± SE) parr-to-smolt largest difference in temperature at release between cohorts survival ranged from a low of 9.6 ± 0.5% (n = 2 cohorts) in was observed between cohorts 2001-1 (mean ± SD, 11.3 ± 2001 to a high of 81.7 ± 4.6% (n = 2 cohorts) in 1999. PARR GROWTH AND PARR-TO-SMOLT SURVIVAL 1213
TABLE 2. Mean ± SE parr-to-smolt survival (%) estimated between PIT-tagging during rearing and passage at Lower Granite Dam for 26 cohorts of PIT-tagged natural fall Chinook salmon subyearlings, 1998–2010 (n = total number in cohort). Mean values of the factors: parr growth (mm/d ± SD, n = number of tagged parr recaptured in parentheses), transition-zone velocity (km/d), parr length (mm, range in parentheses), reservoir velocity (km/d), and reservoir temperature (◦C) are also given.
Parr Transition Parr Reservoir Reservoir Year Cohort n Survival growth zone velocity length velocity temperature 1998 1 314 65.4 ± 4.2 1.1 ± 0.2 (57) 223.3 76 (60–110) 33.2 16.1 2 314 50.8 ± 3.7 1.0 ± 0.4 (53) 113.4 72 (60–106) 24.0 19.0 1999 1 459 86.2 ± 3.7 1.2 ± 0.3 (76) 158.4 78 (60–132) 26.5 16.4 2 459 77.1 ± 4.5 1.4 ± 0.3 (94) 158.9 68 (60–115) 24.1 16.7 2000 1 190 54.2 ± 6.7 1.3 ± 0.2 (33) 81.4 79 (60–115) 20.7 16.1 2 190 36.4 ± 4.7 1.2 ± 0.2 (64) 65.8 77 (60–101) 16.0 17.9 2001 1 99 10.1 ± 3.0 1.2 ± 0.2 (6) 34.2 70 (60–98) 11.4 18.8 2999.1± 6.7 1.2 ± 0.1 (6) 29.9 66 (60–77) 9.1 18.6 2002 1 360 28.5 ± 3.7 1.0 ± 0.2 (62) 76.6 68 (60–102) 22.5 16.3 2 360 28.7 ± 2.9 1.2 ± 0.2 (106) 78.8 67 (60–97) 19.0 17.2 2003 1 863 34.7 ± 2.2 0.8 ± 0.1 (200) 117.3 68 (60–112) 28.0 15.5 2 864 39.0 ± 2.1 1.2 ± 0.3 (157) 123.6 67 (60–92) 19.6 17.1 2004 1 585 15.7 ± 1.6 0.8 ± 0.2 (72) 99.8 66 (60–102) 21.6 15.9 2 586 27.5 ± 1.9 1.2 ± 0.3 (78) 83.8 68 (60–98) 18.5 16.8 2005 1 1,117 47.6 ± 4.0 1.0 ± 0.2 (176) 106.2 66 (60–88) 22.0 14.7 2 1,118 44.2 ± 8.0 1.1 ± 0.3 (119) 100.4 64 (60–97) 19.3 16.0 2006 1 472 73.6 ± 9.7 1.0 ± 0.2 (27) 204.1 70 (60–100) 33.3 14.9 2 473 46.2 ± 9.8 1.1 ± 0.4 (53) 186.0 67 (60–97) 27.7 16.2 2007 1 500 25.6 ± 10.8 0.9 ± 0.2 (42) 57.7 65 (60–103) 17.5 15.8 2 500 18.6 ± 11.6 1.2 ± 0.2 (87) 58.2 66 (60–115) 15.3 17.4 2008 1 1,256 61.4 ± 6.8 1.0 ± 0.2 (321) 95.6 71 (60–109) 36.4 13.6 2 1,256 47.9 ± 5.7 1.1 ± 0.3 (274) 117.0 69 (60–108) 28.7 15.5 2009 1 799 42.8 ± 9.4 0.7 ± 0.3 (16) 147.2 69 (60–101) 38.2 14.6 2 800 28.3 ± 9.8 0.5 ± 0.1 (2) 161.9 63 (60–86) 32.8 15.7 2010 1 341 43.8 ± 15.7 0.8 ± 0.2 (52) 92.2 68 (60–107) 37.2 13.7 2 341 32.0 ± 8.0 1.0 ± 0.3 (38) 218.5 68 (60–97) 29.1 15.5
Factors for Between-Cohort and Annual 1 than for cohort 2 (Table 2). Mean reservoir temperature was Differences in Survival cooler more frequently (12 of 13 years) during the migration of Mean parr growth (10 of 13 years) and mean transition-zone cohort 1 than during the migration of cohort 2. Mixed modeling velocity (7 of 13 years) were lower more frequently for cohort treating mean parr length, reservoir velocity, and reservoir tem-
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 1 than for cohort 2 despite the higher parr-to-smolt survival perature as covariates supported these three factors as sources typically observed for cohort 1 (Table 2). These results were of between-cohort differences in parr-to-smolt survival. The opposite of what was expected under the premise that increases AICc value for mixed model 4 that included cohort as a fixed in both parr growth and transition zone velocity would increase term and mean parr length, reservoir velocity, and temperature parr-to-smolt survival within a year. Therefore, we examined as covariates had only a slightly lower AICc value than mixed these variables together as covariates in mixed modeling to model 5, which did not include the fixed term for cohort determine if they explained annual rather than between-cohort (Table 3). This suggested that mean parr length, reservoir differences. The AICc value for mixed model 2 that included velocity, and reservoir temperature accounted for a large portion year as a random term and mean parr growth and transition of the between-cohort difference in parr-to-smolt survival. zone velocity as covariates had a higher AICc value than mixed model 3 that did not include the random term for year (Table 3). The Relative Influence of the Factors This supported mean parr growth and transition-zone velocity on Parr-to-Smolt Survival factors for annual differences in parr-to-smolt survival. Of the 31 potential regression models, four included factors Mean parr length (11 of 13 years) and mean reservoir that were problematically collinear. Of the 27 remaining velocity (13 of 13 years) were higher more frequently for cohort candidate regression models, 18 had -values greater than 10 1214 CONNOR AND TIFFAN
2010 - 2 (24) model predicted parr-to-smolt survival from parr growth during 2010 - 1 (12) 2009 - 2 (24) rearing and mean reservoir velocity (Table 4; Figure 4). 2009 - 1 (53) * 2008 - 2 (113) ***** ** 2008 - 1 (123) * 2007 - 2 (13) 2007 - 1 (23) DISCUSSION 2006 - 2 (41) 2006 - 1 (55) 2005 - 2 (94) Assumptions and Limitations 2005 - 1 (219) * * We assumed that inadvertent tagging of unmarked hatchery 2004 - 2 (120) ***** 2004 - 1 (66) No detection data smolts did not affect our analysis. Only small percentages of the Cohort ** * 2003 - 2 (193) ***** ** 2003 - 1 (112) * Chinook salmon we tagged had the potential to be unmarked 2002 - 2 (50) * 2002 - 1 (45) hatchery fish (1.2% in 2005, 4.1% in 2008). Further, blind tests 2001 - 2 (3) * * 2001 - 1 (7) showed that our field crew classified origin accurately (97.6% 2000 - 2 (38) ** ** in 2005, 95.9% in 2008; Tiffan and Connor 2011) in portions of 2000 - 1 (34) * 1999 - 2 (163) ***** the Snake River where large percentages of the fish tagged had 1999 - 1 (163) ** 1998 - 2 (83) * * *** the potential to be unmarked hatchery fish. We made all of the 1998 - 1 (90) * assumptions inherent to the single-release survival model (see Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Smith et al. 2003 for details). The accuracy and precision of Month survival estimates made for subyearlings with the single-release survival model is an important area of future research. We also FIGURE 2. Detection date distributions at Lower Granite Dam in the year of release for the 1998–2010 cohorts of PIT-tagged natural fall Chinook salmon assumed that winter passage and passage the year after tagging subyearlings (cohort 1, white boxes; cohort 2, grey boxes). The asterisks are were not factors for between-cohort trends in parr-to-smolt the earliest and latest outliers, the whiskers extend to early and late nonoutliers survival. We conducted a conservative sensitivity analysis that define the migration period, the left side of the box is the 25th percentile, (available from W. P. Connor) to evaluate this assumption. The the vertical line in the box is the median, the right side of the box is the 75th sensitivity analysis indicated that parr-to-smolt survival might percentile, and the number in parentheses along the y-axis is the number of detections made the year of release. The light grey polygon spans the date range have been higher than we reported for cohorts 1998-2 (up 4.8 in which the PIT-tag detection system at Lower Granite Dam was dewatered. percentage points), 1999-2 (up 6.3 percentage points), and Detection dates of fish at Lower Granite Dam the year after release are shown 2002-2 (up 2.7 percentage points), but the increases were not as grey circles. large enough to influence the outcomes of statistical analyses or change our conclusions. We also assumed that tag loss and indicating that these regression models included sets of factors posttagging effects were low and equal between cohorts based that were not informative. Thus, there were nine final regression on the work of Prentice et al. (1990a, tag loss and mortality models, and they explained 68–79% of the variability in parr-to- <1%) that was highly applicable to our study given the large smolt survival (Table 4). Parr-to-smolt survival was predicted overlap in tagging date, temperature at release, and fish size to increase as all of the factors increased except for reservoir observed between cohorts and among years. temperature, which was inversely related to parr-to-smolt We limited our study to the Snake River upper reach that sup- survival (see slope coefficients, Table 4). Mean parr growth was ports one of the four main spawning aggregates of Snake River the most influential factor, followed closely by mean reservoir basin fall Chinook salmon listed for protection under the U.S. velocity, and then mean parr length, mean transition zone Endangered Species Act (NMFS 1992). The natural subyearling velocity, and mean reservoir temperature (Figure 3). The best parr from the Snake River upper reach generally emerge earlier, Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 TABLE 3. Results from mixed model 1 fitted to test for between-cohort (fixed effect) and annual differences (random effect) in parr-to-smolt survival of natural fall Chinook salmon subyearlings estimated between PIT-tagging during rearing and passage at Lower Granite Dam during 1998–2010 (N = 26 cohorts). Mixed models 2 and 3 were fitted to determine if mean parr growth and mean transition-zone velocity explained annual variation. Mixed models 4 and 5 were fitted to determine if mean parr length, reservoir velocity, and reservoir temperature explained between-cohort variation. Akaike’s information criterion corrected for second-order bias (AICc) was used for model selection.
Mixed model Fixed term Random term Covariates AICc 1 Cohort Year 83.5 2 Cohort Year Parr growth, transition zone velocity 76.4 3 Cohort Parr growth, transition zone velocity 75.4 4 Cohort Parr growth, transition zone velocity, parr length, 58.3 reservoir velocity, reservoir temperature 5 Parr growth, transition zone velocity, parr length, 58.7 reservoir velocity, reservoir temperature PARR GROWTH AND PARR-TO-SMOLT SURVIVAL 1215
TABLE 4. The final ordinary least-squares multiple regression models selected in analyses to describe the relative influence of mean parr growth (Gro), transition zone velocity (Tvel), parr fork length (FL), reservoir velocity (Rvel), and reservoir temperature (R◦C) on parr-to-smolt survival of natural fall Chinook salmon subyearlings (N = 26 cohorts) estimated between PIT-tagging during rearing and passage at Lower Granite Dam during 1998–2010. Models are given in the order of rank (best model first). Statistical notation: K, number of parameters including the intercept; AIC, Akaike’s information criterion; AICc, AIC corrected (−1/2 ) for second-order bias; , simple differences between the best model’s AICc and the AICc of the remaining candidate models; e , the model likelihoods; wi, Akaike weights.
2 (−1/2 ) Regression model KR AIC AICc e wi −14.23 + 10.62Gro + 1.98Rvel 3 0.75 −6.10 −5.01 0.00 1.00 0.363 −20.35 + 9.08Gro + 1.01FL + 1.84Rvel 4 0.76 −5.89 −3.99 1.02 0.60 0.218 −13.21 + 9.95Gro + 0.11Tvel + 1.68Rvel 4 0.76 −5.33 −3.43 1.58 0.45 0.165 −20.02 + 8.05Gro + 1.16FL + 0.13Tvel + 1.45Rvel 5 0.78 −5.77 −2.77 2.24 0.33 0.119 −19.89 + 9.08Gro + 1.03FL + 1.82Rvel − 0.13R◦C 5 0.76 −3.90 −0.90 4.11 0.13 0.047 −12.37 + 9.96Gro + 0.12Tvel + 1.64Rvel − 0.18R◦C 5 0.76 −3.35 −0.35 4.66 0.10 0.035 −15.38 + 7.71Gro + 1.39FL + 0.17Tvel + 1.13Rvel − 1.25R◦C 6 0.79 −4.39 0.03 5.04 0.08 0.029 −6.07 + 5.13Gro + 2.22FL + 0.34Tvel − 3.71R◦C 5 0.75 −2.36 0.64 5.65 0.06 0.022 −12.36 + 3.00FL + 0.32Tvel − 2.42R◦C 4 0.68 2.09 3.99 9.00 0.01 0.004 Sum = 2.76
grow faster, and migrate seaward earlier than the natural sub- salmonids, but many of these factors are difficult to measure yearling parr from the other three spawning aggregates (Connor and incorporate into a single analysis. Therefore, we selected et al. 2002, 2003b; Connor and Burge 2003). Consequently, factors a priori that could be measured and were biologically some of the findings in this paper might not be applicable across supportable based on the literature. Parr growth might have the Snake River basin. We also limited our study to natural subyearling parr longer than 59 mm; thus, we probably overes- timated the true parr-to-smolt survival of the entire size range of 3.0 parr. Further, we captured fish by beach-seining knowing that 2.0 capture efficiency could not be measured and probably varied by beach-seining station, across flows, and among different sizes 1.0 of fish. Until technology evolves that allows representative sam- 0.0 pling and tagging of newly emergent fry in large river settings, beach-seining combined with PIT tag technology will continue -1.0 to provide useful information for research and management. We also acknowledge that a multitude of environmental and -2.0
biological factors influence survival of juvenile anadromous Adjusted parr-to-smolt survival -3.0 -0.3 -0.2 -0.1 0.0 0.1 0.2 Adjusted parr growth 1.2 4.0 1.0
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 3.0
0.8 2.0 1.0 0.6 0.0
0.4 -1.0 Total Akaike weight Total -2.0 0.2 -3.0
0.0 Adjusted parr-to-smolt survival -4.0 Parr Length Tvel Rvel Degrees -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 growth Predictor variable Adjusted reservoir velocity
FIGURE 3. Total Akaike weights for evaluating the relative response in parr- FIGURE 4. Partial regression plots showing the relations between adjusted to-smolt survival of natural fall Chinook salmon estimated between PIT-tagging parr-to-smolt survival of natural subyearling fall Chinook salmon estimated during rearing and passage at Lower Granite Dam during 1998–2010 to mean between PIT-tagging during rearing and passage at Lower Granite Dam during parr growth, parr length, transition zone velocity (Tvel), reservoir velocity 1998–2010 (N = 26 cohorts) and adjusted mean parr growth (top panel) and (Rvel), and reservoir temperature (degrees). adjusted mean reservoir velocity (bottom panel). 1216 CONNOR AND TIFFAN
been identified as the most influential factor for parr-to-smolt Healey 1982; sockeye salmon O. nerka, West and Larkin 1987; survival because it was explicitly measured on individual fish, fall Chinook salmon, Smith et al. 2003; spring Chinook salmon, whereas the other factors studied including transition-zone Zabel and Achord 2004; pink salmon O. gorbuscha,Mossetal. velocity were only indices of the conditions actually experi- 2005). Larger subyearling parr could have been less suscepti- enced by the fish. Though we measured reservoir temperature ble to predation initially after tagging than smaller subyearling and calculated the migration temperature index as described parr for at least three reasons. Larger parr were probably better by Connor et al. (2003a), reservoir temperature had relatively swimmers (Webb 1976; Taylor and McPhail 1985) and closer to little influence on survival in our study. We believe the lack of a exceeding predator gape limitations (Miller et al. 1988; Krueger strong temperature effect was the result of limiting our analysis et al. 2011). Larger parr were more likely than smaller parr to the earliest portion of the run that is not subjected to as wide a to move offshore into faster, deeper water to feed, migrate, temperature range in the reservoir (1.4 ± 0.2◦C [mean ± SE] and become pelagic (Chapman and Bjornn 1969; Lister and from Table 2) as is experienced by the full spectrum of the run Genoe 1970; Everest and Chapman 1972; Dauble et al. 1989; (e.g., 2.1 ± 0.1◦C [mean ± SE] from Connor et al. 2003a). Connor et al. 2003c). Therefore, parr tagged at longer lengths We do not believe our temperature results are applicable to late would spend less time after tagging in the presence of shoreline- migrants, especially those that must grow and survive under oriented predators. We believe that the higher reservoir veloci- the warmest summer temperatures in order to become yearling ties usually experienced by cohort 1 also reduced exposure time ocean entrants that contribute to adult returns (e.g., Connor to predators in the reservoir by affecting higher rates of down- et al. 2005). Finally, we emphasize that characterizing variables stream movement. The cooler reservoir temperatures usually ex- as having strictly between-cohort or annual survival effects is perienced by cohort 1 probably resulted in rates of predation that oversimplified and the true effects are more integrated. were lower than for cohort 2 because consumption by predators is temperature dependent (Rieman et al. 1991; Vigg et al. 1991). Parr-to-Smolt Survival Parr-to-smolt survival of cohorts 1 and 2 differed widely Parr-to-smolt survival of Chinook salmon subyearlings was across the 13 years studied (maximum difference, 72.1 per- typically higher for cohort 1 (usually the earlier migrants) than centage points). Our analyses attributed the annual differences for cohort 2 (usually the later migrants), which is consistent in parr-to-smolt survival to parr growth and transition-zone with the findings of Connor et al. (2003a) and the weekly velocity, both of which were directly proportional to survival. trends reported by Smith et al. (2003). We identified differences As previously mentioned, growth rate reflects environmental in mean parr length (usually longer for cohort 1), reservoir and biological conditions that affect survival by regulating the velocity (higher for cohort 1), and reservoir temperature amount of time required to become a smolt, exceed predator (usually cooler for cohort 1) as factors for between-cohort gape sizes, and attain the swimming ability needed to elude differences in parr-to-smolt survival. The difference in mean predators. Increases in transition-zone velocity would reduce parr length was affected by skewed catch date distributions exposure time to predators, as discussed for reservoir velocity, for above-average-sized fish and the method used to delineate but increases in transition-zone velocity might also influence cohorts. In most years, the catch date distributions for above- survival in another way. Coutant and Whitney (2006) hypoth- average-sized fish were skewed to the right (data available at: esized that the perception of turbulence enables fish to orient to www.ptagis.org). Assuming a constant length at emergence velocities that promote net downstream movement. The switch and a constant growth rate between fry emergence and tagging in migrational disposition and subsequent delay in the transition placed a greater proportion of the larger fish into cohort 1 than zone reported for subyearlings by Tiffan et al. (2009) supported into cohort 2 as was usually observed by Connor et al. (2003a). this hypothesis. It is possible that the extent of cue loss is
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 The between-cohort differences in mean reservoir velocity and inversely proportional to transition-zone velocity. As a result, temperature were caused by the differences in the observed a reduction in cue loss would increase survival by reducing migration periods and seasonal runoff and temperature patterns. the time fish were in the presence of northern pikeminnow and Velocities in Lower Granite Reservoir decline and temperatures smallmouth bass that occupy transitional habitats. increase as the migration progresses. Thus, cohort 1 experi- We conclude that the results in this paper support measures enced higher mean velocities in the reservoir and cooler mean to increase velocity within the migration season when subyear- reservoir temperatures than were experienced by cohort 2. ling parr and smolts are present in Lower Granite Reservoir. We suggest that differences in mean parr length, reservoir The study also provides new empirical evidence that upholds the velocity, and reservoir temperature between cohorts 1 and 2 long-held belief that any factor that reduces growth of juvenile affected differences in susceptibility to predation by northern salmonids during freshwater rearing has the potential to reduce pikeminnow Ptychocheilus oregonensis (Shively et al. 1996) and juvenile survival. The specific factors for decreased growth smallmouth bass Micropterus dolomieu (Nelle 1999; Naughton across years will vary depending on the anadromous population et al. 2004). Many researchers have speculated that direct rela- under consideration. Three factors are globally applicable, tions between size of juvenile anadromous salmonids and sur- namely, fish management practices, land use, and climate vival may be associated with predation (chum salmon O. keta, change. The demand placed on fishery resources has increased PARR GROWTH AND PARR-TO-SMOLT SURVIVAL 1217
as the human population has increased. Hatcheries have become Chapman, D. W. 1966. Food and space as regulators of salmonid populations in an increasingly important tool for sating human needs and streams. American Naturalist 100:345–357. conserving imperiled populations (Schramm and Piper 1995). Chapman, D. W., and T. C. Bjornn. 1969. Distribution of salmonids in streams, with special reference to food and feeding. Pages 153–176 in T. G. North- Density-dependent competition for food and space between cote, editor. Symposium on salmon and trout streams. University of British hatchery and natural fish (e.g., Kostow 2009) may be a large fac- Columbia, Vancouver. tor that results in reduced natural fish productivity when hatch- Close, T. L., and C. S. Anderson. 1992. Dispersal, density-dependent growth, ery offspring greatly outnumber their natural counterparts. Land and survival of stocked steelhead fry in Lake Superior tributaries. North use practices have influenced changes in freshwater ecosystems American Journal of Fisheries Management 12:728–735. Connor, W. P., and H. L. Burge. 2003. Growth of wild subyearling fall Chinook that at the most fundamental level have altered the food web salmon in the Snake River. North American Journal of Fisheries Management that is the foundation of fish growth (ISAB 2011). Alterations to 23:594–599. stream temperature and flow projected over time under a suite Connor, W. P., H. L. Burge, R. Waitt, and T. C. Bjornn. 2002. Juvenile life of climate change scenarios (e.g., Parry et al. 2007) would prob- history of wild fall Chinook salmon in the Snake and Clearwater rivers. ably influence both growth during rearing and survival during North American Journal of Fisheries Management 22:703–712. Connor, W. P., H. L. Burge, J. R. Yearsley, and T. C. Bjornn. 2003a. Influence of seaward migration. Continued consideration of the effects of an- flow and temperature on survival of wild subyearling fall Chinook salmon in thropogenic activity on growth of juveniles from both imperiled the Snake River. North American Journal of Fisheries Management 23:362– and healthy populations of anadromous salmonids is warranted. 375. Connor, W. P., C. E. Piston, and A. P. Garcia. 2003b. Temperature during incubation as one factor affecting the distribution of Snake River fall Chi- ACKNOWLEDGMENTS nook salmon spawning areas. Transactions of the American Fisheries Society We thank our colleagues at the U.S. Fish and Wildlife Ser- 132:1236–1243. vice’s Idaho Fisheries Resource Office whose efforts contributed Connor, W. P., S. G. Smith, T. Andersen, S. M. Bradbury, D. C. Burum, E. E. Hockersmith, M. L. Schuck, G. W. Mendel, and R. M. Bugert. 2004. Postre- to the success of this study. This study (and many other stud- lease performance of hatchery yearling and subyearling fall Chinook salmon ies we have conducted) would not have been possible without released into the Snake River. North American Journal of Fisheries Manage- personnel of the Pacific States Marine Fisheries Commission, ment 24:545–560. including D. Marvin (through 2010) and N. Tancreto (2011), Connor, W. P., J. G. Sneva, K. F. Tiffan, R. K. Steinhorst, and D. Ross. 2005. who helped operate and maintain the Columbia Basin PIT-Tag Two alternative juvenile life history types for fall Chinook salmon in the Snake River basin. Transactions of the American Fisheries Society 134: Information System. The statistical review by K. Steinhorst and 291–304. peer-review by G. McMichael, B. Muir, and the journal’s as- Connor, W. P., R. K. Steinhorst, and H. L. Burge. 2003c. Migrational behavior sociate editor and editor improved this manuscript. This study and seaward movement of wild subyearling fall Chinook salmon in the Snake was funded by the Bonneville Power Administration and ad- River. North American Journal of Fisheries Management 23:414–430. ministered by D. Docherty and J. George under project number Cormack, R. M. 1964. Estimates of survival from the sighting of marked animals. Biometrika 51:429–438. 199102900 to meet reasonable and prudent alternatives (2004 Coutant, C. C. 2001. Turbulent attraction flows for guiding juvenile salmonids at Biological Opinion; 2008 Biological Opinion, 55.4, 50.3, 65.2). dams. Pages 57–77 in C. C. Coutant, editor. Behavioral technologies for fish Any use of trade, firm, or product names is for descriptive pur- guidance. American Fisheries Society, Symposium 26, Bethesda, Maryland. poses only does not imply endorsement by the U.S. Government. Coutant, C. C., and R. R. Whitney. 2006. Hydroelectric system development: The findings and conclusions in this article are those of the au- effects on juvenile and adult migration. Pages 249–324 in R. N. Williams, editor. Return to the river: restoring salmon to the Columbia River. Elsevier thors and do not necessarily represent the views of the U.S. Fish Academic Press, Amsterdam. and Wildlife Service. DART (Data Access in Real Time). 2011. Columbia River DART. Columbia Basin Research, University of Washington, Seattle. Available: www.cbr.washington.edu/dart/dart.html. (September 2011).
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 REFERENCES Dauble, D. D., T. L. Page, and R. W. Hanf. 1989. Spatial distribution of juvenile Achord, S., G. M. Matthews, O. W. Johnson, and D. M. Marsh. 1996. Use salmonids in the Hanford Reach, Columbia River. U.S. National Marine of passive integrated transponder (PIT) tags to monitor migration timing of Fisheries Service Fishery Bulletin 87:775–790. Snake River Chinook salmon smolts. North American Journal of Fisheries Dielman, T. E. 1996. Applied regression analysis for business and economics. Management 16:302–313. Wadsworth, Belmont, California. Achord, S., R. W. Zabel, and B. P. Sanford. 2007. Migration timing, growth, and Ebersole, J. L., P. J. Wigington Jr., J. P. Baker, M. A. Cairns, M. R. Church, B. P. estimated parr-to-smolt survival rates of wild Snake River spring–summer Hansen, B. A. Miller, H. R. LaVigne, J. E. Compton, and S. G. Leibowitz. Chinook salmon from the Salmon River basin, Idaho, to the lower Snake 2006. Juvenile coho salmon growth and survival across stream network sea- River. Transactions of the American Fisheries Society 136:142–154. sonal habitats. Transactions of the American Fisheries Society 135:1681– Beckman, B. R., W. W. Dickhoff, W. S. Zaugg, C. Sharpe, S. Hirtzel, R. Schrock, 1697. D. A. Larsen, R. D. Ewing, A. Palmisano, C. B. Schreck, and C. V. W. Everest, F. H., and D. W. Chapman. 1972. Habitat selection and spatial inter- Mahnken. 1999. Growth, smoltification, and smolt-to-adult return of spring action by juvenile Chinook salmon and steelhead trout in two Idaho streams. Chinook salmon from hatcheries on the Deschutes River, Oregon. Transac- Journal of the Fisheries Research Board of Canada 29:91–100. tions of the American Fisheries Society 128:1125–1150. Geist, D. R., Z. Deng, R. P. Mueller, S. R. Brink, and J. A. Chandler. 2010. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel Survival and growth of juvenile Snake River fall Chinook salmon exposed to inference: a practical information-theoretic approach. Springer Science, New constant and fluctuating temperatures. Transactions of the American Fisheries York. Society 139:92–107. 1218 CONNOR AND TIFFAN
Healey, M. C. 1982. Timing and relative intensity of size-selective mortality of Quinn, T. P., and N. P. Peterson. 1996. The influence of habitat complexity juvenile chum salmon (Oncorhynchus keta) during early sea life. Canadian and fish size on over-winter survival and growth of individually marked Journal of Fisheries and Aquatic Sciences 39:952–957. juvenile coho salmon (Oncorhynchus kisutch) in Big Beef Creek, Washington. ISAB (Independent Scientific Advisory Board). 2011. Columbia River basin Canadian Journal of Fisheries and Aquatic Sciences 53:1555–1564. food webs: developing a broader scientific foundation for fish and wildlife Rieman, B. E., R. C. Beamesderfer, S. Vigg, and T. P. Poe. 1991. Estimated restoration. Northwest Power and Conservation Council, Portland, Oregon. loss of juvenile salmonids to predation by northern squawfish, walleyes, and Available: www.nwcouncil.org/library/isab/2011-1/. (September 2011). smallmouth bass in John Day Reservoir, Columbia River. Transactions of the Kostow, K. 2009. Factors that contribute to the ecological risks of salmon and American Fisheries Society 120:448–458. steelhead hatchery programs and some mitigating strategies. Reviews in Fish Schramm, H. L., Jr., and R. G. Piper, editors. 1995. Uses and effects of cultured Biology and Fisheries 19:9–31. fishes in aquatic ecosystems. American Fisheries Society, Symposium 15, Krueger, D. M., E. S. Rutherford, and D. M. Mason. 2011. Influence of predation Bethesda, Maryland. mortality on survival of Chinook salmon parr in a Lake Michigan tributary. Shively, R. S., T. P. Poe, and S. T. Sauter. 1996. Feeding response by northern Transactions of the American Fisheries Society 140:147–163. squawfish to a hatchery release of juvenile salmonids in the Clearwater River, Lister, D. B., and H. S. Genoe. 1970. Stream habitat utilization by cohabiting Idaho. Transactions of the American Fisheries Society 125:230–236. underyearlings of Chinook (Oncorhychus tshawytscha) and coho (O. kisutch) Sigler, J. W., T. C. Bjornn, and F. H. Everest. 1984. Effects of chronic turbidity salmon in the Big Qualicum River, British Columbia. Journal of the Fisheries on density and growth of steelheads and coho salmon. Transactions of the Research Board of Canada 27:1215–1224. American Fisheries Society 113:142–150. Littell, R. C., G. A. Milliken, W. W. Stroup, and R. D. Wolfinger. 1996. SAS Skalski, J. R., S. G. Smith, R. N. Iwamoto, J. G. Williams, and A. Hoffman. 1998. system for mixed models. SAS Institute, Cary, North Carolina. Use of passive integrated transponder tags to estimate survival of migrant Miller, T. J., L. B. Crowder, J. A. Rice, and E. A. Marchall. 1988. Laval size and juvenile salmonids in the Snake and Columbia rivers. Canadian Journal of recruitment mechanisms in fishes: toward a conceptual framework. Canadian Fisheries and Aquatic Sciences 55:1484–1493. Journal of Fisheries and Aquatic Sciences 45:1657–1670. Smith, S. G., W. D. Muir, E. E. Hockersmith, R. W. Zabel, R. J. Graves, C. V. Moss, J. H., D. A. Beauchamp, A. D. Cross, K. W. Myers, E. V. Farley Jr., J. M. Ross, W. P. Connor, and B. D. Arnsberg. 2003. Influence of river conditions Murphy, and J. H. Helle. 2005. Evidence for size-selective mortality after the on survival and travel time of Snake River subyearling fall Chinook salmon. first summer of ocean growth by pink salmon. Transactions of the American North American Journal of Fisheries Management 23:939–961. Fisheries Society 134:1313–1322. Smith, S. G., W. D. Muir, J. G. Williams, and J. R. Skalski. 2002. Factors Muir, W. D., S. G. Smith, J. G. Williams, E. E. Hockersmith, and J. R. Skalski. associated with travel time and survival of migrant yearling Chinook salmon 2001. Survival estimates for migrant yearling Chinook salmon and steelhead and steelhead in the lower Snake River. North American Journal of Fisheries tagged with passive integrated transponders in the lower Snake and lower Management 22:385–405. Columbia rivers, 1993–1998. North American Journal of Fisheries Manage- Taylor, E. B., and J. D. McPhail. 1985. Burst swimming and size-related preda- ment 21:269–282. tion of newly emerged coho salmon Oncorhynchus kisutch. Transactions of Naughton, G. P., D. H. Bennett, and K. B. Newman. 2004. Predation on juvenile the American Fisheries Society 114:546–551. salmonids by smallmouth bass in the Lower Granite Reservoir system, Snake Tiffan, K. F., and W. P. Connor. 2011. Distinguishing between natural and River. North American Journal of Fisheries Management 24:534–544. hatchery Snake River fall Chinook salmon subyearlings in the field using Nelle, R. D. 1999. Smallmouth bass predation on juvenile fall Chinook salmon in body morphology. Transactions of the American Fisheries Society 140:21– the Hells Canyon Reach of the Snake River, Idaho. Master’s thesis. University 30. of Idaho, Moscow. Tiffan, K. F., T. J. Kock, W. P. Connor, F. Mullins, and R. K. Steinhorst. 2012. NMFS (National Marine Fisheries Service). 1992. Threatened status for Downstream movement of fall Chinook salmon juveniles in the lower Snake Snake River spring/summer Chinook salmon, threatened status for Snake River reservoirs during winter and early spring. Transactions of the American River fall Chinook salmon. Federal Register 57:78(22 April 1992):14653– Fisheries Society 141:285–293. 14663. Tiffan, K. F., T. J. Kock, C. A. Haskell, W. P. Connor, and R. K. Steinhorst. 2009. Orciari, R. D., G. H. Leonard, D. J. Mysling, and E. C. Schluntz. 1994. Survival, Water velocity, turbulence, and migration rate of subyearling fall Chinook growth, and smolt production of Atlantic salmon stocked as fry in a southern salmon in the free-flowing and impounded Snake River. Transactions of the New England stream. North American Journal of Fisheries Management American Fisheries Society 138:373–384. 14:588–606. Vigg, S., T. P. Poe, L. A. Prendergast, and H. C. Hansel. 1991. Rates of con- Ott, R. L. 1993. An introduction to statistical methods and data analysis, 4th sumption of juvenile salmonids and alternative prey fish by northern squaw- edition. Wadsworth, Belmont, California. fish, walleyes, smallmouth bass, and channel catfish in John Day Reservoir,
Downloaded by [Department Of Fisheries] at 20:00 25 September 2012 Parry, M. L., O. F. Canziani, J. P. Palutikof, P. J. van der Linden, and Columbia River. Transactions of the American Fisheries Society 120:421– C. E. Hanson, editors. 2007. Contribution of working group II to the 438. fourth assessment report of the intergovernmental panel on climate change, Webb, P. W. 1976. The effect of size on the fast-start performance of rain- 2007. Cambridge University Press, New York. Available: www.ipcc.ch/ bow trout Salmo gairdneri, and a consideration of piscivorous predator-prey publications and data/ar4/wg2/en/contents.html. (September 2011). interactions. Journal of Experimental Biology 65:157–177. Prentice, E. F., T. A. Flagg, and C. S. McCutcheon. 1990a. Feasibility of using Wedemeyer, G. A., R. L. Saunders, and W. C. Clarke. 1980. Environmen- implantable passive integrated transponder (PIT) tags in salmonids. Pages tal factors affecting smoltification and early marine survival of anadromous 317–322 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. Jester Jr., salmonids. U.S. National Marine Fisheries Service Marine Fisheries Review E. D. Prince, and G. A. Winans, editors. Fish-marking techniques. American 42(6):1–14. Fisheries Society, Symposium 7, Bethesda, Maryland. West, C. J., and P. A. Larkin. 1987. Evidence for size-selective mortality of juve- Prentice, E. F., T. A. Flagg, C. S. McCutcheon, and D. F. Brastow. 1990b. nile sockeye salmon (Oncorhynchus nerka) in Babine Lake, British Columbia. PIT-tag monitoring systems for hydroelectric dams and fish hatcheries. Pages Canadian Journal of Fisheries and Aquatic Sciences 44:712–721. 323–334 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. Jester Jr., Yanke, J. A. 2006. Effects of passive integrated transponder (PIT) tags and E. D. Prince, and G. A. Winans, editors. Fish-marking techniques. American elevated water temperatures on survival, growth, and physiology of Snake Fisheries Society, Symposium 7, Bethesda, Maryland. River fall Chinook salmon subyearlings (Oncorhynchus tshawytscha). Mas- PTAGIS (PIT Tag Information System). 2011. Columbia River basin PIT tag ter’s thesis. University of Idaho, Moscow. information system. Pacific States Marine Fisheries Commission, Portland, Zabel, R. W., and S. Achord. 2004. Relating size of juveniles to survival within Oregon. Available: www.ptagis.org. (September 2011). and among populations of Chinook salmon. Ecology 85:795–806. This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:01 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Swimming Depth, Behavior, and Survival of Atlantic Salmon Postsmolts in Penobscot Bay, Maine Mark D. Renkawitz a , Timothy F. Sheehan a & Graham S. Goulette b a National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Northeast Fisheries Science Center, 166 Water Street, Woods Hole, Massachusetts, 02543, USA b National Oceanic and Atmospheric Administration, Northeast Fisheries Science Center, Maine Field Station, 17 Godfrey Drive, Orono, Maine, 04473, USA
Version of record first published: 30 Jul 2012.
To cite this article: Mark D. Renkawitz, Timothy F. Sheehan & Graham S. Goulette (2012): Swimming Depth, Behavior, and Survival of Atlantic Salmon Postsmolts in Penobscot Bay, Maine, Transactions of the American Fisheries Society, 141:5, 1219-1229 To link to this article: http://dx.doi.org/10.1080/00028487.2012.688916
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. Transactions of the American Fisheries Society 141:1219–1229, 2012 American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.688916
ARTICLE
Swimming Depth, Behavior, and Survival of Atlantic Salmon Postsmolts in Penobscot Bay, Maine
Mark D. Renkawitz* and Timothy F. Sheehan National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Northeast Fisheries Science Center, 166 Water Street, Woods Hole, Massachusetts 02543, USA Graham S. Goulette National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Northeast Fisheries Science Center, Maine Field Station, 17 Godfrey Drive, Orono, Maine 04473, USA
Abstract To gain information on postsmolt dynamics of emigrating Atlantic salmon Salmo salar through Penobscot Estuary and Penobscot Bay, Maine, we conducted a telemetry experiment in 2005. We implanted 26 salmon smolts with ultrasonic depth tags, and monitored movement activity and fish passage with linear detection arrays through 44.2 km of the estuary and 45.5 km of the bay. During daylight in the bay, greater than 95% of the detections occurred in water depths of 5 m or less, but depths to 37 m were recorded. At night, 99% of the detections were in the top 5 m of the water column and maximum depth was 9 m. Overall survival was 39% and was highest for smaller fish and those released earlier in the smolt run, when river discharge was greater. Rapid emigration (i.e., approximately 1 km/h) and preferential surface orientation improved survival. These results verify that postsmolts are primarily surface oriented in the waters of Penobscot Bay and that they may experience high rates of nearshore mortality despite their short residence time. Detailed emigration and behavioral data such as these allow scientists and managers to delineate areas of high mortality to develop strategies that improve survival, and provide marine spatial planners information to minimize impacts of coastal zone development.
Marine survival indices for many populations of Atlantic In freshwater, smolts occupy deeper water during the day salmon Salmo salar across the North Atlantic have declined and move in the surface layers at night (Fried et al. 1978; over the last 20 years (Friedland et al. 1993; Hansen and Quinn McCleave 1978; Hansen and Jonsson 1985; Aarestrup et al.
Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 1998; Chaput et al. 2005), and current marine abundance re- 2002) as antipredator responses to piscine and avian preda- mains among the lowest on record (ICES 2011). Extensive mor- tion threats (Solomon 1982). Whether movement behavior and tality during the smolt (i.e., the beginning of seaward migration depth occupancy continues during the postsmolt phase in estu- in freshwater) and postsmolt stages (i.e., from estuary entry un- aries and further out to sea remains uncertain. Postsmolts have til the end of the first winter at sea; Allan and Ritter 1977) are been sampled with trawls from the surface to depths of 10 m implicated in the declines (Ritter 1989; Dieperink et al. 2002; (Levings et al. 1994; Shelton et al. 1997; Lacroix and Knox Potter et al. 2003; Kocik et al. 2009). Improved survival during 2005; Sheehan et al. 2011; Sheehan et al., in press) in coastal the marine phase is necessary for population recovery, mak- waters and with gill nets at sea (Dutil and Coutu 1988; Reddin ing the identification and quantification of potential survival and Short 1991). They have also been found in the stomachs bottlenecks priorities for scientists so managers can implement of surface-feeding seabirds (Dieperink et al. 2002; Montevec- mitigation plans. chi et al. 2002), marine mammals (Middlemas et al. 2003),
*Corresponding author: [email protected] Received November 28, 2011; accepted April 19, 2012 1219 1220 RENKAWITZETAL.
and larger pelagic fishes (Bigelow and Schroeder 1953; Scott freshwater–saltwater mixing zone advances and retreats up to and Scott 1988). However, typically epibenthic foragers such 20 km, depending on river discharge and tidal stage (Haefner as cod and pollock (family Gadidae), skates (family Rajidae), 1967). Penobscot Bay contains two main channels separated by and certain sharks also consume postsmolts (Hvidsten and Lund Islesboro Island and several smaller islands (Xue et al. 2000). 1988; Scott and Scott 1988; SOAFD 1993), which may indicate Generally, the surface waters in the western channel are influ- occasional occupancy of deeper water. Direct measurement of enced more strongly by river discharge, while the waters in the swimming depths and diving frequency of emigrating juveniles eastern channel are more strongly influenced by ocean currents is limited (Plantalech Manel-La et al. 2009). Active tracking of (Xue et al. 2000). juvenile Atlantic salmon indicates primarily surface orientation Telemetry receiver network.—VEMCO ultrasonic VR2 re- in riverine, estuarine, and coastal waters, although frequent div- ceivers were used to track emigrating salmon implanted with ing activity has been noted (McCleave 1978; Spicer et al. 1995; pressure sensor transmitters through Penobscot Estuary and Davidsen et al. 2008) and direct measurements of 10 m (LaBar Penobscot Bay (Figure 1). Stationary receivers (n = 57) were et al. 1978) and 50 m (Reddin et al. 2006) are documented. deployed in 15 linear arrays (nine in the estuary and six in the To investigate the mechanisms responsible for population bay) in conjunction with the U.S. Geological Survey (USGS) declines, researchers have incorporated nonlethal sampling Maine Cooperative Fish and Wildlife Research Unit to moni- (e.g., surface trawling) and tracking techniques (e.g., ultrasonic tor fish movement. Receivers were moored close to the bottom telemetry) during the early marine phase that minimize the dele- substrate in the estuary and suspended 10 m below the water terious impact on already-depleted study populations. While ad- surface in the bay. Since receivers had a detection radius of vances in surface trawling allow for capture and live release of 500 m, single receivers were deployed when the distance be- postsmolts after sampling (Holst and McDonald 2000; Rikard- tween opposing points of land were less than 1 km and multiple sen et al. 2004; Lacroix and Knox 2005; Sheehan et al. 2011), receivers were deployed when distances exceeded this thresh- variations in expected fish behavior (i.e., surface layer orien- old. When multiple receivers were necessary, spacing was ap- tation) could bias analysis of the catch data. Passive telemetry proximately 500 m between receivers to ensure that detection allows for the simultaneous tracking of multiple individual fish radii overlapped, thereby improving detection efficiency. Ar- through riverine, estuarine, and coastal waters (Voegeli et al. rays were deployed at approximately 3–5-km intervals through 1998), providing valuable information on migration dynamics, the estuary, while spacing was greater in the bay to allow for including swimming speeds, emigration rates, migration routes, increased seaward detection ability. Total linear coverage was and survival (Lacroix et al. 2004b; Økland et al. 2006; Thorstad 89.7 km (44.2 km in the estuary, 45.5 km in the bay). et al. 2007; Kocik et al. 2009). This information is necessary to Surgeries and releases.—Fish in this study were age-1 answer specific behavioral and ecological questions (Holbrook advanced growth hatchery smolts reared at the U.S. Fish et al. 2011) that inform management. and Wildlife Service’s Green Lake National Fish Hatchery Our goal was to utilize ultrasonic telemetry to directly mea- (GLNFH) in Ellsworth, Maine. Transmitters (VEMCO V9P-6 sure and quantify postsmolt swimming depth during emigra- L-69kHz-S256, 9 mm in diameter, 38 mm in length, and weigh- tion through Penobscot Bay, Maine. We obtained estimates of ing 4.6 g in air and 2.2 g in water) were implanted in 26 fish survival, swimming speed, migration behavior, and route delin- at GLNFH on 9 May 2005 using the surgical methods detailed eations through the bay. Additionally, we used the data to evalu- in Kocik et al. (2009). Each transmitter emitted a unique signal ate the ability of a surface trawl survey (Sheehan et al. 2011) to and pressure measurement at least once every 50 s, which the sample an emigrating postsmolt population. These data provide receivers recorded along with the date and time of detection. novel insights into the behavior of migrating postsmolts through The transmitters had an accuracy of 2.5 m to a depth of 50 m.
Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 the nearshore environment, and the information obtained will Fish greater than 170-mm FL were selected for tag implan- enable scientists and managers to mitigate losses by natural and tation to reduce transmitter effect on smolt behavior (Lacroix anthropogenic causes. et al. 2004a). Fish were measured (mm) and weighed (g) before release. Smolts recovered at the hatchery for 2 or 9 d prior to trans- METHODS port and release below Veazie Dam, the last passage barrier in Study site.—The Penobscot River drains an area of the system (Figure 1). Two groups of 13 fish were released: 22,196 km2 (Haefner 1967). During this study, mean ± SD release group 1 (RG1) on 11 May 2005 at 1210 hours, and river discharge was 697 ± 285 m3/s, minimum was 391 m3/s, release group 2 (RG2) on 18 May 2005 at 0950 hours. Water and maximum was 1,385 m3/s (USGS 2007; http://waterdata. temperatures at the time of release were the same (i.e., 11.0◦C usgs.gov/, gauge number 01034500). Penobscot Estuary (Fig- and 11.5◦C on 11 May and 18 May, respectively) and within ure 1) stretches over 51 km seaward from Veazie Dam, hav- 1◦C of tank temperatures during transport, well below tempera- ing a mean depth of 9 m and a maximum depth of 31 m. tures known to induce loss of smolt characteristics (McCormick Salinity in the upper estuary is generally negligible and in- et al. 1999). However, river discharge was almost two times creases gradually with seaward progression (Mitnik 2002). The greater on 11 May (1,011 m3/s) than on 18 May (538 m3/s; SWIMMING DEPTH OF ATLANTIC SALMON POSTSMOLTS 1221 Downloaded by [Department Of Fisheries] at 20:01 25 September 2012
FIGURE 1. Map of Penobscot Estuary and Penobscot Bay, Maine, illustrating the locations of telemetry arrays (I–XIa–c), the release site, the freshwater–saltwater mixing zone, and the three potential egress passages available to emigrating postsmolts. 1222 RENKAWITZETAL.
USGS 2007). Releases were intended to bracket the smolt run recorded by the receivers. No distinct patterns of missed detec- according to historic in-river monitoring data (National Marine tions were detected with any of the arrays. Detection efficiencies Fisheries Service, unpublished data) and during daylight to co- were not obtained for array XIa–c. incide with bulk releases of hatchery smolts into the system. Data analysis.—All statistical analyses were conducted in Biological Characteristics Minitab, release 13.1 (Minitab, Inc., State College, Pennsylva- Mean ± SD smolt FL and weight of the study population was nia) using the general linear model routine, and results were 193.2 ± 9.0 mm (range = 172–206 mm) and 75.3 ± 10.5 g, re- evaluated after verifying model assumptions (i.e., normality, spectively (range = 53.7–91.3 g), and overlapped with the over- independence, and homogeneity). An empirically derived de- all mean length and weight distribution of the hatchery popula- tection efficiency (DE) for each array was determined by identi- tion in 2005 (length = 180.5 ± 12.3 mm, range = 108–232 mm; fying missed detections of unique fish at an array subsequently weight = 71.3 ± 16.9 g, range = 35.4–133.8 g). Mean ± detected at seawards arrays. We calculated DE according to the SD transmitter length-to-smolt length and weight proportions formula were 3.2 ± 0.5% and 20.8 ± 1.0%, respectively, slightly greater than sizes recommended by Lacroix et al. (2004a). Fish DEi = Di /D(i→i+n), in RG1 were statistically smaller than those in RG2 in mean length ± SD (one-way ANOVA: F1, 24 = 8.79, p = 0.007; where Di is the number of unique detections at the ith array, RG1 = 183.5 ± 7.1 mm, RG2 = 195.0 ± 9.0 mm) and weight and D(i→i + n) is the total number of unique detections at array i ± SD (one-way ANOVA: F1, 24 = 9.82, p = 0.005; RG1 = through i + n (i.e., the most seaward array). Detection efficiency 63.9 ± 7.8 g, RG2 = 75.3 ± 10.5 g). Significantly more fish was not calculated for array XIa–c because this was the most sea- successfully emigrated from RG1 (61.5%) than RG2 (15.4%; ward array in the telemetry network and missed detections could one-way ANOVA: F1, 25 = 7.85, p = 0.010). Even though 80% not be quantified. Postsmolts were considered successful if de- (8 of 10) of successful migrants were less than 190-mm FL tected at the outermost array (i.e., array XIa–c), and estimates (mean = 183.9 ± 8.25-mm FL) and 56% (9 of 16) of unsuccess- of survival were considered minimal because equipment failure ful migrants were greater than 190-mm FL (mean = 191.0 ± (e.g., transmitter malfunction, etc.) and missed detections at the 9.12-mm FL), migration success was not influenced by length outermost array were not quantified. (ANOVA: F1, 24 = 4.00, p = 0.057) or weight (ANOVA: F1, 24 = All depth detection data were grouped in 1-m bins for each 2.32, p = 0.141; successful = 65.7 ± 11.3 g; unsuccessful = fish at an array to evaluate the vertical usage of the water column. 72.1 ± 11.3 g). Mean detections of each fish at an individual array were also evaluated to determine if swimming depth influenced survival or Swimming Depth differed with habitat (estuary or bay), between release groups, Postsmolts were primarily detected in the top 5 m of the and with light condition. water column throughout their migration (Figure 2). Greater Emigration rate (km/h) was defined as the total time (h) an depths were detected more frequently in the estuary than the bay, individual postsmolt required to traverse the study area over the yet maximum depths were greatest in the bay. In the estuary, total linear midchannel distance (km) between arrays. Swim- 86.7% of the total detections were in the top 5 m of the water ming speed was calculated as the travel time (i.e., the difference column: 99.3% at dawn, 83.7% during day, 86.4% during night, between the last detection at an upstream array and the first de- and 94.6% at dusk conditions. In the bay, 98.2% of the total tection at a seaward array) over the linear midchannel distance detections occurred in the top 5 m: 100.0% at dawn, 95.1% between arrays and was calculated in both body lengths per during day, 99.9% during night, and 84.4% at dusk.
Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 second (bl/s) and kilometers per hour (km/h) for each fish. While mean swimming depths indicated postsmolts were Migration data were also characterized by daylight condi- generally surface oriented, maximum swimming depths indi- tions as defined on the median day of the study period. Dawn cated they utilized the entire water column (Figure 3). Differ- was defined as 0501–0700 hours (i.e., 1 h on either side of sun- ences in depths were detected between successful and unsuc- rise), day as 0701–1900 hours, dusk as 1901–2100 hours (i.e., cessful postsmolts from the different release groups. Overall, 1 h on either side of sunset), and night as 2101–0500 hours. All RG1 was detected at slightly shallower mean depths ± SE than detections were assigned one light condition. RG2 (1.33 ± 0.03 m and 2.71 ± 0.05 m, respectively; one-way ANOVA: F1, 7752 = 378.72, p < 0.001). Within RG1, successful ± RESULTS migrants were detected at slightly shallower mean depths SE than unsuccessful migrants (1.05 ± 0.03 m and 1.85 ± Detection Efficiency 0.06 m, respectively; one-way ANOVA: F1, 3326 = 144.68, Empirically derived detection efficiencies ranged from 88% p < 0.001). In RG2, successful postsmolts were also detected at to 100%. Of the known missed detections, 13 of 14 occurred in slightly shallower depths than unsuccessful fish (2.45 ± 0.12 m the estuary. High river discharge may have moved fish through and 2.74 ± 0.05 m, respectively; one-way ANOVA: F1, 4424 = the detection fields in the estuary before a signal could be 8.77, p = 0.003). Although no clear patterns among postsmolts SWIMMING DEPTH OF ATLANTIC SALMON POSTSMOLTS 1223 Proportion of detections (%) 0 10203040500 102030405060708090 0
5 + + + 10 + + + + + 15 + + + + + + 20 + 25 + 30
+ Detection depth (m) 35 Estuary + Bay 40
FIGURE 2. Detection depths of emigrating Atlantic salmon postsmolts in Penobscot Estuary (arrays I–VIII) and Penobscot Bay (arrays VIII–XI), Maine,in 2005. Proportions are based on 19,217 detections in the estuary and 1,306 detections in the bay (note scales for x-axis and y-axis differ); crosses denote depth detection proportions less than 0.5%.
were evident, individualized diving patterns in the bay lasted for various durations to depths of approximately 35 m.
Emigration Routes Twenty-one (80.8%) postsmolts emigrated through the es- Distance from release (km) tuary and were detected in Penobscot Bay (i.e., at array IX). Overall, 10 fish (38.5%) successfully emigrated to the outer- 0 102030405060708090 most arrays (Figure 4); nine fish emigrated through the western 0 passage, and one emigrated through the eastern passage (Fig- 5 ure 5). Through the middle bay, three fish were detected passing array Xa and 12 postsmolts were detected passing array Xb.One 10 postsmolt was not detected on either side of Islesboro (i.e., by array Xa or Xb), but successfully emigrated via array XIa.No Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 15 postsmolts were detected leaving the bay through the central 20 passage array XIb. Of the 16 unsuccessful postsmolts, five were Estuary Bay last detected in the estuary: one at array II, one at array V, and 25 two at array VIII. Eleven postsmolts were last detected in the bay: five at array IX and six at array X . 30 a–b Swimming Speeds and Emigration Time Detection depth (m) 35 Successful postsmolts initiated emigration immediately upon 40 release, whereas unsuccessful fish delayed emigration for sev- eral hours (Figure 6). All postsmolts that reached a distance FIGURE 3. Mean (solid symbols) and maximum (open symbols) swimming of 50 km from the release site within approximately 50 h suc- depths (m) of successful (circles) and unsuccessful (triangles) Atlantic salmon cessfully emigrated through the system. The mean emigration postsmolts emigrating through Penobscot Estuary and Penobscot Bay, Maine, ± ± in 2005. The shaded region represents maximum seafloor depth at the telemetry rate SD for successful postsmolts was 1.00 0.20 km/h arrays, and the vertical line is the approximate freshwater–saltwater transition (maximum = 1.38 km/h, minimum = 0.77 km/h) and 2.25 ± zone. 1.70 km/h (maximum = 5.93 km/h, minimum = 0.46 km/h) 1224 RENKAWITZETAL.
100 River Discharge and Tides The 5-d mean ± SD river discharge postrelease was 867 ± 137 m/s for RG1 and 487 ± 63 m/s for RG2. On outgoing tides, 80 6 of 12 postsmolts that entered Penobscot Bay successfully emigrated through the system (five from RG1 and one from 60 RG2), and four of nine postsmolts (three from RG1 and one from RG2) successfully emigrated through the system on incoming es mate (%) es mate Estuary tides. 40
DISCUSSION 20 Bay Our data provide novel results that quantify the vertical distri- Survival butions of Atlantic salmon postsmolts 45 km seaward. Hatchery- 0 reared postsmolts primarily occupied the surface waters in 020406080100Penobscot Bay throughout the day, indicating that nearshore sur- face trawling is an appropriate survey method for sampling the Distance from release (km) emigrating population in the Gulf of Maine. The size distribu- tion of the study population overlapped with the hatchery smolt FIGURE 4. Minimum survival estimates of two groups of emigrating Atlantic population (75%; National Oceanic and Atmospheric Adminis- salmon postsmolts through Penobscot Estuary and Penobscot Bay, Maine, in tration, unpublished data) and the emigrating postsmolt popu- May 2005. Triangles represent RG1 (n = 13), while circles represent RG2 (n = 13). The overall survival is represented by the squares. lation from smolt stocked origins (90%; Sheehan et al. 2011). Since approximately 95% of the emigrating cohort originates from hatchery smolt production (Sheehan et al. 2011), the re- sults of this study likely reflect the overall emigration dynamics for unsuccessful postsmolts. The mean ± SD number of days of the Penobscot Hatchery population. We detected evidence of for successful postsmolt emigration (i.e., to array XI )was a–c individualized postsmolt diving behaviors, sometimes proximal 3.4 ± 0.7 d (maximum = 4.9, minimum = 2.7) and was con- to the bottom substrate. Stomach content analysis (Renkawitz sistent between release groups (RG1 = 3.41 ± 0.78 d, RG2 = and Sheehan 2011) and temperature records of recovered data 3.43 ± 0.63 d). storage tags (Reddin et al. 2006) suggest this behavior may be Mean ± SD swimming speeds (bl/s) were not consistent for foraging or diving to evade avian predators (Solomon 1982; between arrays, and individual speeds were variable. Mean ± Hansen and Johnsson 1985; Mather 1998; Aarestrup et al. 2002). SD swimming speeds were more variable in the upper estu- Alternatively, predation threats from piscine sources from be- ary (arrays I–III; 0.97 ± 0.83 bl/s) than in the lower estuary low may force some postsmolts closer to the surface, and sur- (arrays IV–VII; 0.82 ± 0.46 bl/s). After postsmolts entered vival of those that remain deeper may be at greater risk of the bay, swimming speeds decreased with offshore progression predation. Vertical positioning in the water column to occupy (Figure 7), potentially a result of the decreased influence of di- favorable environmental regimes (i.e., temperature and salin- rectional river discharge. Mean ± SD swimming speeds of suc- ity) is also possible (Reddin et al. 2006; Plantalech Manel-La cessful (4.04 ± 2.79 bl/s) and unsuccessful (4.14 ± 2.68 bl/s) et al. 2009) and may explain why estimates of survival were postsmolts were similar in the estuary and the bay (ANCOVA: greater for surface-oriented postsmolts than postsmolts at deeper F1, 190 = 3.18, p = 0.076), and they did not differ between the Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 depths. release groups (RG1 = 4.09 ± 2.66 bl/s, RG2 = 4.10 ± 2.81 Migratory behavior was consistent with overall patterns ob- bl/s; ANCOVA: F = 0.23, p = 0.633). 1, 190 served in previous telemetry studies in coastal Maine systems (Fried et al. 1978; LaBar et al. 1978; Kocik et al. 2009). Most Movement and Light Condition successful postsmolts migrated via the Western Passage into the The tendency for nocturnal estuarine movement decreased Gulf of Maine, suggesting that this may be the primary migra- with seaward progression (Figure 8). Successful migrants tion corridor and point of egress through Penobscot Bay (Shee- moved under dark conditions in the upper estuary (i.e., arrays han et al. 2011). Postsmolts relied less on nocturnal movement I–VI), but once in the lower estuary (array VII) movement oc- with seaward progression in brackish water (Moore et al. 1995; curred irrespective of light conditions. While unsuccessful mi- Aprahamian and Jones 1997; Kocik et al. 2009). The nocturnal grant behavior followed a similar pattern, a higher proportion movement in the upper estuary, coupled with the occupancy of the movements persisted during daylight conditions in the of deeper water during daylight, is suggestive of active visual estuary. All successful postsmolts exited the study area dur- predator avoidance strategies that enhance survival (Moore et al. ing daylight conditions: dawn (20%), daylight (70%), and dusk 1995; Davidsen et al. 2008). The observed patterns of move- (10%). ment could have also been a result of stocking timing (i.e., SWIMMING DEPTH OF ATLANTIC SALMON POSTSMOLTS 1225 Downloaded by [Department Of Fisheries] at 20:01 25 September 2012
FIGURE 5. Emigration corridors of Atlantic salmon postsmolts in Penobscot Bay, Maine, in 2005. Numbers in boxes represent the number of postsmolts that were detected passing the associated arrays. One postsmolt passed through array XIa and was not detected at the preceding arrays. No smolts were detected passing through array XIb. 1226 RENKAWITZETAL.
180 100 160 Estuary 140 80 120 100 60 80 60 40 Time (hours) Time 40 20 20 Bay 0 0 102030405060708090 0 Standard I II III IV V VI VII VIII IV X XI Propor on of movement (%) movement of Propor on Distance (km) Array
FIGURE 6. Mean cumulative emigration rates (h) of successful (solid grey FIGURE 8. Proportion of movement of Atlantic salmon postsmolts by light triangles) and unsuccessful (solid black circles) Atlantic salmon postsmolts condition (dawn = light grey; day = white; dusk = dark grey; night = black) emigrating through Penobscot Estuary and Penobscot Bay, Maine, in 2005. through Penobscot Estuary (arrays I–VIII) and Penobscot Bay (arrays IX–XI), Dashed lines represent the SEs of successful (grey) and unsuccessful (black) Maine, in 2005. The light standard represents the amount of time each light postsmolts. The solid vertical line denotes the approximate freshwater–saltwater condition persisted in a 24-h d. The high proportion of movement at the first transition zone. array was likely an artifact of the proximity to the release site and not active migration. all stocking around midday). Alternatively, since the fish were from hatchery origin without previous experience under natural High mortality rates through estuaries and at specific river conditions, behavioral mechanisms may have played a role nearshore locations may result from an abundant predator suite (Johnsson et al. 2001). (Jepsen et al. 2006), the stress associated with transitioning from freshwater to salt water (Larsson 1985; Lacroix and McCurdy 1996; Dieperink et al. 2002), or through a combi- ) 15.0 nation of effects. In this study, survival was generally higher in -1 the estuary than the bay. High river discharge in 2005 may have 12.5 facilitated rapid movement through the estuary and higher sur- vival (Aarestrup et al. 2002). Entry into the bay did not depend 10.0 on tidal stage. While movement through the bay was rapid for successful fish (approximately 2 d), the survival of postsmolts that entered the bay was low. This may have resulted from fac- 7.5 tors such as an inability to adjust to new osmoregulatory and
Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 + metabolic requirements (Hoar 1988; McCormick et al. 1998; + + 5.0 + + Stefansson et al. 2003) or the inability to meet increased energy + + demands upon ocean entry (Farmer et al. 1978). Alternatively, 2.5 + compromised physical condition from osmotic stress may have + + altered normal defensive responses, making postsmolts more 0.0 susceptible to predation (Larsson 1985; Jarvi¨ 1989). Although not statistically significant, emigration success was greater for Swimming speed (bl·sec Inter-array distance (seaward) smaller than larger postsmolts. This may have resulted from a size mismatch with the co-occurring species complex (Sheehan FIGURE 7. Swimming speeds (bl/s) of Atlantic salmon postsmolts in Penob- et al. 2011); the predation buffer was low for larger postsmolts, scot Estuary and Penobscot Bay, Maine, in 2005. The x-axis represents distances resulting in higher predation and lower survival in this study between telemetry arrays and is not to scale. The boxes and whiskers represent population. the fifth and 95th percentiles upper and lower quartiles, and the median. Crosses represent mean swimming speed between each array, and data points (repre- River discharge, tidal stage, and the direction and magni- sented by a degree symbol) are presented to the left of the boxes. Black points tude of sea surface currents generally dictate migration speeds are from arrays in the estuary, grey represents arrays in the bay. through estuaries and embayments (Moore et al. 1995; Lacroix SWIMMING DEPTH OF ATLANTIC SALMON POSTSMOLTS 1227
and McCurdy 1996). Fish utilize these features to expedite tial survival bottlenecks priorities for scientists so managers can downstream migration and improve seaward survival (Hosmer implement mitigation plans. et al. 1979; Hvidsten and Hansen 1988). Consistent with other studies (Lacroix et al. 2005; Hyvarinen¨ et al. 2006; Hedger 2008), rapid emigration was critical for survival in Penobscot ACKNOWLEDGMENTS Bay, given that all successful postsmolts traveled approximately This work is dedicated to our friend and colleague, Edward 50 km within 50 h of release. Even though postsmolt survival Hastings. Reference to trade names does not imply endorsement was higher and river discharge was greater for RG1 than RG2, by the U.S. Government. survival through the estuary was generally similar. However, the benefit of rapid transit due to high river discharge likely dissi- pated with offshore progression, where survival estimates were REFERENCES lowest. This suggests mortality during the early marine phase is Aarestrup, K., C. Nielsen, and A. Koed. 2002. Net ground speed of downstream migrating radio-tagged Atlantic salmon (Salmo salar L.) and brown trout higher than previously assumed (Potter et al. 2003) and could (Salmo trutta L.) smolts in relation to environmental factors. Hydrobiologia have a particularly large influence on the overall marine survival 483:95–102. of this population (Ritter 1989; Dieperink et al. 2002). Allan, I. R. H., and J. A. Ritter. 1977. Salmonid terminology. ICES Journal of The variation in emigration time is suggestive of individual Marine Science 37:293–299. acclimatization periods and movement patterns. The general Aprahamian, M. W., and G. O. Jones. 1997. The seaward movement of Atlantic salmon smolts in the Usk estuary, Wales, as inferred from power station change in swimming speeds through the estuary and with off- catches. Journal of Fish Biology 50:442–444. shore progression indicates that river and ocean hydrology influ- Bigelow, H. B., and W. C. Schroeder. 1953. Fishes of the Gulf of Maine. U.S. ences postsmolt transit speed (Thorstad et al. 2004; Økland et al. Fish and Wildlife Service Fishery Bulletin 74. 2006; Thorstad et al. 2007). The drastic reduction in swimming Chaput, G., C. M. Legault, D. G. Reddin, F. Caron, and P. G. Amiro. 2005. speeds at array III may have been the result of environmental Provision of catch advice taking account of non-stationarity in productivity of Atlantic salmon (Salmo salar L.) in the northwest Atlantic. ICES Journal factors (such as a strong tidal influence) or may have been where of Marine Science 62:131–143. fish first detected the elevated saline conditions that occur prox- Davidsen, J. G., N. Plantalech Manel-la, F. Økland, O. H. Diserud, E. B. imate to the array (Haefner 1967; Mitnik 2002). The reduced Thorstad, B. Finstad, R. Sivertsgard,˚ R. S. McKinley, and A. H. Rikard- variation in swimming speed in the bay (i.e., from array IX to sen. 2008. Changes in swimming depths of Atlantic salmon Salmo salar arrays XI ) indicates that once postsmolts adjusted to ocean post-smolts relative to light intensity. Journal of Fish Biology 73:1065–1074. a–c Dieperink, C., B. D. Bak, L. F. Pedersen, M. I. Pedersen, and S. Pedersen. conditions, migration speeds became steady and active com- 2002. Predation on Atlantic salmon and sea trout during their first days as pared with fluctuating and passive in the estuary, as suggested postsmolts. Journal of Fish Biology 61:848–852. by Martin et al. (2009). Dutil, J. D., and J. M. Coutu. 1988. Early marine life of Atlantic salmon, Salmo Detailed emigration data allows scientists and managers salar, postsmolts in the northern Gulf of St. Lawrence. U.S. National Marine to delineate areas of high mortality and develop strategies Fisheries Service Fishery Bulletin 86:197–212. Farmer, G. J., J. A. Ritter, and D. Ashfield. 1978. Seawater adaptation and parr- that improve survival. Quantifying the vertical distribution of smolt transformation of juvenile Atlantic salmon, Salmo salar. Journal of the postsmolts in the water column and the reasons why they locate Fisheries Research Board of Canada 35:93–100. there is important to understanding nearshore ecology and usage Fried, S. M., J. D. McCleave, and G. W. LaBar. 1978. Seaward migration of patterns. Generally, smaller fish released early when river dis- hatchery-reared Atlantic salmon, Salmo salar, smolts in the Penobscot River charge was higher had faster emigration rates and higher survival estuary, Maine: riverine movements. Journal of the Fisheries Research Board of Canada 35:76–87. than those released later at a larger size when river discharge Friedland, K. D., D. G. Reddin, and J. F. Kocik. 1993. Marine survival of North and emigration speeds were lower. Fish that swam closer to the American and European Atlantic salmon: effects of growth and environment.
Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 surface had higher survival than those that swam deeper, possi- ICES Journal of Marine Science 50:481–492. bly due to the greater risk of predation from piscine threats from Haefner, P. A., Jr. 1967. Hydrography of the Penobscot River (Maine) estuary. below than avian threats from above. Future work would benefit Journal of the Fisheries Research Board of Canada 24:1553–1571. Hansen, L. P., and B. Jonsson. 1985. Downstream migration of hatchery-reared from larger sample sizes under variable conditions designed to smolts of Atlantic salmon (Salmo salar L.) in the River Imsa, Norway. Aqua- answer specific questions of interest. culture 45:237–248. These data enable quantitative assessment of potential im- Hansen, L. P., and T. P. Quinn. 1998. The marine phase of the Atlantic salmon pacts to Atlantic salmon postsmolts in areas of coastal zone (Salmo salar) life cycle, with comparisons to Pacific salmon. Canadian Jour- development (i.e., in regions where undersea mining, dredging, nal of Fisheries and Aquatic Sciences 55(Supplement 1):104–118. Hedger, R. D., F. Martin, D. Hatin, F. Caron, F. G. Whoriskey, and J. J. Dodson. and tidal energy projects, etc.) are being considered (Pelc and 2008. Active migration of wild Atlantic salmon Salmo salar smolt through a Fujita 2002). This study provides managers with direct informa- coastal embayment. Marine Ecology Progress Series 355:235–246. tion that can be used for ecological and environmental impact Hoar, W. S. 1988. The physiology of smolting salmonids. Pages 275–343 in assessments to facilitate preemptive mitigation of deleterious W. S. Hoar and D. J. Randall, editors. Fish physiology, 2nd edition. Academic anthropogenic influences on emigrating postsmolts. Improved Press, New York. Holbrook, C. M., M. T. Kinnison, and J. Zydlewski. 2011. Survival of migrating survival during the marine phase is necessary for population Atlantic salmon smolts through the Penobscot River, Maine: a prerestoration recovery, making the identification and quantification of poten- assessment. Transactions of the American Fisheries Society 140:1255–1268. 1228 RENKAWITZETAL.
Holst, J. C., and A. McDonald. 2000. FISH-LIFT: a device for sampling live Mather, M. E. 1998. The role of context-specific predation in understanding fish with trawls. Fisheries Research 48:87–91. patterns exhibited by anadromous salmon. Canadian Journal of Fisheries and Hosmer, M. J., J. G. Stanley, and R. W. Hatch. 1979. Effects of hatchery Aquatic Sciences 55(Supplement 1):232–246. procedures on later return of Atlantic salmon to rivers in Maine. Progressive McCleave, J. D. 1978. Rhythmic aspects of estuarine migration of hatchery- Fish-Culturist 41:115–119. reared Atlantic salmon (Salmo salar) smolts. Journal of Fish Biology 12:559– Hvidsten, N. A., and L. P. Hansen. 1988. Increased recapture rate of adult 570. Atlantic salmon, Salmo salar L., stocked as smolts at high water discharge. McCormick, S. D., R. A. Cunjak, B. Dempson, M. F. O’Dea, and J. B. Carey. Journal of Fish Biology 32:153–154. 1999. Temperature-related loss of smolt characteristics in Atlantic salmon Hvidsten, N. A., and R. A. Lund. 1988. Predation on hatchery-reared and wild (Salmo salar) in the wild. Canadian Journal of Fisheries and Aquatic Sciences smolts of Atlantic salmon, Salmo salar L., in the estuary of River Orkla, 56:1649–1658. Norway. Journal of Fish Biology 33:121–126. McCormick, S. D., L. P. Hansen, T. P. Quinn, and R. L. Saunders. 1998. Move- Hyvarinen,¨ P., P. Suuronen, and T. Laaksonen. 2006. Short-term movements of ment, migration, and smolting of Atlantic salmon (Salmo salar). Canadian wild and reared Atlantic salmon smolts in a brackish water estuary: prelimi- Journal of Fisheries and Aquatic Sciences 55(Supplement 1):77–92. nary study. Fisheries Management and Ecology 13:399–401. Middlemas, S. J., J. D. Armstrong, and P. M. Thompson. 2003. The significance ICES (International Council for the Exploration of the Sea). 2011. Re- of marine mammal predation on salmon and sea trout. Pages 43–60 in D. port of the working group on North Atlantic salmon (WGNAS). ICES, Mills, editor. Salmon at the edge. Blackwell Scientific Publications, Oxford, CM 2011/ACOM:09, Copenhagen. Available: www.ices.dk/reports/ACOM/ UK. 2011/WGNAS/wgnas 2011 final.pdf. (August 2011). Mitnik, P. 2002. Penobscot River data report, May 2002. Maine Department of Jarvi,¨ T. 1989. Synergistic effect on mortality in Atlantic salmon, Salmo salar, Environmental Protection, Report DEPLW-0484, Augusta. smolt caused by osmotic stress and presence of predators. Environmental Montevecchi, W. A., D. K. Cairns, and R. A. Myers. 2002. Predation on marine- Biology of Fishes 26:149–152. phase Atlantic salmon (Salmo salar) by gannets (Morus bassanus)inthe Jepsen, N., E. Holthe, and F. Økland. 2006. Observations of predation on salmon northwest Atlantic. Canadian Journal of Fisheries and Aquatic Sciences and trout smolts in a river mouth. Fisheries Management and Ecology 13:341– 59:602–612. 343. Moore, A., E. C. E. Potter, N. J. Milner, and S. Bamber. 1995. The migratory Johnsson, J. I., J. Hojesj¨ o,¨ and I. A. Fleming. 2001. Behavioural and heart behaviour of wild Atlantic salmon (Salmo salar) smolts in the estuary of rate responses to predation risk in wild and domesticated Atlantic salmon. the River Conwy, North Wales. Canadian Journal of Fisheries and Aquatic Canadian Journal of Fisheries and Aquatic Sciences 58:788–794. Sciences 52:1923–1935. Kocik, J. F., J. P. Hawkes, T. F. Sheehan, P. A. Music, and K. F. Beland. Økland, F., E. B. Thorstad, B. Finstad, R. Sivertsgard,˚ N. Plantalech, N. Jepsen, 2009. Assessing estuarine and coastal migration and survival of wild At- and R. S. McKinley. 2006. Swimming speeds and orientation of wild Atlantic lantic salmon smolts from the Narraguagus River, Maine using ultrasonic salmon post-smolts during the first stage of the marine migration. Fisheries telemetry. Pages 293–310 in A.Haro,K.L.Smith,R.A.Rulifson,C.M. Management and Ecology 13:271–274. Moffitt, R. J. Klauda, M. J. Dadswell, R. A. Cunjak, J. E. Cooper, K. L. Pelc, R., and R. M. Fujita. 2002. Renewable energy from the ocean. Marine Beal, and T. S. Avery, editors. Challenges for diadromous fishes in a dynamic Policy 26:471–479. global environment. American Fisheries Society, Symposium 69, Bethesda, Plantalech Manel-la, N., E. B. Thorstad, J. G. Davidsen, F. Økland, R. Maryland. Sivertsgard,˚ R. S. McKinley, and B. Finstad. 2009. Vertical movements of LaBar, G. W., J. D. McCleave, and S. M. Fried. 1978. Seaward migration of Atlantic salmon post-smolts relative to measures of salinity and water temper- hatchery-reared Atlantic salmon (Salmo salar) smolts in the Penobscot River ature during the first phase of the marine migration. Fisheries Management estuary, Maine: open-water movements. ICES Journal of Marine Science and Ecology 16:147–154. 38:257–269. Potter, E. C. E., N. O’Maoileidigh,´ and G. Chaput. 2003. Marine mortality of Lacroix, G. L., and D. Knox. 2005. Distribution of Atlantic salmon (Salmo Atlantic salmon, Salmo salar L.: methods and measures. Canadian Science salar) postsmolts of different origins in the Bay of Fundy and Gulf of Maine Advisory Secretariat Research Document 2003/101. and evaluation of factors affecting migration, growth, and survival. Canadian Reddin, D. G., P. Downton, and K. D. Friedland. 2006. Diurnal and nocturnal Journal of Fisheries and Aquatic Sciences 62:1363–1376. temperatures for Atlantic salmon postsmolts (Salmo salar L.) during their Lacroix, G. L., D. Knox, and P. McCurdy. 2004a. Effects of implanted dummy early marine life. U.S. National Marine Fisheries Service Fishery Bulletin acoustic transmitters on juvenile Atlantic salmon. Transactions of the Amer- 104:415–428. ican Fisheries Society 133:211–220. Reddin, D. G., and P. B. Short. 1991. Postsmolt Atlantic salmon (Salmo salar) Lacroix, G. L., D. Knox, and M. J. W. Stokesbury. 2005. Survival and behaviour in the Labrador Sea. Canadian Journal of Fisheries and Aquatic Sciences
Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 of post-smolt Atlantic salmon in coastal habitat with extreme tides. Journal 48:2–6. of Fish Biology 66:485–498. Renkawitz, M. D., and T. F. Sheehan. 2011. Feeding ecology of early marine Lacroix, G. L., and P. McCurdy. 1996. Migratory behaviour of post-smolt phase Atlantic salmon Salmo salar post-smolts. Journal of Fish Biology Atlantic salmon during initial stages of seaward migration. Journal of Fish 79:356–373. Biology 49:1086–1101. Rikardsen, A. H., M. Haugland, P. A. Bjørn, B. Finstad, R. Knudsen, J. B. Lacroix, G. L., P. McCurdy, and D. Knox. 2004b. Migration of Atlantic salmon Dempson, J. C. Holst, N. A. Hvidsten, and M. Holm. 2004. Geographical postsmolts in relation to habitat use in a coastal system. Transactions of the differences in marine feeding of Atlantic salmon post-smolts in Norwegian American Fisheries Society 133:1455–1471. fjords. Journal of Fish Biology 64:1655–1679. Larsson, P. O. 1985. Predation on migrating smolt as a regulating factor in Baltic Ritter, J. A. 1989. Marine migration and natural mortality of North American salmon, Salmo salar L., populations. Journal of Fish Biology 26:391–397. Atlantic salmon Salmo salar L. Canadian Manuscript Report of Fisheries and Levings, C. D., N. A. Hvidsten, and B. Ø. Johnsen. 1994. Feeding of Atlantic Aquatic Sciences 2041. salmon (Salmo salar L.) postsmolts in a fjord in central Norway. Canadian Scott, W. B., and M. G. Scott. 1988. Atlantic fishes of Canada. Canadian Bulletin Journal of Zoology 72:834–839. of Fisheries and Aquatic Sciences 219. Martin, F., R. D. Hedger, J. J. Dodson, L. Fernandes, D. Hatin, F. Caron, and Sheehan, T. F., D. G. Reddin, G. Chaput, and M. D. Renkawitz. In F. G. Whoriskey. 2009. Behavioural transition during the estuarine migration press. SALSEA North America: a pelagic ecosystem survey targeting At- of wild Atlantic salmon (Salmo salar L.) smolt. Ecology of Freshwater Fish lantic salmon in the northwest Atlantic. ICES Journal of Marine Science. 18:406–417. DOI: 10.1093/icesjms/fss052. SWIMMING DEPTH OF ATLANTIC SALMON POSTSMOLTS 1229
Sheehan, T. F., M. D. Renkawitz, and R. W. Brown. 2011. Surface trawl survey Thorstad, E. B., F. Økland, B. Finstad, R. Sivertsgard,˚ P. A. Bjørn, and R. S. for U.S. origin Atlantic salmon Salmo salar. Journal of Fish Biology 79:374– McKinley. 2004. Migration speeds and orientation of Atlantic salmon and 398. sea trout post-smolts in a Norwegian fjord system. Environmental Biology of Shelton, R. G. J., W. R. Turrell, A. Macdonald, I. S. McLaren, and N. T. Nicoll. Fishes 71:305–311. 1997. Records of post-smolt Atlantic salmon, Salmo salar L., in the Faroe- Thorstad, E. B., F. Økland, B. Finstad, R. Sivertsgard,˚ N. Plantalech, P. A. Bjørn, Shetland Channel in June 1996. Fisheries Research 31:159–162. and R. S. McKinley. 2007. Fjord migration and survival of wild and hatchery- SOAFD (Scottish Office Agriculture and Fisheries Department). 1993. Ma- reared Atlantic salmon and wild brown trout post-smolts. Hydrobiologia rine laboratory Aberdeen annual review, 1991–1992. SOAFD, Aberdeen, 582:99–107. Scotland, UK. USGS (U.S. Geological Survey). 2007. USGS real-time water data for Maine. Solomon, D. J. 1982. Smolt migration in Atlantic salmon, Salmo salar L., and USGS, National Water Information System, Reston, Virginia. Available: wa- sea trout, Salmo trutta L. Pages 196–203 in E. L. Brannon and E. O. Salo, terdata.usgs.gov/me/nwis. (December 2007). editors. Proceedings of symposium on salmon and trout migratory behavior. Voegeli, F. A., G. L. Lacroix, and J. M. Anderson. 1998. Development of University of Washington, School of Fisheries, Seattle. miniature pingers for tracking Atlantic salmon smolts at sea. Hydrobiologia Spicer, A. V., J. R. Moring, and J. G. Trial. 1995. Downstream migratory behav- 371–372:35–46. ior of hatchery-reared, radio-tagged Atlantic salmon (Salmo salar) smolts in Xue, H., Y. Xu, D. Brooks, N. R. Pettigrew, and J. Wallinga. 2000. the Penobscot River, Maine, USA. Fisheries Research 23:255–266. Modeling the circulation in Penobscot Bay, Maine. Pages 1112–1127 Stefansson, S. O., B. T. Bjornsson,¨ K. Sundell, G. Nyhammer, and S. D. in M. L. Spaulding and H. L. Butler, editors. Estuarine and coastal McCormick. 2003. Physiological characteristics of wild Atlantic salmon modeling: proceedings of the 6th international conference on estuar- post-smolts during estuarine and coastal migration. Journal of Fish Biology ine and coastal modeling. American Society of Civil Engineers, Reston, 63:942–955. Virginia. Downloaded by [Department Of Fisheries] at 20:01 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:02 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Variation in Acute Thermal Tolerance within and among Hatchery Strains of Brook Trout Jenni L. McDermid a , Friedrich A. Fischer b , Mohammed Al-Shamlih c , William N. Sloan d , Nicholas E. Jones b & Chris C. Wilson b a Wildlife Conservation Society Canada, Trent University, 2140 East Bank Drive, Peterborough, Ontario, K9J 7B8, Canada b Ontario Ministry of Natural Resources, Aquatic Research and Development Section, Trent University, 2140 East Bank Drive, Peterborough, Ontario, K8J 7B8, Canada c Environmental and Life Sciences Graduate Program, Trent University, 2140 East Bank Drive, Peterborough, Ontario, K8J 7B8, Canada d Ontario Ministry of Natural Resources, Codrington Fisheries Research Facility, 15 Fish Hatchery Road, Codrington, Ontario, K0K 1R0, Canada Version of record first published: 30 Jul 2012.
To cite this article: Jenni L. McDermid, Friedrich A. Fischer, Mohammed Al-Shamlih, William N. Sloan, Nicholas E. Jones & Chris C. Wilson (2012): Variation in Acute Thermal Tolerance within and among Hatchery Strains of Brook Trout, Transactions of the American Fisheries Society, 141:5, 1230-1235 To link to this article: http://dx.doi.org/10.1080/00028487.2012.688917
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. Transactions of the American Fisheries Society 141:1230–1235, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.688917
NOTE
Variation in Acute Thermal Tolerance within and among Hatchery Strains of Brook Trout
Jenni L. McDermid* Wildlife Conservation Society Canada, Trent University, 2140 East Bank Drive, Peterborough, Ontario K9J 7B8, Canada Friedrich A. Fischer Ontario Ministry of Natural Resources, Aquatic Research and Development Section, Trent University, 2140 East Bank Drive, Peterborough, Ontario K8J 7B8, Canada Mohammed Al-Shamlih Environmental and Life Sciences Graduate Program, Trent University, 2140 East Bank Drive, Peterborough, Ontario K8J 7B8, Canada William N. Sloan Ontario Ministry of Natural Resources, Codrington Fisheries Research Facility, 15 Fish Hatchery Road, Codrington, Ontario K0K 1R0, Canada Nicholas E. Jones and Chris C. Wilson Ontario Ministry of Natural Resources, Aquatic Research and Development Section, Trent University, 2140 East Bank Drive, Peterborough, Ontario K8J 7B8, Canada
conditions will be better equipped to handle these anticipated Abstract changes. The ability of coldwater species and populations to respond to temperature increases associated with climate change will de- Climate change is a significant concern to ectotherms, as pend on the existing adaptive potential within and among pop- temperature directly influences the physiological processes of ulations. Brook trout Salvelinus fontinalis is a valued coldwater metabolism, scope for activity, and gas exchange (Fry 1947, species that has been widely stocked across its native range as well as extensively introduced in western North America. We in- 1957; van der Have and de Jong 1996; Gunther et al. 2007). vestigated the thermal tolerance of the three primary brook trout Coldwater fish species, particularly inland populations with Downloaded by [Department Of Fisheries] at 20:02 25 September 2012 hatchery strains used in Ontario (Dickson Lake, Lake Nipigon, finite adaptive resources, may be particularly vulnerable to and Hill’s Lake strains) and the thermal tolerance of a brook trout warmer temperatures (McCullough et al. 2009) because the abil- subspecies, Aurora trout S. fontinalis timagamiensis; all strains ity of populations and species to respond to changing ecological were reared in a common hatchery environment. In addition to comparing the strains’ responses to acute thermal stress, we and climatic conditions is limited by their local and regional also examined variability in temperature tolerance among fam- adaptive resources (Stockwell et al. 2003; Willi et al. 2006; ilies within several of these strains. Evidence for significant dif- Latta 2008). Warmwater species, such as smallmouth bass Mi- ferences in temperature tolerance was observed both within and cropterus dolomieu, are predicted to respond more favorably among the strains, with Aurora trout showing the least capacity to climate change than cold-adapted groups, such as salmonids to cope with higher temperatures. The results of this study sug- gest that thermal performance of brook trout populations will (e.g., brook trout Salvelinus fontinalis; Parmesan 2005). Further- be under substantial selective pressure as water temperatures more, the adaptive potential of a species or population requires increase and that strains with existing tolerances for warmer sufficient variability in the trait under selection to enable an
*Corresponding author: [email protected] Received December 4, 2011; accepted April 19, 2012
1230 NOTE 1231
TABLE 1. Location, lake size, and maximum depth of the source water bodies for brook trout and Aurora trout strains in the Ontario hatchery system. The numbers of individual fish and families used for critical thermal maximum tests are shown.
Lake size Maximum Strain Source water body Location (ha) depth (m) Individuals Families Dickson Lake Dickson Lake 45◦48N, 78◦13W 1,000 18.6 287 12 Lake Nipigon Lake Nipigon 49◦50N, 88◦30W 484,800 165 288 13 Hill’s Lake Hill’s Lake Fish Culture Station 44◦57N, 77◦00W NA NA 288 NA Aurora trout Whirligig Lake 47◦22N, 80◦38W 11 9.1 284 12 Whitepine Lake 47◦23N, 80◦38W 77 21.3
adaptive response (Stockwell et al. 2003). In general, northern Lacustrine brook trout populations are ideal candidates for species may have an additional adaptive disadvantage due to investigations of population-specific differences in thermal re- reduced genetic resources stemming from finite ancestral pop- sponse. Within Ontario, three hatchery strains of brook trout ulations in glacial refugia and postcolonization limits on local are used for stocking across the province to provide angling population sizes (Bernatchez and Wilson 1998) as well as histor- opportunities and for introductions and rehabilitative stocking ical bottlenecks (e.g., Aurora trout S. fontinalis timagamiensis). (Table 1; Kerr 2006). These strains originated from different na- Brook trout are native to coldwater streams and lakes in east- tive environments and were bred for different purposes (OMNR ern North America and have been widely introduced in west- 2005; Kerr 2006). The Hill’s Lake strain is considered domes- ern North America and Europe (Scott and Crossman 1973). ticated after an unknown number of generations in captivity at Brook trout actively seek out groundwater upwellings as ther- Pennsylvania hatcheries followed by approximately 30 gener- mal refugia (Biro 1998) and are dependent on coldwater seeps ations at the Ontario Ministry of Natural Resources (OMNR) for multiple life stages and developmental events (Ridgway and Hill’s Lake Fish Culture Station (OMNR 2005). The Lake Nip- Blanchfield 1998; Ridgway 2008). The distribution of brook igon strain was established over 10 generations ago from a wild trout will be limited by coldwater habitat loss or disruption as- collection in Lake Nipigon, one of largest and deepest inland sociated with climate change (Meisner 1990a, 1990b). Thermal lakes in northern Ontario (OMNR 2005). The Dickson Lake habitat loss can result in sublethal effects, including reduced strain was established less than a decade ago (2002–2004) as a feeding, conversion efficiency, and growth (Brett 1971; Selong regionally representative wild strain from Dickson Lake, a mod- et al. 2001; Wehrly et al. 2007), and has even led to mortal- erately large, shallow lake in Algonquin Park (OMNR 2005). ity events in wild brook trout populations (Gunn and Snucins Aurora trout have also been maintained in the provincial hatch- 2010). Several studies have investigated thermal performance of ery system for over half a century as part of an effort to prevent brook trout (McCauley 1958; Benfey et al. 1997; Galbreath et al. their extirpation (Scott and Crossman 1973; Aurora Trout Re- 2004) and Aurora trout (Sale 1962), with evidence that upper covery Team 2006). Aurora trout are considered severely inbred, temperature tolerances in brook trout range from 28◦Cto30◦C as all contemporary fish are descended from nine individuals (Lee and Rinne 1980; Benfey et al. 1997; Selong et al. 2001). that were captured after acidification of their native lakes in Most of these studies were focused on investigating interspecific the 1950s (Balon 1995; Aurora Trout Recovery Team 2006). differences in thermal tolerance among salmonid species rather These last nine wild Aurora trout were rescued from two small, than intraspecific variation. McCauley (1958) examined perfor- shallow south-central Ontario lakes to which they were native
Downloaded by [Department Of Fisheries] at 20:02 25 September 2012 mance differences between two geographically distant strains (Table 1), and their gametes were used to establish a broodstock of brook trout but failed to find differences between the strains. at the Hill’s Lake Fish Culture Station. After liming and lake Recently, population-specific responses in thermal perfor- rehabilitation to mitigate acidification, subsequent generations mance have been observed across geographically proximate of Aurora trout were stocked back into their native lakes and populations of grayling Thymallus thymallus (Haugen and other nearby lakes (Snucins et al. 1995; Aurora Trout Recovery Vøllestad 2000; Kavanagh et al. 2010) and sockeye salmon On- Team 2006), including Alexander Lake. To prevent domestica- corhynchus nerka (Eliason et al. 2011), whereas other studies tion of the Aurora trout broodstock, the hatchery population is have failed to find evidence of intraspecific variation in ther- maintained by using biannual spawn collections of Aurora trout mal performance (brown trout Salmo trutta: Jensen et al. 2000; from Alexander Lake (OMNR 2005). Atlantic salmon Salmo salar: Jonsson et al. 2001; Arctic char In this study, we examined the acute thermal tolerances and Salvelinus alpinus: Larsson et al. 2005). An understanding of responses of Ontario hatchery strains of brook trout and Aurora fishes’ ability to cope with and adapt to a changing climate trout. We assessed the variability in thermal tolerance among has been identified as a significant knowledge gap (Wilson and and within strains to investigate coping and potential adaptive Mandrak 2004; McCullough et al. 2009) and will be essential in- capacity within and among populations of this ecologically and formation for the sustainable management of fishes (Latta 2008). economically significant coldwater species. If thermal tolerance 1232 MCDERMID ET AL.
is linked to local environmental conditions, it is reasonable to removed from the tank and euthanized with tricaine methane- expect that the different origins and histories of these hatchery sulfonate (MS-222; Sigma-Aldrich, St. Louis, Missouri), and strains will result in population-specific differences in thermal the time spent at 26◦C was recorded. performance, although their shared recent thermal history in For each strain, 12–13 families representing approximately a common hatchery environment may homogenize ancestral 285 fish were tested (Table 1). Each trial consisted of eight differences. Furthermore, the potential adaptive capacity within yearlings from each strain that were placed together in the ex- Aurora trout may be hampered by their historical population perimental tank at ambient temperature and were starved for bottleneck and potential inbreeding. 24 h to allow acclimation and to eliminate differences in re- sponses due to metabolic state. In each trial, the eight fish from each strain (except Hill’s Lake) were identified to family. Fish METHODS from individual families were generally included in three differ- Strains and family rearing.—All brook trout strains included ent trials. Families were randomly assigned to trials to minimize in this study (Hill’s Lake, Lake Nipigon, and Dickson Lake confounding of family and trial effects. strains and Aurora trout broodstock) are maintained at the Statistical analyses.—An ANCOVA was used to examine OMNR Hill’s Lake Fish Culture Station but were brought to differences in CTM between strains, with body size as a co- the OMNR Fisheries Research Facility (Codrington, Ontario) variate. In total, 36 trials containing each of the four strains for the temperature experiments. Families of brook trout and were conducted. The time to LE was averaged by stock for Aurora trout were produced during the 2006 fall spawning sea- each trial to prevent pseudoreplication, and these averages were son from strains at the OMNR Fisheries Research Facility. Each used in the ANCOVA. To examine adaptive potential in thermal family was derived from single-pair matings within each strain; tolerance within strains, we explored existing variation among families were held separately except for Hill’s Lake fish, which families by examining the CTM of the different families within were received as pooled juveniles from the Hill’s Lake Fish strains when this information existed. The Hill’s Lake strain was Culture Station and thus could not be identified to family. After excluded from this comparison, since the pooling of individu- hatching, age-0 fish were reared in family units at equivalent als prevented the assessment of performance differences among densities. All fish were reared at ambient surface water temper- families. For the other strains, family-level variability was ex- ature and were fed to satiation several times per day. amined with an ANOVA among families within each separate Experimental design.—Acute temperature tolerance exper- strain. Relationships between strains and between families were iments were conducted with yearling brook trout and Aurora explored post hoc by use of the Tukey–Kramer honestly sig- trout in the winter of 2008. The critical thermal maximum nificant difference comparison of means (at P < 0.05), which (CTM) method involves exposing experimental fish to progres- corrects for multiple comparisons. sively higher water temperatures until a behavioral or physio- All statistical analyses were performed using JMP software logical endpoint, such as loss of equilibrium (LE) or death, is version 4.0 (SAS Institute, Inc., Cary, North Carolina); the data reached (Becker and Genoway 1979). We used a slight modifi- met the individual assumptions (independence and normality) cation of the CTM method, Elliott’s (1981) hybrid CTM chal- of the statistical analyses. lenge, in which the acute temperature tolerance of each fish was defined as the amount of time to LE while in the zone of thermal resistance (Fry 1947). The temperature for the zone of thermal RESULTS resistance was estimated in a pilot study by using a mixed sam- Yearling brook trout and Aurora trout showed significant ple of individuals from all of the strains to be compared. This differences in thermal tolerance among strains (F3, 138 = 35.95,
Downloaded by [Department Of Fisheries] at 20:02 25 September 2012 pilot sample was exposed to a steady increase in temperature P < 0.001). These differences were independent of fish mass ◦ at a rate of 0.15 C per minute. The temperature at which the (F1, 138 = 0.59, P = 0.45); inclusion of body length or Fulton’s first fish in the sample exhibited LE (26◦C) was defined as the condition factor as covariates also was not significant (data not zone of thermal resistance and was then used in all subsequent shown). All populations differed from one another, except the CTM trials. Temperature was steadily increased to the zone of Aurora trout and Lake Nipigon strains, which did not differ sig- thermal resistance and was held static until the final fish experi- nificantly based on the Tukey–Kramer honestly significant dif- enced LE. Use of the time to LE allowed for a greater ability to ference comparison. The Hill’s Lake and Dickson Lake strains differentiate between fish within a trial (Galbreath et al. 2004, (mean time to LE = 127 and 116 min, respectively) were able to 2006). endure elevated temperatures significantly longer than the Lake The tolerance experiments were carried out in a 172-L tank, Nipigon strain (mean time to LE = 87 min), whereas Aurora where the temperature was increased to 26◦C over the course trout (mean time to LE = 76 min) were least able to withstand of 1.5 h. Dissolved oxygen concentrations were monitored and elevated temperatures (Figure 1). maintained above 6 mg/L by aeration. The thermal tolerance Within-strain variability as measured by the CV (SD/mean) time was recorded for each fish as the time spent in the zone was similar among strains (CV = 0.18–0.20) except the of thermal resistance prior to LE. Upon reaching LE, fish were Lake Nipigon strain, which had greater variation (CV = 0.25; NOTE 1233
Interestingly, differences in thermal tolerance persisted despite the generations spent in the hatchery environment; thus, differ- ences in performance are likely attributable to heritable (genetic) differences among strains. The observed differences in temper- ature tolerance reflect historically different thermal regimes and geographic origins. Strains with more southerly origins (Hill’s Lake and Dickson Lake strains) were better able to cope with warmer conditions than brook trout from the more northerly Lake Nipigon strain, suggesting some degree of ancestral local adaptation for temperature tolerance. Levels of within-strain variability also differed among strains. Brook trout from the Dickson Lake and Lake Nipigon strains showed variability among family groups, whereas the inbred Aurora trout had minimal variability among the family groups. The ability of a population to respond to changing selection increases with its inherent variation for the trait FIGURE 1. Box-and-whisker plots of time to the loss of equilibrium at 25◦C being selected upon, as long as the shift in selection does not for brook trout: (a) Dickson Lake strain, (b) Lake Nipigon strain, (c) Hill’s Lake exceed the existing expressed range for that trait (Stockwell strain, and (d) Aurora trout (line within box = mean; ends of box =±SE; ends et al. 2003). As such, wild brook trout from Dickson Lake of whiskers =±SD). Values for individual families within a strain are shown may have a greater ability to adapt to warmer waters than fish to the left of the vertical line in each panel; pooled data for the strain are shown from Lake Nipigon, which in turn may have greater adaptive to the right of the vertical line. The grey dashed horizontal line at 100 min is included for reference only. potential than Aurora trout. Aurora trout not only showed limited within-strain variability but also showed lower overall Figure 1). Despite this greater variation, the Lake Nipigon strain temperature performance than the Dickson Lake and Hill’s still had lower overall temperature tolerance than the other brook Lake strains. Although Aurora trout are considered a subspecies trout strains (mean time to LE = 87 min). Examination of within- of brook trout, they occupy habitats similar to those of brook strain variability for the Dickson Lake, Lake Nipigon, and Au- trout (Aurora Trout Recovery Team 2006), and a previous rora trout strains showed that there was statistically significant study by Sale (1962) concluded that the upper lethal limit for variation among the families tested within each strain (P < 0.05; Aurora trout consistently followed that of brook trout. The Figure 1). The Dickson Lake and Lake Nipigon strains showed reduced tolerance of Aurora trout to elevated temperatures greater within-strain variation than Aurora trout, in which only and the limited within-strain variability likely resulted from two families were responsible for the statistically significant their historical bottleneck and subsequent inbreeding (Balon within-strain differences (Figure 1). Mean CTM values for the 1995). This combined evidence indicates that Aurora trout have Lake Nipigon and Dickson Lake families varied by as much as less potential to adapt to predicted temperature increases and 43 and 36 min to LE, respectively, whereas Aurora trout fami- may experience more severe impacts from the reductions in lies exhibited less than half the variation observed in the brook coldwater habitat that are likely to result from climate change. trout strains (17 min to LE; Figure 1). Family-level variability Adaptation in a species’ thermal tolerance is likely an ex- also differed among strains, as some families exhibited a nar- tremely complex process and cannot be evaluated rigorously row range of temperature tolerances, whereas other families had with a simple test of thermal tolerance as was implemented here. Downloaded by [Department Of Fisheries] at 20:02 25 September 2012 broader ranges (Figure 1). Expressed thermal performance is the phenotypic expression of underlying physiological processes, which are influenced by rearing conditions and acclimation as well as levels of standing DISCUSSION genetic variation relating to expression of the trait (Stockwell This study adds to the growing body of evidence for in- et al. 2003; Willi et al. 2006). Mapping of quantitative trait loci traspecific variation in thermal performance (Fields et al. 1987; in other salmonids have identified loci associated with thermal Haugen and Vøllestad 2000; Myrick and Cech 2004; Kavanagh tolerance (Jackson et al. 1998; Perry et al. 2001; Somorjai et al. et al. 2010; Eliason et al. 2011). In contrast to other studies 2003); as knowledge of the genetic architecture of brook trout that have shown only limited intraspecific variation in thermal improves with mapping homologies (Timusk et al. 2011) and biology for temperate freshwater fishes (Elliott and Klemetsen as high-resolution panels of genetic markers become available, 2002; Larsson et al. 2005; Forseth et al. 2009), our study reveals future studies may be able to tease out the underlying genetic that brook trout and Aurora trout exhibit strain-specific temper- factors and true adaptive potential within and among brook trout ature tolerance. Because all of the study strains were maintained populations. in a common hatchery environment, they may have responded The observed differences in thermal tolerance among the to the rearing temperatures in the hatchery; if so, we would ex- tested strains appeared to be more strongly linked to their pect all strains to perform similarly to the Hill’s Lake strain. ancestry rather than to recent hatchery history, suggesting that 1234 MCDERMID ET AL.
their differing physiological responses to thermal stress are Balon, E. K. 1995. Threatened fishes of the world: Aurora form of the brook heritable and slow to change across generations. To some extent, charr, Salvelinus fontinalis (Mitchill, 1814) (Salmonidae). Environmental fish are able to behaviorally respond to elevated temperatures Biology of Fishes 43:107–108. Becker, C. D., and R. G. Genoway. 1979. Evaluation of the critical thermal by seeking more preferable temperatures (e.g., thermal refuges) maximum for determining thermal tolerance of freshwater fish. Environmen- but only to the extent that these coldwater areas persist (Meisner tal Biology of Fishes 4:245–256. 1990a, 1990b; Biro 1998). Brook trout in Ontario have already Benfey, T. J., L. E. McCabe, and P. Pepin. 1997. Critical thermal maxima of been observed to suffer mortalities linked with warming waters diploid and triploid brook charr, Salvelinus fontinalis. Environmental Biology (Gunn and Snucins 2010). This limited flexibility in thermal of Fishes 49:259–264. Bernatchez, L., and C. C. Wilson. 1998. Comparative phylogeography of Nearc- performance has management implications for stocking in the tic and Palearctic fishes. Molecular Ecology 7:431–452. face of climate change. For example, supplementation stocking, Biro, P. A. 1998. Staying cool: behavioral thermoregulation during summer by especially from nonnative sources, is known to slow adaptation young-of-year brook trout in a lake. Transactions of the American Fisheries to higher temperatures (McCullough et al. 2009). Widespread Society 127:212–222. supplemental stocking is not advocated; however, stocking Brett, J. R. 1971. Energetic responses of salmon to temperature: a study of some thermal relations in the physiology and freshwater ecology of sockeye salmon will continue to be necessary for populations that already (Oncorhynchus nerka). American Zoologist 11:99–113. rely on supplemental stocking for population maintenance. De Stasio, B. T., Jr., D. K. Hill, J. M. Kleinhans, N. P. Nibbelink, and J. J. Fraser (1981, 1989) determined that domesticated fish from the Magnuson. 1996. Potential effects of global climate change on small north- Hill’s Lake strain had limited success in comparison with Lake temperate lakes: physics, fish, and plankton. Limnology and Oceanography Nipigon and Dickson Lake brook trout. Ashford and Danzmann 41:1136–1149. Eliason, E. J., T. D. Clark, M. J. Hague, L. M. Hanson, Z. S. Gallagher, K. M. (2001) suggested that the lower success of the Hill’s Lake strain Jeffries, M. K. Gale, D. A. Patterson, S. G. Hinch, and A. P. Farrell. 2011. as a stocking source was attributable to the strain’s low repro- Differences in thermal tolerance among sockeye salmon populations. Science ductive success. The Lake Nipigon strain is also not a preferred 332:109–112. candidate for supplemental stocking in warmer habitats due to Elliott, J. M. 1981. Some aspects of thermal stress in freshwater teleosts. Pages the lower thermal performance of this strain. Climatic or latitu- 209–245 in A. D. Pickering, editor. Stress and fish. Academic Press, New York. dinal differences between Lake Nipigon and southern stocked Elliott, J. M., and A. Klemetsen. 2002. The upper critical thermal limits for brook trout populations could lead to a decreased ability to alevins of Arctic charr from a Norwegian lake north of the Arctic Circle. respond to the warmer conditions that are predicted by climate Journal of Fish Biology 60:1338–1341. change models (Stefan et al. 1993; De Stasio et al. 1996). Based Fields, R., S. S. Lowe, C. Kaminski, G. S. Whitt, and D. P. Philipp. 1987. on the present study and other studies, the Dickson Lake strain Critical and chronic thermal maxima of northern and Florida largemouth bass and their reciprocal F1 and F2 hybrids. Transactions of the American is likely the best candidate for stream and lake supplementation Fisheries Society 116:856–863. in southern Ontario because of its thermal performance and Forseth, T., S. Larsson, A. J. Jensen, B. Jonsson, I. Naslund,¨ and I. Berglund. recent wild origin. In general, our results suggest that if 2009. Thermal growth performance of juvenile brown trout Salmo trutta: brook trout populations decline as water temperatures increase no support for thermal adaptation hypotheses. Journal of Fish Biology 74: (Meisner 1990a), stocking and rehabilitation plans should 133–149. Fraser, J. M. 1981. Comparative survival and growth of planted wild, incorporate consideration of the thermal performance and hybrid, and domestic strains of brook trout (Salvelinus fontinalis)in adaptive potential of source strains and populations. Ontario lakes. Canadian Journal of Fisheries and Aquatic Sciences 38:1672– 1684. Fraser, J. M. 1989. Establishment of reproducing populations of brook trout after ACKNOWLEDGMENTS stocking of interstrain hybrids in Precambrian Shield lakes. North American Funding and logistic support for this project were provided Journal of Fisheries Management 9:352–363. by the OMNR. Scott Ferguson and Vaughan Jamieson (OMNR) Fry, F. E. J. 1947. Effects of the environment on animal activity. University of Downloaded by [Department Of Fisheries] at 20:02 25 September 2012 Toronto Press, Toronto. provided valuable assistance with rearing and maintaining the Fry, F. E. J. 1957. The aquatic respiration of fish. Pages 1–63 in M. E. Brown, experimental populations and with data collection. Brad Stitt editor. The physiology of fishes. Academic Press, New York. (Trent University) provided constructive feedback on an earlier Galbreath, P. F., N. D. Adams, and T. H. Martin. 2004. Influence of heating version of the manuscript; comments from John Plumb and an rate on measurement of time to thermal maximum in trout. Aquaculture anonymous reviewer additionally improved the paper. This work 241:587–599. Galbreath, P.F., N. D. Adams, L. W. Sherrill III, and T. H. Martin. 2006. Thermal was carried out in accordance with the Animals for Research tolerance of diploid versus triploid rainbow trout and brook trout assessed Act under OMNR Animal Care Protocol 2008-65. by time to chronic lethal maximum. Environmental Biology of Fishes 75: 183–193. Gunn, J., and E. Snucins. 2010. Brook charr mortalities during extreme temper- REFERENCES ature events in Sutton River, Hudson Bay lowlands, Canada. Hydrobiologia Ashford, B. D., and R. G. Danzmann. 2001. An examination of the post-planting 650:79–84. success of hatchery brook charr possessing different mitochondrial DNA Gunther, S. J., R. D. Moccia, and D. P. Bureau. 2007. Patterns of growth haplotypes. Journal of Fish Biology 59:544–554. and nutrient deposition in lake trout (Salvelinus namaycush), brook trout Aurora Trout Recovery Team. 2006. Recovery strategy for the Aurora trout (Salvelinus fontinalis) and their hybrid, F1 splake (Salvelinus namaycush × (Salvelinus fontinalis timagamiensis) in Canada. Fisheries and Oceans Salvelinus fontinalis) as a function of water temperature. Aquaculture Nutri- Canada, Species at Risk Act Recovery Strategy Series, Ottawa. tion 13:230–239. NOTE 1235
Haugen, T. O., and L. A. Vøllestad. 2000. Population differences in early life- Perry, G. M. L., R. G. Danzmann, M. M. Ferguson, and J. P. Gibson. 2001. history traits in grayling. Journal of Evolutionary Biology 13:897–905. Quantitative trait loci for upper thermal tolerance in outbred strains of rainbow Jackson, T. R., M. M. Ferguson, R. G. Danzmann, A. G. Fishback, P. E. Ihssen, trout (Oncorhynchus mykiss). Heredity 86:333–341. M. O’Connell, and T. J. Crease. 1998. Identification of two QTL influencing Ridgway, M. S. 2008. A roadmap for coasters: landscapes, life histories, and the upper temperature tolerance in three rainbow trout (Oncorhynchus mykiss) conservation of brook trout. Transactions of the American Fisheries Society half-sib families. Heredity 80:143–151. 137:1179–1191. Jensen, A. J., T. Forseth, and B. O. Johnsen. 2000. Latitudinal variation in growth Ridgway, M. S., and P. J. Blanchfield. 1998. Brook trout spawning areas in of young brown trout Salmo trutta. Journal of Animal Ecology 69:1010–1020. lakes. Ecology of Freshwater Fish 7:140–145. Jonsson, B., T. Forseth, A. J. Jensen, and T. F. Næsje. 2001. Thermal performance Sale, P. F. 1962. A note on the lethal temperature of the Aurora trout, Salvelinus of juvenile Atlantic salmon, Salmo salar L. Functional Ecology 15:701–711. timagamiensis. Canadian Journal of Zoology 40:367–369. Kavanagh, K. D., T. O. Haugen, F. Gregersen, J. Jernvall, and L. A. Vøllestad. Scott, W. B., and E. J. Crossman. 1973. Freshwater fishes of Canada. Fisheries 2010. Contemporary temperature-driven divergence in a Nordic freshwater Research Board of Canada, Ottawa. fish under conditions commonly thought to hinder adaptation. BMC Evolu- Selong, J. H., T. E. McMahon, A. V. Zale, and F. T. Barrows. 2001. Effect tionary Biologyy [online serial] 10:article 350. of temperature on growth and survival of bull trout, with application of an Kerr, S. J. 2006. An historical review of fish culture, stocking and fish transfers improved method for determining thermal tolerance in fishes. Transactions in Ontario, 1865–2004. Ontario Ministry of Natural Resources, Fish and of the American Fisheries Society 130:1026–1037. Wildlife Branch, Peterborough. Snucins, E. J., J. M. Gunn, and W. Keller. 1995. Restoration of the Aurora trout Larsson, S., T. Forseth, I. Berglund, A. J. Jensen, I. Naslund,¨ J. M. Elliott, and to its acid-damaged native habitat. Conservation Biology 9:1307–1311. B. Jonsson. 2005. Thermal adaptation of Arctic charr: experimental studies Somorjai, I. M., R. G. Danzmann, and M. M. Ferguson. 2003. Distribution of growth in eleven charr populations from Sweden, Norway and Britain. of temperature tolerance quantitative trait loci in Arctic charr (Salvelinus Freshwater Biology 50:353–368. alpinus) and inferred homologies in rainbow trout (Oncorhynchus mykiss). Latta, R. G. 2008. Conservation genetics as applied evolution: from genetic Genetics 165:1443–1456. pattern to evolutionary process. Evolutionary Applications 1:84–94. Stefan, H. G., M. Hondzo, and X. Fang. 1993. Lake water quality modeling Lee, R. M., and J. N. Rinne. 1980. Critical thermal maxima of five trout species for projected future climate scenarios. Journal of Environmental Quality 22: in the southwestern United States. Transactions of the American Fisheries 417–431. Society 109:632–635. Stockwell, C. A., A. P. Hendry, and M. T. Kinnison. 2003. Contemporary McCauley, R. W. 1958. Thermal relations of geographic races of Salvelinus. evolution meets conservation biology. Trends in Ecology and Evolution 18: Canadian Journal of Zoology 36:655–662. 94–101. McCullough, D. A., J. M. Bartholow, H. I. Jager, R. L. Beschta, E. F. Cheslak, Timusk, E. R., M. M. Ferguson, H. K. Moghadam, J. D. Norman, C. C. Wilson, M. L. Deas, J. L. Ebersole, J. S. Foott, S. L. Johnson, K. R. Marine, M. G. and R. G. Danzmann. 2011. Genome evolution in the fish family Salmonidae: Mesa, J. H. Petersen, Y. Souchon, K. F. Tiffan, and W. A. Wurtsbaugh. 2009. generation of a brook charr genetic map and comparisons among charrs Research in thermal biology: burning questions for coldwater stream fishes. (Arctic charr and brook charr) with rainbow trout. BMC Genetics [online Reviews in Fisheries Science 17:90–115. serial] 12:article 68. Meisner, J. D. 1990a. Effect of climatic warming on the southern margins of van der Have, T. M., and G. de Jong. 1996. Adult size in ectotherms: temper- the native range of brook trout, Salvelinus fontinalis. Canadian Journal of ature effects on growth and differentiation. Journal of Theoretical Biology Fisheries and Aquatic Sciences 47:1065–1070. 183:329–340. Meisner, J. D. 1990b. Potential loss of thermal habitat for brook trout, due Wehrly, K. E., L. Wang, and M. Mitro. 2007. Field-based estimates of to climatic warming, in two southern Ontario streams. Transactions of the thermal tolerance limits for trout: incorporating exposure time and tem- American Fisheries Society 119:282–291. perature fluctuation. Transactions of the American Fisheries Society 136: Myrick, C. A., and J. J. Cech Jr. 2004. Temperature effects on juvenile anadro- 365–374. mous salmonids in California’s Central Valley: what don’t we know? Reviews Willi, Y., J. Van Buskirk, and A. A. Hoffmann. 2006. Limits to the adaptive in Fish Biology and Fisheries 14:113–123. potential of small populations. Annual Review of Ecology, Evolution, and OMNR (Ontario Ministry of Natural Resources). 2005. Fish culture stocks Systematics 37:433–458. catalogue. OMNR, Peterborough. Wilson, C. C., and N. Mandrak. 2004. History and evolution of lake trout in Parmesan, C. 2005. Biotic response: range and abundance changes. Pages 41–55 Shield lakes: past and future challenges. Pages 21–35 in J. Gunn, R. Steedman, in T. E. Lovejoy and L. Hannah, editors. Climate change and biodiversity. and R. Ryder, editors. Boreal Shield watersheds: lake trout ecosystems in a Yale University Press, New Haven, Connecticut. changing environment. Lewis/CRC Press, Boca Raton, Florida. Downloaded by [Department Of Fisheries] at 20:02 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:03 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
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Use of Radiotelemetry and Direct Observations to Evaluate Sea Lion Predation on Adult Pacific Salmonids at Bonneville Dam Matthew L. Keefer a , Robert J. Stansell b , Sean C. Tackley c , William T. Nagy b , Karrie M. Gibbons b , Christopher A. Peery d & Christopher C. Caudill a a Department of Fish and Wildlife Sciences, University of Idaho, 975 6th Street, Moscow, Idaho, 83844-1136, USA b U.S. Army Corps of Engineers, Fisheries Field Unit, Post Office Box 150, Cascade Locks, Oregon, 97014, USA c U.S. Army Corps of Engineers, Portland District, Environmental Resources Branch, Post Office Box 2946, Portland, Oregon, 97208, USA d U.S. Fish and Wildlife Service, Idaho Fishery Resource Office, 276 Dworshak Complex Drive, Ahsahka, Idaho, 83520, USA Version of record first published: 30 Jul 2012.
To cite this article: Matthew L. Keefer, Robert J. Stansell, Sean C. Tackley, William T. Nagy, Karrie M. Gibbons, Christopher A. Peery & Christopher C. Caudill (2012): Use of Radiotelemetry and Direct Observations to Evaluate Sea Lion Predation on Adult Pacific Salmonids at Bonneville Dam, Transactions of the American Fisheries Society, 141:5, 1236-1251 To link to this article: http://dx.doi.org/10.1080/00028487.2012.688918
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. Transactions of the American Fisheries Society 141:1236–1251, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.688918
ARTICLE
Use of Radiotelemetry and Direct Observations to Evaluate Sea Lion Predation on Adult Pacific Salmonids at Bonneville Dam
Matthew L. Keefer* Department of Fish and Wildlife Sciences, University of Idaho, 975 6th Street, Moscow, Idaho 83844-1136, USA Robert J. Stansell U.S. Army Corps of Engineers, Fisheries Field Unit, Post Office Box 150, Cascade Locks, Oregon 97014, USA Sean C. Tackley U.S. Army Corps of Engineers, Portland District, Environmental Resources Branch, Post Office Box 2946, Portland, Oregon 97208, USA William T. Nagy and Karrie M. Gibbons U.S. Army Corps of Engineers, Fisheries Field Unit, Post Office Box 150, Cascade Locks, Oregon 97014, USA Christopher A. Peery U.S. Fish and Wildlife Service, Idaho Fishery Resource Office, 276 Dworshak Complex Drive, Ahsahka, Idaho 83520, USA Christopher C. Caudill Department of Fish and Wildlife Sciences, University of Idaho, 975 6th Street, Moscow, Idaho 83844-1136, USA
Abstract
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 Management of protected species becomes increasingly complex when one protected population negatively affects another. This occurs along coastlines and in rivers and estuaries of the U.S. Pacific Northwest, where protected marine mammals prey on threatened and endangered Pacific salmonids Oncorhynchus spp. Over 9 years, we observed a growing aggregation of California sea lions Zalophus californianus and Steller sea lions Eumetopias jubatus preying upon adult Chinook salmon O. tshawytscha and steelhead O. mykiss at Bonneville Dam on the Columbia River. Both before and concurrent with the observation study, we monitored radio-tagged salmon at Bonneville Dam and during their upriver spawning migrations. Springtime sea lion abundance steadily increased from 2002 to 2010 and the aggregation formed earlier each winter. The principal prey species in winter were resident white sturgeon Acipenser transmontanus and migratory steelhead and then shifted to predominantly Chinook salmon when the spring run arrived. Observation-based estimates of salmonid consumption from January to May varied 12-fold among years (0.4–4.9%, mean = 2.6% of adult salmonids counted at the dam), and radiotelemetry results corroborated these estimates. The highest proportional impact was in winter and early spring. As salmonid abundance increased, per capita consumption by sea lions increased (Type II functional response) but individual salmonid risk decreased (due
*Corresponding author: [email protected] Received December 19, 2011; accepted April 17, 2012
1236 SEA LION PREDATION AT A HYDROELECTRIC DAM 1237
to prey swamping). Population-specific risk analyses indicated predation was substantially higher for early-timed than for late-timed salmon populations. The most at-risk group included Snake River and upper Columbia River Chinook salmon listed as threatened under the U.S. Endangered Species Act. These predation indices should help managers simultaneously tasked with salmon recovery and marine mammal management.
Management of threatened or endangered populations is annual aggregation and small numbers of Pacific harbor seals challenging in light of the interactions between wildlife needs, have visited seasonally for several decades. The pinnipeds pri- human demands on natural resources, and legal obligations as- marily feed on Chinook salmon O. tshawytscha, steelhead O. sociated with species protection. Management complexity is mykiss, white sturgeon Acipenser transmontanus, and Pacific further increased when one protected population or species neg- lamprey Entosphenus tridentatus at the dam. Substantial por- atively impacts other at-risk populations. This is the case along tions of the Columbia River salmon and steelhead runs are the west coast of North America, where harbor seals Phoca ESA-listed (NMFS 2011a), whereas both white sturgeon and vitulina and sea lions (Otariidae), which are protected by the Pacific lamprey populations are of regional conservation con- 1972 U.S. Marine Mammal Protection Act (MMPA), prey on cern (Rieman and Beamesderfer 1990; Clemens et al. 2010). In anadromous Pacific salmonids Oncorhynchus spp., which are response to perceived negative predation impacts on fish con- protected by the 1973 U.S. Endangered Species Act (ESA). centrated below Bonneville Dam, management agencies have Legal protection afforded pinnipeds by the MMPA, includ- conducted a variety of nonlethal sea lion hazing activities (i.e., ing cessation of bounty and predator control programs, is largely boat chasing, cracker shells, rubber bullets, underwater percus- credited with population recovery and recent range expansion sive devices) and in 2005 built removable sea lion exclusion from California to southeast Alaska. The eastern population of structures at fishway entrances to prevent sea lions from enter- threatened Steller sea lions Eumetopias jubatus, for example, ing fish ladders. After extensive legal and scientific review by has increased at an annual average rate of more than 3% since the Pinniped–Fishery Interaction Task Force, targeted Califor- the 1970s and their range has expanded northward (Pitcher et al. nia sea lion trapping began near the dam in 2007 and lethal 2007). Similarly, counts of juvenile California sea lions Zalo- removal began in 2008 (NMFS 2011b). Trapped animals have phus californianus and adult harbor seals indicate rapid popula- been branded, transferred to holding facilities (including zoos), tion growth rates starting in the mid-1970s (Jeffries et al. 2003; or euthanized based on guidelines set by Section 120 of the Brown et al. 2005; Lowry and Forney 2005). Some regional MMPA and interpreted by the Task Force. pinniped populations have probably reached carrying capacity Two multiyear research programs at Bonneville Dam provide (Carretta et al. 2007). In contrast, about half of the 52 evolution- a unique opportunity to gain insight into pinniped behaviors and arily significant units (ESUs) of U.S. west coast salmonids are predation impacts on anadromous salmonids at the site. The first currently ESA-listed and have substantially reduced abundance was a pinniped observation study (2002–2010) that enumerated relative to historic levels (Good et al. 2005). Few of the listed sea lions and seals near the dam and documented their surface- salmonid populations are considered to be recovered. feeding predation events (e.g., Tackley et al. 2008; Stansell Differences in recovery trajectories for pinnipeds and et al. 2010). The second was a Chinook salmon radiotelemetry salmonids have led to increasing numbers of local and regional study (1996–2010) that monitored spring- and summer-run Chi- management conflicts (e.g., Jeffries and Scordino 1997; Baraff nook salmon passage through the tailrace and fishways at Bon- and Loughlin 2000; Weise and Harvey 2005; Wright et al. 2007). neville Dam and on their subsequent upstream homing migration Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 One of the most visible and contentious conflicts involves sea (e.g., Keefer et al. 2004a; Caudill et al. 2007). This paper inte- lion predation on ESA-protected salmonids in the Columbia grates results from the two studies. Specific objectives were to River at the base of Bonneville Dam. Although some harbor (1) summarize sea lion and seal abundance and aggregation tim- seals were present in the lower ∼300 km of the Columbia River ing, (2) estimate total pinniped predation rates on salmonids at historically (Lyman et al. 2002), there are no reports of signifi- several temporal scales using salmonid count data as an index cant sea lion aggregations near Bonneville Dam (river kilometer of availability, and (3) evaluate whether predation risks dif- 235) from the six decades after dam completion in 1938. That fer among Chinook salmon populations. We hypothesized that changed in the early 2000s, when male California sea lions be- Chinook salmon risk would vary as a function of heritable gan congregating at the base of the dam to feed on dense concen- among-population differences in migration timing (e.g., Quinn trations of migratory and resident fishes (Stansell et al. 2010). et al. 2000; Keefer et al. 2004b; Waples et al. 2004). More Since 2002, California sea lion abundance in the Bonneville specifically, we expected that salmon populations with rela- Dam tailrace has steadily increased and they have arrived at tively early migration timing would be most at risk because the the dam progressively earlier in the year (see Results). In ad- relative predator density (number of predators / number of prey) dition, increasing numbers of Steller sea lions have joined the has been highest early in the spring. 1238 KEEFER ET AL.
METHODS each area to ensure that animals and predation events were not Pinniped enumeration and surface feeding observation.— counted twice. In addition, observers rotated among locations Seals and sea lions typically bring large prey, such as adult and shifts to minimize bias among observers, and experienced salmonids, to the surface to facilitate handling and consumption. observers worked to ensure correct predator and prey identi- Surface observation is therefore a useful tool for assessing pin- fication during daily quality-control visits. Daily observations niped diet composition and minimum large prey consumption were also made at favored sea lion haul-out sites in the tail- rates in near-shore environments (Brown and Mate 1983; Roffe race study area to identify individuals and refine abundance and Mate 1984). In this study, two observers at each of the three estimates. major Bonneville Dam tailrace areas (Figure 1) used binocu- Prey categories included Chinook salmon, steelhead, uniden- lars to observe pinnipeds and identify individual California and tified salmonid (2002–2007), Pacific lamprey, white sturgeon, Steller sea lions when possible (from brands, tags, or other dis- and unidentified fish. Daily estimates of the numbers of tinguishing marks). Hourly and daily information recorded by salmonids consumed (i.e., “catch”) were made by combining observers included observation area, estimated number of pin- the observed catch at each of the three tailrace sites. These nipeds present by species, individual pinnipeds identified, and values were then expanded for daytime hours when observers prey items identified to species or genus. Regular observations were not present using linear interpolation for each species generally began at the hour of sunrise and ended at the hour and site. For days without observations, catch was interpo- of sunset. Randomly assigned 1-h breaks occurred in the morn- lated using data from the day before and the day after the ing and afternoon each observation day in 2009–2010; breaks missed observation day or days; no estimates were made for were haphazard in earlier years. Observers used two-way radios dates before the onset of observation within year. The ex- to confirm the presence or absence of individual pinnipeds in panded estimates were considered minimum measures for the Downloaded by [Department Of Fisheries] at 20:03 25 September 2012
FIGURE 1. Columbia River pinniped study area downstream from Bonneville Dam showing the tailrace areas where pinnipeds were observed, radio-tagged Chinook salmon release sites (black stars), and aerial radio antenna sites (white stars). Inset map shows the U.S. portion of the Columbia and Snake river basins, including main-stem dam locations and approximate locations of the 32 lower Columbia River (gray circles), upper Columbia River (black circles), andSnake River (white circles) spring–summer Chinook salmon populations discussed in the text. SEA LION PREDATION AT A HYDROELECTRIC DAM 1239
observation area. Additional “adjusted estimates” included 2004. Complete migration histories of radio-tagged spring and unidentified fish assigned to likely species plus estimated night- summer Chinook salmon in those years were used to assign time catch based on night observations (described in Stansell fish to individual populations and to calculate population- et al. 2009). specific migration timing distributions at Bonneville Dam. Observation effort varied among years based on staff avail- Distributions identified the relative differences in timing and ability and sea lion presence. Regular daily observations be- abundance among populations (see Keefer et al. 2004b), but gan in March (2002–2003, 2005), February (2004), or January radio-tagged salmon were not a random sampling of all popula- (2006–2010) and continued until mid to late May each year. tions in all years (i.e., some fish were selected because they were Progressively earlier start dates were a response to earlier sea of known origin). We chose therefore to focus on population- lion arrival dates in the Bonneville tailrace. End dates were as- specific migration timing distributions rather than abundance to sociated with zero or near-zero pinniped presence. Observations estimate differences in relative risk among populations. For this occurred on nearly every day in 2003 and 2006–2008, whereas purpose, sampling within population was relatively unbiased there were gaps on some weekends in 2002, 2004, 2005, 2009, because sampling of known-origin fish was not biased early or and 2010. late in the migrations. The radiotelemetry study years included Chinook salmon tagging and monitoring.—Adult spring- and early-, late- and average-timed runs and resulting timing distri- summer-run Chinook salmon were collected and radio-tagged butions probably captured much of the interannual variability at the adult fish facility at Bonneville Dam over 14 years (1996– for each population. 1998, 2000–2004, 2006–2007, and 2009–2010). Descriptions Data analyses.—Direct measurements of daily pinniped pre- of fish capture, anesthetization, intragastric tagging methods, dation rate (i.e., the proportion consumed) were not possible handling and release protocols, and transmitter specifications because the number of prey available in the Bonneville tail- are provided in Keefer et al. (2004a, 2005). Briefly, the trap race on any given day was never known. Instead, counts of facility was located adjacent to the north-shore fish ladder. A salmonids passing Bonneville Dam fish ladders were used as picket lead diverted upstream migrants from the ladder to the a relative index of salmonid abundance and prey availabil- trap where a flume system diverted fish selected for tagging into ity. Upstream-migrant salmonids were counted and identified an anesthetic tank. Radio-tagged fish recovered in a 2,275-L tank to species 16 h per day, every day, by subcontractors to the filled with oxygenated river water and were then transported to U.S. Army Corps of Engineers in all years (data archived at: release sites either ∼10 km downstream from the dam or in the www.nwp.usace.army.mil/environment/fishdata). The only ex- Bonneville Dam forebay (Figure 1). ception was that counting did not begin until March in 2002. Chinook salmon were tagged in approximate proportion to Use of count data as an index for salmonid abundance was daily counts at Bonneville Dam in 1996–1998 and 2000–2003 most realistic in the spring when Chinook salmon and steelhead (Keefer et al. 2004b). Starting in 2004, sample sizes were smaller were actively migrating upstream. The counts were less repre- and tagging was less proportional to abundance. Instead, salmon sentative of salmonid abundance in the tailrace in January and were tagged throughout the spring and at varying rates in sum- February. In these months, steelhead were the primary salmonid mer. Radiotelemetry study objectives differed among years and counted at the dam, but many additional steelhead were believed consequently the demographics of annual samples differed. In to be holding in the tailrace (i.e., overwintering, Keefer et al. 1996–1998, tagged salmon were a near-random sample of the 2008a). runs passing the dam. In subsequent years, a portion of each We estimated an index of pinniped predation rate on sample was selected based on the presence of passive inte- salmonids at weekly, seasonal (winter, spring), and annual grated transponder (PIT) tags that identified natal basin. These scales. Using the pinniped observation data set, weekly preda-
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 “known-origin” salmon were used to address objectives related tion estimates were calculated by dividing the estimated num- to specific populations or management activities. ber of salmonids caught each week (Catchweek) by the number Radio-tagged salmon were monitored with an extensive counted passing the dam in the same week (Countweek). This array of fixed-location aerial and underwater antennas and basic estimate required no assumptions about whether captured with mobile-tracking units. In all years, fish that entered the fish would have passed the dam in the same statistical week, al- Bonneville Dam tailrace were detected using aerial antennas though median tailrace and dam passage times for radio-tagged on both sides of the river (Figure 1). Bonneville Dam fish- Chinook salmon were 0.9–1.4 d (Keefer et al. 2004a). Sea- way entrances, ladders, and exits were monitored with un- sonal (winter = weeks 1–11, spring = weeks 12–22) and an- derwater antennas. Detection efficiency of the combined Bon- nual (1 January to 31 May) predation estimates treated captured neville array was near 100% for tagged salmon that entered fish as part of the salmonid count total [i.e., predation rate = fishways or passed the dam in all years. Tailraces and fish- CatchTotal / (CountTotal + CatchTotal)] based on the assump- ways at 7–8 upstream Columbia and Snake River dams were tion that almost all salmonids captured in the tailrace would monitored, and aerial antennas were deployed in most ma- have passed upstream eventually in the absence of predators jor tributaries (details in Keefer et al. 2004a, 2004b). Up- (Keefer et al. 2005). Uncertainty associated with overwintering stream monitoring effort was highest in 1996–1998 and 2000– steelhead abundance and behavior resulted in relatively lower 1240 KEEFER ET AL.
confidence in winter versus spring predation rate estimates. abundance (Countweek) estimates. Preliminary data examination More specifically, the basic method may have resulted in over- following the guidelines in Holling (1966) and Real (1977) indi- estimation of predation in winter when steelhead migration ac- cated a Type II response, where per capita consumption rapidly tivity was reduced. increased as more salmonids became available and then be- It was not possible to directly estimate predation rates for came asymptotic at higher salmonid densities. We modeled this radio-tagged spring Chinook salmon. Instead, we calculated the response using a hyperbolic Michaelis–Menten function (Real proportions of tagged salmon that failed to pass Bonneville 1977). This nonlinear equation takes the form of y = ax/(b + Dam, with the assumption that a portion of failed passage was x), where y = per capita consumption and x = salmonid abun- due to predation. Weekly failure estimates were calculated by dance. We fit the model using a least-squares, Gauss–Newton dividing the number of tagged salmon that failed to pass the dam process where parameters a and b were estimated through an (Failweek) by the number that was detected in the Bonneville tail- iterative convergence function (PROC NLIN; SAS 2000). race (Tailraceweek). Weeks were defined by salmon release dates The multiyear, population-specific migration timing data for from mid-March to 31 May. Two annual estimates were calcu- Chinook salmon were used to estimate the relative risk of pin- lated: FailTotal / TailraceTotal and FailTotal / FishwayTotal, where niped predation. For each population separately, the proportion the denominators were the total numbers of salmon detected that passed Bonneville Dam each week was multiplied by the in the tailrace and at fishways, respectively, and the numera- associated mean weekly predation rate estimates from the ob- tors were the numbers within those categories that did not pass servation data (i.e., mean Catchweek / Countweek). The products the dam. The tailrace antennas were downstream from the pin- were then summed for each population to generate a relative niped observation area while fishway antennas were at the up- risk index. stream margin of the observation area. We therefore anticipated that radiotelemetry-based predation estimates would potentially RESULTS be higher (tailrace) and lower (fishway) than the observation- based estimates. In all estimates, radio-tagged salmon reported Pinniped Observations harvested by anglers or recorded in downstream hatcheries In nine pinniped observation years, the Bonneville tailrace or tributaries were treated as successful (i.e., they evaded was surveyed for a total of 23,954 h (mean = 2,662 h/year; predation). Table 1). The minimum number of unique animals observed per Radiotelemetry-based predation estimates differed in impor- year ranged from 31 in 2002 to 166 in 2010 (mean = 97). There tant ways from the observation-based estimates. First, the cause were more identified California sea lions (annual minimum = of dam passage failure for radio-tagged salmon was unknown in 30–104) than Steller sea lions (0–75) or harbor seals (0–3) in all almost all cases. Attributing mortality from other sources to pin- years. nipeds would clearly inflate predation estimates. We addressed Several measures of pinniped presence at Bonneville Dam this by calculating some baseline passage failure estimates us- indicated rapid growth of the aggregation through the study ing radio-tagged spring Chinook salmon from years prior to the period and changing species composition, especially in recent pinniped aggregation years (1996–1998, 2000–2001). Second, years. First, mean daily minimum counts of California sea li- predation on radio-tagged fish could have occurred day or night ons steadily increased from 2002 to 2006 and then decreased or downstream from the observation areas. These effects would from 2008 to 2010 coincident with the removal of 40 repeat- tend to produce higher estimates than those based on daytime visit animals (Figure 2a). Mean daily minimum counts of Steller surface observation in the tailrace study area (note that a small sea lions showed low overall presence from 2000 to 2005, and adjustment was made for night-time predation based on lim- then a steady increase to approximately one-half of the iden-
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 ited observation effort). The combined radiotelemetry (probably tified pinniped population by 2010. Second, annual maximum maximum) and observation (probably minimum) results there- daily counts of California and Steller sea lions ranged from 14 fore represent a potential range of seasonal predation impacts in to 52 and 0 to 53, respectively (Figure 2b). Year-to-year pat- the Bonneville Dam tailrace area. terns in maximum counts were similar to those for mean daily Relationships between observed predation and salmonid minimums. Third, the number of pinniped days (sum of daily abundance were evaluated in two ways. To test for effects of observed totals) increased linearly from 318 in 2002 to 3,252 in salmonid abundance on the number of salmonids consumed 2010 (r = 0.97, P < 0.001; Figure 2c). (i.e., density-dependent and prey-swamping effects), we used Pinnipeds were observed progressively earlier in the year a log-log regression model where the dependent variable was through time (Figure 3). Within year, Steller sea lions arrived Catchweek and the independent variable was Countweek.Data in larger numbers earlier (on average) than California sea lions, were log10 transformed to improve normality and more easily although many individually identifiable animals of both species display results because both terms varied seasonally by orders of occasionally moved in and out of the observation area. Seasonal magnitude. To test for a functional sea lion predation response timing differences among sea lion species coincided in part with (Holling 1959), we used weekly per capita consumption esti- availability of their primary prey species (i.e., resident white mates (Catchweek / mean pinnipeds observedweek) and salmonid sturgeon for Steller sea lions and migratory Chinook salmon for SEA LION PREDATION AT A HYDROELECTRIC DAM 1241
TABLE 1. Annual numbers of days and total hours of pinniped observation in the Bonneville Dam tailrace and minimum numbers of individual sea lions and seals identified from marks or brands in 2002–2010. Note: a total of 40 California sea lions were removed from the aggregation in 2008–2010.
Minimum number of unique animals Year Days observed Total hours California sea lions Stellar sea lions Harbor seals Total pinnipeds 2002 44 662 30 0 1 31 2003 84 1,356 104 3 2 109 2004 72 553 99 3 2 104 2005 54 1,108 81 4 1 86 2006 100 3,647 72 11 3 87 2007 121 4,433 71 9 2 82 2008 131 5,131 82 39 2 123 2009 103 3,455 54 26 2 82 2010 101 3,609 89 75 2 166
California sea lions). Peak pinniped abundance was in April or Salmonid Counts and Radio-Tagging early May in all years, and typically preceded peak Chinook In the pinniped observation years (2002–2010), total Jan- salmon passage at Bonneville Dam by several days. Almost all uary to May salmonid counts at Bonneville Dam ranged from sea lions exited the study area by late May presumably to begin 81,274 to 284,733 fish (mean = 162,780; Table 2; Figure 2d). migrations to breeding grounds (e.g., Wright et al. 2010). Counts were dominated by adult (mean = 86%) and jack (pre- cocious male) spring Chinook salmon (mean = 10%). The steelhead contribution ranged from 3 to 6% (mean = 4%) and <0.2% of salmonids counted were sockeye salmon (O. nerka). Some Chinook salmon were counted each month, but the an- nual spring run typically started in March or early April and peak daily counts were in the tens of thousands (Figure 3). Dates when the first 10% of the Chinook salmon had passed ranged from 1 April to 2 May (mean = 19 April) and me- dian passage dates ranged from 21 April to 11 May (mean = 2 May). In contrast, steelhead were counted passing the dam on almost all days and were the only salmonid counted in many weeks in January and February. Daily steelhead counts at the dam were typically in the tens to hundreds (January–May maximums = 100–248 fish). Median steelhead passage dates in the January–May period ranged from 21 March to 3 May (mean = 13 April). In 2002–2010, 3,285 radio-tagged spring Chinook salmon
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 were released downstream from Bonneville Dam and were used for predation evaluations. Of these, 3,193 (97%) moved upstream into the tailrace study area and 3,051 (93%) ap- proached or entered a fishway. In the study years prior to high pinniped abundance (1996–1998, 2000–2001), 3,376 spring Chinook salmon were recorded in the Bonneville tailrace and 3,354 (99%) approached fishways. FIGURE 2. Annual estimates of California sea lions (open squares), Steller Population-specific migration timing distributions for sea lions (open triangles), and all pinnipeds (filled circles) counted at Bon- Chinook salmon were calculated using migration histories from neville Dam, 2002–2010. Panels show: (a) mean daily minimum counts, (b) spring–summer Chinook salmon released in 1996–1998 and daily maximum counts, and (c) total pinniped days in thousands (i.e., the sum of 2000–2004 (Table 2). A total of 6,384 salmon were released daily pinniped counts). Panel (d) shows total January–May counts of salmonids (1,000s) at Bonneville Dam (gray = adult and jack Chinook salmon, black = (Table 2) and 5,229 (83%) returned to spawning tributaries steelhead and sockeye salmon). Stars (panels a and b) denote years when or hatcheries or could otherwise be assigned to upriver pop- California sea lions were removed from the aggregation. ulations. Migration timing distributions for 32 populations 1242 KEEFER ET AL.
TABLE 2. Numbers of adult and jack Chinook salmon, adult steelhead, and adult sockeye salmon counted passing fishways at Bonneville Dam plus the numbers of adult spring (March–May) and summer (June–July) Chinook salmon collected and radio-tagged, 1996–2010. Radiotelemetry data from 1996 to 1998 and 2000–2004 were used to estimate population-specific migration timing distributions. Radiotelemetry data from 2002 to 2004, 2006–2007, and 2009–2010 were used to estimate pinniped predation rates downstream from Bonneville Dam.
Number of radio-tagged Number of salmonids counted at Bonneville Dama adult Chinook salmon Chinook salmon Steelhead Sockeye salmon Total Year Adult Jack Adult Adult salmonids Spring Summer 1996 51,265 4,636 4,996 1 60,898 702 150 1997 114,071 963 5,279 11 120,324 680 334 1998 38,342 775 2,933 2 42,052 678 279 1999 38,574 8,692 2,442 0 49,708 2000 178,336 21,259 4,252 140 203,897 801 331 2001 391,842 14,172 7,274 32 413,288 896 316 2002 269,520 6,477 8,734 2 284,733 913b 304b 2003 195,770 14,258 7,904 2 217,934 806 384 2004 170,308 8,885 7,443 134 186,770 349 199 2005 74,053 4,288 2,886 47 81,274 2006 96,458 2,908 5,688 9 105,063 351 7 2007 66,646 16,606 5,188 36 88,476 300 7 2008 125,587 17,552 4,352 53 147,544 2009 114,544 66,630 4,829 57 186,060 376 223 2010 244,423 12,612 9,966 170 267,171 447 153
a1 January to 31 May. b20–28% released upstream were not used in predation analyses. Downloaded by [Department Of Fisheries] at 20:03 25 September 2012
FIGURE 3. Daily numbers of unique pinnipeds observed (black line) in the tailrace study area, total salmonids counted passing fishways (gray shaded area), and radio-tagged adult spring Chinook salmon released downstream from Bonneville Dam (vertical black bars, relative scale: sample sizes are given in Table 2). SEA LION PREDATION AT A HYDROELECTRIC DAM 1243
(mean = 149 salmon per population) were calculated from this that entered the tailrace and 1.1% for salmon that approached aggregate sample. fishways (Table 3). In years with substantial pinniped aggrega- tions, mean failure estimates were 5.2% and 3.3%, respectively. By subtraction, mean predation estimates were 3.6% for radio- Annual Salmonid Predation Estimates tagged salmon that entered the tailrace and 2.2% for those that In the observation study, which applied only to the near-dam approached fishways. As expected, these values bracketed the tailrace area, adjusted annual estimates of salmonid consump- 2.6% mean predation estimate in the observation study. tion by pinnipeds ranged from 1,010 to 6,542 fish (mean = 3,845; Table 3). This predation represented 0.4–4.9% (mean = 2.6%) of the salmonids counted at Bonneville Dam each year Weekly Salmonid Predation Estimates from January to May. Annual percentage of salmonids con- In the pinniped observation study, the mean weekly number sumed was negatively correlated with the total January–May of salmonids consumed by pinnipeds steadily increased from salmonid count at Bonneville Dam (r =−0.84, P = 0.005). <10 per week in January and early February to >500 per week Chinook salmon were the primary salmonid prey in all years, in late April and early May (Figure 4). Mean consumption esti- accounting for 85–99% (mean = 93%) of the observed salmonid mates then declined to near zero by the end of May. Numbers of catch. Steelhead were the most captured salmonid in winter. salmonids consumed per week varied among years, but seasonal Mean annual salmonid consumption per capita ranged from 33 patterns were generally similar. to 74 fish (mean = 45) for California sea lions and from 0 to 18 Mean weekly estimates of the proportion of salmonids caught fish (mean = 6) for Steller sea lions. (Catchweek / Countweek) indicated that individual salmonid pre- In the radiotelemetry study years preceding the pinniped dation risk substantially decreased as salmonid abundance in- aggregation, mean dam passage failure rates (i.e., preaggre- creased. In winter, catch proportions ranged from 0.00 in week gation baseline rates) were 1.6% for spring Chinook salmon 1 to 1.30 in week 4 (grand mean = 0.51, grand median = 0.25;
TABLE 3. Estimated numbers and percentages of salmonids consumed by pinnipeds in the Bonneville Dam tailrace. In the 2002–2010 pinniped observation study, expanded consumption estimates included interpolations for missed daylight observation hours and adjusted estimates additionally included catch classed as “unknown salmonid” and estimated nighttime catch. Dam passage failure rates for radio-tagged spring Chinook salmon were for the tagged populations that entered the tailrace and were recorded approaching fishways. The 1996–2001 estimates were prior to the pinniped aggregation and serve as a baseline (Baseline mean). The 2002–2010 estimates were during the pinniped observation years (New mean). The new mean minus the baseline mean (Difference) is the estimated predation effect.
Observations Radio-tagged salmon Expanded consumption Adjusted consumption dam-passage failure (%) Estimated Percentage of Estimated Percentage of Year Consumption (N) total count (%) Consumption (N) total count (%) Tailrace Fishway 1996 2.5 1.2 1997 1.4 1.2 1998 1.3 0.6 2000 1.5 1.4 Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 2001 1.3 1.0 Baseline mean 1.6 1.1 2002 1,010 0.4 1,010 0.4 0.9 0.3 2003 2,329 1.1 2,329 1.1 5.1 3.1 2004 3,533 1.9 3,533 1.9 7.9 5.8 2005 2,920 3.4 2,920 3.5 2006 3,023 2.8 3,520 3.2 6.2 4.0 2007 3,859 4.2 4,507 4.9 8.0 4.0 2008 4,466 2.9 5,099 3.3 2009 4,489 2.4 5,134 2.7 5.2 3.7 2010 6,081 2.2 6,542 2.4 3.2 2.1 New mean 2.3 3,845 2.6 5.2 3.3 Difference 3.6 2.2 1244 KEEFER ET AL.
Figure 4). Numbers of salmonids observed to be consumed ex- ceeded the number of salmonids counted passing Bonneville Dam in 9 of 59 (15%) winter observation weeks, reflecting limited dam passage by salmonids and extended time steelhead were holding downstream. In spring, mean Catchweek / Countweek estimates ranged from 0.01 late in the run (weeks 20–22) to 0.35 in week 13 (grand mean = 0.12, grand median = 0.03). By com- parison, mean Failweek / Tailraceweek estimates for radio-tagged spring Chinook salmon ranged from 0.03 in week 22 to 0.08 in week 14 (grand mean = 0.05, grand median = 0.04; Figure 4). As salmonid abundance increased, weekly estimates of the total number of salmonids consumed increased (Figure 5). In the Catchweek regression model, log10-transformed consumption estimates increased as log10-transformed salmonid abundance increased (r2 = 0.48, F = 130.0, P < 0.001, n = 142 weeks with nonzero catch). This relationship was also significant when winter (r2 = 0.24, F = 14.3, P < 0.001) and spring (r2 = 0.12, F = 12.2, P < 0.001) data were analyzed separately. FIGURE 4. Top panel shows distributions of weekly predation estimates: The nonlinear Michaelis–Menten model indicated that per Catchweek / Countweek in the observation data (gray box plots) and Failweek capita consumption by sea lions rapidly increased from near- / Tailraceweek in the springtime radiotelemetry data (open box plots). Bottom panel shows mean ( ± SD) weekly numbers of salmonids observed caught by zero salmonids consumed per week at the lowest salmonid den- pinnipeds in the Bonneville tailrace (expanded estimates). Vertical lines separate sity to approximately 16 salmonids per week when salmonid winter weeks (1 January–18 March) from spring weeks (19 March–31 May). abundance was ∼5,000 salmonids counted (Figure 6). The Box plots show median, quartile, 10th, and 90th percentiles and the outlying model became asymptotic at higher salmonid densities, with per data in week 4 are not shown. capita consumption estimates increasing to ∼19 salmonids per week at the highest salmonid densities (i.e., ∼80,000 counted; Downloaded by [Department Of Fisheries] at 20:03 25 September 2012
FIGURE 5. Relationship between an index of predation rate [weekly number of salmon caught, log10(Catchweek)] and an index of prey density [the total weekly number of salmonids counted at Bonneville Dam, log10(Countweek)]. Each point represents one weekly estimate from winter (•) and spring (◦), 2002–2010. Regression model is log10(Catchweek) = [0.3744 × log10(Countweek)] + 0.8254. Ten weeks with zero observed catch were excluded because log10(0) is undefined. SEA LION PREDATION AT A HYDROELECTRIC DAM 1245
FIGURE 6. Relationship between the mean per capita pinniped consumption rate (number of salmonids caught per pinniped per week) and an index of prey density (the total weekly number of salmonids counted at Bonneville Dam, Countweek). The functional predation response line was generated using a hyperbolic, Michaelis–Menten equation where: consumption rate = (18.65 × Countweek) / (732.9 + Countweek). Each point represents one weekly estimate, 2002–2010. One outlier was excluded and there were no weeks with zero salmonids.
Figure 6). The model parameter estimates converged and the expect that predator–prey interactions at the site will continue model was significant (F = 398.9, df = 2, P < 0.0001, n = to be dynamic. Sea lion abundance and overall presence in the 151 weeks), though we noted that there was considerable vari- study area steadily increased as experienced animals returned, ability in weekly estimates across all salmonid densities and the new animals were recruited to the aggregation, and many of asymptote fell well below several data points at high salmonid the predators arrived progressively earlier each year. Harass- density (Figure 6). ment efforts, begun in full in 2006, did not substantially slow the trend of increasing total sea lion presence, although the Population-Specific Predation Risk California sea lion removals did probably reduce abundance of The population-specific analyses indicated that individual this species temporarily. These management actions undoubt- predation risk varied among Chinook salmon populations by an edly added a complicating factor to our abundance and pre- order of magnitude or more. Relative risk estimates were highest dation results. Most of the 40 animals removed were multi- for the early migrating spring-run populations and were lowest year visitors who had begun to arrive earlier each year and for those with large summer-run components (Figure 7). The Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 stay longer, so their removal may ultimately skew or alter the highest modeled estimates were for stocks from the Clearwater current trends. Although our analyses were limited to the use and Salmon rivers (Snake River basin), the Umatilla and De- of an index of prey availability rather than direct estimates schutes rivers (lower Columbia River basin), and the Icicle River of prey density, we are confident that observation-based esti- (upper Columbia River basin). The lowest estimated risk was mates of salmonid predation rates were reasonably accurate, for populations in the Okanogan and Wenatchee rivers (upper particularly given the general convergence with radiotelemetry- Columbia River basin), the South Fork Salmon, upper Salmon, based estimates. The estimated percentage of the January– and Imnaha rivers (Snake River basin), and the Klickitat and May salmonid run consumed by pinnipeds provided com- Hood rivers (lower Columbia River basin). The remaining pop- pelling evidence for prey density effects on both annual and ulations, primarily from the lower Columbia River and Snake weekly predation rates. Although some steelhead were con- River basins, had intermediate risk estimates. sumed throughout the winter and spring, the vast majority of the salmonid predation was on spring Chinook salmon. Our DISCUSSION primary hypothesis,that predation risk would differ among Chi- Size, composition, and behaviors of the sea lion feeding ag- nook salmon populations,was supported by the data. The high- gregation at Bonneville Dam changed through time, and we est proportional predation rates occurred in winter and early 1246 KEEFER ET AL.
FIGURE 7. Left panel: Migration timing distributions calculated using 5,229 radio-tagged spring and summer Chinook salmon from 32 upriver populations in the lower Columbia River, Snake River, and Columbia River upstream from the Snake River confluence, 1996–1998 and 2000–2004. Distributions show 5th, 25th, 50th, 75th, and 95th percentiles. Right panel: the relative risk ( ± 1 SE) of predation by pinnipeds, estimated by multiplying weekly mean predation rate estimates from the pinniped observation study (i.e., mean Catchweek / Countweek) by population-specific migration timing distributions. Warm Springs Hatchery and Pelton Dam Trap are in the Deschutes River basin. Ringold Hatchery is adjacent to the Columbia River main stem. Dworshak Hatchery is in the Clearwater River basin. Downloaded by [Department Of Fisheries] at 20:03 25 September 2012
spring, and consequently estimated predation risk was high- range expansion (Carretta et al. 2007) or declines in traditional est for early migrating Chinook salmon populations. These prey species in the estuary or ocean (e.g., NMFS 2009), or included several groups currently listed as threatened under both. California sea lions also may have originally responded to the ESA. large Columbia River salmonid runs in the early 2000s. Spring Chinook salmon counts at Bonneville Dam were several times Sea Lion Aggregation higher in 2000–2002 (mean =∼295,000) than in the previous The recent sea lion aggregations at Bonneville Dam appear 30 years (1970–1999 mean =∼90,000; U.S. Army Corps of to be unprecedented, with no historical observations or archae- Engineers 2009). Regardless of what first attracted sea lions to ological records of otariids found along the Columbia River the dam, the seasonally abundant resident and migratory prey farther than about 150 km upstream from the Pacific Ocean probably explains the persistent return and growth of the ag- (Lyman et al. 2002). Possible explanations for the novel Bon- gregation. Similar prey-driven marine mammal concentrations neville aggregation include sea lion population recovery and commonly recur in coastal habitats (e.g., Piatt and Methven SEA LION PREDATION AT A HYDROELECTRIC DAM 1247
1992; Simila¨ et al. 1996; Sigler et al. 2009) as well as in estuar- neville tailrace was unknown (especially in winter) and prey ine and freshwater systems (e.g., Roffe and Mate 1984; Carter switching or other types of selection may have occurred. For et al. 2001; Wright et al. 2007). example, in a separate evaluation of the radiotelemetry data that Movement data from male California sea lions tagged with included steelhead and fall Chinook salmon, Naughton et al. satellite transmitters in the lower Columbia River (Wright et al. (2011) found that larger fish had higher rates of pinniped-related 2010) suggest that repeat visits to Bonneville Dam were prob- injuries than did smaller fish at Bonneville Dam (especially aby learned foraging behaviors. All tagged sea lions trapped at among steelhead). This suggests that pinnipeds either selected or previously observed at the dam subsequently revisited, while larger salmonids, that larger individuals escaped at higher rates, those never observed at the dam remained well downstream or that smaller individuals were more likely to die from injuries. (Wright et al. 2010). This is consistent with our observations of Somewhat unexpectedly, year-to-year changes in total pin- repeat visits by many individual animals. For example, among niped days (i.e., + 26% in 2005 and −11% in 2007) and other California sea lions readily distinguished by unique marks or measures of pinniped presence suggest that aggregation size was brands, an average of 56% was present in consecutive years and not particularly sensitive to January–May salmonid abundance. others were recorded in as many as 8 years (Stansell et al. 2010). This was probably because many sea lions arrived at Bonneville Several older, repeat-visit California sea lions captured dispro- Dam well before the spring Chinook salmon run. This suggests portionately high numbers of salmonids at Bonneville Dam in that sea lions were not directly responding to prey abundance most years. These animals tended to be present for extended when they entered the Bonneville tailrace study area, perhaps periods and probably developed selective foraging behaviors. because the predator–prey dynamics were still in initial devel- Such specialization has been documented in other sea lion pre- opmental stages or because sea lion migration to the site was dation studies (e.g., Jeffries and Scordino 1997) and can be simply a learned behavior. bioenergetically rewarding at prey-rich sites. This was clearly In a majority of study years, both the numbers of salmonids demonstrated by two California sea lions that fed extensively at consumed per week and per capita consumption rates by sea Bonneville Dam and weighed 660 and 522 kg when captured lions increased as salmonid abundance increased and the per (B. Wright, Oregon Department of Fish and Wildlife, personal capita consumption leveled off, consistent with a functional pre- communication). These animals were believed to be the largest dation response. Other pinniped predation studies (e.g., Bailey California sea lions ever recorded (typical adult males weigh and Ainley 1981–1982; Middlemas et al. 2006) have reported 200–400 kg, Riedman 1990). Type III functional responses, where consumption rapidly in- How new animals were recruited to the aggregation is un- creases after a threshold prey level is exceeded and then be- known. Recruits presumably used social cues from experienced comes asymptotic when predators reach satiation. Our results or dominant animals (e.g., Conradt and Roper 2005) to locate more closely resembled a Type II response, with rapidly increas- the Bonneville site. The most likely source for new recruits was ing per capita consumption as salmonid abundance increased, from the loosely affiliated sea lions at downstream haul-out sites even at low salmonid abundance, followed by an asymptote at in and near the Columbia River estuary. It is also possible that higher salmonid density (see Figure 6). The lack of evidence for recruits were associated with repeat-visit Bonneville animals at a threshold effect (i.e., a Type III response) may have been be- a larger geographic scale, perhaps including haul-outs along the cause there were few alternate prey sources at Bonneville Dam Pacific coast or more distant rookeries (e.g., Lowry and Forney or because there were two predator species. Regardless, very 2005; Wright et al. 2010). Wherever the source, understanding low salmonid density appeared to be associated with high pre- aggregation recruitment mechanisms will be needed to predict dation risk. At higher densities, the Michaelis–Menten model future sea lion population dynamics at the dam and in the lower results suggested a satiation threshold near 19 salmonids per
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 Columbia River. sea lion per week (approximately three fish per day), although many weekly estimates were higher. Predator Behaviors While the general Michaelis–Menten model was useful for Within year, the predation observation data suggested that showing broad predation patterns by the aggregation overall, both California and Steller sea lions opportunistically fed on there was considerable variation in catch rates across salmonid the most available prey, but may select for size or species, or densities. Other predation responses may have had statistical both. In winter, California sea lions targeted steelhead (typically support had we estimated salmonid consumption for individual ∼0.5–0.9 m long, 2–6 kg) and Pacific lampreys (∼0.5–0.8 m, sea lions or for each species separately given general differences 0.4–0.7 kg) whereas the larger Steller sea lions favored subadult in arrival timing and behavior. Similarly, behaviors such as par- white sturgeon (∼0.7–1.5 m, 4–55 kg). Both species shifted to tial prey consumption (e.g., Gende et al. 2001), prey stealing, Chinook salmon (∼0.6–0.9 m, 5–8 kg) when the spring run and aggregation responses to hazing and California sea lion re- arrived. Within sea lion species, these patterns were consistent moval all could potentially affect how the per capita predation with predation as a function of preferred prey encounter rates data were interpreted. We note that identifying a specific pre- (e.g., Werner and Hall 1974; Middlemas et al. 2006; Garrott dation response curve was not a primary study objective. How- et al. 2007). However, relative prey availability in the Bon- ever, the model results should be useful for projecting sea lion 1248 KEEFER ET AL.
predation estimates under scenarios with different predator and swamping effects (e.g., Ims 1990) and substantially reduced prey densities at Bonneville Dam. individual salmonid predation risk.
Population-Specific Predation Risk Salmonid Predation Estimates The population risk model clearly indicated that Chinook Results from both the observation and radiotelemetry studies salmon from early migrating populations were most at risk of provided some concrete estimates of the direct mortality im- sea lion predation. These included salmon from the Clearwa- pacts of the sea lion aggregation. At the annual scale, consump- ter, Salmon, Icicle, Deschutes, and Umatilla rivers. Of these, tion percentages were negatively correlated with total salmonid the Salmon and Icicle river salmon populations are listed as counts despite the large fluctuations in predator and prey popu- threatened under the ESA (Good et al. 2005). Importantly, the lations. The highest observed consumption (3.5% in 2005 and risk estimates were based on the migration histories of thou- 4.9% in 2007) was in the two study years with the lowest sands of Chinook salmon tagged over multiple years, but they January–May salmonid counts (81,274 and 88,476 fish, respec- were nonetheless only a relative measure and impacts on indi- tively). The highest passage failure rate for radio-tagged fish was vidual populations certainly will differ each year. At a broad also in 2007 when 8.0% of the salmon that entered the tailrace scale, we think the multiyear data set captured many of the failed to pass (none were radio-tagged in 2005). Both results among-population migration timing differences because indi- suggest that sea lions were efficient hunters at low salmonid vidual stocks consistently migrated at the same time each year density and that predation impacts will probably be higher in relative to the overall run (Keefer et al. 2004b). Therefore, while years with small salmonid run size. However, it is unknown the initiation of the Columbia River spring Chinook salmon run whether this will be a linear relationship or if there will be a varies by several weeks as a function of environmental and run threshold at low salmonid density or low resident prey density composition effects (Keefer et al. 2008b; Anderson and Beer whereupon sea lions leave the study area. 2009), the populations present early in each migration are simi- The observed preemptory arrival by the sea lion aggrega- lar across years. This conclusion is broadly supported by studies tion would directly contribute to higher predation impacts in showing population-specific heritability of migration timing in years with small Chinook salmon returns, although reductions salmonids (e.g., Quinn et al. 2000, 2011; Waples et al. 2004) in either per capita consumption rates or sea lion residence and suggests that sea lion predation will probably continue to times would reduce these impacts. If consumption remained disproportionately affect early timed Columbia River popula- relatively constant, the 2009–2010 aggregations that ate an es- tions. Over several generations, relatively high predation early timated ∼5,100–6,500 salmonids would have a 12–15% impact in the spring may result in directional selection, as observed in on a run of 42,000 fish (as in 1998, Table 2). A predation impact human fisheries (e.g., Consuegra et al. 2005; Quinn et al. 2007), of this magnitude would probably trigger significant legal and and reduced population size. management actions. More generally, sea lion predation esti- Ideally, the population risk model would have matched mates at Bonneville Dam need to be considered in the context weekly predation estimates with weekly salmonid composi- of total adult salmonid mortality. The mean annual predation tion estimates from the same year. However, such data are estimate of 2.6%, primarily of spring Chinook salmon, was rarely available, even in intensively monitored systems like the considerably lower than human harvest rates for this population Columbia River. We did not attempt to make these types of es- group. Keefer et al. (2005) reported that mean harvest of spring– timates because population-specific migration timing data were summer Chinook salmon was 8.7% in the main-stem Columbia not collected in the 2005–2010 radiotelemetry studies and be- River hydrosystem above Bonneville Dam and an additional cause weekly sample sizes in 2002–2004 were prohibitively
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 5.9% in Columbia River tributaries upstream from Bonneville small. Nonetheless, population-specific risks certainly differ Dam. Additional human harvest and additional sea lion pre- from the aggregate population risk as measured by the January– dation occur downstream from the Bonneville study area, and May observation data. At a minimum, predation as a percentage these cumulative impacts need to be addressed in management of individual population size varied several-fold between the and conservation plans. earliest- and latest-timed Chinook salmon populations. Both studies indicated that few salmonids were captured dur- More explicit quantification of population-specific risk will ing winter and early spring but catch rates as a proportion of require significantly improved in-season run composition data the fish counts were high. In part, this was because pinniped : and continued refinement of corresponding predation rates. Such salmonid ratios were highest early in the year before the spring information would allow managers to directly estimate the po- run arrived. In addition, ectothermic salmonids may have been tential effects of sea lion predation on long-term population- especially vulnerable to predation at this time given physiolog- specific growth rates (i.e., lambda, λ) for Chinook salmon and ical constraints on escape rates at low water temperatures (e.g., steelhead. For example, Chinook salmon population growth Domenici and Blake 1997). During periods of peak salmonid models reported for Columbia River stocks by McClure et al. abundance, the weekly predation estimates (i.e., per capita pre- (2003) and Kareiva et al. (2000) indicated that adult mortal- dation risk for salmonids) were far lower, indicating likely prey- ity had relatively modest effects on population growth rates SEA LION PREDATION AT A HYDROELECTRIC DAM 1249
compared with mortality in other life stages. However, stocks was supported by the concordance of the estimates from the with small effective population size, like some of the threatened observation and radiotelemetry data sets. early run Chinook salmon populations in our study (Waples et al. Estimating predation effects on ESA-listed Columbia River 2010), may be relatively more sensitive to even small increases salmonid populations will continue to be challenging given in adult mortality. The predation indices we developed could be inter- and intraannual variability in population-specific abun- used to quantify such population-level impacts. dance and migration timing. This challenge is elevated in large, Compared with spring Chinook salmon, very little is known multistock systems like the Columbia River, where dozens of about the source populations of the steelhead that congregate salmon and steelhead populations migrate concurrently and in the Bonneville tailrace in winter and spring. They are pre- ESA-listed fish are mixed with hatchery-origin and unlisted wild sumably a mix of winter-run fish that return to lower Columbia fish. Fish tagging and monitoring projects like the one described River tributaries (Busby et al. 1996) and overwintering summer- here can provide valuable information on relative risk, but ad- run fish that return to sites throughout the basin (e.g., Keefer ditional assessment methodologies will be needed to produce et al. 2008a). Portions of each of these populations are currently quantitative, population-specific estimates. designated as threatened under the ESA. Given the apparent In conclusion, the Bonneville sea lion aggregation has gar- vulnerability of steelhead to pinniped predation in winter and nered international attention because of the conflict among early spring, genetic or other assessments of which populations groups that advocated protection for both the predator and prey are present in the Bonneville tailrace during this period may be species involved. The opposing trajectories of pinniped and warranted. Similarly, pinnipeds consumed significant numbers salmonid populations along the Pacific coast suggest that the of subadult white sturgeon (∼6,100 fish in the tailrace) and adult incidence of such conflicts will probably increase and managers Pacific lamprey (∼2,900 fish) over the nine observation study will be faced with the difficult task of balancing among com- years (Stansell et al. 2010). Given conservation concerns and un- peting conservation objectives. Among the most contentious certainty about the size of these populations, further evaluations pinniped management actions to date was the 2008–2010 re- of population-level predation effects may be necessary. moval of 40 high-impact, repeat-visit California sea lions from the Bonneville aggregation as a measure to reduce predation. The effectiveness of this removal is not yet well understood, Uncertainties in part because of the concurrent rapid increase in Steller sea Uncertainty about the scope of the predation impacts on ESA- lions. Two apparent behavioral changes during this transitional listed Pacific salmon and steelhead has prompted development period were a shift to more frequent short-duration visits by of new sea lion hazing techniques, lethal removal, and a variety younger California sea lions and an increase in prey stealing of methods (e.g., bioenergetic modeling, satellite telemetry, scat (Stansell et al. 2010). Hundreds of kleptoparasitism events were analysis) to refine predation estimates both at Bonneville Dam observed in 2008–2010, with the larger Steller sea lions taking and downstream. As with each of these techniques, our preda- prey from California sea lions. Such aggressive interactions may tion estimation methods relied on several critical assumptions. reduce per capita predation impacts by limiting foraging time First, we were quite confident that the observation data provided (e.g., Abrams and Ginzburg 2000) but may also result in higher reasonable estimates of pinniped abundance near the dam given total consumption if stolen prey is replaced by new captures. the regularity and duration of our effort. Second, we think es- It remains to be seen whether the changing composition and timates of which prey were consumed (i.e., salmonids versus behaviors of the sea lion aggregation will produce a net increase other species) were also reasonable given that most predation or decrease in predation. The dynamic and evolving situation at events were readily observable and that pinnipeds predictably Bonneville Dam highlights the value of long-term monitoring
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 bring large prey to the surface. However, we did not attempt of protected populations and the need for careful consideration to quantify interobserver differences in prey species identifica- of management actions and their potential outcomes. tion or predation event detection. Double-blind experiments or other methods to measure observer-related bias could increase confidence in future predation observation studies. Third, we ACKNOWLEDGMENTS were less confident that adult salmonid counts at Bonneville We thank all the observers in the pinniped observation pro- Dam were a consistently reliable index of abundance in the dam gram. They spent many hours in extreme weather conditions tailrace. As described above, the difference between fish counts to obtain this important information for regional managers. We at the dam and fish abundance in the tailrace probably varied also thank the U.S. Department of Agriculture for supplying weekly and seasonally, and particularly during winter when mi- agents to employ harassment activities to keep sea lions away grants were less active. However, by averaging over weekly from fishways, and the Oregon Department of Fish and Wildlife, timescales, the index should have broadly reflected the density Washington Department of Fish and Wildlife, and Columbia of salmonids in the tailrace through the run season and it was River Inter-Tribal Fish Commission for supplying boats for unlikely that using the count data as an index of run size se- harassment activities. Special thanks to the U.S. Army Corps riously biased our annual predation estimates. This conclusion of Engineers for designing and deploying sea lion exclusion 1250 KEEFER ET AL.
devices, which have kept sea lions out of the Bonneville Dam Domenici, P., and R. W. Blake. 1997. The kinematics and performance of fish fishways since their installation in 2006. We are similarly in- fast-start swimming. Journal of Experimental Biology 200:1165–1178. debted to individuals at many organizations for gathering and Garrott, R. A., J. E. Bruggeman, M. S. Becker, S. T. Kalinowski, and P. J. White. 2007. Evaluating prey switching in wolf-ungulate systems. Ecological compiling the Chinook salmon data used in this study. We Applications 17:1588–1597. especially appreciate the personnel at the many cooperating Gende, S. M., T. P. Quinn, and M. F. Willson. 2001. Consumption choice by hatcheries as well as anglers and tribal fishers for returning ra- bears feeding on salmon. Oecologia 127:372–382. dio transmitters. We also thank T. Bjornn, B. Burke, M. Moser, Good, T. P., R. S. Waples, and P. Adams, editors. 2005. Updated status of L. Stuehrenberg, A. Matter, K. Frick, and T. Bohn (National federally listed ESUs of West Coast salmon and steelhead. NOAA Technical Memorandum NMFS-NWFSC-66. Marine Fisheries Service) for their collaboration on the teleme- Holling, C. S. 1959. Some characteristics of simple types of predation and try project. Funding was provided by the U.S. Army Corps of parasitism. Canadian Entomologist 91:385–398. Engineers; we thank D. Clugston, M. Shutters, R. Dach, M. Holling, C. S. 1966. The functional response of invertebrate predators to prey Langeslay, and T. Mackey for their support. At the University density. Memoirs of the Entomological Society of Canada 48:1–86. of Idaho, T. Bjornn, R. Ringe, K. Tolotti, M. Jepson, S. Lee, C. Ims, R. A. 1990. On the adaptive value of reproductive synchrony as a predator- swamping strategy. American Naturalist 136:485–498. Boggs, T. Reischel, G. Naughton, W. Daigle, M. Morasch, T. Jeffries, S., J. Huber, J. Calambokidis, and J. Laake. 2003. Trends and status of Dick, D. Joosten, and C. Morat helped with telemetry project harbor seals in Washington State: 1978–1999. Journal of Wildlife Manage- oversight, field operations, and collection and processing of ment 67:207–218. telemetry data. We also thank the two anonymous reviewers Jeffries, S., and J. Scordino. 1997. Efforts to protect a winter steelhead run and the journal’s associate editor whose constructive comments from California sea lions at the Ballard locks. Pages 107–115 in G. Stone, J. Goebel, and S. Webster, editors. Pinniped populations, eastern north Pacific: greatly improved the manuscript. status, trends, and issues. New England Aquarium, Boston. Kareiva, P., M. Marvier, and M. McClure. 2000. Recovery and management REFERENCES options for spring/summer Chinook salmon in the Columbia River basin. Abrams, P. A., and L. R. Ginzburg. 2000. The nature of predation: prey de- Science 290:977–979. pendent, ratio dependent or neither? Trends in Ecology and Evolution 15: Keefer, M. L., C. T. Boggs, C. A. Peery, and C. C. Caudill. 2008a. Overwintering 337–341. distribution, behavior, and survival of adult summer steelhead: variability Anderson, J. J., and W. N. Beer. 2009. Oceanic, riverine, and genetic influ- among Columbia River populations. North American Journal of Fisheries ences on spring Chinook salmon migration timing. Ecological Applications Management 28:81–96. 19:1989–2003. Keefer, M. L., C. A. Peery, T. C. Bjornn, M. A. Jepson, and L. C. Stuehrenberg. Bailey, K. M., and D. G. Ainley. 1981–1982. The dynamics of California sea 2004a. Hydrosystem, dam, and reservoir passage rates of adult Chinook lion predation on Pacific hake. Fisheries Research 1:163–176. salmon and steelhead in the Columbia and Snake rivers. Transactions of the Baraff, L. S., and T. R. Loughlin. 2000. Trends and potential interactions be- American Fisheries Society 133:1413–1439. tween pinnipeds and fisheries of New England and the U.S. West Coast. U.S. Keefer, M. L., C. A. Peery, and C. C. Caudill. 2008b. Migration timing of National Marine Fisheries Service Marine Fisheries Review 62(4):1–39. Columbia River spring Chinook salmon: effects of temperature, river dis- Brown, R. F., and B. R. Mate. 1983. Abundance, movements, and feeding habits charge, and ocean environment. Transactions of the American Fisheries So- of harbor seals, Phoca vitulina, at Netarts and Tillamook bays, Oregon. U.S. ciety 137:1120–1133. National Marine Fisheries Service Fishery Bulletin 81:291–301. Keefer, M. L., C. A. Peery, W. R. Daigle, M. A. Jepson, S. R. Lee, C. T. Brown, R. F., B. E. Wright, S. D. Riemer, and J. Laake. 2005. Trends in Boggs, K. R. Tolotti, and B. J. Burke. 2005. Escapement, harvest, and un- abundance and current status of harbor seals in Oregon: 1977–2003. Marine known loss of radio-tagged adult salmonids in the Columbia River–Snake Mammal Science 21:657–670. River hydrosystem. Canadian Journal of Fisheries and Aquatic Sciences 62: Busby, P. J., T. C. Wainwright, G. J. Bryant, L. J. Lierheimer, R. S. Waples, 930–949. F. W. Waknitz, and I. V. Lagomarsino. 1996. Status review of West Coast Keefer, M. L., C. A. Peery, M. A. Jepson, K. R. Tolotti, T. C. Bjornn, and steelhead from Washington, Idaho, Oregon, and California. NOAA Technical L. C. Stuehrenberg. 2004b. Stock-specific migration timing of adult spring– Memorandum NMFS-NWFSC-27. summer Chinook salmon in the Columbia River basin. North American Jour- Carretta, J. V., K. A. Forney, M. S. Lowry, J. Barlow, J. Baker, B. Hanson, and nal of Fisheries Management 24:1145–1162.
Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 M. M. Muto. 2007. U.S. Pacific marine mammal stock assessments: 2007. Lowry, M. S., and K. A. Forney. 2005. Abundance and distribution of California NOAA Technical Memorandum NMFS-SWFSC-414. sea lions (Zalophus californianus) in central and northern California during Carter, T. J., G. J. Pierce, J. R. G. Hislop, J. A. Houseman, and P. R. Boyle. 1998 and summer 1999. U.S. National Marine Fisheries Service Fishery 2001. Predation by seals on salmonids in two Scottish estuaries. Fisheries Bulletin 103:331–343. Management and Ecology 8:207–225. Lyman, R. L., J. L. Harpole, C. Darwent, and R. Church. 2002. Prehistoric Caudill, C. C., W. R. Daigle, M. L. Keefer, C. T. Boggs, M. A. Jepson, B. J. occurrence of pinnipeds in the lower Columbia River. Northwestern Naturalist Burke, R. W. Zabel, T. C. Bjornn, and C. A. Peery. 2007. Slow dam passage in 83:1–6. adult Columbia River salmonids associated with unsuccessful migration: de- McClure, M. M., E. E. Holmes, B. L. Sanderson, and C. E. Jordan. 2003. layed negative effects of passage obstacles or condition-dependent mortality? A large-scale, multispecies status assessment: anadromous salmonids in the Canadian Journal of Fisheries and Aquatic Sciences 64:979–995. Columbia River basin. Ecological Applications 13:964–989. Clemens, B. J., T. R. Binder, M. F. Docker, M. L. Moser, and S. A. Sower. 2010. Middlemas, S. J., T. R. Barton, J. D. Armstrong, and P. M. Thompson. Similarities, differences, and unknowns in biology and management of three 2006. Functional and aggregative responses of harbour seals to changes in parasitic lampreys of North America. Fisheries 35:580–594. salmonid abundance. Proceedings of the Royal Society of London B 273: Conradt, L., and T. J. Roper. 2005. Consensus decision making in animals. 193–198. Trends in Ecology and Evolution 20:449–456. NMFS (National Marine Fisheries Service). 2009. Endangered and threatened Consuegra, S., C. Garc´ıa de Leaniz,´ A. Serdio, and E. Verspoor. 2005. Selective wildlife and plants: proposed threatened status for southern distinct popula- exploitation of early running fish may induce genetic and phenotypic changes tion segment of eulachon. Federal Register 74:48(13 March 2009):10857– in Atlantic salmon. Journal of Fish Biology 67(Supplement 1):129–145. 10876. SEA LION PREDATION AT A HYDROELECTRIC DAM 1251
NMFS (National Marine Fisheries Service). 2011a. Endangered and threatened Sigler, M. F., D. J. Tollit, J. J. Vollenweider, J. F. Thedinga, D. J. Csepp, J. N. species: 5-year reviews for 17 evolutionarily significant units and distinct pop- Womble, M. A. Wong, M. J. Rehberg, and A. W. Trites. 2009. Steller sea lion ulation segments of Pacific salmon and steelhead. Federal Register 76:157(15 foraging response to seasonal changes in prey availability. Marine Ecology August 2011):50448–50449. Progress Series 388:243–261. NMFS (National Marine Fisheries Service). 2011b. Marine mammals; pinniped Simila,¨ T., J. C. Holst, and I. Christensen. 1996. Occurrence and diet of killer removal authority. Federal Register 76:176(12 September 2011):56167– whales in northern Norway: seasonal patterns relative to the distribution 56171. and abundance of Norwegian spring-spawning herring. Canadian Journal of Naughton, G. P., M. L. Keefer, T. S. Clabough, M. A. Jepson, S. R. Lee, C. A. Fisheries and Aquatic Sciences 53:769–779. Peery, and C. C. Caudill. 2011. Influence of pinniped-caused injuries on the Stansell, R. J., K. M. Gibbons, and W. T. Nagy. 2010. Evaluation of pin- survival of adult Chinook salmon (Oncorhynchus tshawytscha) and steelhead niped predation on adult salmonids and other fish in the Bonneville trout (Oncorhynchus mykiss) in the Columbia River basin. Canadian Journal Dam tailrace, 2008–2010. U.S. Army Corps of Engineers, Cascade Locks, of Fisheries and Aquatic Sciences 68:1615–1624. Oregon. Piatt, J. F., and D. A. Methven. 1992. Theshold foraging behavior of baleen Stansell, R. J., S. C. Tackley, W. T. Nagy, and K. M. Gibbons. 2009. Evaluation whales. Marine Ecology Progress Series 84:205–210. of pinniped predation on adult salmonids and other fish in the Bonneville Pitcher, K. W., P. F. Olesiuk, R. F. Brown, M. S. Lowry, S. J. Jeffries, J. L. Sease, Dam tailrace. U.S. Army Corps of Engineers, 2009 Field Report, Cascade W. L. Perryman, C. E. Stinchcomb, and L. F. Lowry. 2007. Abundance and Locks, Oregon. distribution of the eastern North Pacific Steller sea lion (Eumetopias jubatus) Tackley, S. C., R. J. Stansell, and K. M. Gibbons. 2008. Pinniped predation on population. U.S. National Marine Fisheries Service Fishery Bulletin 107:102– adult salmonids and other fish in the Bonneville Dam tailrace, 2005–2007. 115. U.S. Army Corps of Engineers, Cascade Locks, Oregon. Quinn, T. P., S. Hodgson, L. Flynn, R. Hilborn, and D. E. Rogers. 2007. Direc- USACE (U.S. Army Corps of Engineers). 2009. Annual fish passage report. tional selection by fisheries and the timing of sockeye salmon (Oncorhynchus USACE, U.S. Army Engineer Districts, Portland, Oregon, and Walla Walla, nerka) migrations. Ecological Applications 17:731–739. Washington. Quinn, T. P., M. J. Unwin, and M. T. Kinnison. 2000. Evolution of temporal Waples, R. S., D. W. Jensen, and M. McClure. 2010. Eco-evolutionary dynamics: isolation in the wild: genetic divergence in timing of migration and breeding fluctuations in population growth rate reduce effective population size in by introduced Chinook salmon populations. Evolution 54:1372–1385. Chinook salmon. Ecology 91:902–914. Quinn, T. P., M. J. Unwin, and M. T. Kinnison. 2011. Contemporary diver- Waples, R. S., D. J. Teel, J. M. Myers, and A. R. Marshall. 2004. Life-history gence in migratory timing of naturalized populations of Chinook salmon, divergence in Chinook salmon: historic contingency and parallel evolution. Oncorhynchus tshawytscha, in New Zealand. Evolutionary Ecology Research Evolution 58:386–403. 13:45–54. Weise, M. J., and J. T. Harvey. 2005. Impact of the California sea lion (Zalophus Real, L. A. 1977. The kinetics of functional response. American Naturalist californianus) on salmon fisheries in Monterey Bay, California. U.S. National 111:289–300. Marine Fisheries Service Fishery Bulletin 103:685–696. Riedman, M. 1990. The pinnipeds: seals, sea lions, and walruses. University of Werner, E. E., and D. J. Hall. 1974. Optimal foraging and the size selec- California Press, Berkeley. tion of prey by the bluegill sunfish (Lepomis macrochirus). Ecology 55: Rieman, B. E., and R. C. Beamesderfer. 1990. White sturgeon in the lower 1042–1052. Columbia River: is the stock overexploited? North American Journal of Fish- Wright, B. E., S. D. Riemer, R. F. Brown, A. M. Ougzin, and K. A. Bucklin. eries Management 10:388–396. 2007. Assessment of harbor seal predation on adult salmonids in a Pacific Roffe, T. J., and B. R. Mate. 1984. Abundances and feeding habits of pinnipeds Northwest estuary. Ecological Applications 17:338–351. in the Rogue River, Oregon. Journal of Wildlife Management 48:1262–1274. Wright, B. E., M. J. Tennis, and R. F. Brown. 2010. Movements of male SAS (Statistical Analysis Systems). 2000. SAS/STAT user’s guide, version 8. California sea lions captured in the Columbia River. Northwest Science SAS Institute, Cary, North Carolina. 84:60–72. Downloaded by [Department Of Fisheries] at 20:03 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:04 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
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Largemouth Bass Selected for Differential Vulnerability to Angling Exhibit Similar Routine Locomotory Activity in Experimental Ponds Thomas R. Binder a b , Michael A. Nannini c , David H. Wahl c , Robert Arlinghaus d e , Thomas Klefoth d , David P. Philipp f & Steven J. Cooke b a Great Lakes Fishery Commission, Hammond Bay Biological Station, 11199 Ray Road, Millersburg, Michigan, 49759, USA b Fish Ecology and Conservation Physiology Laboratory, Department of Biology and Institute of Environmental Science, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, K1S 5B6, Canada c Illinois Natural History Survey, Sam Parr Biological Station, 6401 Meacham Road, Kinmundy, Illinois, 62854, USA d Department of Biology and Ecology of Fishes, Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Müggelseedamm 310, 12587, Berlin, Germany e Inland Fisheries Management Laboratory, Faculty of Agriculture and Horticulture, Humboldt-Universität zu Berlin, Invalidenstrasse 42, 10115, Berlin, Germany f Illinois Natural History Survey, University of Illinois, 1816 South Oak Street, Champaign, Illinois, 61820, USA Version of record first published: 30 Jul 2012.
To cite this article: Thomas R. Binder, Michael A. Nannini, David H. Wahl, Robert Arlinghaus, Thomas Klefoth, David P. Philipp & Steven J. Cooke (2012): Largemouth Bass Selected for Differential Vulnerability to Angling Exhibit Similar Routine Locomotory Activity in Experimental Ponds, Transactions of the American Fisheries Society, 141:5, 1252-1259 To link to this article: http://dx.doi.org/10.1080/00028487.2012.688919
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. Transactions of the American Fisheries Society 141:1252–1259, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.688919
ARTICLE
Largemouth Bass Selected for Differential Vulnerability to Angling Exhibit Similar Routine Locomotory Activity in Experimental Ponds
Thomas R. Binder* Great Lakes Fishery Commission, Hammond Bay Biological Station, 11199 Ray Road, Millersburg, Michigan 49759, USA; and Fish Ecology and Conservation Physiology Laboratory, Department of Biology and Institute of Environmental Science, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada Michael A. Nannini and David H. Wahl Illinois Natural History Survey, Sam Parr Biological Station, 6401 Meacham Road, Kinmundy, Illinois 62854, USA Robert Arlinghaus Department of Biology and Ecology of Fishes, Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Muggelseedamm¨ 310, 12587 Berlin, Germany and Inland Fisheries Management Laboratory, Faculty of Agriculture and Horticulture, Humboldt-Universitat¨ zu Berlin, Invalidenstrasse 42, 10115 Berlin, Germany Thomas Klefoth Department of Biology and Ecology of Fishes, Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Muggelseedamm¨ 310, 12587 Berlin, Germany David P. Philipp Illinois Natural History Survey, University of Illinois, 1816 South Oak Street, Champaign, Illinois 61820, USA Steven J. Cooke Fish Ecology and Conservation Physiology Laboratory, Department of Biology and Institute of Environmental Science, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada Downloaded by [Department Of Fisheries] at 20:04 25 September 2012
Abstract A growing body of work is focused on attempting to understand the biological mechanism(s) by which some fish are highly vulnerable to angling while others are not. We used electromyogram telemetry to monitor routine locomotory activity of artificially selected largemouth bass Micropterus salmoides in experimental ponds to test two potential explanatory hypotheses: (1) that the difference in angling vulnerability between high-vulnerability (HV) bass and low-vulnerability (LV) bass is related to a difference in routine activity level between the two groups, and (2) that the difference in vulnerability between HV and LV bass is related to a difference in the diel activity pattern
*Corresponding author: [email protected] Received January 19, 2012; accepted April 18, 2012
1252 ANGLING AND LOCOMOTORY ACTIVITY IN LARGEMOUTH BASS 1253
displayed by each group (e.g., LV fish are more active at night, a time where there is typically little bass fishing effort). Neither hypothesis was supported by our results. Differences in vulnerability to angling in artificially selected lines of largemouth bass were not related to inherent differences in routine locomotory activity in our ponds. Mean daily activity levels were close to 5% of maximum swim speed in both groups, which we estimated to reflect a mean swimming distance of approximately 5,875 m (range = 1,280–9,670 m) per day. There was also no difference in the diel pattern of activity displayed by the two groups. Both HV and LV bass displayed a significant diurnal activity pattern: 16% and 19% higher activity levels during the day than at night, respectively. These results contribute to the ongoing efforts to understand the behavioral basis of vulnerability to angling in largemouth bas and other fish species.
There is an increasing concern over the potential evolu- (Redpath et al. 2009), which was later explained by a 10% higher tionary consequences of fishing-induced phenotypic selection standard metabolic rate (SMR) in HV bass (Redpath et al. 2010). (e.g., Policansky 1993; Walsh et al. 2006; Jørgensen et al. 2007; Furthermore, there was evidence of reduced aerobic and anaer- Kuparinen and Merila¨ 2007; Theriault´ et al. 2008; Enberg et al. obic capacity in the LV bass (Redpath et al. 2010). In addition, 2009). Indeed, there is now abundant evidence that size-selective Cooke et al. (2007) identified significant differences in obligate harvest in commercial fisheries has contributed to changes in parental care behavior exhibited by nesting males, HV males ex- the life history traits of heavily fished populations (for review hibiting more intense parental care and expending more energy see Law 2000; Heino and Godø 2002; Jørgensen et al. 2007; in guarding the nest. High vulnerability males were also found Hutchings and Fraser 2008). In addition to size-selectivity that is to behave more aggressively against potential predators than LV common in most fishing gear, some fishing gear, especially those males. Because metabolism and aggression are often correlated gears working passively and depending on active fish movement (Metcalfe et al. 1995; Ros et al. 2006; Huntingford et al. 2010), to encounter the gear, may preferentially catch specific behav- the elevated aggression level exhibited by HV males represents ioral phenotypes (Heino and Godø 2002). As a result, fishing a plausible mechanism by which the greater vulnerability to can also influence the evolution of life history traits through lure-based recreational fishing (which was also used during the correlated behavioral mechanisms (Lewin et al. 2006; Biro and selection process; Philipp et al. 2009) can be explained. Post 2008; Uusi-Heikkila¨ et al. 2008). Moreover, because be- One further mechanism that could explain the difference in havioral traits may have larger heritability compared with life vulnerability to angling between HV and LV largemouth bass is history traits, evolution by means of behavioral selection may differences in locomotory activity. This is because vulnerability occur more rapidly (Uusi-Heikkila¨ et al. 2008), providing an to lure-based angling should, at least in part, be a function of advantage for behavioral phenotypes less vulnerable to fishing the probability of encountering the angler’s lure. Under the (Cooke et al. 2007; Philipp et al. 2009). assumption that HV and LV bass both inhabit the same habitats, Like commercial fishing, recreational fishing targets a one would then predict that individuals that are more active specific suite of behaviors that make some fish more likely to be should have a higher probability of encountering a lure or bait angled than others. Vulnerability to angling has been linked to and thus being captured. Similarly, differences in vulnerability personality traits like boldness (Wilson et al. 2011) and aggres- to angling could result from differences in the expression of diel sion (Suski and Philipp 2004; Cooke et al. 2007), and may also activity, as long as anglers preferentially fish during particular be affected by learning and other cognitive abilities (Beukema times of the day. In this case, vulnerable fish need not be more 1970; Raat 1985; Askey et al. 2006), all of which may influence active in general but instead are more likely to be active during Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 the fitness of an individual. Although it has largely been ignored, those times of the day that they are being targeted by anglers. recreational fishing-induced selection, therefore, has the poten- Because most largemouth bass fishing happens during the day, tial to exert strong selective pressure on a population, which will as did the fishing for the selection experiment by Philipp et al. lead to evolutionary adaptations as long as the targeted traits are (2009), one would predict HV bass to be significantly more heritable and not the result of unknown environmental factors. active at day compared with LV bass. A long-term artificial selection study by Philipp et al. (2009) The purpose of this study was to test the hypothesis that the on largemouth bass Micropterus salmoides provides empirical difference in vulnerability to angling between the two artificially evidence that vulnerability to angling is a complex heritable selected lines of largemouth bass is due to inherent differences trait, exhibiting a realized heritability (h2) of 0.15. Within just in routine locomotory activity. We employed electromyogram three generations of truncated selection, changes in several (EMG) telemetry in a 4,000-m2 experimental pond to compare physiological and behavioral traits have been noted as correlated both the level and diel pattern of activity displayed by individ- responses to selection for vulnerability to angling. For example, uals from the two selected lines. We made two nonmutually age-1 low-vulnerability (LV) bass had a 9–17% higher realized exclusive predictions. First, if the difference in vulnerability to growth rate than their high-vulnerability (HV) counterparts angling between the two lines is related to differences in activity 1254 BINDER ET AL.
level, then HV bass should display a greater overall level of in length) were inserted parallel to one another (∼1 cm apart), activity than LV bass. Second, if the difference in vulnerability into the band of red muscle running along the lateral line using to angling between the two lines is related to differences in a 12-gauge plunger device. The external antenna wire of the the period during which the fish are most active, then the two transmitter was fed out through the open incision, which was groups should display different diel swimming activity patterns. then closed with four simple, interrupted monofilament sutures. Fish were then transferred back to holding tanks to recover from the anesthetic. All surgeries were performed by a single METHODS surgeon to control for surgery effects (Cooke et al. 2003). Experimental animals.—This study used a total of 29 adult We did not attempt to assign a sex to the fish at the time of largemouth bass (15 HV and 14 LV; age 2 + , mixed sexes, surgery because the gonads were reduced at this time of the mean ± SE TL = 337 ± 5 mm, mean ± SE mass = 681 ± year and were not easily seen through the incision without the 26 g). The fish used in the present study belonged to the F5 risk of causing trauma to the internal organs. We did record a generation (bred in 2007), having experienced three generations sex for each of the fish in trial 2 at the end of the trial when the of truncated selection according to vulnerability to angling and transmitters were removed, but, unfortunately, transmitters were two further generations without selection. The selection process removed from the fish in trial 1 before sex could be recorded. began in 1977 in Ridge Lake, an experimental reservoir in Fox Electromyogram tag calibration.—The EMG tags used in Ridge State Park, Charleston, Illinois, as described completely this study transmitted a unitless EMG value ranging from 0 in Philipp et al. (2009). Briefly, between 1977 and 1980, all to 50 that can be used to estimate the swim speed of fishes angling in Ridge Lake was controlled and catch histories were (Cooke et al. 2004; Brown et al. 2007). The caveat is that, maintained for each largemouth bass that was landed. At the although the relationship between EMG value and swim speed end of the 4-year study, the lake was drained and the bass is approximately linear (Cooke et al. 2004), the slope and were categorized based on the number of times they had been intercept of the relationship varies by tag, fish, and position of caught by anglers. Bass that were never caught were used as the electrodes (Beddow and McKinley 1999; Geist et al. 2002; broodstock to establish an LV line, and bass that were captured Brown et al. 2007). For this reason, the EMG tags must be four or more times within a single year were used as broodstock individually calibrated after implantation if the values are to be to establish an HV line. High vulnerability and LV bass were accurately used to estimate swimming speeds. 2 separately bred in two 800-m brood ponds. The F1 fish were Electromyogram tags in this study were calibrated relative to fin-clipped to identify parental line and experimentally angled, the range of EMG value obtained for each fish. In this way, EMG separated into LV and HV groups based on the number of times values were converted to percent of maximum swim speed, they were caught, and then bred to establish an F2 generation. rather than an absolute swimming speed, using the equation This process was repeated through the F3 generation. 2 The F fish used in this study were stocked into 1,200-m (EMG − EMGmin) 5 % max activity = × 100, ponds at the Sam Parr Biological Station (near Kinmundy, (EMGmax − EMGmin) Illinois) approximately 1 year before the current study. High- vulnerability and LV bass (identified by pectoral fin clips) of where EMG is the EMG value to be converted, EMGmin is the both sexes were mixed equally into two groups for use in two minimum EMG value for a given fish that summed greater than separate trials for assessment of activity metrics (15 bass in 1% of the total number of EMG values for that fish (assumption trial 1 and 14 bass in trial 2). The day before each trial began, was that this value represents the EMG level when the fish the holding pond was drained and the bass were moved to were at rest in the pond), and EMGmax was the greatest EMG
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 1.25-m-diameter holding tanks to await surgical implantation value obtained during a calibration procedure at the end of the of the EMG transmitters. trial where the bass were stimulated to burst swim (maximum Surgeries began the next morning at approximately exertion) while being captured in a dip net and held at the 0900 hours. Bass were netted individually from the holding surface for up to 60 s. tank and anesthetized in a 50 mg/L solution of clove oil (9:1, This method of calibrating EMG tags could yield biased clove oil: ethanol). Once stage 4 anesthesia was achieved, activity estimates if the variation in minimum and maximum the fish was measured for total length and mass, and then EMG values among individuals is the result of different min- transferred to a wetted foam surgical table. The gills of the bass imum and maximum swim speeds (i.e., behavioral differences) were infused with aerated water containing light anesthetic rather than random differences in tag sensitivity and electrode (20 mg/L solution of clove oil) for the duration of the surgery position. Such bias would be evident from consistently higher (4–5 min). The EMG transmitters (Lotek Wireless; model mean minimum or maximum EMG values in one experimental CEMG2-R11-18, 11 × 54 mm, 10 g in air, transmission rate = group relative to the other. Thus, to validate our use of this 30/min) were implanted through a ∼3-cm-long incision made calibration technique to compare routine activity between just left of the ventral midline of the fish, immediately anterior experimental groups, we used two-sample t-tests to compare to the pelvic girdle. The two gold-tipped electrodes (10 mm the mean minimum, maximum, and range of EMG values ANGLING AND LOCOMOTORY ACTIVITY IN LARGEMOUTH BASS 1255
10 be reliable in terms of comparing among-individual differences A in routine locomotory activity level. Experimental details.—Following implantation and recov- 8 ery of all fish used in a given trial (mid to late afternoon on the day of tagging), the group was transferred to a 4,000-m2 6 experimental pond. A total of two 7-d trials were run in fall 2009. Trial 1 ran from 23 September 2009 to 30 September 2009, and trial 2 ran from 2 October 2009 to 9 October 2009. 4 Water temperatures within the experimental pond ranged from 19◦Cto22◦C during trial 1, and from 14◦Cto18◦C during trial Minimum EMG 2 2. In both trials, naturally occurring aquatic invertebrates and stocked free-swimming fathead minnows Pimephales promelas and juvenile bluegill Lepomis macrochirus acted as a forage 0 base for the bass. The large number of forage fish remaining 50 at the end of the study period when the ponds were drained B suggests that food availability was not limited. A telemetry receiver (Lotek SRX-400 series) and three- 40 element Yagi antenna were placed at one end of the pond to record the transmitted EMG signals. The receiver cycled 30 through the unique tag frequencies, logging tag transmissions for 30 s at each frequency. Given the sample size of 14–15 fish in each trial, this meant that the activity level of each fish 20 was monitored for 30 s about 8 times/h for the duration of the trial. One exception occurred during trial 1: on the night of Maximum EMG 10 26 September 2009, there was a 6-h power interruption that resulted in a loss of records during that period. After each trial, the experimental pond was drained, and 0 fish were collected and held temporarily in a small raceway 50 near the pond. Each fish individually underwent the calibration C procedure described above to determine the EMG value trans- mitted during maximum exertion (EMGmax). Following the 40 procedure, the fish were euthanized and the transmitters were recovered. 30 Statistical analyses.—Activity records were analyzed over a 4-d period following an acclimation period of 32–34 h. Activity records began at midnight the day after the fish were released 20 and ended at midnight the night the ponds began draining (draining started at ∼0100 hours). Two individuals from trial Range of EMG 10 1 and a third from trial 2 fell victim to predation by resident
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 great blue herons Ardea herodias during the study and were consequently eliminated from analysis. A fourth individual 0 from trial 2 was also eliminated from analysis because one of HV LV its electrodes became dislodged during the trial. Two-sample t-tests were used to compare the mean length Group of LV and HV fish and to test for differences in activity level FIGURE 1. (A) Minimum, (B) maximum, and (C) range of EMG values for between male and female bass in trial 2 (six males and six HV and LV largemouth bass. Bars display mean values ± SE. There was females; sex was not recorded in trial 1). Comparison of mean no significant difference in any of these parameters between HV and LV bass routine activity level between LV and HV fish and between (t-test: P > 0.451 for all comparisons). night (1800 to 0559 hours) and day (0600 to 1759 hours) was accomplished using a nested linear mixed model. The model between HV and LV fish. No evidence of bias was found among included treatment group (LV or HV), time of day (day or any of the variables (Figure 1; t-test: t = 0.698, 0.337, and night), and their interaction as fixed effects. Fish identification −0.139; P = 0.492, 0.739, and 0.451 for minimum, maximum, was nested within trial (random effects) to account for repeated and range, respectively). As a result, we assume our method to measures, and fish length was a covariate. 1256 BINDER ET AL.
All statistics were performed use JMP statistical software 8 (version 4.0.4; SAS Institute, Cary, North Carolina). P-values < A 0.05 were considered significant, and all means are reported as mean ± SE. 6
RESULTS 4 Low vulnerability bass were slightly larger, on average, than HV bass (346 versus 333 mm), but the difference was =− =
not significant (t-test: t 1.788, P 0.087). There was no % max swim speed relationship between activity level and fish length (nested linear 2 mixed model: F = 0.1418, P = 0.710), nor was there sufficient evidence to conclude a significant effect of sex on activity level (t-test: t =−1.931, P = 0.083). The latter result, however, may 0 8 be due to the fact that we only recorded sex for individuals in B trial 2 and, consequently, sample size was low for this test. The activity level of largemouth bass in this study was generally low, relative to their maximum swimming speed. 6 Mean activity levels ranged from 1.1% to 8.3% of maximum swim speed. The overall activity level of HV and LV bass did not differ significantly from one another. Mean activity levels 4 were 5.1 ± 0.5% (median = 4.8%, range = 2.7–8.3%) and 4.3 ± 0.5% (median = 4.0%, range = 1.1–8.0%) of maximum % max swim speed swim speed in HV and LV bass, respectively (Figure 2; nested 2 linear mixed model: F = 1.055, P = 0.316). Though activity occurred during both day and night, both groups displayed a significant diurnal activity pattern, activity 0 level during the day being significantly greater than at night 2300 h 0700 h 1500 h 2300 h (Figure 3; nested linear mixed model: F = 61.141, P < 0.001). Time of day Activity levels tended to peak by a couple of hours after dawn and were at their lowest by approximately 3 h after dusk. High- FIGURE 3. Diel activity pattern of (A) HV and (B) LV largemouth bass. Vertical bars display mean activity level (% of maximum swim speed) over vulnerability bass were 16% more active in the day than at night each hour of the day for the duration of the study. Dark horizontal bars identify nighttime hours, and light horizontal bars identify daytime hours. Both groups displayed a significant diurnal activity pattern (nested linear mixed model: 6 P < 0.001).
5 (mean difference = 0.75 ± 0.15% of maximum swim speed), and LV bass were 19% more active during the day than at night 4 (mean difference = 0.77 ± 0.16% of maximum swim speed). Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 3 DISCUSSION 2 Selection for vulnerability to angling in largemouth bass did not result in observed differences in their routine locomotory % max swim speed 1 activity in experimental ponds. We tested two possible explana- tory hypotheses related to routine activity: (1) that the difference in angling vulnerability between HV and LV bass is related to a 0 difference in the level of routine activity between the two groups, HV LV and (2) that the difference in vulnerability between HV and LV Group bass is related to a difference in the diel activity pattern expressed by the two groups. Neither hypothesis was supported by our re- FIGURE 2. Mean activity levels of HV and LV largemouth bass. Bars display sults. High-vulnerability and LV bass displayed equal levels of mean activity (% of maximum swim speed) ± SE. There was no significant difference in mean activity level between HV and LV bass (nested linear mixed activity and, while HV and LV bass did each display a significant model: P = 0.316). diel activity pattern, the pattern was the same for both groups. ANGLING AND LOCOMOTORY ACTIVITY IN LARGEMOUTH BASS 1257
It is possible that this negative result is due to a reversal behavior through the mechanism of varying hunger levels. If this of evolutionary change over the two generations (F4 and F5) is the case, then one could predict that differences in foraging ac- without selection for angling vulnerability. Indeed, genetically tivity would be greatest when feeding opportunities are limited. based reversal of fishing-induced evolution has been demon- In our study, the presence of numerous stocked prey items and strated in Atlantic silversides Menidia menidia (Conover et al. limited refuge habitat for prey likely meant that prey were read- 2009; Salinas et al., in press). The silversides were exposed to ily available to all bass and, consequently, HV bass may not have size-selective fishing for five generations and then monitored needed to maintain higher activity levels to meet their energetic for an additional five generations after selection was halted, requirements. In addition, swimming could have been restricted at which time there was already evidence of evolutionary in the small ponds or the ponds might not have been large enough changes being reversed (i.e., fish body size was increasing). to induce a need to search for food, thereby constraining possi- The authors predicted full recovery from fisheries selection in bilities for activity differences to be expressed. Therefore, while approximately 12 generations. We believe complete reversal of there appears not to be inherent differences in locomotor activity evolutionary change is unlikely in our study for two reasons. between HV and LV bass, it would be necessary to compare the First, as far as we are aware, there are no extrinsic selection activity levels of HV and LV fish in a more competitive system factors in the experimental ponds that would cause a rapid before we can rule out the possibility of an indirect relationship shift in activity level. The study by Conover et al. (2009) did between locomotor activity and vulnerability to angling. demonstrate that phenotypic traits can reverse in the absence of High-vulnerability and LV bass both displayed diurnal extrinsic selection factors; however, the reversal was relatively activity patterns, activity levels during the day being 16–19% slow (only partial reversal over five generations), so complete higher than during the night. These results are consistent with reversal of phenotypic differences in locomotor activity over several previous studies of diel activity in largemouth bass, just two generations seems unlikely. Second, other phenotypic although the magnitude of the difference between daytime and behavioral differences between these two lines in terms of nighttime activity levels was more subtle in this study than in aggression and nest defense were evident in F4 generation fish some others (Warden and Lorio 1975; Reynolds and Casterlin (Cooke et al. 2007; Nannini et al. 2011) and were also present in 1976). This result can be attributed to seasonal differences F5 fish (Sutter 2010). We therefore conclude that vulnerability in the expression of diel activity in largemouth bass. Several to angling is, in fact, not related to inherent differences in independent studies have reported seasonality with respect routine locomotor activity in largemouth bass held in ponds. to diel activity in largemouth bass. For example, Demers Mean activity level in this study was approximately 5% of et al. (1996) found that the elevation of daytime activity in maximum (burst) swim speed in both HV and LV bass. Maxi- largemouth bass was strongest between July and September, mum burst speeds of approximately 4 body lengths/s have been and was reduced in October as water temperature and day observed in free-swimming, similarly sized smallmouth bass length decreased. Similarly, Hanson et al. (2007) observed a M. dolomieu (Peake and Farrell 2004). If we assume a similar clear diurnal activity pattern in free-swimming largemouth bass maximum burst speed for the largemouth bass in this study, then in April, but not in either January or November. estimated mean swimming speed was approximately 6.8 cm/s Our approach with respect to the implementation of EMG and the estimated mean distance traveled each day was approxi- telemetry to study the activity of free-swimming largemouth mately 5,875 m (range = 1,280–9,670 m). These values are con- bass produced two limitations that, if addressed in future sistent with values obtained from free-swimming largemouth studies, would allow researchers to take greater advantage of bass in Warner Lake, near Kingston, Ontario (Hanson et al. the full benefits of this technology. First, calibrating EMG 2007). Calculated daily swimming distances for those fish were tags on an absolute scale (such as would be done in a swim
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 7,300 m in April and decreased to 2,700 m by mid-November. tunnel; e.g., Brown et al. 2007) rather than a relative scale (as Our swimming distance estimate may seem somewhat high was done in this study) would increase precision and allow for relative to the fall value for Warner Lake bass (Hanson et al. construction of accurate bioenergetics models for comparing 2007), but it is likely that the slight elevation is a result of the groups of fish. Second, by monitoring the activity of bass in warmer water temperatures at our more-southern field site. groups, the swimming behavior of individual fish was recorded The lack of difference in activity level between HV and LV for only about 4 min/h. Censored behavior of individuals bass is somewhat surprising given the 10% higher SMR and 16% may have limited our ability to detect subtle differences in higher metabolic scope for activity previously observed in the activity, although this is a limitation with most other telemetry HV bass, relative to their LV counterparts (Redpath et al. 2010). techniques as well. Nonetheless, our approach did have two Higher metabolic demand in HV bass can come at a cost in terms obvious benefits over traditional position telemetry. First, EMG of growth if food is limited (Redpath et al. 2009) but, contrary telemetry can provide more-accurate instantaneous estimates to what one might predict, did not appear to support higher ac- of movement rates than positional telemetry because it is tivity levels (i.e., foraging activity) in our system. Nannini et al. not based on the assumption of linear movement between (2011) hypothesized that the differences in energetic require- two points over time (Cooke et al. 2001). Second, activity ments between HV and LVbass influence differences in foraging estimation using positional telemetry requires either that there 1258 BINDER ET AL.
are several listening stations present, often at a prohibitively REFERENCES inflated financial cost, or that the fish be followed for some time Arnason,´ E., U. B. Hernandez, and K. Kristinsson. 2009. Intense habitat-specific period, which can alter the behavior of the fish and may bias the fisheries-induced selection at the molecular Pan I locus predicts imminent results. collapse of a major cod fishery. PLoS (Public Library of Science) ONE [online serial] 4(5):article e5529. DOI: 10.1371/journal.pone.0005529. In conclusion, the behavioral basis of artificial selection Askey, P. J., S. A. Richards, J. R. Post, and E. A. Parkinson. 2006. Linking for angling vulnerability in largemouth bass seem unrelated angling catch rates and fish learning under catch-and-release regulations. to inherent differences in routine locomotory activity. By North American Journal of Fisheries Management 26:1020–1029. contrast, angling vulnerability in this species is clearly related Beddow, T. A., and R. S. McKinley. 1999. Importance of electrode positioning to metabolism and aggression (Suski and Philipp 2004; Cooke in biotelemetry studies estimating muscle activity in fish. Journal of Fish Biology 54:819–831. et al. 2007; Redpath et al. 2010) and may be influenced in part Beukema, J. J. 1970. Acquired hook-avoidance in the pike Esox lucius L. fished by differences in foraging behavior (Nannini et al. 2011). One with artificial and natural baits. Journal of Fish Biology 2:155–160. possibility that has not yet been tested, however, is that the Biro, P. A., and J. R. Post. 2008. Rapid depletion of genotypes with fast growth difference in vulnerability to angling is a result of differences and bold personality traits from harvested fish populations. Proceedings of in the spatial ecology (e.g., home range size and preferred the National Academy of Sciences of the USA 105:2919–2922. Brown, R. S., C. P. Tatara, J. R. Stephenson, and B. A. Berejikian. 2007. habitat) of HV and LV bass. Anglers rarely fish all available Evaluation of a new coded electromyogram transmitter for studying swim- habitats but rather target specific habitats that are traditionally ming behavior and energetics in fish. North American Journal of Fisheries known for high angling success. As a result, by selectively Management 27:765–772. targeting specific habitats, anglers might inadvertently select Conover, D. O., S. B. Munch, and S. A. Arnott. 2009. Reversal of evolutionary for habitat-related behavioral phenotypes (Arnason´ et al. 2009; downsizing caused by selective harvest of large fish. Proceedings of the Royal Society of London B 276:2015–2020. Jakobsdottir´ et al. 2011; Parsons et al. 2011). Indeed, in a study Cooke, S. J., C. M. Bunt, J. F. Schreer, and D. H. Wahl. 2001. Comparison of of the relationship between boldness and angling vulnerability several techniques for mobility and activity estimates of smallmouth bass in in bluegill, Wilson et al. (2011) found that bolder sunfish (as lentic environments. Journal of Fish Biology 58:573–587. determined by a standardized refuge emergence test) were more Cooke, S. J., B. D. S. Graeb, C. D. Suski, and K. G. Ostrand. 2003. Effects of su- likely to approach a baited hook from open water, whereas shy ture material on incision healing, growth and survival of juvenile largemouth bass implanted with miniature radio transmitters: case study of a novice and sunfish tended to approach the hook from a refuge. If differ- experienced fish surgeon. Journal of Fish Biology 62:1366–1380. ences in spatial ecology and habitat use were to exist between Cooke, S. J., C. D. Suski, K. G. Ostrand, D. H. Wahl, and D. P. Philipp. 2007. HV and LV selected bass, these differences could also account Physiological and behavioral consequences of long-term artificial selection for diverging foraging tactics and energetic constraints (Savino for vulnerability to recreational angling in a teleost fish. Physiological and and Stein 1989), both of which have previously been observed Biochemical Zoology 80:480–490. Cooke, S. J., E. B. Thorstad, and S. G. Hinch. 2004. Activity and energetics between these two lines (Cooke et al. 2007; Redpath et al. of free-swimming fish: insights from electromyogram telemetry. Fish and 2010; Nannini et al. 2011). Research into the spatial ecology of Fisheries 5:21–52. the two lines of bass, therefore, represents a logical next step in Demers, E., R. S. McKinley, A. H. Weatherley, and D. J. McQueen. 1996. discovering the behavioral mechanisms on which this selection Activity patterns of largemouth and smallmouth bass determined with elec- regime acts. tromyogram biotelemetry. Transactions of the American Fisheries Society 125:434–439. Enberg, K., C. Jørgensen, E. S. Dunlop, M. Heino, and U. Dieckmann. 2009. Implications of fisheries-induced evolution for stock rebuilding and recovery. Evolutionary Applications 2:394–414. ACKNOWLEDGMENTS Geist, D. R., R. S. Brown, K. Lepla, and J. Chandler. 2002. Practical application We thank the staff of the Sam Parr Biological Station (E. of electromyogram radiotelemetry: the suitability of applying laboratory- Giebelstein and M. Porto) for their assistance in collecting, acquired calibration data to field data. North American Journal of Fisheries Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 maintaining, and tagging the bass during the study. We also Management 22:474–479. Hanson, K. C., S. J. Cooke, C. D. Suski, G. Niezgoda, F. J. S. Phelan, acknowledge T. Redpath for her early efforts to study the lo- R. Tinline, and D. P. Philipp. 2007. Assessment of largemouth bass comotory activity of bass. Funding support was provided by (Micropterus salmoides) behaviour and activity at multiple spatial and tempo- the Canada Foundation for Innovation, the Ontario Ministry of ral scales utilizing a whole-lake telemetry array. Hydrobiologia 582:243–256. Research and Innovation, the Natural Sciences and Engineering Heino, M., and O. R. Godø. 2002. Fisheries-induced selection pressures Research Council of Canada, the Illinois Natural History Survey in the context of sustainable fisheries. Bulletin of Marine Science 70: 639–656. and the University of Illinois, the Gottfried-Wilhelm-Leibniz- Huntingford, F. A., G. Andrew, S. Mackenzie, D. Morera, S. M. Coyle, Association in the Adaptfish Project, the German Federal Min- M. Pilarczyk, and S. Kadri. 2010. Coping strategies in a strongly schooling istry for Education and Research, Program for Social-Ecological fish, the common carp Cyprinus carpio. Journal of Fish Biology 76:1576– Research in the Project Besatzfisch (grant 01UU0907), and 1591. the Deutsche Bundestiftung Umwelt. The procedures used in Hutchings, J. A., and D. J. Fraser. 2008. The nature of fisheries- and farming- induced evolution. Molecular Ecology 17:294–313. this study were conducted in accordance with the policies of Jakobsdottir,´ K. B., H. Pardoe, A.´ Magnusson,´ H. Bjornsson,¨ C. Pampoulie, the Canadian Council on Animal Care, as administered by the D. E. Ruzzante, and G. Marteinsdottir.´ 2011. Historical changes in geno- Carleton University Animal Care Committee (AUP B07-04). typic frequencies at the Pantophysin locus in Atlantic cod (Gadus morhua) ANGLING AND LOCOMOTORY ACTIVITY IN LARGEMOUTH BASS 1259
in Icelandic waters: evidence of fisheries-induced selection? Evolutionary Redpath, T. D., S. J. Cooke, C. D. Suski, R. Arlinghaus, P. Couture, D. H. Wahl, Applications 4:562–573. and D. P. Philipp. 2010. The metabolic and biochemical basis of vulnerability Jørgensen, C., K. Enberg, E. S. Dunlop, R. Arlinghaus, D. S. Boukal, K. Brander, to recreational angling after three generations of angling-induced selection in B. Ernande, A. G. Gardmark,˚ F. Johnston, S. Matsumura, H. Pardoe, K. Raab, a teleost fish. Canadian Journal of Fisheries and Aquatic Sciences 67:1983– A. Silva, A. Vainikka, U. Dieckmann, M. Heino, and A. D. Rijnsdorp. 2007. 1992. Ecology: managing evolving fish stocks. Science 318:1247–1248. Reynolds, W. W., and M. E. Casterlin. 1976. Activity rhythms and light intensity Kuparinen, A., and J. Merila.¨ 2007. Detecting and managing fisheries-induced preferences of Micropterus salmoides and M. dolomieui. Transactions of the evolution. Trends in Ecology and Evolution 22:652–659. American Fisheries Society 105:400–403. Law, R. 2000. Fishing, selection, and phenotypic evolution. ICES Journal of Ros, A. F. H., K. Becker, and R. F. Oliveira. 2006. Aggressive behaviour and Marine Science 57:659–668. energy metabolism in a cichlid fish, Oreochromis mossambicus. Physiology Lewin, W. C., R. Arlinghaus, and T. Mehner. 2006. Documented and poten- and Behavior 89:164–170. tial biological impacts of recreational fishing: insights for management and Salinas, S., K. O. Perez, T. A. Duffy, S. J. Sabatino, L. A. Hice, S. B. conservation. Reviews in Fisheries Science 14:305–367. Munch, and D. O. Conover. In press. The response of correlated traits fol- Metcalfe, N. B., A. C. Taylor, and J. E. Thorpe. 1995. Metabolic rate, so- lowing cessation of fishery-induced selection. Evolutionary Applications. cial status and life-history strategies in Atlantic salmon. Animal Behaviour DOI:10.1111/j.1752-4571.2012.00243.x. 49:431–436. Savino, J. F., and R. A. Stein. 1989. Behavior of fish predators and their prey: Nannini, M. A., D. H. Wahl, D. P. Philipp, and S. J. Cooke. 2011. The influence habitat choice between open water and dense vegetation. Environmental Bi- of selection for vulnerability to angling on foraging ecology in largemouth ology of Fishes 24:287–293. bass Micropterus salmoides. Journal of Fish Biology 79:1017–1028. Suski, C. D., and D. P. Philipp. 2004. Factors affecting the vulnerability to Parsons, D. M., M. A. Morrison, J. R. McKenzie, B. W. Hartill, R. Bian, and angling of nesting male largemouth and smallmouth bass. Transactions of the C. Francis. 2011. A fisheries perspective of behavioural variability: differ- American Fisheries Society 133:1100–1106. ences in movement behaviour and extraction rate of an exploited sparid, Sutter, D. A. H. 2010. Impact of angling induced selection on aggression, snapper (Pagrus auratus). Canadian Journal of Fisheries and Aquatic Sci- nest guarding behavior and reproductive success of male largemouth bass ences 68:632–642. (Micropterus salmoides). Master’s thesis. Humboldt-Universitat¨ zu Berlin, Peake, S. J., and A. P. Farrell. 2004. Locomotory behaviour and post-exercise Berlin. physiology in relation to swimming speed, gait transition and metabolism in Theriault,´ V., E. S. Dunlop, U. Dieckmann, L. Bernatchez, and J. J. Dodson. free-swimming smallmouth bass (Micropterus dolomieu). Journal of Experi- 2008. The impact of fishing-induced mortality on the evolution of alternative mental Biology 207:1563–1575. life-history tactics in brook charr. Evolutionary Applications 1:409–423. Philipp, D. P., S. J. Cooke, J. E. Claussen, J. B. Koppelman, C. D. Suski, and Uusi-Heikkila,¨ S., C. Wolter, T. Klefoth, and R. Arlinghaus. 2008. A behavioral D. P. Burkett. 2009. Selection for vulnerability to angling in largemouth bass. perspective on fishing-induced evolution. Trends in Ecology and Evolution Transactions of the American Fisheries Society 138:189–199. 23:419–421. Policansky, D. 1993. Fishing as a cause of evolution in fishes. Pages 2–18 in Walsh, M. R., S. B. Munch, S. Chiba, and D. O. Conover. 2006. Maladap- T. K. Stokes, J. M. McGlade, and R. Law, editors. The exploitation of evolving tive changes in multiple traits caused by fishing: impediments to population resources. Springer-Verlag, Heidelberg, Germany. recovery. Ecology Letters 9:142–148. Raat, A. J. P. 1985. Analysis of angling vulnerability of common carp, Cyprinus Warden, R. L., Jr., and W. J. Lorio. 1975. Movements of largemouth bass carpio L., in catch-and-release angling ponds. Aquaculture Research 16: (Micropterus salmoides) in impounded waters as determined by underwater 171–187. telemetry. Transactions of the American Fisheries Society 104:696–702. Redpath, T. D., S. J. Cooke, R. Arlinghaus, D. H. Wahl, and D. P. Philipp. Wilson, A. D. M., T. R. Binder, K. P. McGrath, S. J. Cooke, and J. G. J. Godin. 2009. Life-history traits and energetic status in relation to vulnerability to 2011. Capture technique and fish personality: angling targets timid bluegill angling in an experimentally selected teleost fish. Evolutionary Applications sunfish, Lepomis macrochirus. Canadian Journal of Fisheries and Aquatic 2:312–323. Sciences 68:749–757. Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:04 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
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Survival and Growth of Juvenile Pacific Lampreys Tagged with Passive Integrated Transponders (PIT) in Freshwater and Seawater Matthew G. Mesa a , Elizabeth S. Copeland a , Helena E. Christiansen a , Jacob L. Gregg b , Sean R. Roon b & Paul K. Hershberger b a U.S. Geological Survey, Western Fisheries Research Center, Columbia River Research Laboratory, 5501 Cook-Underwood Road, Cook, Washington, 98605, USA b U.S. Geological Survey, Western Fisheries Research Center, Marrowstone Marine Field Station, 616 Marrowstone Point Road, Nordland, Washington, 98358, USA Version of record first published: 30 Jul 2012.
To cite this article: Matthew G. Mesa, Elizabeth S. Copeland, Helena E. Christiansen, Jacob L. Gregg, Sean R. Roon & Paul K. Hershberger (2012): Survival and Growth of Juvenile Pacific Lampreys Tagged with Passive Integrated Transponders (PIT) in Freshwater and Seawater, Transactions of the American Fisheries Society, 141:5, 1260-1268 To link to this article: http://dx.doi.org/10.1080/00028487.2012.686951
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. Transactions of the American Fisheries Society 141:1260–1268, 2012 American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.686951
ARTICLE
Survival and Growth of Juvenile Pacific Lampreys Tagged with Passive Integrated Transponders (PIT) in Freshwater and Seawater
Matthew G. Mesa,* Elizabeth S. Copeland, and Helena E. Christiansen U.S. Geological Survey, Western Fisheries Research Center, Columbia River Research Laboratory, 5501 Cook-Underwood Road, Cook, Washington 98605, USA Jacob L. Gregg, Sean R. Roon, and Paul K. Hershberger U.S. Geological Survey, Western Fisheries Research Center, Marrowstone Marine Field Station, 616 Marrowstone Point Road, Nordland, Washington 98358, USA
Abstract Tagging methods are needed for both adult and juvenile life stages of Pacific lampreys Lampetra tridentata to better understand their biology and factors contributing to their decline. We developed a safe and efficient technique for tagging juvenile Pacific lampreys with passive integrated transponder (PIT) tags. We tested the short-term survival of PIT-tagged juvenile lampreys in freshwater at four temperatures (9, 12, 15, and 18◦C) and their long-term growth and survival in seawater. For both experiments there was little to no tag loss, and juvenile lampreys in freshwater showed high survival at all temperatures at 7 d (95–100%) and 14 d (88–100%) posttagging. Prolonged holding (40 d) resulted in significantly lower survival (28–79%) at warmer temperatures (12–18◦C). For juvenile lampreys tagged in freshwater and then transitioned to seawater, survival was 97% for tagged fish until day 94, and at the end of 6 months, survival was about 58% for both tagged and control fish. About half of the tagged and control fish that survived in seawater grew, but there was no difference in growth between the two groups. In freshwater, but not in seawater, most fish that died had an aquatic fungal infection. In both experiments, survival increased with increasing fish length at tagging. Our results indicate that tags similar in size to a 9-mm PIT tag are a feasible option for tagging metamorphosed juvenile lampreys migrating downstream and that when fungal infections are mitigated—as in seawater—long-term (at least 6 months) survival of tagged juvenile lampreys is high.
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 Passive integrated transponder (PIT) technology allows for the most important and useful marking technologies available the identification of individuals and provides information on to scientists today. habitat use, movements, growth, and survival of fishes. In Despite the many advantages of PIT tags for marking fish, salmonids, for example, PIT tags have been used to track the mi- they have rarely been used to study the biology of Pacific lam- grations of juveniles (Skalski 1998; Zabel et al. 2005; Williams preys Lampetra tridentata. This is unfortunate because popu- 2008), determine the effects of mitigation strategies on down- lations of these fish in the Columbia River basin have greatly stream passage rates (Marsh et al. 1999; Williams et al. 2001), declined (Close et al. 2002), and data from PIT tag studies and estimate travel time between dams (Zabel 2002), long-term could be useful for identifying problems with downstream pas- survival, and adult return rates (Zabel and Williams 2002; Keefer sage at dams and providing information about homing, ocean et al. 2008). There are many other examples of the use of PIT dispersal, growth rates, and population structure. Passive inte- tags in fisheries research and management, and they are one of grated transponder tags have been used in combination with
*Corresponding author: [email protected] Received October 31, 2011; accepted April 16, 2012
1260 PIT-TAGGED JUVENILE PACIFIC LAMPREYS 1261
radio transmitters to study the dam passage characteristics of per day. Water was either heated to 9◦C with an immersion adult lampreys (Moser et al. 2002), but they have not been ex- heater or to 12, 15, or 18◦C using single-pass electric heaters. tensively used with juvenile fish. Although the small size of PIT Heated water was passed through packed columns to dissipate tags enables the marking of very small fish (Baras et al. 1999; excess dissolved gases. Juvenile lampreys were acclimated for Archdeacon et al. 2009; Dixon and Mesa 2011), they have not 1 week at the test temperatures (two tanks per temperature) been used in juvenile lampreys because of the limited internal before starting the experiment. body cavity of these fish, their fragile nature, and lack of an We randomly selected a tank, removed all fish from it, and effective implantation technique. placed the fish in insulated coolers with aeration. Fish were Two earlier studies (Schreck et al. 1999; Mueller et al. 2006) anesthetized in groups of about four to eight in a solution of made some headway in developing PIT-tagging protocols for tricaine methanesulfonate (MS-222, 250 mg/L) buffered with metamorphosed juvenile Pacific lampreys (henceforth referred an equal amount of sodium bicarbonate. When fish started to to as juvenile lampreys) and identified some key issues to resolve become unresponsive, individual fish were removed from the before extensive tagging occurs in the field. The two studies anesthetic, weighed to the nearest 0.1 g, and measured to the used different tag sizes (9- versus 12-mm PIT tags) and differ- nearest 1 mm. Each fish was then placed on its right side under ent tagging procedures that required a dissecting microscope or a magnifying light in a groove cut into a moist closed-cell foam multiple needles and a PIT tag injector plus suturing. A simple, pad saturated with 150 µL/L of the water conditioner PolyAqua standardized protocol that could be used in the field is lacking. (Kordon, Hayward, California). We made a 2–3-mm-long inci- These studies were also done at different water temperatures, sion 20 mm posterior to the gill pores on the left lateral side and Mueller et al. (2006) suggested that higher water temper- with a 3.0-mm microsurgical scalpel (15◦ blade; AngioTech, atures lead to increased mortality of tagged fish. A systematic Vancouver, British Columbia), inserted a PIT tag (9 × 2 mm, study of the survival of PIT-tagged juvenile lampreys in wa- Biomark, Boise, Idaho) through the incision by hand, and guided ter temperatures typically found in the Columbia River during the tag anteriorly (Figure 1). Control fish were treated similarly their out-migration is needed. Finally, both studies indicated a to tagged fish but were not subjected to an incision or tag in- problem with fungal infections in juvenile lampreys leading to sertion. The entire tagging procedure (including weighing and mortality, but the cause of these infections was unclear. measuring) took about 45–60 s per fish. Although Schreck et al. (1999) and Mueller et al. (2006) After handling, juvenile lampreys were transferred to a made substantial contributions to the development of PIT- recovery bucket with aeration before being returned to their tagging techniques for juvenile lampreys, many questions must original tanks. In the end, we returned 55–60 control and 46–60 be answered before any tagging technology is implemented in tagged juvenile lampreys to each tank for a total of 101–120 the field. Here, we used a new, simple PIT-tagging procedure fish per tank. After tagging was complete, fish were held for 40 (Mesa et al. 2011) and tested the survival of PIT-tagged juvenile d, and each tank was provided a constant food supply of 9–11 Pacific lampreys held at four different temperatures. We also hatchery-reared yearling coho salmon Oncorhynchus kisutch. determined long-term growth and survival of PIT-tagged juve- We used coho salmon because they were readily available nile lampreys in seawater and confirmed that active swimming and worked well for us as food for juvenile lampreys in did not promote tag loss. Collectively, our research should previous experiments. We monitored the juvenile lampreys for provide information about temperature thresholds for tagging juvenile lampreys and their long-term survival in seawater.
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 METHODS Effect of temperature on survival of PIT-tagged juvenile Pacific lampreys in freshwater.—Outmigrating juvenile Pacific lampreys were collected from the smolt bypass facility at John Day Dam on the Columbia River during May–June 2010 and transported to the Columbia River Research Laboratory in Cook, Washington. Fish were placed into eight 1.2-m-diameter circu- lar tanks (140–160 fish per tank) that received water from the Little White Salmon River. Water was maintained at 12◦C un- der a simulated natural photoperiod with overhead incandescent lights. Fish were treated with 50 mg/L hydrogen peroxide (ac- tive compound) in static water for 30 min twice a week before tagging to inhibit fungal growth. FIGURE 1. PIT-tagging of juvenile Pacific lamprey, showing (A) insertion of a PIT tag by hand and (B) a PIT-tagged juvenile lamprey with the tag (black Juvenile lampreys were transitioned to four test temperatures arrow) anterior to the incision at 40 d posttagging. [Figure available online in ◦ ◦ (9, 12, 15, and 18 C) by adjusting the water temperature 1–2.5 C color.] 1262 MESA ET AL.
mortality and scanned the bottom of the tanks for shed PIT were removed from the tank, scanned for a PIT tag, and weighed tags 5 d per week. Dead juvenile lampreys were removed and measured. After 180 d, we euthanized all the fish, identified from the tank, weighed and measured, scanned for a PIT tag, them to treatment group, and weighed and measured them. and the date and presence of fungal infection were noted. If We again used Kaplan–Meier curves to describe the survival tagged, the tag number, condition of the incision, and general of juvenile lampreys in each treatment, pooling the data from location of the tag inside the body cavity were also recorded. replicate tanks for analysis. We compared curves between the Juvenile lampreys were not treated with fungal inhibitors during treatments using the log-rank (Mantel–Cox) test and then used the experiment. After 40 d, all surviving juvenile lampreys logistic regression to determine the effect of initial size on sur- were weighed, measured, and scanned for tags. For tagged fish, vival of tagged fish. To analyze growth, we compared the mean incision healing and tag location were recorded. initial and final sizes between and within each group using a We compared the initial size of control and tagged juve- one-way ANOVA. To calculate the percent change in mass for nile lampreys within a temperature treatment using two-sample individual PIT-tagged fish, we divided the final mass of each t-tests with Welch’s correction and used a one-way ANOVA fish by its initial mass. to determine whether the initial size of control or tagged fish Effect of swimming on tag loss in PIT-tagged juvenile Pacific differed among temperatures. Kaplan–Meier curves were used lampreys.—Juvenile Pacific lampreys were collected from the to describe the survival of tagged and untagged fish within a smolt bypass facility at the John Day Dam in May 2011 and temperature at 7, 14, and 40 d, and fish in replicate tanks were transported to the Columbia River Research Laboratory. They pooled for analysis. We compared curves among treatments us- were maintained at 12◦C as described above for lampreys col- ing the log-rank (Mantel–Cox) test, and analyses were done lected in 2010 and held in freshwater. We treated them twice using GraphPad Prism software (GraphPad Software, La Jolla, weekly with hydrogen peroxide beginning about 3 weeks after California). To determine whether size of fish at the time of tag- transfer owing to increased incidence of fungal growth. ging influenced survival, we used logistic regression analysis The swim trials were performed in a 4 × 1.2-m oval tank and tested the significance of the coefficients with a likelihood with an adjustable flow-inducing propeller, and the water depth ratio test (LRT; R Software, R Development Core Team 2011). was 32 cm. Four groups of five or six fish were weighed, mea- We could not do this for control fish because they could not be sured, PIT-tagged as described above, and allowed to recover for individually identified. The level of significance for all tests was 20–24 h in a perforated 19-L bucket suspended in the swim tank. 0.05. A 20–24-h recovery period was chosen because juvenile lam- Long-term survival of PIT-tagged juvenile Pacific lampreys preys would not swim actively when tested 30 min after surgery, in seawater.—Juvenile Pacific lampreys (n = 120) were tagged the longer recovery period allowed initiation of wound healing, shortly after capture at the smolt bypass facility at the John Day and others at our laboratory have used similar procedures for Dam in April 2010. They were removed from a holding raceway recovery of juvenile salmonids after surgical implantation of ra- in groups of three to four, weighed and measured, and tagged dio transmitters (Adams et al. 2001; Beeman et al. 2007). After as described previously. An equal number of control fish were holding, juvenile lampreys were gently poured from the bucket handled, weighed, and measured but not tagged. After handling, into a straight 1.31-m section of the tank with plastic mesh juvenile lampreys were transferred to a recovery bucket with (hole size, 6.4 × 6.4 mm) on either end to protect them from aeration for about 20 min before being moved to a flow-through the propeller. The bottom and sides of the tank were also cov- tank with unheated river water (11◦C). ered with plastic mesh (hole size, 3 × 4.4 mm) to prevent the After 7 d, the juvenile lampreys were placed in an insu- fish from attaching. We swam fish at a water velocity of 15 cm/s lated tank with supplemental oxygen and transported to the U.S. for 3 h. This velocity was chosen because it was about half
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 Geological Survey’s Marrowstone Marine Field Station in Nord- the 5-min sustained swimming speed (37 cm/s) of PIT-tagged land, Washington (about an 8-h drive). Upon arrival, each fish juvenile lampreys (Mueller et al. 2006). Juvenile lampreys that was carefully netted, scanned for a tag, and arbitrarily assigned rested against the downstream mesh were forced to swim by to one of two 2.4-m-diameter tanks filled with static freshwater gentle nudging with a net. After a trial, we transferred the fish (12◦C). We placed similar numbers of control and tagged fish to a 19-L aquarium for 4–14 d. Fish were monitored for tag loss in each tank. For 72 h, ambient seawater was added at a rate of during and after the swim trials. 0.8 L/min and then increased to 1.5 L/min. Juvenile lampreys were in 100% seawater (29–31‰) within 96 h of transport. Dur- RESULTS ing the 180-d holding period, fish were fed with a constant sup- ply of 4–7 adult rainbow trout O. mykiss in each tank. The trout Effect of Temperature on Survival of PIT-tagged Juvenile were originally from Clear Springs Trout Farm, Buhl, Idaho, and Pacific Lampreys in Freshwater weighed about 1 kg each. We used them because they were sur- During the experiment, temperature fluctuations were within plus from previous experiments and readily available. We mon- the following ranges for the four groups: 9 ± 1.8◦C, 12 ± itored the juvenile lampreys daily for mortality and scanned the 1.1◦C, 15 ± 1.1◦C, and 18 ± 0.3◦C. No tags were shed, and bottom of each tank for shed PIT tags. Dead juvenile lampreys only 2.2% of incisions had an abnormality (i.e., protruding PIT-TAGGED JUVENILE PACIFIC LAMPREYS 1263
TABLE 1. Mean initial length and mass of Pacific lampreys at four test temperatures. Asterisks denote a significant difference in mean values between control and tagged fish within a temperature as determined by a two-sample t-test. Means for control or tagged fish in a column without a letter in common are significantly different (control and tagged fish were analyzed separately using one-way ANOVA).
Total length (mm) Mass (g) Test temperature Treatment n Mean (SD) Range Mean (SD) Range 9◦C Control 116 132.5 (7.3) z 123–156 3.2 (0.6) z 2.2–5.3 Tagged 115 135.6 (8.5)* z 125–163 3.3 (0.7) z 2.3–6.9 12◦C Control 110 137.7 (8.8) y 124–158 3.5 (0.7) y 2.1–5.7 Tagged 116 140.2 (9.6)* y 125–168 3.6 (0.8) y 2.4–6.3 15◦C Control 119 134.4 (7.7) z 123–155 3.3 (0.6) zy 2.4–5.3 Tagged 119 135.9 (8.2) z 125–160 3.3 (0.6) z 2.3–4.9 18◦C Control 110 135.1 (9.5) zy 120–181 3.3 (0.9) z 2.2–9.3 Tagged 104 134.2 (7.8) z 124–160 3.1 (0.6) z 2.3–5.8
viscera, signs of infection, or a poorly healed incision). For test: P < 0.05). At 40 d, survival curves of control fish differed surviving juvenile Pacific lampreys at day 40, 70% of tags significantly between all pairwise comparisons at different remained anterior to the incision, 12% were posterior to the temperatures (log-rank tests: P < 0.05), except for 15◦C and incision, and 18% were located at the incision. The initial 18◦C. Survival of fish was higher at 9◦C than at any other lengths and masses of fish used in the four temperature temperature and lower than at any other temperature at 12◦C treatments differed slightly, depending on group (Table 1). At (Figure 2). The same results also occurred in tagged fish. At all 9◦C and 12◦C, the mean initial lengths, but not weights, of temperatures, most of the fish that died had an aquatic fungal tagged fish were significantly greater than those of control fish infection. Survival of tagged fish to 40 d posttagging was (9◦C: t = 2.89, df = 223, P = 0.004; 12◦C: t = 2.05, df = 236, significantly influenced by fish length at time of tagging (LRT: P = 0.04; Table 1). There were no differences in initial length or χ2 = 8.63, df = 1, P = 0.003), and survival increased directly mass between tagged and control fish at any other temperature. with fish length (Figure 3). Within control groups, the mean initial length of fish at 12◦C was significantly greater than values of fish at 9◦C and 15◦Cbut Long-term Survival of PIT-tagged Juvenile Pacific not 18◦C(F = 7.80; df = 3, 467; P < 0.0001); the mean initial Lampreys in Seawater mass of fish at 12◦C was greater than values of fish at 9◦C and During the 7 d of holding after tagging juvenile Pacific 18◦C but not 15◦C(F = 4.46; df = 3, 467; P = 0.004; Table 1). lampreys at the smolt bypass facility at the John Day Dam, For tagged groups, fish at 12◦C were significantly larger than six tags were shed within the first 4 d. Thereafter, no tags were fish at other temperatures (length: F = 10.55; df = 3, 457; P < shed. During the first week in seawater, three fish died. No 0.0001; mass: F = 9.09; df = 3, 456; P < 0.0001; Table 1). further mortality occurred until day 94 (1 August). At the end Mean survival of juvenile Pacific lampreys after 7 d of 6 months, survival was about 58% for each group and did (95–100%) and 14 d (88–100%) was high for all groups, and not differ significantly (log-rank test: χ2 = 0.0007, df = 1, there was no significant difference in survival between tagged P = 0.9796; Figure 4). Fish length at tagging significantly Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 and control fish (Figure 2). Prolonged laboratory holding (40 d) influenced survival (LRT: χ2 = 4.69, df = 1, P = 0.03), and in- resulted in decreased survival of both control and tagged fish at creased length improved survival (Figure 4). Juvenile lampreys warm temperatures (12–18◦C) and a high incidence of fungal in- were first observed feeding on day 23 (22 May), and feeding fections (Figure 2). At 40 d, survival curves between tagged and activity generally increased over time. At the time we ended the control fish were significantly different only at 15◦C (log-rank experiment (25 October), about 12–15 fish were feeding in each test: χ2 = 6.42, df = 1, P = 0.01; Figure 2) because one replicate tank. The final mean ± SD length and mass of control (160.1 ± of control fish had about 20% lower survival than the other 26.5 mm, 6.24 ± 4.07 g) and tagged (157.5 ± 29.5 mm, 6.37 ± group. Survival curves of control fish at 7 d differed significantly 6.04 g) fish were significantly greater than values of control only between 12◦C and all other temperatures (log-rank tests: (134.5 ± 8.4 mm, 3.49 ± 0.68 g) and tagged (134.9 ± P < 0.05). Survival curves of tagged fish at 7 d only differed 7.0 mm, 3.48 ± 0.61 g) fish at the start of the experiment between 9◦C and 12◦C (log-rank test: P < 0.05). At 14 d, (ANOVA: length: F = 53.21; df = 3, 370; P < 0.0001; mass: survival curves of both control and tagged fish did not differ F = 23.93; df = 3, 370; P < 0.0001). There were no significant between 9◦C and 15◦C or between 12◦C and 18◦C. All other differences between sizes of tagged and untagged lampreys at pairwise comparisons were significantly different (log-rank the end of the experiment (Figure 5). About 59% of PIT-tagged 1264 MESA ET AL.
FIGURE 2. Survival of untagged control and PIT-tagged juvenile Pacific lampreys reared in freshwater at four test temperatures. Dashed vertical lines indicate 7- and 14-d time points.
fish that survived over 180 d appeared to have fed since they increased in mass from 1% to 1,149%. The remaining 41% of fish lost 2–40% of their body weight and apparently did not feed.
Effect of Swimming on Tag Loss in PIT-tagged Juvenile Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 Pacific Lampreys We tested 21 juvenile lampreys with a mean ± SD length and mass of 151 ± 7 mm and 4.6 ± 0.8 g. All fish tested swam for the full test period (3 h), though most required repeated prompting to swim continuously. No tags were lost during or after the trials, and incisions healed normally.
DISCUSSION In this study, we used a quick, safe, and effective technique for PIT-tagging juvenile Pacific lampreys that we believe im- proved upon methods previously described by Schreck et al. (1999), Quintella et al. (2005), and Mueller et al. (2006). Our FIGURE 3. Survival based on initial length at tagging of PIT-tagged juvenile technique required no sutures and resulted in very high tag re- Pacific lampreys reared in freshwater at four test temperatures for 40 d. tention, no detectable effect on survival, and identification of PIT-TAGGED JUVENILE PACIFIC LAMPREYS 1265
FIGURE 4. Survival of juvenile Pacific lampreys reared in seawater for 180 d. (A) Survival of untagged control and PIT-tagged juvenile lampreys. (B) Survival of PIT-tagged juvenile lampreys based on initial length at tagging.
possible size thresholds for tagging. The method we used for implanting juvenile lampreys with PIT tags used an incision FIGURE 5. Box plots of the initial length and mass of untagged (control) and lateral to the mid-ventral line because preliminary work at our PIT-tagged juvenile Pacific lampreys reared in seawater for 180 d. The boxes represent the 25th and 75th quartiles. The whiskers represent the 5th and 95th Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 laboratory indicated that this technique resulted in less evis- ceration of the digestive tract through the wound edges than percentiles, and the line is the median. Outliers are shown as solid dots. Groups without letters in common are statistically different. an incision directly on the mid-ventral line (Mesa et al. 2011). We also used 3.0-mm constant depth microsurgical blades that facilitated a consistent, easy, and safe surgery. We agree with per fish) but required three different needles—one to puncture a Schreck et al. (1999) that magnification is useful for this type hole in the body cavity, another to enlarge the hole, and a final of surgery. They used a dissecting microscope whereas we used needle to inject the PIT tag into the body cavity. These proce- a magnifying light, both of which could be adapted for field dures resulted in high tag retention and minimal tag effects, but situations. The procedures described by Quintella et al. (2005) we feel they lack the simplicity required for large-scale field and Mueller et al. (2006) seemed too complicated for use in field operations. situations. Quintella et al. (2005) implanted PIT tags into young During preliminary work (Mesa et al. 2011) and this study, sea lampreys Petromyzon marinus with a procedure that used a we implanted PIT tags into over 700 juvenile Pacific lampreys, 3-mm incision, a single suture, and a parafilm “bandage” that and only 6 tags were shed. Our range of tag loss (0–5%) was fixed with cyanocrylate glue and took 3–5 min to complete. was similar to values from other studies of young lampreys Mueller et al. (2006) used a procedure that was quicker (30–60 s (Quintella et al. 2005; Mueller et al. 2006) and from a variety 1266 MESA ET AL.
of teleost fishes, including juvenile bullhead Cottus gobio tra prophylactic treatments could have more severely damaged (Bruyndoncx et al. 2002), Chinook salmon O. tshawytscha the skin and mucous layer in fish at 12◦C and left them more (Knudsen et al. 2009), and the Rio Grande silvery minnow vulnerable to other stressors. Second, the juvenile lampreys at Hybognathus amarus (Archdeacon et al. 2009). We also showed 12◦C were larger on average than fish at any other temperature, that active swimming 20–24 h posttagging did not result in tag suggesting that they were older and perhaps undergoing phys- loss even though the incisions were unsutured. That long-term iological changes associated with the transition to seawater— tag loss does not occur is a critical assumption in studies that i.e., “smoltification.” Indeed, an increased sensitivity to stress use tags and is something that remains to be tested in lampreys during smoltification can occur in juvenile salmonids (Specker tagged as juveniles. 1982; Barton et al. 1985). Finally, since nearly all fish that died Our results showed that survival of juvenile Pacific lampreys had a fungal infection, the unusually high mortality of fish at in freshwater was high at 7 and 14 d, influenced by fish size and 12◦C could be due to optimal fungal growth at that temperature. temperature, and independent of treatment group, since tagged Most aquatic fungi, however, have optimal growth temperatures and untagged fish showed consistently similar trends at all tem- higher than 12◦C (Rozek and Timberlake 1979; Koeypudsa et al. peratures. Small fish from both tagged and untagged groups 2005). Based on the survival curves seen at 9, 15, and 18◦C, died, making it difficult to establish a size threshold for tagging we think that an effect specific to the juvenile lampreys held at based on mortality. Our surgeon reported that implanting the 12◦C (whether for the reasons suggested above or others) signif- 9-mm PIT tags was relatively difficult in fish from about 120– icantly increased the mortality at 12◦C with prolonged holding. 130 mm but became progressively easier and faster as fish size Although further analysis of the survival of juvenile lampreys increased. Mueller et al. (2006) concluded that juvenile lam- PIT-tagged at 12◦C is needed to clarify these results, we feel preys greater than 120 mm were suitable for being implanted confident that they can be safely handled at this temperature. with 12-mm PIT tags, but based on our results, we suggest a In freshwater, most juvenile lampreys apparently died from more conservative approach would be to use a fish-size thresh- aquatic fungal infections, which have been a recurring problem old for tagging of near 135 mm and a 9-mm PIT tag. Experi- in laboratory studies of juvenile Pacific lampreys (Schreck et al. enced surgeons should be able to effectively tag fish that are 1999; Mueller et al. 2006). Our results showed that rearing fish only130 mm. in seawater clearly inhibited fungal infections and resulted in Survival of tagged and untagged juvenile Pacific lampreys at very little mortality over a period of 3 months. Thus, for fish 14 d was highest in fish held at 9◦C and 15◦C and declined by that are tagged and released in the wild, fungal infections are 5–12% in fish held at 12◦C and 18◦C. With prolonged holding, likely to be minimal or nonexistent if they can reach the estuary survival was still very high at 9◦C but declined by 20–25% in or ocean promptly. Based on juvenile salmonid migration rates fish held at warmer temperatures (15–18◦C; except for fish at or water particle travel times, we believe that most juvenile 12◦C, see below). Our results at 40 d are similar to those of Pacific lampreys tagged in the Columbia River basin would Mueller et al. (2006), who reported 98% survival of PIT-tagged make it to seawater in time to prevent fungal infections. For and untagged juvenile lampreys held in cold water and only example, during years of normal flow, the median travel time 40–50% survival of fish gradually acclimated to and held at for juvenile Chinook salmon from John Day Dam to Bonneville 19–23◦C. Based on our work and that of Mueller et al. (2006), Dam, the first dam on the Columbia River, is about 2–2.5 d. In it would probably be safe to work with juvenile lampreys at low flow years, median travel time can be up to 6.5 d. Further, 9–10◦C but not at temperatures of 19◦C or higher. Our survival water particle transit time from John Day Dam to Bonneville results at 7–14 d indicated that juvenile lampreys can also be Dam is only about 1.5–4 d (J. McCann, Fish Passage Center, tagged at temperatures between these two extremes, but they personal communication). Thus, if juvenile lampreys move as
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 should migrate to seawater within 14 d after handling for best fast as juvenile salmon, or if they do not swim and move at least survival. To determine whether juvenile lampreys can stay in as fast as water flow, we expect that most fish tagged at John freshwater for longer periods of time will require resolving two Day Dam would be in seawater well within 7 d and certainly issues: (1) the unexpectedly low survival we saw at 12◦C with within 14 d—when survival of our fish was still high. prolonged holding and (2) the proximate cause of fish mortalities Problems with fungal infections in juvenile Pacific lampreys in our study, which we suspect were due to fungal infections. held captive for long periods may be an artifact of laboratory As previously mentioned, there was unusually high mortality holding and have little relevance to field situations. Fungal in- at 12◦C by 40 d—about double that seen at higher temperatures. fections in fish are generally thought to be a result of secondary This surprised us, because nothing in our experience—and that invaders following injury or stress (Piper et al. 1982), and the of others—indicated that working with juvenile lampreys at high incidence of infection in juvenile lampreys held in the lab- 12◦C was a problem. A number of factors may have contributed oratory may be linked to the stresses associated with capture and to the unusual mortality at 12◦C. First, most of the fish from the transport, anesthetic solutions, handling and captivity, or some 12◦C tanks were collected earlier and held 11 d longer than fish combination thereof. Although we observed fungal infections from the other tanks and were treated with hydrogen peroxide on some juvenile lampreys when they were first collected, and twice before we began treatments on the other fish. These ex- Schreck et al. (1999) also noted that 25% of juvenile lampreys PIT-TAGGED JUVENILE PACIFIC LAMPREYS 1267
collected at the John Day Dam in early June 1999 had visible Beeman, J. W., N. Bower, S. Juhnke, L. Dingmon, M. van den Tillaart, and T. signs of fungal infection, detailed examinations of juvenile Thomas. 2007. Effects of antenna length and material on output power and Pacific lampreys at John Day Dam in 2011 showed that only detection of miniature radio transmitters. Hydrobiologia 582:221–229. Bruyndoncx, L., G. Knaepkens, W. Meeus, L. Bervoets, and M. Eens. 2002. The 10 of 4,200 fish had evidence of fungus (G. Kovalchuk, Pacific evaluation of passive integrated transponder (PIT) tags and visible implant States Marine Fish Commission, personal communication). elastomer (VIE) marks as new marking techniques for the bullhead. Journal Thus, fungal infections on juvenile lampreys can occur in of Fish Biology 60:260–262. the wild, but we suspect they may be worsened by handling Close, D. A., M. S. Fitzpatrick, and H. W. Li. 2002. The ecological and cul- and long-term captivity of fish that should be bound for the tural importance of a species at risk of extinction, Pacific lamprey. Fisheries 27(7):19–25. estuary and ocean. More conclusive evidence of whether fungal Dixon, C. J., and M. G. Mesa. 2011. Survival and tag loss in Moapa White River infections in juvenile lampreys related to tagging and handling springfish implanted with passive integrated transponder tags. Transactions are a problem for fish released in the wild will come from of the American Fisheries Society 140:1375–1379. future mark–recapture studies. Goodman, D. H., S. B. Reid, M. F. Docker, G. R. Haas, and A. P. Kinziger. The development of safe, efficient, and effective techniques 2008. Mitochondrial DNA evidence for high levels of gene flow among popu- lations of a widely distributed anadromous lamprey Entosphenus tridentatus for PIT-tagging juvenile lampreys is essential to continued study (Petromyzontidae). Journal of Fish Biology 72:400–417. of their biology, distribution, and migration. Future research us- Keefer, M. L., C. C. Caudill, C. A. Peery, and S. R. Lee. 2008. Transporting juve- ing PIT tags in juvenile Pacific lampreys should focus on gain- nile salmonids around dams impairs adult migration. Ecological Applications ing a greater understanding of migration timing, dam passage 18:1888–1900. issues, growth rates, dispersal, homing (or lack thereof), and Knudsen, C. M., M. V. Johnston, S. L. Schroder, W. J. Bosch, D. E. Fast, and C. R. Strom. 2009. Effects of passive integrated transponder tags on smolt- their genetic population structure (Goodman et al. 2008; Lin to-adult recruit survival, growth, and behavior of hatchery spring Chinook et al. 2008). Our research has resulted in an effective protocol salmon. North American Journal of Fisheries Management 29:658–669. for PIT-tagging juvenile lampreys, which may also be useful Koeypudsa, W., P. Phadee, J. Tangtrongpiros, and K. Hatai. 2005. Influence for tagging these fish with other devices, such as miniature of pH, temperature and sodium chloride concentration on growth rate of acoustic transmitters. We would support the implementation of Saprolegnia sp. Journal of Scientific Research at Chulalongkorn University 30:123–130. PIT-tagging programs for juvenile lampreys in the Columbia Lin, B., Z. Zhang, Y. Wang, K. P. Currens, A. Spidle, Y. Yamazaki, and River basin, similar to efforts that currently exist for salmonids. D. A. Close. 2008. Amplified fragment length polymorphism assessment of genetic diversity in Pacific lampreys. North American Journal of Fisheries Management 28:1182–1193. ACKNOWLEDGMENTS Marsh, D. M., G. M. Matthews, S. Achord, T. E. Ruehle, and B. P. Sandford. We thank Chris Walker of the Columbia River Research Lab- 1999. Diversion of salmonid smolts tagged with passive integrated transpon- oratory for conducting the logistic regression analyses, Lilith ders from an untagged population passing through a juvenile collection sys- Taylor and Rachael Wade of the Marrowstone Marine Field tem. North American Journal of Fisheries Management 19:1142–1146. Mesa, M. G., E. S. Copeland, and H. E. Christiansen. 2011. Development Station for technical assistance, and Annette Peterson, Lisa of standard protocols for tagging juvenile lampreys with passive integrated Weiland, Joe Warren, and Lisa Gee of the Columbia River transponder (PIT) tags. Report to the U.S. Army Corps of Engineers, Contract Research Laboratory for technical assistance, as well as Sean W66QKZ03335311, Portland, Oregon. Tackley of the U.S. Army Corps of Engineers for administer- Moser, M. L., P. A. Ocker, L. C. Stuehrenberg, and T. C. Bjornn. 2002. ing this project. Reviews by Tom Archdeacon, Brien Rose, and Passage efficiency of adult Pacific lampreys at hydropower dams on the lower Columbia River, USA. Transactions of the American Fisheries Society an anonymous reviewer improved the manuscript. Any use of 131:956–965. trade names is for descriptive purposes only and does not imply Mueller, R. P., R. A. Moursund, and M. D. Bleich. 2006. Tagging juvenile endorsement by the U.S. Government. Pacific lamprey with passive integrated transponders: methodology, short- term mortality, and influence on swimming performance. North American
Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 Journal of Fisheries Management 26:361–366. REFERENCES Piper, R. G., I. B. McElwain, L. E. Orme, J. P. McCraren, L. G. Fowler, and J. R. Adams, N. S., G. E. Johnson, D. W. Rondorf, S. M. Anglea, and T. Wik. Leonard. 1982. Fish hatchery management. U.S. Fish and Wildlife Service, 2001. Biological evaluation of the behavioral guidance structure at Lower Washington, D.C. Granite Dam on the Snake River, Washington in 1998. Pages 145–160 in Quintella, B. R., N. O. Andrade, R. Espanhol, and P. R. Almeida. 2005. The use C. C. Coutant, editor. Behavioral technologies for fish guidance. American of PIT telemetry to study movements of ammocoetes and metamorphosing Fisheries Society, Symposium 26, Bethesda, Maryland. sea lampreys in river beds. Journal of Fish Biology 66:97–106. Archdeacon, T. P., W. J. Remshardt, and T. L. Knecht. 2009. Comparison of two R Development Core Team. 2011. R: a language and environment for statisti- methods for implanting passive integrated transponders in Rio Grande silvery cal computing. R Foundation for Statistical Computing, Vienna. Available: minnow. North American Journal of Fisheries Management 29:346–351. www.R-project.org. (February 2011). Baras, E., L. Westerloppe, C. Melard,´ J. C. Philippart, and V. Benech.´ 1999. Rozek, C. E., and W. E. Timberlake. 1979. Optimum conditions for syn- Evaluation of implantation procedures for PIT-tagging juvenile Nile tilapia. chronous oospore production in crosses of Achlya ambisexualis. Experimental North American Journal of Aquaculture 61:246–251. Mycology 3:378–381. Barton, B. A., C. B. Schreck, R. D. Ewing, A. R. Hemmingsen, and R. Patino.˜ Schreck, C. B., M. S. Fitzpatrick, and D. L. Lerner. 1999. Determination 1985. Changes in plasma cortisol during stress and smoltification in coho of passage of juvenile lamprey: development of a tagging protocol. Ore- salmon, Oncorhynchus kisutch. General and Comparative Endocrinology gon Cooperative Fish and Wildlife Research Unit, Oregon State University, 59:468–471. Corvallis. 1268 MESA ET AL.
Skalski, J. R. 1998. Estimating season-wide survival rates of outmigrating rivers hydropower system, 1966–1980 and 1993–1999. North American Jour- salmon smolt in the Snake River, Washington. Canadian Journal of Fish- nal of Fisheries Management 21:310–317. eries and Aquatic Sciences 55:761–769. Zabel, R. W. 2002. Using “travel time” data to characterize the behavior of Specker, J. L. 1982. Interrenal function and smoltification. Aquaculture 28: migrating animals. American Naturalist 159:372–387. 59–66. Zabel, R. W., T. Wagner, J. L. Congleton, S. G. Smith, and J. G. Williams. 2005. Williams, J. G. 2008. Mitigating the effects of high-head dams on the Columbia Survival and selection of migrating salmon from capture–recapture models River, USA: experience from the trenches. Hydrobiologia 609:241–251. with individual traits. Ecological Applications 15:1427–1439. Williams, J. G., S. G. Smith, and W. D. Muir. 2001. Survival estimates for down- Zabel, R. W., and J. G. Williams. 2002. Selective mortality in Chinook salmon: stream migrant yearling juvenile salmonids through the Snake and Columbia what is the role of human disturbance? Ecological Applications 12:173–183. Downloaded by [Department Of Fisheries] at 20:04 25 September 2012 This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:06 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Genetic Analysis Reveals Dispersal of Florida Bass Haplotypes from Reservoirs to Rivers in Central Texas Jesse W. Ray a b , Martin Husemann a , Ryan S. King a c & Patrick D. Danley a a Department of Biology, Baylor University, One Bear Place 97388, Waco, Texas, 76798, USA b The Institute of Ecological, Earth, and Environmental Sciences, Baylor University, One Bear Place 97205, Waco, Texas, 76798, USA c Center for Reservoir and Aquatic Systems Research, Baylor University, One Bear Place 97178, Waco, Texas, 76798, USA
Version of record first published: 30 Jul 2012.
To cite this article: Jesse W. Ray, Martin Husemann, Ryan S. King & Patrick D. Danley (2012): Genetic Analysis Reveals Dispersal of Florida Bass Haplotypes from Reservoirs to Rivers in Central Texas, Transactions of the American Fisheries Society, 141:5, 1269-1273 To link to this article: http://dx.doi.org/10.1080/00028487.2012.690814
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. Transactions of the American Fisheries Society 141:1269–1273, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.690814
NOTE
Genetic Analysis Reveals Dispersal of Florida Bass Haplotypes from Reservoirs to Rivers in Central Texas
Jesse W. Ray* Department of Biology, Baylor University, One Bear Place 97388, Waco, Texas 76798, USA and The Institute of Ecological, Earth, and Environmental Sciences, Baylor University, One Bear Place 97205, Waco, Texas 76798, USA Martin Husemann Department of Biology, Baylor University, One Bear Place 97388, Waco, Texas 76798, USA Ryan S. King Department of Biology, Baylor University, One Bear Place 97388, Waco, Texas 76798, USA and Center for Reservoir and Aquatic Systems Research, Baylor University, One Bear Place 97178, Waco, Texas 76798, USA Patrick D. Danley Department of Biology, Baylor University, One Bear Place 97388, Waco, Texas 76798, USA
M. punctulatus, largemouth bass M. salmoides, Guadalupe bass Abstract M. treculii, Florida bass M. floridanus, and smallmouth bass M. We analyzed the genetic structure of putative largemouth dolomieu. The first three of these species are native to Texas, bass Micropterus salmoides populations in nonstocked, wadeable while the latter two are nonnative and are stocked as sport fish. streams of central Texas. Mitochondrial D-loop sequences were generated for 69 fish sampled for this project. In addition, 27 large- The status of the Florida bass is still under debate. Although mouth bass and Florida bass M. floridanus specimens provided by many consider the Florida bass to be a valid species (Kassler the TexasParks and Wildlife Department (TPWD) were sequenced. et al. 2002; Near et al. 2003, 2004; Eschmeyer 2012), species The TPWD samples represented stock lineages as well as wild fish designation has not been completely accepted (Nelson et al. from outside of the sampling region. Our analyses revealed the 2004). Here, we side with the former perspective in using a full presence of both largemouth bass and Florida bass mitochondrial haplotypes at all sampling locations. Haplotypes of the nonnative species designation for the Florida bass. Florida bass accounted for 26% of all haplotypes. The presence Largemouth bass and Florida bass are the most extensively
Downloaded by [Department Of Fisheries] at 20:06 25 September 2012 of Florida bass haplotypes at the sampling locations indicates that managed black basses in Texas. Recent trends in fisheries man- the influence of stocking reaches far beyond managed reservoirs. agement have involved a shift to the stocking of Florida bass The admixture of nonnative genetic material can increase genetic in lieu of largemouth bass across much of North America, as diversity of native populations, but outbreeding depression, com- petition, and other negative impacts are of concern. Although the Florida bass attain a larger maximum size than largemouth stocking of nonnative Florida bass in reservoirs may enhance fish- bass. These species are closely related and hybridize readily ing opportunities, it also has the ability to alter stream systems that when in natural sympatry (Bailey and Hubbs 1949) and when are directly connected to stocked reservoirs. stocked outside of their native ranges (Gelwick et al. 1995). The results of such stocking events have been variable and de- The black basses Micropterus spp. currently include nine bated (Philipp and Whitt 1991; Maceina and Murphy 1992; recognized species (Near and Koppelman 2009), some of which Philipp 1992), but data from a Texas reservoir indicate that are among the most popular sport fishes in North America. Florida bass introgression into native largemouth bass popu- Five species of Micropterus occur in Texas: spotted bass lations can occur rapidly, with nonnative alleles found in the
*Corresponding author: jesse [email protected] Received August 7, 2011; accepted April 24, 2012 1269 1270 RAY ET AL.
FIGURE 1. Maps of (a) Texas, showing the locations of the Trinity River and Brazos River watersheds; and (b) the Trinity and Brazos River tributaries, displaying sampling sites (black dots) and known Florida bass stocking locations (fish symbols; HOG = Hog Creek; HARR = Harris Creek; DUFF = Duffau Creek; NBOS = North Bosque River; NOLR = Nolan River; CFTR = Clear Fork [Trinity River]).
majority of largemouth bass in less than 4 years (Maceina et al. Trinity and Brazos River drainages with haplotypes found in 1988). populations from Illinois and Florida and in Texas Parks and While the stocking of nonnative fish species occurs world- Wildlife Department (TPWD) stocks. Our results illustrate wide, often little is known about the long-term impacts of such the geographic extent to which Florida bass haplotypes have actions. Generally, fisheries with high numbers of introduced spread from known stocking locations and show that the species appear to be less stable than those with mostly na- stocking of nonnative fish into reservoirs has impacts on aquatic tive fauna (Moyle 1986). Negative impacts of the stocking and communities that are distant to the actual stocking location. release of baitfish are documented for many North American species, including the red shiner Cyprinella lutrensis (Walters
Downloaded by [Department Of Fisheries] at 20:06 25 September 2012 et al. 2008; Blum et al. 2010) and kokanee Oncorhynchus nerka METHODS (Morgan et al. 1978). Impacts of the large-scale introductions of Study sites and sampling techniques.—Samples of putative largemouth bass and Florida bass include the muddling of his- largemouth bass were collected at five sites within the Brazos torical species boundaries (Philipp et al. 1983) and obscuring River system and at one site in the Trinity River system of the native range of largemouth bass (Boschung and Mayden (Figure 1). The Brazos River locations include Hog Creek, 2004). Data on largemouth bass and Florida bass introductions Harris Creek, Duffau Creek, North Bosque River, and Nolan are abundant, yet relatively little is known about watershed-scale River; the Trinity River site lies on Clear Fork. All sampling lo- impacts of fish stocking in Texas. cations are tributaries of stocked reservoirs, spanning distances We analyzed mitochondrial D-loop sequences of putative of 5–80 km from stocking sites. largemouth bass populations in wadeable streams from the Specimens were collected by using a backpack electrofisher Brazos and Trinity River watersheds in central Texas and tested (Smith-Root Model LR-24) and seine nets (4.6 × 1.8 m or for the presence of Florida bass haplotypes. These streams 1.8 × 1.8 m) during the course of long-term fish community represent nonstocked locations that are connected to stocked monitoring projects (Pease et al. 2011; Stanley et al. 2012). reservoirs by varying distances. We compared mitochondrial All fish were initially identified as native largemouth bass. At haplotypes of largemouth bass and Florida bass found in the the time of sampling, we had no reason to believe that the NOTE 1271
stocking of distant reservoirs was influencing sampling locations magnetic beads (Bangs Laboratories) and a 96-ring solid-phase on wadeable streams, and there were no apparent morphological reversible immobilization plate (Agencourt). The purified PCR differences among sampled fish. Florida bass and largemouth products were sequenced by the Yale University Sequencing bass are difficult to distinguish in the field by using simple Facility (New Haven, Connecticut). Sequences were inspected, morphological traits, especially for juveniles, as there is signif- trimmed, and aligned using Geneious version 5.0.3 (Drummond icant overlap in their phenotypes (Kassler et al. 2002). Small et al. 2006). Median joining networks were constructed under fish were retained, while larger specimens were identified, fin default conditions by using Network (Bandelt et al. 1999). clipped, and released. Fin clips were stored in 99.8% ethanol. In addition to field samples, reference samples of Florida bass and RESULTS largemouth bass lineages stocked in Texas were obtained from In total, 69 sequences trimmed to 883 bp in length were ob- TPWD (D. Lutz-Carrillo, TPWD, A. E. Wood Laboratory). The tained from the fish sampled for this project. An additional 27 TPWD also provided samples of wild-caught largemouth bass sequences were generated from TPWD samples: 10 Florida bass from Texas and Illinois and of wild Florida bass captured in from the stock lineage, 8 largemouth bass from the stock lin- Florida. The TPWD samples served as a baseline for comparing eage, 4 wild-caught Florida bass, 3 wild-caught largemouth bass the haplotypes of our black bass samples with the haplotypes of from Texas, and 2 wild-caught largemouth bass from Illinois. known stocked lineages and nonnative wild lineages. All sequences were submitted to GenBank (accession num- DNA methods.—Genomic DNA was extracted from frozen bers JN979571-JN979602 and JN979661-JN979724). Within or ethanol-preserved tissue by using a Qiagen DNeasy the 69 samples obtained from our sampling effort, 51 haplotypes Blood and Tissue Kit (Qiagen, Inc., Valencia, California) were found, 16 of which corresponded to Florida bass. Three in accordance with the manufacturer’s protocol for tissue single-base insertion–deletion events were found. None of the samples. For amplification and sequencing of the mitochondrial insertion–deletion events was unique, and none of them resulted D-loop control region, we used the primers CR-F (5 -GGATT in the creation of a new haplotype. The median joining net- TTAACCCYCACCMCT-3 ) and CR-R (5 -TTCTAGGGCTCA work, which included regional and TPWD samples, showed two TCTTAACATCTTC-3 ) with an M13-41 adapter (Husemann distinct groups corresponding to largemouth bass and Florida et al., in press). The PCR products were tested on a 1% agarose bass (Figure 2). Florida bass haplotypes were found at all gel stained with Gel Red (0.1 × , Biotium) and photographed. sampling locations, with no appreciable pattern related to the The PCR products were purified using solid-phase reversible distance from known stocking sites. Due to the uniparental in- immobilization (DeAngelis et al. 1995) with carboxylated heritance of mitochondria, only maternal lineages are sampled Downloaded by [Department Of Fisheries] at 20:06 25 September 2012
FIGURE 2. Median joining network of Florida bass Micropterus floridanus and largemouth bass M. salmoides haplotypes. Black dots represent intermediate unsampled (inferred) haplotypes, and circles represent sampled haplotypes, with circle size proportional to sample size (shown inside circle for all n > 1). Color corresponds to sample origin (location codes are defined in Figure 1). The dotted line between Florida bass and largemouth bass networks represents a gap of 30 mutational steps. 1272 RAY ET AL.
and hybrid individuals cannot be differentiated from nonhy- studies using nuclear markers will help to distinguish between brids. At this time, we cannot determine whether our samples these scenarios or combinations thereof. represent a hybridizing group of largemouth bass and Florida Fifty-one different haplotypes were recovered from our sam- bass or two distinct co-occurring species. ple of 69 fish, indicating high levels of genetic diversity in largemouth bass and Florida bass in central Texas. While high genetic diversity is often a sign of healthy populations, changes DISCUSSION in the genetic makeup of populations due to hybridization may The largemouth bass has a long and nearly continuous stock- reduce genetic diversity over the long term and can even lead ing history in the state of Texas, but recent stocking efforts have to extinction (Rhymer and Simberloff 1996). Hybridization due transitioned from largemouth bass to Florida bass. The Florida to anthropogenic translocation is considered a major threat to bass is widely considered a better sport fish because it attains a fish species worldwide (Allendorf et al. 2001) and may lead greater maximum size, which strongly influences the stocking to outbreeding depression, which has been shown to occur in of this species across much of North America. Although stocked mixed lineages of largemouth bass and Florida bass (Philipp reservoirs are commonly studied (e.g., Maceina et al. 1988), the et al. 2002; Goldberg et al. 2005). impact of fish stocking on stream networks that are connected The spread of nonnative alleles into native populations is a to these reservoirs remains largely unknown. Our sampling lo- common theme in fisheries and wildlife management, particu- cations represent nonstocked, wadeable streams, all of which larly with regard to heavily managed species such as the large- are connected to stocked reservoirs at varying distances. Under- mouth bass (Gelwick et al. 1995; Johnson and Fulton 1999) standing the genetic impact of stocking is particularly important, and salmonids (Apostolidis et al. 2008; Campos et al. 2008; as largemouth bass and Florida bass readily hybridize (Bailey Dawnay et al. 2011). The presence of Florida bass haplotypes and Hubbs 1949; Gelwick et al. 1995) and outbreeding depres- in all sampled largemouth bass populations in central Texas sion has been documented across lineages of largemouth bass highlights the large-scale impacts that fish stocking can have on (Philipp et al. 2002; Goldberg et al. 2005). the genetic makeup of fish populations. While it is evident that Of the 69 fish sampled for this project, 51 fish had haplotypes largemouth bass in central Texas have high genetic diversity, the corresponding to largemouth bass and 18 fish (26%) had haplo- portions of diversity that are attributable to natural versus an- types that corresponded to Florida bass. The largest proportion thropogenic sources are unknown. Further studies using nuclear of Florida bass haplotypes occurred in the Nolan River, a tribu- markers would be useful for differentiating hybrids from pure tary that drains directly into Lake Whitney. In contrast, the four lineages of Florida bass and largemouth bass in Texas. Under- remaining sample sites within the Brazos River drainage are standing the extent of Florida bass dispersal and hybridization tributaries of Lake Waco and had a lower incidence of Florida with largemouth bass would elucidate the mechanisms of intro- bass haplotypes. Both Lake Whitney and Lake Waco have been gression and would shed light on the impacts of Florida bass stocked with largemouth bass and Florida bass over the past stocking on native largemouth bass populations in central Texas. 40 years (www.tpwd.state.tx.us; accessed 10 November 2011). All stocking events since the 1980s have involved the stocking of Florida bass. Lake Whitney received nearly 10 times the num- ACKNOWLEDGMENTS ber of Florida bass fingerlings than did Lake Waco since this We thank E. Hooser, J. Back, and J. Taylor for help in fish col- transition, and the higher stocking effort may explain the larger lections, and we thank Dijar Lutz-Carrillo (TPWD, A. E. Wood proportion of Florida bass haplotypes in populations upstream. Laboratory) for providing numerous samples. We are grateful to Our results suggest that the stocking of Florida bass has a E. Rapstine and S. Namkung for processing and organizing tis-
Downloaded by [Department Of Fisheries] at 20:06 25 September 2012 significant influence on fish populations far beyond the stocked sue samples and to B. Bartram, D. Lewis, and J. Tibbs (TPWD) reservoirs. We found Florida bass haplotypes in Duffau Creek, for providing survey data, technical advice, and comments on which is situated more than 80 km upstream from the clos- previous versions of the manuscript. Jesse W. Ray and Martin est documented stocking location. Although Florida bass and Husemann contributed equally to this article. Funding for this largemouth bass haplotypes are distinct, we were not able to project was provided by a grant from the Texas Commission genetically differentiate between hybrids and pure lineages by on Environmental Quality (Contract 582-6-578-80304) to Ryan using the D-loop marker, and morphological species delimi- S. King, a Baylor University Summer Undergraduate Research tation was not possible because the majority of the sampled Fellowship to E. Hooser and Ryan S. King, and additional fund- individuals were juveniles. Therefore, the origin of fish con- ing from Baylor University to Patrick D. Danley and Ryan S. taining Florida bass mitochondrial DNA cannot be definitively King. pinpointed. Florida bass alleles in fish sampled outside of the reservoirs presumably arose from one of two sources: (1) direct REFERENCES dispersal of stocked fish and their descendants from the reser- Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg. 2001. The prob- voirs into surrounding streams or (2) hybridization with and lems with hybrids: setting conservation guidelines. Trends in Ecology and introgression into native largemouth bass populations. Further Evolution 16:613–622. NOTE 1273
Apostolidis, A. P., M. J. Madeira, M. M. Hansen, and A. Machordom. 2008. Maceina, M. J., and B. R. Murphy. 1992. Comments: stocking Florida large- Genetic structure and demographic history of brown trout (Salmo trutta) mouth bass outside its native range. Transactions of the American Fisheries populations from the southern Balkans. Freshwater Biology 53:1555–1566. Society 121:686–688. Bailey, R. M., and C. L. Hubbs. 1949. The black basses (Micropterus)of Morgan, M. D., S. T. Threlkeld, and C. R. Goldman. 1978. Impact of the Florida with description of a new species. Occasional Papers of the Museum introduction of kokanee (Oncorhynchus nerka) and opossum shrimp (Mysis of Zoology University of Michigan 516. relicta) on a subalpine lake. Journal of the Fisheries Research Board of Canada Bandelt, H. J., P. Forster, and A. Rohl.¨ 1999. Median-joining networks for 35:1572–1579. inferring intraspecific phylogenies. Molecular Biology and Evolution 16: Moyle, P. B. 1986. Fish introductions into North America: patterns and ecolog- 37–48. ical impact. Pages 27–43 in H. A. Mooney and J. A. Drake, editors. Ecol- Blum, M. J., D. M. Walters, N. M. Burkhead, B. J. Freeman, and B. A. Porter. ogy of biological invasions of North America and Hawaii. Springer-Verlag, 2010. Reproductive isolation and the expansion of an invasive hybrid swarm. New York. Biological Invasions 12:2825–2836. Near, T. J., D. I. Bolnick, and P.C. Wainwright. 2004. Investigating phylogenetic Boschung, H. T., Jr., and R. L. Mayden. 2004. Fishes of Alabama. Smithsonian relationships of sunfishes and black basses (Actinopterygii: Centrarchidae) Books, Washington, D.C. using DNA sequences from mitochondrial and nuclear genes. Molecular Campos, J. L., D. Posada, and P. Moran.´ 2008. Introgression and genetic struc- Phylogenetics and Evolution 32:344–357. ture in northern Spanish Atlantic salmon (Salmo salar L.) populations ac- Near, T. J., T. W. Kassler, J. B. Koppelman, C. B. Dillman, and D. P. cording to mtDNA data. Conservation Genetics 9:157–169. Philipp. 2003. Speciation in North American black basses, Micropterus Dawnay, N., L. Dawnay, R. N. Hughes, R. Cove, and M. I. Taylor. 2011. (Actinopterygii: Centrarchidae). Evolution 57:1610–1621. Substantial genetic structure among stocked and native populations of the Near, T. J., and J. B. Koppelman. 2009. Species diversity, phylogeny, and phy- European grayling (Thymallus thymallus, Salmonidae) in the United logeography of Centrarchidae. Pages 1–38 in S. J. Cooke and D. P. Philipp, Kingdom. Conservation Genetics 12:731–744. editors. Centrarchid fishes: diversity, biology, and conservation. Blackwell DeAngelis, M. M., D. G. Wang, and T. L. Hawkins. 1995. Solid-phase reversible Scientific Publications, Oxford, UK. immobilization for the isolation of PCR products. Nucleic Acids Research Nelson, J. S., E. J. Crossman, H. Espinosa-Perez,´ L. T. Findley, C. R. Gilbert, 23:4742–4743. R. N. Lea, and J. D. Williams. 2004. Common and scientific names of fishes Drummond, A. J., M. Kearse, J. Heled, R. Moir, T. Thierer, B. Ashton, A. from the United States, Canada, and Mexico. American Fisheries Society, Wilson, and S. Stones-Havas. 2006. Geneious v5.0.3: DNA sequence analy- Special Publication 29, Bethesda, Maryland. sis software for biologists. Biomatters, Auckland, New Zealand. Available: Pease, A. A., J. M. Taylor, K. O. Winemiller, and R. S. King. 2011. Mul- www.geneious.com. (August 2011). tiscale environmental influences on fish assemblage structure in central Eschmeyer, W. N., editor. 2012. Catalog of fishes. California Academy Texas streams. Transactions of the American Fisheries Society 140:1409– of Sciences, San Francisco. Available: research.calacademy.org/research/ 1427. ichthyology/catalog/fishcatmain.asp. (March 2012). Philipp, D. P. 1992. Comments: stocking largemouth bass outside its Gelwick, F. P., E. R. Gilliland, and W. J. Matthews. 1995. Introgression of native range. Transactions of the American Fisheries Society 121: the Florida largemouth bass genome into stream populations of northern 688–691. largemouth bass in Oklahoma. Transactions of the American Fisheries Society Philipp, D. P., W. F. Childers, and G. S. Whitt. 1983. A biochemical genetic 124:550–562. evaluation of the northern and Florida subspecies of largemouth bass. Trans- Goldberg, T. L., E. C. Grant, K. R. Inendino, T. W. Kassler, J. E. Claussen, and actions of the American Fisheries Society 112:1–20. D. P. Philipp. 2005. Increased infectious disease susceptibility resulting from Philipp, D. P., J. E. Claussen, T. W. Kassler, and J. M. Epifanio. 2002. Mixing outbreeding depression. Conservation Biology 19:455–462. stocks of largemouth bass reduces fitness through outbreeding depression Husemann, M., J. Ray, R. S. King, E. A. Hooser, and P. D. Danley. In press. Pages 349–363 in D. P. Philipp and M. S. Ridgway, editors. Black bass: ecol- Comparative biogeography reveals differences in population genetic structure ogy, conservation, and management. American Fisheries Society, Symposium of five species of stream fish. 31, Bethesda, Maryland. Johnson, R. L., and T. Fulton. 1999. Persistence of Florida largemouth bass Philipp, D. P., and G. S. Whitt. 1991. Survival and growth of northern, Florida, alleles in a northern Arkansas population of largemouth bass, Micropterus and reciprocal F1 hybrid largemouth bass in central Illinois. Transactions of salmoides Lacep` ede.´ Ecology of Freshwater Fish 8:35–42. the American Fisheries Society 120:58–64. Kassler, T. W., J. B. Koppelman, T. J. Near, C. B. Dillman, J. M. Rhymer, J. M., and D. Simberloff. 1996. Extinction by hybridization and intro- Levengood, D. L. Swofford, J. L. VanOrman,J. E. Claussen, and D. P.Philipp. gression. Annual Review of Ecology and Systematics 27:83–109. 2002. Molecular and morphological analyses of the black basses: implications Stanley, C. E., J. M. Taylor, and R. S. King. 2012. Coupling fish commu-
Downloaded by [Department Of Fisheries] at 20:06 25 September 2012 for taxonomy and conservation. Pages 291–322 in D. P. Philipp and M. S. nity structure with instream flow and habitat connectivity between two hy- Ridgway, editors. Black bass: ecology, conservation, and management. drologically extreme years. Transactions of the American Fisheries Society American Fisheries Society, Symposium 31, Bethesda, Maryland. 141:1000–1015. Maceina, M. J., B. R. Murphy, and J. J. Isely. 1988. Factors regulating Florida Walters, D. M., M. J. Blum, B. Rashleigh, B. J. Freeman, B. A. Porter, and largemouth bass stocking success and hybridization with northern largemouth N. M. Burkhead. 2008. Red shiner invasion and hybridization with black- bass in Aquilla Lake, Texas. Transactions of the American Fisheries Society tail shiner in the upper Coosa River, USA. Biological Invasions 10:1229– 117:221–231. 1242. This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:08 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Mercury in Groupers and Sea Basses from the Gulf of Mexico: Relationships with Size, Age, and Feeding Ecology Derek M. Tremain a & Douglas H. Adams a a Florida Fish and Wildlife Conservation Commission, Fish and Wildlife Research Institute, 1220 Prospect Avenue, Suite 285, Melbourne, Florida, 32901, USA Version of record first published: 09 Aug 2012.
To cite this article: Derek M. Tremain & Douglas H. Adams (2012): Mercury in Groupers and Sea Basses from the Gulf of Mexico: Relationships with Size, Age, and Feeding Ecology, Transactions of the American Fisheries Society, 141:5, 1274-1286 To link to this article: http://dx.doi.org/10.1080/00028487.2012.683232
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. Transactions of the American Fisheries Society 141:1274–1286, 2012 C American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.683232
ARTICLE
Mercury in Groupers and Sea Basses from the Gulf of Mexico: Relationships with Size, Age, and Feeding Ecology
Derek M. Tremain* and Douglas H. Adams Florida Fish and Wildlife Conservation Commission, Fish and Wildlife Research Institute, 1220 Prospect Avenue, Suite 285, Melbourne, Florida 32901, USA
Abstract We analyzed total mercury concentration in muscle tissue of 15 serranid species (n = 1,401 fish) collected from the Gulf of Mexico and also developed a comprehensive model relating two commonly used mercury analysis methods. There was considerable interspecific and intraspecific variability in mercury within groupers and sea basses. Mean mercury concentration for individual species ranged from 0.03 to 0.91 mg/kg wet weight across all size ranges, and for legally harvestable grouper and sea bass concentrations were 0.32 mg/kg and 0.09 mg/kg, respectively. Mercury in individual fish ranged from 0.01 to 3.3 mg/kg. Approximately 23% of all grouper samples analyzed contained mercury at concentrations that exceeded the U.S. Environmental Protection Agency’s methylmercury consumption guidance criterion for humans (0.3 mg/kg). Mercury in legally harvestable individuals of some species exceeded 0.5 mg/kg, and harvestable black grouper Mycteroperca bonaci had a mean mercury concentration greater than 1.0 mg/kg. A positive relationship was observed between total mercury content and both fish length and fish age. For species that feed principally on fishes (e.g., gag M. microlepis and scamp M. phenax), mean mercury concentration was greater than that for species that feed mainly on invertebrates (e.g., red grouper Epinephelus morio and sand perch Diplectrum formosum). Although overall mean mercury concentrations were relatively low, for six species the maximum mercury concentration was at a level that has been associated with sublethal effects on fish physiology and with human health risks. Current grouper management regulations in the Gulf of Mexico select for a fishery of the largest and oldest individuals, which have the greatest mercury burdens. The species- and size-specific mercury data from this study can be used to inform at-risk human populations, refine regional fish consumption advisories, and further our understanding of mercury bioaccumulation in these important fishery species.
The Serranidae (groupers and sea basses) are a diverse family (Shapiro 1987; Bullock and Smith 1991; Crabtree and Bullock of demersal species, occurring worldwide in warm-temperate, 1998). Groupers and sea basses are ecologically important Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 tropical, and subtropical waters. In the Gulf of Mexico, they are components of continental shelf ichthyofaunas (Bullock and ecologically and economically important species and exhibit a Smith 1991) and, as a group, function as prey and predator in wide range of life history patterns (Bullock and Smith 1991). estuarine, nearshore, and offshore food webs. Many species are Most serranid species reside near reef or reeflike habitats, upper-trophic-level piscivores or opportunistic carnivores. although some species such as gag Mycteroperca microlepis, Groupers and sea basses support major recreational and red grouper Epinephelus morio, goliath grouper E. itijara, and commercial fisheries in the U.S. waters of the Gulf of Mexico. sand perch Diplectrum formosum may reside within estuarine In 2010, reported recreational and commercial landings of waters during early life history stages before migrating to groupers and sea basses from the United States exceeded 7.1 offshore habitats (Bullock and Smith 1991). Complex repro- million kg, of which more than 3.6 million kg (50.7%) was ductive strategies that include synchronous or protogynous from the Gulf of Mexico (National Marine Fisheries Ser- hermaphroditism have been observed in many serranid species vice, Fisheries Statistics Division, personal communication).
*Corresponding author: [email protected] Received October 20, 2011; accepted April 2, 2012
1274 MERCURY IN GROUPERS AND SEA BASSES 1275
Approximately 98% of the reported recreational and com- human health concerns. Therefore, the Florida Fish and Wildlife mercial catch of these species in the Gulf of Mexico Conservation Commission–Fish and Wildlife Research Institute were landed along Florida’s west coast (National Marine (FWC–FWRI) analyzed total mercury (THg) concentration in a Fisheries Service, Fisheries Statistics Division, personal representative collection of these species from the eastern Gulf communication). of Mexico. The objectives of this study were to determine vari- Mercury, a toxic metallic element, can bioaccumulate in fish ability in THg concentrations in the edible muscle tissue of the tissue, and humans who consume fish can consume significant various species of grouper and sea basses from nearshore and levels of mercury. Methylmercury (MeHg) is the form of offshore waters of the Gulf of Mexico and to examine the rela- mercury most toxic to humans (NRC 2000), and the majority of tionships between THg and fish size, age, and feeding ecology. total mercury in fish muscle tissue (>95%) is in the monomethyl We also investigated the relationship between THg concen- form (CH3Hg) (Grieb et al. 1990; Bloom 1992; Hammerschmidt trations determined by two commonly used mercury analysis and Fitzgerald 2006). Inorganic mercury enters the coastal and techniques directly applied to marine and estuarine fish tissues. marine aquatic environment through atmospheric deposition and riverine input (Mason and Sheu 2002; Balcolm et al. 2004), and it can be sequestered in sediments where it is methylated METHODS by sulfate-reducing bacteria (Gilmour and Henry 1991; Heyes Groupers and sea basses were collected from nearshore and et al. 2006). The majority of MeHg in marine organisms is a offshore waters of the eastern Gulf of Mexico from 1991 to 2008. result of MeHg production in coastal and continental shelf sed- Approximately 74% of all samples were collected from 2006 iments (Hammerschmidt and Fitzgerald 2004), water column to 2008 in association with FWC–FWRI Fisheries-Independent methylation of bioavailable inorganic mercury in subsurface, Monitoring (FIM) program research cruises with trawls, traps, low-oxygen waters (Sunderland et al. 2009; Hammerschmidt seines, or hooked sampling gear. Most samples collected before and Bowman 2012), and possibly other deep-ocean sources 2006 were obtained from FWC–FWRI Fisheries-Dependent (Kraepiel et al. 2003). Methylation rates in coastal sediments Monitoring (FDM) biostatistical sampling programs, which vary temporally and spatially and depend on many factors, received fish directly from the commercial and recreational including bioturbation, sediment hydrodynamics, sediment fisheries operating within the study area. Although the general chemistry, and the effects of physicochemical, physiological, region of capture for these FDM specimens was known, specific and ecological influences on mercury transport from sediments geographic coordinates were not available. The geographic to organisms (Chen et al. 2008). Passive diffusion, advection, range of our collections extended from approximately the and resuspension can transfer MeHg from surface sediments to DeSoto Canyon off Pensacola in the Florida panhandle the water column (Bloom et al. 1999). Uptake of MeHg from (30.2◦N, 87.2◦W) to the Marquesas Keys (24.4◦N, 82.0◦W) and the water column by marine phytoplankton is the initial step encompassed habitats that extended across the West Florida of the entry and biomagnification of mercury in the marine food Shelf into shallow coastal waters and, in some cases, into the web (Baeyens et al. 2003; IAEA 2004; Hammerschmidt and lower reaches of Florida’s Gulf coast estuaries (Figure 1). Fitzgerald 2006), and MeHg bioconcentrated by phytoplankton Fish collected during FIM research cruises were processed is integrated into pelagic (Mathews and Fisher 2008) and for mercury analysis at sea directly after capture. Fish sampled benthic (Lawrence and Mason 2001) prey. The general ability directly from the commercial and recreational fisheries were of fish to metabolize and eliminate mercury after uptake from placed on ice and returned to the laboratory for processing. their environment is limited (Trudel and Rasmussen 1997). At least one measure of fish length (standard, fork, or total Therefore, mercury levels are greatest in larger and older length) was recorded for each individual. Fish total length
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 individuals and among upper-trophic-level pelagic (Cai et al. (TL) was used in all subsequent analyses; however, when TL 2007; Adams 2010) and demersal (Adams and Onorato 2005; measurements were unavailable, they were calculated from Bank et al. 2007) fish species in the Gulf of Mexico. species-specific length–length regression equations available Fish consumption has been positively correlated with in published scientific literature or from FWC–FWRI. We mercury levels in humans (Choy et al. 2002; Hightower and removed and processed sagittal otoliths from representative Moore 2003), and consumption of contaminated fish is the most subsamples of the four most abundant species collected (i.e., important known source of human exposure to mercury in the red grouper, gag, scamp M. phenax, and sand perch) to estimate United States (NRC 2000). There have been recent concerns age according to the methods outlined in Hood and Schlieder regarding elevated levels of mercury in grouper species from (1992) and Crabtree and Bullock (1998). U.S. and international waters; however, mercury-related data for We collected axial muscle tissue for mercury analysis this varied group of fish has often been broadly classified under from the left dorsal area above the lateral line and anterior a general “grouper” category (Sunderland 2007; USFDA 2012) to the origin of the first dorsal fin of each fish using a clean and not categorized by species. A more complete understanding stainless-steel knife. White muscle tissue taken from this of mercury contamination in grouper and sea bass species region is representative of the portion of fish consumed by is necessary to provide insight into ecosystem condition and humans (Adams and McMichael 2001). Care was taken to 1276 TREMAIN AND ADAMS
FIGURE 1. Eastern Gulf of Mexico study area, showing the distribution of sites (open circles) at which fish were collected for mercury analysis.
ensure that the sample made no contact with the outer layer triplicate tissue samples, and certified fish tissue reference ma- of the specimen, blood, scales, or surrounding surfaces. Tissue terial (e.g., TORT-2, DORM-2, DOLT-2, or DOLT-3 obtained Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 samples were immediately placed in sterile polyethylene vials from the National Research Council of Canada) for each group and stored at −20◦C until analyzed. of 20 CVAAS or 10 CAAS samples analyzed (USEPA 1991). Total mercury concentration of each axial muscle sample In addition, duplicate matrix spikes were completed for each was determined at the Florida Department of Environmental group of 20 CVAAS or 40 CAAS samples analyzed. Mercury Protection (FDEP) Chemistry Laboratory by wet digestion concentrations were measured as milligrams per kilogram wet and cold vapor atomic absorption spectrometry (CVAAS; weight; however, reporting precision differed slightly between U.S. Environmental Protection Agency [EPA] Method 245.6) methods. Results of CAAS analyses were rounded to a precision (USEPA 1991) or at FWC–FWRI by thermal decomposition level of 0.01 mg/kg. The FDEP laboratory results from CVAAS (combustion), amalgamation, and atomic absorption spectrom- analyses were reported to only two significant figures, which etry (CAAS; EPA Method 7473) (USEPA 1998), or both. The truncated precision at THg levels greater than 1.0 mg/kg. CAAS analysis was completed with a calibrated DMA-80 Before 2006, all tissue samples were analyzed for THg at Direct Mercury Analyzer (Milestone, Shelton, Connecticut). FDEP using CVAAS. Beginning in 2006, most tissue samples Quality control procedures for CVAAS and CAAS methods were analyzed at FWC–FWRI using CAAS; however, we included analysis of laboratory method blanks, duplicate or continued to use CVAAS for a portion of our samples based MERCURY IN GROUPERS AND SEA BASSES 1277
on quality assurance requirements or the unique data needs volume (%V), percent by number (%N), percent by frequency of individual research projects. The two analytical methods of occurrence (%O), index of relative importance (IRI; Pinkas produce comparable but not identical determinations of THg et al. 1971), and percent IRI (%IRI; Cortes´ 1997). Taxa whose (Cizdziel et al. 2002; Lowery et al. 2007). Therefore, before overall%IRI was less than 0.1% were removed from the data analyzing the combined data set, we used regression analyses summaries. to relate CVAAS and CAAS results based on split samples from We used linear regression to describe the relationships a comprehensive phylogenetic, ontogenetic, and trophic spec- between THg and fish length and age. Mercury data regressed trum of marine and estuarine fish species, including groupers against fish length were log transformed to meet or improve and sea basses. We included species other than serannids in the limitations of normality and homoscedasticity. Differences this methodological comparison to increase sample size and in THg between the groupers and sea basses and between resolution over the broad range of THg concentrations observed legal and sublegal size-classes of species were analyzed in our data. Owing to the presence of outlier sample data and using the Mann–Whitney rank sum test (alpha = 0.05) for heteroscedastic residuals that increased with an increase in nonparametric data. Legal and sublegal management size the independent variable (THg by CAAS method), we used limits were based on current federal regulations for the Gulf fixed-weight regression methods with weights corresponding of Mexico (FWC 2012; GMFMC 2012). We used the Pearson to the inverse of the presumed variance (SAS 1989). Weights product moment correlation to relate species’ mean THg to the proportional to CAAS−1.0 produced the largest model r2 relative importance of fish prey (%IRI: Actinopterygii) in their and the smallest root mean square error (MSE) values. The diet. resulting regression equation was used to convert all our CAAS results into CVAAS units for subsequent analysis and reporting. RESULTS To assess the dietary importance of fish and invertebrates to common grouper and sea bass in the eastern Gulf of Mexico, CVAAS versus CAAS Regressions we analyzed stomach-content data for red grouper, gag, scamp, We analyzed by each method a total of 1,566 split samples and sand perch collected by the FIM program in the study area that represented 40 marine and estuarine fish species from 16 = from 2004 to 2008. We used these data to relate the general families and a broad range of mercury values (THg(CVAAS) dietary composition of each species with its mean mercury con- 0.01 to 3.2 mg/kg) (Table 1). Families with lower THg (mean < centration. Diet and mercury samples were not always collected THg(CVAAS) 0.25 mg/kg) included the Coryphaenidae (dol- from the same individual; however, all samples were collected phinfishes), Lutjanidae (snappers), Mugilidae (mullets), and across the study region according to scientifically randomized Paralichthyidae (sand flounders). Families with moderate THg = procedures and represent the same segment of the population (THg(CVAAS) 0.25 to 0.5 mg/kg) included the Centropomidae defined by the size ranges of individuals collected. Specifically, (snooks), Serranidae (groupers and sea basses), Pomatomidae each sample (diet or mercury) was collected alternately at ran- (bluefish Pomatomus saltatrix), Carangidae (jacks), and Scom- > domly preselected trawl or hooked-gear transect stations over a bridae (mackerels). Families with elevated THg (THg(CVAAS) stratified depth gradient, and individual fish culled for analysis 0.5 mg/kg) included pelagic and epibenthic predators such as at each station were preselected by their random measurement the Sphyrnidae (hammerhead sharks), Rachycentridae (cobia order. Stomach contents of fish caught by hooked gears at deeper Rachycentron canadum), and members of the Sciaenidae stations, which were often represented by larger individuals, (primarily large adult red drum Sciaenops ocellatus). Mercury were not included in dietary analyses owing to hooked-gear or values for species within these families often extended across these broad mercury concentration categories as evidenced by Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 barotrauma-related bias (e.g., regurgitation, stomach eversion). All stomach samples were processed according to standardized large standard deviations around their mean. FIM procedures. Stomachs were removed, fixed in 10% forma- Regression equations that defined the relationship between lin for at least 48 h, rinsed, and preserved for storage in 50% CVAAS- and CAAS-derived THg indicated that there was isopropyl alcohol. For processing, each stomach was opened and a very strong linear relationship between the two analytical assigned a fullness category (i.e., empty; trace quantity; slightly methods and that CAAS concentrations were 12–13% greater full; full, not distended; distended; distended and overflowing). than those determined by CVAAS methods (Figure 2). The Each stomach was emptied and its contents sorted, identified to relationship between CVAASand CAAS THg for these samples the lowest possible taxonomic level, and measured by volume was defined by the following regression equations: (mm3), and prey items were counted. Owing to differential lev- = . + . × els of prey digestion and identification, we grouped individual THg(CVAAS) 0 001 (0 880 THg(CAAS)); prey items back into their lowest common taxonomic classifi- n = 1,566; r2 = 0.980, and cation before analysis (class for fish, mollusks, annelid worms, = . + . × and sipunculid worms; order for crustaceans and lancets) and THg(CAAS) 0 003 (1 130 THg(CVAAS)); calculated the following metrics for each taxon: percent by n = 1,566; r2 = 0.977. 1278 TREMAIN AND ADAMS
TABLE 1. Summary of fish families and corresponding mercury data used to relate results of cold vapor atomic absorption (CVAAS) and combustion atomic absorption (CAAS) methodologies. Numbers in parentheses represent the number of species analyzed within each family.
Mercury concentration (mg/kg) Mean SD Minimum Maximum Family Number CVAAS CAAS CVAAS CAAS CVAAS CAAS CVAAS CAAS Carcharhinidae (3) 9 0.43 0.46 0.56 0.60 0.13 0.16 1.9 2.04 Sphyrnidae (2) 3 1.49 1.79 1.51 1.66 0.58 0.81 3.2 3.70 Centropomidae (2) 208 0.28 0.33 0.19 0.24 0.05 0.06 1.0 1.55 Serranidae (6) 46 0.28 0.31 0.12 0.15 0.10 0.13 0.65 0.89 Malacanthidae (1) 2 0.22 0.23 0.06 0.07 0.18 0.18 0.26 0.28 Pomatomidae (1) 15 0.33 0.43 0.37 0.50 0.02 0.03 1.1 1.44 Rachycentridae (1) 12 0.93 1.00 0.47 0.56 0.45 0.33 1.8 2.14 Carangidae (4) 41 0.47 0.54 0.33 0.37 0.08 0.09 1.3 1.39 Coryphaenidae (1) 17 0.08 0.09 0.09 0.09 0.03 0.03 0.37 0.36 Lutjanidae (5) 77 0.15 0.18 0.09 0.12 0.05 0.05 0.81 1.10 Gerreidae (1) 2 0.03 0.03 0.01 0.01 0.02 0.02 0.04 0.04 Sciaenidae (3) 909 1.07 1.21 0.64 0.72 0.02 0.02 2.6 3.33 Mugilidae (1) 82 0.01 0.01 <0.01 0.02 <0.01 <0.01 0.02 0.15 Scombridae (6) 120 0.49 0.55 0.35 0.38 0.11 0.11 1.8 1.92 Istiophoridae (1) 1 0.78 0.83 0.78 0.83 0.78 0.83 Paralichthyidae (2) 22 0.14 0.16 0.05 0.06 0.07 0.08 0.24 0.28 Total n 1,566
Mercury and Diet in Groupers and Sea Basses We determined THg in 1,401 serranids collected from 1991 to 2008, which represented 11 grouper and four sea bass species (Table 2). Approximately 74% of these samples (n = 1,031) were collected from 2006 to 2008 during FIM program sampling. The red grouper, gag, and scamp were the most abundant grouper species in our analyses and represent three of the primary fishery species in the eastern Gulf of Mexico (National Marine Fisheries Service, Fisheries Statistics Division, personal communication). The mean THg concentration for all groupers analyzed (Epinephelus, Mycteroperca, Cephalopholis) was 0.22 mg/kg and for all legally harvestable individuals (i.e., individuals not restricted Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 by fishery management size limits or harvest moratoria) was 0.32 mg/kg. Total mercury for individual grouper ranged from 0.01 to 3.30 mg/kg, with both extremes observed in the goliath grouper. The mean THg for nonharvestable and sublegal-sized grouper, which represented more than half of our total samples (57%), was 0.16 mg/kg. The mean THg for legally harvestable individuals ranged from 0.16 mg/kg in graysby Cephalopholis cruentata to 1.01 mg/kg in black grouper M. bonaci.The sand perch was the most abundant sea bass species analyzed; however, the black sea bass Centropristis striata is the only sea bass in our collections for which size restrictions have been FIGURE 2. Regression relationship between cold vapor atomic absorp- placed on the fishery and that supports significant commercial tion spectrometry (CVAAS) and combustion atomic absorption spectrometry (CAAS) mercury results for 1,566 split samples taken from marine and estuar- or recreational fisheries (National Marine Fisheries Service, ine fish species in Florida. Fisheries Statistics Division, personal communication). Total MERCURY IN GROUPERS AND SEA BASSES 1279
TABLE 2. Total mercury concentration, legal harvest status, and length data for groupers and sea basses collected in the eastern Gulf of Mexico from 1991 to 2008. All total mercury levels are reported for dorsal axial muscle tissue as mg/kg wet weight. Total lengths are reported in mm (mm TL). Legal status andharvest restrictions represent management limits in place at the time of publication (FWC 2012; GMFMC 2012). Data for all sea basses, all legal sea basses, all groupers, and all legal groupers is in bold italics.
Mercury concentration (mg/kg) Length (mm TL)
Mini- Maxi- Mini- Maxi- Group and species Status Restrictions n Mean Median SD mum mum Mean mum mum
Sea basses All 358 0.09 0.07 0.07 0.01 0.40 206 76 292 Legal 327 0.09 0.07 0.07 0.01 0.40 207 76 292 Bank sea bass All No limit 40 0.07 0.05 0.08 0.01 0.40 168 76 283 Centropristis ocyurus Black sea bass All 33 0.14 0.12 0.06 0.04 0.31 200 127 292 Centropristis striata Legal ≥254 mm TL 2 0.24 0.24 0.10 0.17 0.31 275 258 292 Sublegal <254 mm TL 31 0.13 0.12 0.05 0.04 0.22 196 127 252 Sand perch All No limit 280 0.09 0.07 0.06 0.01 0.36 214 104 287 Diplectrum formosum Blackear bass All No limit 5 0.03 0.03 0.00 0.03 0.04 120 111 126 Serranus atrobranchus
Groupers All 1,043 0.22 0.16 0.23 0.01 3.30 470 111 2,057 Legal 449 0.32 0.27 0.21 0.04 1.60 605 203 1,373 Graysby All No limit 13 0.16 0.17 0.04 0.07 0.22 283 221 329 Cephalopholis cruentata Rock hind All No limit 6 0.18 0.19 0.05 0.09 0.24 384 308 448 Epinephelus adscensionis Speckled hind All No limit 7 0.20 0.15 0.10 0.12 0.34 510 360 840 Epinephelus drummondhayi Yellowedge grouper All No limit 8 0.23 0.22 0.08 0.13 0.34 398 313 448 Epinephelus flavolimbatus Goliath grouper All No harvest 29 0.64 0.41 0.77 0.01 3.30 906 112 2,057 Epinephelus itajara Red grouper All 559 0.17 0.13 0.11 0.03 0.79 429 163 874 Epinephelus morio Legal ≥508 mm TL 184 0.26 0.22 0.13 0.05 0.79 581 508 874 Sublegal <508 mm TL 372 0.12 0.10 0.07 0.03 0.66 354 163 507 No TL data 3 0.37 0.29 0.24 0.18 0.65 Warsaw grouper All No limit 1 0.24 0.24 0.24 0.24 889 889 889 Epinephelus nigritus Snowy grouper All No limit 7 0.20 0.16 0.19 0.04 0.57 478 203 950 Epinephelus niveatus Black grouper All 13 0.91 0.94 0.45 0.26 1.60 840 505 1,373 Mycteroperca bonaci Legal ≥559 mm TL 11 1.01 1.00 0.41 0.26 1.60 898 581 1,373
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 Sublegal <559 mm TL 2 0.37 0.37 0.07 0.32 0.42 520 505 534 Gag All 351 0.27 0.22 0.18 0.04 1.06 490 111 1,140 Mycteroperca microlepis Legal ≥559 mm TL 175 0.40 0.36 0.17 0.13 1.06 676 561 1,140 Sublegal <559 mm TL 176 0.13 0.12 0.08 0.04 0.39 304 111 558 Scamp All 49 0.19 0.12 0.14 0.04 0.59 481 282 744 Mycteroperca phenax Legal ≥406 mm TL 35 0.24 0.23 0.14 0.07 0.59 533 409 744 Sublegal <406 mm TL 14 0.08 0.06 0.04 0.04 0.17 353 282 405
mercury concentration for all sea bass ranged from 0.01 to Red grouper.—We analyzed THg in 559 red grouper 0.40 mg/kg (mean = 0.09 mg/kg). The mean THg in legally (Table 2). Total length of sampled fish ranged from 163 to harvestable black sea bass was 0.24 mg/kg. The median THg 874 mm (mean = 429 mm). Age was estimated for 261 for all the grouper samples (0.16 mg/kg) was greater than for individuals (TL = 315–871 mm) and ranged from 2 to 17 years the sea basses (0.07 mg/kg) (Mann–Whitney rank sum test: P < (mean [ ± SD] = 5.8 ± 1.9 years), although specimens older 0.001). than 8 years were poorly represented among this subset of 1280 TREMAIN AND ADAMS
individuals (Figure 3). Total mercury concentration of all red Approximately 71% (n = 35) of all scamps collected for grouper ranged from 0.03 to 0.79 mg/kg (mean = 0.17 mg/kg) this study were larger than the current minimum legal size limit and for aged red grouper from 0.051 to 0.79 mg/kg (mean = (406 mm TL, GMFMC 2012). Total mercury concentration 0.18 mg/kg). There was a significant positive relationship for legal-sized fish ranged from 0.07 to 0.59 mg/kg (mean = between THg and fish length (r2 = 0.64, P < 0.001) and 0.24 mg/kg) and for sublegal-sized fish from 0.04 to 0.17 mg/kg between THg and fish age (r2 = 0.45, P < 0.001) (Figure 3). (mean = 0.08 mg/kg). Mercury concentrations in legal-sized Approximately 33% (n = 184) of all red grouper collected scamps were significantly greater than those in sublegal-sized for this study were larger than the current minimum recreational fish (Mann–Whitney rank sum test: P < 0.001). legal size limit (508 mm TL, GMFMC 2012). Total mercury con- We analyzed the diet of 26 scamps, which included 11 fish centrations for legal-sized fish ranged from 0.05 to 0.79 mg/kg (297–516 mm TL, mean = 357 mm) with prey items present in (mean = 0.26 mg/kg) and for sublegal-sized fish from 0.03 their stomach (Table 3). Fish prey were identified as the most to 0.66 mg/kg (mean = 0.12 mg/kg). Mercury concentrations important item (%IRI = 96.4%). Amphipods and several other in legal-sized fish were significantly greater than those in invertebrate groups were also identified in the diet of scamps, sublegal-sized fish (Mann–Whitney rank sum test: P < 0.001). but were much less important. We analyzed the diet of 436 red grouper, which included Sand perch.—We analyzed THg in 280 sand perch (Table 2). 271 fish (169–585 mm TL, mean = 308 mm) with prey items Total length of sampled fish ranged from 104 to 287 mm (mean in their stomach (Table 3). Decapod crustaceans and fish = 214 mm). Age was estimated for 35 individuals (TL = 155– (Actinopterygii) were identified as the most important prey 287 mm) and ranged from 0 to 3 years (mean = 1.3 ± 0.7 years) items (%IRI = 60.6% and 38.7%, respectively). Stomatopods (Figure 3). Total mercury concentration of all sand perch ranged and cephalopods were also present in red grouper stomachs but from 0.01 to 0.36 mg/kg (mean = 0.09 mg/kg) and for aged were much less important (%IRI = 0.3%). sand perch from 0.04 to 0.36 mg/kg (mean = 0.13 mg/kg). Gag.—We analyzed THg in 351 gags (Table 2). Total There was a significant positive relationship between THg and length of sampled fish ranged from 111 to 1,140 mm (mean TL (r2 = 0.55, P < 0. 001) and between THg and fish age (r2 = = 490 mm). Age was estimated for 127 individuals (TL = 0.73, P < 0.001) (Figure 3). There are currently no federal or 134–868 mm) and ranged from 0 to 5 years (mean = 0.9 ± state legal size limits for sand perch in the Gulf of Mexico. 1.2 years), although most of the sampled fish that we aged We analyzed the diet of 18 sand perch (196–339 mm TL, (93%) were less than 3 years old (Figure 3). Total mercury mean = 262 mm), all of which contained prey items in their concentration of all gags ranged from 0.04 to 1.06 mg/kg (mean stomach (Table 3). Decapod crustaceans were identified as the = 0.27 mg/kg), and for aged gags from 0.04 to 0.46 mg/kg most important prey item (%IRI = 76.2%). Mysids and fish (mean = 0.13 mg/kg). There was a significant positive relation- also occurred regularly in the diet of sand perch but were of ship between THg and fish length (r2 = 0.77, P < 0.001) and secondary importance (%IRI = 15% and 6.5%, respectively). between THg and fish age (r2 = 0.72, P < 0.001) (Figure 3). In addition, sand perch consumed small amounts of a diverse Approximately 50% (n = 175) of all gags collected for this assortment of other invertebrate prey items. study were larger than the current minimum legal size limit (559 mm TL, GMFMC 2012). Total mercury concentration Interspecies Mercury–Diet Relationships for legal-sized fish ranged from 0.13 to 1.06 mg/kg (mean = Among the four abundant species for which we analyzed diet 0.40 mg/kg) and for sublegal-sized fish from 0.04 to 0.39 mg/kg composition, mean THg in red grouper, gag, and scamp was (mean = 0.13 mg/kg). Mercury concentrations in legal-sized higher than in sand perch (Kruskal–Wallis analysis of variance gags were significantly greater than those in sublegal-sized fish [ANOVA] on ranks: df = 3, P < 0.001; Dunn’s pairwise multiple < Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 (Mann–Whitney rank sum test: P < 0.001). comparison: P 0.05), but there was not a significant difference We analyzed the diet of 1,142 gags, which included 785 fish in THg between red grouper and scamp. The relationship be- (119–857 mm TL, mean = 232 mm) with prey items in their tween THg and the importance of fish prey (%IRI: Actinoptery- stomach (Table 3). Fish were identified as the most important gii) in these four species is shown in Figure 4. Median THg in prey item (%IRI = 78.3%), followed by decapod crustaceans sand perch, red grouper, and gag increased directly with%IRI (%IRI = 21.6%). for fish (Pearson product moment correlation: n = 3, r = 0.997, Scamp.—We analyzed THg in 49 scamps (Table 2). Total P = 0.0451); however, this relationship was not significant with length of sampled fish ranged from 282 to 744 mm (mean the inclusion of scamp (n = 4, r = 0.630, P = 0.370), which = 481 mm). Age was estimated for 14 individuals (TL = had the highest%IRI for fish but relatively low median THg. 282–590 mm) and ranged from 1 to 12 years (mean = 4.4 ± 3.4 years) (Figure 3). Total mercury concentration of all scamps DISCUSSION ranged from 0.04 to 0.59 mg/kg (mean = 0.19 mg/kg) and for aged scamps from 0.04 to 0.49 mg/kg (mean = 0.15 mg/kg). CVAAS versus CAAS Comparisons There was a significant positive relationship between THg and The two methods used to determine THg in this study fish length (r2 = 0.79, P < 0.001;) and between THg and fish differ primarily in the way mercury is separated and recovered age (r2 = 0.82, P < 0.001) (Figure 3). from the fish tissues. The CVAAS method uses wet chemistry MERCURY IN GROUPERS AND SEA BASSES 1281 Downloaded by [Department Of Fisheries] at 20:08 25 September 2012
FIGURE 3. Linear relationship between (A) total mercury concentration and fish total length (TL) or (B) total mercury concentration and fish age for red grouper, gag, scamp, and sand perch collected in the eastern Gulf of Mexico. 1282 TREMAIN AND ADAMS
TABLE 3. Stomach contents of four grouper and sea bass species collected by the FWC–FWRI in the eastern Gulf of Mexico. Prey species are grouped into their lowest common taxonomic classification. %V = percent by volume; %N = percent by number; %O = percent by occurrence; IRI = index of relative importance; %IRI = IRI standardized by percent.
Fish total length (mm; nonempty) Species Prey taxon %V %N %O IRI %IRI Mean SD Minimum Maximum Red grouper 308 70.7 169 585 n = 436 Decapoda 25.9 60.0 73.8 6,333.9 60.6 Nonempty = 271 Actinopterygii 58.3 28.1 46.8 4,041.0 38.7 Stomatopoda 0.7 4.2 7.4 36.5 0.3 Cephalopoda 14.8 2.0 3.5 26.4 0.3 Gag 232 99.9 119 857 n = 1,142 Actinopterygii 82.1 41.4 81.5 10,058.5 78.3 Nonempty = 785 Decapoda 16.9 43.6 45.9 2775.1 21.6 Scamp 357 64.5 297 516 n = 26 Actinopterygii 94.9 65.4 90.9 14,569.1 96.4 Nonempty = 11 Amphipoda 1.0 19.2 18.2 367.8 2.4 Bivalvia 3.6 3.8 9.1 67.9 0.4 Stomatopoda 0.4 3.8 9.1 38.7 0.3 Decapoda 0.1 3.8 9.1 35.6 0.2 Cephalopoda 0.0 3.8 9.1 35.2 0.2 Sand perch 262 43.3 196 339 n = 18 Decapoda 71.8 50.0 88.9 10,826.2 76.2 Nonempty = 18 Mysida 0.6 29.0 72.2 2138.4 15.0 Actinopterygii 11.2 7.3 50.0 922.0 6.5 Sipunculidea 7.5 1.6 11.1 101.6 0.7 Polychaeta 2.3 1.6 22.2 87.5 0.6 Cephalopoda 4.5 1.6 11.1 67.9 0.5 Isopoda 0.0 1.6 11.1 18.1 0.1 Amphioxiformes 0.4 1.6 5.6 11.0 0.1 Amphipoda 0.1 1.6 5.6 9.3 0.1 Gastropoda 0.9 0.8 5.6 9.2 0.1
techniques to digest tissues and recover mercury, whereas CVAAS. The popularity of the CAAS analysis method has the CAAS method uses a process of thermal decomposition, increased in recent years, in part because it is less expensive, is catalytic reduction, and amalgamation. The linear relationship faster, produces no waste reagents, and requires less laboratory between CVAAS and CAAS methods has been defined in equipment (Cizdziel et al. 2002). Therefore, it is likely that
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 a limited number of studies on freshwater (Cizdziel et al. existing long-term databases will include results from both 2002) and marine (Lowery et al. 2007) species; however, the methodologies as CAAS methods become integrated. The linear applicability of those relationships across a broad range of model we defined relating the CVAAS and CAAS methods marine species was unknown. As with those earlier studies, was based on more than 1,500 samples that represented a broad our CAAS analyses produced greater estimates of THg than phylogenetic, ontogenetic, and trophic spectrum of marine and did our CVAAS methods, but the linear relationships defined estuarine fish species and included a wide range of mercury were different. Cizdziel et al. (2002), who analyzed spiked values. As such, it is a robust model for extrapolating between samples of known mercury concentration, reported values that these two analytical methods and can be used in future studies were approximately 16% greater when determined by CAAS of mercury in marine and estuarine fish species. than by CVAAS. Lowery et al. (2007), analyzing mercury in homogenized fillets of seven marine species from the Gulf Mercury and Diet in Groupers and Sea Basses of Mexico, reported values approximately 18% greater when A broad range of mercury concentrations was associated determined by CAAS than by CVAAS. In our analysis of 40 with the 15 serranid species analyzed in this study. Groupers marine and estuarine species, THg concentrations determined (Epinephelus, Mycteroperca, Cephalopholis) generally con- using CAAS were 12–13% greater than those determined using tained mercury in greater concentrations than did sea basses MERCURY IN GROUPERS AND SEA BASSES 1283
in sample size or fish lengths for each species, but spatial effects related to capture site could have also influenced the observed difference in THg concentrations. Unlike gags, which were collected primarily from central and northwest Florida, most black grouper were collected from southwest Florida and the Florida Keys—areas in which elevated mercury concentrations have been documented for many species (Strom and Graves 2001; Adams et al. 2003; Adams and Onorato 2005). Among species with different growth rates, age-based regression analyses may be more suitable for comparing bioaccumulation rates. Age can serve as an indirect proxy for the underlying factors that may affect mercury accumulation in species over time (i.e., differential influx or assimilation rates associated with ontogenetic shifts in prey selection, habitat, or bioenergetics). Scamps have slower growth rates than gags (Matheson et al. 1986; Hood and Schlieder 1992; Harris et al. 2002), but similar THg concentrations in these congeners may still be expected at defined ages if bioaccumulation rates are also similar. In our investigation, however, the age-based bioac- cumulation rate for gag (defined by the slope of species-specific regression equations) was approximately twice that of scamp. FIGURE 4. Total mercury concentration (lower panel) and relative dietary Spatially, there was some cross-shelf separation in the collection importance of fish prey (upper panel) in four abundant grouper and sea bass locations for these two species. Most of the aged scamps were species from the eastern Gulf of Mexico. The horizontal lines of the box-and- collected in deeper shelf habitats, whereas the majority of aged whisker plot represent the median, 25th and 75th percentiles (box), and 5th and gags were young fish collected near shallow coastal habitats and 95th percentiles (whiskers). Species with different letters in parentheses have adjacent estuaries. Biogeophysical characteristics of estuarine significantly different median mercury values. habitats can promote in situ methylation and biomagnification of mercury (Evans and Crumley 2005; Chen et al. 2008), (Centropristis, Serranus, Diplectrum), which can be related and the different bioaccumulation rates we observed among to differences in their life histories and feeding ecologies. these two sympatric species could reflect the many factors that Although there is overlap in the life history patterns of these influence mercury uptake in their respective habitats. Rates of two broad groups of serranids, in general, the grouper species metal bioaccumulation in species can be related to site-specific we examined live longer and feed at a higher trophic level than geochemical influences and species-specific physiological do the sea basses in this study (Bullock and Smith 1991). Sig- constants such as influx rates from diet or solution, assimilation nificant positive relationships between mercury concentration efficiency, and rates of detoxification and excretion (Luoma and and body size, age, and trophic level have been observed in Rainbow 2005). These physiological constants are unknown many marine fish species from the Gulf of Mexico (Adams et al. for many species, including serranids, and future studies that 2003; Adams and Onorato 2005; Bank et al. 2007; Cai et al. generate these data can help to interpret the ecological effects 2007) and elsewhere. Still, the mean THg concentration in both of mercury in marine systems.
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 the groupers and sea basses was relatively low when compared Our analysis of diet composition among the four most abun- with that in pelagic or upper-level predatory species in the Gulf dant grouper and sea bass species in this study suggested rela- of Mexico, such as mackerels, tunas, and sharks, in which THg tionships between observed THg and feeding ecology. Species often exceeds 1.0 mg/kg (Adams et al. 2003; Cai et al. 2007). that feed principally on fishes (e.g., gag) contained greater mean The direct relationship we observed between fish size and mercury concentrations than did species that feed mainly on in- mercury concentration in serranids has been widely reported vertebrates (e.g., red grouper and sand perch). Members of the for many marine species in the eastern Gulf of Mexico (Adams genus Mycteroperca exhibit morphological and behavioral char- et al. 2003), yet the factors that determine rates of mercury acteristics that would be expected in predominantly piscivorous bioaccumulation in grouper species are poorly understood. species, including elongate body form, canine teeth, and a less Mercury concentration as a function of size differed between intimate association with bottom habitats (Randall 1967). The sympatric species in this study. Among Mycteroperca groupers, black grouper, which had the highest mean THg concentration the gag and black grouper have similar growth rates (Crabtree found in this study, feeds almost exclusively on fish, even during and Bullock 1998), but mean mercury concentrations in the juvenile stage (Randall 1967; Brule´ et al. 2005). However, legal-sized black grouper were considerably greater than in scamp, which also feed almost exclusively on fish, had a mean legal-sized gags. This may be related, in part, to disparities THg that was relatively low among the Mycteroperca groupers. 1284 TREMAIN AND ADAMS
Reasons for this difference are unclear but may be related to dif- important. Wells et al. (2008) found elevated THg concentra- fering mercury levels in available fish prey or species-specific tions in northern Gulf of Mexico juvenile red snapper Lutjanus physiological differences in mercury uptake and assimilation, campechanus in areas in which shrimp are commercially or it may be an artifact of the small sample size for scamp in this trawled, and hypothesized that sediment disturbance as a result study. In contrast to the diet of the groupers, the diet of the sand of trawling increased foraging opportunities at higher trophic perch consisted primarily of crustaceans, and the sand perch levels or increased localized MeHg production and bioavail- exhibited the lowest mean THg among the species for which we ability. We collected grouper and sea bass specimens across a analyzed diet. Although data regarding mercury concentrations wide geographic range using multiple types of gear; however, in invertebrate prey species in Florida waters are limited, some detailed long-term habitat associations could not be assigned to commonly consumed invertebrate prey can contain lower mer- most individuals. Future studies should expand sampling efforts cury concentrations than those found in many fish prey species to obtain a greater number species over a wider bathymetric in the Gulf of Mexico (GBNEP 1992; Jop et al. 1997; Adams distribution, but should also be specifically designed to increase et al. 2003). Food habit studies have not been completed on our understanding of how habitat conditions, ontogenetic shifts many grouper and sea bass species in the Gulf of Mexico, and in prey availability and selection, or physiological determinants studies that have been done for most serranids provide only lim- interact to influence mercury levels in serranid species. ited or qualitative information (Bullock and Smith 1991, and references therein). Many studies of groupers have provided Fishery and Fish Health Implications data on stomach content that are outdated, were obtained using Recent work suggests that mercury may cause sublethal baited gear that may have introduced bias, or involved sampling effects in marine (Adams et al. 2010) and freshwater (Sandhein- designs that were spatially and temporally restricted. For exam- rich and Wiener 2011) fish species. Sublethal effects of mercury ple, the only diet study of legal-sized gag in the Gulf of Mexico have been more extensively studied in freshwater fish species, was conducted nearly 30 years ago using baited hook-and-line and cell damage, tissue damage, changes in biochemical gear (Naughton and Saloman 1985). Still, our results correspond processes, and reduced reproductive success were thought to well with these few published studies despite the prominence of occur in fish with axial muscle methylmercury concentrations relatively small individuals in our samples. Future diet studies of between 0.5 and 1.2 mg/kg wet weight (Sandheinrich and should be expanded to include a broader range of size-classes Wiener 2011). Adams et al. (2010) observed possible sublethal and incorporate stable isotopes to characterize long-term feed- effects (e.g., liver and kidney lesions, alteration of blood bio- ing habits and identify shifts in trophic level through ontogeny. chemistry, and negative neurochemical changes) in marine fish When combined with mercury data for both target species and in the southeastern United States at mean axial muscle mercury their prey, these comprehensive diet data can help clarify the re- concentrations of up to 0.56 mg/kg wet weight, a concentration lationship between mercury bioaccumulation and trophic ecol- that is within the suggested effects range for fish from freshwa- ogy in these important fishery species. ter systems. Although the mean THg concentration for all but Many serranids, including red grouper, gag, and scamp, are two species examined in this study (black grouper and goliath protogynous hermaphrodites, changing from female to male at grouper) were well below 0.5 mg/kg, the maximum observed larger sizes (Hood and Schlieder 1992; Harris et al. 2002), and THg in 6 of the 15 species examined exceeded this possible the sand perch is a synchronous hermaphrodite (Bullock and threshold concentration, and in three species maximum THg Smith 1991). The physiological influences of hermaphroditism was greater than 1.0 mg/kg (goliath grouper, gag, and black on mercury bioaccumulation are unknown; however, other grouper). Specific data on how elevated mercury concentrations aspects of reproduction (e.g., maternal transmission and may affect grouper and sea bass at the individual or population
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 reproductive-based elimination of mercury) are known to be im- level are not available, and further study is needed. portant in mercury cycling for marine fish species with differing Mercury-related data in grouper species have often been reproductive strategies (Walker 1976; Adams and McMichael broadly classified under a general “grouper” category (Sunder- 1999; Alvarez et al. 2006). Most of our samples were collected land 2007; USFDA 2012), in part owing to the manner in which in relatively shallow nearshore and offshore areas, which proba- these fish are marketed by commercial retailers. We found bly resulted in undersampling of the larger and older age-classes considerable interspecific and intraspecific variability in THg of species that migrate to deeper reef habitats offshore as they within groupers of the eastern Gulf of Mexico. Collectively, grow. Consequently, few male specimens were identified in our the groupers examined in this study contained relatively small samples, and comparing sex-specific patterns in mercury bioac- mean THg concentrations; however, our samples for commonly cumulation for these species was not possible in this study. targeted fishery species did not include a full range of upper We defined a significant positive relationship between THg size- and age-classes. Still, nearly one-quarter (23%) of all the and fish size and between THg and age; however, considerable grouper samples we analyzed contained mercury concentrations THg variability within species remained unexplained by these that exceeded the EPA’s methylmercury consumption guidance metrics. Whereas large-scale spatial effects may account for criterion for humans (0.3 mg/kg, USEPA 2001). Mercury con- some of the variability, localized habitat effects may also be centrations increased in groupers as fish attained greater size MERCURY IN GROUPERS AND SEA BASSES 1285
and age. All regulated grouper species in the Gulf of Mexico are offspring in Atlantic croaker (Micropogonias undulatus). Aquatic Toxicology currently managed with minimum size limits, which select for 80:329–337. a fishery consisting of the largest and oldest individuals. There- Baeyens, W., M. Leermakers, T. Papina, A. Saprykin, N. Brion, J. Noyen, M. De Gieter, M. Elskens, and L. Goeyens. 2003. Bioconcentration and fore, public health consumption advisories based on a broad biomagnification of mercury and methylmercury in North Sea and Scheldt classification for the species group may not adequately inform estuary fish. Archives of Environmental Contamination and Toxicology 45: at-risk human populations. Although the fish are commercially 498–508. marketed as “grouper,” recreational fishers who must identify Balcolm, P. H., W. F. Fitzgerald, G. M. Vandal, C. H. Lamborg, C. S. Langer, K. individual species for legal harvest, as well as fisheries and R. Rolfhus, and C. H. Hammerschmidt. 2004. Mercury sources and cycling in the Connecticut River and Long Island Sound. Marine Chemistry 90:53–74. human health agencies that develop consumption advisories, Bank, M. S., E. Chesney, J. P. Shine, A. Maage, and D. B. Senn. 2007. Mercury can benefit from the regional species- or size-specific mercury bioaccumulation and trophic transfer in sympatric snapper species from the information from this study. Further efforts to collect mercury Gulf of Mexico. Ecological Applications 17:2100–2110. and other biological data from larger individuals and older Bloom, N. S. 1992. On the chemical form of mercury in edible fish and ma- age-classes of these fishery species are necessary if we are to un- rine invertebrate tissue. Canadian Journal of Fisheries and Aquatic Sciences 49:1010–1017. derstand the full range of fishery and human health implications. Bloom, N. S., G. A. Gill, S. Cappellino, C. Dobbs, L. McShea, C. Driscoll, R. Mason, and J. Rudd. 1999. Speciation and cycling of mercury in Lavaca Bay, ACKNOWLEDGMENTS Texas, sediments. Environmental Science and Technology 33:7–13. We thank the Florida Department of Environmental Protec- Brule,´ T., E. Puerto-Novelo, E. Perez-D´ ´ıaz, and X. Renan-Galindo.´ 2005. Diet tion Bureau of Laboratories for the analyses that helped make composition of juvenile black grouper (Mycteroperca bonaci) from coastal nursery areas of the Yucatan´ Peninsula, Mexico. Bulletin of Marine Science this study possible. We greatly appreciate the efforts of staff 77:441–452. from the FWC–FWRI Fisheries-Independent Monitoring (FIM) Bullock, L. H., and G. B. Smith. 1991. Seabasses (Pisces: Serranidae). Memoirs Program and Fisheries-Dependent Monitoring Program for their of the Hourglass Cruises 8(Part 2). assistance in collecting fish and processing samples, and staff of Cai, Y., J. R. Rooker, G. A. Gill, and J. P. Turner. 2007. Bioaccumulation of the FWC–FWRI Marine Fisheries Research section for age data. mercury in pelagic fishes from the northern Gulf of Mexico. Canadian Journal of Fisheries and Aquatic Sciences 64:458–469. In addition, D. Chagaris and the FIM Gut Lab provided excel- Chen, C., A. Amirbahman, N. Fisher, G. Harding, C. Lamborg, D. Nacci, and lent assistance with diet data, and R. Kiltie (FWC–FWRI Cen- D. Taylor. 2008. Methylmercury in marine ecosystems: spatial patterns and ter for Biostatistics and Modeling) provided valuable statistical processes of production, bioaccumulation, and biomagnification. Ecohealth assistance. Helpful suggestions for improving the manuscript 5:399–408. were also offered by D. Blewett, D. Chagaris, A. Collins, B. Choy, C. M. Y., C. W. K. Lam, L. T. F. Cheung, C. M. Briton-Jones, L. P. Cheung, and C. J. Haines. 2002. Infertility, blood mercury concentrations Crowder, G. Onorato, and three anonymous reviewers. This and dietary seafood consumption: a case–control study. BJOG: International study was supported in part by funding from the Department Journal of Obstetrics and Gynaecology 109:1121–1125. of the Interior, U.S. Fish and Wildlife Service, Federal Aid in Cizdziel, J. V., T. A. Hinners, and E. M. Heithmar. 2002. Determination of total Sport Fish Restoration, Project F-43, the U.S. Environmental mercury in fish tissues using combustion atomic absorption spectrometry with Protection Agency’s Gulf of Mexico Program, and by state of gold amalgamation. Water, Air, and Soil Pollution 135:355–370. Cortes,´ E. 1997. A critical review of methods of studying fish feeding based on Florida saltwater recreational fishing license monies. analysis of stomach contents: application to elasmobranch fishes. Canadian Journal of Fisheries and Aquatic Sciences 54:726–738. REFERENCES Crabtree, R. E., and L. H. Bullock. 1998. Age, growth, and reproduction of the Adams, D. H. 2010. Mercury in wahoo, Acanthocybium solandri, from offshore black grouper, Mycteroperca bonaci, in Florida waters. U.S. National Marine waters of the southeastern United States and Bahamas. Marine Pollution Fisheries Service Fishery Bulletin 96:735–753. Bulletin 60:139–151. Evans, D. W., and P. H. Crumley. 2005. Mercury in Florida Bay fish: spatial Adams, D. H., and R. H. McMichael Jr. 1999. Mercury levels in four species distribution of elevated concentrations and possible linkages to Everglades
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 of sharks from the Atlantic coast of Florida. U.S. National Marine Fisheries restoration. Bulletin of Marine Science 77:321–345. Service Fishery Bulletin 97:372–379. FWC (Florida Fish and Wildlife Conservation Commission). 2012. Florida Adams, D. H., and R. H. McMichael Jr. 2001. Mercury levels in marine and es- saltwater recreational fishing regulations. FWC, Tallahassee. Available: tuarine fishes of Florida. Florida Marine Research Institute Technical Report myfwc.com/fishing/saltwater/regulations/. (January 2012). TR-6. GBNEP (Galveston Bay National Estuary Program). 1992. Toxic contami- Adams, D. H., R. H. McMichael Jr., and G. E. Henderson. 2003. Mercury levels nant characterization of aquatic organisms in Galveston Bay: a pilot study. in marine and estuarine fishes of Florida: 1989–2001, 2nd edition, revised. GBNEP, Galveston, Texas. Florida Marine Research Institute Technical Report TR-9. Gilmour, C. C., and E. A. Henry. 1991. Mercury methylation in aquatic systems Adams, D. H., and G. V. Onorato. 2005. Mercury concentrations in red drum, affected by acid deposition. Environmental Pollution 71:131–169. Sciaenops ocellatus, from estuarine and offshore waters of Florida. Marine GMFMC (Gulf of Mexico Fishery Management Council). 2012. Recreational Pollution Bulletin 50:291–300. fishing regulations for the Gulf of Mexico federal waters. GMFMC, Tampa, Adams, D. H., C. Sonne, N. Basu, R. Dietz, D. H. Nam, P. S. Leifsson, and Florida. Available: www.gulfcouncil.org/fishing regulations/index.php. A. L. Jensen. 2010. Mercury contamination in spotted seatrout, Cynoscion (March 2012). nebulosus: an assessment of liver, kidney, blood, and nervous system health. Grieb, T. M., G. L. Bowie, C. T. Driscoll, S. P. Gloss, C. L. Schofield, and Science of the Total Environment 408:5808–5816. D. B. Porcella. 1990. Factors affecting mercury accumulation in fish in Alvarez,M.C.,C.A.Murphy,K.A.Rose,I.D.McCarthy,andL.A.Fuiman. the upper Michigan peninsula. Environmental Toxicology and Chemistry 9: 2006. Maternal body burdens of methylmercury impair survival skills of 919–930. 1286 TREMAIN AND ADAMS
Hammerschmidt, C. R., and K. L. Bowman. 2012. Vertical methylmercury NRC (National Research Council). 2000. Toxicological effects of methylmer- distribution in the subtropical North Pacific Ocean. Marine Chemistry 132– cury. National Academy Press, Washington, D.C. 133:77–82. Pinkas, L., M. S. Oliphant, and I. L. K. Iverson. 1971. Food habits of albacore, Hammerschmidt, C. R., and W. F. Fitzgerald. 2004. Geochemical controls on bluefin tuna, and bonito in Californian waters. California Department of Fish the production and distribution of methylmercury in near-shore marine sedi- and Game, Fish Bulletin 152. ments. Environmental Science and Technology 38:1487–1495. Randall, J. E. 1967. Food habits of reef fishes of the West Indies. Studies in Hammerschmidt, C. R., and W.F. Fitzgerald. 2006. Bioaccumulation and trophic Tropical Oceanography 5:665–847. transfer of methylmercury in Long Island Sound. Archives of Environmental Sandheinrich, M. B., and J. G. Wiener. 2011. Methylmercury in freshwater fish: Contamination and Toxicology 51:416–424. recent advances in assessing toxicity of environmentally relevant exposures. Harris, P. J., D. M. Wyanski, D. B. White, and J. L. Moore. 2002. Age, growth, Pages 169–190 in N. Beyer and J. Meador, editors. Environmental contami- and reproduction of scamp, Mycteroperca phenax, in the southwestern North nants in wildlife: interpreting tissue concentrations, 2nd edition. Taylor and Atlantic, 1979–1997. Bulletin of Marine Science 70:113–132. Francis, Boca Raton, Florida. Heyes, A., R. P. Mason, E. H. Kim, and E. Sunderland. 2006. Mercury methy- SAS (Statistical Analysis Systems). 1989. SAS/STAT user’s guide, version 6, lation in estuaries: insights from using measuring rates using stable mercury 4th edition, volume 2. SAS Institute, Cary, North Carolina. isotopes. Marine Chemistry 102:134–147. Shapiro, D. Y. 1987. Reproduction in groupers. Pages 295–327 in J. J. Polovina Hightower, J. M., and D. Moore. 2003. Mercury levels in high-end consumers and S. Ralston, editors. Tropical snappers and groupers: biology and fisheries of fish. Environmental Health Perspectives 111:604–608. management. Westview Press, Boulder, Colorado. Hood, P. B., and R. A. Schlieder. 1992. Age, growth, and reproduction of gag, Strom, D. G., and G. A. Graves. 2001. A comparison of mercury in estuarine fish Mycteroperca microlepis (Pisces: Serranidae), in the eastern Gulf of Mexico. between Florida Bay and the Indian River Lagoon, Florida, USA. Estuaries Bulletin of Marine Science 51:337–352. 24:597–609. IAEA (International Atomic Energy Agency). 2004. Sediment distribution coef- Sunderland, E. M. 2007. Mercury exposure from domestic and imported es- ficients and concentration factors for biota in the marine environment. IAEA tuarine and marine fish in the U.S. seafood market. Environmental Health Technical Report Series 422. Perspectives 115:235–242. Jop, K. M., R. C. Biever, J. L. Hoberg, and S. P. Shepard. 1997. Analysis of Sunderland, E. M., D. P. Krabbenhoft, J. W. Moreau, S. A. Strode, and W. metals in blue crabs, Callinectes sapidus, from two Connecticut estuaries. M. Landing. 2009. Mercury sources, distribution, and bioavailability in the Bulletin of Environmental Contaminants and Toxicology 58:311–317. North Pacific Ocean: insights from data and models. Global Biogeochemical Kraepiel, A. M. L., K. Keller, H. B. Chin, E. G. Malcolm, and F. M. M. Morel. Cycles 23:article GB2010. DOI: 10.1029/2008GB003425. 2003. Sources and variations of mercury in tuna. Environmental Science and Trudel, M., and J. B. Rasmussen. 1997. Modeling the elimination of mercury Technology 37:5551–5558. by fish. Environmental Science and Technology 31:1716–1722. Lawrence, A. L., and R. P. Mason. 2001. Factors controlling the bioaccumula- USEPA (U.S. Environmental Protection Agency). 1991. Determination of tion of mercury and methylmercury by the estuarine amphipod Leptocheirus mercury in tissues by cold vapor atomic absorption spectrometry: method plumulosus. Environmental Pollution 111:217–231. 245.6 (revision 2.3). USEPA, Environmental Monitoring Systems Labora- Lowery, T. A., R. S. Winters, and E. S. Garrett III. 2007. Comparison of total tory, Cincinnati, Ohio. mercury determinations of fish fillet homogenates by thermal decomposition, USEPA (U.S. Environmental Protection Agency). 1998. Mercury in solids and amalgamation, and atomic absorption spectrophotometry versus cold vapor solutions by thermal decomposition, amalgamation, and atomic absorption atomic absorption spectrophotometry. Journal of Aquatic Food Product Tech- spectrometry: draft method 7473. USEPA, Washington, D.C. nology 16:5–15. USEPA (U.S. Environmental Protection Agency). 2001. Water quality criteria Luoma, S. N., and P. S. Rainbow. 2005. Why is metal bioaccumulation so for the protection of human health: methylmercury. USEPA, Report EPA- variable? biodynamics as a unifying concept. Environmental Science and 823-R-01-001, Washington, D.C. Technology 39:1921–1931. USFDA (U.S. Food and Drug Administration). 2012. Mercury levels in Mason, R. P., and G. R. Sheu. 2002. Role of the ocean in the global commercial fish and shellfish (1990–2010). USFDA, Washington, D.C. mercury cycle. Global Biogeochemical Cycles 16(4):article 1093. DOI: Available: www.fda.gov/Food/FoodSafety/Product-SpecificInformation/Sea 10.1029/2001GB001440. food/FoodbornePathogensContaminants/Methylmercury/ucm115644.htm. Matheson, R. H., III, G. R. Huntsman, and C. S. Manooch. 1986. Age, growth, (March 2012). mortality, food, and reproduction of the scamp, Mycteroperca phenax,col- Walker, T. I. 1976. Effects of species, sex, length and locality on the lected off North Carolina and South Carolina. Bulletin of Marine Science mercury content of school shark Galeorhinus australis (Macleay) and 38:300–312. gummy shark Mustelus antarcticus Guenther from south-eastern Aus-
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 Mathews, T., and N. S. Fisher. 2008. Evaluating the trophic transfer of cadmium, tralian waters. Australian Journal of Marine and Freshwater Research 27: polonium, and methylmercury in an estuarine food chain. Environmental 603–616. Toxicology and Chemistry 27:1093–1101. Wells, R. J. D., M. M. Chumchal, and J. H. Cowan Jr. 2008. Effect of trawling Naughton, S. P., and C. H. Saloman. 1985. Food of gag (Mycteroperca mi- and habitat on mercury concentration in juvenile red snapper from the north- crolepis) from North Carolina and three areas of Florida. NOAA Technical ern Gulf of Mexico. Transactions of the American Fisheries Society 137: Memorandum NMFS-SEFC-160. 1839–1850. This article was downloaded by: [Department Of Fisheries] On: 25 September 2012, At: 20:08 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Quantifying Salmon-Derived Nutrient Loads from the Mortality of Hatchery-Origin Juvenile Chinook Salmon in the Snake River Basin Dana R. Warren a b & Michelle M. McClure a a National Oceanic and Atmospheric Administration Northwest Fisheries Science Center, 2725 Montlake Boulevard, Seattle, Washington, 98112, USA b Department of Fisheries and Wildlife, Oregon State University, 104 Nash Hall, Corvallis, Oregon, 97331, USA Version of record first published: 15 Aug 2012.
To cite this article: Dana R. Warren & Michelle M. McClure (2012): Quantifying Salmon-Derived Nutrient Loads from the Mortality of Hatchery-Origin Juvenile Chinook Salmon in the Snake River Basin, Transactions of the American Fisheries Society, 141:5, 1287-1294 To link to this article: http://dx.doi.org/10.1080/00028487.2012.686950
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. Transactions of the American Fisheries Society 141:1287–1294, 2012 American Fisheries Society 2012 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2012.686950
NOTE
Quantifying Salmon-Derived Nutrient Loads from the Mortality of Hatchery-Origin Juvenile Chinook Salmon in the Snake River Basin
Dana R. Warren*1 and Michelle M. McClure National Oceanic and Atmospheric Administration Northwest Fisheries Science Center, 2725 Montlake Boulevard, Seattle, Washington 98112, USA
environments, anadromous salmonids accumulate up to 90% Abstract of their adult body weight in the ocean and return to spawn Hatchery supplementation of anadromous salmon is extensive and die in their natal streams where they can add a substantial across the Pacific Northwest region with millions of juvenile salmon amount of nutrients to otherwise nutrient-poor headwaters. The stocked annually. The influence of hatchery-origin fish as prey items in recipient ecosystems has been explored, but influences of subsidies that these salmon-derived nutrients (SDN) provide to these fish on broader stream nutrient dynamics has not been well- recipient streams can increase growth and biomass of periphy- studied. Salmon-derived nutrients (SDN) associated with the mor- ton, invertebrates, resident fish, juvenile anadromous fish, and tality of adult anadromous salmon provide key subsidies to fresh- riparian vegetation (Cederholm et al. 1999; Gende et al. 2002; water habitats. While a number of studies have estimated current Naiman et al. 2002; Schindler et al. 2003). Given its potential and historic SDN loading from returning wild salmon, SDN con- tributions from the mortality of hatchery-origin juveniles (many to boost stream productivity, a number of studies have worked of which die in the stream prior to emigration) remains largely un- to quantify SDN input from returning adult salmon and to de- known. We conducted a mass balance analysis of SDN input and ex- termine how declines in salmon stocks are likely to influence port via hatchery activities (stocking and broodstock collection) in associated stream ecosystems (Lyle and Elliott 1998; Jonsson the Snake River watershed. Using Chinook salmon Oncorhynchus and Jonsson 2003; Nislow et al. 2004; Scheuerell et al. 2005; tshawytscha as a model species, we accounted for yearly SDN in- put (via hatchery-origin juvenile fish mortality) and export (via Verspoor et al. 2010). In the U.S. Pacific Northwest, nutrient broodstock collections and presmolt growth) over 6 years (2002– flux estimates have focused almost exclusively on wild salmon 2007) in the portion of the Snake River upstream from Lower populations (Cederholm et al. 1999; Scheuerell et al. 2005; Granite Dam accessible to anadromous fish. In the year with high- Moore et al. 2007), leaving the effects of SDN contributions est smolt mortality (2003), hatchery-origin smolt mortality pro- from hatcheries largely unexplored. vided a net input of SDN equivalent to approximately 8,100 return- ing adults. In the year with lowest smolt mortality (2004), hatchery An analysis of SDN budgets that includes hatchery activ- activities collectively yielded a net loss of nutrients. Although the ities must consider a number of additional pathways above Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 mass of SDN from hatchery-origin smolts may be presented in and beyond those included in an analysis of wild fish alone adult equivalencies, functional influences of SDN from hatchery (Figure 1). For example, in managed systems, returning adult smolt mortality are likely to differ. Salmon-derived nutrients from fish are collected for broodstocks to produce juvenile fish that hatcheries enter food webs through largely piscivorous pathways whereas SDN from adult carcasses enter food webs through mul- are out-planted 1 to 2 years later. These broodstock collec- tiple pathways at multiple trophic levels. The SDN from hatchery- tions represent a loss of SDN from streams, but mortality of origin smolts probably influence different components of the food the out-planted juveniles prior to their migration to the ocean web more than do adult carcasses and have the potential to more represents a novel input of SDN to the streams in which fish directly affect predator populations. are stocked. These management-associated inputs and losses of SDN in streams have not been quantified in Pacific North- Nutrient availability can influence fundamental charac- west ecosystems. The relationship between inputs via hatchery teristics and functions of an ecosystem. In many stream fish mortality and export via broodstock collections establish the
*Corresponding author: [email protected] 1Present address: Department of Fisheries and Wildlife, Oregon State University, 104 Nash Hall, Corvallis, Oregon 97331, USA. Received May 20, 2011; accepted April 16, 2012 1287 1288 WARREN AND MCCLURE
FIGURE 1. Conceptual model for transport of salmon-derived nutrients (SDN) in (A) natural and (B) managed stream systems. The aspects of this process addressed in the current mass balance analysis are represented by the darkly shaded boxes. Transport of SDN in the managed system assumes that age-1 hatchery smolts emigrate shortly after stocking and that only age-0 fish have time to add biomass within the system prior to emigration.
potential for a net loss or a net gain of SDN as a result of hatchery modified and extensively managed. We focus on Chinook activities. In this study we focus on quantifying nutrient fluxes salmon in this study because (1) they are an economically and of nitrogen (N) and phosphorus (P) associated with broodstock ecologically important species that historically dominated much collections and hatchery smolt mortality from a mass balance of the Snake River watershed but have experienced steady and perspective. We do not address the food web effects of stock- significant population declines over the last century (Petrosky ing, which have been well reviewed in earlier papers (Eby et al. et al. 2001), and (2) the majority of salmon stocking in the 2006; Kostow 2009; Gozlan et al. 2010; Ellis et al. 2011). portion of the Snake River accessible to anadromous fish is con- Stocking has become a common fisheries management tool ducted with juvenile Chinook salmon (although there are con- to enhance the recovery of endangered species and to create siderable stocking efforts for other anadromous salmonids in the fishable populations in small streams, large rivers, and lakes. basin as well). Within the Snake River basin, Chinook salmon Nutrient input from stocking is generally assumed to be neg- are separated into two Evolutionarily Significant Units (ESUs; ligible; however, studies that have tested this assumption by distinct population segments of Chinook salmon that are treated quantifying and evaluating these inputs are rare (Nislow et al. separately under the U.S. Endangered Species Act). These
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 2004). In the present study, we evaluate potential nutrient in- two ESUs—the Snake River Spring–Summer (SRSS) Chinook put or loss associated with stocking in the Snake River basin, salmon ESU (a stream-type lineage) and the Snake River Fall a system with high stocking effort and hatchery-origin smolt (SRF) Chinook salmon ESU(an ocean-type lineage)—are both mortality rates that regularly exceed 40%. listed as threatened under the Endangered Species Act, and are subject to separate but extensive hatchery supplementation. Many of the stocks currently used for these efforts are derived METHODS from and belong to the ESUs. Because we focus our annual es- We evaluated SDN loading and loss associated with removal timate of SDN from hatchery activities on basin-wide potential and stocking of Chinook salmon Oncorhynchus tshawytscha in for SDN loading, we do not distinguish between spring–summer the portions of the Snake River upstream of Lower Granite Dam versus fall Chinook salmon ESUs in the overall mass balance that are accessible to anadromous fish, which encompass parts of analysis. In our site-specific analysis of SDN input, we focused Idaho, Oregon, and Washington. This tributary to the Columbia only on stocking of the spring–summer ESU. The majority of River contains numerous anadromous fish populations, many juvenile Chinook salmon stocked in the Snake River are from of which are supplemented by hatcheries. The main stem of the spring–summer ESU, and stocking events for these fish tend the Snake River and many of its larger tributaries are a highly to be the largest biomass additions. NOTE 1289
We used data on juvenile fish passage collected at the up- River system (Scheuerell et al. 2005). We did, however, account permost passable dam on the Snake River, Lower Granite Dam for growth of age-0 fish stocked upstream from LGD. The to- (LGD), located at river kilometer 172 (river mile 107) of the tal mass of SDN lost from the system owing to growth and Snake River in Garfield, Washington. Chinook salmon that subsequent emigration of these hatchery-origin age-0 Chinook spawn in the Snake River upstream from LGD pass a total salmon from the SRF Chinook salmon ESU was calculated by of eight large hydroelectric dams, four on the main stem of the multiplying mean accrued mass per fish in that year by the Columbia River and four on the lower main stem of the Snake number of hatchery-origin age-0 individuals that passed LGD River. These dams and associated passage facilities create pas- in the fall and the fraction of total juvenile Chinook biomass sage bottlenecks, but they are also ideal points for counting fish that is either N or P. We used age-0 summer survival estimates (Scheuerell et al. 2005). A subset of the emigrating smolts that from FPC reports and initial stocking densities to estimate the pass through each system are counted daily and this value is number of age-0 hatchery-origin fish that migrated in the fall. used to estimate total passage through each dam facility. There We assumed a 41% increase in smolt biomass between stock- are 10 hatcheries and 31 stocking sites upstream of LGD; in ing and out-migration based on data from Connor et al. (2008). recent years between 12.4 million and 15.5 million juvenile Chinook salmon eggs are rarely out-planted so we did not in- Chinook salmon have been out-planted annually. The majority clude hatchery-origin eggs in this mass balance study as was of juveniles (74% to 87% by number and 91% to 97% by mass) done in Nislow et al. (2004). are stocked as age-1 smolts. We identified two primary areas of uncertainty in this analy- Mass balance calculations.—We calculated the net annual sis associated with (1) collection efforts in hatchery broodstock flux of SDN to the Snake River associated with stocking juvenile traps and (2) the fate of excess and spawned out salmon car- salmon upstream of LGD using a mass balance approach. The casses from hatcheries. Some hatcheries collect additional fish annual hatchery-activity SDN flux was estimated from stocking at their traps above and beyond broodstock requirements in or- mortality (input), broodstock collection (output), and the growth der to reduce interaction between wild and hatchery-origin adult of age-0 hatchery fish (output) in each of six consecutive years fish. The number of fish collected in excess of broodstock needs (2002–2007) (Figure 1). We focused here on direct input and is inconsistently reported. This represents an unaccounted-for output processes associated with out-planting and broodstock removal of SDN from streams in our analysis. However, as the collection in the reaches upstream of LGD (dark gray boxes value of nutrient subsidies from salmon has been increasingly in Figure 1b). We therefore did not include input via returning promoted, a growing number of hatcheries are making carcasses adults of hatchery origin in this analysis. Estimates of down- (both excess fish and “spawned-out individuals”) available for stream passage of hatchery-origin smolts at LGD, stocking lo- nutrient addition back into the stream if they are pathogen free cations upstream of LGD, stocking levels at each stocking site, (Idaho Department of Fish and Game et al. 2008). Unfortu- the mean mass of hatchery smolts each year, and smolt mortal- nately, the number of carcasses returned to the stream is also ity were obtained from The Fish Passage Center (FPC) website unavailable in aggregate and inconsistently reported in individ- (www.fpc.org/) (FPC 2007) or were calculated from data pro- ual reports. This represents an unaccounted-for addition back vided by the FPC. A complete census of hatchery broodstock to the stream in these systems. In the absence of useable data collections that includes all sites in all years is not available for for either excess adult removal or carcass additions back to Snake River hatcheries upstream of LGD. Therefore, in order the stream, we assume that these processes balanced out in to estimate broodstock collection numbers, we used the number our analysis with broodstock carcasses returned to the stream, of hatchery smolts that were out-planted in a given year and thereby compensating for excess adult collection at broodstock back-calculated the number of adults that would be needed to traps.
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 produce this number of hatchery fish from collections 2 years Nutrients may also be lost from the system at stocking sites prior. For this estimate we assumed that, on average, each mat- via bioturbation of sediments in the presence of a high den- ing pair produced approximately 3,200 smolts for release in the sity of fish (Moore et al. 2004, 2007; Holtgrieve and Schindler Snake River (one adult per 1,600 stocked smolts at the target 1:1 2011). Juvenile fish do not engage in redd construction so their sex ratio used by most hatcheries) (Idaho Department of Fish influence via bioturbation is likely to be lower than is it for and Game et al. 2008). The estimated mass of an individual adult fish; however, when stocked at high densities, bioturba- adult Chinook salmon collected for broodstocks was taken from tion effects are clearly possible. For the purpose of this analysis, Scheuerell et al. (2005) for full-sized adults (5.5 kg). The pro- we did not consider export via bioturbation as its influence is portion of adult and juvenile Chinook salmon that is P (adult, highly dependent upon stream size, stream substrate conditions, 0.0038; juvenile, 0.0043) and N (0.03 for adult and juvenile) and fish densities. were taken from Scheuerell et al. (2005), Moore et al. (2004), Because broodstock collections and the stocking of hatchery and Thomas et al. (2003). smolts occur at different times of year, an annual mass bal- We assumed relatively rapid emigration and therefore no ance may not reflect realized nutrient inputs to the system via significant biomass accumulation before passing LGD for fish stocking. In addition, stocking itself is conducted across a range stocked at age 1 (primarily SRSS smolts) within the Snake of stream sizes with varying degrees of potential dilution of 1290 WARREN AND MCCLURE
nutrients from hatchery-origin smolts. We therefore conducted growth of age-0 fish together were thus greater than inputs as- an additional analysis that focused specifically on SDN input sociated with smolt mortality. Nutrient inputs during years with via stocking mortality across the 31 sites where spring–summer low smolt survival appear substantial on an annual basis, and P Chinook salmon smolts are stocked in spring. The majority of inputs can be equivalent to as many as 8,100 returning adults in a spring–summer Chinook salmon stocking effort in the Snake year. According to FPC estimates, between 38,000 and 110,000 River occurs in the spring. For a representative year (2006), we returning adult Chinook salmon (wild and hatchery-origin com- estimated potential increases in stream nutrient concentrations bined) passed LGD annually from 2002 to 2007. On an annual for the duration of stocking, assuming 10% mortality at the input basis, the maximum SDN loads from juvenile hatchery Chi- site over the duration of the stocking effort (equation 1). nook salmon could therefore represent a substantial increase in SDN upstream of LGD (increases of 7% to 21% above P loads Ns · ms · Ms · Ps provided by all adults in years with a net input of hatchery-origin [nutrient] = (1) Q · 86,000 · d SDN). It is important to note, however, that while the mass of Where [nutrient] represents the potential change in nutrient hatchery-origin SDN may be relatively large, the ecological ef- concentration due to hatchery-origin smolt mortality at a stock- fects of SDN input from hatchery smolts are unlikely to match ing site; Ns indicates the total number of smolts stocked at a SDN input from adult mortality on a 1:1 basis. Many of the given location; ms indicates the mean mass of fish stocked; Ms key pathways for nutrient assimilation into the stream ecosys- indicates the hatchery smolt mortality rate (as a proportion) at tem differ between SDN from hatchery-origin juvenile Chinook the stocking site—we used a value of 0.1 in Figure 3; Ps in- salmon and SDN from returning adult Chinook salmon that dicates the proportion of stocked fish mass represented by the spawn and die in the stream (including adults of hatchery origin nutrient of interest (e.g., nitrogen or phosphorous at 0.03 and or wild origin). Juvenile fish are often eaten whole by individual 0.0043, respectively); Q indicates discharge in cubic meters per predators at or near the top trophic position in the stream (Muir second during stocking; 86,400 accounts for the number of sec- et al. 2001; Kostow 2009; Monzyk et al. 2009). In contrast, onds in a day and d indicates the number of days over which adult carcasses generally support organisms at a broad range stocking occurred. Although total mortality estimates for Chi- of trophic positions (Chaloner et al. 2002; Wipfli et al. 2010). nook salmon smolts between stocking and LGD was about 40% In addition, most stocking occurs at a different time of year in 2006, we used a local mortality rate of 10% for this analysis than spawning does, and the type of streams into which fish are because most hatchery smolt mortality in these systems occurs stocked differ from those in which wild adult fish spawn. downstream from the initial stocking location (Muir et al. 2001; Salmon carcasses can increase secondary production in Smith et al. 2003). Stocking efforts lasted anywhere from 1 to 39 streams both directly via consumption of the carcass itself (Bilby d, but most were quite short, with a median stocking duration et al. 1998; Chaloner and Wipfli 2002; Claeson et al. 2006) and of 2.5 d. We also estimated potential maximum nutrient con- indirectly via release of biologically reactive inorganic nutrients, centration increases due to stocking separately for each of the which promote primary production and subsequent increases in 31 stocking sites using total nutrient input estimates (mass of grazer biomass (Wipfli et al. 1998; Johnston et al. 2004; Cak nutrients), mean discharge at those sites during the month that et al. 2008). Inorganic nutrients released as carcasses decay stocking occurred, and the number of days over which stocking support primary production. The carcass itself can also support occurred (total volume of water). This calculation assumes all stream macroinvertebrates and other detritvores that feed on the nutrients from dead fish become available in a biologically re- flesh (Bilby et al. 1998; Chaloner and Wipfli 2002; Janetski active inorganic form. This assumption provides an upper limit et al. 2009). If smolts and fry do die and leave carcasses in the
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 on potential nutrient availability associated with hatchery-origin stream, the carbon and nutrient subsidies from these fish are SDN. likely to enter the stream ecosystem in a manner similar to that of retained adult carcasses. However, given the smaller size of juvenile carcasses relative to adult carcasses, dead smolts and RESULTS AND DISCUSSION dead fry persist for a much shorter period of time in the stream Broodstock collections and the out-planting of juvenile leading to a shorter period of influence on the stream food web Chinook salmon collectively yielded a net input of nutrients (Elliott 1997). to the portion of the Snake River upstream of LGD accessible to More importantly, while some juvenile hatchery-origin fish anadromous fish in 5 of the 6 years studied (Figure 2). Year-to- die shortly after stocking and do indeed occur as carcasses in year variability in juvenile survival was the primary influence on the stream, most of the mortality for stocked smolts occurs as a whether these hatchery activities were a net source or a net sink result of predation with direct consumption of an entire fish and for nutrients. In the 1 year that stocking activities did not yield a no remaining carcass (Muir et al. 2001; Monzyk et al. 2009), a net nutrient input (2004), smolt survival was high owing to rel- fundamentally different route than that of nutrients from adult atively high stream and main-stem flows (FPC 2007). The mass carcasses. This substantial input from hatchery smolts for pis- of nutrients removed via broodstock collections and in-stream civorous predators can have cascading effects on the stream NOTE 1291 Downloaded by [Department Of Fisheries] at 20:08 25 September 2012
FIGURE 2. Panel A indicates the annual input and output of nitrogen (N) and phosphorus (P) via mortality of Chinook salmon smolts stocked upstream from Lower Granite Dam and panel B indicates the equivalent number of adult salmon for each year’s addition and loss of nutrients (assuming adults are 0.38% Pand have a mean mass of 5.5 kg, per Scheuerell et al. 2005).
food web (Kostow 2009). This includes bottom-up support of which can then promote primary production (Vanni 2002). The increased predator populations, which increases predation risk consumption of a larger number of hatchery-origin juveniles for native smolts (Eby et al. 2006; Kostow 2009; Gozlan et al. therefore has the potential to lead to elevated rates of inorganic 2010; Ellis et al. 2011). Predation upon hatchery-origin juvenile nutrient availability. These excreted nutrients are likely to be Chinook salmon also has the potential to indirectly affect nu- released in low amounts though, and they are likely to be ac- trient dynamics via predator excretion. Not all of the nutrients cessed and incorporated into the ecosystem in different times from a given prey item are assimilated. Some of the nutrients and places than are adult carcasses, which function as small are excreted as highly available forms of inorganic nutrients, localized point sources for inorganic nutrients. 1292 WARREN AND MCCLURE
FIGURE 3. Potential increases in stream nutrient concentration at each of 31 stocking sites in the portion of the Snake River accessible to anadromous fish upstream of Lower Granite Dam assuming 10% mortality of spring–summer hatchery-origin Chinook salmon smolts at each stocking site. Estimates of potential elevated nitrogen (N) concentrations are illustrated by the lightly shaded bars and estimates of potential elevated phosphorus (P) concentrations are illustrated by the solid bars.
Most juvenile Chinook salmon stocking occurs in mid to to greater than 3 µg/L. Although streams in the Snake River late spring, particularly for hatchery fish in the spring–summer ecosystem differ from those in the Fraser River, this result ESU. This is also a time period when primary production in suggests that elevating stream P concentrations by ≥2 µg/L forested streams is generally highest—stream temperatures are could affect in-stream primary production under the right con- increasing but leaf-out has not yet occurred (Roberts et al. 2007). ditions. The duration of any increase may be limited though. Our analysis of site-by-site nutrient concentration changes rep- Estimates of potential maximum increases in nutrient concen- resents maximum potential increases in nutrient availability at tration are calculated only over the stocking period and con- a time when elevated nutrients have the greatest potential to centrations will probably return to prestocking levels relatively make a difference; however, these potential increases in con- soon after stocking ends. In the absence of measurements of Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 centration assume that all nutrients from dead hatchery fish are stream primary production, we cannot address this effect di- immediately available. We found a wide range in the potential rectly in the current study. Whether SDN from hatchery-origin increase in stream nutrient concentrations during the stocking fish influences primary production or any other aspect of the period from a high of 117.7 µg/L of N and 16.9 µg/L of P to stream ecosystem over short or long time periods will depend a low of less than 0.001 µg/L of N and less than 0.0001 µg/L upon the abundance of live fish, the number and persistence of P (mean P and N of 2.3 and 16.1 µg/L, respectively, and of carcasses, and ultimately, the chemical and physical condi- median of 1.9 and 0.27 µg/L, respectively; Figure 3). For the tions in the recipient stream (Ambrose et al. 2004; Mitchell and majority of sites, nutrient inputs in spring 2006 led to poten- Lamberti 2005; Chaloner et al. 2007). tial inputs of less than 10 µg/L of N and 2 µg/L of P. In an The age-0 fish stocked in this system are from the SRF assessment of periphyton biomass and chlorophyll a in salmon Chinook salmon ESU, and typically out-migrate as subyear- streams of the Fraser River system, British Columbia, Verspoor lings after a period of growth in the stream. This subset of et al. (2010) found that net primary production was signifi- stocking activity yielded a net loss of nutrients in all 6 years cantly and positively related to prespawning soluble reactive evaluated. Because the fish are stocked at a smaller size, the nu- phosphorous concentrations that ranged from less than 1 µg/L trient input associated with their mortality is much lower than it NOTE 1293
is for age-1 individuals. In addition, the within-system growth ity, stream geomorphology, background nutrient concentrations, and subsequent migration of age-0 fish from streams contributed and riparian and upland land use (Ambrose et al. 2004; Janetski substantially to annual nutrient losses associated with stocking et al. 2009; Harvey and Wilzbach 2010; Verspoor et al. 2010; activities for this ESU. In age-0 fish, which gain substantial Holtgrieve and Schindler 2011). A whole ecosystem approach biomass in the stream before they emigrate, the P and N that will be needed as we move beyond this initial mass balance they accumulated is removed from the system rather than recy- analysis to a broad assessment of the fate and influence of SDN cled within the system as would occur with a resident fish. This from hatchery-origin fish. supports work by Nislow et al. (2004) suggesting that the age and stage of development at which stocking occurs can strongly influence the SDN nutrient budget. The mortality rates evalu- ACKNOWLEDGMENTS ated in this study do not account for terrestrial predation. Avian D.R.W. and M.M.M. conceived of this idea together. Ini- and other land-based predators remove fish and their nutrients tial analysis and writing were conducted by D.R.W. with from the system and therefore from the input portion of the ba- substantial input and revision on all aspects of the paper pro- sic mass balance analysis used here (Collis et al. 2001). We do vided by M.M.M. We thank John Baily, Neil Bettez, Chris not have a measure of terrestrial predator mortality at these sites Caudill, Robbins Church, Laura Cowger, Ann Gannam, Damon so the SDN input associated with the simple smolt mortality is Holzer, Jeff Jorgensen, Matt Keefer, Greg Kovalchuck, Kate probably an overestimate in the current analysis. Macneale, Doug Marsh, David Noakes, Julie Pett-Ridge, Mike While SDN from wild adult salmon has been well studied, Rust, Mark Scheuerell, Keith Nislow, and two anonymous re- SDN from hatchery-origin smolts has not. Overall, we demon- viewers who provided valuable input and important information strated in this study that the impact of SDN from stocking ju- for this project. Funding was provided by the National Research venile Chinook salmon varied with age and stage of released Council’s Post-Doctoral Fellowship Program and The National fish, year-specific smolt survival rates, and local conditions. In Oceanographic and Atmospheric Administration (NOAA). This most years hatchery activities lead to a net input of nutrients work does not reflect the official views of NOAA or any other and the variability in SDN fluxes is driven primarily by the agency or funding source. survival of hatchery-origin spring–summer smolts. When smolt survival was low there was a net influx of SDN to streams as a result of hatchery activities. When smolt survival was high, REFERENCES SDN removal via broodstock collections and in-stream growth Ambrose, H. E., M. A. Wilzbach, and K. W. Cummins. 2004. Periphyton of fall Chinook salmon juveniles exceeded total SDN inputs via response to increased light and salmon carcass introduction in northern in-stream mortality of hatchery-origin juveniles. This resulted California streams. Journal of the North American Benthological Society in a net loss of SDN from the system via hatchery activities. In 23:701–712. Bilby, R. E., B. R. Fransen, P. A. Bisson, and J. K. Walter. 1998. Response of addition, the total mass and duration of a stocking event rela- juvenile coho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus tive to stream discharge strongly influences maximum potential mykiss) to the addition of salmon carcasses to two streams in southwestern nutrient increases, and few sites will experience potential con- Washington, U.S.A. Canadian Journal of Fisheries and Aquatic Sciences centration increases larger than 10 µg/L of N and 2 µg/L of P 55:1909–1918. for the duration of stocking. Cak, A. D., D. T. Chaloner, and G. A. Lamberti. 2008. Effects of spawning salmon on dissolved nutrients and epilithon in coupled stream-estuary systems There is indication that differences in nutrient flux due to of southeastern Alaska. Aquatic Sciences 70:169–178. stocking activities vary between the two ESUs examined, in Cederholm, J. C., M. D. Kunze, T. Murota, and A. Sibatani. 1999. Pacific association with their life history. Snake River Fall Chinook salmon carcasses: essential contributions of nutrients and energy for aquatic
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 salmon—ocean-type fish that begin out-migration earlier in and terrestrial ecosystems. Fisheries 24(10):6–15. their life cycle—experience substantial growth as they move Chaloner, D. T., G. A. Lamberti, A. D. Cak, N. L. Blair, and R. T. Edwards. 2007. Inter-annual variation in responses of water chemistry and epilithon to downstream. In addition, these smaller juveniles provide fewer Pacific salmon spawners in an Alaskan stream. Freshwater Biology 52:478– nutrients when they die. As a result, the full range of stock- 490. ing activities, including mortality of hatchery juveniles, nutrient Chaloner, D. T., and M. S. Wipfli. 2002. Influence of decomposing Pacific loss associated with in-stream growth of hatchery-origin juve- salmon carcasses on macroinvertebrate growth and standing stock in south- niles, and broodstock collection associated with this ESU alone eastern Alaska streams. Journal of the North American Benthological Society 21:430–442. consistently yielded a net loss of SDN. Chaloner, D. T., M. S. Wipfli, and J. P. Caouette. 2002. Mass loss and macroin- In addition to input or export values, the pathway by which vertebrate colonisation of Pacific salmon carcasses in south-eastern Alaskan hatchery-origin SDN enters the stream food web is a key con- streams. Freshwater Biology 47:263–273. sideration in this or any mass balance analysis. The influence Claeson, S. M., J. L. Li, J. E. Compton, and P. A. Bisson. 2006. Response of of hatchery-origin SDN on the broader stream ecosystem is nutrients, biofilm, and benthic insects to salmon carcass addition. Canadian Journal of Fisheries and Aquatic Sciences 63:1230–1241. dependent upon a number of local factors including the na- Collis, K., D. D. Roby, D. P. Craig, B. A. Ryan, and R. D. Ledgerwood. 2001. ture of hatchery smolt mortality at a stocking site, predator Colonial waterbird predation on juvenile salmonids tagged with passive inte- population size and density, food web dynamics, fish activ- grated transponders in the Columbia River estuary: vulnerability of different 1294 WARREN AND MCCLURE
salmonid species, stocks, and rearing types. Transactions of the American to Lower Granite Dam on the Snake River. Transactions of the American Fisheries Society 130:385–396. Fisheries Society 138:1093–1108. Connor, W. P., B. D. Arnsberg, S. G. Smith, D. M. Marsh, and W. D. Muir. Moore, J. W., D. E. Schindler, J. L. Carter, J. Fox, J. Griffiths, and G. W. 2008. Postrelease performance of natural and hatchery subyearling fall Holtgrieve. 2007. Biotic control of stream fluxes: spawning salmon drive Chinook salmon in the Snake and Clearwater rivers. 2006 Annual Report nutrient and matter export. Ecology 88:1278–1291. to the Bonneville Power Administration, Projects 1983350003, 199102900, Moore, J. W., D. E. Schindler, and M. D. Scheuerell. 2004. Disturbance and 199801004, Portland, Oregon. of freshwater habitats by anadromous salmon in Alaska. Oecologia 139: Eby, L. A., W. J. Roach, L. B. Crowder, and J. A. Stanford. 2006. Effects 298–308. of stocking-up freshwater food webs. Trends in Ecology and Evolution 21: Muir, W. D., S. G. Smith, J. G. Williams, E. E. Hockersmith, and J. R. Skalski. 576–584. 2001. Survival estimates for migrant yearling Chinook salmon and steelhead Elliott, J. M. 1997. An experimental study on the natural removal of dead trout tagged with passive integrated transponders in the lower Snake and lower fry in a lake district stream. Journal of Fish Biology 50:870–877. Columbia rivers, 1993–1998. North American Journal of Fisheries Manage- Ellis, B. K., J. A. Stanford, D. Goodman, C. P. Stafford, D. L. Gustafson, D. A. ment 21:269–282. Beauchamp, D. W. Chess, J. A. Craft, M. A. Deleray, and B. S. Hansen. 2011. Naiman, R. J., R. E. Bilby, D. E. Schindler, and J. M. Helfield. 2002. Pacific Long-term effects of a trophic cascade in a large lake ecosystem. Proceedings salmon, nutrients, and the dynamics of freshwater and riparian ecosystems. of the National Academy of Sciences of the USA 108:1070–1075. Ecosystems 5:399–417. FPC (Fish Passage Center). 2007. 2007 annual report of the Columbia basin fish Nislow, K. H., J. D. Armstrong, and S. McKelvey. 2004. Phosphorus flux due and wildlife authority. FPC, Portland, Oregon. to Atlantic salmon (Salmo salar) in an oligotrophic upland stream: effects Gende, S. M., R. T. Edwards, M. F. Willson, and M. S. Wipfli. 2002. Pacific of management and demography. Canadian Journal of Fisheries and Aquatic salmon in aquatic and terrestrial ecosystems. BioScience 52:917–928. Sciences 61:2401–2410. Gozlan, R. E., J. R. Britton, I. Cowx, and G. H. Copp. 2010. Current knowl- Petrosky, C. E., H. A. Schaller, and P. Budy. 2001. Productivity and survival edge on non-native freshwater fish introductions. Journal of Fish Biology 76: rate trends in the freshwater spawning and rearing stage of Snake River 751–786. Chinook salmon (Oncorhynchus tshawytscha). Canadian Journal of Fisheries Harvey, B. C., and M. A. Wilzbach. 2010. Carcass addition does not enhance and Aquatic Sciences 58:1196–1207. juvenile salmonid biomass, growth, or retention in six northwestern California Roberts, B. J., P. J. Mulholland, and W. R. Hill. 2007. Multiple scales of streams. North American Journal of Fisheries Management 30:1445–1451. temporal variability in ecosystem metabolism rates: results from 2 years of Holtgrieve, G. W., and D. E. Schindler. 2011. Marine-derived nutrients, bio- continuous monitoring in a forested headwater stream. Ecosystems 10:588– turbation, and ecosystem metabolism: reconsidering the role of salmon in 606. streams. Ecology 92:373–385. Scheuerell, M. D., P. S. Levin, R. W. Zabel, J. G. Williams, and B. L. Sanderson. Idaho Department of Fish and Game, USFWS (U.S. Fish and Wildlife Service), 2005. A new perspective on the importance of marine-derived nutrients to Shoshone-Bannock-Tribes, Idaho Power Company, and Nez Perce Tribe. threatened stocks of Pacific salmon (Oncorhynchus spp.). Canadian Journal 2008. 2008 Annual operating plan for fish production programs in the Salmon of Fisheries and Aquatic Sciences 62:961–964. River basin. USFWS, Pacific Region Hatchery Review, Snake River Re- Schindler, D. E., M. D. Scheuerell, J. W. Moore, S. M. Gende, T. B. Francis, views SR-073, Portland, Oregon. Available: www.fws.gov/pacific/Fisheries/ and W. J. Palen. 2003. Pacific salmon and the ecology of coastal ecosystems. Hatcheryreview/snakedocs.html. (May 2011). Frontiers in Ecology and the Environment 1:31–37. Janetski, D. J., D. T. Chaloner, S. D. Tiegs, and G. A. Lamberti. 2009. Pacific Smith, S. G., W. D. Muir, E. E. Hockersmith, R. W. Zabel, R. J. Graves, C. V. salmon effects on stream ecosystems: a quantitative synthesis. Oecologia Ross, W. P. Connor, and B. D. Arnsberg. 2003. Influence of river conditions 159:583–595. on survival and travel time of Snake River subyearling fall Chinook salmon. Johnston, N. T., E. A. MacIsaac, P. J. Tschaplinski, and K. J. Hall. 2004. Effects North American Journal of Fisheries Management 23:939–961. of the abundance of spawning sockeye salmon (Oncorhynchus nerka)on Thomas, S. A., T. V. Royer, G. W. Minshall, and E. Snyder. 2003. Assessing nutrients and algal biomass in forested streams. Canadian Journal of Fisheries the historic contribution of marine-derived nutrients to Idaho streams. Pages and Aquatic Sciences 61:384–403. 41–55 in J. G. Stockner, editor. Nutrients in salmonid ecosystems: sustaining Jonsson, B., and N. Jonsson. 2003. Migratory Atlantic salmon as vectors for the production and biodiversity. American Fisheries Society, Symposium 34, transfer of energy and nutrients between freshwater and marine environments. Bethesda, Maryland. Freshwater Biology 48:21–27. Vanni, M. J. 2002. Nutrient cycling by animals in freshwater ecosystems. Annual Kostow, K. 2009. Factors that contribute to the ecological risks of salmon and Review of Ecology and Systematics 33:341–370. steelhead hatchery programs and some mitigating strategies. Reviews in Fish Verspoor, J. J., D. C. Braun, and J. D. Reynolds. 2010. Quantitative links between
Downloaded by [Department Of Fisheries] at 20:08 25 September 2012 Biology and Fisheries 19:9–31. Pacific salmon and stream periphyton. Ecosystems 13:1020–1034. Lyle, A. A., and J. M. Elliott. 1998. Migratory salmonids as vectors of carbon, Wipfli, M. S., J. P. Hudson, and J. P. Caouette. 1998. Influence of salmon nitrogen and phosphorus between marine and freshwater environments in carcasses on stream productivity: response of biofilm and benthic macroin- north-east England. Science of the Total Environment 210–211:457–468. vertebrates in southeastern Alaska, U.S.A. Canadian Journal of Fisheries and Mitchell, N. L., and G. A. Lamberti. 2005. Responses in dissolved nutrients Aquatic Sciences 55:1503–1511. and epilithon abundance to spawning salmon in southeast Alaska streams. Wipfli, M. S., J. P. Hudson, J. P. Caouette, N. L. Mitchell, J. L. Lessard, R. A. Limnology and Oceanography 50:217–227. Heintz, and D. T. Chaloner. 2010. Salmon carcasses increase stream produc- Monzyk, F. R., B. C. Jonasson, T. L. Hoffnagle, P. J. Keniry, R. W. Carmichael, tivity more than inorganic fertilizer pellets: a test on multiple trophic levels and P. J. Cleary. 2009. Migration characteristics of hatchery and natural in streamside experimental channels. Transactions of the American Fisheries spring Chinook salmon smolts from the Grande Ronde River basin, Oregon, Society 139:824–839.