This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:34 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Impacts of Diet on Thiamine Status of Lake Ontario American Eels John D. Fitzsimons a , Scott B. Brown b g , Lisa R. Brown b , Guy Verreault c , Rémi Tardif c , Ken G. Drouillard d , Scott A. Rush e & Jana R. Lantry f a Fisheries and Oceans Canada , Great Lakes Laboratory of Fisheries and Aquatic Sciences , 867 Lakeshore Road, Burlington , Ontario , L7R 4A6 , Canada b Environment Canada, Science and Technology Branch , National Water Research Institute , 867 Lakeshore Road, Burlington , Ontario , L7R 4A6 , Canada c Ministère des Ressources Naturelles et de la Faune , 186 rue Fraser, Riviére-du-Loup , Québec , G5R 1C8 , Canada d Great Lakes Institute for Environmental Research, Department of Biological Sciences , University of Windsor , 401 Sunset Avenue, Windsor , Ontario , N9B 3P4 , Canada e Department of Wildlife, Fisheries, and Aquaculture , Mississippi State University , Box 9690, Mississippi State, Mississippi , 39762 , USA f New York State Department of Environmental Conservation , Cape Vincent Fisheries Research Station , 541 East Broadway, Cape Vincent , New York 13618 , USA g Deceased Published online: 02 Sep 2013.

To cite this article: John D. Fitzsimons , Scott B. Brown , Lisa R. Brown , Guy Verreault , Rémi Tardif , Ken G. Drouillard , Scott A. Rush & Jana R. Lantry (2013) Impacts of Diet on Thiamine Status of Lake Ontario American Eels, Transactions of the American Fisheries Society, 142:5, 1358-1369, DOI: 10.1080/00028487.2013.811100 To link to this article: http://dx.doi.org/10.1080/00028487.2013.811100

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1358–1369, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.811100

ARTICLE

Impacts of Diet on Thiamine Status of Lake Ontario American Eels

John D. Fitzsimons* Fisheries and Oceans Canada, Great Lakes Laboratory of Fisheries and Aquatic Sciences, 867 Lakeshore Road, Burlington, Ontario L7R 4A6, Canada Scott B. Brown1 and Lisa R. Brown Environment Canada, Science and Technology Branch, National Water Research Institute, 867 Lakeshore Road, Burlington, Ontario L7R 4A6, Canada Guy Verreault and Remi´ Tardif Ministere` des Ressources Naturelles et de la Faune, 186 rue Fraser, Riviere-du-Loup,´ Quebec´ G5R 1C8, Canada Ken G. Drouillard Great Lakes Institute for Environmental Research, Department of Biological Sciences, University of Windsor, 401 Sunset Avenue, Windsor, Ontario N9B 3P4, Canada Scott A. Rush Department of Wildlife, Fisheries, and Aquaculture, Mississippi State University, Box 9690, Mississippi State, Mississippi 39762, USA Jana R. Lantry New York State Department of Environmental Conservation, Cape Vincent Fisheries Research Station, 541 East Broadway, Cape Vincent, New York 13618, USA

Abstract The Lake Ontario–upper St. Lawrence River (LOUSL) population of American Eels Anguilla rostrata (hereafter, “eels”) was once one of the most important groups of the species but is now in a state of serious decline. Given that thiamine deficiency has been observed in almost all of the top predators in Lake Ontario, we assessed the Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 potential that a diet-induced thiamine deficiency associated with consumption of Alewives Alosa pseudoharengus could be involved. Muscle thiamine was measured in eels from throughout the LOUSL corridor and was compared with putative threshold effect levels established from the literature. Mirex concentrations were used to separate Lake Ontario-resident eels from non-Lake Ontario-resident eels. Stable isotope analyses of eel muscle samples and potential prey, including Alewives, were combined with mixing model software to infer the diets consumed by Lake Ontario eels. Although residence in Lake Ontario was associated with a significant decline in muscle thiamine concentration, estimated Alewife consumption by eels was unexpectedly low. Instead, mixing model results indicated that crayfish and Round Goby Neogobius melanostomus were the major prey. Both taxa are known to contain thiaminase and have the potential to cause thiamine deficiency, but there are no reports of thiamine deficiency associated with Round Goby consumption, thus implicating crayfish. There was no recovery in thiamine levels prior to the initiation of oceanic migration. As a result, thiamine levels of Lake Ontario-resident eels prior to migration were only slightly above putative threshold effect levels for pathological and behavioral effects in Japanese Eels Anguilla japonica,but

*Corresponding author: fi[email protected] 1Deceased. Received August 24, 2012; accepted May 28, 2013 Published online September 2, 2013

1358 THIAMINE STATUS OF AMERICAN EELS 1359

this would require confirmation with American Eels. Since thiamine levels are expected to decline further during migration, additional effects on eel behavior, reproduction, and survival seem probable. Because of panmixia, such effects—when combined with the relatively high reproductive potential of Lake Ontario-resident eels—may have consequences for the entire species.

The Lake Ontario–upper St. Lawrence River (LOUSL) pop- mercial landings (COSEWIC 2006) indicate that this region ulation of American Eels Anguilla rostrata (hereafter, “eels”) contributed a substantial portion (27–67%) of the total North has experienced dramatic declines in abundance in recent years American spawning output of American Eels. (Castonguay et al. 1994; Axelsen 1997; Casselman et al. 1997); Hydropower dams and overfishing are considered to be the because of panmixia, these declines have the potential to im- two largest sources of mortality for LOUSL eels. The Moses– pact the entire American Eel species (Velez-Espino´ and Koops Saunders and Beauharnois–Les Cedres hydropower dams lo- 2010). The declines are reflected in both CPUE of adults in the cated on the St. Lawrence River (completed in 1958 and 1961, river (de Lafontaine et al. 2010) and reductions in the numbers respectively) represent a major impediment to both upstream of juvenile eels ascending the eel ladder at the Moses–Saunders and downstream migrations of eels in the LOUSL corridor Hydroelectric Dam (Casselman et al. 1997). Casselman et al. (Verreault and Dumont 2003; COSEWIC 2006; Cairns et al. (1997) reported that after 1982, the number of juvenile eels 2007; de Lafontaine et al. 2009, 2010). However, these two ascending the ladder (used as a recruitment index) reached a hydropower facilities alone cannot account for the dramatic record low. The recruitment index fell to its lowest value in decline in abundance of the LOUSL eel population. It was es- 1999 (MacGregor et al. 2008), representing a thousand-fold de- timated that eel mortality due to passage through the turbines crease in abundance relative to that observed in 1982. at the Moses–Saunders and Beauharnois–Les Cedres dams av- The large population of eels that historically resided in the eraged 22% per dam (Verreault and Dumont 2003). Although LOUSL corridor (Figure 1; Casselman 2003) appeared to com- collectively a large source of mortality at approximately 40% plete the entire portion of the adult phase of their life cycle overall, turbine-related mortality alone would have been insuffi- in freshwater up until the point of their spawning migration, cient to completely block access either upstream to Lake Ontario as indicated by otolith Sr:Ca ratios (Goodwin 1999; Thibault or downstream to the ocean. et al. 2007; J. D. Fitzsimons, unpublished data). There is still In the past, fishing mortality of eels was high: eel landings uncertainty as to the habitat occupied within the corridor and in Ontario increased rapidly from 1,435 metric tons in 1950 to the relative dietary contributions from freshwater and marine re- 13,240 metric tons in 1960 and 195,000 metric tons in 1975 sources. Otolith Sr:Ca ratios can confirm freshwater residence of (Cairns et al. 2008), corresponding to a 100-fold increase in eels (Thibault et al. 2007), but such indications do not preclude mortality over 25 years. Although commercial harvest remained the possibility that eels supplement their diets with seasonally relatively stable at 104–123 metric tons between 1984 and 1993, migratory marine prey (Garman and Macko 1998). Such a strat- harvest declined thereafter by about 10 metric tons/year, reach- egy may be one means by which freshwater eels can maintain ing 12 metric tons by 2002—2 years before the closure of the strict catadromy (Cairns et al. 2009). commercial fishery in 2004 (Mathers and Stewart 2009). The LOUSL corridor represented a significant area for Collectively, the combined mortality resulting from the ef- growth of juvenile eels and was considered to be especially fects of hydropower dams and overfishing is estimated at 75%.

Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 important in maintaining the overall abundance of this panmic- As a result, a portion of this population—albeit much reduced— tic species (Cotˆ e´ et al. 2009) for several reasons. The freshwater should have been able to migrate to the Sargasso Sea, spawn runoff of the St. Lawrence River (10,000 m3/s) represents ap- successfully, and support ongoing recruitment (Verreault and proximately 19% of the total freshwater runoff in the species’ Dumont 2003), especially given the high fecundity and com- range (Castonguay et al. 1994). The LOUSL system is easily pensatory ability of this group (DeAngelis et al. 1990; Haro the largest freshwater feeding area anywhere in the eel’s range. et al. 2000). There are, however, other factors potentially con- Available data indicate that the LOUSL population is composed tributing to eel mortality, and much less is known about their almost exclusively of female eels (Vladykov 1966; Dutil et al. effects; such factors include habitat loss and alteration, oceanic 1985; Couillard et al. 1997). The size at migration increases conditions, parasitism, and organic pollution (Haro et al. 2000). with latitude, and LOUSL eels are some of the largest and most The composition of the fish community may also be involved; fecund in North America (Barbin and McCleave 1997; Oliveira since the 1900s, the Lake Ontario fish community has been dom- 1999; Tremblay 2009; Jessop 2010). Pre-collapse estimates of inated by nonnative Alewives Alosa pseudoharengus.Alewives the reproductive contribution of American Eels from the St. entered the Great Lakes in the 1800s and have been associated Lawrence River–Lake Ontario component based on a linear re- with declines in a variety planktivores and piscivores, result- lationship between water discharge and recruitment plus com- ing in a general reduction in fishery productivity, although the 1360 FITZSIMONS ET AL.

FIGURE 1. Map of locations where American Eels were collected in the Lake Ontario–upper St. Lawrence River corridor (A = yellow eels; B = juvenile eels; C = upstream Ontario silver eels; D = downstream silver eels [includes downstream Ontario silver eels and downstream other silver eels]).

extent to which Alewives may have affected eels that are resident also contain thiaminase (e.g., Rainbow Smelt Osmerus mordax, in Lake Ontario is unclear (Smith 1970). Round Goby Neogobius melanostomus, and crayfish; Rutledge Clupeids such as Alewives are known to contain high thi- and Ying 1972; Tillitt et al. 2005; Honeyfield et al. 2012). How- aminase activity, and when they comprise a sufficiently high ever, relatively little is known about the quantitative importance proportion of a predator’s diet, they can cause thiamine (vita- of any of these prey taxa for eel diets or their possible effects min B1) deficiency (Neilands 1947; Tillitt et al. 2005; Honey- on the thiamine status and health of the eels. Given that the field et al. 2007; Riley and Evans 2008; Wistbacka and Bylund Lake Ontario eel population has shown dramatic declines from Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 2008). A thiamine deficiency associated with high Alewife con- historic levels (Casselman et al. 1997), there is keen interest by sumption affects top predators throughout Lake Ontario and fishery management agencies to identify sources of mortality elsewhere in the Great Lakes; a reduction in thiamine below and, to the extent possible, reduce or eliminate them. species-specific threshold levels results in behavioral changes To evaluate the thiamine status of Lake Ontario eels, we col- and death of adult salmonines (Morito et al. 1986; Fitzsimons lected eels from four separate locations in the LOUSL corridor and Brown 1998; Fitzsimons et al. 1999, 2005; Brown et al. and measured their muscle thiamine concentrations. The four 2005a, 2005b; Honeyfield et al. 2005; Johnston et al. 2005). groups were (1) upstream migratory juvenile eels that had not yet For larval stages of these same salmonines, thiamine deficiency entered or fed in Lake Ontario, (2) yellow eels that were resident leads to mortality known as early mortality syndrome, as well as and feeding in Lake Ontario, (3) downstream migratory silver reduced growth, foraging, and predator avoidance (Brown et al. eels that had fed previously in Lake Ontario and were collected 1998b; Fitzsimons et al. 2009). Lake Ontario eels are known to in the St. Lawrence River 100 km downstream of Lake Ontario consume Alewives and other prey (J. Casselman, Queens Uni- prior to their oceanic migration, and (4) downstream migratory versity, personal communication; J. Bowlby, Ontario Ministry silver eels that were collected in the St. Lawrence River 600 km of Natural Resources, personal communication), some of which downstream of Lake Ontario. Samples at the downstream-most THIAMINE STATUS OF AMERICAN EELS 1361

location represented a mixture of Lake Ontario-resident eels and after eels were euthanized with clove oil were processed in the non-Lake Ontario-resident eels; therefore, we used concentra- same manner as frozen samples, with the exception that once tions of the chlorinated hydrocarbon mirex to separate the two the muscle sample was removed, it was immediately frozen on groups for comparison of their thiamine levels. Mirex is an in- dry ice and stored in a chest freezer (−20◦C) until analyzed for dustrial chemical for which the only known source on the Great thiamine. The heads of all eels were removed, and their sagittal Lakes consists of industrial discharges on two Lake Ontario trib- otoliths were extracted, mounted, and aged according to the utaries. Mirex has previously been used to determine the origin methods of Verreault et al. (2009). of eels in the LOUSL corridor (Dutil et al. 1985; Couillard et al. Estimation of prey contribution to eel diets.—To establish 1997). For American Eels collected in 2007 from the LOUSL diets of eels while they resided in Lake Ontario, we used mixing corridor, including Lake Ontario and the lower St. Lawrence model software to obtain diet estimates based on stable isotope River, mirex levels in Lake Ontario eels were significantly el- analyses of silver eel muscle and potential Lake Ontario prey. evated over that of eels known to be resident in the lower St. Standard stomach content analysis was not feasible because sil- Lawrence River (Byer et al. 2013). Because silver eels have ver eels have stopped feeding (Pankhurst and Sorensen 1984). stopped feeding (Pankhurst and Sorensen 1984), we used these Most of the potential prey samples that were used to establish eels to establish the relative importance of different prey taxa the diets of Lake Ontario eels with stable isotopes and mixing in the diets of Lake Ontario eels by using stable isotopes and models (see below) came from a concurrent study that exam- mixing models. Measured thiamine levels in silver eels were ined trophic relationships for Lake Trout Salvelinus namaycush compared with published biological thresholds to establish the in Lake Ontario (Fitzsimons, unpublished data). The concurrent potential effects of feeding in Lake Ontario on eel migration and study included samples of Alewives, Rainbow Smelt, Round reproduction. Goby, and Slimy Sculpin Cottus cognatus that were collected during 2005–2006, coincident with the period in which eels were collected from the St. Lawrence River. Due to the dif- METHODS ficulty in using conventional sampling gear to collect crayfish, Eel sample collection and processing.—To determine the which constitute a known component of Lake Ontario eels’ diets thiamine status of eels prior to their entry into Lake Ontario, (J. Casselman, Queens University, personal communication), we collected juvenile eels during July 2005 from the Moses– the stable isotope signature for this taxon was established by Saunders Dam eel ladder (hereafter, “juvenile eels”; Figure 1; using crayfish from the stomach contents of Smallmouth Bass Marcogliese and Casselman 2009). To represent eels that were Micropterus dolomieu that were collected with gill nets in east- resident in Lake Ontario, we collected adult yellow eels in June ern Lake Ontario during August 2010. Because diets have not 2003 with a 10-m trap net that was set adjacent to a rocky pier been reported for juvenile eels in the St. Lawrence River below at Port Weller near the downstream exit of the Welland Canal Lake Ontario or for yellow eels resident in the St. Lawrence in western Lake Ontario (hereafter, “yellow eels”; Figure 1). River, we had no knowledge of their potential prey; as a result, Adult silvering or silver eels (McGrath et al. 2003) in the initial we did not collect prey and use their stable isotope signatures in stages of their downstream migration from Lake Ontario to the a mixing model for juvenile eels or yellow eels as was done for Atlantic Ocean were collected in September 2005 from the tail- Lake Ontario-resident eels. race area of the Moses–Saunders Dam, located approximately Thiamine analysis.—Thiamine concentration was analyzed 100 km downstream of Lake Ontario (hereafter, “upstream On- by the high-performance liquid chromatography method of tario silver eels”). One month later in 2005, silver eels were also Brown et al. (1998a). This method uses fluorescence detection collected at La Pocatiere by using a weir situated approximately to separate three thiamine vitamers (free thiamine, thiamine Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 600 km downstream of Lake Ontario (hereafter, “downstream monophosphate, and thiamine pyrophosphate); the three vita- silver eels”). Because the foraging area used by eels sampled at mers were quantified individually but were summed and re- La Pocatiere was more uncertain than that of eels collected fur- ported as total thiamine. ther upstream (Castonguay et al. 1989), we used muscle mirex Mirex analysis.—Samples of eel muscle were analyzed for concentrations (see below; Couillard et al. 1997) to differentiate mirex via the methods described by Braune et al. (2007). Briefly, between eels that had previously resided and fed in Lake Ontario tissue homogenates were ground, spiked with recovery surro- and eels that had not. gate standards, and extracted with dichloromethane : hexane After collection, eels were either (1) immediately placed (50:50% volume/volume). Sample cleanup was performed by in a freezer (−20◦C) and processed dead (juvenile eels and gel permeation chromatography followed by activated Florisil upstream Ontario silver eels); or (2) euthanized with an overdose chromatography. Chemical analysis was conducted by using of clove oil and processed immediately after death (yellow eels a Hewlett-Packard 5890 gas chromatograph with a Hewlett- and downstream silver eels). All frozen samples were measured Packard 5973 mass selective detector that was operated in the for weight ( ± 1 g) and TL ( ± 1 cm) in the frozen state, and electron impact mode and using selected ion monitoring. For ev- a frozen muscle sample was removed for thiamine, mirex, and ery batch of five samples injected, two organochlorine pesticide stable isotope analyses (see below). Samples taken immediately standards (Supelco, Bellefonte, Pennsylvania), a method blank, 1362 FITZSIMONS ET AL.

12 Cayuga Brown Trout Salmo trutta (CBT; δ13C =−25.1‰; δ15N 10 = 17.4‰; 12.2% N, 54.9% C), and the SDs of replicate CBT δ13 δ15 8 samples were 0.05% for C and 0.18% for N. Stable iso- topes were analyzed on a Finnigan MAT Delta Plus continuous- 6 flow elemental analyzer at Cornell University’s Stable Isotope Count 4 Facility. 2 Statistics.—Among-location (among-group) variation in thi- amine, stable isotopes, TL, weight, age, and mirex concentration 0 was examined using one-way ANOVAs. When among-site dif- ferences existed for any of the measured variables, they were identified using Tukey’s honestly significant difference test. Re- Range of log mirex (mg/kg) for class lationships between muscle thiamine, mirex, or stable isotopes and measures of length, weight, or age within and among sam- FIGURE 2. Histogram of American Eel counts by log(mirex level) class pling locations were examined by using a combination of linear among downstream silver eels collected from the St. Lawrence River at La regression and nonlinear regression. All data were converted to Pocatiere. logarithms before statistical analysis due to the large ranges in values and the resultant non-normal distributions and heteroge- and in-house reference tissue (Great Lakes Institute for Envi- neous variances. Log-transformed data met the assumptions of ronmental Research [GLIER] Detroit River fish homogenate) normality and homogeneity of variance. were also analyzed. Organochlorine pesticides, which included The proportional contributions of potential prey items to the mirex, were determined by using an external quantification ap- diets of putative Lake Ontario eels (e.g., upstream Ontario sil- proach as described by Drouillard and Norstrom (2003)—that ver eels and downstream Ontario silver eels) were evaluated is, using an average response factor for equivalent compounds in using mixing models run in MixSIR (Inger and Bearhop 2008), organochlorine secondary standard mixtures. The nominal de- a computational program that can provide estimates of the rel- tection limit was 0.3 ng/g of wet weight. Blanks and reference ative contributions of sources to an organism’s diet (Moore and tissues that were quantified during each batch of sample ex- Semmens 2008). Using uninformative priors and estimates of tractions were in compliance with the normal quality assurance uncertainty associated with mixing model inputs, each MixSIR procedures instituted by GLIER’s Organic Analytical Labora- model ran for 109 iterations, resulting in convergence on the tory (certified by the Canadian Association for Environmental posterior source contributions of the different prey items con- Analytical Laboratories). tributing to the eels’ diets. The maximum importance ratio was We used a histogram (Figure 2) of the numbers of eels in 17 below 0.001, suggesting that our models were effective in esti- classes of log(muscle mirex concentration) to visually discrim- mating the true posterior density (Moore and Semmens 2008). inate the presence of a bimodal distribution and the extent of All mixing models involve underlying assumptions about each mode. This method was used to assign eels to a putative which food resources to include and which trophic fractiona- non-Lake Ontario group (hereafter, “downstream other silver tion factors to apply. Our decision to include Alewives, Rain- eels”) based on the lower mode or to a Lake Ontario feeding bow Smelt, Round Goby, Slimy Sculpin, and crayfish in our group (hereafter, “downstream Ontario silver eels”) based on mixing models was based in part on agency records relevant the upper mode. to the eels’ diets, but since these records were extremely lim- Stable isotope analysis.—Stable isotope analyses of 13C and ited, we also included prey fish that were found to be important Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 15N were conducted on lipid-extracted samples to eliminate the for salmonines in Lake Ontario and that might also be con- effects of lipid on fractionation (Post et al. 2007). This was sumed by eels (Rand and Stewart 1998; Lantry 2001; Dietrich done primarily to eliminate the effects of lipids on δ13C rather et al. 2006; Paterson et al. 2009; Rush et al. 2012). Although than δ15N and to expedite sample processing, although lipid Threespine Sticklebacks Gasterosteus aculeatus and Emerald extraction appears to have little effect on the fractionation of Shiner Notropis atherinoides are also present in Lake Ontario δ15N (Murry et al. 2006). Lipids were extracted and quantified (Mills et al. 2003), their abundances when eel collections were using methanol and chloroform by following the methods of made were extremely low, so these species were not included Folch et al. (1957) as revised by Post and Parkinson (2001) and (Dietrich et al. 2006). For all modeled contributions, we used Arrington et al. (2006). isotope values obtained from lipid-extracted eel samples (this Lipid-extracted samples were dried for over 48 h at 40– study) and from potential prey items (Table 1). We used iso- ◦ 50 C and were ground into a fine powder. All stable isotope topic discrimination factors for Lake Trout muscle (δ13C: 0.05 values are reported in the delta (δ) notation: δ13Corδ15N = ± 0.23‰; δ15N: 3.49 ± 1.08‰) because no isotopic discrimi- 13 12 15 14 [(Rsample/Rstandard)–1] × 1,000, where R is C/ Cor N/ N. nation factors exist for eels; the Lake Trout, like the American The global standards are Pee Dee belemnite for δ13C and atmo- Eel, is a top predator, and the factors for Lake Trout are similar spheric nitrogen for δ15N. The working standard for animals was to those reported for a broad range of species (Vander Zanden THIAMINE STATUS OF AMERICAN EELS 1363

TABLE 1. Mean ( ± SD) stable isotope signatures (δ15Nandδ13C) for taxa 16 that were potentially consumed by American Eels in Lake Ontario (fishes: from Fitzsimons, unpublished data; crayfish: this study). 14

Taxon δ15N(‰) δ13C(‰) 12 Round Goby 15.50 ± 1.17 −23.22 ± 1.46 10 Slimy Sculpin 17.41 ± 0.60 −24.66 ± 0.67 8 Alewife 13.67 ± 0.73 −23.63 ± 1.13 6 Rainbow Smelt 15.95 ± 0.83 −23.95 ± 0.53 Crayfish 9.10 ± 1.70 −17.10 ± 0.50 4 Muscle thiamine (nmol/g) 2

0 and Rasmussen 2001; Post 2002). The results of the MixSIR models are presented as medians with 5th and 95th credibility 010203040 intervals. Age (years)

FIGURE 3. Relationship between muscle thiamine concentration and age of RESULTS American Eels: juvenile eels (diamonds), yellow eels (circles), upstream Ontario silver eels (squares), and downstream Ontario silver eels (triangles). Line shows Collections the fit to the data. There was significant among-location (among-group) vari- ation in weight (F4, 85 = 741.9, P < 0.0001), TL (F4, 85 = δ13 δ15 13.70, P < 0.0001), and age (F = 52.0, P < 0.001) of eels stream other silver eels). Similarly, neither C nor Nwas 4, 85 > (Table 2). Juvenile eels were significantly (P < 0.05) lighter related to TL for any of these groups (P 0.05). than eels from all other groups; average weights of the other groups were not significantly different. Similarly, juvenile eels Thiamine were significantly (P < 0.05) shorter than eels from all other There was significant among-location variation in muscle = < groups, but average TL was not significantly different among thiamine concentration (F4, 85 18.4, P 0.001; Table 2). Thi- the other groups. The mean age ± SE of juvenile eels (7.1 ± amine concentrations in juvenile eels were significantly higher 0.4 years) was significantly younger than that of yellow eels than those in all other groups. Downstream other silver eels had (17.5 ± 0.6 years). Mean ages of the upstream Ontario silver significantly higher thiamine concentrations than yellow eels, eels (26.2 ± 2.1 years), downstream Ontario silver eels (24.0 ± upstream Ontario silver eels, and downstream Ontario silver 2.0 years), and downstream other silver eels (25.2 ± 1.1 years) eels, whereas the latter three groups did not significantly differ were similar but significantly (P < 0.05) greater than that of in thiamine concentration. There was no relationship between juvenile eels or yellow eels; the exception was that the mean thiamine and age within any given group of eels. Among juvenile age of downstream other silver eels was similar to the mean age eels, yellow eels, upstream Ontario silver eels, and downstream of yellow eels. Location had no effect on either δ13Corδ15N Ontario silver eels, there was a significant exponential decline = for eels collected from the St. Lawrence River (e.g., upstream in thiamine concentration with increasing age (F1, 46 70.4, Ontario silver eels, downstream Ontario silver eels, and down- r2 = 0.60, P < 0.001; Figure 3). Downloaded by [Department Of Fisheries] at 21:34 27 October 2013

TABLE 2. Summary of collection year, sample size (N), TL, total weight, age, lipid percentage, and muscle δ13C, δ15N, thiamine concentration, and mirex concentration for American Eels collected from the Lake Ontario–St. Lawrence River corridor (ND = no data). Means are presented with SEs (range is shown in parentheses).

TL Weight Age Lipid δ13C δ15N Thiamine Mirex Group Year N (cm) (g) (years) (%) (‰) (‰) (nmol/g) (ng/g)

Juvenile eels 2005 17 40.1 ± 12.7 900 ± 10.0 7.1 ± 0.4 ND ND ND 6.75 ± 0.56 ND (34–50) (30–152) (5–12) (3.56–13.89) Yellow eels 2003 15 101.0 ± 2.5 2,320 ± 190 17.5 ± 0.6 ND ND ND 1.73 ± 0.30 ND (82–108) (34–50) (14–22) (0.55–4.14) Upstream Ontario 2005 9 105.6 ± 3.6 2,390 ± 130 26.2 ± 2.1 10.0 ± 1.6 −17.8 ± 0.6 15.8 ± 0.3 1.67 ± 0.25 8.0 ± 2.4 silver eels (93–128) (1,760–2,840) (17–35) (5.0–17.8) (−20.6 to −14.7) (14.5–18.3) (0.72–3.11) (2.1–21.9) Downstream Ontario 2005 10 106.2 ± 2.0 2,793 ± 173 24.0 ± 2.0 13.7 ± 1.2 −19.3 ± 0.5 16.2 ± 0.02 1.59 ± 0.26 27.6 ± 3.1 silver eels (84–111) (1,676–3,784) (13–36) (6.7–19.2) (−23.3 to −17.2) (15.2–16.9) (0.62–2.69) (16.4–46.1) Downstream other 2005 40 98.1 ± 1.4 2,143 ± 120 25.2 ± 1.1 12.2 ± 0.6 −18.1 ± 0.4 15.2 ± 0.2 3.65 ± 0.52 2.6 ± 0.4 silver eels (80–113) (568–3,494) (9–34) (6.7–23.8) (−26.5 to −14.0) (10.1–17.5) (0.75–15.87) (0.1–7.3) 1364 FITZSIMONS ET AL.

Diet Analysis The most parsimonious explanation for low muscle thiamine Based on estimates from the mixing model that used the sta- concentrations in Lake Ontario-resident eels relative to non- ble isotope signatures of eels (Table 2) and their potential prey Lake Ontario-resident eels was the high consumption of crayfish (Table 1), crayfish comprised the highest proportion (83.2%; rather than consumption of Alewives as expected. The role of credibility interval = 60.5–93.5%) of the diet for upstream crayfish in the diets and thiamine dynamics of Lake Ontario eels Ontario silver eels. Round Goby contributed the next-highest would, however, require additional study for confirmation. Of diet proportion (14.6%; credibility interval = 0.4–38.6%) for the two primary taxa contributing to the diets of Lake Ontario upstream Ontario silver eels, whereas the combined amounts eels as indicated by our analysis, Round Goby exhibit a broad of Alewives, Slimy Sculpin, and Rainbow Smelt in the diet range in thiaminase activity (Tillitt et al. 2005; Honeyfield et al. were less than 1%. Downstream Ontario silver eels’ diets were 2012). However, their thiaminase activity is too low or their similar to those of upstream Ontario silver eels; proportional consumption is too low, as a diet consisting of Round Goby contributions to the diet were similar between crayfish (48.8%; was not associated with thiamine deficiency in Lake Michigan credibility interval = 42.0–55.1%) and Round Goby (43.1%; Lake Trout, whereas the consumption of Alewives was asso- credibility interval = 39.8–58.1%). The combined amount of ciated with thiamine deficiency (Jaroszewska et al. 2009). The Alewives, Slimy Sculpin, and Rainbow Smelt in the diets of crayfish Procambarus clarkii reportedly contains thiaminase, downstream Ontario silver eels was 2%. but there are no reports of thiamine deficiency in fish related to the consumption of crayfish (Rutledge and Ying 1972). In the Great Lakes, the crayfishes Orconectes propinquus, Orconectes DISCUSSION virilis, and Orconectes rusticus all readily consume dreissenid Our data provide clear evidence that Lake Ontario eels— mussels, which are a known source of elevated thiaminase ac- like Lake Ontario Chinook Salmon Oncorhynchus tshawytscha, tivity that is 25-fold higher than the thiaminase activity reported Coho Salmon Oncorhynchus kisutch, Lake Trout, Rainbow for Lake Michigan Alewives (Tillitt et al. 2009). However, it Trout Oncorhynchus mykiss, Brown Trout, and Walleyes Sander is not known whether the thiaminase found in dreissenids is vitreus (Johnston et al. 2005; Fitzsimons et al. 2007)—are af- accumulated by predators like crayfish or Round Goby (Ray fected by thiamine deficiency, but unexpectedly this was unre- and Corkum 1997). Even if the elevated thiaminase activity of lated to eels’ consumption of Alewives. Thiamine concentra- Round Goby is related to their consumption of dreissenids, it tions in downstream other silver eels were significantly higher appears that levels were insufficient to cause thiamine deficiency than those in downstream Ontario silver eels. Given the similar- in Lake Trout in Lake Michigan (Jaroszewska et al. 2009). No ity in weight, TL, and age of downstream other silver eels and information is available on thiaminase activity in Great Lakes downstream Ontario silver eels, it is highly unlikely that size- crayfish, and more work would be required to confirm levels or age-related factors can explain the difference in thiamine and their potential to cause thiamine deficiency. Thiamine lev- levels between the two groups. Both groups showed lower thi- els in Round Goby are adequate for fish nutrition (Honeyfield amine relative to juvenile eels; the decline was less than twofold et al. 2012), and while thiamine levels in decapods like crayfish for downstream other silver eels, whereas it was over fourfold appear sufficient to support fish nutrition (Niimi et al. 1997), the for downstream Ontario silver eels. The declines in thiamine thiamine levels in Great Lakes crayfish have not been reported. for yellow eels and upstream Ontario silver eels were simi- The difference in thiamine levels between juvenile eels and lar in magnitude to those for downstream Ontario silver eels. downstream other silver eels does not appear to involve the pres- Not all of the difference in thiamine levels between Lake On- ence of a marine clupeid in the diet for downstream other silver tario eels and juvenile eels can be attributed to the effects of eels. Seasonal spawning emigrations of thiaminase-rich marine Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 diet, and some apparently involved changes that were related clupeids (Neilands 1947) in the St. Lawrence River have been to ontogeny or other factors. Ontogenetic changes in muscle documented to extend as far upstream as Montreal´ and last for at thiamine concentration have also been observed in Lake Trout least 2 months (Maltais et al. 2010). Eels that are resident in the with no suspected thiaminase in their diets, although the mag- lower St. Lawrence River could potentially feed upon the clu- nitude of these ontogenetic changes was much smaller than peids involved in these migrations, but the 2-month migration that observed in Lake Trout exhibiting an ontogenetic change period seems insufficient to substantially affect thiamine levels, in the proportion of Alewives in the diet (Fitzsimons, unpub- and the thiamine levels would be expected to recover once the lished data). Evidently, for Lake Ontario eels, thiamine defi- clupeids ceased to be part of the diet (Fitzsimons et al. 2010). ciency is not restricted to the oldest individuals; thiamine levels Moreover, there was no indication that δ13C varied between ju- in yellow eels that had been resident in Lake Ontario for ap- venile eels and downstream other silver eels, which would have proximately 10 years (based on the average age of juvenile eels been expected if downstream other silver eels had fed heavily collected at the eel ladder) were indistinguishable from those on marine prey as opposed to freshwater prey. A predictable of either upstream Ontario silver eels or downstream Ontario gradient of 13C exists across the marine–freshwater ecotone silver eels, which were, on average, 6 years older than yellow due to relative depletion of 13C in freshwater (Fry and Sherr eels. 1984; Fry 2002). Fish feeding in different locations along this THIAMINE STATUS OF AMERICAN EELS 1365

salinity gradient should therefore have contrasting carbon iso- the only other lake-resident eel population in which diet has tope signatures. Harrod et al. (2005) demonstrated significant been examined (Facey and LaBar 1981). The percentages of differences in 13C among European Eels Anguilla anguilla feed- fish (average = 48%) and crayfish (average = 49%) in diets ing on freshwater, brackish-water, or marine prey. for upstream Ontario silver eels and downstream Ontario silver The presence of Round Goby in the diets of Lake Ontario eels eels were higher than those observed for Lake Champlain eels; as reflected in the results of the mixing models is consistent with the percent occurrence of fish and crayfish (combined) in the the increasing abundance of Round Goby in Lake Ontario and diet averaged 26% for Lake Champlain eels, and the remaining with a benthic feeding mode (Walsh et al. 2007). Abundance 48% of the diet mostly comprised non-crayfish invertebrates. of Round Goby in Lake Ontario was relatively low in 2005, The difference between Lake Ontario eels and Lake Champlain when the eels were collected, and their abundance increased by eels may be attributable to the larger size of Lake Ontario eels; over threefold between 2005 and 2008 (M. Walsh, U.S. Geolog- the average size of Lake Champlain eels (67 cm; Facey and ical Survey [USGS], personal communication). However, Lake Labar 1981) was almost 30 cm less than the average size of eels Trout that were collected in 2004 from eastern Lake Ontario had examined in this study. already incorporated Round Goby into their diet (Dietrich et al. Thiamine levels of Lake Ontario eels appeared to be estab- 2006). The Lake Ontario eels’ diet as inferred from our study lished in Lake Ontario, well before their oceanic migration to results suggests that the feeding mode of eels while they inhabit the Sargasso Sea. Once the eels left Lake Ontario, thiamine sta- Lake Ontario is primarily benthic, reflecting their preferred habi- tus showed no appreciable change between eels collected in the tat. This, along with a declining trend in Alewife abundance in St. Lawrence River 100 km downstream of Lake Ontario and Lake Ontario, may explain the lack of Alewife consumption those collected 600 km downstream of the lake. Eels collected as indicated by the mixing model results (Mills et al. 2003). in the St. Lawrence River ranged in age from 24 to 26 years Round Goby and crayfish, the primary prey of eels based on and were 6–8 years older than eels collected from Lake Ontario, the mixing model results, are found primarily in rocky benthic where the average age was 18 years. The lack of change in thi- habitat (Ray and Corkum 2001; Fitzsimons et al. 2002), which amine status between eels at locations 600 km and 100 km from is the same habitat used by eels (Casselman 2008; J. Cassel- the Gulf of St. Lawrence may be due to a reduction or cessa- man, Queens University, personal communication; Fitzsimons, tion of feeding associated with degeneration of the alimentary unpublished data). At the Port Weller collection location, scuba tract, which occurs just prior to or during the migration of sil- divers observed that eels were closely associated with nearshore ver (maturing) eels (Pankhurst and Sorensen 1984). As a result, rocky habitat, where they swam through the interstices (J. D. thiamine levels in Lake Ontario eels may be highly dependent Fitzsimons, unpublished observations). Similarly, an abundance on the nature of the Lake Ontario food web (Mills et al. 2003), index for eels in eastern Lake Ontario was based on the num- which has undergone considerable change due to multiple in- ber of eels (caught by electrofishing) that were associated with vasions by nonnative aquatic species but with largely unknown large rock and boulder habitat in the vicinity of Main Duck and consequences for resident fishes. Yorkshire islands (Casselman 2003). Unlike thiamine deficiency in salmonines (Brown et al. Depending on the relative importance of Round Goby in the 2005a), much less is known about the effects of thiamine defi- diets of Lake Ontario eels in the future and assuming that they are ciency on eels (Hashimoto et al. 1970; Arai et al. 1972). Muscle not the cause of thiamine deficiency, Round Goby may improve thiamine levels that are required by American Eels to complete the thiamine status of eels relative to a situation in which crayfish their oceanic migration and thiamine levels that are sufficient dominate the diet. Expanded utilization of Round Goby would to support ovary development have not been identified. The thi- be seemingly enhanced by the preference of eels for interstitial amine concentrations in downstream Ontario silver eels (1.6 Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 spaces in rock and rubble substrate, the same habitat preferred by nmol/g; but not in downstream other silver eels: 3.6 nmol/g) Round Goby (Johnson et al. 2005). In addition, because of their appeared to be close to a putative threshold effect level, but relatively high preferred temperature (e.g., 16.7◦C; Barila and this is based on a single laboratory study (Hashimoto et al. Stauffer 1980), eels may be able to make use of Round Goby 1970) for another eel species, so our conclusions require con- that occur primarily in shallower, warmer waters during the firmation. In that study, after Japanese Eels Anguilla japon- summer (Walsh et al. 2007). Recently, Round Goby have been ica were fed clams containing thiaminase (Neilands 1947), the found in the stomachs of eels occupying shallow nearshore areas whole-body thiamine levels in the eels declined (Hashimoto of Lake Ontario (Joshua Stacey, Queens University, personal et al. 1970). Hemorrhaging was observed in Japanese Eels with communication). Additional work will be required to determine muscle thiamine concentrations of approximately 1.5 nmol/g, the extent to which Round Goby can contribute to an increase and reduced growth and survival were evident in eels that had in thiamine levels of Lake Ontario eels. muscle thiamine concentrations of approximately 0.8 nmol/g. For the two eel groups that were evaluated with the mixing Japanese Eels on the clam diet also lost their appetite, exhibited model (upstream Ontario silver eels and downstream Ontario reduced growth, ataxia, and abnormal swimming; at a very ad- silver eels), the inferred diet showed a higher proportion of fish vanced stage of thiamine deficiency, they developed rigidity of and crayfish than was reported for eels from Lake Champlain, the trunk and severe flexion. It is not known whether sensitivity 1366 FITZSIMONS ET AL.

to thiamine deficiency is similar among anguillids or how repre- For species that, like eels, have large geographic distribu- sentative the data for Japanese Eels may be for American Eels. tions, there are few natural or anthropogenic processes that Nevertheless, among salmonines, sensitivity to thiamine defi- can simultaneously affect population dynamics across the entire ciency shows relatively little variation across life stages or across range, but the effects of such processes are likely exacerbated by species (Brown et al. 2005b; Fitzsimons et al. 2007); thresholds panmixia. Among the natural processes, oceanic changes in the for larval and adult mortality varied by less than twofold among Sargasso Sea and the spread of the infectious nematode Anguil- species. Moreover, the threshold muscle thiamine concentration licoloides crassus (Moser et al. 2001) seem to be of increasing associated with mortality of adult salmonines was quite sim- concern for eel populations, and it would appear that thiamine ilar to that reported by Hashimoto et al. (1970) for Japanese deficiency can be added to this list. Large-scale perturbations in Eels. the Sargasso Sea have the potential to impact the entire Ameri- The demands of oceanic migration and an ongoing need for can Eel species as a result of disrupted reproductive success and thiamine could easily put Lake Ontario-resident eels into a more reduced larval survival. The spread of A. crassus (Moser et al. severe thiamine deficiency than that observed in eels collected 2001) is hypothesized to be one of the ecological factors driving from the lower St. Lawrence River just prior to migration. This the widespread decline in American Eel abundances (Haro et al. could have negative implications for adult survival and repro- 2000) as a result of reductions in growth (Thomas and Ollevier ductive success. At an estimated swimming speed of approx- 1992) and swimming ability (Sprengel and Luchtenberg¨ 1991). imately 40 km/d (Tesch 1974; Van Den Thillart et al. 2004), Reduced growth and reduced swimming ability will adversely a Lake Ontario eel that has left the lower St. Lawrence River affect migrating silver eels. Similarly, thiamine deficiency has would take approximately 90 d or 3 months to reach the Sargasso the potential to (1) reduce the swimming ability of eels as they Sea—a distance of 3,500 km (Tremblay 2009). Since eels stop return to the Sargasso Sea (Ketola et al. 2005, 2009) and (2) af- feeding and assimilating nutrients prior to migration (Pankhurst fect reproductive success through thiamine deposition into the and Sorensen 1984; Chow et al. 2010), they must rely entirely ovaries (Honeyfield et al. 2005) that is insufficient to support on body stores of thiamine that accumulated prior to migration normal larval development (Brown et al. 1998a; Fitzsimons et al. to support the thiamine requirements of routine metabolism and 2009). Low thiamine levels in larval stages can negatively affect swimming as well as oogenesis. For Chinook Salmon (D. Hon- the use of the yolk sac and transition to exogenous feeding as eyfield, USGS, unpublished data) and Atlantic Salmon Salmo well as the ability to forage effectively and to avoid predation salar (Fynn-Aikins et al. 1998), thiamine clearance rates dur- (Fitzsimons et al. 2009, 2012). Although thiamine deficiency ing starvation ranged from 0.110 to 0.818 nmol·g−1·month−1. has only been documented for eels that are resident in Lake Thiamine clearance rates for Japanese Eels during starvation Ontario, during years of high abundance this eel population were higher, estimated at 1.01 nmol·g−1·month−1 based on data constituted 54–67% of the annual spawn output from Canadian reported by Hashimoto et al. (1970), but this would require con- waters, and so it demographically made a large contribution to firmation for American Eels. Nevertheless, assuming that the the species (COSEWIC 2006). mean thiamine level of 1.6 nmol/g for downstream Ontario sil- To conclude, Lake Ontario American Eels are affected by ver eels represents the concentration at the start of the estimated an apparent crayfish-diet-induced thiamine deficiency, although 3-month-long oceanic migration and assuming a thiamine clear- more work would be required to confirm the role of crayfish in ance rate of 1.01 nmol·g−1·month−1, we predict that many Lake the diet and in thiamine dynamics. The thiamine deficiency has Ontario eels would have (1) insufficient muscle thiamine to the potential to affect oceanic migration of adult eels as well complete migration, (2) insufficient thiamine to deposit into the as their reproductive success in the Sargasso Sea. Because of developing ovaries for normal egg and larval development, or (3) panmixia and because of the large demographic contribution Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 both, before succumbing to the effects of thiamine deficiency. attributed to the Lake Ontario subpopulation, such effects may In contrast, because the thiamine concentration in downstream have consequences for the entire species. Incorporation of these other silver eels (3.7 nmol/g) averaged almost 2 nmol/g higher effects into a panmictic modeling framework (Velez-Espino´ and than that in downstream Ontario silver eels, the downstream Koops 2010) will be necessary to develop an understanding of other silver eels would have higher thiamine reserves to draw the range of thiamine deficiency effects within the range of on to support migration and ovarian development and, as a re- ecological and environmental processes that control variation in sult, would make a greater contribution to recruitment. Using the abundance and life history of American Eels. the same thiamine clearance rate of 1.01 nmol·g−1·month−1, downstream other silver eels are predicted to have sufficient thi- amine to complete migration. Additional work with American ACKNOWLEDGMENTS Eels held under simulated migratory conditions would be re- We would like to thank Georgina Williston, Bill Williston, quired to confirm this supposition. Preliminary work indicates Bud Timmins, and Dan Walsh for their assistance in processing that thiamine levels in 44–53% of the Lake Ontario yellow eels samples. Kevin McGrath kindly provided eel samples from the and silver eels were so low as to impair swimming performance Moses–Saunders Dam at Cornwall. We thank Cornell University (D. Honeyfield, USGS, personal communication). for conducting the stable isotope analysis. THIAMINE STATUS OF AMERICAN EELS 1367

REFERENCES station at Cornwall, Ontario, 1974–1995. Pages 161–169 in R. H. Peterson, Arai, S., T. Nose, and Y. Hashimoto. 1972. Qualitative requirements of young editor. The American Eel in eastern Canada: stock status and management eels Anguilla japonica for water soluble vitamins and their deficiency symp- strategies: proceedings of eel workshop, January 13–14, 1997, Quebec City, toms. Bulletin of Freshwater Fisheries Research Laboratory (Tokyo) 22:69– QC. Canadian Technical Report of Fisheries and Aquatic Sciences 2196. 83. Castonguay, M., J. D. Dutil, and C. Desjardins. 1989. Distinction between Arrington, D. A., B. K. Davidson, K. O. Winemiller, and C. A. Layman. 2006. American Eels (Anguilla rostrata) of different geographic origins on the basis Influence of life history and seasonal hydrology on lipid storage in three of their organochlorine contaminant levels. Canadian Journal of Fisheries and neotropical fish species. Journal of Fish Biology 68:1347–1361. Aquatic Sciences 46:836–843. Axelsen, F. 1997. The status of the American Eel (Anguilla rostrata) stock Castonguay, M., P. V. Hodson, C. M. Couillard, M. J. Eckersley, J. D. Dutil, in Quebec. Pages 121–133 in R. H. Peterson, editor. The American Eel and G. Verreault. 1994. Why is recruitment of the American Eel, Anguilla in eastern Canada: stock status and management strategies: proceedings of rostrata, declining in the St. Lawrence River and Gulf? Canadian Journal of eel workshop, January 13–14, 1997, Quebec City, QC. Canadian Technical Fisheries and Aquatic Sciences 51:479–488. Report of Fisheries and Aquatic Sciences 2196. Chow, S., H. Kurogi, S. Katayama, D. Ambe, M. Okazaki, T. Watanabe, T. Barbin, G. P.,and J. D. McCleave. 1997. Fecundity of the American Eel Anguilla Ichikawa, M. Kodama, J. Aoyama, A. Shinoda, S. Watanabe, K. Tsukamoto, rostrata at 45◦N in Maine, U.S.A. Journal of Fish Biology 51:840–847. S. Miyazaki, S. Kimura, Y. Yamada, K. Nomura, H. Tanaka, Y. Kazeto, K. Barila, T. Y., and J. R. Stauffer Jr. 1980. Temperature behavioral responses of the Hata, T. Handa, A. Tawa, and N. Mochioka. 2010. Japanese Eel Anguilla American Eel, Anguilla rostrata (Lesueur), from Maryland. Hydrobiologia japonica do not assimilate nutrition during the oceanic spawning migra- 74:49–51. tion: evidence from stable isotope analysis. Marine Ecology Progress Series Braune, B. M., M. L. Mallory, H. G. Gilchrist, R. J. Letcher, and K. G. Drouillard. 402:233–238. 2007. Levels and trends of organochlorines and brominated flame retardants COSEWIC (Committee on the Status of Endangered Wildlife in Canada). in ivory gull eggs from the Canadian Arctic, 1976 to 2004. Science of the 2006. COSEWIC assessment and status report on the American Eel Anguilla Total Environment 378:403–417. rostrata in Canada. COSEWIC, Ottawa. Available: www.sararegistry.gc.ca/ Brown, S. B., J. D. Fitzsimons, D. C. Honeyfield, and D. E. Tillitt. 2005a. virtual sara/files/cosewic/sr american eel e.pdf. (August 2012). Implications of thiamine deficiency in Great Lakes salmonines. Journal of Cotˆ e,´ C. L., M. Castonguay, G. Verreault, and L. Bernatchez. 2009. Differential Aquatic Animal Health 17:113–124. effects of origin and salinity rearing conditions on growth of glass eels of Brown, S. B., J. D. Fitzsimons, V. P. Palace, and L. Vandenbyllaardt. 1998a. the American Eel Anguilla rostrata: implications for stocking programmes. Thiamine and early mortality syndrome in Lake Trout. Pages 18–25 in G. Journal of Fish Biology 74:1934–1948. McDonald, J. D. Fitzsimons, and D. C. Honeyfield, editors. Early life stage Couillard, C. M., P. V. Hodson, and M. Castonquay. 1997. Correlations between mortality syndrome in fishes of the Great Lakes and Baltic Sea. American pathological changes and chemical contamination in American Eels, Anguilla Fisheries Society, Symposium 21, Bethesda, Maryland. rostrata, from the St. Lawrence River. Canadian Journal of Fisheries and Brown, S. B., D. C. Honeyfield, J. G. Hnath, M. Wolgamood, S. V.Marcquenski, Aquatic Sciences 54:1916–1927. J. D. Fitzsimons, and D. E. Tillitt. 2005b. Thiamine status in adult salmonines DeAngelis, D. L., L. W. Barnthouse, W. Van Winkle, and R. G. Otto. 1990. in the Great Lakes. Journal of Aquatic Animal Health 17:59–64. A critical appraisal of population approaches in assessing fish community Brown, S. B., D. C. Honeyfield, and L. Vandenbyllaardt. 1998b. Thiamine health. Journal of Great Lakes Research 16:576–590. analysis in fish tissues. Pages 73–81 in G. McDonald, J. D. Fitzsimons, and de Lafontaine, Y., P. Gagnon, and B. Cotˆ e.´ 2010. Abundance and individual size D. C. Honeyfield, editors. Early life stage mortality syndrome in fishes of of American Eel (Anguilla rostrata) in the St. Lawrence River over the past the Great Lakes and Baltic Sea. American Fisheries Society, Symposium 21, four decades. Hydrobiologia 647:185–198. Bethesda, Maryland. de Lafontaine, Y., M. Lagace,´ F. Gingras, D. Labonte,´ F. Marchand, and E. Byer, J. D., M. Lebeuf, M. Alaee, B. R. Stephen, S. Trottier, S. Backus, Lacroix. 2009. Decline of the American Eel in the St. Lawrence River: ef- M. Keir, C. M. Couillard, J. Casselman, and P. V. Hodson. 2013. Spatial fects of local hydroclimatic conditions on CPUE indices. Pages 207–228 in trends of organochlorinated pesticides, polychlorinated biphenyls, and poly- J. M. Casselman and D. K. Cairns, editors. Eels at the edge: science, sta- brominated diphenyl ethers in Atlantic anguillid eels. Chemosphere 90:1719– tus, and conservation concerns. American Fisheries Society, Symposium 58, 1728. Bethesda, Maryland. Cairns, D. K., D. L. Omilusik, P. H. Leblanc, E. G. Atkinson, D. S. Moore, and Dietrich, J. P., B. J. Morrison, and J. A. Hoyle. 2006. Alternative ecological N. McDonald. 2007. American Eel abundance indicators in the southern Gulf pathways in the eastern Lake Ontario food web: Round Goby in the diet of of St. Lawrence. Canadian Data Report of Fisheries and Aquatic Sciences Lake Trout. Journal of Great Lakes Research 32:395–400.

Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 1192. Drouillard, K. G., and R. J. Norstrom. 2003. The influence of diet properties Cairns, D. K., D. A. Secor, W. E. Morrison, and J. A. Hallett. 2009. Salinity- and feeding rates on PCB toxicokinetics in the ring dove. Archives of Envi- linked growth in anguillid eels and the paradox of temperate-zone catadromy. ronmental Contamination and Toxicology 44:97–106. Journal of Fish Biology 74:2094–2114. Dutil, J. D., B. Legar´ e,´ and C. Desjardins. 1985. Discrimination d’un stock de Cairns, D. K., V. Tremblay, F. Caron, J. M. Casselman, G. Verreault, B. M. poisson, l’anguille (Anguilla rostrata), basee´ sur la presence´ d’un produit Jessop, Y. de Lafontaine, R. G. Bradford, R. Verdon, P. Dumont, Y. Mailhot, chimique de synthese,` le mirex. Canadian Journal of Fisheries and Aquatic J. Zhu, A. Mathers, K. Oliveira, K. Benhalima, J. P. Dietrich, J. A. Hallett, and Sciences 42:455–458. M. Lagace.´ 2008. American Eel abundance indicators in Canada. Canadian Facey, D. E., and G. W. LaBar. 1981. Biology of American Eels in Lake Cham- Data Report of Fisheries and Aquatic Sciences 1207. plain, Vermont.Transactions of the American Fisheries Society 110:396–402. Casselman, J. M. 2003. Dynamics of resources of the American Eel, Anguilla Fitzsimons, J. D., and S. B. Brown. 1998. Reduced egg thiamine levels in inland rostrata: declining abundance in the 1990s. Pages 255–274 in K. Aida, K. and Great Lakes Lake Trout and their relationship with diet. Pages 160–171 in Tsukamoto, and K. Yamauchi, editors. Eel biology. Springer-Verlag, Tokyo. G. McDonald, J. D. Fitzsimons, and D. C. Honeyfield, editors. Early life stage Casselman, J. M. 2008. Quantitative electrofishing protocol for sampling lo- mortality syndrome in fishes of the Great Lakes and Baltic Sea. American cations in the lower Ottawa and Mississippi rivers, fall 2008. AFishESci, Fisheries Society, Symposium 21, Bethesda, Maryland. Mallorytown, Ontario. Fitzsimons, J. D., S. Brown, L. Brown, D. Honeyfield, J. He, and J. E. Johnson. Casselman, J. M., L. A. Marcogliese, and P. V. Hodson. 1997. Recruitment 2010. Increase in Lake Trout reproduction in Lake Huron following the index for the upper St. Lawrence River and Lake Ontario eel stock: a re- collapse of Alewife: relief from thiamine deficiency or larval predation? examination of eel passage at the R.H. Saunders hydroelectric generating Aquatic Ecosystem Health and Management 13:73–84. 1368 FITZSIMONS ET AL.

Fitzsimons, J. D., S. B. Brown, D. C. Honeyfield, and J. G. Hnath. 1999. A review Jaroszewska, M., B. J. Lee, K. Dabrowski, S. Czesny, J. Rinchard, P. Trzeciak, of early mortality syndrome (EMS) in Great Lakes salmonids: relationship and B. Wilczynska.´ 2009. Effects of vitamin B1 (thiamine) deficiency in with thiamine deficiency. Ambio 28:9–15. Lake Trout (Salvelinus namaycush) alevins at hatching stage. Comparative Fitzsimons, J. D., S. B. Brown, B. Williston, G. Williston, L. R. Brown, K. Biochemistry and Physiology 154A:255–262. Moore, D. C. Honeyfield, and D. E. Tillitt. 2009. Influence of thiamine Jessop, B. M. 2010. Geographic effects on American Eel (Anguilla rostrata) deficiency on Lake Trout larval growth, foraging, and predator avoidance. life history characteristics and strategies. Canadian Journal of Fisheries and Journal of Aquatic Animal Health 21:302–314. Aquatic Sciences 67:326–346. Fitzsimons, J. D., D. L. Perkins, and C. C. Krueger. 2002. Sculpins and crayfish Johnson, T. B., M. Allen, L. D. Corkum, and V. A. Lee. 2005. Comparison in Lake Trout spawning areas in Lake Ontario: estimates of abundance and of methods needed to estimate population size of Round Gobies (Neogobius egg predation on Lake Trout eggs. Journal of Great Lakes Research 28:421– melanostomus) in western Lake Erie. Journal of Great Lakes Research 31:78– 436. 86. Fitzsimons, J. D., B. Williston, P. Amcoff, L. Balk, C. Pecor, H. G. Ketola, J. Johnston, T. A., L. M. Miller, D. M. Whittle, S. B. Brown, M. D. Wiegand, P. Hinterkopf, and D. C. Honeyfield. 2005. The effect of thiamine injection A. R. Kapuscinski, and W. C. Leggett. 2005. Effects of maternally trans- on upstream migration, survival, and thiamine status of putative thiamine- ferred organochlorine contaminants on early life survival in a freshwater fish. deficient Coho Salmon. Journal of Aquatic Animal Health 17:48–58. Environmental Toxicology and Chemistry 24:2594–2602. Fitzsimons, J. D., B. Williston, L. Vandenbyllaardt, A. El-Shaarawi, and S. B. Ketola, H. G., T. L. Chiotti, R. S. Rathman, J. D. Fitzsimons, D. C. Honeyfield, Brown. 2012. Use of a thiamine antagonist to evaluate the effects of thiamine P. J. Van Dusen, and G. E. Lewis. 2005. Thiamine status of Cayuga Lake deficiency on Lake Trout embryonic development. Journal of Great Lakes Rainbow Trout and its influence on spawning migration. North American Research 38:236–242. Journal of Fisheries Management 25:1281–1287. Fitzsimons, J. D., B. Williston, G. Williston, L. Brown, A. El-Shaarawi, L. Ketola, H. G., J. H. Johnson, J. Rinchard, F. J. Verdoliva, M. E. Penney, A. W. Vandenbyllaardt, D. Honeyfield, D. Tillitt, M. Wolgamood, and S. B. Brown. Greulich, and R. C. Lloyd. 2009. Effect of thiamine status on probability of 2007. Egg thiamine status of Lake Ontario salmonines 1995–2004 with em- Lake Ontario Chinook Salmon spawning in the upper or lower sections of phasis on Lake Trout. Journal of Great Lakes Research 33:93–103. Salmon River, New York. North American Journal of Fisheries Management Folch, J., M. Lees, and G. H. Sloane Stanley. 1957. A simple method for 29:895–902. the isolation and purification of total lipides from animal tissues. Journal of Lantry, J. R. 2001. Spatial and temporal dynamics of predation by Lake Ontario Biological Chemistry 226:497–509. trout and salmon. Master’s thesis. College of Environmental Science and Fry, B. 2002. Conservative mixing of stable isotopes across estuarine salinity Forestry, State University of New York, Syracuse. gradients: a conceptual framework for monitoring watershed influences on MacGregor, R., A. Mathers, P. Thompson, J. M. Casselman, J. M. Dettmers, S. downstream fisheries production. Estuaries 25:264–271. LaPan, T. C. Pratt, and B. Allen. 2008. Declines of American Eel in North Fry, B., and E. B. Sherr. 1984. δ13C measurements as indicators of carbon America: complexities associated with bi-national management. Pages 357– flow in marine and freshwater ecosystems. Contributions in Marine Science 381 in M. G. Schechter, N. J. Leonard, and W. W. Taylor, editors. International 27:13–47. governance of fisheries ecosystems: learning from the past, finding solutions Fynn-Aikins, K., P. R. Bowser, D. C. Honeyfield, J. D. Fitzsimons, and H. for the future. American Fisheries Society, Bethesda, Maryland. G. Ketola. 1998. Effect of dietary amprolium on tissue thiamin and Cayuga Maltais, E., G. Daigle, G. Colbeck, and J. J. Dodson. 2010. Spawning dynamics syndrome in Atlantic Salmon. Transactions of the American Fisheries Society of American Shad (Alosa sapidissima) in the St. Lawrence River, Canada– 127:747–757. USA. Ecology of Freshwater Fish 19:586–594. Garman, G. C., and S. A. Macko. 1998. Contribution of marine-derived or- Marcogliese, L. A., and J. M. Casselman. 2009. Long-term trends in size and ganic matter to an Atlantic coast, freshwater, tidal stream by anadromous abundance of juvenile American Eels ascending the upper St. Lawrence clupeid fishes. Journal of the North American Benthological Society 17:277– River. Pages 191–205 in J. M. Casselman and D. K. Cairns, editors. Eels 285. at the edge: science, status, and conservation concerns. American Fisheries Goodwin, K. R. 1999. American Eel subpopulation characteristics in the Po- Society, Symposium 58, Bethesda, Maryland. tomac River drainage, Virginia. Master’s thesis. Virginia Polytechnic Institute Mathers, A., and T. J. Stewart. 2009. Management of American Eels in Lake and State University, Blacksburg. Ontario and the upper St. Lawrence River. Pages 359–366 in J. M. Casselman Haro, A., W. Richkus, K. Whalen, A. Hoar, W. D. Busch, S. Lary, T. Brush, and D. K. Cairns, editors. Eels at the edge: science, status, and conservation and D. Dixon. 2000. Population decline of the American Eel: implications concerns. American Fisheries Society, Symposium 58, Bethesda, Maryland. for research and management. Fisheries 25(9):7–16. McGrath, K. J., J. Bernier, S. Ault, J. D. Dutil, and K. Reid. 2003. Differenti-

Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 Harrod, C., J. Grey, T. K. McCarthy, and M. Morrissey. 2005. Stable isotope ating downstream migrating American Eels Anguilla rostrata from resident analyses provide new insights into ecological plasticity in a mixohaline pop- eels in the St. Lawrence River. Pages 315–327 in D. A. Dixon, editor. Biol- ulation of European Eel. Oecologia 144:673–683. ogy, management, and protection of catadromous eels. American Fisheries Hashimoto, Y., S. Arai, and T. Nose. 1970. Thiamine deficiency symptoms ex- Society, Symposium 33, Bethesda, Maryland. perimentally induced in the eel. Bulletin of the Japanese Society of Scientific Mills, E. L., J. M. Casselman, R. Dermott, J. D. Fitzsimons, G. Gal, K. T. Holeck, Fisheries 36:791–797. J. A. Hoyle, O. E. Johannsson, B. F. Lantry, J. C. Makarewicz, E. S. Millard, Honeyfield, D. C., M. E. Daniels, L. R. Brown, M. T. Arts, M. G. Walsh, and S. B. I. F. Munawar, M. Munawar, R. O’Gorman, R. W. Owens, L. G. Rudstam, Brown. 2012. Survey of four essential nutrients and thiaminase activity in five T. Schaner, and T. J. Stewart. 2003. Lake Ontario: food web dynamics in a Lake Ontario prey fish species. Journal of Great Lakes Research 38:11–17. changing ecosystem (1970–2000). Canadian Journal of Fisheries and Aquatic Honeyfield, D. C., J. P. Hinterkopf, J. D. Fitzsimons, D. E. Tillitt, J. L. Zajicek, Sciences 60:471–490. and S. B. Brown. 2005. Development of thiamine deficiencies and early Moore, J. W., and B. X. Semmens. 2008. Incorporating uncertainty and prior mortality syndrome in Lake Trout by feeding experimental and feral fish information into stable isotope mixing models. Ecology Letters 11:470–480. diets containing thiaminase. Journal of Aquatic Animal Health 17:4–12. Morito, C. L. H., D. H. Conrad, and J. W. Hilton. 1986. The thiamin deficiency Honeyfield, D. C., C. S. Vandergoot, P. W. Bettoli, J. P. Hinterkopf, and J. L. signs and requirement of Rainbow Trout (Salmo gairdneri, Richardson). Fish Zajicek. 2007. Thiamine and fatty acid content of Walleye tissue from three Physiology and Biochemistry 1:93–104. southern U.S. reservoirs. Journal of Aquatic Animal Health 19:84–93. Moser, M. L., W. S. Patrick, and J. U. Crutchfield Jr. 2001. Infection of American Inger, R., and S. Bearhop. 2008. Applications of stable isotope analyses to avian Eels, Anguilla rostrata, by an introduced nematode parasite, Anguillicola ecology. Ibis 150:447–461. crassus, in North Carolina. Copeia 2001:848–853. THIAMINE STATUS OF AMERICAN EELS 1369

Murry, B. A., J. M. Farrell, M. A. Teece, and P. M. Smyntek. 2006. Effect der Deutschen Wissenschaftlichen Kommission fuer Meeresforschung 23: of lipid extraction on the interpretation of fish community trophic relation- 181–197. ships determined by stable carbon and nitrogen isotopes. Canadian Journal Thibault, I., J. J. Dodson, F. Caron, W. N. Tzeng, Y. Iizuka, and J. C. Shiao. of Fisheries and Aquatic Sciences 63:2167–2172. 2007. Facultative catadromy in American Eels: testing the conditional strategy Neilands, J. B. 1947. Thiaminase in aquatic animals of Nova Scotia. Journal of hypothesis. Marine Ecology Progress Series 344:219–229. the Fisheries Research Board of Canada 7a:94–99. Thomas, K., and F. Ollevier. 1992. Paratenic hosts of the swimbladder nematode Niimi, A. J., C. J. C. Jackson, and J. D. Fitzsimons. 1997. Thiamine dynamics Anguillicola crassus. Diseases of Aquatic Organisms 13:165–174. in aquatic ecosystems and its biological implications. Internationale Revue Tillitt, D. E., S. C. Riley, A. N. Evans, S. J. Nichols, J. L. Zajicek, J. Rinchard, der Gesamten Hydrobiologie und Hydrographie 82:47–56. C. A. Richter, and C. C. Krueger. 2009. Dreissenid mussels from the Great Oliveira, K. 1999. Life history characteristics and strategies of the American Lakes contain elevated thiaminase activity. Journal of Great Lakes Research Eel, Anguilla rostrata. Canadian Journal of Fisheries and Aquatic Sciences 35:309–312. 56:795–802. Tillitt, D. E., J. L. Zajicek, S. B. Brown, L. R. Brown, J. D. Fitzsimons, D. C. Pankhurst, N. W., and P. W. Sorensen. 1984. Degeneration of the alimentary Honeyfield, M. E. Holey, and G. M. Wright. 2005. Thiamine and thiaminase tract in sexually maturing European Anguilla anguilla (L.) and American Eels status in forage fish of salmonines from Lake Michigan. Journal of Aquatic Anguilla rostrata (LeSueur). Canadian Journal of Zoology 62:1143–1149. Animal Health 17:13–25. Paterson, G., D. M. Whittle, K. G. Drouillard, and G. D. Haffner. 2009. Declin- Tremblay, V. 2009. Reproductive strategy of female American Eels among five ing Lake Trout (Salvelinus namaycush) energy density: are there too many subpopulations in the St. Lawrence River watershed. Pages 85–102 in J. M. salmonid predators in the Great Lakes? Canadian Journal of Fisheries and Casselman and D. K. Cairns, editors. Eels at the edge: science, status, and con- Aquatic Sciences 66:919–932. servation concerns. American Fisheries Society, Symposium 58, Bethesda, Post, D. M. 2002. Using stable isotopes to estimate trophic position: models, Maryland. methods, and assumptions. Ecology 83:703–718. Van Den Thillart, G., V. Van Ginneken, F. Korner,¨ R. Heijmans, R. Van Der Post, D. M., C. A. Layman, D. A. Arrington, G. Takimoto, J. Quattrochi, and Linden, and A. Gluvers. 2004. Endurance swimming of European Eel. Journal C. G. Montana.˜ 2007. Getting to the fat of the matter: models, methods and of Fish Biology 65:312–318. assumptions for dealing with lipids in stable isotope analyses. Oecologia Vander Zanden, M. J., and J. B. Rasmussen. 2001. Variation in δ15Nandδ13C 152:179–189. trophic fractionation: implications for aquatic food web studies. Limnology Post, J. R., and E. A. Parkinson. 2001. Energy allocation strategy in young fish: and Oceanography 46:2061–2066. allometry and survival. Ecology 82:1040–1051. Velez-Espino,´ L. A., and M. A. Koops. 2010. A synthesis of the ecologi- Rand, P. S., and D. J. Stewart. 1998. Prey fish exploitation, salmonine produc- cal processes influencing variation in life history and movement patterns of tion, and pelagic food web efficiency in Lake Ontario. Canadian Journal of American Eel: towards a global assessment. Reviews in Fish Biology and Fisheries and Aquatic Sciences 55:318–327. Fisheries 20:163–186. Ray, W. J., and L. D. Corkum. 1997. Predation of zebra mussels by Round Verreault, G., W. Dargere, and R. Tardif. 2009. American Eel movements, Gobies, Neogobius melanostomus. Environmental Biology of Fishes 50:267– growth, and sex ratio following translocation. Pages 129–136 in J. M. Cas- 273. selman and D. K. Cairns, editors. Eels at the edge: science, status, and con- Ray, W. J., and L. D. Corkum. 2001. Habitat and site affinity of the Round servation concerns. American Fisheries Society, Symposium 58, Bethesda, Goby. Journal of Great Lakes Research 27:329–334. Maryland. Riley, S. C., and A. N. Evans. 2008. Phylogenetic and ecological character- Verreault, G., and P. Dumont. 2003. An estimation of American Eel escapement istics associated with thiaminase activity in Laurentian Great Lakes fishes. from the upper St. Lawrence River and Lake Ontario in 1996 and 1997. Transactions of the American Fisheries Society 137:147–157. Pages 243–251 in D. A. Dixon, editor. Biology, management, and protection Rush, S. A., G. Paterson, T. B. Johnson, K. G. Drouillard, G. D. Haffner, C. E. of catadromous eels. American Fisheries Society, Symposium 33, Bethesda, Hebert,M.T.Arts,D.J.McGoldrick,S.M.Backus,B.F.Lantry,J.R.Lantry, Maryland. T. Schaner, and A. T. Fisk. 2012. Long-term impacts of invasive species on a Vladykov, V. D. 1966. Remarks on the American Eel (Anguilla rostrata native top predator in a large lake system. Freshwater Biology 57:2342–2355. LeSueur): sizes of elvers entering streams; the relative abundance of adult Rutledge, J. E., and L. C. Ying. 1972. Reduction of antithiamine activity in males and females; and present economic importance of eels in North Amer- crayfish by heat treatments. Journal of Food Science 37:497–498. ica. Internationale Vereinigung fur¨ theoretische und angewandte Limnologie Smith, S. H. 1970. Species interactions of the Alewife in the Great Lakes. Verhandlungen 16:1007–1017. Transactions of the American Fisheries Society 99:754–765. Walsh, M. G., D. E. Dittman, and R. O’Gorman. 2007. Occurrence and food

Downloaded by [Department Of Fisheries] at 21:34 27 October 2013 Sprengel, G., and H. Luchtenberg.¨ 1991. Infection by endoparasites reduces habits of the Round Goby in the profundal zone of southwestern Lake Ontario. maximum swimming speed of European Smelt Osmerus eperlanus and Eu- Journal of Great Lakes Research 33:83–92. ropean Eel Anguilla anguilla. Diseases of Aquatic Organisms 11:31–35. Wistbacka, S., and G. Bylund. 2008. Thiaminase activity of Baltic salmon prey Tesch, F. W. 1974. Speed and direction of Silver and Yellow eels, An- species: a comparison of net- and predator-caught samples. Journal of Fish guilla anguilla, released and tracked in the open North Sea. Berichte Biology 72:787–802. This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:35 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 Species and Life History Affect the Utility of Otolith Chemical Composition for Determining Natal Stream of Origin for Pacific Salmon Christian E. Zimmerman a , Heidi K. Swanson b , Eric C. Volk c & Adam J. R. Kent d a U.S. Geological Survey , Alaska Science Center , 4210 University Drive, Anchorage , Alaska , 99508 , USA b Department of Biology , University of Waterloo , 200 University Avenue West, Waterloo , Ontario , NZL 361 , Canada c Alaska Department of Fish and Game , 333 Raspberry Road, Anchorage , Alaska , 99518 , USA d Department of Geosciences , Oregon State University , 104 Wilkinson Hall, Corvallis , Oregon , 97331 , USA Published online: 02 Sep 2013.

To cite this article: Christian E. Zimmerman , Heidi K. Swanson , Eric C. Volk & Adam J. R. Kent (2013) Species and Life History Affect the Utility of Otolith Chemical Composition for Determining Natal Stream of Origin for Pacific Salmon, Transactions of the American Fisheries Society, 142:5, 1370-1380, DOI: 10.1080/00028487.2013.811102 To link to this article: http://dx.doi.org/10.1080/00028487.2013.811102

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1370–1380, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.811102

ARTICLE

Species and Life History Affect the Utility of Otolith Chemical Composition for Determining Natal Stream of Origin for Pacific Salmon

Christian E. Zimmerman* U.S. Geological Survey, Alaska Science Center, 4210 University Drive, Anchorage, Alaska 99508, USA Heidi K. Swanson Department of Biology, University of Waterloo, 200 University Avenue West, Waterloo, Ontario NZL 361, Canada Eric C. Volk Alaska Department of Fish and Game, 333 Raspberry Road, Anchorage, Alaska 99518, USA Adam J. R. Kent Department of Geosciences, Oregon State University, 104 Wilkinson Hall, Corvallis, Oregon 97331, USA

Abstract To test the utility of otolith chemical composition as a tool for determining the natal stream of origin for salmon, we examined water chemistry and otoliths of juvenile and adult Chum Salmon Oncorhynchus keta and Coho Salmon O. kisutch from three watersheds (five rivers) in the Norton Sound region of Alaska. The two species are characterized by different life histories: Coho Salmon rear in freshwater for up to 3 years, whereas Chum Salmon emigrate from freshwater shortly after emergence. We used laser ablation (LA) inductively coupled plasma (ICP) mass spectrometry (MS) to quantify element: Ca ratios for Mg, Mn, Zn, Sr, and Ba, and we used multicollector LA-ICP-MS to determine 87Sr:86Sr ratios in otolith regions corresponding to the period of freshwater residence. Significant differences existed in both water and otolith elemental composition, suggesting that otolith composition could be used to discriminate the natal origin of Coho Salmon and Chum Salmon but only when 87Sr:86Sr ratios were included in the discriminant function analyses. The best discriminant model included 87Sr:86Sr ratios, and without 87Sr:86Sr ratios it was difficult to discriminate among watersheds and rivers. Classification accuracy was 80% for Coho Salmon and 68% for Chum Salmon, indicating that this method does not provide sufficient sensitivity to estimate straying rates of Pacific salmon at the scale we studied. Downloaded by [Department Of Fisheries] at 21:35 27 October 2013

Identification of the natal origins of fish is fundamental to ture reflects a balance between genetic drift within populations understanding population dynamics and population structure and gene flow among populations. Homing of adult spawners (Secor 2010). In the study of population dynamics, natal origin to their natal habitat (i.e., philopatry) results in breeding pop- is the critical piece of information that allows for the matching ulations that are (1) adapted to local conditions and (2) demo- of fishery catch with production (Cadrin and Secor 2009). For graphically and genetically isolated from one another (Hendry geographically structured populations such as Pacific salmon, et al. 2004; Utter et al. 2009). Straying or dispersal decreases the identification of natal origin is needed in studies of straying and variance among local breeding populations (Barton and Whit- homing. Pacific salmon are typically structured as geograph- lock 1997), and the quantification of dispersal capabilities and ically distinct populations in partial genetic isolation. This struc- patterns is a critical step in examining both genetic structure and

*Corresponding author: [email protected] Received January 24, 2013; accepted May 29, 2013 Published online September 2, 2013

1370 NATAL ORIGIN OF PACIFIC SALMON 1371

metapopulation dynamics in animal populations (Wiens 1996; Sr, and Ba), individual fish could be classified to streams with an Ims and Yoccoz 1997). accuracy of 82%. These studies indicate that otolith composition Chemical composition of otoliths has been used to exam- could provide a powerful tool for assessing connectivity among ine natal origin and connectivity among populations of marine populations of anadromous Pacific salmon. fishes (e.g., Thresher 1999; Rooker et al. 2003; Miller et al. Several studies have demonstrated that elemental or isotope 2005). Because otoliths grow throughout the life of the fish composition of otoliths can differ among the natal streams used and are both conservative and metabolically inert, elements or by anadromous salmonids. For example, Veinott and Porter compounds that are incorporated into the calcium carbonate ma- (2005) compared elemental signatures using four elements and trix are permanently retained and thus act as an environmental determined that otolith signatures of Atlantic Salmon Salmo monitor and archive (Campana 1999; Thresher 1999). Compo- salar from three streams in Newfoundland, Canada, differed sition of elements within otoliths is generally determined by sufficiently to permit determination of the natal stream of origin. the composition of ambient water (Campana 1999; Elsdon and Similarly, the utility of 87Sr:86Sr for determining natal stream Gillanders 2003; Wells et al. 2003; Zimmerman 2005), and when has been demonstrated for Chinook Salmon O. tshawytscha in coupled with the chronologically resolved structure of otoliths, the Sacramento River–San Joaquin River basin, California (In- the otolith chemical composition can be used to indicate envi- gram and Weber 1999; Barnett-Johnson et al. 2008), and in ronmental conditions or residency during specific life stages, the Columbia River basin (Barnett-Johnson et al. 2010). Sohn specific years, or both (Campana 1999). et al. (2005) used eight elements to discriminate among Chum Multi-elemental analyses of otoliths have been used to iden- Salmon O. keta from three rivers in Korea and suggested that tify natal origins, habitat associations, and stock structure in multi-elemental analyses of otolith chemical composition could a variety of marine fish species (e.g., Campana et al. 1994; be used to identify the natal origin of Chum Salmon captured Thorrold et al. 2001; Ruttenberg and Warner 2006). Thorrold at sea; however, this has not been investigated for wild Chum et al. (2001) used otolith microchemistry to examine natal hom- Salmon. ing in Weakfish Cynoscion regalis among estuaries on the east To further investigate whether otolith chemical composition coast of the United States. Those authors found that 60–81% of can be used to discriminate among natal streams of origin and spawning Weakfish were spawning in their natal estuary. Such to estimate straying in Pacific salmon species, we examined an approach would be very useful for quantification of straying variability in ambient water chemistry, chemical composition rates and assessment of connectivity among metapopulations of of otoliths in juvenile salmon, and chemical composition of ju- anadromous and freshwater fishes. venile growth zones in otoliths of adult salmon from several Although the use of otolith chemical composition as a tool rivers in the Norton Sound region of western Alaska. Rivers to assess connectivity has not been reported as extensively for and watersheds were selected to represent the greatest possible freshwater fishes as for marine fishes, it has been used in a degree of geologic variability (which should be reflected in wa- variety of contexts. Isotopes of strontium (87Sr and 86Sr) have ter chemistry and otolith composition). These watersheds are been used to examine salmonid movement among tributaries and underlain by differing geology, which is an important prereq- natal origin (stream of origin; Kennedy et al. 1997, 2000; Ingram uisite for otolith microchemical studies (Barnett-Johnson et al. and Weber 1999; Barnett-Johnson et al. 2005). For a variety of 2008; Elsdon et al. 2008). The Nome River watershed (Fig- freshwater and anadromous fishes, multi-elemental signatures ure 1) drains a region containing Precambrian crystalline rocks have been used to determine natal stream of origin (Sohn et al. that were formed between 570 × 106 and 3.6 × 109 years be- 2005; Veinott and Porter 2005; Veinott et al. 2012), connectivity fore present. The Unalakleet River watershed is dominated by or movement among tributaries or lake habitats (Brazner et al. younger Cretaceous rocks that were formed 65 × 106 to 136 × Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 2004; Clarke et al. 2007; Marklevitz et al. 2011), and origin 106 years before present. The Fish River watershed is a mixture of fish that are stocked in or transferred to lakes and streams of young and old Precambrian rocks (between 570 × 106 and (Coghlan et al. 2007; Gibson-Reinemer et al. 2009). Milton and 3.6 × 109 years old) and quaternary deposits that were formed Chenery (2001) used otolith composition and genetic analyses within the last 2 × 106 years. Given this heterogeneity in rock to examine population structure of an anadromous shad, the ages and types, we thought it likely that elemental composition Hilsa Tenualosa ilisha. Using eight elements to compare otolith of stream water would differ among sites and that this would be microchemistry among Hilsa spawning locations, Milton and reflected in the elemental and Sr isotope signatures within the Chenery (2001) were able to distinguish among locations, but freshwater growth region of otoliths. Chum Salmon and Coho they found that movement among locations (straying) was so Salmon O. kisutch were selected as contrasting species to il- high that three distinct spawning populations within the Bay of lustrate how reliance on freshwater as juveniles may affect the Bengal could be treated as a single breeding population or stock. utility of otolith-based methods of discriminating natal origin; Wells et al. (2003) quantified molar ratios of Mg, Mn, Sr, and Ba Chum Salmon typically migrate to sea immediately after emer- to Ca in the first summer growth region of otoliths in Westslope gence (Salo 1991), whereas Coho Salmon rear in freshwater for Cutthroat Trout Oncorhynchus clarkii lewisi from the Coeur up to 3 years (Sandercock 1991). In the Norton Sound region d’Alene River, Idaho. Based on the use of three elements (Mn, of Alaska, Coho Salmon spawning populations typically consist 1372 ZIMMERMAN ET AL.

165° W 160° W

FISH WATERSHED 65° N

NOME WATERSHED Fish Alaska River Niukluk River Nome River 65° N

UNALAKLEET WATERSHED 64° N

North River Unalakleet Norton Sound River

64° N

63° N Chiroskey River N

63° N 0s40 80 160 Kilometer

165° W 160° W

FIGURE 1. Map of the study area in Norton Sound, Alaska.

of three year-classes (ages 1.1, 2.1, and 3.1), and Chum Salmon in freshwater, whereas Chum Salmon had emerged within the spawning populations consist of up to four year-classes (ages preceding weeks. Fish were frozen at the end of each day in 0.2, 0.3, 0.4, and 0.5; Kent et al. 2008). We hypothesized that the field, and otoliths were removed from thawed fish within 1 otolith-based discrimination of the stream of origin is possible month of capture. Adult Chum Salmon and Coho Salmon were for Coho Salmon but that short freshwater residence times and collected in August and September 2005 from either subsistence lingering maternal signals in Chum Salmon would prevent the fisheries or as carcasses found on river margins, and our target adequate accrual of otolith material beyond any maternal influ- sample was 25 adult fish per species from the Nome, Fish, and ences (such as those described by Kalish 1990), thereby hinder- Niukluk rivers. Given the pilot nature of this study and the lack ing accurate discrimination. Finally, we assessed whether the of sampling opportunities in all rivers, we only collected adult Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 addition of 87Sr:86Sr isotope ratios to standard otolith elemental samples from three rivers to test the utility of the method. All composition analyses would improve discriminatory power. otoliths were stored dry in plastic vials for up to 6 months before preparation and analysis. Water chemistry.—Three to five surface water samples METHODS (30 mL) were collected from each river at the same times and Study area.—Juvenile Chum Salmon and Coho Salmon were locations as juvenile fish sampling. Water samples were filtered collected from five rivers in three watersheds (Nome, Fish, and through 0.45-µm membrane filters before being acidified to pH Unalakleet River watersheds) that drain to Norton Sound in less than 2 with quartz-distilled nitric acid. In the laboratory, western Alaska (Figure 1). The hydrology of rivers in this re- samples were diluted from 1 to 6 mL with 1% quartz-distilled gion is dominated by snowmelt and summer rainfall, with peak nitric acid and then were analyzed for Ca, Mg, Mn, Sr, and flows occurring between June and September. Juvenile salmon Ba with a Varian Liberty 150 inductively coupled plasma (ICP) were captured in June 2005 by using baited minnow traps and optical emission spectrometer at the Keck Collaboratory for pole seines at multiple sites within each river. Our target sam- Plasma Spectrometry, Oregon State University. Concentrations ple size was 25 fish of each species in each river. At the time were calculated from emission intensities and the intensities of of sampling, juvenile Coho Salmon had spent at least 1 year standard solutions. Accuracy of the method was verified by NATAL ORIGIN OF PACIFIC SALMON 1373

running a National Institute of Standards and Technology dard. Mean 87Sr:86Sr corresponding to otolith growth deposited (NIST) certified freshwater reference material (NIST 1643c) during the first summer (and outside of the nucleus) was used at the start and end of the analysis session and after every five to characterize each fish. analyses. Data analysis.—Statistical analyses were conducted in R Otolith analysis.—Otolith preparation followed the methods version 2.14.1 (R Development Core Team 2011) and the Sta- described by Zimmerman and Reeves (2002) and Donohoe and tistical Analysis Systems version 9.1.3 (SAS 2003). To assess Zimmerman (2010). One sagittal otolith from each fish was the degree of variation in water chemistry at our study sites, mounted (sulcus side down) with Crystal Bond 509 on a micro- we compared elemental concentrations of Ba, Mg, Mn, and scope cover slip, attached on one edge to a standard microscope Sr among rivers and among watersheds by using multivariate slide. The otolith was ground in the sagittal plane to the level of ANOVA (MANOVA) followed by both multivariate and uni- the nucleus with 2,000-grit sandpaper. The mounting medium variate pairwise contrasts. Analyses were performed on loge- was heated, and the otolith was turned sulcus side up. The otolith transformed data to meet homogeneity of variance and nor- was then ground with 2,000-grit sandpaper in the sagittal plane mality assumptions. For MANOVAs and multivariate pairwise into the nucleus and was polished with a slurry of 0.05-µmalu- contrasts, we report results of Pillai’s trace, which is the most mina paste. The cover slip was then cut with a scribe so that robust test (in comparison with Wilks’ lambda, the Hotelling– several prepared otoliths could be mounted on a single petro- Lawley trace, and Roy’s greatest root) when assumptions are graphic slide for analysis (Donohoe and Zimmerman 2010). not met (Gotelli and Ellison 2004). Univariate pairwise com- Analyses of otolith chemistry were conducted at the Keck parisons were achieved with Tukey’s test. Collaboratory for Plasma Spectrometry. Elemental analy- Analyses of otolith constituents were performed on loge- ses were conducted using a Thermal Elemental PQ Excell transformed data to meet the assumptions of homogeneity of quadropole ICP mass spectrometer connected to a New Wave variance and normality. Chemical composition of otoliths (Sr Research deep ultraviolet (DUV) 193-nm argon fluoride laser. isotope ratios and element: Ca ratios) for the otolith growth Analyses were conducted with a 30-µm-diameter spot size and region corresponding to the first summer (and outside of the a pulse rate of 15 Hz. All samples were taken from a tran- nucleus) was compared among rivers by using MANOVAs and sect that began in the core of the otolith and terminated at the pairwise contrasts. This was done for each species, and data were otolith’s edge. Background levels were measured for 30 s prior transformed as described above. We then used linear discrim- to otolith ablation and were subtracted from the measurements inant function analysis to determine whether multi-elemental obtained during otolith ablation. Count rates for each analyte and Sr isotope signatures could be used to classify fish to the isotope (24Mg, 55Mn, 66Zn, 88Sr, and 138Ba) were normalized watershed or river of origin. Discriminant function models were to 43Ca to account for differences in instrument sensitivity and constructed for both Chum Salmon and Coho Salmon at the river ablation rate (Campana et al. 1997). Each otolith analysis was scale and watershed scale by using the juvenile otolith data. paired with an analytical transect on a polished sample of NIST For juveniles of each species and hierarchical grouping (wa- 612 glass standard to allow for calculations of element concen- tershed versus river), a discriminant function was constructed trations. Mean elemental data corresponding to otolith growth using (1) all elemental and isotope data; (2) only elemental data; deposited during the first summer (and outside of the nucleus) and (3) only Sr:Ca and 87Sr:86Sr ratios. Discriminatory power were used to characterize each fish. was compared among models by using Wilks’ lambda and a Otolith 87Sr:86Sr data were collected on a second otolith tran- cross-validated leave-one-out approach to classify each fish to sect via the methods of Miller and Kent (2009). Multicollector its location of origin (Wells et al. 2000; Gibson-Reinemer et al. (MC) laser ablation (LA) ICP mass spectrometry (MS) instru- 2009). Accuracy of the classifications determined by discrimi- Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 mentation included the New Wave Research DUV 193-nm ex- nant functions was compared with that expected by chance alone cimer laser (see above) and a NuPlasma MC-ICP mass spec- under the assumption that random chance will result in correct trometer. We followed the general method of Woodhead et al. classifications with a percentage that is inversely proportional to (2005) to correct for potential krypton and rubidium (Rb) in- the number of groups classified (White and Ruttenberg 2007). terferences and to monitor calcium argide or dimer formation. Otoliths from adult salmon were classified using the baseline Background interferences by krypton isotopes and contributions discriminant function constructed with the juvenile otolith data, from any other gas species present within the plasma and sweep and classifications were compared with the capture locations. gas supplies were corrected by measuring an on-peak baseline prior to the ablation of otoliths. Measured backgrounds were subtracted from the intensities measured during otolith ablation. RESULTS Mass biases were corrected by reference to an 87Sr:86Sr ratio of 0.1194, and we corrected for isobaric interference of 87Rb on Water Chemistry 87Sr by measuring beam intensity for 85Rb and calculating the Water chemistry (defined here as Mg, Mn, Sr, and Ba con- contribution of 87Rb. A deep-sea gastropod collected from the centrations) varied significantly among watersheds and among Gulf of Mexico was used as an in-house marine carbonate stan- rivers (MANOVAs: F>8, 52 > 6.83, P < 0.0001; Figure 2). 1374 ZIMMERMAN ET AL.

icantly higher in the Unalakleet River than in the North River or Niukluk River (Tukey’s tests: P < 0.05). Overall, it appeared that water chemistry reflected regional variations in geology (as described above), and we therefore expected that analyses of otolith composition among rivers and among watersheds would be informative.

Composition and Discrimination of Otoliths from Juveniles Otolith composition within the first-summer growth region varied significantly among watersheds and among rivers for both juvenile Chum Salmon and juvenile Coho Salmon (MANOVAs: F>5, 112 = 9.21, P < 0.0001; Figure 3). For both species, dif- ferences among watersheds reflected variability in 87Sr:86Sr and Sr:Ca (Tukey’s tests: P < 0.05), whereas for Coho Salmon there was also significant among-watershed variability in Mg:Ca (Tukey’s tests: P < 0.05; Figure 3). Similar to the observations for water chemistry, Chum Salmon in the Nome and Niukluk rivers had similar otolith composition (i.e., with all elements considered together; F5, 108 = 1.30, P = 0.27), but all other pairwise differences were significant (F5, 108 > 4.66, P < 0.0007). In contrast, the compo- sition of otoliths in Coho Salmon differed significantly among all rivers (F5, 110 > 3.03, P = 0.0133). For both species, differ- ences among rivers reflected variability in 87Sr:86Sr and Sr:Ca (Tukey’s tests: P < 0.05; Figure 3). For Coho Salmon, there was additional among-river variability in Mg:Ca and Ba:Ca (Tukey’s tests: P < 0.05; Figure 3). When constructing discriminant functions for analysis among watersheds, we found that the first two discriminant func- tions described 100% of the variation in both Coho Salmon and Chum Salmon for all combinations of analytes: (1) all element: Ca ratios and the 87Sr:86Sr ratio; (2) all element: Ca ratios; and (3) only Sr:Ca and 87Sr:86Sr ratios. For juveniles of both Chum Salmon and Coho Salmon, the full model including all element: Ca ratios and 87Sr:86Sr ratios provided the best discrimination FIGURE 2. Mean ( ± SE) concentrations (ppm) of water constituents in among watersheds, as indicated by the lowest Wilks’ lambda six Alaskan rivers. Letters indicate significant univariate pairwise differences values (Table 1) and by the overall classification rates. For both among rivers, whereas numbers indicate significant univariate pairwise dif- Chum Salmon and Coho Salmon, the discriminant function that

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 < ferences among watersheds (Tukey’s tests: P 0.05). Rivers grouped within was developed by using only Sr:Ca and 87Sr:86Sr ratios was watersheds are indicated by the horizontal lines at the top of the plot (Nome River, Fish River, and Unalakleet River watersheds, respectively). only slightly less successful at discriminating among watersheds (Table 1), whereas discriminant functions that were constructed using only the element: Ca ratios provided the least ability to Univariate pairwise comparisons showed that differences among discriminate among watersheds (Table 1). For Coho Salmon, the watersheds were driven by variability in Ba, Mg, and Sr con- first discriminant function, which was constructed based on all centrations (Tukey’s tests: P < 0.05); there were no significant element: Ca ratios and 87Sr:86Sr ratios, clearly separated fish that differences in Mn (Tukey’s test: P > 0.05; Figure 2). were captured in the Unalakleet River watershed (Chiroskey and When all elements were considered together for among-river North rivers) from fish that were collected in the Fish River and comparisons, the adjacent Nome and Niukluk rivers had similar Nome River watersheds (Figure 4a). Patterns for Chum Salmon water chemistry (F4, 22 = 4.45, P = 0.09), despite being located were similar but less pronounced (Figure 4b). in different watersheds. Pairwise differences between all other At the among-river scale, the first two discriminant func- rivers were significant (F4, 22 > 4.5, P < 0.01; Figure 2). Once tions explained 98.9% of the variation for Coho Salmon and again, these differences were largely driven by variations in Ba, 95.9% of the variation for Chum Salmon (Figure 5a, b). For the Mg, and Sr (Figure 2), although Mn concentrations were signif- discriminant functions using all element and isotope data, the NATAL ORIGIN OF PACIFIC SALMON 1375

0.716 1.4 A Coho Salmon B 0.714 Chum Salmon 1.2

B

0.712 AB ) 1.0 A -3 Sr

86 0.8 0.710 a a a Sr/

87 0.6 Mg/Ca (x10 0.708 c 0.4 b 0.706 C C B 0.2 BC AB AC A 0.704 0.0 Nome Niukluk Fish North Chiroskey Nome Niukluk Fish North Chiroskey

0.6 1.0 C D 0.5 0.8

) 0.4 ) -4 -5 0.6 0.3 0.4 Zn/Ca (x10 Mn/Ca (x 10 0.2 A AB 0.2 AB 0.1 B B

0.0 0.0 Nome Niukluk Fish North Chiroskey Nome Niukluk Fish North Chiroskey

2.0 1.4 E F

1.2 a C 1.5 a a

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 b 1.0 ) ) -5 -3 D 0.8 b b 1.0 ab A ab 0.6 ab A Sr/Ca (x10 Ba/Ca (x10 a AB B B 0.4 0.5 BBAB 0.2

0.0 0.0 Nome Niukluk Fish North Chiroskey Nome Niukluk Fish North Chiroskey

FIGURE 3. Mean ratios ( + 95% confidence interval) of analytes measured in the freshwater growth region of otoliths from juvenile Coho Salmon and Chum Salmon. Capital letters indicate significant univariate pairwise differences between rivers for Coho Salmon, and lowercase letters indicate significant pairwise differences between rivers for Chum Salmon (Tukey’s tests: P < 0.05). 1376 ZIMMERMAN ET AL.

TABLE 1. Overall classification rates and Wilks’ lambda (λ) for discriminant function analyses of otolith chemical composition in juvenile Coho Salmon and Chum Salmon examined at the watershed level (Nome, Fish, and Unalakleet River watersheds, Alaska).

Juvenile Coho Salmon Juvenile Chum Salmon Analysis Classification rate Wilks’ λ Classification rate Wilks’ λ All element: Ca ratios and the 87Sr:86Sr ratio 0.93 0.0510 0.81 0.2008 Element: Ca ratios only 0.83 0.1996 0.65 0.5588 Sr:Ca and 87Sr:86Sr ratios only 0.92 0.0640 0.82 0.2309 Downloaded by [Department Of Fisheries] at 21:35 27 October 2013

FIGURE 4. Bivariate plots of discriminant function scores from the otolith FIGURE 5. Bivariate plots of discriminant function scores from the otolith model constructed based on all elemental and isotope data at the watershed scale model constructed based on all elemental and isotope data at the river scale for for (A) juvenile Coho Salmon and (B) juvenile Chum Salmon (triangles = Nome (A) juvenile Coho Salmon and (B) juvenile Chum Salmon (triangles = Nome River watershed; diamonds = Fish River watershed; squares = Unalakleet River River; solid gray circles = Niukluk River; open circles = Fish River; open watershed). squares = North River; solid gray squares = Chiroskey River). NATAL ORIGIN OF PACIFIC SALMON 1377

TABLE 2. Classification matrix (river scale) based on otolith chemical com- TABLE 3. Classification matrix (watershed scale) based on otolith chemical position in juvenile Coho Salmon and Chum Salmon (overall classification rates composition in adult Coho Salmon and Chum Salmon collected from the Nome for the full data set = 0.80 for Coho Salmon and 0.68 for Chum Salmon). Values and Fish River watersheds (adults were not captured in the Unalakleet River in bold italics represent correct classification to the river of origin. watershed). Valuesin bold italics represent correct classification to the watershed of capture. Actual river of origin Predicted river Watershed of capture of origin Nome Niukluk Fish North Chiroskey Predicted watershed of origin Nome River Fish River Juvenile Coho Salmon Nome 0.81 0.20 0.16 0.00 0.00 Adult Coho Salmon Niukluk 0.07 0.68 0.26 0.00 0.00 Nome River 0.78 0.30 Fish 0.11 0.12 0.58 0.00 0.00 Fish River 0.18 0.68 North 0.00 0.00 0.00 0.87 0.00 Unalakleet River 0.04 0.02 Chiroskey 0.00 0.00 0.00 0.13 1.00 Adult Chum Salmon Juvenile Chum Salmon Nome River 0.45 0.31 Nome 0.72 0.17 0.17 0.00 0.16 Fish River 0.23 0.69 Niukluk 0.08 0.58 0.22 0.00 0.00 Unalakleet River 0.32 0.00 Fish 0.16 0.25 0.57 0.00 0.00 North 0.04 0.00 0.00 0.91 0.20 Chiroskey 0.00 0.00 0.04 0.09 0.64 level, 53% of adult Chum Salmon collected in the Nome River watershed were classified as originating from outside of that overall jack-knifed classification accuracy was 80% for Coho watershed, and 30% of adult Chum Salmon captured in the Fish Salmon and 68% for Chum Salmon. Proportion of misclassi- River watershed were classified as originating from outside of fied fish varied from 9% (North River) to 43% (Fish River) for that watershed (Table 3). When analyzed at the river level, 52% juvenile Coho Salmon and from 0% (Chiroskey River) to 42% of adult Chum Salmon captured in the Nome River were clas- (Fish River) for juvenile Chum Salmon. For Coho Salmon, mis- sified as originating outside of that river, 44% of adult Chum classifications were typically with nearest neighbors (Table 2). Salmon captured in the Niukluk River were classified as orig- That is, Coho Salmon from the Nome, Niukluk, and Fish rivers inating outside of that river, and 33% of adult Chum Salmon were not misclassified as originating from the North River or captured in the Fish River were classified as originating outside Chiroskey River and vice versa (Figure 1; Table 2). This was not of that river (Table 4). the case with Chum Salmon, as there were misclassifications be- tween samples from the two furthest watersheds (Table 2). For example, 4% of Nome River Chum Salmon were misclassified TABLE 4. Classification matrix (river scale) based on otolith chemical com- as North River fish, and 20% of Chiroskey River Chum Salmon position in adult Coho Salmon and Chum Salmon that were collected from the Nome, Niukluk, and Fish rivers. Values in bold italics represent correct were misclassified as Nome River fish (see Figure 1 for location classification to the river of capture. reference; Table 2). River of capture Classification of Otoliths from Adults Predicted river Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 In total, 23, 24, and 26 adult Coho Salmon and 31, 25, and of origin Nome Niukluk Fish 24 adult Chum Salmon were collected from subsistence fish- Adult Coho Salmon eries in the Nome, Niukluk, and Fish rivers, respectively. At Nome 0.78 0.33 0.31 the watershed level, 22% of adult Coho Salmon collected in Niukluk 0.18 0.54 0.31 the Nome River watershed were classified as originating from Fish 0.00 0.13 0.38 outside of that watershed, and 32% of adult Coho Salmon cap- North 0.00 0.00 0.00 tured in the Fish River watershed were classified as originating Chiroskey 0.04 0.00 0.00 from outside of that watershed (Table 3). When analyzed at the Adult Chum Salmon river level, 22% of adult Coho Salmon that were captured in the Nome 0.48 0.40 0.29 Nome River were classified as originating outside of the Nome Niukluk 0.03 0.56 0.67 River, 45% of adult Coho Salmon captured in the Niukluk River Fish 0.13 0.04 0.00 were classified as originating outside of that river, and 61% of North 0.20 0.00 0.00 adult Coho Salmon captured in the Fish River were classified Chiroskey 0.16 0.00 0.04 as originating outside of that river (Table 4). At the watershed 1378 ZIMMERMAN ET AL.

DISCUSSION would rear in freshwater in the wild. This artificially long rear- Results from our study suggest that otolith “tags” may be a ing period would allow for greater accrual of freshwater-derived useful tool for determining the natal river or watershed of origin otolith material. In another western Alaska river (Kuskokwim (i.e., provenance) of Pacific salmon, but several caveats must be River), juvenile Chum Salmon that were collected in the estuary carefully considered before applying this technique (Wells et al. showed no indication of freshwater growth (i.e., there was no 2003; Elsdon et al. 2008). First, the accuracy of discrimination decline in otolith Sr:Ca from a maternal signal to a freshwater will depend on variability in water chemistry among the wa- level), and many fry still had yolk reserves when captured at the tersheds or rivers of interest, and such variability is ultimately river mouth (Hillgruber et al. 2007; C. E. Zimmerman, personal driven by variability in the underlying geology. For this reason, observation). Given the confounding issues of maternal signals we found that the proportion of misclassifications was lower and a short duration of freshwater rearing, we argue that otolith among watersheds than among rivers (because rivers within chemical composition is not a robust means of identifying natal the same watershed have similar underlying geology and wa- stream of origin for wild Chum Salmon. This is simply an issue ter chemistry), and we suggest that underlying geology should related to the life history of the species. We suspect that the be considered and analyses of water chemistry should be con- same issue exists for Pink Salmon O. gorbuscha, which spend ducted before analyzing otolith composition. This is particularly even less time in freshwater and frequently spawn just upstream important in regions of relatively homogeneous geology. of salt water (Heard 1991). Life history traits may also limit the utility of this tool. We Inclusion of 87Sr:86Sr ratios in the analyses greatly increased were able to distinguish among natal rivers for Coho Salmon our ability to discriminate among watersheds and among natal with relatively high confidence because juvenile Coho Salmon rivers. Although facilities with LA-ICP-MS instrumentation are remain in their natal rivers for up to three winters before mi- becoming relatively common, it is less common to find facilities grating to sea. This allows for substantial accrual of freshwater- with the capability of measuring isotope ratios (i.e., MC-LA- derived otolith material, providing an unambiguous freshwater ICP-MS instrumentation). As a result, it would be beneficial if region to sample in the otoliths of adults. A relatively protracted element: Ca ratios alone were capable of facilitating discrim- juvenile rearing period in freshwater allows for the sampling ination among natal rivers for salmonids. Although the use of of otolith freshwater growth regions that are unlikely to be af- element: Ca ratios has been demonstrated to be feasible in some fected by maternal material. Chum Salmon, on the other hand, cases (e.g., Wells et al. 2003; Veinott and Porter 2005; Veinott migrate to sea immediately after emergence from the gravel. Our et al. 2012), our models based only on element: Ca ratios were results (relatively poor discrimination among rivers or among not as robust as the models that included 87Sr:86Sr ratios. This watersheds and a high proportion of misclassifications) indicate indicates that the analytes required for discrimination among that while in freshwater, Chum Salmon do not deposit enough sites likely vary among regions and will differ depending on the maternally independent otolith material to enable use of otolith question at hand. We suggest that pilot studies examining water tags for determination of natal origin. chemistry and otolith elemental and isotopic variability be in- Although Chum Salmon migrate from freshwater immedi- corporated into study designs; this will improve both scientific ately after emergence and the possibility of confounding mater- and economic efficiencies. nal material being present is high (Kalish 1990; Volk et al. 2000; Although we found that otolith composition was sufficiently Zimmerman and Reeves 2002), Arai and Hirata (2006) demon- different among watersheds and rivers to allow classification of strated differences in Mg, Zn, Sr, and Ba between freshwater natal origin, this method was not robust enough to allow for and seawater growth regions in the otoliths of Chum Salmon. estimation of straying rates among rivers in the Norton Sound However, examination of a “typical” profile of Sr, as presented region. Based on classifications of the otoliths from adults, we Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 by Arai and Hirata (2006), suggests that there were in fact estimated that 22–32% of adult Coho Salmon had provenance maternal influences throughout the time period identified as outside of the watershed from which they were collected (i.e., freshwater growth (i.e., elevated Sr at the start of the transect they were strays). These values are relatively high when com- and a gradual decline until the fish moved to seawater). Sohn pared with those of other wild populations. Labelle (1992), for et al. (2005) examined otolith elemental composition in Chum example, reported overall straying rates of approximately 4.7% Salmon juveniles collected from three hatcheries in Korea and for Coho Salmon on Vancouver Island, British Columbia. In found significant differences among sites. Using a discriminant that study, straying rates were typically greater than 2% but function approach similar to the one used in our study, Sohn did range as high as 40% in one case. The misclassification et al. (2005) argued that otolith composition could be used to rates for juvenile Coho Salmon captured in the Nome, Niukluk, identify stocks of Chum Salmon captured in the ocean. The ju- and Fish rivers were similarly high (Table 2), indicating that venile salmon examined by Sohn et al. (2005) ranged in mean our estimates of straying could simply be misclassifications of length from 43 to 82 mm, whereas the lengths of Chum Salmon adults that actually originated from the same streams in which that we examined ranged from 36 to 43 mm. This suggests that they were captured. Although our classification rates for juve- the juvenile Chum Salmon studied by Sohn et al. (2005) were nile salmon (our baseline) are similar to those reported in other held in hatcheries for a longer period of time than Chum Salmon studies (Wells et al. 2000; Brazner et al. 2004; Gibson-Reinemer NATAL ORIGIN OF PACIFIC SALMON 1379

et al. 2009), they may not be sufficiently precise to allow for Barnett-Johnson, R., T. E. Pearson, F. C. Ramos, C. B. Grimes, and the examination of straying. However, this does not preclude R. B. MacFarlane. 2008. Tracking natal origins of salmon using iso- the use of otolith chemical composition for examining stray- topes, otoliths, and landscape geology. Limnology and Oceanography 53: 1633–1642. ing at other locations. For example, Veinott and Porter (2005) Barnett-Johnson, R., F. C. Ramos, C. B. Grimes, and R. B. MacFarlane. 2005. and Veinott et al. (2012) reported classification accuracies of Validation of Sr isotopes in otoliths by laser ablation multicollector induc- 83–100% (with most near 100%) for populations of Atlantic tively coupled plasma mass spectrometry (LA-MC-ICPMS): opening avenues Salmon and Brown Trout Salmo trutta from streams at the same in fisheries science applications. Canadian Journal of Fisheries and Aquatic spatial scales we examined. In our study, adjacent rivers did not Sciences 62:2425–2430. Barnett-Johnson, R., D. J. Teel, and E. Casillas. 2010. Genetic and otolith have sufficient geological differences and resulting water chem- isotopic markers identify salmon populations in the Columbia River at istry differences to allow for a meaningful analysis of straying broad and fine geographic scales. Environmental Biology of Fishes 89: at that spatial scale. 533–546. In summary, for both Coho Salmon and Chum Salmon, Barton, N. H., and M. C. Whitlock. 1997. The evolution of metapopulations. the chemical composition of otoliths was sufficiently different Pages 183–214 in I. A. Hanski and M. E. Gilpin, editors. Metapopula- tion biology: ecology, genetics, and evolution. Academic Press, San Diego, among watersheds to allow for reasonable classification of na- California. tal river at the watershed level within Norton Sound. Patterns in Brazner, J. C., S. E. Campana, and D. K. Tanner. 2004. Habitat fingerprints for geology were not distinct enough to allow for robust discrimina- Lake Superior coastal wetlands derived from elemental analysis of Yellow tion among rivers, however. At the scale of geologic diversity in Perch otoliths. Transactions of the American Fisheries Society 133:692–704. this study, it appears that the inclusion of 87Sr:86Sr ratios is nec- Cadrin, S. X., and D. H. Secor. 2009. Accounting for spatial population structure in stock assessment: past, present, and future. Pages 405–426 in R. J. Beamish essary for discriminating among watersheds and among rivers; and B. J. Rothschild, editors. The future of fisheries science in North America. 87 86 without the Sr: Sr ratios, we would not have been able to Springer-Verlag, Berlin. discriminate among watersheds. Differentiation and classifica- Campana, S. E. 1999. Chemistry and composition of fish otoliths: pathways, tion among rivers were also affected by life history. Our ability mechanisms and applications. Marine Ecology Progress Series 188:263–297. to discriminate the natal origin of Chum Salmon was hindered Campana, S. E., A. J. Fowler, and C. M. Jones. 1994. Otolith elemental fin- gerprinting for stock identification of Atlantic Cod (Gadus morhua)using by the fact that these fish do not rear in freshwater for a suf- laser ablation ICPMS. Canadian Journal of Fisheries and Aquatic Sciences ficient period to develop a strong freshwater otolith signature 51:1942–1950. that is free from maternal influences. As such, it is not possible Campana, S. E., S. R. Thorrold, C. M. Jones, D. Gunther,¨ M. Tubrett, H. Lon- to differentiate Chum Salmon from these rivers. Misclassifica- gerich, S. Jackson, N. M. Halden, J. M. Kalish, P. Piccoli, H. de Pontual, tion proportions from this study should not be used to infer H. Troadec, J. Panfili, D. H. Secor, K. P. Severin, S. H. Sie, R. Thresher, W. J. Teesdale, and J. L. Campbell. 1997. Comparison of accuracy, precision, straying rates without further investigation. First, a multiyear and sensitivity in elemental assays of fish otoliths using the electron micro- study should be conducted to determine the temporal stability probe, proton-induced X-ray emission, and laser ablation inductively coupled of otolith signatures. If there is significant temporal variability, plasma mass spectrometry. Canadian Journal of Fisheries and Aquatic Sci- it would be inadvisable to use juvenile salmon collected in a ences 54:2068–2079. single year as a baseline for adults collected in the same year (as Clarke, A. D., K. H. Telmer, and J. M. Shrimpton. 2007. Elemental analysis of otoliths, fin rays and scales: a comparison of bony structures to provide was done in this study). Genetic analyses should also be used population and life-history information for the Arctic Grayling (Thymallus as a complementary approach to estimate straying. arcticus). Ecology of Freshwater Fish 16:354–361. Coghlan, S. M., Jr., M. S. Lyerly, T. R. Bly, J. S. Williams, D. Bowman, and R. Hannigan. 2007. Otolith chemistry discriminates among hatchery-reared and ACKNOWLEDGMENTS tributary-spawned salmonines in a tailwater system. North American Journal We thank Karen Dunmall, Henry Oyoumick, Jack Koutchak, of Fisheries Management 27:531–541.

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 Jr., Tom Gray, and B. J. Gray for logistical support and assistance Donohoe, C. J., and C. E. Zimmerman. 2010. A method of mounting multiple in the field. We are also grateful to Shiway Wang for preparing otoliths for beam-based microchemical analyses. Environmental Biology of Fishes 89:473–477. the otoliths for analysis and for conducting the initial data re- Elsdon, T. S., and B. M. Gillanders. 2003. Relationship between water and otolith duction and to Andy Ungerer for analyzing the water samples. elemental concentrations in juvenile Black Bream Acanthopagrus butcheri. We appreciate the helpful comments provided by Brian Wells, Marine Ecology Progress Series 260:263–272. Lance Campbell, and an anonymous reviewer. This study was Elsdon, T. S., B. K. Wells, S. E. Campana, B. M. Gillanders, C. M. Jones, K. E. funded by the Arctic–Yukon–Kuskokwim Sustainable Salmon Limburg, D. H. Secor, S. R. Thorrold, and B. D. Walther. 2008. Otolith chem- istry to describe movements and life-history parameters of fishes: hypotheses, Initiative and the U.S. Geological Survey. Use of trade, prod- assumptions, limitations, and inferences. Oceanography and Marine Biology: uct, or firm names is for descriptive purposes only and does not An Annual Review 46:297–330. imply endorsement by the U.S. Government. Gibson-Reinemer, D. K., B. M. Johnson, P. J. Martinez, D. L. Winkelman, A. E. Koenig, and J. D. Woodhead. 2009. Elemental signatures in otoliths of hatch- ery Rainbow Trout (Oncorhynchus mykiss): distinctiveness and utility for REFERENCES detecting origins and movement. Canadian Journal of Fisheries and Aquatic Arai, T., and T. Hirata. 2006. Determination of trace elements in otoliths of Sciences 66:513–524. Chum Salmon Oncorhynchus keta by laser ablation-ICP-mass spectrometry. Gotelli, N. J., and A. M. Ellison. 2004. A primer of ecological statistics. Sinauer, Fisheries Science 72:977–984. Sunderland, Massachusetts. 1380 ZIMMERMAN ET AL.

Heard, W. R. 1991. Life history of Pink Salmon (Oncorhynchus gorbuscha). Salo, E. O. 1991. Life history of Chum Salmon (Oncorhynchus keta). Pages Pages 119–230 in C. Groot and L. Margolis, editors. Pacific salmon life 231–309 in C. Groot and L. Margolis, editors. Pacific salmon life histories. histories. University of British Columbia Press, Vancouver. University of British Columbia Press, Vancouver. Hendry, A. P., V. Castric, M. T. Kinnison, and T. P. Quinn. 2004. The evolution Sandercock, F. K. 1991. Life history of Coho Salmon (Oncorhynchus kisutch). of philopatry and dispersal: homing versus straying in salmonids. Pages 52– Pages 395–445 in C. Groot and L. Margolis, editors. Pacific salmon life 91 in A. P. Hendry and S. C. Stearns, editors. Evolution illuminated: salmon histories. University of British Columbia Press, Vancouver. and their relatives. Oxford University Press, New York. SAS (Statistical Analysis Systems). 2003. SAS/STAT user’s guide, version Hillgruber, N., C. E. Zimmerman, S. E. Burril, and L. J. Haldorson. 2007. Early 9.1.3. SAS Institute, Cary, North Carolina. marine ecology of juvenile Chum Salmon (Oncorhynchus keta) in Kuskok- Secor, D. H. 2010. Is otolith science transformative? new views on fish migra- wim Bay, Alaska. Final Report to North Pacific Research Board, Project tion. Environmental Biology of Fishes 89:209–220. R0327, Anchorage, Alaska. Available: doc.nprb.org/web/03 prjs/r0327 Sohn, D., S. Kang, and S. Kim. 2005. Stock identification of Chum Salmon (On- final.pdf. (January 2013). corhynchus keta) using trace elements in otoliths. Journal of Oceanography Ims, R. A., and N. G. Yoccoz. 1997. Studying transfer processes in metapop- 61:305–312. ulations: emigration, migration, and colonization. Pages 247–266 in I. A. Thorrold, S. R., C. Latkoczy, P. K. Swart, and C. M. Jones. 2001. Natal homing Hanski and M. E. Gilpin, editors. Metapopulation biology: ecology, genetics, in a marine fish metapopulation. Science 291:297–299. and evolution. Academic Press, San Diego, California. Thresher, R. E. 1999. Elemental composition of otoliths as a stock delineator in Ingram, B. L., and P.K. Weber. 1999. Salmon origin in California’s Sacramento– fishes. Fisheries Research 43:165–204. San Joaquin river system as determined by otolith strontium isotopic compo- Utter, F. M., M. V. McPhee, and F. W. Allendorf. 2009. Population genetics and sition. Geology 27:851–854. the management of Arctic-Yukon-Kuskokwim salmon populations. Pages Kalish, J. M. 1990. Use of otolith microchemistry to distinguish the progeny of 97–123 in C. C. Krueger and C. E. Zimmerman, editors. Pacific salmon: sympatric anadromous and non-anadromous salmonids. U.S. National Marine ecology and management of western Alaska’s populations. American Fish- Fisheries Service Fishery Bulletin 88:657–666. eries Society, Symposium 70, Bethesda, Maryland. Kennedy, B. P., J. D. Blum, C. L. Folt, and K. H. Nislow. 2000. Using natural Veinott, G., and R. Porter. 2005. Using otolith microchemistry to distinguish strontium isotopic signatures as fish markers: methodology and application. Atlantic Salmon (Salmo salar) parr from different natal streams. Fisheries Canadian Journal of Fisheries and Aquatic Sciences 57:2280–2292. Research 71:349–355. Kennedy, B. P., C. L. Folt, J. D. Blum, and C. P. Chamberlain. 1997. Natural Veinott, G., P. A. H. Westley, L. Warner, and C. F. Purchase. 2012. Assigning isotope markers in salmon. Nature 387:766–767. origins in a potentially mixed-stock recreational Sea Trout (Salmo trutta) Kent, S., G. Knuepfer, and L. Neff. 2008. Salmonid escapements at fishery. Ecology of Freshwater Fish 21:541–551. Kwiniuk, Niukluk and Nome rivers, 2007. Alaska Department of Fish and Volk, E. C., A. Blakley, S. L. Schroder, and S. M. Kuehner. 2000. Otolith Game, Fishery Data Series 08-57, Anchorage. Available: www.adfg.alaska. chemistry reflects migratory characteristics of Pacific salmonids: using otolith gov/FedAidpdfs/FDS08-57.pdf. (January 2013). core chemistry to distinguish maternal associations with sea and freshwaters. Labelle, M. 1992. Straying patterns of Coho Salmon (Oncorhynchus kisutch) Fisheries Research 46:251–266. stocks from southeast Vancouver Island, British Columbia. Canadian Journal Wells, B. K., B. E. Rieman, J. L. Clayton, D. L. Horan, and C. M. of Fisheries and Aquatic Sciences 49:1843–1855. Jones. 2003. Relationships between water, otolith, and scale chemistries Marklevitz, S. A. C., B. J. Fryer, D. Gonder, Z. Yang, J. Johnson, A. Moerke, and of Westslope Cutthroat Trout from the Coeur d’Alene River, Idaho: Y.E. Morbey. 2011. Use of otolith chemistry to discriminate juvenile Chinook the potential application of hard-part chemistry to describe movements Salmon (Oncorhynchus tshawytscha) from different wild populations and in freshwater. Transactions of the American Fisheries Society 132:409– hatcheries in Lake Huron. Journal of Great Lakes Research 37:698–706. 424. Miller, J. A., M. A. Banks, D. Gomez-Uchida, and A. L. Shanks. 2005. A Wells, B. K., S. R. Thorrold, and C. M. Jones. 2000. Geographic variation in comparison of population structure in Black Rockfish (Sebastes melanops) trace element composition of juvenile Weakfish scales. Transactions of the as determined with otolith microchemistry and microsatellite DNA. Canadian American Fisheries Society 129:889–900. Journal of Fisheries and Aquatic Sciences 62:2189–2198. White, J. W., and B. I. Ruttenberg. 2007. Discriminant function analysis in ma- Miller, J. A., and A. J. R. Kent. 2009. The determination of maternal run time in rine ecology: some oversights and their solutions. Marine Ecology Progress juvenile Chinook Salmon (Oncorhynchus tshawytscha) based on Sr/Ca and Series 329:301–305. 87Sr/86Sr within otolith cores. Fisheries Research 95:373–378. Wiens, J. A. 1996. Wildlife in patchy environments: metapopulations, mosaics, Milton, D. A., and S. R. Chenery. 2001. Can otolith chemistry detect the pop- and management. Pages 53–84 in D. R. McCullough, editor. Metapopulations

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 ulation structure of the shad Hilsa Tenualosa ilisha? comparison with the and wildlife conservation. Island Press, Washington, D.C. results of genetic and morphological studies. Marine Ecology Progress Se- Woodhead, J., S. Swearer, J. Hergt, and R. Maas. 2005. In situ Sr-isotope ries 222:239–251. analysis of carbonates by LA-MC-ICP-MS: interference corrections, high R Development Core Team. 2011. R: a language and environment for statisti- spatial resolution and an example from otolith studies. Journal of Analytical cal computing. R Foundation for Statistical Computing, Vienna. Available: Atomic Spectrometry 20:22–27. www.R-project.org./. (January 2013). Zimmerman, C. E. 2005. Relationship of otolith strontium-to-calcium ratios and Rooker, J. R., D. H. Secor, V. S. Zdanowicz, G. De Metrio, and L. O. Relini. salinity: experimental validation for juvenile salmonids. Canadian Journal of 2003. Identification of Atlantic Bluefin Tuna (Thunnus thynnus) stocks from Fisheries and Aquatic Sciences 62:88–97. putative nurseries using otolith chemistry. Fisheries Oceanography 12:75–84. Zimmerman, C. E., and G. H. Reeves. 2002. Identification of steelhead and Ruttenberg, B. I., and R. R. Warner. 2006. Spatial variation in the chemical resident Rainbow Trout progeny in the Deschutes River, Oregon, revealed composition of natal otoliths from a reef fish in the Galapagos´ Islands. Marine with otolith microchemistry. Transactions of the American Fisheries Society Ecology Progress Series 328:225–236. 131:986–993. This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:35 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 Anadromous Sea Lampreys Recolonize a Maine Coastal River Tributary after Dam Removal Robert Hogg a , Stephen M. Coghlan Jr. a & Joseph Zydlewski b a Department of Wildlife Ecology , University of Maine , 5575 Nutting Hall, Orono , Maine , 04469 , USA b U.S. Geological Survey, Maine Cooperative Fish and Wildlife Research Unit , University of Maine , 5575 Nutting Hall, Orono , Maine , 04469 , USA Published online: 02 Sep 2013.

To cite this article: Robert Hogg , Stephen M. Coghlan Jr. & Joseph Zydlewski (2013) Anadromous Sea Lampreys Recolonize a Maine Coastal River Tributary after Dam Removal, Transactions of the American Fisheries Society, 142:5, 1381-1394, DOI: 10.1080/00028487.2013.811103 To link to this article: http://dx.doi.org/10.1080/00028487.2013.811103

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1381–1394, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.811103

ARTICLE

Anadromous Sea Lampreys Recolonize a Maine Coastal River Tributary after Dam Removal

Robert Hogg* and Stephen M. Coghlan Jr. Department of Wildlife Ecology, University of Maine, 5575 Nutting Hall, Orono, Maine 04469, USA Joseph Zydlewski U.S. Geological Survey, Maine Cooperative Fish and Wildlife Research Unit, University of Maine, 5575 Nutting Hall, Orono, Maine 04469, USA

Abstract Sedgeunkedunk Stream, a third-order tributary to the Penobscot River, Maine, historically supported several anadromous fishes, including the Atlantic Salmon Salmo salar, Alewife Alosa pseudoharengus, and Sea Lamprey Petromyzon marinus. However, two small dams constructed in the 1800s reduced or eliminated spawning runs entirely. In 2009, efforts to restore marine–freshwater connectivity in the system culminated with removal of the lowermost dam, thus providing access to an additional 4.6 km of lotic habitat. Because Sea Lampreys utilized accessible habitat prior to dam removal, they were chosen as a focal species with which to quantify recolonization. During spawning runs of 2008–2011 (before and after dam removal), individuals were marked with PIT tags and their activity was tracked with daily recapture surveys. Open-population mark–recapture models indicated a fourfold increase in the annual abundance of spawning-phase Sea Lampreys, with estimates rising from 59 ± 4(N ± SE) before dam removal (2008) to 223 ± 18 and 242 ± 16 after dam removal (2010 and 2011, respectively). Accompanying the marked increase in annual abundance was a greater than fourfold increase in nesting sites: the number of nests increased from 31 in 2008 to 128 and 131 in 2010 and 2011, respectively. During the initial recolonization event (i.e., in 2010), Sea Lampreys took 6 d to move past the former dam site and 9 d to expand into the furthest upstream reaches. Conversely, during the 2011 spawning run, Sea Lampreys took only 3 d to penetrate into the upstream reaches, thus suggesting a potential positive feedback in which larval recruitment into the system may have attracted adult spawners via conspecific pheromone cues. Although more research is needed to verify the migratory pheromone hypothesis, our study clearly demonstrates that small-stream dam removal in coastal river systems has the potential to enhance recovery of declining anadromous fish populations. Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 Dams are ubiquitous throughout the world, providing hydro- the impounded sediment downstream (Hart et al. 2002; Gardner electric power generation, flood control, municipal water sup- et al. 2013). plies, and recreational opportunities. Historically, dams were The Penobscot River is Maine’s largest river, and the wa- built without forethought of their ecological impacts, and some tershed once supported as many as 11 co-evolved diadromous early dams have outlived their utility. Dams constructed without species (Saunders et al. 2006). However, 113 dams throughout fish passage systems have blocked anadromous fish migrations the watershed have severed marine–freshwater connectivity and and are a leading cause of fish declines in Maine and around have led to declines in all sea-run fishes (PRRT 2012). Efforts the world (Limburg and Waldman 2009). Dams contribute to to restore marine–freshwater connectivity along the Penobscot declines in the biodiversity and productivity of stream system River are underway, with main-stem dam removal projects an- fauna (Freeman et al. 2003), and dam removal may provide ticipated to occur from June 2012 to November 2013 (PRRT rapid (<1 year) ecosystem responses if high-water events move 2012).

*Corresponding author: [email protected] Received March 2, 2013; accepted May 23, 2013 Published online September 2, 2013 1381 1382 HOGG ET AL.

only anadromous species known to consistently spawn in Sedge- unkedunk Stream prior to dam removal and now serves as a focal species for evaluating the short-term efficacy of restoration efforts. Anadromous Sea Lampreys begin their life history in fresh- water streams and rivers, where fertilized eggs settle into gravel and cobble nests. Embryos incubate for 3–8 d before the lar- vae (ammocoetes) emerge, drift downstream, settle in silty sub- strate, and filter feed for as many as 8 years (Beamish 1980). After this prolonged period of larval filter feeding, ammocoetes undergo a suite of behavioral, physiological, and morphological changes as they prepare to leave the freshwater environment. This transformation is likely triggered by maturation to a mini- mum body length of 120 mm, body mass of 3 g, and condition factor of 1.5, in combination with the accumulation of sufficient lipid reserves to ensure survival during the 10–11-month non- trophic metamorphosis period (Jones 2007). These transformers (or macrophthalmia) develop large eyes, an oral disk, and salt- water tolerance as they exit freshwater and become parasitic in the open ocean. Sea Lampreys are parasitic in the Atlantic FIGURE 1. Locations of Sedgeunkedunk Stream (Penobscot County, Maine), Ocean for 2–3 years, after which they cease feeding and migrate Fields Pond, barriers that were removed as part of the Sedgeunkedunk Stream back into freshwater rivers to spawn (Beamish 1980). Restoration Project (e.g., former Mill Dam), and natural landmarks that were Anadromous lampreys select spawning streams by cueing identified as potential barriers to Sea Lamprey range expansion. on temperature and flow (Andrade et al. 2007; Keefer et al. 2009; Binder et al. 2010), but chemical compounds released by Sedgeunkedunk Stream, a small tributary to the Penobscot the ammocoetes are extremely influential (Wagner et al. 2009; River below head of tide, typifies the small streams in Maine Vrieze et al. 2010). The observed reliance upon chemical and that have been impacted by dams (Figure 1). Recent restoration environmental cues in selection of spawning habitat—instead of efforts in Sedgeunkedunk Stream have provided opportunities philopatry as exhibited by Atlantic Salmon (Hansen and Quinn to assess fish community responses to dam removal, and the 1998) or Alewives (Jessop 1994)—suggests that Sea Lampreys system provides ideal conditions for predicting the recovery may recolonize newly accessible habitat more rapidly than the of other upstream tributaries that are influenced by main-stem other historically cohabitating anadromous species of Sedge- Penobscot River dam removals (Gardner et al. 2012, 2013). unkedunk Stream. The rapid expansion of Sea Lampreys into Sedgeunkedunk Stream is one of only three major tributaries the upper Laurentian Great Lakes during the 1930s after the flowing into the Penobscot River downstream of the lowermost construction of navigation channels further demonstrates the main-stem dam (i.e., Veazie Dam), which is slated for removal species’ ability to exhibit rapid colonization (Smith and Tibbles during 2013–2014 (PRRT 2012). Therefore, recovery of anadro- 1980). mous species in Sedgeunkedunk Stream may provide a glimpse Because Sea Lampreys in the upper Great Lakes parasitize of predicted restoration outcomes on the main-stem Penobscot valuable sport fishes, research in North America has largely Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 River. been driven by mitigating negative impacts upon recreational Efforts to restore marine–freshwater connectivity in Sedge- and commercial fisheries (Christie and Goddard 2003). How- unkedunk Stream culminated in August 2009 with the removal ever, within their native range, anadromous Sea Lampreys are of the lowermost dam, Mill Dam, at stream kilometer 0.7 a focus of concern due to decreasing runs. Declines and local (Figure 1), allowing access to an additional 4.7 km of high- extirpations of Sea Lampreys in Europe have been documented quality spawning and rearing habitat for Sea Lampreys Petromy- (Renaud 1997), and the species has received conservation at- zon marinus and federally endangered Atlantic Salmon Salmo tention on both sides of the Atlantic Ocean (Maitland 2003; salar. Additionally, the removal of Mill Dam, coupled with CRASC 2011). Additionally, in recognition of the unique eco- construction of a rock–ramp fishway that bypassed the remnants logical functions that anadromous species may perform, current of the former Meadow Dam, provided a corridor for migrating restoration efforts have shifted away from single-species ap- Alewives Alosa pseudoharengus to access lentic spawning habi- proaches to more-community-based and ecosystem-based ap- tat in Fields Pond (Figure 1). Previous studies within Sedgeunke- proaches. Operating under this community-based paradigm, re- dunk Stream indicated the annual occurrence of Sea Lamprey source managers have recognized that Sea Lampreys may be spawning runs, which were limited to the lower 700 m of stream an ecologically important constituent of stream ecosystems and below Mill Dam (Gardner et al. 2012). The Sea Lamprey was the that the Sea Lamprey’s recovery may therefore be critical to TRIBUTARY RECOLONIZATION BY SEA LAMPREYS 1383

restoration of native anadromous fish assemblages in Maine Fishway (44◦4405N, 68◦4556W) and flows 5.3 km down- (Saunders et al. 2006). stream to the confluence of the Penobscot River near head of Sea Lampreys are semelparous and die within days after tide at river kilometer (rkm) 36.5 (44◦4608N, 68◦4706W). spawning (Beamish 1980). Postspawning mortality typically The lower 90-m reach of Sedgeunkedunk Stream experiences occurs during periods of declining discharge and increasing tidal fluctuations due to its proximity with the Penobscot River summer temperatures, thereby translating into rapid carcass de- head of tide. The Sedgeunkedunk Stream watershed drains composition. Therefore, Sea Lamprey carcasses may provide approximately 5,400 ha and includes several ponds in the pulses of marine-derived nutrients in otherwise oligotrophic headwaters. The watershed is mostly forested, but some urban headwater streams at a favorable time to support instream pro- and industrial development exists, primarily in downstream duction (Nislow and Kynard 2009; Guyette 2012). Sea Lamprey reaches. The relatively low-gradient stream has a median spawners use their suctorial disk mouths to rearrange gravel and bank-full width of approximately 5 m, with a peak discharge of cobble substrate during nest construction. Essentially, they ex- 5m3/s immediately after early spring ice-out and a base flow cavate rocks from the tails of pools and deposit them slightly discharge of 0.1 m3/s during late summer. The lowermost dam downstream to form pit-and-mound nest structures. Pairs of (Mill Dam; 44◦4555N, 68◦4647W) was located 700 m up- male and female individuals spawn from a remnant “anchor stream of the Penobscot River confluence and 610 m upstream rock” in the pit, where they vibrate vigorously against one an- of head of tide. Although the Meadow Dam Fishway provides other and release gametes. Finally, the fertilized eggs settle marine–freshwater connectivity between the Atlantic Ocean downstream; although only approximately 15% of these eggs and Fields Pond, access through the fishway is inconsequential are ultimately deposited in the mound, a high proportion of the for Sea Lampreys because their spawning requirements limit mound-deposited eggs (85–90%) survive to hatch (Smith and them to lotic habitats. Therefore, this study was focused on the Marsden 2009). Spawning-related activities detach fine sedi- 5.2-km reach of lotic habitat from the Meadow Dam Fishway ments from coarser substrates (Kircheis 2004), and these modi- downstream to the Sedgeunkedunk Stream head of tide. How- fications to streambed topography may reduce substrate armor- ever, we note that a 4-m-high natural waterfall (Tannery Falls) ing and embeddedness, similar to the effects of redd-building located at rkm 4.8 may be a substantial barrier to Sea Lamprey Pacific salmon Oncorhynchus spp. (Montgomery et al. 1996). migration, especially during low-flow years (Figure 1). Hence, nest construction and spawning activities by Sea Lam- preys may “condition” the spawning habitat for Atlantic Salmon METHODS (Kircheis 2004; Saunders et al. 2006), provide prey in the form of displaced eggs and dislodged benthic invertebrates (Scott Mark–Recapture Surveys and Crossman 1985), and potentially create physical structure Sea Lamprey capture and tagging.—Our methods were sim- for drift-feeding fishes. ilar to those of Gardner et al. (2012). As migrating Sea Lam- The present study expands on previous research conducted preys entered Sedgeunkedunk Stream, they were captured with prior to dam removal (Gardner et al. 2012) and serves to quantify an Indiana-style trap net (fyke net) anchored 90 m upstream the efficacy of dam removal as a restoration tool. The primary from the confluence with the Penobscot River. The 2.5-m-long focus of this study was the hypothesized expansion of Sea Lam- fyke net was constructed of 3-mm square mesh, with a 1.3- × preys into previously inaccessible habitat of Sedgeunkedunk 1.6-m (height × width) mouth and a 1-m-diameter cod end. Stream. Our project goal was to compare and contrast the abun- The trap was centered longitudinally in the 0.8-m-deep thal- dance, distribution, and behavior of spawning Sea Lampreys weg ( ± 0.2 m, dependent on tidal cycle and discharge), and the before and after dam removal. Specifically, our objectives were wings of the trap spanned the entire 4.5-m width of the stream Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 to (1) provide annual estimates of spawning-phase Sea Lam- ( ± 0.5 m, dependent on tidal cycle and discharge). We deployed preys by using mark–recapture data; (2) quantify and compare the fyke net from 15 May to 26 June 2010 and from 22 May to 6 the distributions and abundances of nesting sites before and after July 2011. Two submersible light-emitting diode (LED) lamps dam removal; (3) characterize attributes, behaviors, and move- were sewn into the entrances of the fyke net during 2011 to ment patterns of spawning-phase Sea Lampreys in response to increase trap efficiency (Purvis et al. 1985). dam removal; and (4) describe annual patterns in timing of the Upon capture, each Sea Lamprey received two tags. A full- Sea Lamprey spawning run as related to stream temperature and duplex (12- × 2-mm) PIT tag was implanted within the dorsal discharge. musculature via a hypodermic injector, and an externally visible t-bar anchor tag (uniquely coded) was inserted into the dorsal musculature on the opposite side to assess PIT tag retention STUDY AREA on future dates. We recorded the mass, length, and sex of each Sedgeunkedunk Stream is a third-order tributary to the Sea Lamprey before release. Fully mature Sea Lampreys ex- Penobscot River (Penobscot County, Maine) and flows through hibit sexual dimorphism, and males are accurately identified by the town of Orrington and the city of Brewer (Figure 1). the presence of a thickened dorsal ridge, or “rope” (Hardisty Sedgeunkedunk Stream drains Fields Pond at the Meadow Dam and Potter 1971). However, this dorsal characteristic may not 1384 HOGG ET AL.

be fully developed in early arriving males. Therefore, to verify during 2010 and 2011. The bridge location provided ideal sex, we used a suite of primary and secondary sexual character- conditions, with relatively uniform depth across a fixed stream istics, including the gentle expression of gametes, female post width of 4.2 m. We used a propeller-driven current velocity anal fin development, and presence of the male genital papilla meter (Swoffer Model 2100) and a U.S. Geological Survey or “penis,” in addition to the dorsal rope (Percy et al. 1975). (USGS) top-set wading rod to gauge the stream at minimum If we lacked confidence in sex determination based on all of 1-week intervals. Individual gauging measurements were these characteristics, we recorded the sex of the individual as regressed to average daily water levels to estimate a continuous unconfirmed. Out-of-water processing generally took less than daily discharge record. A third-order polynomial (R2 = 0.998, 40 s, and individuals were allowed to re-acclimate in live wells P < 0.001) was used to build an average daily discharge for at least 15 min prior to their release back into the stream. No curve for 2010. Because the standpipe was damaged by ice adverse effects were witnessed after the tagging process, and scour in 2011 and was subsequently replaced, a second-order recently tagged fish were often observed building nests within polynomial (R2 = 0.984, P < 0.001) was used for that year. hours of tagging. Spawning surveys.—We conducted daily surveys on foot to Data Analysis track the activity of tagged individuals and to identify Sea Lam- Data set.—Unless otherwise stated, we incorporated prey nests along the entire reach of stream from the fyke net archived 2008 pre-dam-removal data from Gardner et al. (2012) to the Meadow Dam Fishway. Foot surveys were performed by to perform direct post-dam-removal comparisons. We note that two crews: one crew worked upstream from the fyke net, and due to flood conditions throughout the month of June in 2009, no the other worked downstream from the fishway. Surveys gen- Sea Lamprey spawning activity was observed during the 2009 erally began shortly after dawn and no later than 0700 hours. pre-dam-removal season (Gardner et al. 2012). Therefore, we Surveys were completed by 1800 hours. We captured nontagged report no data from 2009. All means are reported with SEs, and individuals with dip nets or by hand and processed them as de- statistical tests were conducted using the Statistical Analysis scribed previously. A portable PIT tag antenna coupled with a Systems version 9.2 (SAS 2010) at the significance level α of battery-powered reader was used to identify previously tagged 0.05 unless otherwise noted. individuals without repeated handling (Hill et al. 2006), thus Abundance estimates.—We estimated abundance of spawn- minimizing the disruption of spawning activity (Gardner et al. ing Sea Lampreys in Sedgeunkedunk Stream for 1 year be- 2012). Upon each Sea Lamprey encounter, we recorded the in- fore dam removal (2008) and 2 years after dam removal (2010 dividual’s identity (unique tag code), tag retention, condition and 2011) by using a Jolly–Seber population analysis (POPAN) (live or dead; carcass recoveries were recorded as “losses on model developed for open populations (Arnason and Schwarz capture”), behavior, nest attendance, and location. 1999) in program MARK (White and Burnham 1999). We Nest surveys.—We marked each nest location with a coded recorded carcasses as losses on capture (Schwarz et al. 1993), stake driven into the streambank and recorded Universal Trans- and after enumeration, carcasses were removed from the analy- verse Mercator (UTM) coordinates with a handheld GPS de- ses. The POPAN model is appropriate for estimating the abun- vice (eTrex Legend H; Garmin, Inc.). All UTM waypoints were dance of spawning Sea Lampreys in Sedgeunkedunk Stream ground-truthed at a later date and were found to be within 5 m because the following assumptions were likely met: (1) animals of the respective nest locations. Many Sea Lampreys exhibit retained their tags throughout the duration of the studies; (2) photophobic, nocturnal behavior and abandon nests in favor of tags were read properly; (3) sampling was consistent with daily sheltered areas during daylight hours (Kelso and Gardner 2000). encounter histories (sampling was not instantaneous, but model Furthermore, male lampreys typically initiate nest construction, developers claim that departures of less than 2–3 d are small Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 but a male will often abandon a particular nest if he is not joined enough to avoid violation; Schwarz et al. 1993); (4) the study promptly by a receptive female (Manion and Hanson 1980). area was held constant; and (5) constant trap and survey efforts Therefore, nest identifications were based on obvious substrate provided equal catchability between marked and unmarked an- disturbances in addition to direct observations of spawning ac- imals at each sampling occasion (Pledger and Efford 1998). We tivity. used Akaike’s information criterion corrected for small sample sizes (AICc) to evaluate and select the best candidate models for Spawning Run Timing: Temperature and Discharge each spawning run (Burnham and Anderson 2002). Candidate We adopted the remote stream gauging methodology models included the following as parameters: the probability of of Lundquist et al. (2005) and deployed a combination capture ( pcap), probability of apparent survival (), and prob- pressure and temperature sensor (Solinst Levelogger Junior; ability of entering the study system ( pent). Parameters were www.solinst.com), which was encased in a protective polyvinyl set to vary at daily time steps or to remain constant, but be- chloride standpipe anchored to a concrete bridge in Sedgeunke- cause models incorporating time-dependent capture parameters dunk Stream at rkm 0.6. We programmed the levelogger to are inherently unable to estimate abundance at the first or last record temperature and water level continuously at 1-h intervals sampling occasion (Schwarz et al. 1993), we limited pcap to a from May to November (the expected onset of winter icing) constant value in all model iterations. TRIBUTARY RECOLONIZATION BY SEA LAMPREYS 1385

Nesting site distributions.—In the context of newly avail- measurement tool in ArcGIS version 9.3 (ESRI, Inc., Redlands, able habitat, we expected a relatively slow rate of Sea Lamprey California). In total, we observed 57% (2008: n = 27), 48% range expansion due to their purported reliance upon conspe- (2010: n = 63), and 49% (2011: n = 76) of tagged Sea Lam- cific chemical cues (ammocoete pheromones) for selection of preys after the initial capture date; although most of the tagged spawning habitat. We anticipated comparatively fewer nests up- individuals were subsequently detected only once, some were stream of the former Mill Dam during the first year of recol- detected as many as six times after the tagging date. We ob- onization (2010), but we expected to observe an increase in served a small percentage of same-day repeat detections (<5%) upstream nesting sites during 2011 after larval recruitment from and found that most of those individuals fell back relatively the prior year. To test these hypotheses, we conducted chi-square short distances downstream (range = 12–242 m). For consis- (χ2) goodness-of-fit tests for 2010 and 2011; the null hypothe- tency, we removed all same-day fallback distances and used ses stated that nesting site selection would be proportionately only the furthest upstream daily detection in analyses. Because equivalent to the habitat available in the historically accessible initial capture locations were so variable, we limited statisti- reach downstream of the former Mill Dam and in the newly cal analysis of movement patterns to detections of maximum accessible upstream reach. upstream distances (Max rkm) for individuals that were en- We used the following rationale for setting up the χ2 anal- countered at least two times. A two-factor ANOVA on ranked yses. Sea Lampreys accessed approximately 3,000 m2 (610-m Max rkm data incorporating the year × gender interaction was length × 5-m median width) of lotic habitat prior to the August used to explore differences in movement patterns. We did not 2009 removal of Mill Dam. Restoration efforts provided an include the 2008 Max rkm data in statistical analysis because additional 23,000 m2 (4.6 km of stream length) of available lotic the presence of Mill Dam limited the potential for Sea Lamprey habitat for Sea Lampreys, but 2010 drought conditions confined movements to downstream reaches below rkm 0.7 (Gardner Sea Lampreys to habitat below a beaver dam located near rkm et al. 2012). However, we do report gender-specific 2008 me- 4.0 (Figure 1). Therefore, Sea Lampreys accessed 17,000 m2 dians and ranges of Max rkm for pre- and post-dam-removal (3.4 km of stream length), and the return to a more typical flow comparisons. Furthermore, we simply plotted gender-specific regime during 2011 resulted in an additional 1,000 m2 of habitat point measurements of upstream and downstream movements ending at Tannery Falls (Figure 1). We note that although the for individuals that were detected more than once to elucidate distance between the beaver dam and Tannery Falls is approx- the gender-specific movement patterns that occurred after dam imately 800 m in stream length, the beaver dam impoundment removal. renders approximately 600 m of stream unsuitable for spawning. Spawning run timing: temperature and discharge.—To de- Sea Lamprey capture, biological measures, and behavior.— tect interannual stream temperature variation, we used a one- We surmised that individual Sea Lampreys that were captured way ANOVA with year as the factor and average daily temper- in the fyke net would be heavier than those that were tagged ature during the Sea Lamprey spawning period as the response further upstream. Additionally, we anticipated that the 2011 ad- variable. Additionally, a Student’s t-test assuming unequal vari- dition of LED lights at the fyke net entrance would increase ances was employed to detect interannual variation in stream the trap efficiency. Therefore, we employed two-way ANOVA discharge. The use of Student’s t-test was appropriate because models incorporating the year × trap interaction as a factor to in- we only estimated discharge during the 2010 and 2011 spawning vestigate differences in size distributions of males and females runs. separately. Furthermore, we questioned whether the sex ratio We used generalized least-squares (GLS) regression models would be skewed toward exploratory males during recoloniza- to explore how run timing was related to both temperature and tion after dam removal, so we used χ2 goodness-of-fit tests to discharge. Daily counts of initial Sea Lamprey captures in the Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 determine whether there were differences in gender distribu- study system were used as the response variable. Mean daily tions. If recolonization was driven by the exploratory behavior discharge, change in discharge from the previous day, mean of males, we would expect to see more male-initiated instances daily temperature, and change in temperature from the previ- of nest construction. Therefore, we organized active nest ob- ous day were used as predictor variables. Generalized least- servation data into categories of single, paired, or communal squares modeling offers an alternative to ordinary linear regres- nesting behaviors. For the 2010 and 2011 spawning seasons, we sion by accounting for correlative, non-independent residuals used χ2 goodness-of-fit tests to compare the observed gender of (Trepanier´ et al. 1996) and has been used for comparable time a single Sea Lamprey against an expected equal probability of series migration data (Anderson and Quinn 2007). We followed the individual being male or female. Additionally, χ2 tests were the GLS modeling protocol of Anderson and Quinn (2007) by used to compare the observed genders of paired Sea Lampreys specifying the error structure as a first-order autoregressive pro- against expected equal probabilities that pairs were engaged in cess, and we utilized maximum likelihood techniques for pa- either male–female courtship or same-sex nest construction. rameter estimation (R Development Core Team 2010). Reported To quantify individual movements, successive detections GLS P-values are from t-tests of each environmental predictor were organized chronologically and minimum pathway dis- variable, and reported R2 values were calculated by comparing tances traveled between detections were estimated with the the log-likelihood of each fitted model to the log-likelihood 1386 HOGG ET AL.

TABLE 1. Total number of Sea Lampreys captured, number captured in the fyke-net trap (percentage of annual captures is shown in parentheses), count of nests identified, number of males (M), number of females (F), number of individuals with unconfirmed gender (U), observed gender ratio (M:F), run duration, mean average daily temperature, and mean average daily discharge during annual spawning runs before dam removal (2008) and after dam removal (2010 and 2011) in Sedgeunkedunk Stream, Maine. Means are presented with SEs; variables that are significantly different at α = 0.05 are in bold italics. No Sea Lampreys were observed in the system during 2009 due to flood conditions.

Captures Year Total Trap Nests M F U M:F Days Temperature (◦C) Discharge (m3/s) 2008 47 16 (34%) 31 26 21 – 1.24 10 19.3 ± 0.6 – 2009 0 0 (0%) – – – – – – – – 2010 131 39 (30%) 128 72 50 9 1.44 24 19.0 ± 0.4 0.21 ± 0.02 2011 156 72 (46%) 131 86 67 3 1.28 20 19.9 ± 0.4 0.28 ± 0.02

of the null (intercept-only) model (Nagelkerke 1991). Sea nual run size abundances based on approximately threefold in- Lamprey spawning runs in Maine are extremely abbreviated creases in the number of individuals tagged during both of the and usually last between 3 and 6 weeks (Kircheis 2004). There- years after dam removal (Figure 2). fore, to pinpoint discharge and temperature influences on daily counts of immigrating spawners, we limited our analyses to a Nesting Site Distributions period extending from 1 week before the initial detection to the Evidence of Sea Lamprey nesting was not observed upstream date of the final detection for both the 2010 and 2011 spawning of the former Mill Dam until the sixth day of the 2010 spawning runs (n = 31 d in 2010; n = 33 d in 2011). run, when we observed a previously tagged male engaged in solitary nest construction less than 100 m beyond the remnant structure. Sea Lampreys penetrated further upstream at a pace RESULTS of approximately 400 m/d until exhibiting a burst of activity on the ninth day of the 2010 spawning run. On that day, we marked Sea Lamprey Capture and Abundance Estimates multiple nests in newly colonized habitat, including a nest that We observed a considerable increase in the number of Sea was located slightly downstream of a beaver dam near rkm 4 Lampreys captured and the duration of spawning runs after the (Figure 1). This nest marked the extent of Sea Lamprey range removal of Mill Dam (Table 1). Post-dam-removal captures were expansion during the 2010 spawning run (Figure 3). Whereas 2.8 times greater (in 2010) and 3.3 times greater (in 2011) than Sea Lampreys took 9 d to access a 4-km extent of linear stream the 2008 pre-dam-removal captures (Table 1). Spawning run habitat during the 2010 spawning run, they expanded their range durations in Sedgeunkedunk Stream more than doubled after 800 m further to Tannery Falls at rkm 4.8 (Figure 3) in just 3 d the removal of Mill Dam (Table 1). We note that a female Sea during the 2011 spawning run. Lamprey was captured in the trap net on 10 June 2011, but it was Post-dam-removal abundances of Sea Lamprey nesting sites a victim of snapping turtle Chelydra serpentina predation upon increased over 400% relative to the 2008 pre-dam-removal capture (fyke-net bycatch). Therefore, we considered the 2011 count. Nest counts rose from 31 in 2008 to 128 and 131 in spawning run duration to be a 20-d period because no subsequent 2010 and 2011, respectively (Table 1). Sea Lamprey nesting evidence of spawning was observed in the system until 16 June sites were predominately located in historically accessible habi- Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 2011. The 2011 trap efficiency increased by 12% and 16% in tat downstream of the former Mill Dam during both 2010 and comparison with 2008 and 2010, respectively (Table 1). The 2011. Forty-eight (38%) of the 128 nests observed during 2010 number of males was greater than the number of females in and 39 (30%) of the 131 nests observed during 2011 occurred all 3 years, but 2010 was the only spawning run in which a in approximately 15% of the available habitat during both years statistically significant gender bias was observed ( p = 0.046; (Figure 3). The χ2 analyses revealed that a Sea Lamprey nest Table 1). was more likely to be observed downstream of the former Mill The POPAN models incorporating a constant probability of Dam than in newly accessible upstream reaches ( p < 0.001 for capture {pcap(.)}, a constant probability of apparent survival both years). {(.)}, and a time-dependent probability of entering the system {pent(t)} consistently had the most support among the candi- Sea Lamprey Biological Measures and Behavior date models based on AICc and model weighting (wi) scores There were no discernible differences in Sea Lamprey (Table 2). The {pcap(.), (.), pent(t)} models estimated annual length between genders or among years, but body mass was spawning run sizes of 59 ± 4 (mean ± SE) in 2008, 223 ± 18 comparatively lighter during the 2010 spawning run than during in 2010, and 242 ± 16 in 2011 (Figure 2). The {pcap(.), (.), the other years (Table 3). The two-way ANOVA for body mass pent(t)} models reasonably estimated a fourfold increase in an- of males indicated a difference among years ( p = 0.047), TRIBUTARY RECOLONIZATION BY SEA LAMPREYS 1387

TABLE 2. Akaike’s information criterion corrected for small sample sizes (AICc), Akaike difference (AICc; difference in AICc value between the ith model and the best model), Akaike weight (wi; the relative probability that the ith model is the best model), number of parameters (Ki), and abundance estimates (N ± SE) for candidate models used to estimate annual Sea Lamprey spawning run sizes in Sedgeunkedunk Stream. Candidate models are defined by three distinct probabilities: the probability of capture ( pcap), probability of apparent survival (), and probability of entering the study system ( pent), where individual probabilities are set either to vary by daily time step (t) or to remain constant throughout the annual study period (.). The pcap(.) and (.) parameter values (± SE) are given for the best-fit models. No Sea Lampreys were observed in the system during 2009 due to flood conditions.  Model AICc AICc wi Ki N ± SE pcap(.) (.) 2008 (before dam removal) pcap(.), (.), pent(t) 296.82 0.00 0.99 12 59 ± 40.63± 0.06 0.80 ± 0.04 pcap(.), (t), pent(t) 306.85 10.03 0.01 21 56 ± 4 pcap(.), (t), pent(.) 7,013.71 6,716.89 0.00 11 361 ± 111 pcap(.), (.), pent(.) 7,022.70 6,725.88 0.00 3 144 ± 20 2010 (after dam removal) pcap(.), (.), pent(t) 676.84 0.00 1.00 21 223 ± 18 0.30 ± 0.03 0.79 ± 0.03 pcap(.), (t), pent(t) 703.98 27.14 0.00 43 226 ± 21 pcap(.), (t), pent(.) 12,031.93 11,355.09 0.00 20 2,722 pcap(.), (.), pent(.) 17,586.68 16,909.84 0.00 2 3,081 ± 438 2011 (after dam removal) pcap(.), (.), pent(t) 654.00 0.00 0.98 17 242 ± 16 0.41 ± 0.04 0.76 ± 0.03 pcap(.), (t), pent(t) 662.15 7.92 0.02 37 326 ± 153 pcap(.), (.), pent(.) 3,471.69 2,817.46 0.00 2 4,414 ± 261 pcap(.), (t), pent(.) Non-estimable

but no effects were detected at the trap level ( p = 0.512) or ( p = 0.951). Post hoc comparisons revealed that a gender- for the interaction between year and trap ( p = 0.825). Post specific difference in mass of females existed only between the hoc comparisons revealed that a gender-specific difference in 2010 and 2011 spawning runs ( p = 0.039; Table 3). the mass of males existed only between the 2008 and 2010 We observed active Sea Lamprey nests on 84 and 90 separate spawning runs ( p = 0.05; Table 3). Likewise, the two-way occasions during 2010 and 2011, respectively. It was common ANOVA for body mass of female Sea Lampreys indicated a to observe Sea Lampreys engaging in communal nesting behav- difference among years ( p = 0.04), but no effects were detected ior. Communal nest attendance ranged between three and eight atthetraplevel(p = 0.323) or for the trap × year interaction individuals, with a maximum of three males or seven females at a given nest. Of the active nests that were observed during 2010 and 2011, 25% (n = 22) and 29% (n = 26), respectively, were communal (Figure 4). The χ2 analyses for the 2010 season 250 Number Estimated revealed that if only a single Sea Lamprey was observed on a Number Tagged nest, that individual was more likely to be a male ( p = 0.008); if we observed a pair, they were more likely to be engaged in

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 200 strictly male–female courtship ( p = 0.003; Figure 4). However, χ2 analyses for the 2011 season revealed that if only one Sea 150 Lamprey was observed on a nest, it was just as likely to be a male as to be a female; if a pair was observed, this did not necessarily 100 translate into strict male–female courtship (Figure 4). Female Sea Lampreys that were detected at least once 50

Sea lamprey spawning run (N) Sea lamprey spawning subsequent to the initial tagging procedure penetrated further NA upstream than their male counterparts during both post- dam-removal years, as indicated by median Max rkm values 2008 2009 2010 2011 ( p = 0.025; Table 4). As mentioned previously, no evidence of spawning activity beyond remnants of the former Mill Dam FIGURE 2. Sea Lamprey spawning run size (± SE; gray bars) per year in (rkm 0.610) was observed during the first several days of the Sedgeunkedunk Stream, as estimated from mark–recapture encounter histories 2010 recolonization event, and exploratory plots of all detected of tagged individuals (black diamonds). Dashed vertical line represents the August 2009 removal of Mill Dam. Sea Lampreys were not observed in the upstream and downstream movements aid in illustrating this system during the 2009 spawning season. phenomenon (Figure 5). Upstream movements increased 1388 HOGG ET AL.

(Figure 5). These patterns coincide with the statistical difference 20 2008 Dam Site observed between the Max rkm values for males and females. 16 BD Falls Spawning Run Timing: Temperature and Discharge 12 Spawning run timing was variable among years and did not appear to be influenced by the presence or absence of Mill Dam. 8 Annual spawning runs began as early as 1 June in 2010 and as late as 18 June in 2008. Spawning run duration, however, did ap- 4 pear to be affected by the dam and was at least 10 d longer in both post-dam-removal years (Table 1). There was no discernible dif- ference in average daily water temperature during pre- and post- 20 dam-removal spawning runs (Table 1). However, average daily 2010 discharge was comparatively greater during the 2011 spawning 16 run than during the 2010 run ( p = 0.024; Table 1). Although GLS models revealed no significant relationships linking daily 12 Sea Lamprey counts with temperature, change in temperature, discharge, or change in discharge during the 2010 and 2011 8 spawning runs ( p > 0.05), the 2011 discharge model indicated that a descending limb in the hydrograph had marginal explana- 2 = = 4 tory power (R 0.27, P 0.055) in describing the arrival of Sea Lampreys into Sedgeunkedunk Stream during the 2011 spawning run. Sea lamprey nest abundance 20 2011 DISCUSSION 16 Abundance Estimates 12 The primary objective of this study was to document abun- dance patterns of spawning-phase Sea Lampreys as they re- 8 sponded to the August 2009 removal of Mill Dam in Sedge- unkedunk Stream. The spring of 2010 was the first opportunity 4 in over a century for migrating adult Sea Lampreys to access historic spawning habitat beyond the former Mill Dam. Sea Lampreys responded rapidly to this opportunity by recoloniz- 0.1 0.5 0.9 1.3 1.7 2.1 2.5 2.9 3.3 3.7 4.1 4.5 ing approximately 3.3 km of newly accessible habitat during Kilometers upstream of trap 2010 and expanding their range by an additional 0.8 km in 2011 (Figure 3). In correspondence with an approximate sixfold in- FIGURE 3. Distributions of Sea Lamprey nesting sites in Sedgeunkedunk crease in available habitat, our most well-supported POPAN Stream, as observed before dam removal (2008: n = 31) and after dam removal (2010: n = 128; 2011: n = 131). Dashed vertical lines demark the former models estimated a nearly fourfold increase in the abundance Mill Dam (Dam Site), the beaver dam (BD), and Tannery Falls (Falls). No of migrating adult Sea Lampreys: from 59 fish in 2008 before Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 Sea Lampreys were observed in the system during the 2009 (pre-dam-removal) dam removal to 223 fish in 2010 and 242 fish in 2011 after dam spawning season due to flood conditions. removal (Figure 2). Our POPAN estimates of annual Sedgeunkedunk Stream spawners were biologically plausible given the seasonally pre- rapidly after 7 June 2010; interestingly, females appeared to dictable and semelparous nature of Sea Lamprey spawning move upstream at a faster rate than males (Figure 5). Down- events. First, the constant pcap parameters were plausible given stream movements were not prevalently detected in the system that our survey protocols were extremely consistent within and until after nesting was observed in the upstream reaches directly among years; traps were set in advance of all spawning runs, and below the beaver dam (located near rkm 4) on 9 June 2010. Early daily foot surveys were conducted at regular hours in advance movements during the 2011 spawning run did not appear to fol- of and throughout the entire duration of each spawning run. low the pattern observed during the previous year. Instead, there Although there was considerable interannual variation among were multiple movements greater than 0.6 km/d detected within the pcap parameters (Table 2), this variation was expected given the first week of the 2011 spawning run (Figure 5). Additionally, the increased amount of survey habitat after dam removal and the most of the rapid upstream movements during the second increased trap efficiency during 2011. The pcap was greatest in week of the 2011 spawning run were detected from females 2008 (0.63; Table 2), when surveys were limited to 0.610 km of TRIBUTARY RECOLONIZATION BY SEA LAMPREYS 1389

TABLE 3. Mean body length (mm) and mass (g) of Sea Lampreys that were tagged in Sedgeunkedunk Stream during annual spawning runs in 2008–2011. Means are presented with SE (n is given in parentheses); asterisks indicate variables that were significantly different from one another at α = 0.05. No Sea Lampreys were observed in the system during 2009 due to flood conditions.

Trap-captured fish Upstream-captured fish All captured fish Year Length (mm) Mass (g) Length (mm) Mass (g) Length (mm) Mass (g) Males 2008 607 ± 21 (10) 750 ± 60 (8) 631 ± 12 (16) 740 ± 60 (11) 622 ± 11 (26) 740 ± 40 (19)* 2010 634 ± 9 (27) 640 ± 60 (27) 622 ± 6 (44) 590 ± 30 (44) 627 ± 5 (71) 610 ± 30 (71)* 2011 636 ± 7 (43) 670 ± 30 (43) 632 ± 7 (39) 650 ± 20 (39) 634 ± 5 (82) 660 ± 20 (82) Pooled 632 ± 5 (80) 670 ± 30 (78) 627 ± 4 (99) 630 ± 20 (94) 629 ± 3 (179) 650 ± 20 (172) Females 2008 600 ± 19 (5) 700 ± 70 (4) 614 ± 16 (13) 680 ± 60 (8) 610 ± 12 (18) 680 ± 50 (12) 2010 602 ± 13 (12) 630 ± 50 (12) 610 ± 7 (37) 590 ± 30 (37) 608 ± 6 (49) 600 ± 20 (49)* 2011 636 ± 8 (27) 720 ± 30 (27) 618 ± 7 (38) 660 ± 20 (38) 625 ± 6 (65) 690 ± 20 (65)* Pooled 622 ± 7 (44) 700 ± 30 (43) 614 ± 5 (88) 630 ± 20 (83) 617 ± 4 (132) 650 ± 20 (126)

accessible habitat, but pcap declined precipitously in 2010 (0.30; arrival to the spawning grounds, and the observed similarity be- Table 2), when available habitat increased by nearly 600% and tween interannual values (range = 0.76–0.80) reflected this trap efficiency was at its lowest level (30%). Probability of cap- phenomenon (Table 2). Finally, the time-dependent pent param- ture rose to an intermediate level in 2011 ( pcap = 0.41; Table 2), eters were plausible given the observed daily variation in the and this increase may be explained by the greater trap effi- number of individuals tagged within and among years. ciency (46%). Secondly, the constant parameters were plau- Historical Sea Lamprey abundance data for the Penobscot sible given the semelparous nature of spawning Sea Lampreys; River and its tributaries are lacking (Kircheis 2004; Saunders spawning-related mortality typically occurs within a week upon et al. 2006), so we do not know whether the observed fourfold increase in spawning Sea Lampreys in Sedgeunkedunk Stream after dam removal represents a return to historic levels. Sea Lamprey spawning runs in Lake Ontario tributaries with com- parable discharge (mean annual discharge < 1.0 m3/s) were estimated over the course of 8 years (1997–2004) to range be- tween 207 ± 30 individuals (N ±SE) in a 1.22-km reach of Port Britain Creek and 798 ± 98 individuals in a 0.55-km reach of

TABLE 4. Gender-specific medians (range in parentheses) of observed max- imum upstream movements (Max river kilometer [rkm]) by tagged Sea Lam- preys that were detected in Sedgeunkedunk Stream at least once after the initial tagging procedure. Sea Lampreys tagged during 2008 were excluded from sta-

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 tistical analysis because the presence of Mill Dam limited their movements to the lower 0.610 km of stream. Asterisks indicate gender-specific median values that were significantly different from one another at α = 0.05.

Year Max rkm detected Males 2008 (n = 14) 0.341 (0.146–0.529) 2010 (n = 34) 0.875 (0.000–3.865) 2011 (n = 39) 0.407 (0.000–3.737) 2010–2011 (n = 73) 0.621 (0.000–3.865)* Females FIGURE 4. Abundances of active Sea Lamprey nests observed during the 2010 = and 2011 annual spawning runs in Sedgeunkedunk Stream. Nests are categorized 2008 (n 13) 0.381 (0.146–0.529) by the gender and number of individuals in attendance (M = males; F = females; 2010 (n = 28) 2.232 (0.056–3.690) Single = solitary individual; Paired = two individuals; Communal = three or 2011 (n = 36) 2.049 (0.000–3.725) more individuals). Asterisks indicate distributions that are significantly different 2010–2011 (n = 64) 2.050 (0.000–3.725)* from 1:1 at α = 0.05. 1390 HOGG ET AL.

3 per nest ratios. Annual number of Sea Lampreys per nest dis- 2010 played minimal variation, ranging from a high of 1.9 in 2008 to a 2 low of 1.7 in 2010, with a median value of 1.8 calculated for the 2011 data. The consistent relationship between raw nest counts 1 and POPAN estimates (R2 = 0.998; P = 0.001) suggests that nest enumeration may provide a proxy for abundance estimates 0 when mark–recapture studies are cost prohibitive. Sea Lampreys in Sedgeunkedunk Stream continued to pre- -1 dominately select nesting locations downstream of the former Mill Dam in both years after dam removal. However, the 2011 -2 Females nesting site distribution trended towards a more equitable lon- Males gitudinal distribution of nesting sites. Nest abundances down- -3 stream of the former Mill Dam declined from 48 nests (38%) in 3 2010 to 39 nests (30%) in 2011, and 20 (15%) of the nests ob- 2011b) 2011 served in 2011 were in the lowermost 200 m of stream. With the 2 exception of those 20 nests, the 2011 longitudinal distribution of Sea Lamprey nesting sites was more evenly dispersed than 1 the 2010 distribution (Figure 3).

Minimum Distance Detected (km/day) 0 Sea Lamprey Capture, Biological Measures, and Behavior Our fyke-net trap was not as effective as we had anticipated, -1 but the addition of waterproof LED lights sewn into the entrance during 2011 did improve trap performance. Our study was not -2 designed with the intention of statistically examining the effec- tiveness of illuminated traps, but we did desire to improve trap -3 efficiency as a means of intercepting Sea Lampreys as they en-

n n tered the system, thereby reducing the disturbance of spawning -Jun -Jun activities. Additionally, a standardized tagging location could 1-May 7-Ju 1-Ju 05-Jul 3 0 14 2 28 have potentially improved our ability to investigate movement Midpoint Date of Detected Movement patterns. For these purposes, we report limited success in Sea Lamprey capture by use of illuminated trap entrances. FIGURE 5. Detected minimum pathway distances of tagged Sea Lampreys in Variation in trap efficiency may explain the observed differ- Sedgeunkedunk Stream during the 2010 (upper panel) and 2011 (lower panel) ences in 2010 gender-specific body masses compared with the spawning runs. Point measurements were standardized to kilometers per day by using the midpoint day of successive observations; positive values represent other 2 years. Although trap efficiency was comparable in 2008 upstream movements, and negative values represent downstream movements. and 2010 (34% and 30%, respectively), the body mass of males in 2010 was lower than that in 2008, and this discrepancy may Shelter Valley Creek (Binder et al. 2010). Port Britain Creek, have resulted from the striking difference in total captures be- which has a mean annual discharge of 0.5 m3/s (Binder et al. tween the 2 years. Masses were recorded upon initial capture, 2010), appears to be somewhat comparable to Sedgeunkedunk which sometimes occurred far upstream of the trap (maximum Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 Stream. Given that Port Britain Creek regularly hosts spawning distance from trap = 4 km) for 45 males in 2010, whereas in Sea Lamprey densities of approximately 0.03 fish/m2, one may 2008 only 13 males were initially captured at a maximum dis- anticipate comparable densities in Sedgeunkedunk Stream after tance of just 0.6 km. The analysis had low discriminatory power the system experiences multiple year-classes of ammocoete re- at the trap level (10%) and for the trap × year interaction (8%). cruitment. Therefore, an annual spawning run of over 500 Sea Therefore, the disparity in male body mass between 2008 and Lampreys may be realistic for the 18,000 m2 of lotic habitat that 2010 may simply be a reflection of the small sample size in 2008 are presently available in Sedgeunkedunk Stream. combined with variable capture locations in both years. Accompanying the overall rise in trap efficiency between Nesting Site Distributions 2010 and 2011 (30% and 46%, respectively), the number of ripe The observed number of Sea Lamprey nests increased more female Sea Lampreys that were initially captured in the trap than fourfold during the 2 years after the removal of Mill Dam, more than doubled: from only 12 females in 2010 to 28 females essentially mirroring the nearly fourfold increase in our POPAN in 2011. Females that were initially captured upstream of the abundance estimates. An interesting pattern was uncovered af- trap were usually intercepted after bouts of spawning and likely ter combining the raw nest count data with the corresponding experienced some degree of decreased body mass due to the POPAN estimates in the development of annual Sea Lamprey release of gametes and due to starvation. The absolute increase TRIBUTARY RECOLONIZATION BY SEA LAMPREYS 1391

in captures of prespawning females from the trap may explain gender-specific difference provides additional support. In why the body mass of females was greater in 2011 than in 2010, describing the reproductive life histories of anadromous Pacific but again the analysis had low discriminatory power at the trap salmon populations, Morbey (2000) defined protandry as “the level (17%) and for the trap × year interaction (6%). earlier arrival of males to the spawning grounds than females.” Sea Lamprey range expansion into previously inaccessible Morbey (2000) argued that protandry is a valuable reproductive habitat during the 2010 spawning run appeared to be driven strategy for male salmon because they are semelparous, and by the exploratory behavior of males. Male Sea Lampreys are intraspecific competition for access to spawning females is sensitive to a larval migratory pheromone that serves as a con- fierce due to the semelparous life history. Sea Lampreys share specific cue, drawing migrants toward tributaries with habitats many mating system attributes with Pacific salmon, so it that are adequate for offspring rearing (Wagner et al. 2009). follows that protandry may be an equally valuable strategy for In turn, females are sensitive to the male mating pheromone, a them as well. Our data suggest that Sea Lampreys exhibited bile acid compound released by spermiating males that attracts protandry during the 2010 recolonization event and that the females toward the vicinity of potential mates (Siefkes et al. phenomenon of protandry provides a parsimonious explanation 2005). This conspecific pheromone communication system for the male-biased sex ratio, a statistical preponderance of may explain why Sea Lampreys took 6 d to move just 0.65 km solitary males at nesting sites, and a relatively slow progression past the former dam site in 2010. The lack of ammocoetes of movement into previously unoccupied habitats. and associated larval migratory pheromone signals from newly accessible upstream reaches likely provided little motivation Spawning Run Timing: Temperature and Discharge for spawning males to venture into the previously inaccessible Our intensive monitoring of Sea Lamprey spawning runs habitat. However, males display antagonistic behavior during in Sedgeunkedunk Stream revealed that in all years studied, the establishment of nesting territory (Manion and Hanson spawning-phase migrants arrived at least 2–4 weeks later in this 1980); therefore, range expansion may have resulted from brief stream than in most streams of the lower Penobscot River water- exploratory searches for vacant spawning habitats. shed (O. Cox, Maine Department of Marine Resources, Bangor, Whereas it took Sea Lampreys 6 d to expand their range be- personal communication). Additionally, the 2009 spawning sea- yond the former Mill Dam and an additional 3 d to penetrate son appeared anomalous, as Gardner et al. (2012) were unable the furthest upstream reaches during 2010, activity extended to detect Sea Lampreys entering Sedgeunkedunk Stream at all, throughout the system up to the Tannery Falls boundary in only even though Sea Lampreys were found in neighboring tribu- 3 d during 2011. Perhaps a cohort of 1-year-old ammocoetes that taries. The lack of detections in 2009 could be attributable to were spawned during the 2010 run settled into rearing habitats the extreme precipitation events throughout the month of June, upstream of the former Mill Dam, subsequently releasing larval when unusually high discharge in the lower portion of Sedge- pheromones that cued the 2011 adult migrants immediately to unkedunk Stream may have inhibited spawning activities. Re- the furthest upstream reaches. Prior lines of evidence from Sea gardless, the initiation of spawning activity in Sedgeunkedunk Lamprey studies in the Great Lakes suggested that adult spawn- Stream was extremely variable among years, occurring as early ing runs were extremely responsive to ammocoete populations. as 1 June during 2010 and as late as 18 June during 2008. Moore and Schleen (1980) reported that the removal of ammo- Sea Lamprey spawning activities in Maine typically occur coetes from a stream reduced the number of spawning adults during late May and early June, when mean daily water temper- in subsequent migrations. Additionally, Sorensen and Vrieze atures range between 17◦C and 19◦C (Kircheis 2004). However, (2003) found that streams with relatively large ammocoete pop- our data show that in all 3 years, Sedgeunkedunk Stream tem- ulations attracted larger adult spawning runs than neighboring peratures exceeded this range for periods of days to weeks prior Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 streams with smaller larval populations. Therefore, the increased to the arrival of spawning-phase Sea Lampreys. Mean daily tem- activity observed in the upstream reaches of Sedgeunkedunk peratures were 19◦C or greater during all three spawning runs Stream during the early stages of the 2011 spawning run may (Table 1), thus suggesting additional or alternative environmen- have resulted from the prior year’s establishment of ammocoete tal cues to migration in Sedgeunkedunk Stream. Perhaps the recruits and the subsequent release of larval conspecific chem- observed difference in stream discharge between the 2010 and ical cues. Although we lack ammocoete data with which to 2011 spawning runs (Table 1) can partially explain some of the confirm the larval migratory pheromone hypothesis for Sedge- variability in run timing. unkedunk Stream, the subtle differences between the 2010 and Binder et al. (2010) found significant stream-dependent dif- 2011 nesting site distributions provide indirect support. ferences in the relative importance of environmental variables The male-biased sex ratio and the observed prevalence of as predictors of Sea Lamprey spawning runs in six Lake Ontario active nests occupied by single males during the 2010 spawning tributaries. Although water temperature was the best predictor run further support the contention that range expansion was among all six streams, water level—a surrogate measure for driven by the exploratory behavior of males. We also detected stream discharge—was an equally reliable explanatory variable a prevalence of shorter maximum upstream movements (Max but only in the two smallest streams, Port Britain Creek and rkm) among males in comparison with females, and this subtle Shelter Valley Creek (Binder et al. 2010). As alluded to earlier, 1392 HOGG ET AL.

these two streams compare well with Sedgeunkedunk Stream, semelparous life history of the Sea Lamprey translates into and results from our GLS modeling exercises support the find- consistent delivery of marine-derived nutrients at a crucial time ings of Binder et al. (2010) regarding the importance of water of the year when many aquatic organisms are at the peak of level in relation to Sea Lamprey migratory activity. Our GLS re- their growing seasons. Additionally, the nest-building activities sults, although not significant at an α level of 0.05, indicated that of Sea Lampreys have the potential to condition habitat that stream discharge during 2011 had some explanatory power in has been degraded by decades of increased sedimentation. The describing the arrival of spawning migrants to Sedgeunkedunk literature is replete with evidence suggesting that redd-digging Stream. Pacific salmon improve the quality of riverine habitats by Perhaps temperature and discharge must reach a com- sweeping fine sediments downstream, coarsening the stream bination of threshold levels before Sea Lampreys enter the bed, and reducing cobble embeddedness (Montgomery et al. spawning grounds. Close inspection of Sedgeunkedunk Stream 1996). Sea Lamprey nest construction may produce similar hydrographs in relation to daily Sea Lamprey counts offers effects in coastal New England systems, and ongoing research a simplistic explanation regarding the observed variation in in Sedgeunkedunk Stream is currently addressing these ques- timing of the annual spawning runs. Mean daily temperatures tions. Very little is understood regarding the community-level throughout the peak of the 2010 spawning run (1–14 June) were effects of recurring Sea Lamprey spawning disturbances, and within the 17–19◦C range reported for Maine streams (Kircheis this study provided us with the impetus to explore their role 2004), while discharge was consistently below 0.45 m3/s as streambed “conditioners.” The daily spawning surveys during the same period. In contrast, the 2011 hydrograph was conducted during this study allowed us to catalog exact nesting vastly different. The 2011 mean daily discharge was above locations and to return to those sites periodically to measure 1.0 m3/s until midway through the first week of June and did microhabitat characteristics that are influenced by Sea Lamprey not decline below 0.45 m3/s until 12 June. Consequently, the spawning disturbances. The synergistic interactions of multiple peak of the 2011 spawning run (21 June–1 July) was delayed co-evolved diadromous and freshwater fishes may be a nec- in comparison with 2010, and mean daily temperatures were essary ingredient for the recovery of high-functioning aquatic consistently above 20◦C during that period. The complete ecosystems throughout Maine and northern New England absence of spawning Sea Lampreys in Sedgeunkedunk Stream (Saunders et al. 2006), and this study provides a platform during the flood of 2009 and the relative delay in 2011 run with which to begin addressing the potentially overlooked role timing are not surprising given that high-discharge events have played by anadromous Sea Lampreys within their native range. inhibited migratory activity in other anadromous (Masters et al. 2006) and potamodromous lampreys (Malmqvist 1980). Additionally, because Sedgeunkedunk Stream converges ACKNOWLEDGMENTS with the Penobscot River near head of tide, late-arriving We would like to thank the collaborators and those who as- migrants may display phenotypic plasticity in their migratory sisted with this project. Field technicians from the University of behavior as a response to a relatively short (36.5-km) upstream Maine included Megan Patridge, Gabe Vachon,Meghan Nelson, migration distance. Quinn and Adams (1996) reasoned that Mary Banker, Ryan Haley, Morgan Burke, Jake Kwapiszeski, anadromous fishes that spawn shortly after entering freshwater Andy O’Malley, Phill Adams, Jake Poirier, Dylan Wingfield, are more likely than long-distance migrants to have evolved Chelsea Wagner, Ethan Lamb, and Trevor Violette. Volunteers adaptations in response to fluctuating temperature because fish and alternate field help included Cory Gardner, Silas Ratten, that migrate short distances likely experience the same condi- Wes Ashe, Ed Hughes, Ann Grote, Ian Kiraly, Margaret Guyette, tions as developing larvae. Whereas spawning Sea Lampreys Adam Derkacz, Sarah Drahovzal, Dan Stich, Doug Sigourney, Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 enter the Fort River (tributary to the Connecticut River at rkm Quentin Tuckett, Rena Carey, John Wood, Anthony McLaugh- 159) consistently earlier in the season (Nislow and Kynard lin, Cody Kent, Jesse Hogg, Barbara Shrewsberry, Greg Innes, 2009), late migrants to Sedgeunkedunk Stream may be display- Derek Trunfio, Kira Fizell, Ben Emmott, Ana Rapp, Rich May, ing an adaptation that favors arrival at the spawning grounds Chuck Attean, Imre Kormendy, Margo Relford, and Marius consistent with temperatures that are optimal (18.4◦C; range = Mutel. The Town of Orrington, City of Brewer, Bob’s Kozy 15.5–21.1◦C) for embryonic development (Smith et al. 1968). Korner, The Brookside Grill, Grave’s Dryland Marine, Rick Violette, and Bob Fennell provided logistical help and access Conclusions to study sites. We also thank Rory Saunders (National Oceanic Our study has clearly demonstrated that restorative dam and Atmospheric Administration [NOAA]), Joshua Royte (The removal projects have the potential to enhance recovery of Nature Conservancy), and Dan Hayes (Michigan State Uni- declining anadromous fish populations by providing access to versity) for their assistance. This paper benefited from help- habitats that are necessary for the completion of migratory life ful reviews by Kevin Simon, Theodore Castro-Santos, and two histories. The Sea Lampreys’ rapid response to dam removal in anonymous reviewers. This work was supported in part by an Sedgeunkedunk Stream may produce a multitude of beneficial award from Maine Sea Grant. The views expressed herein are effects by providing sorely missed ecological services. The those of the authors and do not necessarily reflect the views of TRIBUTARY RECOLONIZATION BY SEA LAMPREYS 1393

Maine Sea Grant. This work was also supported in part by the At- Hill, M. S., G. B. Zydlewski, J. D. Zydlewski, and J. M. Gasvoda. 2006. Develop- lantic Salmon Federation, Penobscot Valley Audubon Chapter, ment and evaluation of portable PIT tag detection units: PITpacks. Fisheries NOAA, University of Maine, and the USGS Maine Coopera- Research 77:102–109. Jessop, B. M. 1994. Homing of Alewives (Alosa pseudoharengus) and Blueback tive Fish and Wildlife Research Unit. Sampling was conducted Herring (A. aestivalis) to and within the Saint John River, New Brunswick, under Institutional Animal Care and Use Committee Protocol as indicated by tagging data. Canadian Technical Report of Fisheries and Number A2011-06-03. Mention of trade names does not imply Aquatic Sciences 2015. endorsement by the U.S. Government. Jones, M. L. 2007. Toward improved assessment of Sea Lamprey population dynamics in support of cost-effective Sea Lamprey management. Journal of Great Lakes Research 33(Supplement 2):35–47. Keefer, M. L., M. L. Moser, C. T. Boggs, W. R. Daigle, and C. A. Peery. REFERENCES 2009. Variability in migration timing of adult Pacific Lamprey (Lampetra Anderson, J. H., and T. P. Quinn. 2007. Movements of adult Coho Salmon (On- tridentata) in the Columbia River, U.S.A. Environmental Biology of Fishes corhynchus kisutch) during colonization of newly accessible habitat. Cana- 85:253–264. dian Journal of Fisheries and Aquatic Sciences 64:1143–1154. Kelso, J. R. M., and W. M. Gardner. 2000. Emigration, upstream move- Andrade, N. O., B. R. Quintella, J. Ferreira, S. Pinela, I. Povoa,´ S. Pedro, ment, and habitat use by sterile and fertile Sea Lampreys in three Lake and P. R. Almeida. 2007. Sea Lamprey ( petromyzon marinus L.) spawning Superior tributaries. North American Journal of Fisheries Management 20: migration in the Vouga River basin (Portugal): poaching impact, preferential 144–153. resting sites and spawning grounds. Hydrobiologia 582:121–132. Kircheis, F. W. 2004. Sea Lamprey Petromyzon marinus Linnaeus 1758. U.S. Arnason, A. N., and C. J. Schwarz. 1999. Using POPAN-5 to analyse banding Fish and Wildlife Service, White Paper, Carmel, Maine. data. Bird Study 46(Supplement 1):157–168. Limburg, K. E., and J. R. Waldman. 2009. Dramatic declines in North Atlantic Beamish, F. W. H. 1980. Biology of the North American anadromous Sea diadromous fishes. BioScience 59:955–965. Lamprey, Petromyzon marinus. Canadian Journal of Fisheries and Aquatic Lundquist, J. D., M. D. Dettinger, and D. R. Cayan. 2005. Snow-fed streamflow Sciences 37:1924–1943. timing at different basin scales: case study of the Tuolumne River above Binder, T. R., R. L. McLaughlin, and D. G. McDonald. 2010. Relative impor- Hetch Hetchy, Yosemite, California. Water Resources Research [online serial] tance of water temperature, water level, and lunar cycle to migratory activity 41(7):W07005. in spawning-phase Sea Lampreys in Lake Ontario. Transactions of the Amer- Maitland, P. S. 2003. Ecology of the River, Brook and Sea lamprey. Conserving ican Fisheries Society 139:700–712. Natura 2000 Rivers, Ecology Series 5, English Nature, Peterborough, UK. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel Malmqvist, B. 1980. The spawning migration of the Brook Lamprey, Lampetra inference: a practical information-theoretic approach, 2nd edition. Springer- planeri Bloch, in a south Swedish stream. Journal of Fish Biology 16:105– Verlag, New York. 114. Christie, G. C., and C. I. Goddard. 2003. Sea Lamprey international symposium Manion, P. J., and L. H. Hanson. 1980. Spawning behavior and fecundity of (SLIS II): advances in the integrated management of Sea Lamprey in the lampreys from the upper three Great Lakes. Canadian Journal of Fisheries Great Lakes. Journal of Great Lakes Research 29(Supplement 1):1–14. and Aquatic Sciences 37:1635–1640. CRASC (Connecticut River Atlantic Salmon Commission). 2011. Connecticut Masters, J. E. G., M. H. Jang, K. Ha, P. D. Bird, P. A. Frear, and M. C. Lucas. River basin anadromous fisheries restoration. U.S. Fish and Wildlife Service, 2006. The commercial exploitation of a protected anadromous species, the Connecticut River Coordinator’s Office, Sunderland, Massachusetts. Avail- River Lamprey (Lampetra fluviatilis L.), in the tidal River Ouse, north-east able: www.fws.gov/r5crc/. (April 2012). England. Aquatic Conservation: Marine and Freshwater Ecosystems 16:77– Freeman, M. C., C. M. Pringle, E. A. Greathouse, and B. J. Freeman. 2003. 92. Ecosystem-level consequences of migratory faunal depletion caused by dams. Montgomery, D. R., J. M. Buffington, N. P. Peterson, D. Schuett-Hames, and Pages 255–266 in K. E. Limburg and J. R. Waldman, editors. Biodiversity, T. P. Quinn. 1996. Stream-bed scour, egg burial depths, and the influence of status, and conservation of the world’s shads. American Fisheries Society, salmonid spawning on bed surface mobility and embryo survival. Canadian Symposium 35, Bethesda, Maryland. Journal of Fisheries and Aquatic Sciences 53:1061–1070. Gardner, C., S. M. Coghlan Jr., and J. Zydlewski. 2012. Distribution and abun- Moore, H. H., and L. P. Schleen. 1980. Changes in spawning runs of Sea dance of anadromous Sea Lamprey spawners in a fragmented stream: current Lamprey ( petromyzon marinus) in selected streams of Lake Superior af- status and potential range expansion following barrier removal. Northeastern ter chemical control. Canadian Journal of Fisheries and Aquatic Sciences

Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 Naturalist 19:99–110. 37:1851–1860. Gardner, C., S. M. Coghlan Jr., J. Zydlewski, and R. Saunders. 2013. Distri- Morbey, Y. 2000. Protandry in Pacific salmon. Canadian Journal of Fisheries bution and abundance of stream fishes in relation to barriers: implications and Aquatic Sciences 57:1252–1257. for monitoring stream recovery after barrier removal. River Research and Nagelkerke, N. J. D. 1991. A note on a general definition of the coefficient of Applications 29:65–78. determination. Biometrika 78:691–692. Guyette, M. Q. 2012. Responses of Atlantic Salmon stream communities to Nislow, K. H., and B. E. Kynard. 2009. The role of anadromous Sea Lamprey in marine-derived nutrients. Doctoral dissertation. University of Maine, Orono. nutrient and material transport between marine and freshwater environments. Hansen, L. P., and T. P. Quinn. 1998. The marine phase of the Atlantic Pages 485–494 in A.Haro,K.L.Smith,R.A.Rulifson,C.M.Moffitt,R.J. Salmon (Salmo salar) life cycle, with comparisons to Pacific salmon. Klauda, M. J. Dadswell, R. A. Cunjak, J. E. Cooper, K. L. Beal, and T. S. Canadian Journal of Fisheries and Aquatic Sciences 55(Supplement 1): Avery, editors. Challenges for diadromous fishes in a dynamic global envi- 104–118. ronment. American Fisheries Society, Symposium 69, Bethesda, Maryland. Hardisty, M. W., and I. C. Potter. 1971. The general biology of adult lampreys. Percy, R., J. F. Leatherland, and F. W. H. Beamish. 1975. Structure and ul- Pages 127–206 in M. W. Hardisty and I. C. Potter, editors. The biology of trastructure of the pituitary gland in the Sea Lamprey, Petromyzon mar- lampreys, volume 1. Academic Press, London. inus at different stages in its life cycle. Cell and Tissue Research 157: Hart, D. D., T. E. Johnson, K. L. Bushaw-Newton, R. J. Horwitz, A. T. Bednarek, 141–164. D. F. Charles, D. A. Kreeger, and D. J. Velinsky. 2002. Dam removal: Pledger, S., and M. Efford. 1998. Correction of bias due to heterogeneous cap- challenges and opportunities for ecological research and river restoration. ture probability in capture–recapture studies of open populations. Biometrics BioScience 52:669–682. 54:888–898. 1394 HOGG ET AL.

PRRT (Penobscot River Restoration Trust). 2012. Penobscot River restoration Siefkes, M. J., S. R. Winterstein, and W. Li. 2005. Evidence that 3-keto petromy- trust website. PRRT, Augusta, Maine. Available: www.penobscotriver.org/. zonol sulphate specifically attracts ovulating female Sea Lamprey, Petromy- (April 2012). zon marinus. Animal Behaviour 70:1037–1045. Purvis, H. A., C. L. Chudy, E. L. King Jr., and V. K. Dawson. 1985. Response of Smith, A. J., J. H. Howell, and G. W. Piavis. 1968. Comparative embryology of spawning-phase Sea Lampreys ( petromyzon marinus) to a lighted trap. Great five species of lampreys of the upper Great Lakes. Copeia 1968:461–469. Lakes Fishery Commision Technical Report 42:15–25. Smith, B. R., and J. J. Tibbles. 1980. Sea Lamprey ( petromyzon marinus)in Quinn, T. P., and D. J. Adams. 1996. Environmental changes affecting the lakes Huron, Michigan, and Superior: history of invasion and control, 1936– migratory timing of American Shad and Sockeye Salmon. Ecology 77:1151– 78. Canadian Journal of Fisheries and Aquatic Sciences 37:1780–1801. 1162. Smith, S. J., and J. E. Marsden. 2009. Factors affecting Sea Lamprey egg R Development Core Team. 2010. R: a language and environment for statisti- survival. North American Journal of Fisheries Management 29:859–868. cal computing. R Foundation for Statistical Computing, Vienna. Available: Sorensen, P. W., and L. A. Vrieze. 2003. The chemical ecology and potential www.R-project.org/. (March 2013). application of the Sea Lamprey migratory pheromone. Journal of Great Lakes Renaud, C. B. 1997. Conservation status of Northern Hemisphere lampreys Research 29(Supplement 1):66–84. (Petromyzontidae). Journal of Applied Ichthyology 13:143–148. Trepanier,´ S., M. A. Rodr´ıguez, and P. Magnan. 1996. Spawning migrations SAS (Statistical Analysis Systems). 2010. SAS/STATR user’s guide. SAS In- in landlocked Atlantic Salmon: time series modelling of river discharge and stitute, Cary, North Carolina. water temperature effects. Journal of Fish Biology 48:925–936. Saunders, R., M. A. Hachey, and C. W. Fay. 2006. Maine’s diadromous fish Vrieze, L. A., R. Bjerselius, and P. W. Sorensen. 2010. Importance of the community: past, present, and implications for Atlantic Salmon recovery. olfactory sense to migratory Sea Lampreys Petromyzon marinus seeking Fisheries 31:537–547. riverine spawning habitat. Journal of Fish Biology 76:949–964. Schwarz, C. J., R. E. Bailey, J. R. Irvine, and F. C. Dalziel. 1993. Estimating Wagner, C. M., M. B. Twohey, and J. M. Fine. 2009. Conspecific cueing in salmon spawning escapement using capture–recapture methods. Canadian the Sea Lamprey: do reproductive migrations consistently follow the most Journal of Fisheries and Aquatic Sciences 50:1181–1197. intense larval odour? Animal Behaviour 78:593–599. Scott, W. B., and E. J. Crossman. 1985. Freshwater fishes of Canada. Gordon White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation Soules, Vancouver. from populations of marked animals. Bird Study 46(Supplement 1):120–139. Downloaded by [Department Of Fisheries] at 21:35 27 October 2013 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:36 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Upper Thermal Tolerances of Rio Grande Cutthroat Trout under Constant and Fluctuating Temperatures Matthew P. Zeigler a , Stephen F. Brinkman b , Colleen A. Caldwell c , Andrew S. Todd d , Matthew S. Recsetar e & Scott A. Bonar f a Department of Fish, Wildlife, and Conservation Ecology , New Mexico State University , 2980 South Espina Street, Las Cruces , New Mexico , 88003 , USA b Colorado Parks and Wildlife , 317 West Prospect Road, Fort Collins , Colorado , 80526 , USA c U.S. Geological Survey, New Mexico Cooperative Fish and Wildlife Research Unit , New Mexico State University , 2980 South Espina Street, Las Cruces , New Mexico , 88003 , USA d U.S. Geological Survey, Crustal Geophysics and Geochemistry Science Center , Box 25046, Mail Stop 964D, Denver Federal Center, Denver , Colorado , 80225 , USA e Arizona Cooperative Fish and Wildlife Research Unit, School of Natural Resources and the Environment , University of Arizona , 104 Biosciences East, Tucson , Arizona , 85721 , USA f U.S. Geological Survey, Arizona Cooperative Fish and Wildlife Research Unit, School of Natural Resources and the Environment , University of Arizona , 104 Biosciences East, Tucson , Arizona , 85721 , USA Published online: 02 Sep 2013.

To cite this article: Matthew P. Zeigler , Stephen F. Brinkman , Colleen A. Caldwell , Andrew S. Todd , Matthew S. Recsetar & Scott A. Bonar (2013) Upper Thermal Tolerances of Rio Grande Cutthroat Trout under Constant and Fluctuating Temperatures, Transactions of the American Fisheries Society, 142:5, 1395-1405, DOI: 10.1080/00028487.2013.811104 To link to this article: http://dx.doi.org/10.1080/00028487.2013.811104

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1395–1405, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.811104

ARTICLE

Upper Thermal Tolerances of Rio Grande Cutthroat Trout under Constant and Fluctuating Temperatures

Matthew P. Zeigler* Department of Fish, Wildlife, and Conservation Ecology, New Mexico State University, 2980 South Espina Street, Las Cruces, New Mexico 88003, USA Stephen F. Brinkman Colorado Parks and Wildlife, 317 West Prospect Road, Fort Collins, Colorado 80526, USA Colleen A. Caldwell U.S. Geological Survey, New Mexico Cooperative Fish and Wildlife Research Unit, New Mexico State University, 2980 South Espina Street, Las Cruces, New Mexico 88003, USA Andrew S. Todd U.S. Geological Survey, Crustal Geophysics and Geochemistry Science Center, Box 25046, Mail Stop 964D, Denver Federal Center, Denver, Colorado 80225, USA Matthew S. Recsetar Arizona Cooperative Fish and Wildlife Research Unit, School of Natural Resources and the Environment, University of Arizona, 104 Biosciences East, Tucson, Arizona 85721, USA Scott A. Bonar U.S. Geological Survey, Arizona Cooperative Fish and Wildlife Research Unit, School of Natural Resources and the Environment, University of Arizona, 104 Biosciences East, Tucson, Arizona 85721, USA

Abstract The Rio Grande Cutthroat Trout Oncorhynchus clarkii virginalis is the southernmost subspecies of Cutthroat Trout, and as with the other subspecies, stream temperature regulates growth, reproductive success, distribution, and Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 survival. An understanding of the upper thermal tolerance of Rio Grande Cutthroat Trout is important for developing water temperature standards and for assessing suitable habitat for reintroduction and management. Hatch success of Rio Grande Cutthroat Trout eggs was determined under static temperatures. The thermal requirements of fry and juveniles were also assessed under static and fluctuating temperature regimes using the acclimated chronic exposure method. Egg hatch success was 46–70% from 6◦Cto16◦C but declined significantly at 18◦Cand20◦C. Maximum growth of fry that were fed to satiation occurred at 15.3◦C. The 30-d ultimate upper incipient lethal temperature (UUILT) was 22.6◦C for fry and 21.7◦C for juveniles. Survival during fluctuating temperature experiments was dependent upon the daily maximum temperature and the daily fluctuation. The upper thermal limits for Rio Grande Cutthroat Trout were lower than those of Rainbow Trout O. mykiss but similar to those of other Cutthroat Trout subspecies. The low UUILT of Rio Grande Cutthroat Trout relative to some salmonids may increase the risk of deleterious effects brought about by a changing climate, habitat alteration, and sympatric nonnative salmonids, which are known to outcompete Cutthroat Trout at temperatures above the species’ optimal range. Daily mean water temperatures near the Rio Grande Cutthroat Trout’s optimal growth temperature of 15◦C would be suitable

*Corresponding author: [email protected] Received March 12, 2013; accepted May 28, 2013 Published online September 2, 2013 1395 1396 ZEIGLER ET AL.

for reintroduction of this subspecies. Depending on the daily temperature fluctuation, daily maximum temperatures within reintroduction streams and current habitat should remain at or below 25◦C to ensure long-term persistence of a Rio Grande Cutthroat Trout population. This information will aid in establishing water quality standards to protect habitat where the subspecies currently occurs.

Similar to other subspecies of Cutthroat Trout Oncorhynchus fied exposure time and acclimation temperature (Beitinger et al. clarkii, the Rio Grande Cutthroat Trout O. clarkii virginalis is 2000). Using a range of acclimation and exposure temperatures, restricted to a small fraction (12%) of its historic range (Alves the upper ILT can be calculated as the most extreme tempera- et al. 2008). Introduction of nonnative salmonids and habitat ture at which 50% of the population could survive indefinitely destruction have caused declines in the subspecies’ distribution (Jobling 1981). (Pritchard and Cowley 2006). Recently, the Rio Grande Cut- Although both methods are widely used to establish species’ throat Trout was listed as a candidate for protection under the upper thermal limits, each has disadvantages, including the Endangered Species Act of 1973 (U.S. Federal Register 2008). stochastic effects of two independent variables (time and tem- Current management of this subspecies focuses on establishing perature) in the CTMax method (Beitinger et al. 2000) and new populations or expanding current populations to ensure fu- the confounding effects of handling on thermal stress in the ture persistence. Although stream temperature plays a vital role ILT method (Bennett and Judd 1992). Both methods also fail in the distribution, establishment, and persistence of Cutthroat to measure the effects of prolonged elevated temperatures that Trout populations (Harig and Fausch 2002; Dunham et al. can occur for several weeks during the summer (Caissie 2006). 2003; de la Hoz Franco and Budy 2005), thermal requirements The acclimated chronic exposure (ACE) method, a hybrid of specific to the Rio Grande Cutthroat Trout have not been inves- the ILT and CTMax methods, tests the thermal limits of fishes tigated. Current water temperature standards in New Mexico by using more ecologically relevant conditions (Zale 1984; Se- and Colorado were developed using laboratory-based thermal long et al. 2001; Bear et al. 2007). In the ACE method, accli- tolerance data for nonnative Cutthroat Trout subspecies (i.e., mation temperatures are gradually increased by 1◦C per day Bonneville Cutthroat Trout O. clarkii utah, Lahontan Cutthroat until test temperatures are reached, and the final test temper- Trout O. clarkii henshawi, Snake River Fine-Spotted Cutthroat atures are maintained for long periods (30–60 d). Survival is Trout O. clarkii behnkei, Westslope Cutthroat Trout O. clarkii recorded, and the ultimate upper ILT (UUILT; temperature that lewisi, and Yellowstone Cutthroat Trout O. clarkii bouvieri; is lethal to 50% of the population) for the test period is esti- Todd et al. 2008). Variation—albeit small—in response to mated (Selong et al. 2001). The increased duration of the test elevated temperatures among the Cutthroat Trout subspecies enables the measurement of sublethal effects (e.g., a reduc- (Wagner et al. 2001) may mean that these current standards tion in growth) in addition to survival (Widmer et al. 2006a; are not adequate to protect native Rio Grande Cutthroat Trout Bear et al. 2007). throughout the subspecies’ range. Static temperature treatments are used to measure the upper Thermal preferences and limits developed in laboratory set- thermal limits of fish in both the ACE and ILT methodolo- tings are often used to assess the suitability of potential habitat gies. Laboratory and field studies indicate that if temperatures for coldwater species and to establish water quality standards for subsequently return to lower, nonlethal levels, fish can survive protecting habitats (McCullough et al. 2001; Todd et al. 2008). short-term exposure to high temperatures that would otherwise

Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 Traditional laboratory studies assess thermal stress by manip- be lethal under long-term static conditions (Johnstone and Rahel ulating both temperature and exposure time. The most widely 2003; Schrank et al. 2003). Daily minimum temperatures may used methods of testing upper thermal tolerances of fishes in- provide a reprieve from otherwise stressful elevated tempera- clude the critical thermal maximum (CTMax) and the incipient tures experienced during diel temperature maxima (Dickerson lethal temperature (ILT). The CTMax method assesses acute and Vinyard 1999). The flexibility of the ACE method allows thermal limits by subjecting fish to a linear and rapid increase of for the application of diel cyclic temperatures for long periods temperature (e.g., 0.3◦C/min) from an acclimation temperature (>7 d); therefore, this method represents the most ecologically until a sublethal endpoint is reached, such as the loss of equi- relevant measure of a species’ temperature limits. However, librium or the onset of spasms (Beitinger et al. 2000). The ILT ascertaining a useful metric that describes a species’ thermal method evaluates the lethality of chronic temperatures by mea- limits when the fish are subjected to a diel fluctuation is diffi- suring survival for a predetermined amount of time (typically 7 cult because of the complexity of variables that describe a diel d) after an abrupt transfer of fish from an acclimation tempera- temperature cycle (i.e., phase duration, magnitude of the cycle, ture to environmentally relevant temperature treatment(s) (Ben- temperature minimum, and temperature maximum). Laboratory nett and Judd 1992). The median lethal temperature (LT50) is systems that are designed to mimic diel temperature fluctuations derived from survival data and is reported as the ILT for a speci- are also difficult to build and operate (Widmer et al. 2006b). THERMAL TOLERANCES OF RIO GRANDE CUTTHROAT TROUT 1397

In this study, a series of experiments was used to assess of eight head tanks in the static temperature experiment and the upper thermal tolerance of early life stages of Rio Grande three head tanks in the fluctuating temperature experiment, in Cutthroat Trout. The objectives were to estimate the effects which temperature was controlled using temperature program- of temperature on the viability and hatch rate of Rio Grande mers (Series 16B; Love Controls, Michigan City, Indiana) and Cutthroat Trout eggs and to determine the effects of static and aquarium heaters. Each test aquarium received water from the fluctuating temperatures on survival and growth of Rio Grande head tanks at a rate of 50 mL/s. Data loggers (Onset Computer Cutthroat Trout fry and juveniles. These results will directly Corp., Bourne, Massachusetts) recorded temperatures in test benefit management of the subspecies through the develop- aquaria at 1-h intervals during the experiments. Temperatures ment of water temperature standards and protective thermal were stable during all experiments and deviated little from set limits. points, with an average SD of 0.3◦C from the set point for each experiment. Mean daily temperatures during fluctuating temper- ature experiments were within 0.2◦C of set points. Fry were fed METHODS soft-moist trout starter (Rangen, Inc., Buhl, Idaho) five times Experimental background and fish care.—A series of ex- per day by using automatic feeders (Fish Mate, Conroe, Texas), periments was conducted at the Colorado Parks and Wildlife and this diet was supplemented with live brine shrimp Artemia (CPW) Aquatic Toxicology Laboratory, Fort Collins, to assess spp. nauplii (Argent Chemical Laboratories, Redmond, Wash- the effects of temperature on hatch rates of Rio Grande Cut- ington). Feeding rates were adjusted to ensure that fry were fed throat Trout eggs and the effects of static and fluctuating tem- in excess of satiation. Aquaria were cleaned daily to remove peratures on survival and growth of fry. A second experiment waste and uneaten food. Dissolved oxygen (mg/L), ammonia was conducted at the University of Arizona (UA) Environmen- (as total N, mg/L), and nitrite (mg/L) were monitored in test tal Research Laboratory, Tucson, to assess the effects of static aquaria throughout the entire length of each experiment and did and fluctuating temperatures on survival of juvenile Rio Grande not exceed allowable thresholds. Cutthroat Trout. Although the experiments were conducted in Age-0 fish were obtained from Seven Springs Hatchery (New separate settings, the test conditions and methods were similar Mexico Game and Fish, Jemez Springs) for use in the study at (Table 1). the UA Environmental Research Laboratory. The thermal testing Rio Grande Cutthroat Trout eggs were obtained from ma- facility used a recirculating water system with two head tanks ture, ripe adults that were captured in Haypress Lake (Mineral maintained at 10◦C and 35◦C. Water from both head tanks was County, Colorado). Eggs were stripped, fertilized, and water mixed to obtain test temperatures using computer-controlled In- hardened in the field and were then transported to the CPW tellefaucets (Hass Manufacturing Company, Averill Park, New Aquatic Toxicology Laboratory. Upon arrival, eggs were treated York). Water was delivered to 75-L test tanks for 3 min of each for fungus with formalin at 1,600 mg/L for 15 min (Piper et al. half-hour at a rate of 4 L/min. A recirculating water system 1952). The thermal testing facility used a flow-through system pumped water from test tanks through a biofilter, particle filters,

TABLE 1. Comparison of methods used in static and fluctuating temperature experiments with Rio Grande Cutthroat Trout at the Colorado Parks and Wildlife (CPW) Aquatic Toxicology Laboratory and at the University of Arizona (UA) Aquatic Ecology Laboratory.

Method CPW UA Static temperature Pre-test acclimation 14 d at 14◦C 14dat14◦C Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 Increase rate 1◦C per day 1◦C per day Fish size 0.18 g (14 d post-swim-up) 2.65 g (SD = 1.13 g) Life stage Fry Juvenile Density 0.47 g/L 1.13 g/L Test temperatures 10, 12, 14, 16, 18, 20, 22, 24, 26◦C 17, 19, 22, 24, 26, 28◦C Experiment duration 60 d 30 d Fluctuating temperature Pre-test acclimation 7 d at 20◦C 14dat14◦C Increase rate 1◦C per day 1◦C per day Fish size 0.88 g (110 d posthatch) 2.65 g (SD = 1.13 g) Life stage Fry Juvenile Density 0.97 g/L 1.13 g/L Test temperatures 18–22◦C and 15–25◦C 16–22, 19–25, 17–27, 21–27, and 19–29◦C Experiment duration 30 d 30 d 1398 ZEIGLER ET AL.

and a 390-W ultraviolet sterilizer, and then back to the two head no longer observed in the study. Testing began in August 2010 tanks. Each test tank contained an air stone, a small powerhead and ended in October 2010. pump (to continuously mix the water), and a thermocouple. The At the UA Environmental Research Laboratory, 30 age-0 fish thermocouple in each test tank recorded temperature at 10-min (mean weight = 2.6 g) were netted, measured (to the nearest intervals and was integrated with Labview software (National mm TL), weighed (to the nearest 0.1 g), and placed into each of Instruments, Austin, Texas) for controlling tank temperatures. 18 randomly selected, 75-L tanks. Fish were acclimated at 14◦C Temperatures during both static and fluctuating temperature ex- for a minimum of 14 d. After the acclimation period, tempera- periments were consistent and deviated little from set points. ture was increased at a rate of 1◦C per day until experimental The deviation from the set point averaged 0.4◦C (range = 0.1– temperatures were reached. Temperature increases from accli- 0.9◦C) for static temperatures and 0.2◦C (range = 0.0–0.3◦C) for mation temperatures were staggered to allow all treatments to fluctuating temperatures during the experiments. Fish were fed reach their target temperatures on the same day. After temper- daily to satiation using BioVita Starter (Bio-Oregon, Longview, ature treatments were reached, they were maintained for 30 d. Washington). All test tanks were cleaned daily to remove waste Treatments were 17, 19, 22, 24, 26, and 28◦C; each treatment and uneaten food. Dissolved oxygen (mg/L), ammonia (as total was replicated three times. Testing began in October 2010 and N, mg/L), and nitrite (mg/L) were monitored daily through- ended in December 2010. out all test tanks for the entire experiment and did not exceed Fluctuating temperature experiments.—At the CPW Aquatic allowable thresholds. Toxicology Laboratory, survival and growth of fry were assessed Egg experiment.—Twenty eggs were distributed to incuba- at a static temperature of 20◦C and under two temperature fluc- tion cups constructed from 1,000-µm-mesh nylon screen that tuations: ± 2◦C and ± 5◦C. Both fluctuation treatments had a was affixed to polyvinyl chloride pipe sections (2.5 × 2.5 × daily mean temperature of 20◦C (i.e., 20 ± 2◦C = 18–22◦C; 7.5 cm) by use of aquarium-grade silicone adhesive. A single 20 ± 5◦C = 15–25◦C) and were replicated three times. After incubation cup was suspended in a 7.6-L glass aquarium and acclimation to 20◦C for 7 d, six groups of 21 fry (mean weight = received 40 mL of water per minute from one of eight aerated, 0.88 g) were randomly selected and distributed among six 19-L temperature-controlled head tanks. The initial temperature in aquaria. Fry were added to test tanks at 1200 hours on the rising each experimental unit was 12◦C and was adjusted up or down limb of the fluctuation cycle when test temperatures reached over 24 h until the test temperature was reached. Survival and 20◦C. Water from temperature-controlled head tanks was de- hatch rates were determined at test temperatures of 6, 8, 10, 12, livered to test aquaria at a rate of 90 mL/min. The minimum 14, 16, 18, and 20◦C; each test temperature was replicated three temperature of the fluctuation occurred at 0600 hours, and max- times. Data loggers (Onset) in each egg incubation cup recorded imum temperature of the fluctuation occurred at 1800 hours. water temperatures at 1-h intervals. Egg mortality and hatching Seven fry from each tank were subsampled after 10, 20, and 30 were monitored and recorded daily. d. Subsampled fry were euthanized, blotted dry with a paper Static temperature experiments.—At the CPW Aquatic Tox- towel, and weighed to the nearest 0.001 g. icology Laboratory, 20 fry (mean weight = 0.18 g) were Survival of juveniles exposed to ± 3◦C and ± 5◦C fluctua- randomly selected and distributed into one of twenty-seven tions was assessed at the UA Environmental Research Labora- 7.6-L glass aquaria. Each aquarium received water at a rate tory. Fluctuation treatments had daily mean temperatures of 19, of 50 mL/min from one of nine aerated, temperature-controlled 22, and 24◦C; however, the 19◦C trials included only a ± 3◦C head tanks. Treatments were 10, 12, 14, 16, 18, 20, 22, 24, and fluctuation because of the cooling limitations of the system. The 26◦C, and each treatment was replicated three times. The tem- five temperature treatments were each replicated three times. At peratures of the head tanks were initially set at 14◦C for 14 d and the start of the experiment, 30 fish (mean weight = 2.65 g) were Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 then were adjusted to target temperatures at a rate of 1◦C per day. randomly selected, measured (to the nearest mm TL), weighed Adjustments from acclimation temperatures were staggered so (to the nearest 1.0 g), and placed into one of 15 randomly se- that all treatments achieved the target temperature on the same lected, 75-L tanks. Fish were acclimated to 14◦C for a minimum day. Five fry from each tank were subsampled without replace- of 14 d. Temperatures in test tanks were increased at a rate of ment to assess growth after target temperatures were attained 1◦C per day until the daily mean of each treatment (19, 22, or on day 0, and fry were subsampled again at 20, 40, and 60 d. 24◦C) was reached. Once the daily mean was reached, temper- Subsampled fry were terminally anesthetized (Finquel, Argent ature fluctuations were initiated. The minimum temperature oc- Laboratories, Redmond, Washington), blotted dry with a paper curred at 0200 hours, and the maximum temperature occurred towel, and weighed to the nearest 0.001 g. Subsampling without at 1400 hours. Testing began in October 2010 and ended in replacement reduced the cumulative effects of handling stress December 2010. and minimized density-dependent effects among the experimen- Statistical analysis.—Egg mortality and hatch rates were tal units. Although subsampling reduced the power to detect arcsine–square root transformed and analyzed with ANOVA. temperature-related mortality in the latter stages of the test, it is When statistical differences were observed, means from the dif- improbable that removal significantly affected the UUILT be- ferent temperature treatments were compared by using Tukey’s cause the first subsampling event occurred after mortality was test (α = 0.05). Mean growth rate (mg·fish−1·d−1) in each THERMAL TOLERANCES OF RIO GRANDE CUTTHROAT TROUT 1399

aquarium was determined by using linear regression of mean 0.0001). Only 1 of 60 eggs hatched at 20◦C, and the sac fry died fry weight as a function of time. Growth rates were plotted the next day. Mean number of degree-days to hatch decreased against mean temperature, and a second-order polynomial re- with increasing temperature (Figure 1). gression line was fitted to the data to determine the temperature of maximum growth. The UUILT was calculated as the LT50 Static Temperature Experiments (temperature considered lethal to 50% of the fish) at the end Survival of Rio Grande Cutthroat Trout fry was high (≥95%) of each experiment by using the trimmed Spearman–Karber during acclimation and did not differ among temperature treat- method (Hamilton et al. 1977). ments. Temperature-related mortality of fry occurred shortly after target temperatures were reached: all fry died within 15dat24◦C and within 5 d at 26◦C (Figure 2). During the RESULTS 60-d test, survival of fry held at constant temperatures (10◦C ◦ Egg Experiment and 22 C) varied from 87% to 100%. Survival of juveniles dur- ing acclimation and temperature ramping was high (100%) for Hatch success of Rio Grande Cutthroat Trout eggs was 46– ◦ 70% at test temperatures of 6–16◦C but was significantly re- all treatments except the 28 C treatment, in which all fish died ◦ ◦ before the target temperature was attained. Mortality in the static duced (<22%) at 18 C and 20 C (Figure 1; F7, 16 = 27.3; P < 26◦C treatment began on day 1, and all juveniles died by day 4 (Figure 2). Mortality of juveniles did not begin until day 4 in Downloaded by [Department Of Fisheries] at 21:36 27 October 2013

FIGURE 1. Average hatch success (%; ± SD; upper panel) and mean degree- days (◦C) to hatch (lower panel) for Rio Grande Cutthroat Trout eggs in relation to temperature. Mean degree-days were calculated by summing the daily mean FIGURE 2. Daily mean survival (%) of Rio Grande Cutthroat Trout fry (upper temperature until hatch for each egg within an experimental unit and then panel) and juveniles (lower panel) at static temperatures of 22, 24, and 26◦C. calculating the average for that unit. Note that the x-axis (time [day]) scale differs between graphs. 1400 ZEIGLER ET AL.

FIGURE 3. Survival ( ± 95% confidence interval) in relation to temperature FIGURE 4. Growth rate (mg/d) of Rio Grande Cutthroat Trout fry in relation for fry (gray circles) and juveniles (black circles) of Rio Grande Cutthroat Trout. to temperature over a 60-d period. Dashed lines indicate the 95% confidence Each circle represents the temperature that was lethal to 50% of the test fish interval of the regression line. (LT50) for the given exposure time (day). Note that only the first 30 d of data for fry are depicted because no mortality occurred after day 15. Survival of juveniles during the acclimation and ramping period was 100%. Mortality in the 24 ± 5◦C treatment began ◦ the 22◦C treatment and until day 3 in the 24◦C treatment. All within a few minutes of reaching 29 C on the first day, and all fish died by day 12 in the 24◦C static treatment. Survival rates juveniles were dead within 30 min (data not shown). In both the ◦ ◦ of juveniles in the 17◦C and 19◦C static temperature treatments 22 ± 5 C and 24 ± 3 C treatments, mortality began on day 1 ◦ were high (>96%) for the entire 30-d experiment. (Figure 5). All fish in the 22 ± 5 C treatment died by day 5, ◦ The LT50 for fry rapidly decreased with time, from 25.7◦C and all fish in the 24 ± 3 C treatment died by day 7 (Figure 5). ◦ (95% confidence interval [CI] = 25.2–26.2◦C) on day 3 to Although mortality in the 22 ± 3 C treatment began on day 2, 22.6◦C on day 15 (Figure 3). No mortality occurred after day over 50% of the juveniles survived to day 17 of the experiment, 15 at any temperature. The 30-d UUILT for fry (initial mean and some fish survived the entire 30 d (Figure 5). Survival was ◦ weight = 0.18 g) was 22.6◦C. The LT50 followed a similar 100% in the 19 ± 3 C treatment. pattern for juveniles, declining rapidly from 24.2◦C (95% CI = 23.9–24.4◦C) on day 3 to 22.3◦C (95% CI = 21.9–22.6◦C) by Sublethal Effects day 15. The 30-d UUILT for juveniles (initial mean weight = In addition to decreased growth at higher temperatures, two ◦ = ◦ other sublethal effects were noted. Although fry survival was 2.65 g) was 21.7 C (95% CI 21.3–22.0 C; Figure 3). ◦ Growth of fry was linear over the 60-d test at constant temper- not altered at 22 C, severe scoliosis occurred in 50% of the atures. The R2 for regressions of mean fry weight versus time fry by 40 d and in 75% of the fry by 60 d. Three juveniles in ± ◦ in individual aquaria ranged from 0.91 to 0.99. Growth rate the 22 3 C treatment developed scoliosis, but they survived ◦ Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 ranged from a low of 5.3 mg/d at 21.9 C to a high of 43 mg/d the entire 30-d study. Fungus Saprolegnia spp. developed in at 15.0◦C. Growth rates increased at temperatures from 10◦C juveniles by day 1 of the higher temperature treatments. The ◦ ◦ = ± to 15 C and then declined at temperatures of 18 C or higher incidence of Saprolegnia spp. was 26% (SE 4%) in the 22 ◦ = ◦ (Figure 4). Estimated maximum growth of fry occurred at 3 C fluctuating treatment, 25% (SE 4%) in the 22 C static ◦ = ◦ 15.3 C. treatment, 23% (SE 3%) in the 24 C static treatment, and 7% (SE = 4%) in the 28◦C static treatment. However, Saprolegnia spp. was not observed in fish that were exposed to the 17◦C and Fluctuating Temperature Experiments 19◦C static treatments or the 19 ± 3◦C fluctuating treatment. Survival of Rio Grande Cutthroat Trout fry was 92% in the static 20◦C treatment, 97% in the 20 ± 2◦C fluctuating treatment, and 100% in the 20 ± 5◦C fluctuating treatment DISCUSSION (Figure 5). Mean growth rates were not significantly differ- Thermal tolerances of Rio Grande Cutthroat Trout (fry and ent among treatments (F2, 3 = 1.53; P = 0.28) and averaged juveniles) were within the range of published values for other 60 mg·fish−1·d−1 for the static 20◦C treatment, 58 mg·fish−1·d−1 Cutthroat Trout subspecies. The 7-d UUILTs of Rio Grande for the 20 ± 2◦C treatment, and 45 mg·fish−1·d−1 for the 20 ± Cutthroat Trout fry (24.7◦C) and juveniles (23.4◦C) were near 5◦C treatment. the lower bounds of reported ranges of upper thermal limits for THERMAL TOLERANCES OF RIO GRANDE CUTTHROAT TROUT 1401

The ACE method provides ecologically applicable thermal limits in comparison with traditional short-term static methods by better mimicking the thermal conditions experienced by fish in lotic and lentic environments, where high temperatures typ- ically last for several weeks during the summer (Selong et al. 2001; Bear et al. 2007). During acclimation, the slow increase in temperature to targeted levels allows the fish to acclimate to environmentally realistic temperature changes. Extending the test duration past the traditional 7 d combines an assessment of the cumulative effects of temperature and exposure time, such as delayed mortality or other sublethal influences (i.e., decreased growth, disease, and scoliosis). For example, the 30-d UUILT was 2◦C lower than the 7-d UUILT for both fry and juveniles. Westslope Cutthroat Trout and Bull Trout Salvelinus confluen- tus also exhibited delayed mortality and concomitant decreases in UUILTs at extended test durations (Selong et al. 2001; Bear et al. 2007). The addition of treatments with fluctuating temperatures in- creases our understanding of the upper thermal limits of Rio Grande Cutthroat Trout by subjecting the fish to environmentally relevant conditions (Zeigler et al. 2013). We demonstrated that Rio Grande Cutthroat Trout tolerated fluctuating temperatures that were lethal under static temperature exposures. Survival of fry was not affected by the 20 ± 5◦C fluctuating treatment even though temperatures exceeded the 30-d UUILT more than 40% of the time. In addition, no mortality occurred in the 19 ± 3◦C treatment while daily maximum temperature exceeded the 30-d UUILT. These results are consistent with other lab- oratory experiments (Dickerson and Vinyard 1999; Johnstone and Rahel 2003; Widmer et al. 2006a) and field observations (Schrank et al. 2003) in which brief exposures to lethal temper- atures were tolerated if fish had the opportunity to recover at FIGURE 5. Daily mean survival (%) of Rio Grande Cutthroat Trout fry that lower temperatures. Juvenile Rio Grande Cutthroat Trout were were exposed to 20◦C (static), 20 ± 2◦C, and 20 ± 5◦C temperature treatments unable to tolerate fluctuating temperatures when daily maxima for 30 d (upper panel); and mean survival of juveniles that were exposed to 19 were too high and when the duration of exposure was too long ◦ ◦ ◦ ◦ ± 3 C, 22 ± 3 C, 22 ± 5 C, and 24 ± 3 C treatments for 30 d (lower panel). (i.e., the 22 ± 5◦C, 24 ± 3◦C, and 24 ± 5◦C treatments). Further investigation into the physiological effects of fluctu- salmonids (McCullough 1999), similar to the 7-d UUILTs of ating temperatures on a species’ upper thermal limits is needed Bonneville Cutthroat Trout (24.2◦C; Johnstone and Rahel 2003) to clarify how survival is affected by the interaction of max- Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 and Westslope Cutthroat Trout (24.1◦C; Bear et al. 2007) and imum temperature and fluctuating temperature. Experimental the 4-d UUILT of Lahontan Cutthroat Trout (23.3–23.6◦C; Vigg design should focus on mimicking environmental conditions and Koch 1980). Of the UUILTs from these previous studies, (i.e., similar fluctuations and daily maxima) with scenarios of only that of Westslope Cutthroat Trout was obtained using the increasing daily means and maxima to determine how increases ACE method, thus hindering direct comparisons among the sub- in temperature, either anthropogenic or naturally produced, af- species. Compared with nonnative salmonids that occur within fect the species. Managers can then develop metrics to evaluate the subspecies’ current range, the 7-d UUILT of Rio Grande stream thermal conditions that include exposure time and the Cutthroat Trout was lower than the 7-d UUILT of Rainbow daily temperature range to assess suitable habitat. This study Trout O. mykiss (26.0◦C; Bear et al. 2007), which was also used daily temperature fluctuations similar to those observed obtained by using the ACE method. Although the 7-d UUILT in habitat that is currently occupied by Rio Grande Cutthroat for Rio Grande Cutthroat Trout was similar to those reported Trout (Table 2; Zeigler et al. 2013). Acclimation procedures for Brown Trout Salmo trutta (24.7◦C; Elliott 1981) and Brook for determining thermal limits under fluctuating temperatures Trout Salvelinus fontinalis (24.5◦C; McCormick et al. 1972), must also be standardized to allow for meaningful comparisons differences in acclimation procedures, fish sizes, and ages pre- among species. One possible procedure for standardizing treat- clude direct comparisons. ments of fluctuating temperature would be to increase the static 1402 ZEIGLER ET AL.

TABLE 2. Information on daily temperature ranges during the summer period (July 1–September 30, 2010 and 2011) in several streams (within Colorado and New Mexico) that support populations of Rio Grande Cutthroat Trout. Daily temperature ranges in this table represent the total daily temperature fluctuation. For comparison with temperature fluctuations used in the experiment, the experimental fluctuation must be multiplied by 2. For instance, the daily range for the ± 3◦C fluctuation is 6.0◦C. Further information on collection of these stream temperature data is provided by Zeigler et al. (2013).

Maximum 7-d average Minimum 7-d average Average summer daily range (◦C) daily range (◦C) daily range (◦C) Stream State 2010 2011 2010 2011 2010 2011 Alamosito Creek CO 4.76 4.89 2.67 2.26 3.74 3.40 Cuates Creek CO 3.01 3.37 1.76 1.56 2.44 2.50 Jaroso Creek CO 5.44 6.84 3.00 2.85 4.19 4.25 North Fork Trinchera Creek CO 6.11 7.13 3.15 3.55 4.53 5.37 Rhodes Gulch CO 6.06 6.99 3.01 3.14 4.50 4.62 San Francisco Creek CO 6.42 7.55 2.99 2.75 4.66 4.37 Wagon Creek CO 6.44 8.07 3.98 3.61 5.37 5.80 West Indian Creek CO 7.59 9.36 4.42 4.64 5.79 6.93 Little Vermejo Creek NM 11.36 12.34 5.99 5.92 9.61 9.20 Ricardo Creek NM 7.67 7.90 3.86 3.97 5.86 5.73 Oiser Creek CO 9.70 10.69 5.83 4.06 7.43 6.75 Jack’s Creek NM 5.89 6.15 2.62 2.67 4.43 4.00 Canones˜ Creek NM 6.85 7.18 3.73 3.43 5.35 4.73 Cave Creek CO 4.11 5.75 1.99 2.04 2.76 3.26 East Middle Creek CO 7.36 8.38 4.20 2.69 5.37 4.90 Jack’s Creek CO 10.73 7.16 5.22 2.46 8.17 4.39 Prong Creek CO 7.88 8.44 3.21 3.30 6.02 5.45 Cabresto Creek NM 6.39 7.80 3.53 2.88 4.94 5.01 Columbine Creek NM 4.26 5.12 2.27 2.25 3.49 3.69 Comanche Creek NM 12.16 13.85 5.75 5.79 9.13 10.24 Costilla Creek NM 13.61 14.83 8.50 7.59 11.68 11.18 East Fork Costilla Creek CO 12.10 13.48 7.85 7.13 10.47 10.30 Italianos Creek NM 5.26 5.44 2.95 2.47 4.14 3.80 Little Costilla Creek NM 8.13 10.13 4.00 4.70 5.72 6.96 Policarpio Creek NM 6.70 6.09 2.98 2.72 4.80 4.60 Powderhouse Creek NM 8.82 8.04 5.07 4.06 6.55 5.56 San Cristobal Creek NM 3.40 4.00 1.85 1.82 2.48 2.72 Vidal Creek NM 11.57 11.18 4.88 5.73 8.95 8.70 West Fork Costilla Creek CO 10.59 11.22 6.80 4.82 8.30 7.71 Downloaded by [Department Of Fisheries] at 21:36 27 October 2013

acclimation temperature slowly (1◦C per day) until the mini- (see Snieszko 1970). Not surprisingly, increased incidence of mum or mean temperature of the fluctuation is reached. This Saprolegnia spp. infections and scoliosis in Rio Grande Cut- procedure would allow fish to fully acclimate to the test temper- throat Trout were observed at temperatures near the 30-d UUILT. atures, which would better mimic conditions experienced by fish Susceptibility to disease is not only dependent on the presence in the wild. Although this would increase the length of the exper- of a pathogen but also on the life cycle and thermal requirements iment, the importance of the acclimation procedure on tempera- of the disease. Although the coupled effects of increased tem- ture tolerances has been well established (Beitinger and Bennett perature and disease are difficult to determine in wild salmonid 2000), and its standardization for fluctuating temperature exper- populations, high temperatures and increased disease virulence iments would be beneficial in allowing comparisons between were observed to cause declines in a wild Brown Trout popula- species. tion (Hari et al. 2006). Careful consideration must be given to Sublethal effects other than reduced growth must be con- the sublethal effects of temperature in wild fish populations to sidered when assessing thermal effects on fish. Environmental evaluate how climate change and increasing temperatures affect stressors reduce a fish’s resistance to opportunistic pathogens population viability. THERMAL TOLERANCES OF RIO GRANDE CUTTHROAT TROUT 1403

Ecological Implications Wagner et al. 2001; Recsetar et al. 2012). Differences in up- Responses to temperature vary among Cutthroat Trout sub- per thermal tolerances due to the source of the test fish have species (Wagner et al. 2001; Myrick 2008) and even among also been observed in some species (Fields et al. 1987), al- populations of the same subspecies (Vigg and Koch 1980; Dri- though differences among salmonid species have demonstrated nan et al. 2012; Underwood et al. 2012). The upper thermal mixed results (see McCullough et al. 2009). Egg hatch success limits of Rio Grande Cutthroat Trout were similar to those of and timing (degree-days) of Rio Grande Cutthroat Trout were other Cutthroat Trout subspecies. Interestingly, the maximum in agreement with results for other salmonids, indicating that growth temperature of Rio Grande Cutthroat Trout fry was higher temperatures decreased hatch success and hatch timing 15.3◦C, similar to that of Colorado River Cutthroat Trout O. (Baird et al. 2002). Differences in thermal limits among Cut- clarkii pleuriticus (15.3–16.4◦C; Brandt 2009), slightly higher throat Trout merit further research given the critical status of the than those of Yellowstone Cutthroat Trout and Snake River Cut- majority of subspecies, their susceptibility to climate change throat Trout (14.5–14.7◦C; Myrick 2008), and higher than that (Wenger et al. 2011), and their importance in establishing cold- of Westslope Cutthroat Trout (13.6◦C; Bear et al. 2007). Dif- water temperature standards (Todd et al. 2008). ferences in temperature at maximum growth indicate an appar- The fundamental thermal niche, defined as the range from ◦ ◦ ent latitudinal pattern. The Rio Grande Cutthroat Trout has the 3 Clowerto1C higher than the optimal growth temperature ◦ southernmost distribution among Cutthroat Trout subspecies, (Christie and Regier 1998), is 13.3–16.3 C for Rio Grande Cut- and as such, this subspecies can be expected to be more warm- throat Trout. Although the fundamental niche was calculated adapted than other subspecies. Strong support exists for local under ideal conditions in a laboratory setting, one can expect thermal adaptation within Cutthroat Trout populations (Drinan that as temperature exceeds the thermal limits, a decrease in indi- et al. 2012). The latitudinal trend suggests that subspecies of vidual growth and a reduction in population viability will occur. Cutthroat Trout have become thermally adapted to the temper- Decreased food availability and the presence of nonnative fishes ature regimes where they reside, effectively maximizing their may shift the fundamental thermal niche to lower temperatures growth at the temperature most commonly experienced by each (Wootton 1998; Taniguchi and Nakano 2000). Coupling of labo- subspecies. ratory findings with field observations of realized thermal niches While support exists for local thermal adaptation of maxi- could resolve the thermal requirements of targeted species (Huff mum growth, the UUILTs of Cutthroat Trout subspecies do not et al. 2005). Although there is no empirical evidence describ- exhibit a similar latitudinal pattern. Adaptation to temperature ing the preferred thermal niche of Rio Grande Cutthroat Trout, for growth raises the question as to why adaptation to elevated historic and current distributions provide ancillary support that ◦ temperatures is not manifested in thermal tolerances as well the subspecies’ suitable temperature range is 19.0 C and below (see McCullough et al. 2009). In salmonids, for example, there (Haak et al. 2010). This is similar to the thermal range observed ◦ appears to be little variation in upper thermal tolerances among in our study. Daily maximum temperatures near 25 C appear to genera (McCullough 1999). Size-mediated differences in ther- be near the upper limits for Rio Grande Cutthroat Trout when mal limits within a species (Recsetar et al. 2012; Underwood exposed to fluctuating temperatures. Survival of fish that are ex- ◦ et al. 2012) may be related to biomechanical (surface area : posed to daily maximum temperatures of 25 C may depend on volume ratio) processes. the diel temperature range and the amount of time the fish spend at cooler temperatures during the day. Although the upper ther- Applicability of Laboratory Tests mal limits of Rio Grande Cutthroat Trout are similar to those of Laboratory assessments of thermal limits eliminate compli- other Cutthroat Trout subspecies, the low upper thermal limits cating conditions, such as disease or food limitations. This study of the Rio Grande Cutthroat Trout increase this subspecies’ risk Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 combined a series of experiments that examined the upper ther- of deleterious effects from increased stream temperature caused mal tolerances of Rio Grande Cutthroat Trout early life stages by a changing climate and habitat alterations. The ability of (eggs and fry) and juveniles. Mortality of fry and juveniles was nonnative salmonids to outcompete Cutthroat Trout at temper- rapid at static temperatures of 24◦C and 26◦C. The UUILTs of atures above a subspecies’ optimal limit (De Staso and Rahel the fry and juveniles tracked closely through time, whereas the 1994; Bear et al. 2007) will also lead to increased negative in- UUILT of juveniles was about 1◦C lower, presumably due to teractions among fishes as stream temperatures increase in the mortality of juveniles at 22◦C, which did not occur among fry. future. Differences in the UUILT between the two life stages may have been caused by the sources of test fish, the observed Sapro- legnia spp. infections in juveniles, and differences in the sizes ACKNOWLEDGMENTS of test fish (initial mean weights = 0.18 g versus 2.65 g). Al- A Crockett and P. Forsburg assisted with data collection and though larger fish are considered more thermally sensitive to the care of embryos and fry for experiments conducted at the high water temperatures than smaller fish (Meeuwig et al. 2004; CPW Aquatic Toxicology Laboratory, Fort Collins. R. Galindo Underwood et al. 2012), an effect of size on thermal tolerance and S. Hall assisted with experiments conducted at the UA is not consistent in the literature (De Staso and Rahel 1994; Environmental Research Laboratory. We also thank K. Patten, 1404 ZEIGLER ET AL.

M. Sloane, and T. Jacobson (New Mexico Department of Game Dunham, J., R. Schroeter, and B. Rieman. 2003. Influence of maximum water and Fish) for providing study fish used at the UA Environmental temperature on occurrence of Lahontan Cutthroat Trout within streams. North Research Laboratory. The study at the CPW Aquatic Toxicol- American Journal of Fisheries Management 23:1042–1049. Elliott, J. M. 1981. Some aspects of thermal stress on freshwater teleosts. Pages ogy Laboratory was supported in part by the U.S. Fish and 209–245 in A. D. Pickering, editor. Stress and fish. Academic Press, New Wildlife Service (Federal Aid Grant F-243). Funding for the York. study at the UA laboratory was provided by the U.S. Geolog- Fields, R., S. S. Lowe, C. Kaminski, G. S. Whitt, and D. P. Philipp. 1987. Critical ical Survey’s Science Support Partnership. Additional funding and chronic thermal maxima of northern and Florida Largemouth Bass and was provided by the Agriculture Experiment Station and De- their reciprocal F1 and F2 hybrids. Transactions of the American Fisheries Society 116:856–863. partment of Fish, Wildlife, and Conservation Ecology at New Haak, A. L., J. E. Williams, D. Isaak, A. Todd, C. C. Muhlfeld, J. L. Kershner, Mexico State University. Any use of trade names is for descrip- R. E. Gresswell, S. W. Hostetler, and H. M. Neville. 2010. The potential tive purposes only and does not imply endorsement by the U.S. influence of changing climate on the persistence of salmonids of the inland Government. This study was performed under the auspices of West. U.S. Geological Survey, Open-File Report 2010-1236, Reston, Virginia. New Mexico State University’s Institutional Animal Care and Available: pubs.usgs.gov/of/2010/1236. (November 2010). Hamilton, M. A., R. C. Russo, and R. V. Thurston. 1977. Trimmed Spearman- Use Committee (Number 2009-006). Karber method for estimating median lethal concentrations in toxicity bioas- says. Environmental Science and Technology 11:714–719. Hari, R. E., D. M. Livingstone, R. Siber, P. Burkhardt-Holm, and H. Guttinger.¨ REFERENCES 2006. Consequences of climatic change for water temperature and Brown Alves, J. E., K. A. Patten, D. E. Brauch, and P. M. Jones. 2008. Range-wide Trout populations in Alpine rivers and streams. Global Change Biology status of Rio Grande Cutthroat Trout (Oncorhynchus clarki virginalis): 2008. 12:10–26. Colorado Parks and Wildlife, Denver. Available: wildlife.state.co.us. Harig, A. L., and K. D. Fausch. 2002. Minimum habitat requirements for es- (September 2011). tablishing translocated Cutthroat Trout populations. Ecological Applications Baird, H. B., C. C. Krueger, and D. C. Josephson. 2002. Differences in incubation 12:535–551. period and survival of embryos among Brook Trout strains. North American Huff, D. D., S. L. Hubler, and A. N. Borisenko. 2005. Using field data to estimate Journal of Aquaculture 64:233–241. the realized thermal niche of aquatic vertebrates. North American Journal of Bear, E. A., T. E. McMahon, and A. V. Zale. 2007. Comparative ther- Fisheries Management 25:346–360. mal requirements of Westslope Cutthroat Trout and Rainbow Trout: im- Jobling, M. 1981. Temperature tolerance and the final preferendum—rapid plications for species interactions and development of thermal protec- methods for the assessment of optimum growth temperatures. Journal of tion standards. Transactions of the American Fisheries Society 136:1113– Fish Biology 19:439–455. 1121. Johnstone, H. C., and F. J. Rahel. 2003. Assessing temperature tolerance of Beitinger, T. L., and W. A. Bennett. 2000. Quantification of the role of acclima- Bonneville Cutthroat Trout based on constant and cycling thermal regimes. tion temperature in temperature tolerance of fishes. Environmental Biology Transactions of the American Fisheries Society 132:92–99. of Fishes 58:277–288. McCormick, J. H., K. E. F. Hokanson, and B. R. Jones. 1972. Effects of Beitinger, T. L., W. A. Bennett, and R. W. McCauley. 2000. Temperature tol- temperature on growth and survival of young Brook Trout, Salvelinus erances of North American freshwater fishes exposed to dynamic changes in fontinalis. Journal of the Fisheries Research Board of Canada 29:1107– temperature. Environmental Biology of Fishes 58:237–275. 1112. Bennett, W. A., and F. W. Judd. 1992. Comparison of methods for determining McCullough, D. A. 1999. A review and synthesis of effects of alterations to the low temperature tolerance: experiments with Pinfish, Lagodon rhomboides. water temperature regime on freshwater life stages of salmonids, with special Copeia 1992:1059–1065. reference to Chinook Salmon. U.S. Environmental Protection Agency, Report Brandt, M. M. 2009. Optimal starter diets and culture conditions for Colorado EPA 910-R-99-010, Seattle. River Cutthroat Trout (Oncorhynchus clarkii pleuriticus). Master’s thesis. McCullough, D. A., J. M. Bartholow, H. I. Jager, R. L. Beschta, E. F. Cheslak, Colorado State University, Fort Collins. M. L. Deas, J. L. Ebersole, J. S. Foott, S. L. Johnson, K. R. Marine, M. G. Caissie, D. 2006. The thermal regime of rivers: a review. Freshwater Biology Mesa, J. H. Petersen, Y. Souchon, K. F. Tiffan, and W. A. Wurtsbaugh. 2009. 51:1389–1406. Research in thermal biology: burning questions for coldwater stream fishes.

Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 Christie, G. C., and H. A. Regier. 1988. Measures of optimal thermal habitat Reviews in Fisheries Science 17:90–115. and their relationship to yields for four commercial fish species. Canadian McCullough, D. A., S. Spalding, D. Sturdevant, and M. Hicks. 2001. Summary Journal of Fisheries and Aquatic Sciences 45:301–314. of technical literature examining the physiological effects of temperature on de la Hoz Franco, E. A., and P. Budy. 2005. Effects of biotic and abiotic factors salmonids. U.S. Environmental Protection Agency, Report EPA-910-D-01- on the distribution of trout and salmon along a longitudinal stream gradient. 005, Washington, D.C. Environmental Biology of Fishes 72:379–391. Meeuwig, M. H., J. B. Dunham, J. P. Hayes, and G. L. Vinyard. 2004. Ef- De Staso, J., III, and F. J. Rahel. 1994. Influence of water temperature on fects of constant and cyclical thermal regimes on growth and feeding of interactions between juvenile Colorado River Cutthroat Trout and Brook juvenile Cutthroat Trout of variable sizes. Ecology of Freshwater Fish 13: Trout in a laboratory stream. Transactions of the American Fisheries Society 208–216. 123:289–297. Myrick, C. A. 2008. Development and evaluation of starter diets and culture Dickerson, B. R., and G. L. Vinyard. 1999. Effects of high chronic temper- conditions for three subspecies of Cutthroat Trout and Gila Trout. Western atures and diel temperature cycles on the survival and growth of Lahon- Regional Aquaculture Center, Annual Progress Report, University of Wash- tan Cutthroat Trout. Transactions of the American Fisheries Society 128: ington, Seattle. 516–521. Piper, R. G., I. B. McElwain, L. E. Orme, J. P. McCraren, L. G. Fowler, and J. R. Drinan, D. P., A. V. Zale, M. A. H. Webb, M. L. Taper, B. B. Shep- Leonard. 1952. Fish hatchery management. U.S. Fish and Wildlife Service, ard, and S. T. Kalinowski. 2012. Evidence of local adaptation in Wests- Washington, D.C. lope Cutthroat Trout. Transactions of the American Fisheries Society 141: Pritchard, V. L., and D. E. Cowley. 2006. Rio Grande Cutthroat Trout (On- 872–880. corhynchus clarkii virginalis): a technical conservation assessment. U.S. THERMAL TOLERANCES OF RIO GRANDE CUTTHROAT TROUT 1405

Forest Service, Rocky Mountain Region, Species Conservation Project, Vigg, S. C., and D. L. Koch. 1980. Upper lethal temperature range of Lahontan Hamilton, Montana. Available: www.fs.fed.us/r2/projects/scp/assessments. Cutthroat Trout in waters of different ionic concentration. Transactions of the (September 2008). American Fisheries Society 109:336–339. Recsetar, M. S., M. P.Zeigler, D. L. Ward, S. A. Bonar, and C. A. Caldwell. 2012. Wagner, E. J., R. E. Arndt, and M. Brough. 2001. Comparative tolerance of four Relationship between fish size and upper thermal tolerance. Transactions of stocks of Cutthroat Trout to extremes in temperature, salinity, and hypoxia. the American Fisheries Society 141:1433–1438. Western North American Naturalist 61:434–444. Schrank, A. J., F. J. Rahel, and H. C. Johnstone. 2003. Evaluating laboratory- Wenger, S. J., D. J. Isaak, C. H. Luce, H. M. Neville, K. D. Fausch, J. B. Dunham, derived thermal criteria in the field: an example involving Bonneville D. C. Dauwalter, M. K. Young, M. M. Elsner, B. E. Rieman, A. F. Hamlet, Cutthroat Trout. Transactions of the American Fisheries Society 132: and J. E. Williams. 2011. Flow regime, temperature, and biotic interactions 100–109. drive differential declines of trout species under climate change. Proceedings Selong, J. H., T. E. McMahon, A. V. Zale, and F. T. Barrows. 2001. Effect of the National Academy of Sciences of the USA 108:14175–14180. of temperature on growth and survival of Bull Trout, with application of an Widmer, A. M., C. J. Carveth, S. A. Bonar, and J. R. Simms. 2006a. Upper improved method for determining thermal tolerance in fishes. Transactions temperature tolerance of Loach Minnow under acute, chronic, and fluctuating of the American Fisheries Society 130:1026–1037. thermal regimes. Transactions of the American Fisheries Society 135:755– Snieszko, S. F., editor. 1970. A symposium on diseases of fishes and shellfishes. 762. American Fisheries Society, Special Publication 5, Washington, D.C. Widmer, A. M., C. J. Carveth, J. W. Keffler, and S. A. Bonar. 2006b. De- Taniguchi, Y., and S. Nakano. 2000. Condition-specific competition: implica- sign of a computerized, temperature-controlled, recirculating aquaria system. tions for the altitudinal distribution of stream fishes. Ecology 81:2027–2039. Aquacultural Engineering 35:152–160. Todd, A. S., M. A. Coleman, A. M. Konowal, M. K. May, S. Johnson, N. K. Wootton, R. J. 1998. Ecology of teleost fishes. Kluwer, Boston. M. Vieira, and J. F. Saunders. 2008. Development of new water temperature Zale, A. V.1984. Applied aspects of the thermal biology, ecology, and life history criteria to protect Colorado’s fisheries. Fisheries 33:433–443. of the Blue Tilapia, Tilapia aurea (Pisces: Cichlidae). Doctoral dissertation. Underwood, Z. E., C. A. Myrick, and K. B. Rogers. 2012. Effect of acclimation University of Florida, Gainesville. temperature on the upper thermal tolerance of Colorado River Cutthroat Zeigler, M. P., A. S. Todd, and C. A. Caldwell. 2013. Water temperature and Trout Oncorhynchus clarkii pleuriticus: thermal limits of a North American baseflow discharge of streams throughout the range of Rio Grande Cut- salmonid. Journal of Fish Biology 80:2420–2433. throat Trout in Colorado and New Mexico—2010 and 2011. U.S. Geo- U.S. Office of the Federal Register. 2008. Status review for Rio Grande Cutthroat logical Survey, Open-File Report 2013-1051, Reston, Virginia. Available: Trout. Federal Register 73:94(14 May 2008):27900–27926. pubs.usgs.gov/of/2013/1051/. (May 2013). Downloaded by [Department Of Fisheries] at 21:36 27 October 2013 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:37 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 Body Size and Growth Rate Influence Emigration Timing of Oncorhynchus mykiss Ian A. Tattam a d , James R. Ruzycki b , Hiram W. Li a & Guillermo R. Giannico c a Oregon Cooperative Fishery Research Unit, Department of Fisheries and Wildlife , Oregon State University, 104 Nash Hall , Corvallis , Oregon , 97331 , USA b Oregon Department of Fish and Wildlife, Eastern Oregon University, 203 Badgley , One University Boulevard , LaGrande , Oregon , 97850 , USA c Department of Fisheries and Wildlife , Oregon State University , 104 Nash Hall, Corvallis , Oregon , 97331 , USA d Oregon Department of Fish and Wildlife , Post Office Box 9, John Day , Oregon , 97845 , USA Published online: 02 Sep 2013.

To cite this article: Ian A. Tattam , James R. Ruzycki , Hiram W. Li & Guillermo R. Giannico (2013) Body Size and Growth Rate Influence Emigration Timing of Oncorhynchus mykiss , Transactions of the American Fisheries Society, 142:5, 1406-1414, DOI: 10.1080/00028487.2013.815661 To link to this article: http://dx.doi.org/10.1080/00028487.2013.815661

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1406–1414, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.815661

ARTICLE

Body Size and Growth Rate Influence Emigration Timing of Oncorhynchus mykiss

Ian A. Tattam*1 Oregon Cooperative Fishery Research Unit, Department of Fisheries and Wildlife, Oregon State University, 104 Nash Hall, Corvallis, Oregon 97331, USA James R. Ruzycki Oregon Department of Fish and Wildlife, Eastern Oregon University, 203 Badgley, One University Boulevard, LaGrande, Oregon 97850, USA Hiram W. Li Oregon Cooperative Fishery Research Unit, Department of Fisheries and Wildlife, Oregon State University, 104 Nash Hall, Corvallis, Oregon 97331, USA Guillermo R. Giannico Department of Fisheries and Wildlife, Oregon State University, 104 Nash Hall, Corvallis, Oregon 97331, USA

Abstract Juvenile Oncorhynchus mykiss migrate extensively in freshwater during the fall. We used individual tagging to study the spatial origin, influences, and outcomes of fall migration on fish that emigrated from summer rearing tributaries during the fall (early emigrants) and those that did not (late emigrants) in the South Fork John Day River, Oregon. Fall migration amplified body size differences between early and late emigrants. There were more early emigrants from a lower-gradient stream than from a higher-gradient stream. Early emigration was positively related to individual summer growth rate and fall body size. Oncorhynchus mykiss dispersed downstream into higher- order streams during the fall. Early emigrants shifted to an alternative location and experienced significantly greater winter growth than did late emigrants that remained in tributaries. Early emigrants initiated smolt migration sooner the following spring than did late emigrants. Early and late emigration from the South Fork John Day River was associated with asynchronous emigrant-to-adult survival rates. Downloaded by [Department Of Fisheries] at 21:37 27 October 2013

Stream fishes, including salmonids, can be highly mobile multiple behavior has been termed “partial migration” (Jonsson (Gowan et al. 1994; Kahler et al. 2001; Baxter 2002; Bram- and Jonsson 1993). blett et al. 2002; Gowan and Fausch 2002; Roni et al. 2012). Growth rate may influence which individuals become movers However, movement patterns are not uniform among streams and which become stayers (McMillan et al. 2012). Move- nor among individuals (Northcote 1992). Patterns of movement ment can be a density-dependent response wherein smaller, less vary among (Riddell and Leggett 1981) and within (Roni and dominant individuals are forced to emigrate (Chapman 1962; Quinn 2001; Steingr´ımsson and Grant 2003; Roni et al. 2012) Keeley 2001; Bujold et al. 2004; Imre et al. 2004; Griffiths et al. populations. Within a single population, there are both “movers” 2013). Conversely, dominant individuals may volitionally emi- and “stayers” (Leider et al. 1986; Grant and Noakes 1987). This grate in search of higher levels of resources in alternative areas

*Corresponding author: [email protected] 1Present address: Oregon Department of Fish and Wildlife, Post Office Box 9, John Day, Oregon 97845, USA. Received March 5, 2013; accepted June 4, 2013 Published online September 2, 2013

1406 SIZE AND GROWTH RATE INFLUENCE EMIGRATION TIMING 1407

(Armstrong et al. 1997; Roni and Quinn 2001; Gowan and (4) the smolt migration timing differs between early and late Fausch 2002). In other instances, social hierarchies, growth rate, emigrants. and condition factor may not appreciably influence which in- dividuals emigrate (Riddell and Leggett 1981; Giannico and Healey 1998; Kahler et al. 2001). For instance, Riddell and METHODS Leggett (1981) observed different proportions of fall emigrants Study location.—The South Fork John Day River (SFJD) between two streams although growth rate and condition of basin is a fifth-order watershed in northeast Oregon. Anadromy Atlantic Salmon Salmo salar parr were similar between the in the SFJD is limited by a waterfall at river kilometer (rkm; streams. They theorized that higher fall emigration from one measured as distance above the river mouth) 45. We studied stream was an adaptive response to the higher gradient and two tributaries downstream of this waterfall, Black Canyon and cooler temperature in that stream. Similarly, Bjornn (1971) Murderers creeks, and the SFJD downstream from the conflu- found fewer juvenile salmonids emigrated from experimental ence with Murderers Creek (Figure 1). These streams support a channels when large substrate was present, as opposed to small population of Oncorhynchus mykiss with both anadromous and gravel substrate. Hence, the influence of abiotic factors may nonanadromous life history forms. Mature nonanadromous O. override biotic factors in some streams. mykiss, commonly referred to as Rainbow Trout, are present Anadromy complicates the study of local movement because and rarely exceed 200 mm FL (McMillan et al. 2012). The anadromous salmonids will eventually smolt and migrate to the anadromous form is commonly referred to as summer steel- ocean. However, given this eventual migration, the timing and head. Juvenile summer steelhead commonly spend 2 years in the nature of individual migration in freshwater may still be influ- SFJD (range: 1–4 years). Principal downstream migration peri- enced by biotic variables. For example, growth rate (Thorpe ods for O. mykiss within the SFJD are fall (October–December) 1987a, 1987b; Thorpe and Metcalfe 1998; Cucherousset et al. and spring (April–May). During 2003–2004, 36% of the total 2005) and intraspecific competition (Chapman 1962; Hunting- annual emigrants migrated during October–February and 64% ford et al. 1988) may influence seasonal migrations that oc- migrated during February–June (Schultz et al. 2006). During cur prior to smoltification. The relative influence that biotic or late winter to early spring, juvenile summer steelhead undergo environmental factors experienced during the summer have on a physiological transformation referred to as “smolting,” which later movement patterns, such as fall migration, remains unclear facilitates migration to saltwater via the Columbia River. Mi- (Leider et al. 1986; Rodr´ıguez 2002). gration through the Columbia River occurs from April through Fall migrations are a common seasonal behavior for anadro- June. Steelhead spend one to two winters in the Pacific Ocean, mous salmonids (Bjornn 1971; Riddell and Leggett 1981; Roni et al. 2012). These migrations are often from low-order tributaries to higher-order streams. The locations selected for winter rearing influence growth and survival because higher- order streams are typically warmer, which may increase growth (Higgins 1985; Koskela et al. 1997; Morgan and Metcalfe 2001; Murphy et al. 2006) and decrease mortality (Smith and Griffith 1994). Additionally, the timing of smoltification may be advanced by an increased accumulation of degree-days (Zydlewski et al. 2005) in higher-order streams. We studied population-scale and individual-scale migration Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 of Oncorhynchus mykiss in the South Fork of the John Day River, Oregon. Our focus was on the movement patterns ex- pressed by O. mykiss during the fall, as behavior during the fall influences survival in the upcoming winter stress period. We tested four hypotheses concerning “early emigrants” (de- fined herein as individuals that emigrated from summer habi- tat during October–December and migrated downstream to higher-order streams for winter) and “late emigrants” (defined herein as individuals that did not emigrate from summer habi- FIGURE 1. Map of the South Fork John Day River (SFJD) basin showing the tat until after December). Our predictive hypotheses were that location of the PIT tag arrays, rotary screw trap, and tributary sampling universe. (1) the proportion of emigrants from a low-gradient creek dif- Inset shows the location of the SFJD basin in Oregon. Dashed arrow denotes fers from that of a high-gradient creek, (2) the growth rates streamflow direction. Oncorhynchus mykiss were PIT-tagged throughout the during summer differ between individuals who subsequently sampling universe (highlighted portions) of Black Canyon and Murderers creeks during summer 2004 and 2005. Numbered circles denote sentinel reaches where expressed early-emigrant or late-emigrant patterns, (3) the win- capture–recapture of O. mykiss occurred during all seasons. [Figure available in ter growth rates differ between early and late emigrants, and color online.] 1408 TATTAM ET AL.

TABLE 1. Summary of O. mykiss tagged with PIT tags in Murderers and Black Canyon creeks during summer 2004 and summer 2005. Estimated survival (95% confidence interval in parentheses) to John Day Dam (JDD) represents the component of each tagging cohort which survived and initiated anadromous migration.

FL at tagging (mm) Location Year n tagged Mean Median Range Survival to JDD Murderers 2004 1,125 120 117 65–232 19% (16–21%) Black Canyon 2004 1,203 113 109 63–224 16% (14–18%) Murderers 2005 1,521 128 125 65–248 16% (15–18%) Black Canyon 2005 1,449 113 109 62–229 13% (11–15%)

then reenter the Columbia River during July–September. Mature in each valley segment. In addition to our systematic summer adult steelhead spend one winter in freshwater prior to migrating capture and tagging, we randomly selected “sentinel reaches” back to the SFJD and spawning from March–May. The two life within each of the six geomorphic valley segments. Sentinel history forms are visually indistinguishable with the exception reaches encompassed at least five pools and ranged in thalweg of adult steelhead (FL of 500–800 mm), which are larger than length from 99 to 363 m. They were sampled four times per the nonanadromous form. Since all fish we captured and tagged year (June, September, December–January, March–April). The were <250 mm (Table 1) and may have been either form, we three sentinel reaches in each stream were labeled 1, 2, and 3 refer to all tagged individuals as O. mykiss.OftheO. mykiss that in ascending order proceeding upstream from the SFJD (Fig- we captured and marked during our study, similar proportions of ure 1). As O. mykiss from Black Canyon and Murderers creeks the individuals from Black Canyon and Murderers creeks ulti- emigrated into the SFJD during fall and winter, we sampled the mately migrated to the ocean based on our estimates of survival SFJD during December–January and March–April. We captured to John Day Dam on the Columbia River (Table 1; detection O. mykiss at sites representative of available habitat to estimate data from CBR 2012; survival estimates following Paulsen and winter growth rates in the lower SFJD. Fisher 2001). Capture methods varied by season. During June and Septem- The environments of Black Canyon Creek, Murderers Creek, ber, O. mykiss were captured via seining or electrofishing. Dur- and the SFJD differed. Water temperatures in Black Canyon ing December–January and March–April we night snorkeled Creek ranged from 20◦C during summer to 1◦C during win- with handheld dive lights and dipnetted O. mykiss. Captured ter, with minimal ice formation. Water temperatures in Mur- fish were anesthetized and scanned for the presence of a pas- derers Creek ranged from 26◦Cinsummerto0◦C in winter, sive integrated transponder (PIT) tag. If they were not already and surface ice (<20 cm thick) intermittently covered pools tagged, a full-duplex PIT tag (12 × 2.02 mm, 134.2 kHz ISO; and glides. Stream temperatures in the SFJD were similar to Digital Angel Corp., St. Paul, Minnesota) was injected into the Murderers Creek and ranged from 0◦C during winter to 26◦C peritoneal cavity (e.g., Prentice et al. 1990; PTSC 1999). We during summer. Ice formation occurred only in shaded canyon recorded FL (nearest mm) and location of capture (to the chan- sections. Stream flows in Black Canyon Creek ranged from nel unit scale, i.e., pool- or riffle-specific). Fish were recovered 0.3to5.7m3/s. Stream flows in Murderers Creek ranged from and released into their channel unit of capture. 0.07 m3/s during summer to 14.2 m3/s during winter. Stream Recapture and redetection.—We used recaptures at sentinel flows in the SFJD ranged from 0.5 to 70.8 m3/s. reaches of Murderers and Black Canyon creeks to estimate sea- Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 Capture and tagging.—Our sampling frame for Black sonal growth rates for each valley segment. When O. mykiss Canyon and Murderers creeks was defined as the distance of migrated past rkm 10 of the SFJD, we used a 1.52-m-diameter each creek that we could logistically access (Figure 1). Within rotary screw trap (RST; E.G. Solutions, Corvallis, Oregon; Fig- this sampling frame, we identified three distinct geomorphic val- ure 1) to capture them. We concurrently operated a PIT antenna ley segments in each creek (Frissell et al. 1986; Gregory et al. array 80 m upstream of the RST to detect previously PIT-tagged 1991). The Black Canyon Creek valley segments had gradients individuals (Figure 1). of 3.2, 4.8, and 5.7% (downstream to upstream). The Murder- During fall 2005 (September 27 to December 27), we used ers Creek valley segments had gradients of 1.3, 0.8, and 1.3% an array of two PIT tag antennas separated by more than 100 m (downstream to upstream). During summer 2004 and summer to detect O. mykiss emigrating from Murderers Creek. The array 2005, we systematically rotated sampling effort through each of was located 0.9 km upstream of the SFJD (Figure 1). The Mur- these six valley segments. A starting channel unit within each derers Creek PIT array allowed us to determine the direction of valley segment was randomly selected for each day’s sampling. movement and hence estimate the proportion of early emigrants We sampled upstream from the starting channel unit each day, from Murderers Creek independently of migration to the RST. and we sampled habitat units without replacement on subse- The detection efficiency of the Murderers Creek PIT array was quent visits to achieve the greatest spatial coverage possible estimated to be 82%. Logistical constraints precluded us from SIZE AND GROWTH RATE INFLUENCE EMIGRATION TIMING 1409

operating a PIT array on Black Canyon Creek. To avoid bias, planatory variables. Fork length and growth were not correlated we did not use data from the Murderers Creek PIT array and (r = 0.06, P = 0.67). There was evidence of a negative corre- only used data from the RST and PIT array at rkm 10 of the lation between pool depth and substrate size (r =−0.59, P = SFJD (Figure 1) to compare the proportion of early emigrants 0.02). However, this relationship was dominated by one outlier between Murderers and Black Canyon creeks. and, when removed, there was no significant correlation (r = As O. mykiss migrated through the Columbia River, they −0.41, P = 0.15). There were also no significant correlations had a probability of detection at fixed PIT arrays in John Day between biotic (FL, growth) and abiotic (depth, substrate size) Dam (Columbia River rkm 347, 4 km downstream of the John variables (r < 0.25, P ≥ 0.07). Day River confluence; CBR 2012) and mobile PIT arrays in the We jointly analyzed all three sentinel reaches in Murder- Columbia River estuary (Columbia River rkm 75; Ledgerwood ers Creek. Significant serial autocorrelation was present among et al. 2004; CBR 2012). These detections provided migration model residuals. We grouped individuals by channel unit and timing for early-emigrant and late-emigrant individuals. We as- then by length and progressively increased each length grouping sumed detection probability did not differ between groups (early until no significant autocorrelation was present among residuals. emigrant and late emigrant) within a year. Final FL categories were <130, 131–159, and >160 mm. This Statistical analyses.—We used z-tests to compare the propor- binomial logistic regression modeled the number of emigrants tions of O. mykiss migrating past the RST that had previously as a function of the number released in each group, as influenced been PIT-tagged anywhere in the sampling frame of Murder- by group means of each explanatory variable. We used Akaike ers or Black Canyon creeks (Table 1 summarizes the tagged information criterion (AICc) corrected for small sample size to individuals used for these comparisons). Since the RST sub- identify models which best explained the data with the fewest sampled the total migrant population, we estimated capture effi- parameters. ciency of the RST via upstream release and recapture of marked We compared growth for the period of December–January to O. mykiss (Tattam et al. 2013). We used these capture efficien- March–April among reaches. We compared the Black Canyon cies to estimate the abundance of PIT-tagged O. mykiss passing and Murderers creeks sentinel reaches and one reach in the the RST site. During periods when the RST was not operated, SFJD with one-way analysis of variance (ANOVA). We used we used the PIT arrays (efficiency of the arrays was calibrated Benjamini and Hochberg false discovery rate control with with captures at the RST during simultaneous operation) to es- α = 0.05 to control type-1 errors during multiple comparisons timate the abundance of PIT-tagged O. mykiss passing the RST (Verhoeven et al. 2005). site. After this calibration, which accounted for the influence of Finally, we compared the detection date at John Day Dam environmental variables on detection probability, we assumed of early-emigrant and late-emigrant O. mykiss from Murderers equal detection probability at this site for PIT-tagged individuals Creek. We used two-way ANOVA to test for differences in emigrating from our two study streams. mean detection date between groups and between years. We We used logistic regression to analyze fall emigration from used Pearson’s correlation to evaluate the relationship between sentinel reaches in Murderers Creek during 2005. We only in- the detection date at John Day Dam and the detection date in the cluded individuals that were captured in both June and Septem- Columbia River estuary for O. mykiss from all tagging locations ber (n = 54). Each PIT-tagged individual had a binary response in the SFJD. of either early emigrant (migrated past the Murderers Creek PIT array during fall) or late emigrant (not detected at the array during fall). This response was modeled as a function of FL RESULTS (measured in September), specific growth rate during summer Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 (specific growth rates were calculated as mm·mm−1·d−1, and we Prevalence of Early Emigration hereafter refer to specific growth rate as “growth”), substrate in- We estimated that 13.2% and 11.5% of O. mykiss PIT-tagged dex, pool depth, and stream geomorphic reach. Since the avail- in Murderers Creek migrated past rkm 10 of the SFJD dur- ability of winter concealment habitat may influence emigration ing fall 2004 and fall 2005, respectively. An estimated 3.1% rates (Bjornn 1971) in addition to FL and summer growth, we and 3.2% of individuals PIT-tagged in Black Canyon Creek quantified concealment habitat of the closest downstream pool to were early emigrants during 2004 and 2005 (Figure 2). The each individual’s location in September using maximum depth percentage of O. mykiss emigrating from Murderers Creek was and substrate size. We visually estimated substrate size (sand– significantly greater than the percentage emigrating from Black silt, gravel, cobble, boulder) and relative composition (dominant Canyon Creek during both fall 2004 (z = 9.0, P < 0.01) and fall or subdominant). We assigned numeric values, increasing with 2005 (z = 8.6, P < 0.01). The percentage emigrating during fall particle size (1 for sand–silt to 4 for boulder), then a weighted did not differ significantly between years within either Black sum [(1.25 · dominant) + (0.75 · subdominant)] was calcu- Canyon (z = 0.2, P = 0.42) or Murderers creeks (z = 1.3, P = lated for each pool. Higher scores indicated larger substrate and 0.09). Given the low percentage of early emigrants from Black presumably better overwintering habitat. Prior to developing Canyon Creek, we focused solely on Murderers Creek when logistic regression models, we tested for correlation among ex- examining potential correlates of fall emigration. 1410 TATTAM ET AL.

were not included in any of our competing models (Table 2). Interaction terms for FL × reach and growth × reach were in- cluded in two of our competing models. The presence of these interaction terms in our competing models indicated that the influence of FL and growth on early emigration was spatially dependent in Murderers Creek. Larger individuals were less likely to become early emigrants from upstream reaches than from downstream reaches.

Outcome of Early Emigration Early-emigrant and late-emigrant fish had differences in their growth rate during winter. We were unable to recapture PIT- tagged O. mykiss in reach 2 of Black Canyon Creek. Growth rates were significantly different among the remaining five reaches in Black Canyon Creek, Murderers Creek, and the SFJD (F = 26.7, P < 0.001). Mean growth in the SFJD was signif- FIGURE 2. Percentage of early-emigrant Oncorhynchus mykiss from Black 5, 41 Canyon and Murderers creeks during 2004 and 2005. The abundance of early icantly higher than mean growth in all other reaches, excepting emigrants was estimated at a rotary screw trap in the South Fork John Day River reach 2 in Murderers Creek (Figure 3). Within Murderers Creek, (rkm 10). Error bars indicate 95% confidence intervals. reaches 1 and 2 were both significantly different from reach 3 (Figure 3). There were no significant differences in growth Correlates of Early Emigration among reaches 1 and 3 in Black Canyon Creek and reach 3 in In Murderers Creek, our AICc selection identified three com- Murderers Creek (Figure 3). peting models (Table 2). All three competing models included Early emigration from Murderers Creek was associated with FL and reach as explanatory variables. Two of the competing differences in smolt migration timing at John Day Dam. Smolt models included growth, in addition to FL and reach (Table 2). timing was dependent on whether smolts were early emigrants Inclusion of these three explanatory variables in at least one or late emigrants, where early emigrants arrived significantly = < of the competing models indicated their significant association sooner at John Day Dam (F1, 132 17.3, P 0.001). Migration = with early emigration. Explanatory variables describing physi- timing of each group did not differ between years (F1, 132 = cal habitat characteristics of Murderers Creek (pool depth and 0.6, P 0.44) and there was no interaction between life history = = streambed particle size) were deemed not significant as they and year (F1, 132 0.4, P 0.54). In the spring of 2005, mean detection date at John Day Dam for early emigrants (May 3) was significantly earlier (F = 3.9, P = 0.05; Figure 4) TABLE 2. Model selection results for AICc analysis of early emigration from 1, 132 Murderers Creek. Explanatory variables evaluated were as follows: average FL, mean specific growth rate during summer (Growth), sentinel reach where tagged 0.25 (Reach), pool depth, and streambed particle size. Models with a delta AICc of c less than 5 and the null model are presented. The null model has no explanatory Reach 1

variables and serves as a check on the power of the explanatory variables. ·100) 0.20 Reach 2 -1

Multiplication signs indicate first order interactions. ·d Reach 3 b,c

-1 b Delta Model 0.15 Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 Model AICc AICc weight

(FL) + (Growth) + 132.1 0.0 0.29 0.10 a,b + × + (Reach) (FL Reach) a (Growth × Reach) + + 0.05

(FL) (Growth) 132.4 0.3 0.25 Specific Growth Rate (mm·mm a + × (Reach) (FL Reach) nr (FL) + (Reach) + 134.0 1.9 0.11 0.00 (FL × Reach) Black Canyon Murderers Creek SFJD + + (FL) (Growth) (Reach) 134.7 2.6 0.08 FIGURE 3. Comparison of mean specific growth rates during winter 2005 (Growth) + (Reach) + 135.0 2.9 0.07 among five reaches in Murderers and Black Canyon creeks and one reach in (Growth × Reach) the South Fork John Day River (SFJD). Oncorhynchus mykiss were individ- (Growth) + (Reach) 135.1 3.0 0.07 ually marked in December 2004 and recaptured in late March 2005. Letters (FL) + (Reach) 136.2 4.1 0.04 shared among bars indicate reaches that were not significantly different. Error bars indicate 95% confidence intervals and “nr” indicates that no recoveries of (Null) 165.1 33.0 0.00 individually marked fish occurred in this reach. SIZE AND GROWTH RATE INFLUENCE EMIGRATION TIMING 1411

12 reaches in these two streams has since indicated that O. mykiss A) were predominantly of anadromous maternal origin (Mills et al. 10 2012). Maturity sampling in these same reaches indicated that Early-Emigrant Late-Emigrant mature resident males were present (McMillan et al. 2012). Al- 8 though our data cannot distinguish residency from mortality, our lifetime tracking of tagged individuals (Table 1) suggests

6 that an approximately equal percentage of the individuals in each creek ultimately expressed anadromy. Our results demon- strated that, in two streams which had a comparable percentage 4 of anadromous individuals, fish size and growth rates influenced short-term migration patterns. 2 Our observation of lower emigration rates from higher- gradient reaches differs from that of Riddell and Leggett (1981). 0 Differences in stream temperature and growth potential may ex- 20 B) plain why a higher proportion of PIT-tagged fish emigrated from low-gradient than high-gradient reaches. Our lower-gradient Number of Fish Detected Fish of Number stream (Murderers Creek) had the coldest winter stream tem- 15 peratures, which was the inverse of what Riddell and Leggett (1981) observed. They also observed comparable growth rates between streams of differing gradient. However, we found in- 10 dividual summer growth rate was significantly greater in low- gradient Murderers Creek reaches than in high-gradient Black Canyon Creek (I. A. Tattam, unpublished data). Lower growth 5 rates in Black Canyon Creek possibly allowed fewer individu- als to reach length or growth thresholds (Metcalfe et al. 1988; Metcalfe 1998) needed to increase their odds of early emigra-

0 tion (Cucherousset et al. 2005). Additionally, McMillan et al. 10 5 20 5 30 y 5 10 15 20 25 30 e 4 (2012) observed that O. mykiss in Black Canyon Creek had ril ril1 ril ril2 ril a ay ay ay ay ay n Ap Ap Ap Ap Ap - M M M M M M Ju - - - - - 1 - - - - - 1 - il 6 11 16 21 26 ay y 6 11 16 21 26 3 higher lipid levels than those in Murderers Creek. Individuals in pr ril ril ril ril M a ay ay ay ay ay A Ap Ap Ap Ap M M M M M M Murderers Creek appear to be investing in length growth rather Date of Detection than lipid storage (McMillan et al. 2012). Greater fish size and FIGURE 4. Frequency of detection of early-emigrant and late-emigrant On- growth in Murderers Creek likely contributed to a higher pro- corhynchus mykiss at John Day Dam on the Columbia River (rkm 347) detected portion of individuals becoming early emigrants. Alternatively, out-migrating during (A) spring 2005 and (B) spring 2006. Date of detection our lower-gradient valley segments in Murderers Creek may intervals are shared between panels. have had higher sedimentation rates, which reduced interstitial concealment habitat (Cunjak 1996) and hence increased early than for late emigrants (May 10). During spring 2006, mean emigration (Bjornn 1971). However, our regression modeling suggests that FL and growth had greater influence on early detection dates also differed significantly (F1, 132 = 27.4, P < 0.001) between early emigrants (April 30) and late emigrants emigration than the habitat metrics that we measured. We sug- Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 (May 10; Figure 4). The detection date in the Columbia River gest that individual growth rate, and not stream gradient, was estuary was significantly correlated with the detection date at the principal driver of higher early emigration from Murderers John Day Dam during 2005 (r = 0.92, P < 0.001, n = 13) and Creek than Black Canyon Creek. 2006 (r = 0.99, P < 0.001, n = 12). Higher growth rates are typically associated with individu- als that are more “dominant” in the social hierarchy (Metcalfe et al. 1992). The notion that dominant individuals emigrated DISCUSSION from Murderers Creek instead of subdominant individuals seem- The percentage of O. mykiss adopting an early-emigrant strat- ingly contradicts typical experimental results (Chapman 1962; egy differed between our two study streams. The odds of early Keeley 2001; Bujold et al. 2004; Imre et al. 2004). When viewed emigration were associated with fish size and growth during the in the context of partial migration, however, our results become preceding summer. Fewer individuals in Black Canyon Creek more coherent. Variation in growth rates influences life history became early emigrants than in Murderers Creek. The parent- (Metcalfe 1998) and in Atlantic Salmon populations this is man- age of tagged individuals may have influenced our results. We ifested in “upper-modal” and “lower-modal” groups. Upper- did not determine the parentage (resident or anadromous) of the modal individuals are prompted to begin smolting in late sum- O. mykiss that we tagged. Otolith sampling within our study mer or early fall, whereas lower-modal individuals may not be 1412 TATTAM ET AL.

prompted to smolt until the following spring (Huntingford et al. during 2005 ocean entry but lower survival during 2006 ocean 1988; Whitesel 1993; Jonsson et al. 1998). Similarly, individual entry (Wilson et al. 2008). Brown Trout Salmo trutta (Cucherousset et al. 2005) and Brook Trout Salvelinus fontinalis (Morinville and Rasmussen 2003) Conclusions with a higher metabolic demand that could not be sustained in Fall emigration does not appear to be a fixed strategy but small streams emigrated to larger rivers. Chapman et al. (2011) rather a facultative tactic in response to the constraints of the proposed a “fasting endurance hypothesis” to explain why larger rearing environment. Although early emigrants grow faster, and individuals with higher energy requirements undertake seasonal presumably reach a larger size at smoltification, this life history migrations. The faster-growing (as measured by FL) individu- strategy is not exclusively expressed in the population. Interan- als in Murderers Creek likely had a greater metabolic demand nual variation in marine survival, associated with ocean entry during fall as a result of having invested their energy into length timing differences between early-emigrant and late-emigrant growth rather than lipid storage. During the fall–winter tran- individuals, likely creates asynchronous productivity between sition, we suggest that the faster-growing individuals did not these life histories (Hilborn et al. 2003) and prevents any single have sufficient fasting endurance (Chapman et al. 2011) to re- life history from establishing population-level dominance. Thus, main in Murderers Creek through winter and hence adopted an although emigration choice operates at the individual level, the early-emigrant strategy. presence of both early and late emigrants increases the resilience Early emigration compounded the phenotypic differences of the population to changing environmental conditions. Moni- that existed between early-emigrant and late-emigrant O. mykiss toring should estimate the abundance of both life histories, with at the beginning of fall. The initially larger and faster-growing their relative contribution to the population as a whole being an early emigrants subsequently experienced higher growth during important indicator of viability. Identification and protection of winter. Thus, size differences between groups likely further di- all habitats (both summer and winter rearing habitat) utilized by verged following emigration. Some of the early emigrants from both life histories will also be an important management action. Murderers Creek migrated downstream of our RST. We tracked the migration and winter holding locations of some of these individuals with surgically implanted radio transmitters. Most ACKNOWLEDGMENTS individuals migrated less than 20 km downstream of the RST to winter rearing areas (Tattam, unpublished data). We were not We thank S. White, F. Madrin˜an,´ J. Feldhaus, S. Hep- able to recapture and measure these early emigrants immedi- pell, P. Bayley, J. Davis, B. Kingsley, V. Mueller, J. Togstad, ately prior to smoltification for comparison with late emigrants. J. Silbernagel, D. Myers, N. Weber, B. Tattam, and T. Tattam Nonetheless, our data suggest that early emigrants attained a for field assistance. We thank W. Wilson, J. Schricker, T. Goby, larger size at smoltification, which may facilitate greater marine T. Schultz, R. Lamb, D. Bondurant, T. Hartill, and L. Hewlett survival (Bilton et al. 1982; Ward et al. 1989; Tipping 1997). Fu- for operating the screw trap and J. Neal and T. Unterwegner ture fish tagging should focus on following the potential effect of the Oregon Department of Fish and Wildlife for sharing of increased length on smolt-to-adult survival by recapturing observations and hypotheses about fall emigration in the John tagged individuals at sampling facilities in Columbia River hy- Day basin that instigated this work. The work of I. Tattam., H. dropower dams. Li, and G. Giannico was funded by the U.S. Bureau of Recla- Differential smolt arrival timing at the Columbia River estu- mation, Pacific Northwest Region, through M. Newsom. The ary could be traced back to expression of either early-emigrant involvement of J. Ruzycki was supported by the Bonneville or late-emigrant behavior 5–8 months earlier. Physical condi- Power Administration (Project Number 1998-016-00) through Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 tions and biotic communities in the estuary vary on a daily basis J. Baugher, J. Karnezis, and J. Swan. The PIT tags were pro- (Weitkamp et al. 2012). Likewise, ocean conditions can vary at vided by C. Jordan of the National Oceanic and Atmospheric short time scales and peak marine survival rates can occur at Administration. Also, P. Bayley, J. Dunham, D.L.G. Noakes, different ocean entry times in different years (Lundqvist et al. and two anonymous reviewers provided extensive comments 1994). We hypothesize that fall migration patterns ultimately that improved the manuscript. Reference to trade names does manifest in different smolt-to-adult survival rates between early- not imply endorsement by the United States Geological Sur- emigrant and late-emigrant groups. This may create differential vey, Oregon Cooperative Fishery Research Unit, Oregon State parr-to-adult recruitment between tributaries based on the preva- University, or the Oregon Department of Fish and Wildlife. lence of fall emigration. In the case of our study streams, Mur- derers Creek will have a greater proportion of smolts with early ocean entry than Black Canyon Creek. Early ocean entry may REFERENCES be advantageous during some years and disadvantageous dur- Armstrong, J. D., V. A. Braithwaite, and F. A. Huntingford. 1997. Spatial strate- gies of wild Atlantic Salmon parr: exploration and settlement in unfamiliar ing other years (Muir et al. 2006). Early emigrants (and hence areas. Journal of Animal Ecology 66:203–211. early ocean entry smolts) from the SFJD had higher survival Baxter, C. V.2002. Fish movement and assemblage dynamics in a Pacific North- (measured from the SFJD to adult detection at Bonneville Dam) west riverscape. Doctoral dissertation. Oregon State University, Corvallis. SIZE AND GROWTH RATE INFLUENCE EMIGRATION TIMING 1413

Bilton, H. T., D. F. Alderdice, and J. T. Schnute. 1982. Influence of time and Jonsson, N., B. Jonsson, and L. P. Hansen. 1998. Long-term study of the ecology size at release of juvenile Coho Salmon (Oncorhynchus kisutch) on returns at of wild Atlantic Salmon smolts in a small Norwegian river. Journal of Fish maturity. Canadian Journal of Fisheries and Aquatic Sciences 39:426–447. Biology 52:638–650. Bjornn, T. C. 1971. Trout and salmon movements in two Idaho streams as related Kahler, T. H., P. Roni, and T. P. Quinn. 2001. Summer movement and growth to temperature, food, stream flow, cover, and population density. Transactions of juvenile anadromous salmonids in small western Washington streams. of the American Fisheries Society 100:423–438. Canadian Journal of Fisheries and Aquatic Sciences 58:1947–1956. Bramblett, R. G., M. D. Bryant, B. E. Wright, and R. G. White. 2002. Sea- Keeley, E. R. 2001. Demographic responses to food and space competition by sonal use of small tributary and main-stem habitats by juvenile steelhead, juvenile steelhead trout. Ecology 82:1247–1259. Coho Salmon, and Dolly Varden in a southeastern Alaska drainage basin. Koskela, J., J. Pirhonen, and M. Jobling. 1997. Growth and feeding responses of Transactions of the American Fisheries Society 131:498–506. a hatchery population of Brown Trout (Salmo trutta L.) at low temperatures. Bujold, V., R. A. Cunjak, J. P. Dietrich, and D. A. Courtemanche. 2004. Drifters Ecology of Freshwater Fish 6:116–121. versus residents: assessing size and age differences in Atlantic Salmon Ledgerwood, R. D., B. A. Ryan, E. M. Dawley, E. P. Nunnallee, and J. W. (Salmo salar) fry. Canadian Journal of Fisheries and Aquatic Sciences 61: Ferguson. 2004. A surface trawl to detect migrating juvenile salmonids tagged 273–282. with passive integrated transponder tags. North American Journal of Fisheries CBR (Columbia Basin Research). 2012. Columbia River DART (data ac- Management 24:440–451. cess in real time). CBR, University of Washington, Seattle. Available: Leider, S. A., M. W. Chilcote, and J. J. Loch. 1986. Movement and survival of www.cbr.washington.edu/dart/. (March 2013). presmolt steelhead in a tributary and the main stem of a Washington river. Chapman, B. B., C. Bronmark,¨ J. Å. Nilsson, and L. A. Hansson. 2011. The North American Journal of Fisheries Management 6:526–531. ecology and evolution of partial migration. Oikos 120:1764–1775. Lundqvist, H., S. McKinnell, H. Fangstam,¨ and I. Berglund. 1994. The effect Chapman, D. W. 1962. Aggressive behavior in juvenile Coho Salmon as a cause of time, size and sex on recapture rates and yield after river releases of Salmo of emigration. Journal of the Fisheries Research Board of Canada 19:1047– salar smolts. Aquaculture 121:245–257. 1080. McMillan, J. R., J. B. Dunham, G. H. Reeves, J. S. Mills, and C. E. Jordan. 2012. Cucherousset, J., D. Ombredane, K. Charles, F. Marchand, and J. L. Bagliniere.` Individual condition and stream temperature influence early maturation of 2005. A continuum of life history tactics in a Brown Trout (Salmo trutta) Rainbow and steelhead trout, Oncorhynchus mykiss. Environmental Biology population. Canadian Journal of Fisheries and Aquatic Sciences 62:1600– of Fishes 93:343–355. 1610. Metcalfe, N. B. 1998. The interaction between behavior and physiology Cunjak, R. A. 1996. Winter habitat of selected stream fishes and potential in determining life history patterns in Atlantic Salmon (Salmo salar). impacts from land-use activity. Canadian Journal of Fisheries and Aquatic Canadian Journal of Fisheries and Aquatic Sciences 55(Supplement 1): Sciences 53(Supplement 1):267–282. 93–103. Frissell, C. A., W. J. Liss, C. E. Warren, and M. D. Hurley. 1986. A hierarchical Metcalfe, N. B., F. A. Huntingford, and J. E. Thorpe. 1988. Feeding intensity, framework for stream habitat classification: viewing streams in a watershed growth rates, and the establishment of life-history patterns in juvenile Atlantic context. Environmental Management 10:199–214. Salmon Salmo salar. Journal of Animal Ecology 57:463–474. Giannico, G. R., and M. C. Healey. 1998. Effects of flow and food on winter Metcalfe, N. B., P. J. Wright, and J. E. Thorpe. 1992. Relationships between movements of juvenile Coho Salmon. Transactions of the American Fisheries social status, otolith size at first feeding and subsequent growth in Atlantic Society 127:645–651. Salmon (Salmo salar). Journal of Animal Ecology 61:585–589. Gowan, C., and K. D. Fausch. 2002. Why do foraging stream salmonids move Mills, J. S., J. B. Dunham, G. H. Reeves, J. R. McMillan, C. E. Zimmerman, during summer? Environmental Biology of Fishes 64:139–153. and C. E. Jordan. 2012. Variability in expression of anadromy by female On- Gowan, C., M. K. Young, K. D. Fausch, and S. C. Riley. 1994. Restricted corhynchus mykiss within a river network. Environmental Biology of Fishes movement in resident stream salmonids: a paradigm lost? Canadian Journal 93:505–517. of Fisheries and Aquatic Sciences 51:2626–2637. Morgan, I. J., and N. B. Metcalfe. 2001. The influence of energetic require- Grant, J. W. A., and D. L. G. Noakes. 1987. Movers and stayers: foraging tactics ments on the preferred temperature of overwintering juvenile Atlantic Salmon of young-of-the-year Brook Charr, Salvelinus fontinalis. Journal of Animal (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences 58:762– Ecology 56:1001–1013. 768. Gregory, S. V., F. J. Swanson, W. A. McKee, and K. W. Cummins. 1991. An Morinville, G. R., and J. B. Rasmussen. 2003. Early juvenile bioener- ecosystem perspective of riparian zones: focus on links between land and getic differences between anadromous and resident Brook Trout (Salveli- water. BioScience 41:540–551. nus fontinalis). Canadian Journal of Fisheries and Aquatic Sciences 60:

Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 Griffiths, J. R., D. E. Schindler, and L. W. Seeb. 2013. How stock of origin 401–410. affects performance of individuals across a meta-ecosystem: an example Muir, W. D., D. M. Marsh, B. P. Sandford, S. G. Smith, and J. G. Williams. from Sockeye Salmon. PLoS (Public Library of Science) ONE [online serial] 2006. Post-hydropower system delayed mortality of transported Snake River 8(3):e58584. stream-type Chinook Salmon: unraveling the mystery. Transactions of the Higgins, P. J. 1985. Metabolic differences between Atlantic Salmon (Salmo American Fisheries Society 135:1523–1534. salar) parr and smolts. Aquaculture 45:33–53. Murphy, M. H., M. J. Connerton, and D. J. Stewart. 2006. Evaluation of winter Hilborn, R., T. P.Quinn, D. E. Schindler, and D. E. Rogers. 2003. Biocomplexity severity on growth of young-of-the-year Atlantic Salmon. Transactions of the and fisheries sustainability. Proceedings of the National Academy of Sciences American Fisheries Society 135:420–430. of the USA 100:6564–6568. Northcote, T. G. 1992. Migration and residency in stream salmonids: some Huntingford, F. A., N. B. Metcalfe, and J. E. Thorpe. 1988. Choice of feeding ecological considerations and evolutionary consequences. Nordic Journal of station in Atlantic Salmon, Salmo salar, parr: effects of predation risk, season Freshwater Research 67:5–17. and life history strategy. Journal of Fish Biology 33:917–924. Paulsen, C. M., and T. R. Fisher. 2001. Statistical relationship between parr- Imre, I., J. W. A. Grant, and E. R. Keeley. 2004. The effect of food abun- to-smolt survival of Snake River spring–summer Chinook Salmon and in- dance on territory size and population density of juvenile steelhead trout dices of land use. Transactions of the American Fisheries Society 130: (Oncorhynchus mykiss). Oecologia 138:371–378. 347–358. Jonsson, B., and N. Jonsson. 1993. Partial migration: niche shift versus Prentice, E. F., T. A. Flagg, C. S. McCutcheon, D. F. Brastow, and D. C. sexual maturation in fishes. Reviews in Fish Biology and Fisheries 3: Cross. 1990. Equipment, methods, and an automated data-entry station for 348–365. PIT tagging. Pages 335–340 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, 1414 TATTAM ET AL.

D. B. Jester Jr., E. D. Prince, and G. A. Winans, editors. Fish-marking tech- Thorpe, J. E. 1987a. Environmental regulation of growth patterns in juvenile niques. American Fisheries Society, Symposium 7, Bethesda, Maryland. Atlantic Salmon. Pages 463–474 in R. C. Summerfelt and G. E. Hall, editors. PTSC (PIT Tag Steering Committee). 1999. PIT tag marking procedures man- Age and growth of fish. Iowa State University Press, Ames. ual. Columbia Basin Fish and Wildlife Authority, PTSC, Portland, Oregon. Thorpe, J. E. 1987b. Smolting versus residency: developmental conflict in Available: php.ptagis.org/wiki/images/e/ed/MPM.pdf. (March 2013). salmonids. Pages 244–252 in M. J. Dadswell, R. J. Klauda, C. M. Moffitt, Riddell, B. E., and W. C. Leggett. 1981. Evidence of an adaptive basis for R. L. Saunders, R. A. Rulifson, and J. E. Cooper, editors. Common strate- geographic variation in body morphology and time of downstream migration gies of anadromous and catadromous fishes. American Fisheries Society, of juvenile Atlantic Salmon (Salmo salar). Canadian Journal of Fisheries and Symposium 1, Bethesda, Maryland. Aquatic Sciences 38:308–320. Thorpe, J. E., and N. B. Metcalfe. 1998. Is smolting a positive or a negative Rodr´ıguez, M. A. 2002. Restricted movement in stream fish: the paradigm is developmental decision? Aquaculture 168:95–103. incomplete, not lost. Ecology 83:1–13. Tipping, J. M. 1997. Effect of smolt length at release on adult returns of hatchery- Roni, P., T. Bennett, R. Holland, G. Pess, K. Hanson, R. Moses, M. reared winter steelhead. Progressive Fish-Culturist 59:310–311. McHenry, W. Ehinger, and J. Walter. 2012. Factors affecting migration tim- Verhoeven, K. J. F., K. L. Simonsen, and L. M. McIntyre. 2005. Implementing ing, growth, and survival of juvenile Coho Salmon in two coastal Wash- false discovery rate control: increasing your power. Oikos 108:643–647. ington watersheds. Transactions of the American Fisheries Society 141: Ward, B. R., P. A. Slaney, A. R. Facchin, and R. W. Land. 1989. Size-biased 890–906. survival in steelhead trout (Oncorhynchus mykiss): back-calculated lengths Roni, P., and T. P. Quinn. 2001. Effects of wood placement on movements of from adults’ scales compared to migrating smolts at the Keogh River, British trout and juvenile Coho Salmon in natural and artificial stream channels. Columbia. Canadian Journal of Fisheries and Aquatic Sciences 46:1853– Transactions of the American Fisheries Society 130:675–685. 1858. Schultz, T. L., W. H. Wilson, J. R. Ruzycki, R. Carmichael, J. Schricker, and D. Weitkamp, L. A., P.J. Bentley, and M. N. C. Litz. 2012. Seasonal and interannual P.Bondurant. 2006. Escapement and productivity of spring Chinook and sum- variation in juvenile salmonids and associated fish assemblage in open waters mer steelhead in the John Day River basin. Annual Report to the Bonneville of the lower Columbia River estuary. U.S. National Marine Fisheries Service Power Administration, Project 1998-016-00, Portland, Oregon. Available: Fishery Bulletin 110:426–450. pisces.bpa.gov/release/documents/documentviewer.aspx?doc=00005840-4. Whitesel, T. A. 1993. Comparison of juvenile Atlantic Salmon (Salmo salar) (March 2013). reared in a hatchery and introduced into a stream: a two-size-threshold model Smith, R. W., and J. S. Griffith. 1994. Survival of Rainbow Trout during their of smoltification. Canadian Special Publication of Fisheries and Aquatic first winter in the Henrys Fork of the Snake River, Idaho. Transactions of the Sciences 118:239–247. American Fisheries Society 123:747–756. Wilson, W.H., J. Schricker, J. R. Ruzycki, and R. Carmichael. 2008. Productivity Steingr´ımsson, S. O., and J. W. A. Grant. 2003. Patterns and correlates of of spring Chinook Salmon and summer steelhead in the John Day River movement and site fidelity in individually tagged young-of-the-year Atlantic basin. Annual Technical Report to the Bonneville Power Administration, Salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences Project 1998-016-00, Portland, Oregon. Available: pisces.bpa.gov/release/ 60:193–202. documents/documentviewer.aspx?doc=P106434. (March 2013). Tattam, I. A., J. R. Ruzycki, P. B. Bayley, H. W. Li, and G. R. Giannico. Zydlewski, G. B., A. Haro, and S. D. McCormick. 2005. Evidence for cu- 2013. The influence of release strategy and migration history on capture rate mulative temperature as an initiating and terminating factor in downstream of Oncorhynchus mykiss in a rotary screw trap. North American Journal of migratory behavior of Atlantic Salmon (Salmo salar) smolts. Canadian Jour- Fisheries Management 33:237–244. nal of Fisheries and Aquatic Sciences 62:68–78. Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:37 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 Using Seasonal Variation in Otolith Microchemical Composition to Indicate Largemouth Bass and Southern Flounder Residency Patterns across an Estuarine Salinity Gradient Troy M. Farmer a c , Dennis R. DeVries a , Russell A. Wright a & Joel E. Gagnon b a Department of Fisheries and Allied Aquacultures , Auburn University , 203 Swingle Hall, Auburn , Alabama , 36849 , USA b Great Lakes Institute for Environmental Research , University of Windsor , 401 Sunset Avenue, Windsor , Ontario , N9B 3P4 , Canada c Aquatic Ecology Laboratory , Ohio State University , 1314 Kinnear Road, Columbus , Ohio , 43212 , USA Published online: 06 Sep 2013.

To cite this article: Troy M. Farmer , Dennis R. DeVries , Russell A. Wright & Joel E. Gagnon (2013) Using Seasonal Variation in Otolith Microchemical Composition to Indicate Largemouth Bass and Southern Flounder Residency Patterns across an Estuarine Salinity Gradient, Transactions of the American Fisheries Society, 142:5, 1415-1429, DOI: 10.1080/00028487.2013.806348 To link to this article: http://dx.doi.org/10.1080/00028487.2013.806348

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1415–1429, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.806348

ARTICLE

Using Seasonal Variation in Otolith Microchemical Composition to Indicate Largemouth Bass and Southern Flounder Residency Patterns across an Estuarine Salinity Gradient

Troy M. Farmer,*1 Dennis R. DeVries, and Russell A. Wright Department of Fisheries and Allied Aquacultures, Auburn University, 203 Swingle Hall, Auburn, Alabama 36849, USA Joel E. Gagnon Great Lakes Institute for Environmental Research, University of Windsor, 401 Sunset Avenue, Windsor, Ontario N9B 3P4, Canada

Abstract Estuaries are transitional zones where salinity largely controls the distribution of both freshwater and marine species. However, the degree to which freshwater and marine species use these variable habitats as year-round residents or transient migrants is largely unknown. We used otolith strontium : calcium ratios (Sr : Ca) as a marker for salinity exposure to reconstruct lifetime salinity exposure plots for individual Largemouth Bass Micropterus salmoides and Southern Flounder Paralichthys lethostigma collected across a seasonally variable estuarine salinity gradient. Initially, we determined that a definable relationship existed between salinity, water chemistry, and otolith Sr : Ca for both species. We then used otolith Sr : Ca profiles to indicate lifetime salinity exposure and subsequently classified each fish as either a freshwater resident, transient (combination of freshwater and estuarine signals in otolith), or estuarine resident. For Southern Flounder we also used Sr : Ca profiles from the otolith core to classify each individual as having either a freshwater core or estuarine or marine core. For Largemouth Bass, most (88%) individuals in the lower estuary were estuarine residents, whereas most (77%) individuals in the upper estuary were freshwater residents. These data support the hypothesis that adult Largemouth Bass in lower portions of the estuary do not migrate to avoid salinity, but rather remain in lower portions of the estuary throughout life. For Southern Flounder, 45% of individuals had estuarine or marine core Sr : Ca signals, while 55% had freshwater Sr : Ca core signals. Combining core and residency classification patterns revealed that three patterns described 95% of the Southern Flounder collected: (1) freshwater core and freshwater resident (16%), (2) freshwater core and estuarine resident (37%), and (3) estuarine or marine core and estuarine resident (42%). These data demonstrate that both Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 tidal freshwater and low-salinity estuarine habitats are important nursery areas for Southern Flounder.

As transitional zones between freshwater and marine en- 2004). Despite this, the degree to which salinity gradients affect vironments, estuaries have highly variable and extreme abi- habitat use and movement of freshwater, estuarine, and ma- otic conditions (Day et al. 1989). Variation in salinity has long rine fish species is still not well understood. Specifically, for been recognized as a factor controlling estuarine fish distribu- freshwater species there is little published information describ- tions, through its effect on behavior and physiological processes ing the degree to which these species are residents or seasonal (Bulger 1993; Peterson and Meador 1994; Paperno and Brodie transients in low-salinity, oligohaline (salinity 0.5–5‰), and

*Corresponding author: [email protected] 1Present address: Aquatic Ecology Laboratory, Ohio State University, 1314 Kinnear Road, Columbus, Ohio 43212, USA. Received September 13, 2012; accepted May 13, 2013 Published online September 6, 2013 1415 1416 FARMER ET AL.

mesohaline (salinity 5–18‰) estuarine habitats (Peterson and (Nelson 2006). While the use of oligohaline and mesohaline Ross 1991). For resident estuarine and marine species, there is waters by paralichthyid species is well documented (Rogers limited knowledge regarding the preferred salinity range and et al. 1984; Rozas and Hackney 1984; Glass et al. 2008; Nanez-˜ associated habitat of juveniles and subadults using estuaries as James et al. 2009), recent work indicates age-0 and juvenile nursery habitat (Glass et al. 2008). flatfish species may use, (Lowe et al. 2011) or even prefer (Zuc- Freshwater fish species are a major component of tidal fresh- chetta et al. 2010; Lowe et al. 2011), tidal freshwater habitat. water and oligohaline portions of estuaries (Rogers et al. 1984; Specifically, Southern Flounder Paralichthys lethostigma is Rozas and Hackney 1984). In these complex habitats, distri- a paralichthyid that uses a variety of estuarine habitats during its butions of freshwater species are believed to shift with chang- life stages (Stokes 1977; Burke et al. 1991; Fischer and Thomp- ing salinity (Felley 1987), increasing in abundance downstream son 2004; Glass et al. 2008) and is a species for which tidal when freshwater conditions persist and then retreating upstream freshwater areas may provide suitable nursery habitat (Reichert when salinity increases (Felley 1987). For example, centrarchids and van der Veer 1991; Allen and Baltz 1997; Lowe et al. 2011). are particularly abundant in the upstream, freshwater portions While tolerance of Southern Flounder to freshwater or reduced of estuaries in eastern North America (Felley 1987; Rogers salinity depends strongly on life stage (e.g., eggs and larvae ex- et al. 1984; Rozas and Hackney 1984) while their distribution perienced 100% mortality at salinity ≤10‰; Smith et al. 1999), in downstream portions of estuaries is thought to be limited by juvenile growth did not differ between freshwater and higher salinity (Felley 1987), which affects osmoregulation (Meador salinities in the laboratory (Daniels and Borski 1998), indicat- and Kelso 1990a; Susanto and Peterson 1996), reproduction ing the potential for freshwater areas to serve as nursery habitat (Tebo and McCoy 1964), and growth (Meador and Kelso 1990b; for this species. Further, recent otolith microchemical analysis Peterson 1988; Glover et al. 2013). indicated that age-0 Southern Flounder do, in fact, use tidal Quantifying behavioral responses of centrarchids to salin- freshwater habitat and that some individuals use this habitat ex- ity has resulted in mixed findings. In laboratory experiments, clusively during the first year of life (Lowe et al. 2011). While estuarine Largemouth Bass Micropterus salmoides preferred older juvenile and adults have also been collected in freshwater low-salinity waters (≤3‰; Meador and Kelso 1989) and did habitats (Powell and Schwartz 1977; Rogers et al. 1984; Rozas not survive extended durations in elevated salinities (≥12‰; and Hackney 1984), the duration of residence in freshwater por- Meador and Kelso 1990a; Susanto and Peterson 1996). In the tions of the estuary remains unknown for juvenile and subadult field, anecdotal evidence suggests that Largemouth Bass may Southern Flounder (Lowe et al. 2011). migrate to freshwater areas to avoid seasonally elevated salin- Using otolith microchemical analysis of adult Largemouth ity (Swingle and Bland 1974; Meador and Kelso 1989), while Bass and subadult Southern Flounder, we characterized lifetime telemetry of individually tagged Largemouth Bass suggests they salinity experience of individuals across a seasonal salinity gra- move little in response to elevated salinity (Norris et al. 2005). dient. In estuarine and marine studies, the relationship between Recent research indicates age-0 Largemouth Bass can withstand the strontium : calcium ratio (Sr : Ca) in fish otoliths and salin- elevated salinity (Lowe et al. 2009) and experience increased ity has been a useful tool for recreating individual exposure to growth in low-salinity areas relative to freshwater areas, proba- salinity; however, the relationship between ambient water Sr : bly due to increased quality and quantity of food consumption Ca and salinity is complex and likely differs among estuaries (Peer et al. 2006). Given these mixed findings, the question of (Kraus and Secor 2004). While the relationship between otolith whether adult Largemouth Bass use oligohaline and mesohaline and ambient water Sr : Ca is consistently positive (Elsdon et al. habitat as year-round residents or seasonal transients remains 2008), incorporation of elements into the otolith may be af- unresolved. fected by temperature (Bath et al. 2000), physiology (Kalish Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 A second major component of tidal freshwater estuarine fish 1989), and presence of other competing elements of similar va- communities includes the juveniles and subadults of euryhaline lence (de Vries et al. 2005). Hence, we initially quantified the and marine species, spawned in offshore coastal waters, which relationship between salinity, water, and otolith microchemistry subsequently use this habitat as a nursery (Rogers et al. 1984; to validate the use of otolith Sr : Ca patterns to recreate previous Rozas and Hackney 1984). While the use of high-salinity estuar- salinity experience of individuals (Kraus and Secor 2004). ine (i.e., seagrass and salt marsh) nursery habitats has garnered The first objective of this study was to determine if Sr : much research and management effort (Beck et al. 2001), the Ca in the otoliths of Largemouth Bass and Southern Flounder use of tidal freshwater portions of estuaries has not received as could serve as a marker of salinity exposure history. If otolith much attention (Rogers et al. 1984). Subsequently, relatively Sr : Ca served as a marker of salinity exposure, then our next little is known regarding salinity preferences and effects of objective was to use this information to indicate and compare salinity fluctuation on the habitat use by euryhaline and ma- patterns of resident or migratory behavior by adult Largemouth rine species in the tidal freshwater portions of estuaries (Glass Bass and subadult Southern Flounder across a salinity gradient. et al. 2008). For example, paralichthyids (large-tooth flounders) Finally, our last objective was to quantify the effect of freshwater commonly occur across salinity gradients in tropical and tem- river discharge on otolith microchemical composition, thereby perate estuaries along the Atlantic, Indian, and Pacific coasts understanding the degree to which interannual variability in OTOLITH-BASED FISH RESIDENCY PATTERNS 1417

freshwater river discharge may affect the use of otolith Sr : Ca in the Delta in 2005 and 2006. Samples were collected at the to indicate resident or migratory behavior. same sites where fish were collected and, to expand the range of salinities sampled, at two additional sites: Crab Creek and Mullet Point (30◦ 24 59 N, 87◦ 54 32 W; not shown in Fig- METHODS ure 1). Water samples were collected in June (N = 4), August Study Site (N = 4), and October (N = 10) of 2005 and in March (N = This study was conducted in the Mobile–Tensaw River 6) and October (N = 6) of 2006; all sites were not sampled in Delta, Alabama (hereafter referred to as the “Delta”; Figure 1). each month. Samples were taken 1 m below the surface with The Delta drains the Mobile River basin, a large (about a Van Dorn bottle, filtered through a 0.45-µm glass-fiber filter, 114,000 km2), diverse watershed across four physiographic acidified with 1.25 mL of reagent grade nitric acid in the field provinces containing sandstone, shale, and carbonate bedrock (USEPA 1996), and stored in acid-washed polypropylene bot- materials (Fenneman and Johnson 1946). It begins at the tles (125 mL). Samples were transported to the laboratory on ice confluence of the Alabama and Tombigbee rivers, ends at and refrigerated until being shipped to the analytical laboratory the head of Mobile Bay, and is approximately 55 km long for determination of trace metal content. Salinity, temperature, and 10–15 km wide (Swingle et al. 1966). Based on mean and dissolved oxygen profiles were measured (YSI 30 m) in 1-m daily discharge (1850 m3/s), this river system ranks as the increments from the surface to the bottom at all sampling sites fourth largest river system in the USA (Morisawa 1968). and dates. Also, data loggers (Solinist Levelogger 3000R) were Annual variation in flow through the Delta dictates the timing placed at 1 m below the average low tide mark and recorded and magnitude of salinity intrusion. During periods of low salinity readings every 0.5 h at two sites (Figure 1). flow in summer and fall, the Delta can experience saltwater intrusion extending north to the confluence of the Alabama and Otolith Preparation Tombigbee rivers (approximately 8 km north of the study area Sagittal otoliths were removed from the fish with Teflon- shown in Figure 1; Braun and Neugarten 2005). Collection coated forceps and placed in 30% H2O2 for 30 s for cleaning. sites were established to incorporate predominant habitat types After rinsing in deionized, ultra-filtered water (Fisher Scientific, across the upstream to downstream seasonal salinity gradient. Canada), otoliths were dried and stored in individual polyethy- lene vials. For Largemouth Bass, a single otolith was selected at Fish Collection random for elemental analysis; however, for Southern Flounder, Largemouth Bass and Southern Flounder were collected from the right otolith was always used due to asymmetry between 2005 to 2006 at seven sites across an upstream to downstream right and left otoliths in Paralichthys species (Sipe and Chitten- seasonal salinity gradient (Figure 1). Three sites were grouped den 2001; Fischer and Thompson 2004). Otoliths were mounted as upstream, one as middle, and three as downstream. Large- in epoxy resin and a transverse section through the core was re- mouth Bass were collected prior to (March–May) and during moved with a low-speed diamond blade Isomet saw. Transverse (August–October) elevated salinity periods in 2005 and 2006 core sections were mounted on petrographic slides with thermo- with pulsed-DC electrofishing (7.5 GPP, Smith-Root, Vancou- plastic cement and polished with 320-grit, 600-grit, and 800-grit ver, Washington). Southern Flounder were collected at four sites paper until the otolith surface was smooth and the core region (DL, GI, BM, DB) across the upstream to downstream seasonal was exposed. After polishing, a digital image of each otolith salinity gradient during fall 2005 and 2006. In addition to elec- was captured for measurement and age determination. trofishing, trawling was conducted at Southern Flounder sites Methods for otolith cleaning are detailed in Ludsin et al. in fall 2005 using a 4.9-m headrope otter trawl (6.4-mm bar (2006) and Lowe et al. (2009). Briefly, otolith sections were Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 mesh wings and body and 3.2-mm bar mesh cod end). Because mounted onto petrographic slides and sonicated with ultrapure trawling was less efficient than electrofishing (T. M. Farmer, MilliQ water in an ULTRAsonik cleaner (model 57X; Ney Den- unpublished data), trawling was not conducted in fall 2006. tal, Inc., Bloomfield Connecticut) for 10 min. Otoliths were then During each collection period, up to 10 Largemouth Bass and rinsed three times with MilliQ water and allowed to air dry for 10 Southern Flounder per site were retained for otolith micro- 24–48 h in a Class-100 clean room under a laminar-flow hood. chemical analysis. Individuals retained reflected the size range collected in the field. For Southern Flounder, low catch rates at Elemental Analysis and Data Processing some sites prevented us from processing 10 individuals per site. Strontium and calcium concentration in the otoliths from In all, otoliths from 204 Largemouth Bass (150–500 mm TL; both species were quantified using a laser ablation inductively ages 1–11) and 38 Southern Flounder (200–500 mm TL; ages coupled plasma mass spectrometry (LA-ICPMS) system that 1–3) were processed for microchemical analysis. uses a high-energy (up to 2.5 mJ/pulse), ultrafast (130 fs) laser (Integra-C by Quantronics, East Setauket, New York;) Habitat and Water Chemistry and a Thermo-Elemental X7 quadrupole ICPMS located at Water samples were collected for determination of trace ele- the Great Lakes Institute for Environmental Research in Wind- ment concentration prior to and during elevated salinity periods sor, Ontario. For this study, the laser operating conditions were 1418 FARMER ET AL. Downloaded by [Department Of Fisheries] at 21:37 27 October 2013

FIGURE 1. Map of the Mobile–Tensaw River Delta, Alabama, showing the seven sampling site locations by name and abbreviation and the locations of salinity loggers (dark cross) and one additional water chemistry site (crab creek; dark star). Mean daily salinity (calculated from readings taken every 0.5 h at 1 m depth by loggers) is shown in the bottom panels for (A) Gravine Island, and (B) D’Olive Bay. Salinity values at 1 m recorded during monthly fish collections are presented for (C) upstream and middle, and (D) downstream sites. OTOLITH-BASED FISH RESIDENCY PATTERNS 1419

pulse frequency = 100 Hz, energy = 0.075 mJ/pulse, and spot reflect water Sr : Ca over the lifetime of Largemouth Bass and size = 24–29 µm. Laser ablation microsampling was conducted Southern Flounder. as straight-line transects from the core to either the dorsal or ven- tral edge of the otolith, in a direction parallel to the sulcus (i.e., Quantifying Lifetime Salinity Exposure to provide the trace element composition for the entire life of Largemouth Bass.—Sr : Ca was plotted across the otolith the individual). Ablation speed for all otoliths was consistently (i.e., from the core to the edge) to visually compare profiles of set at 5 µm/s via a computer-controlled stage. Calcium was used lifetime salinity experience for individual Largemouth Bass to as the interval standard against which Sr concentrations were expected salinity regimes for a given region. Expected salin- estimated (Campana 1999). A synthetic glass reference stan- ity regimes were determined from field measurement of salin- dard (National Institute of Standards and Technology 610) with ity in this and previous studies in the Delta (Peer et al. 2006; known elemental concentrations for Sr and Ca was analyzed Lowe et al. 2009, 2011; Norris et al. 2010; Glover et al. 2013). twice before and twice after every eight otolith ablations to pro- Individual Largemouth Bass were classified as either freshwa- vide an external standard, measure for instrumental drift, and ter resident, estuarine resident, or transient, based on lifetime obtain estimates of precision (CV) for elemental measurements. salinity exposure plots. Based on a previously determined rela- The background (Ar) carrier gas was analyzed for 60 s prior to tionship between Largemouth Bass otolith Sr : Ca and salinity each otolith ablation to determine a limit of detection (LOD) for (Whitledge et al. 2007), we used a threshold of 2.5 mmol/mol each element from each otolith using the formula of Longerich Sr : Ca to indicate exposure to salinity >2‰ for Largemouth et al. (1996): LOD = mean background count rate + 3(SD). Bass, which we classified as freshwater residents if otolith For both Largemouth Bass and Southern Flounder, 100% of Sr : Ca was consistently <2.5 mmol/mol across all years of otoliths were above LOD levels for Sr and Ca. Additionally, life. Largemouth Bass with a distinct Sr : Ca peak ≥2.5 in both elements were measured with high precision (CV <10%; each year of life were classified as estuarine residents. Individ- Ludsin et al. 2006) in 100% of the otoliths from both species. uals with a combination of freshwater and estuarine signatures Trace metal analysis of water samples was conducted by across years, indicating changing resident status, were classi- solution-mode ICPMS, following a 100-fold dilution with an fied as transient. We scaled transect Sr : Ca data to years using internal standard solution of 1% HNO3 spiked with beryllium, the distance between each annulus and the speed of an individ- indium, and thallium to correct for matrix and instrumental drift ual laser ablation such that patterns of salinity experience were effects on an individual sample basis. Calibration standards for comparable among all individuals. To do so, we assumed that multiple elements were measured before, during, and after each the last Sr : Ca measurement at the otolith edge corresponded sample to provide a measure of analytical precision and accuracy to the date of capture and that the first Sr : Ca measurement at (Lowe et al. 2009). the center of each otolith core corresponded to the hatch date (assumed to be April 1 for Largemouth Bass; DeVries and Frie Relating Salinity, Water and Otolith Microchemistry 1996; Peer et al. 2006). We used nonlinear regression to determine the relationship From the above classifications, we created contingency tables between water Sr : Ca and salinity. To verify that Largemouth and used Fisher’s exact test (R version 2.12.2; R Development Bass and Southern Flounder otolith Sr : Ca generally track water Core Team 2011; Ramsey and Schafer 2002) to determine if Sr : Ca, we regressed water Sr : Ca against otolith edge Sr : Ca to salinity exposure was independent of site and season for Large- characterize recent elemental incorporation into the otolith (the mouth Bass. We tested the null hypothesis that Largemouth mean of all Sr : Ca measurements across the last approximate Bass classifications did not differ between season of collection 20 µm of the otolith, which we assumed to be approximately (i.e., spring or fall) using three 6 × 3 contingency tables. Cat- Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 30 d, based on otolith growth rates; Farmer 2008). Because egories in these tables were site (rows) and season (columns); a daily salinity fluctuations are highly variable (Figure 1A, B), separate table was constructed for each classification type (i.e., about 20-d time lags exist for otolith Sr : Ca to equilibrate freshwater residents, transients, and estuarine residents). If sea- with water Sr : Ca (Elsdon and Gillanders 2005; Lowe et al. son had no effect on the distributions of classifications across 2009) and because exact habitat use and movement pattern of sites, then classifications were combined across seasons for each field collected individuals used in this analysis are unknown, we site. Three new 6 × 3 contingency tables were then constructed did not intend for this method to define the exact relationship to test the null hypothesis that distributions of each classification between water and otolith Sr : Ca. Instead, this method was type was independent of site of collection for Largemouth Bass. used to validate the general principle that otolith Sr : Ca reflects Categories in these tables were site (rows) and observed and large-scale, seasonal patterns in ambient water Sr : Ca in the expected classifications (columns), where expected counts for Delta. Relationships between water Sr : Ca and otolith edge Sr : each site were the total proportion for each classification type Ca were determined for Largemouth Bass (2005: N = 18; 2006: multiplied by the total number of individuals at a given site; a N = 40) and Southern Flounder (2005: N = 11) collected in fall separate table was constructed for each classification type. simultaneously with water samples. If the otolith edge Sr : Ca Southern Flounder.—Because we were interested in both the ratios were related to water concentrations over a wide range early life habitat use (i.e., approximate salinity, indicated by of ages, we could assume that otolith Sr : Ca would generally Sr : Ca, during early life) and the lifetime residency of subadult 1420 FARMER ET AL.

Southern Flounder, we created classifications for two regions of and Tombigbee rivers (gage 02469761) to investigate whether Southern Flounder otoliths. We examined Sr : Ca profiles across annual differences in Sr : Ca were related to flow, which is an the first 100 µm of the otolith transect, to classify early life habi- important factor controlling salinity in the Delta (Braun and tat use as freshwater or estuarine/marine (Lowe et al. 2011). Re- Neugarten 2005). Only Largemouth Bass from the most recent cent research indicates that Southern Flounder sagittal otoliths year-classes (2000–2006; N = 186) were used in this analysis are approximately 25 µm at the initiation of metamorphosis, due to low sample sizes of fish from older year-classes. grow rapidly through metamorphosis, tripling in size, and are approximately 100 µm at the completion of metamorphosis RESULTS (Schreiber et al. 2010). Therefore our early life Sr : Ca pro- file likely spans the early life stages of embryonic development, Relating Salinity, Water, and Otolith Microchemistry hatching, pelagic larval growth, metamorphosis, and settlement. Salinity, measured by continuous data loggers and discrete We used a Sr : Ca threshold of 1.71 mmol/mol (analogous to monthly measurements (Figure 1A–D), was consistently <2‰ 1,500 mg/L Sr) to indicate exposure to salinity >2‰ and sub- at upstream sites throughout 2005 and 2006. At downstream sequently classified individuals with Sr : Ca <1.71 mmol/mol sites, salinity was <2‰ during winter and spring, but was as having freshwater habitat use and individuals with core Sr : elevated (≥2‰) from August–December in 2005 and June– Ca ≥1.71 mmol/mol as having estuarine/marine habitat use. A December in 2006. While salinity varied greatly across sampling Sr : Ca threshold of 1.71 mmol/mol has been used previously sites, temperature and dissolved oxygen were similar across sites for Southern Flounder (Lowe et al. 2011) and is similar to val- (Farmer 2008). ues used to indicate residence in brackish waters for European In water samples, Sr : Ca was positively related to ambient Flounder Platichthys flesus (1,300–1,700 mg/L Sr, Morais et al. salinity (Figure 2). Water Sr : Ca was also positively related to 2011). However, we should note that our threshold Sr : Ca value otolith edge Sr : Ca for both species across years (Figure 3). for Southern Flounder has not been validated by controlled labo- Furthermore, Largemouth Bass otolith edge Sr : Ca was not ratory experiments and, therefore, still represents an assumption related to age (P = 0.8). Because Largemouth Bass spanning a of this study. wide range of ages (1–7) were used for otolith edge analysis, Sr : To determine lifetime residency patterns, we used the same Ca would be expected to reflect water concentrations throughout approach used for Largemouth Bass (i.e., other than a threshold their adult lifetimes. Consequently, Sr : Ca in otoliths of both value of 1.71 mmol/mol) in which each individual Southern Largemouth Bass and Southern Flounder was used to indicate Flounder was classified as either freshwater resident, estuarine salinity exposure. resident, or transient, based on lifetime salinity exposure plots. Scaling of Sr : Ca to years was conducted as for Largemouth Quantifying Lifetime Salinity Exposure Bass with the exception that hatch date was assumed to be Largemouth Bass.—Lifetime Sr : Ca profiles in Large- January 1 for Southern Flounder (Fitzhugh et al. 1996; Glass mouth Bass otoliths displayed two consistently different pat- et al. 2008). terns across the Delta. Individuals collected from upstream, From the above classifications, we created contingency tables and used Fisher’s exact test (R version 2.12.2; R Development Core Team 2011) to determine if lifetime salinity exposure and early-life habitat use were independent of site of collection for Southern Flounder. To test the null hypothesis that early-life habitat use was independent of site of collection, we combined Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 all flounder collected into a 4 × 2 contingency table with site (rows) and early-life habitat use classification (columns) as cate- gories. To test the null hypothesis that lifetime residency patterns were independent of site of collection, three new 4 × 2 contin- gency tables were constructed. Categories in these tables were site (rows) and observed and expected classifications (columns); a separate table was constructed for each classification type.

Assessing the Influence of Freshwater Input Finally, for upstream, middle, and downstream sites, we re- gressed annual geometric mean Sr : Ca from only the age-0 portion (i.e., Sr : Ca measured from the core to the first an- FIGURE 2. Sr : Ca ratios in water samples collected from the Mobile–Tensaw nulus) of Largemouth Bass otoliths against annual mean daily River Delta, Alabama, regressed against salinity measured at 1 m at the time of river discharge. Mean daily river discharge was summed from collection. Samples were collected in June, August, and October of 2005 and in U.S. Geological Survey gages on the Alabama (gage 02428400) March and October of 2006. OTOLITH-BASED FISH RESIDENCY PATTERNS 1421

TABLE 1. Largemouth Bass residency classification by collection site and location (given in parentheses as upstream [US], middle [MD], and downstream [DS]) determined from otolith Sr : Ca transects for individuals (N) collected in the spring and fall 2005–2006 in the Mobile–Tensaw River Delta, Alabama. Values >50% are in bold italics. Distributions of both freshwater and estuarine residents were not independent of collection site (Fisher’s exact test; P < 0.001).

Freshwater Estuarine resident Transient resident Site Site (location) N % N % N % totals TN (US) 22 79 2741428 ML (US) 28 76 9240037 GI (MD) 8 21 15 38 16 41 39 BB (DS) 0 0 4 21 15 79 19 BM (DS) 1 3 5 14 30 83 36 DB(DS)12124396 45

These patterns caused the majority of individuals collected at downstream sites to be classified as estuarine residents, while the majority of individuals at upstream sites were classified as freshwater residents (Table 1). Overall, freshwater and estuarine residents accounted for 82% of all Largemouth Bass collected; a much smaller proportion (18%) was classified as transients. Trends in site-specific residency patterns were consistent across our two seasons of collection (i.e., spring and fall) for freshwater residents (Fisher’s exact test; P = 0.63), transients (P = 0.81) and estuarine residents (P = 0.84), highlighting the temporal stability of these residency patterns. After combining across seasons, Fisher’s exact test indicated that residency classification was not independent of site for freshwater (P < 0.001) and estuarine residents (P < 0.001). Upstream sites were predominated by freshwater residents and downstream sites were predominated by estuarine residents; our middle site had a mixture of freshwater residents, transients, and estuarine residents (Table 1). Southern Flounder. —Analysis of otolith core Sr : Ca indi- cated that 55% of Southern Flounder in this study had a fresh- Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 water core signal (Sr : Ca <1.71 mmol/mol), while 45% had ≥ FIGURE 3. Sr : Ca measured at the edge (the last 20 µm) of otoliths from an estuarine or marine core signal (Sr : Ca 1.71 mmol/mol; Largemouth Bass and Southern Flounder collected in October 2005 and Large- Table 2). Furthermore, core signal was independent of site of mouth Bass in 2006 from the Mobile–Tensaw River Delta, Alabama. Otolith Sr collection for individuals with freshwater (P = 0.53) and es- : Ca is regressed against Sr : Ca from water samples taken concurrently with tuarine or marine core signals (P = 0.44), indicating that an fish collection. Asterisks indicate water Sr : Ca for Dennis Lake (DL), Gravine individual with a freshwater core signal was just as likely to be Island (GI), and Bay Minette (BM) that were averaged from samples collected during early (October 3–4, 2005) and mid-October (October 14–16, 2005). collected at a downstream site as an upstream site, and that the converse was also true for an individual with an estuarine or marine core signal. freshwater sites (TN, ML) had consistently low Sr : Ca Analysis of Southern Flounder lifetime salinity exposure ratios (typically <2.5 mmol/mol) across the entire otolith tran- plots indicated that 79% of individuals were estuarine residents, sect (i.e., freshwater resident pattern), while those collected from as opposed to 18% that were freshwater residents (Table 2). downstream, seasonally mesohaline sites (BB, BM, DB) typi- Only a single Southern Flounder collected at our middle site cally displayed a single peak of relatively high Sr : Ca between (GI) was classified as a transient. As with core signals, lifetime each otolith annulus (i.e., estuarine resident pattern; Figure 4). residency classification was independent of site of collection 1422 FARMER ET AL.

FIGURE 4. Individual Largemouth Bass transect Sr : Ca profiles showing representative patterns for freshwater residents, transients, and estuarine residents across ages. For each individual, the season, year, and site of collection are shown, as are age and TL. The dashed line indicates exposure to salinity >2‰.

for both freshwater (P = 0.27) and estuarine residents (P = described 95% of the Southern Flounder collected: (1) freshwa- 0.70; Table 2). ter core and freshwater residency (16%), (2) freshwater core and Combining core and residency classification patterns re- estuarine residency (37%), and (3) estuarine or marine core and vealed three main lifetime residency patterns (Figure 5) that estuarine residency (42%; Table 2). While Southern Flounder Downloaded by [Department Of Fisheries] at 21:37 27 October 2013

TABLE 2. Southern Flounder residency classification by collection site and location (given in parentheses as upstream [US], middle [MD], and downstream [DS]) determined from otolith Sr : Ca transects for individuals (N) collected in fall 2005–2006 in the Mobile–Tensaw River Delta, Alabama. Values >50% are in bold italics. Both core and residency classifications were independent of collection site (Fisher’s exact test; P > 0.1).

Freshwater core Estuarine or marine core Freshwater Estuarine Freshwater Estuarine resident Transient resident resident Transient resident Site Site (location) N % N % N % N % N % N % totals DL(US)22500 450 00002258 GI(MD)4361943619001911 BM(DS)00002290000571 7 DB(DS)00004330000867 12 OTOLITH-BASED FISH RESIDENCY PATTERNS 1423

FIGURE 5. Individual Southern Flounder transect Sr : Ca profiles showing representative patterns of core and resident classification combinations across collection sites. Core portions of profiles (lighter shade) are differentiated from portions of profiles used for resident classification (darker shade). For each individual, the season, year, and site of collection are shown as are age and TL. The dashed line indicates exposure to salinity >2‰. with freshwater core signals were classified as both freshwa- downstream sites was negatively related to mean annual river ter residents and as estuarine residents, only a single Southern discharge (Figure 6). In each region, low Sr : Ca was observed Flounder with an estuarine or marine core signal was classified in years of high river discharge and high Sr : Ca was observed in

Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 as a freshwater resident. All Southern Flounder with freshwater years of low river discharge (Figure 6). However, the slope of the cores and freshwater residency patterns were collected at up- relationship between otolith Sr : Ca and river discharge became stream (DL) or middle sites (GI; Table 2). Conversely, Southern increasingly negative moving from upstream to downstream Flounder with estuarine or marine cores and estuarine residency sites. patterns were most abundant at downstream sites (BM and DB; Table 2). Despite these trends, the distributions of residency pat- terns (Table 2) were determined to be independent of collection DISCUSSION site (Fisher’s exact test; P > 0.2). Lifetime Sr : Ca profiles in Relating Salinity, Water, and Otolith Microchemistry Southern Flounder revealed the diversity of habitat use patterns We defined the mixing curve relating ambient Sr : Ca to observed across sampling sites during early life and subadult salinity for the Delta. While freshwater Sr : Ca is generally periods (Figure 5). lower than in marine waters, some freshwater systems contain Sr : Ca equal to or greater than those in marine systems (Kraus Assessing the Influence of Freshwater Input and Secor 2004). Therefore, verifying the assumption of low Annual geometric mean Sr : Ca from the age-0 portion of Sr : Ca in freshwater systems is an essential step for any study Largemouth Bass otoliths collected at upstream, middle, and using otolith Sr : Ca as a marker for salinity (Walther and 1424 FARMER ET AL.

ambient water samples. As noted previously, these relationships were not intended to define the exact relationships between water and otolith Sr : Ca. Rather, our goal was to verify that otolith Sr : Ca was generally responsive to large scale, seasonal changes in water Sr : Ca in the Delta. Indeed, the same reasons these plots were not used to indicate exact relationships (i.e., highly variable daily salinity fluctuations, lag time associated with uptake into otoliths, and unknown movement patterns of fish), probably caused very different responses of Sr : Ca at the edges of Largemouth Bass otoliths in 2005 compared with 2006.

Quantifying Lifetime Salinity Exposure Our results indicate how a common freshwater (i.e., Large- mouth Bass) and subadult marine species (i.e., Southern Floun- der) use similar estuarine habitats in differing ways. Largemouth Bass are abundant in both freshwater and low-salinity portions of the Delta, and most individuals appear to be lifetime res- idents in either freshwater or low-salinity estuarine habitats, with little movement between these regions. This was surpris- ing, particularly in low-salinity areas, given that salinity was seasonally elevated and highly variable in these areas and ex- ceeded levels previously thought to limit distributions of this species. Subadult Southern Flounder, a marine species, were also collected in both freshwater and low-salinity habitats of the Delta; however, otolith microchemical analysis indicated that, unlike Largemouth Bass, Southern Flounder may exhibit partial migration between freshwater and low-salinity areas. For exam- ple, 53% of Southern Flounder collected at upstream and middle sites had Sr : Ca patterns indicative of estuarine residents (pos- sibly indicating recent migration to these freshwater habitats), compared with only 19% of Largemouth Bass. Additionally, the finding that 55% of Southern Flounder had a freshwater sig- nal around the time of otolith core formation deserves further attention because it indicates potential use of freshwater areas (possibly in the lower estuary) very early in life. Because both freshwater and subadult marine species are major components of tidal freshwater and oligohaline to mesohaline portions of estuaries (Rogers et al. 1984; Rozas and Hackney 1984), our findings are probably relevant to similar freshwater and marine Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 species in other estuarine systems. FIGURE 6. Geometric mean Sr : Ca ratios (error bars represent 95% con- Largemouth Bass.—While patterns of Sr : Ca in Largemouth fidence intervals) by year for age-0 portions of Largemouth Bass otoliths re- gressed against annual mean daily discharge (ft3/s) for (A) upstream sites (TN, Bass otoliths from downstream sites in the Delta are consistent ML), (B) middle site (GI), and (C) downstream sites (BB, BM, DB). Mean with a lack of movement away from salinity to freshwater habi- daily discharge was averaged from U.S. Geological Survey gages at Coffeeville tats, there is the possibility that our sampling could have failed (Tombigbee River) and Claiborne (Alabama River), Alabama. to detect a partial population shift to freshwater sites, leaving behind the group that we sampled in the lower Delta. Indeed, Limburg 2012). In the Delta, Sr : Ca in freshwater (i.e., salinity this kind of partial migration pattern has been documented in <0.5‰) ranged from 2.1–4.5 mmol/mol, a range encompassing other coastal fish populations (Kraus and Secor 2004; Kerr and the median value for coastal rivers across the USA (about Secor 2010). However, the consistency of estuarine residency 2.5 mmol/mol; Kraus and Secor 2004), and well below Sr : Ca patterns across seasons (i.e., spring’s freshwater period, versus for marine waters (8.5 mmol/mol; Walther and Limburg 2012). fall’s peak salinity) at downstream sites does not support this Because otolith Sr : Ca generally tracked ambient water Sr : scenario. Instead, the consistency of estuarine residency pat- Ca, we defined positive relationships for both Largemouth Bass terns in spring and fall indicates that adult Largemouth Bass are and Southern Flounder between Sr : Ca in the otolith edge and probably not conducting seasonal upstream migrations to avoid OTOLITH-BASED FISH RESIDENCY PATTERNS 1425

salinity exposure in the fall and then returning to downstream pattern reversing later in life (i.e., age-3 and older; Glover et al. sites during freshwater periods the following spring (as some 2013). Bioenergetic modeling indicated that a complex suite of previous field studies have suggested for coastal Largemouth interactions among salinity, water temperature, seasonal food Bass populations; Swingle and Bland 1974). If this were the web dynamics, and size-specific osmoregulation costs were the case, we would have expected to see Largemouth Bass with cause of these spatial trends, indicating that the cost: benefit ra- freshwater residency patterns collected at downstream sites in tio of being an estuarine resident is not only seasonally dynamic the spring, which we did not. From our analysis of temporal and but also changes with age (Glover et al. 2013). When combined spatial patterns in residency classifications, it does not appear with our findings, studies such as these begin to provide answers that movement to avoid salinity is a common feature in this to the question of why freshwater fish species would reside in coastal Largemouth Bass population. brackish habitats that represent the edge of their physiological The lack of movement away from salinity supports the find- tolerance. ings of two previous studies of adult (Norris et al. 2005) and Southern Flounder.—Although Largemouth Bass lifetime age-0 (Lowe et al. 2009) Largemouth Bass in the Delta. Norris resident status was related to site of collection, our results et al. (2005) used acoustic telemetry to track adult Largemouth for Southern Flounder indicated that lifetime resident status Bass movements in response to annual salinity increases. While and early life habitat use were independent of site of collec- no large-scale migration away from salinity was observed, the tion. Although our sample sizes were small, our findings are study covered 2 years of relatively low salinity in the Delta (peak in agreement with those of Lowe et al. (2011), and we believe salinity ∼2‰). Lowe et al. (2009) reconstructed salinity expo- they demonstrate two important findings for Southern Flounder: sure in age-0 Largemouth Bass using otolith Sr : Ca, exactly as (1) a variety of habitats may be used during early life, including this study did for adults, and found that juvenile Largemouth marine, estuarine, and tidal freshwater habitats, and (2) distinct Bass also resided in oligohline and mesohaline habitats dur- migratory contingents may exist for this species. Below, we dis- ing periods of elevated salinity. Our findings expand on those cuss implications from our work, the limitations of our study, of Lowe et al. (2009) and indicate that all ages of adult Large- and suggestions for future research. mouth Bass reside in oligohaline and mesohaline habitats during Our analysis of Sr : Ca in the core (i.e., first 100 µm) of South- annual periods of increased salinity. ern Flounder otoliths, indicated that only 45% of all subadult Previous research has suggested that short-term fluctuations Southern Flounder (ages 1–3) had the generally accepted pat- in salinity may influence the distribution of freshwater fish in tern (Stokes 1977; Burke et al. 1991; Fischer and Thompson estuaries (Perez 1969; Peterson 1988; Moser and Gerry 1989). 2004) of spending early life in higher salinity habitats, such as Otolith Sr : Ca does not provide an instantaneous measure estuarine or offshore marine environments, followed by move- of salinity exposure due to a time lag required for otolith Sr ment to lower salinity areas within the Delta. In contrast, 55% of : Ca to equilibrate with water Sr : Ca (Elsdon and Gillan- subadult Southern Flounder appeared to have spent the major- ders 2005; Melancon et al. 2009; Lowe et al. 2009). We ity of their early life in freshwater or low-salinity habitats. Our measured within-daily salinity fluctuations as high as 15‰, findings clearly are in contrast to previous research on South- which exceeds what was thought to be the tolerance limits of ern Flounder reproduction. Daniels et al. (1996) found complete Largemouth Bass (Bulkley 1975; Peterson 1988), suggesting mortality of eggs and premetamorphic larvae at salinities <10‰ that these observed salinity fluctuations do not exclude Large- in the laboratory, and Smith et al. (1999) found that eggs could mouth Bass from this coastal estuarine habitat; although fine- not maintain buoyancy at salinities <30‰. Cosson et al.(2008) scale tracking would be required to determine this on a daily reported that sperm of male Southern Flounder probably re- scale. quires saline marine water for activation of motility. Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 Finally, extended and repeated annual exposure to salinity has There are a number of possible explanations for the low implications for growth, survival, and reproduction of coastal Sr : Ca in the otolith primordia (i.e., first 25 µm of our core Largemouth Bass populations. Salinity exposure results in in- transect), which spans the egg and pelagic larval stages of creased energetic costs in Largemouth Bass due to the energy life. First, although counter to previous research with Southern requirements of osmoregulation at elevated salinity (Peterson Flounder (Daniels et al. 1996; Smith et al. 1999), spawning 1988; Meador and Kelso 1990a; Glover et al. 2012). It is, there- and hatching could have occurred in low-salinity or freshwater fore, not surprising that studies of coastal Largemouth Bass environments (see also Nissling et al. 2002; Florin and Holund¨ have consistently documented slower growth, smaller maximum 2008; Morais et al. 2011; Morais 2012; Lowe et al. 2012; size, and lower survival in these populations (Meador and Kelso Daverat et al. 2012). However, given the established life history 1990a; Norris et al. 2010; Glover et al. 2013) relative to those (Stokes 1977; Burke et al. 1991; Fischer and Thompson 2004) inland. However, despite these costs, recent studies of juvenile and reproductive physiology (Daniels et al. 1996; Smith et al. (Peer et al. 2006) and adult Largemouth Bass (Norris et al. 1999) of Southern Flounder, we suggest that the burden of proof 2010; Glover et al. 2013) in the Delta found that Largemouth for freshwater spawning is high and has not yet been satisfied. Bass during the first 3 years of life are larger in downstream, To test the hypothesis of freshwater spawning for discrete estuarine areas versus in upstream, freshwater areas with the populations of Southern Flounder, field and laboratory studies 1426 FARMER ET AL.

of reproductive development, spawning, and larval survival are estuarine residence. These findings indicate that separate mi- needed (see Nissling et al. 2002 for an example of such a study). gratory contingents of Southern Flounder may exist. Southern Second, maternal contributions to developing otoliths may Flounder with a freshwater core appear capable of occupying have affected otolith microchemistry during egg and larval either freshwater or estuarine habitats as subadults, but those stages, as has been documented for some salmonids (Kalish with estuarine or marine cores were primarily found to occupy 1990; Volk et al. 2000). While the details of this process are estuarine habitats as subadults. By examining transect Sr : Ca probably species specific, maturing (i.e., gravid) female South- plots, we can infer generalities in habitat use across the lifetime ern Flounder have been captured in freshwater areas of Mobile of Southern Flounder from each potential migratory contingent. Delta during winter (Lowe et al. 2012), prior to the offshore When viewed together, Sr : Ca plots seem to indicate differ- migration for spawning, indicating that a freshwater maternal ential timing of arrival into the Delta freshwater habitats among signal may be a plausible explanation for the low Sr : Ca ob- potential migratory contingents. Most freshwater core and fresh- served in the egg and larval regions of the otolith. Future con- water resident individuals appeared to experience freshwater trolled laboratory studies with Southern Flounder are required earlier in life than some freshwater core and estuarine resident to determine the extent to which, if any, maternally derived mi- individuals and earlier than most estuarine or marine core and crochemical signals are incorporated into the earliest growth estuarine resident individuals. Across all potential migratory regions of the otolith. contingents, there seemed to be a general movement towards Third, we may have failed to sample the otolith primor- low-salinity estuarine and tidal freshwater habitat throughout dia with 100% accuracy given the size of our laser ablation the first 2 years of life. While movement towards low-salinity spot (24–29 µm) relative to this section of the otolith (abourt habitats during the first year of life has been noted in numerous 25 µm). Despite out best efforts to section and polish all otoliths field studies (Powell and Schwartz 1977; Rogers et al. 1984; to expose the primordia, imprecision in some individuals may Burke et al. 1991; Walsh et al. 1999; Furey and Rooker 2013), have occurred due to differences in sample configuration (e.g., our study represents the first, to our knowledge, to quantify this the relatively small primordia were not exposed on the sam- migratory behavior during the first 2 years of life for individual ple surface despite appearing to be based on examination using subadult Southern Flounder. transmitted light microscopy). We suggest future studies inter- Our finding that Southern Flounder appear to use marine, es- ested in precisely quantifying primordial chemical composition tuarine, and tidal freshwater habitats as nursery areas during the use scanning electron microscopy with energy dispersive spec- first 2 years of life highlights habitat connectivity as an impor- troscopy to verify the primordia has been exposed. While all of tant process in the successful recruitment of Southern Floun- these explanations may have contributed to our results, further der, as has been shown for other estuarine-dependent species research will be required to distinguish among them. (Gillanders 2005). Future research should aim to establish the While interpreting Sr : Ca in the earliest portion of the otolith relative contribution of differing migratory contingents to the core transect (i.e., 0–25 µm) is challenging, Sr : Ca later in our adult population. This information would assist in determining otolith core transect (i.e., 25–100 µm) probably reflects envi- if specific habitats contribute disproportionally large numbers ronmental Sr : Ca. Many flatfish species develop tolerance and of recruits to the adult population and therefore deserve special even preferences for low-salinity waters once they are ready consideration for protection. for or began metamorphosis (Daniels et al. 1996; Specker et al. 1999; Hutchinson and Hawkins 2004; Bos and Thiel 2006), Assessing the Influence of Freshwater Input and they can survive both low-salinity and freshwater condi- Several recent studies have highlighted the importance tions after metamorphosis (e.g., Smith et al. 1999). As such, of assessing the temporal stability of otolith microchemical Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 we suggest that low Sr : Ca in the later portion of our otolith markers (Gillanders 2005; Walther and Limburg 2012). In profiles (i.e., 25–100 µm) indicates low-salinity or freshwater the Delta, we found that annual trends in Largemouth Bass habitat use very early in life, either during or soon after meta- otolith Sr : Ca were negatively related to river discharge. For morphosis, and that use of freshwater habitat early in life is a most Largemouth Bass at our upstream and downstream sites, common feature among Southern Flounder in the Delta. The interannual variability in river discharge did not affect our importance of both tidal freshwater and low-salinity estuarine ability to classify these individuals as freshwater or estuarine habitat to juvenile Southern Flounder is also supported by a long residents. However, at the two sites with the highest proportions history of field studies across the range of this species (Powell of transient Largemouth Bass (GI and ML), interannual and Schwartz 1977; Rogers et al. 1984; Burke et al. 1991; Walsh variability in river discharge may have confounded our ability et al. 1999; Furey and Rooker 2013). to correctly classify individuals. Specifically, at our middle When core classification was combined with lifetime res- site (GI), some individuals classified as transients may have, ident status, virtually all (95%) subadult Southern Floun- in reality, been lifetime residents that experienced freshwater der were found to exhibit one of three habitat use patterns: conditions during years of high river discharge versus estuarine (1) freshwater core and freshwater residence, (2) freshwater core conditions during years of low river discharge. Variable river and estuarine residence, and (3) estuarine or marine core and discharge across years may have also been the cause of the OTOLITH-BASED FISH RESIDENCY PATTERNS 1427

single Southern Flounder classified as transient, which was Brian Fryer and Stuart Ludsin for expert advice regarding the also collected at our middle site (GI). Therefore, variable river otolith microchemistry portion of this research. This research discharge across years may have caused us to underestimate the was funded in part by the National Sea Grant College Program number of lifetime residents at these two sites (GI and ML). under National Oceanic and Atmospheric Administration Grant NA16RG2268 through Mississippi-Alabama Sea Grant Con- CONCLUSIONS sortium Project R/CEH-20-PD and by the Alabama Department of Conservation and Natural Resources through Federal Aid in Our study offers the first conclusive evidence from the field Sport Fish Restoration Project F-40-R to DRD and RAW. that adult Largemouth Bass can withstand extended (up to 6 months) and highly variable durations of salinity exposure throughout life. This finding has important implications for REFERENCES management of Largemouth Bass in coastal areas, which has Allen, R. L., and D. M. Baltz. 1997. Distribution and microhabitat use by been increasingly of interest in recent years (Krause 2002; Nor- flatfishes in a Louisiana estuary. Environmental Biology of Fishes 50:85– ris et al. 2010; Glover et al. 2013). The lack of movement by 103. Bath, G. E., S. R. Thorrold, C. M. Jones, S. E. Campana, J. W. McLaren, and Largemouth Bass in response to salinity indicates that coastal J. W. H. Lam. 2000. Strontium and barium uptake in aragonitic otoliths of populations may function more as a mosaic of interconnected marine fish. Geochimica et Cosmochimica Acta 64:1705–1714. lakes and streams, rather than as a single homogenous man- Beck, M. W., K. L. Heck Jr., K. W. Able, D. L. Childers, D. B. Eggleston, B. M. agement unit (Lowe et al. 2009). Furthermore, the finding that Gillanders, B. Halpern, C. G. Hayes, K. Hoshino, T. J. Minello, R. J. Orth, Largemouth Bass are year-round residents in oligohaline and P. F. Sheridan, and M. P. Weinstein. 2001. The identification, conservation, and management of estuarine and marine nurseries for fish and invertebrates. mesohaline habitats and, thus, probably have increased energetic BioScience 51:633–641. costs due to salinity exposure. This has important implications Bos, A. R., and R. Thiel. 2006. Influence of salinity on the migration of postlarval for understanding how life history strategies (e.g., tradeoffs be- and juvenile flounder Pleuronectes flesus L. in a gradient experiment. Journal tween somatic growth and age-at-maturity; Glover et al. 2013) of Fish Biology 68:1411–1420. and other habitat-specific factors such as bioaccumulation of Braun, D., and R. Neugarten. 2005. Mobile–Tensaw River Delta, Alabama, hydrological modifications impact study. Nature Conservancy and Mobile heavy metals (e.g., mercury; Farmer et al. 2010) and persistent Bay Watch, Mobile, Alabama. organic compounds (e.g., PCBs: Daverat et al. 2011) in coastal Bulger, A. J., B. P.Hayden, M. E. Monaco, D. M. Nelson, and M. G. McCormick- fish populations may differ from those of inland, freshwater Ray. 1993. Biologically-based estuarine salinity zones derived from a multi- populations. variate analysis. Estuaries 16:311–322. For Southern Flounder, our findings offer additional evidence Bulkley, R. V. 1975. Chemical and physical effects on the centrarchid basses. Pages 286–294 in R. H. Stroud and H. E. Clepper, editors. Black bass biology that freshwater habitat within estuaries may be important as and management. Sport Fishing Institute, Washington, D.C. nursery habitat during the first and second years of life (Lowe Burke, J. S., J. M. Miller, and D. E. Hoss. 1991. Immigration and settlement et al. 2011). While older juveniles and adults have been collected pattern of Paralichthys dentatus and P. lethostigma in an estuarine nursery in freshwater habitats (Powell and Schwartz 1977; Rogers et al. ground, North Carolina, U.S.A. Netherlands Journal of Sea Research 27:393– 1984; Rozas and Hackney 1984), our study represents, to our 405. Campana, S. E. 1999. Chemistry and composition of fish otoliths: pathways, knowledge, the first to quantify lifetime salinity exposure in mechanisms and applications. Marine Ecology Progress Series 188:263–297. subadults of this species. Our efforts appear to have identified Cosson, J., A. L. Groison, M. Suquet, C. Fauvel, C. Dreanno, and R. Billard. the existence of several different strategies with respect to habi- 2008. Studying sperm motility in marine fish: an overview on the state of the tats used for nursery areas. Additional research is needed to de- art. Journal of Applied Ichthyology 24:460–486. termine if these differences in habitat use represent truly distinct Daniels, H. V., D. L. Berlinsky, R. G. Hodson, and C. V. Sullivan. 1996. Effects of stocking density, salinity, and light intensity on growth and survival of

Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 migratory contingents or if they simply indicate that juvenile and Southern Flounder Paralichthys lethostigma larvae. Journal of the World subadult Southern Flounder exhibit a large degree of plasticity Aquaculture Society 27:153–159. in habitat selection. Furthermore, future laboratory experiments Daniels, H. V., and R. J. Borski. 1998. Effects of low salinity on growth and sur- that control for the effect of maternal and environmental fac- vival of Southern Flounder (Paralichthys lethostigma) larvae and juveniles. tors on otolith microchemical composition would greatly assist Pages 187–191 in W. H. Howell, B. J. Keller, P. K. Park, J. P. McVey, K. Takayanagi, and Y. Uekita, editors. Nutrition and technical development of in interpreting core elemental signatures from otoliths (Morais aquaculture: proceedings of the twenty-sixth U.S.–Japan aquaculture sympo- 2012; Lowe et al. 2012). sium. University of New Hampshire Sea Grant Program, Durham. Daverat, F., P. Morais, E. Dias, J. Babaluk, J. Martin, M. Eon, R. Fablet, C. Pecheyran,´ and C. Antunes. 2012. Plasticity of European Flounder life history ACKNOWLEDGMENTS patterns discloses alternatives to catadromy. Marine Ecology Progress Series We thank Tammy DeVries, Zhaoping Yang, David Glover, 465:267–280. Mike Lowe, Alicia Norris, Mike Leonard, Norman Haley, Chris Daverat, F., N. Tapie, L. Quiniou, R. Maury Brachet, R. Riso, M. Eon, J. Laroche, Kellogg, Mark Fritz, Karen Farmer, Zach DeVries, Emily De- and H. Budzinski. 2011. Otolith microchemistry interrogation of comparative contamination by Cd, Cu and PCBs of eel and flounder, in a large SW France Vries, Benjamin Martin, Ryan Hunter, Andrew Gascho-Landis, catchment. Estuarine, Coastal and Shelf Science 92:332–338. Jennifer Barkman, and Tommy Purcell for processing labora- Day, J. W., Jr., C. A. S. Hall, W. M. Kemp, and A. Ya´nez-Arancibia.˜ 1989. tory samples and conducting field collections. Thanks to both Estuarine ecology. Wiley, New York. 1428 FARMER ET AL.

DeVries, D. R., and R. V. Frie. 1996. Determination of age and growth. Pages Kraus, R. T., and D. H. Secor. 2004. Incorporation of strontium into otoliths 483–512 in B. R. Murphy and D. W. Willis, editors. Fisheries techniques, 2nd of an estuarine fish. Journal of Experimental Marine Biology and Ecology edition. American Fisheries Society, Bethesda, Maryland. 302:85–106. de Vries, M. C., B. M. Gillanders, and T. S. Elsdon. 2005. Facilitation of barium Krause, R. A. 2002. Exploitation of an estuarine Largemouth Bass population in uptake into fish otoliths: influence of strontium concentration and salinity. northwest Florida. Pages 553–558 in D. P.Philipp and M. S. Ridgway, editors. Geochimica et Cosmochimica Acta 69:4061–4072. Black bass: ecology, conservation, and management. American Fisheries Elsdon, T. S., and B. M. Gillanders. 2005. Consistency of patterns between Society, Symposium 31, Bethesda, Maryland. laboratory experiments and field collected fish in otolith chemistry: an ex- Longerich, H. P.,S. E. Jackson, and D. Gunther.¨ 1996. Laser ablation inductively ample and applications for salinity reconstructions. Marine and Freshwater coupled plasma mass spectrometric transient signal data acquisition and an- Research 56:609–617. alyte concentration calculation. Journal of Analytical Atomic Spectrometry Elsdon, T. S., B. K. Wells, S. E. Campana, B. M. Gillanders, C. M. Jones, K. E. 11:899–904. Limburg, D. H. Secor, S. R. Thorrold, and B. D. Walther. 2008. Otolith chem- Lowe, M. R., D. R. DeVries, R. A. Wright, S. A. Ludsin, and B. J. Fryer. 2009. istry to describe movements and life-history parameters of fishes: hypotheses, Coastal Largemouth Bass (Micropterus salmoides) movement in response assumptions, limitations and inferences. Oceanography and Marine Biology: to changing salinity. Canadian Journal of Fisheries and Aquatic Sciences An Annual Review 46:297–330. 66:2174–2188. Farmer, T. M. 2008. Mercury bioaccumulation patterns in two estuarine sportfish Lowe, M. R., D. R. DeVries, R. A. Wright, S. A. Ludsin, and B. J. Fryer. 2011. populations. Master’s thesis. Auburn University, Auburn, Alabama. Otolith microchemistry reveals substantial use of freshwater by Southern Farmer, T. M., R. A. Wright, and D. R. DeVries. 2010. Mercury concentration in Flounder in the northern Gulf of Mexico. Estuaries and Coasts 34:630– two estuarine fish populations across a seasonal salinity gradient. Transactions 639. of the American Fisheries Society 139:1896–1912. Lowe,M.R.,S.A.Ludsin,B.J.Fryer,R.A.Wright,D.R.DeVries,andT.M. Felley, J. D. 1987. Nekton assemblages of three tributaries to the Calcasieu Farmer. 2012. Response to “comment on ‘otolith microchemistry reveals estuary, Louisiana. Estuaries 10:321–329. substantial use of freshwater by Southern Flounder in the northern Gulf of Fenneman, N. M., and D. W. Johnson. 1946. Physiographic divisions of the Mexico’ ” by Pedro Morais. Estuaries and Coasts 35:907–910. conterminous U.S. U.S. Geological Survey, Reston, Virginia. Available: wa- Ludsin, S. A., B. J. Fryer, and J. E. Gagnon. 2006. Comparison of solution- ter.usgs.gov/GIS/metadata/usgswrd/XML/physio.xml. (February 2013). based versus laser ablation inductively coupled plasma mass spectrometry Fischer, A. J., and B. A. Thompson. 2004. The age and growth of Southern for analysis of larval fish otolith microelemental composition. Transactions Flounder, Paralichthys lethostigma, from Louisiana estuarine and offshore of the American Fisheries Society 135:218–231. waters. Bulletin of Marine Science 75:63–77. Meador, M. R., and W. E. Kelso. 1989. Behavior and movements of Largemouth Fitzhugh, G. R., L. B. Crowder, and J. P. Monaghan Jr. 1996. Mechanisms Bass in response to salinity. Transactions of the American Fisheries Society contributing to variable growth in juvenile Southern Flounder (Paralichthys 118:409–415. lethostigma). Canadian Journal of Fisheries and Aquatic Sciences 53:1964– Meador, M. R., and W. E. Kelso. 1990a. Physiological responses of Large- 1973. mouth Bass, Micropterus salmoides, exposed to salinity. Canadian Journal of Florin,A.B.,andJ.Hoglund.¨ 2008. Population structure of flounder (Platichthys Fisheries and Aquatic Sciences 47:2358–2363. flesus) in the Baltic Sea: differences among demersal and pelagic spawners. Meador, M. R., and W. E. Kelso. 1990b. Growth of Largemouth Bass in Heredity 101:27–38. low-salinity environments. Transactions of the American Fisheries Society Furey, N. B., and J. R. Rooker. 2013. Spatial and temporal shifts in suitable 119:545–552. habitat of juvenile Southern Flounder (Paralichthys lethostigma). Journal of Melancon, S., B. J. Fryer, and J. L. Markham. 2009. Chemical analysis of Sea Research 76:161–169. endolymph and the growing otolith: fractionation of metals in freshwater fish Gillanders, B. M. 2005. Using elemental chemistry of fish otoliths to determine species. Environmental Toxicology and Chemistry 28:1279–1287. connectivity between estuarine and coastal habitats. Estuarine, Coastal and Morais, P. 2012. Comments on Lowe et al. “otolith microchemistry reveals Shelf Science 64:47–57. substantial use of freshwater by Southern Flounder in the northern Gulf of Glass, L. A., J. R. Rooker, R. T. Kraus, and G. J. Holt. 2008. Distribution, Mexico.” Estuaries and Coasts 35:904–906. condition, and growth of newly settled Southern Flounder (Paralichthys Morais, P., E. Dias, J. Babaluk, and C. Antunes. 2011. The migration patterns of lethostigma) in the Galveston Bay estuary, TX. Journal of Sea Research the European Flounder Platichthys flesus (Linnaeus, 1758) (Pleuronectidae, 59:259–268. Pisces) at the southern limit of its distribution range: ecological implications Glover, D. C., D. R. DeVries, and R. A. Wright. 2012. Effects of temperature, and fishery management. Journal of Sea Research 65:235–246.

Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 salinity and body size on routine metabolism of coastal Largemouth Bass Morisawa, M. 1968. Streams: their dynamics and morphology. McGraw Hill, Micropterus salmoides. Journal of Fish Biology 81:1463–1478. New York. Glover, D. C., D. R. DeVries, and R. A. Wright. 2013. Growth of Largemouth Moser, M. L., and L. R. Gerry. 1989. Differential effects of salinity changes Bass in a dynamic estuarine environment: an evaluation of the relative effects on two estuarine fishes, Leiostomus xanthurus and Micropogonias undulatus. of salinity, diet, and temperature. Canadian Journal of Fisheries and Aquatic Estuaries 12:35–41. Sciences 70:485–501. Nanez-James,˜ S. E., G. W. Stunz, and S. A. Holt. 2009. Habitat use patterns Hutchinson, S., and L. E. Hawkins. 2004. The relationship between temperature of newly settled Southern Flounder, Paralichthys lethostigma, in Aransas– and the size and age of larvae and peri-metamorphic stages of Pleuronectes Copano Bay, Texas. Estuaries and Coasts 32:350–359. flesus. Journal of Fish Biology 65:448–459. Nelson, J. S. 2006. Fishes of the world; 4th edition. Wiley, Hoboken, New Kalish, J. M. 1989. Otolith microchemistry: validation of the effects of physi- Jersey. ology, age and environment on otolith composition. Journal of Experimental Nissling, A., L. Westin, and O. Hjerne. 2002. Reproductive success in relation Marine Biology and Ecology 132:151–178. to salinity for three flatfish species, Dab (Limanda limanda), Plaice (Pleu- Kalish, J. M. 1990. Use of otolith microchemistry to distinguish the progeny of ronectes platessa), and Flounder (Pleuronectes flesus), in the brackish water sympatric anadromous and non-anadromous salmonids. U.S. National Marine Baltic Sea. ICES Journal of Marine Science 59:93–108. Fisheries Service Fishery Bulletin 88:657–666. Norris, A. J., D. R. DeVries, and R. A. Wright. 2010. Coastal estuaries as Kerr, L. A., and D. H. Secor. 2010. Latent effects of early life history on partial habitat for a freshwater fish species: exploring population-level effects of migration for an estuarine-dependent fish. Environmental Biology of Fishes salinity on Largemouth Bass. Transactions of the American Fisheries Society 89:479–492. 139:610–625. OTOLITH-BASED FISH RESIDENCY PATTERNS 1429

Norris, A. J., R. A. Wright, D. R. DeVries, D. L. Armstrong Jr., and J. Zolczyn- Sipe, A. M., and M. E. Chittenden Jr. 2001. A comparison of calcified structures ski. 2005. Movement patterns of coastal Largemouth Bass in the Mobile– for aging Summer Flounder, Paralichthys dentatus. U.S. National Marine Tensaw River Delta, Alabama: a multi-approach study. Proceedings of the Fisheries Service Fishery Bulletin 99:628–640. Annual Conference Southeastern Association of Fish and Wildlife Agencies Smith, T. I. J., M. R. Denson, L. D. Heyward Sr., W. E. Jenkins, and L. M. 59(2005):200–216. Carter. 1999. Salinity effects on early life stages of Southern Flounder Par- Paperno, R., and R. B. Brodie. 2004. Effects of environmental variables upon alichthys lethostigma. Journal of the World Aquaculture Society 30:236– the spatial and temporal structure of a fish community in a small, freshwater 244. tributary of the Indian River lagoon, Florida. Estuarine, Coastal and Shelf Specker, J. L., A. M. Schreiber, M. E. McArdle, A. Poholek, J. Henderson, Science 61:229–241. and D. A. Bengtson. 1999. Metamorphosis in Summer Flounder: effects of Peer, A. C., D. R. DeVries, and R. A. Wright. 2006. First-year growth and acclimation to low and high salinities. Aquaculture 176:145–154. recruitment of coastal Largemouth Bass (Micropterus salmoides): spatial Stokes, G. M. 1977. Life history studies of Southern Flounder (Paralichthys patterns unresolved by critical periods along a salinity gradient. Canadian lethostigma) and Gulf Flounder (P. albigutta) in the Aransas Bay area Journal of Fisheries and Aquatic Sciences 63:1911–1924. of Texas. Texas Parks and Wildlife Department, Technical Series 25, Perez, K. T. 1969. An orthokinetic response to rates of salinity change in two Austin. estuarine fishes. Ecology 50:454–457. Susanto, G. N., and M. S. Peterson. 1996. Survival, osmoregulation and oxy- Peterson, M. S. 1988. Comparative physiological ecology of centrarchids in gen consumption of YOY coastal Largemouth Bass, Micropterus salmoides hyposaline environments. Canadian Journal of Fisheries and Aquatic Sciences (Lacepede) exposed to saline media. Hydrobiologia 323:119–127. 45:827–833. Swingle, H. A., and D. G. Bland. 1974. A study of the fishes of the coastal Peterson, M. S., and M. R. Meador. 1994. Effects of salinity on freshwater watercourses of Alabama. Alabama Marine Resources Bulletin 10:17–102. fishes in coastal plain drainages in the southeastern U.S. Reviews in Fisheries Swingle, W. E., S. L. Spencer, and T. M. Scott Jr. 1966. Statistics on the sport Science 2:95–121. fishery of the Mobile delta during the period of July 1, 1963, to June 30, 1964. Peterson, M. S., and S. T. Ross. 1991. Dynamics of littoral fishes and decapods Proceedings of the Annual Conference Southeastern Association of Fish and along a coastal river–estuarine gradient. Estuarine, Coastal and Shelf Science Wildlife Agencies 19(1965):439–446. 33:467–483. Tebo, L. B., Jr., and E. G. McCoy. 1964. Effect of sea-water concentration on Powell, A. B., and F. J. Schwartz. 1977. Distribution of Paralichthid flounders the reproduction and survival of Largemouth Bass and Bluegills. Progressive (Bothidae: Paralichthys) in North Carolina estuaries. Chesapeake Science Fish-Culturist 26:99–106. 18:334–339. USEPA (U.S. Environmental Protection Agency). 1996. Method 1669: sampling R Development Core Team. 2011. R: a language and environment for statisti- ambient water for trace metals at EPA water quality criteria levels. USEPA, cal computing. R Foundation for Statistical Computing, Vienna. Available: Office of Water, Washington, D.C. www.R-project.org. (September 2012). Volk, E. C., A. Blakley, S. L. Schroder, and S. M. Kuehner. 2000. Otolith Ramsey, F. L., and D. W. Schafer. 2002. The statistical sleuth: a course in meth- chemistry reflects migratory characteristics of Pacific salmonids: using otolith ods of data analysis, 2nd edition. Duxbury Press, Pacific Grove, California. core chemistry to distinguish maternal associations with sea and freshwaters. Reichert, M. J. M., and H. W. van der Veer. 1991. Settlement, abundance, Fisheries Research 46:251–266. growth and mortality of juvenile flatfish in a subtropical tidal estuary (Georgia, Walsh, H. J., D. S. Peters, and D. P. Cyrus. 1999. Habitat utilization by small U.S.A.). Netherlands Journal of Sea Research 27:375–391. flatfishes in a North Carolina estuary. Estuaries 22:803–813. Rogers, S. G., T. E. Targett, and S. B. Van Sant. 1984. Fish-nursery use in Geor- Walther, B. D., and K. E. Limburg. 2012. The use of otolith chemistry to gia salt-marsh estuaries: the influence of springtime freshwater conditions. characterize diadromous migrations. Journal of Fish Biology 81:796–825. Transactions of the American Fisheries Society 113:595–606. Whitledge, G. W., B. M. Johnson, P. J. Martinez, and A. M. Martinez. 2007. Rozas, L. P., and C. T. Hackney. 1984. Use of oligohaline marshes by fishes and Sources of nonnative centrarchids in the upper Colorado River revealed by macrofaunal crustaceans in North Carolina. Estuaries 7:213–224. stable isotope and microchemical analyses of otoliths. Transactions of the Schreiber, A. M., X. Wang, Y.Tan, Q. Sievers, B. Sievers, M. Lee, and K. Burrall. American Fisheries Society 136:1263–1275. 2010. Thyroid hormone mediates otolith growth and development during Zucchetta, M., A. Franco, P. Torricelli, and P. Franzoi. 2010. Habitat distribution flatfish metamorphosis. General and Comparative Endocrinology 169:130– model for European Flounder juveniles in the Venice Lagoon. Journal of Sea 137. Research 64:133–144. Downloaded by [Department Of Fisheries] at 21:37 27 October 2013 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:38 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Conservation of the Owens Pupfish: Genetic Effects of Multiple Translocations and Extirpations Amanda J. Finger a , Steve Parmenter b & Bernie P. May a a Genomic Variation Laboratory, Department of Animal Science , University of California–Davis , One Shields Avenue, Davis , California , 95616 , USA b California Department of Fish and Wildlife , 407 West Line Street, Bishop , California , 93514 , USA Published online: 06 Sep 2013.

To cite this article: Amanda J. Finger , Steve Parmenter & Bernie P. May (2013) Conservation of the Owens Pupfish: Genetic Effects of Multiple Translocations and Extirpations, Transactions of the American Fisheries Society, 142:5, 1430-1443, DOI: 10.1080/00028487.2013.811097 To link to this article: http://dx.doi.org/10.1080/00028487.2013.811097

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1430–1443, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.811097

ARTICLE

Conservation of the Owens Pupfish: Genetic Effects of Multiple Translocations and Extirpations

Amanda J. Finger* Genomic Variation Laboratory, Department of Animal Science, University of California–Davis, One Shields Avenue, Davis, California 95616, USA Steve Parmenter California Department of Fish and Wildlife, 407 West Line Street, Bishop, California 93514, USA Bernie P. May Genomic Variation Laboratory, Department of Animal Science, University of California–Davis, One Shields Avenue, Davis, California 95616, USA

Abstract The Owens Pupfish Cyprinodon radiosus represents many of the challenges of managing threatened or endangered species in fragmented refuge populations. All six extant populations of the endangered Owens Pupfish were examined to assess how management practices, including serial translocations and founder events, have influenced the genetic diversity of the species and to make recommendations for future management. Four populations were sampled twice with 3–4 years intervening; two additional populations were sampled once. Populations were genotyped at nine microsatellite loci; estimated effective population sizes ranged from 34.2 to 347.8 individuals based on the linkage disequilibrium method and from 10 to 48 using the sibship assignment method. All of the populations were estimated to have undergone severe bottlenecks, and statistically significant pairwise FST values increased during the period between sampling. From this data we infer that the individual refuge populations have differentiated and lost genetic diversity and that without intervention they will continue to do so. For the long-term persistence of this species, we recommend founding new populations composed of 30–50 founders from each of the extant populations, regularly translocating up to 10 migrants per generation among stable populations, and maximizing habitat area and quality.

The habitat for many species is increasingly fragmented and ditional management challenges from a genetic perspective. For Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 altered, isolating populations, reducing their sizes, and mak- example, founder events can occur when establishing new pop- ing them more susceptible to stochastic events and extinc- ulations that reduce genetic diversity, exposing the species to tion (e.g., Gilpin and Soule´ 1986). Increasingly, refuge pop- risks such as a reduction in evolutionary potential (Nei et al. ulations are intentionally created to provide insurance against 1975; Allendorf 1986) or fitness (Frankham et al. 2002; Reed extinction. A sampling of fishes managed this way includes the and Frankham 2003; but see Reed 2010). Further, disjunct refuge Mohave Tui Chub Gila bicolor mohavensis (Chen et al. 2013), populations are often small and may diverge as a result of ge- Railroad Valley Springfish Crenichthys nevadae (USFWS netic drift, possibly leading to the fixation of deleterious alleles 2007), Shoshone Pupfish Cyprinodon nevadensis shoshone and general loss of genetic diversity (Frankham et al. 2002). In (Castleberry et al. 1990) and Leon Springs Pupfish Cyprinidon some cases, refuge populations derive from sequential transloca- bovinus and Pecos Gambusia Gambusia nobilis (Gumm et al. tions and founder events, potentially reducing genetic diversity 2011). However, reliance on refuge populations can create ad- further (Stockwell et al. 1996; Le Corre and Kremer 1998).

*Corresponding author: ajfi[email protected] Received April 13, 2012; accepted May 21, 2013 Published online September 6, 2013

1430 CONSERVATION GENETICS OF THE OWENS PUPFISH 1431

The problems associated with the management of small management and evolutionarily significant units (Moritz 1994; populations are particularly challenging for aquatic species in Hurt and Hedrick 2004); restoring habitat, translocating individ- deserts; many such populations in western North America are uals, removing deleterious nonnative species, and performing threatened, endangered, or already extinct (Moyle and Williams genetic assessment (e.g., Hurt and Hedrick 2004); monitoring 1990). The increased risk of extinction is due, in part, to inten- genetic diversity (e.g., Meffe 1990; Minckley 1995; Vrijenhoek sifying isolation as climate change desiccated connecting lakes 1998), and implementing captive breeding programs (e.g., Vri- and rivers and habitats shrank during the present interglacial pe- jenhoek 1998). Stochastic events may have even greater short- riod (approximately the last 12,000–15,000 years; Reheis et al. term impacts on the survival of species with restricted ranges, 2002; Echelle 2008). Today, aquatic species in the desert often so the creation of refuges and habitat protection are critical; inhabit isolated locations characterized by extreme and fluctuat- however, long-term viability may depend on maintaining ge- ing conditions. Such taxa may be made up of only one or a few netic diversity, promoting adaptation, and preserving evolution- small populations, making them extinction prone (Meffe and ary processes (Moritz 1999). Vrijenhoek 1988). Human activities such as habitat destruction (dewatering, impoundment, or diversion of rivers and springs) and the introduction of exotic species compound this risk of STUDY SPECIES extinction (Meffe 1990). The Owens Pupfish (Miller 1948) is endemic to the Owens The history of the Owens Pupfish Cyprinodon radiosus Valley in Inyo County, California. Owens Pupfish are small demonstrates some of the difficulties of conserving a formerly (<65 mm total length) fish belonging to a group of 16 extant and widespread aquatic species in refuges. The struggle to prevent 3 extinct pupfish species and subspecies found in the American extinction of the Owens Pupfish has a complicated history, Southwest and northern Mexico (the Desert Pupfish complex; involving rangewide decline, multiple extirpations, founding Pister 2001). Once so common they were a food source for events, observed demographic bottlenecks, and serial transloca- Native Americans (Wilke and Lawton 1976), Owens Pupfish tions. In this study, nine microsatellite loci were used to assess were listed as endangered in 1967 (U.S. Office of the Federal the genetic variation within and among refuge populations of Register 1967). Currently, the species persists owing to active Owens Pupfish to evaluate how 47 years of interventions and management and removal of introduced predators in artificial population management have influenced contemporary genetic and seminatural refuges (Table 1; Figure 1). diversity and make management recommendations for this fed- In habitats with constant temperatures, juvenile Owens Pup- erally listed species. fish reach sexual maturity at 3–4 months and can spawn before Conserving rare aquatic species may include research and their first winter, but they rarely live longer than 1 year (Bar- management actions such as clarifying taxonomy and defining low 1961; Soltz and Naiman 1978). In more variable habitats,

TABLE 1. Descriptions of extant Owens Pupfish populations.

Estimated Date(s) Source(s) population of initial No. of of Location Abbreviation Habitat description GPS coordinates size founders founders founders Marvin’s Marsh MM 0.028-ha 37.504211◦N, 100–1,000 1986 1,178 OVNFSa ◦ Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 man-made marsh –118.409978 W Pond D PD 0.004-ha pool in 37.477180◦N, 100–1,000 65 Old BLM ◦ ab cienega wetland –118.402245 W 1987 Spring Pond G–H PG/H 0.012-ha pool in 37.476459◦N, 100–1000 cienega wetland –118.401502◦W BLM Spring BLM 0.069-ha 37.477180◦N, 1,000–10,000 2003 2,881 Mule springbrook –118.402262◦W Springa Well 368 W368 0.020-ha 36.769950◦N, 100–1,000 1986 92 OVNFSa naturalized –118.125289◦W channel below artesian well Lower Mule LMS 0.004-ha 37.106625◦N, 300–400 2007 50 BLM Spring Pond man-made pond –118.201794◦W Spring aExtirpated. OVNFS = Owens Valley Native Fish Sanctuary. bThe BLM Spring population used to establish those in Pond D and Pond G–H was extirpated in 1988, and the current BLM Spring population is of a separate lineage. 1432 FINGER ET AL.

FIGURE 1. Aerial images showing the locations of Owens Pupfish refuge populations in (A) the Owens Valley (Inyo and Mono counties) and (B) Fish Slough (Mono County). [Figure available online in color.]

fish undergo an annual dormancy period and larvae spawned in gle population existing in 1964 (Table 1). The refuges included autumn will mature the following spring (at 6 months), while three modified springs (BLM Spring, Lower Mule Spring Pond, those spawned in spring will mature by early summer. The fish and Warm Springs); a flowing artesian well (Well 368); and in these habitats may live for 2–3 years (Pister 2001). Given two cienega—a marshy area fed by springs in the American the variability of desert aquatic environments, it is probable Southwest—wetlands (Marvin’s Marsh and the Letter Ponds). that Owens Pupfish populations historically underwent large The Letter Ponds cienega is subdivided into seasonally persis- seasonal and interannual variations in size, similar to other tent ponds that we referenced by the letters A through H and Death Valley system pupfishes (Naiman 1976; Sada and Dea- that share a dynamic hydrologic environment characterized by con 1994). Under suitable conditions (e.g., no predatory fishes, intermittent interconnections and sometimes unpredictable des- a permanent water source), Owens Pupfish quickly repopulate a iccation. The population at Lower Mule Spring Pond was estab- habitat following an observed demographic bottleneck (Young lished during the study period using equal numbers of male and Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 1976). female Owens Pupfish captured for this study from BLM Spring Owens Pupfish were thought to be extinct in 1942 (Miller (2007; N = 50), augmented by an entire population salvaged 1969) and did not receive conservation attention until a remnant and removed from Warm Springs in 2008 (N = 18). population of some 200 individuals was documented in 1964 (Miller and Pister 1971). The last approximately 800 descen- dants of the rediscovered fish were rescued from a drying habitat METHODS in 1969 (Pister 1993). Of these, about half were introduced into Nonlethal caudal fin clips were collected from four disjunct BLM Spring, from which all extant populations descend. Since populations of Owens Pupfish (BLM Spring, Marvin’s Marsh, then, refuge populations have been repeatedly founded, extir- the Letter Ponds [represented by Pond D], and Well 368; Table 1; pated, refounded, and at times supplemented with additional Figure 1) in 2007. A second set of samples was collected during fish. California Department of Fish and Wildlife (CDFW) file 2010–2011 from the original four locations plus two additional records show that since 1969, 83 transfers of Owens Pupfish ones: (1) an additional site in the Letter Ponds with individuals involving 31,995 individuals have occurred among a total of 27 from the interconnected Ponds G and H (hereafter, Pond G–H) habitats. During this study, five to six Owens Pupfish popula- and (2) Lower Mule Spring Pond. tions cumulatively occupied 0.125 ha. Each population derived Unbaited minnow traps were set overnight in July and August from two to six sequential translocations tracing back to the sin- among four populations in 2007 (Marvin’s Marsh, BLM Spring, CONSERVATION GENETICS OF THE OWENS PUPFISH 1433

TABLE 2. Microsatellite loci used in this study. Abbreviations are as follows: tests (Rice 1989). The default MCMC parameters were used = = = NA number of alleles, He expected heterozygosity, and Ho observed for both linkage disequilibrium (LD) and Hardy–Weinberg tests heterozygosity. All loci are from Burg et al. (2002). (dememorization number = 1,000; number of batches = 100; Size range number of iterations per batch = 1,000). Locus Dye NA He Ho FIS (bp) We calculated allelic richness (Rs) in HP-Rare (Kalinowski 2005), which uses rarefaction to correct for the increased like- AC9 PET 3 0.299 0.273 0.087 105–115 lihood of detecting rare alleles with increased sample size AC17 6FAM 6 0.643 0.631 0.019 130–154 (Kalinowski 2004). Pairwise FST values were calculated us- AC25 6FAM 4 0.465 0.447 0.038 152–270 ing the method of Reynolds et al. (1983) and Slatkin (1995) in AC29 6FAM 3 0.609 0.589 0.033 149–159 ARLEQUIN using 1,000 permutations. − AC35 NED 4 0.461 0.466 0.011 176–184 A neighbor joining (N–J) tree was used to visualize genetic GATA2 VIC 10 0.839 0.781 0.069 167–203 distances. Such trees do not necessarily portray evolutionary re- GATA26 6FAM 11 0.361 0.354 0.018 196–312 lationships, just differences in the frequencies of alleles. To con- GATA39 NED 13 0.786 0.705 0.103 283–331 struct the N–J tree, the SEQBOOT application in the software GATA73 PET 12 0.767 0.749 0.024 264–320 package PHYLIP, version 3.69 (Felsenstein 1995) was used to simulate 1,000 data sets before calculating Cavalli-Sforza and Edwards (1967) chord distances (DCEs) for comparison be- Pond D, and Well 368), five populations in 2010 (Marvin’s tween all pairs of sites in GENDIST (Felsenstein 1995). The Marsh, BLM Spring, Well 368, Lower Mule Spring Pond, and main assumption behind DCEs is that differences in allele fre- Pond G–H), and one population in 2011 (Pond D) (Table 1). Fin quencies are due to genetic drift only. This approach was chosen tissue was obtained from 28 to 57 individuals per site, depending because it does not assume that population sizes have remained on capture success, and placed in coin envelopes for dry storage. constant or equal over time (Felsenstein 1995) and because Genetic diversity and divergence.—Whole genomic DNA Takezaki and Nei (1996) found that DCEs are more likely to was extracted from fin clips using the Qiagen DNeasy tis- recover true tree topology than other genetic distance estimates. sue kit protocol. Nine microsatellite loci developed for Devils Unrooted N–J trees were constructed with the DCE matrices Hole Pupfish Cyprinodon diabolis (GATA2, GATA26, GATA39, calculated in GENDIST using the NEIGHBOR application in GATA73, AC9, AC17, AC25, AC29, AC35; Burg et al. 2002; PHYLIP (Felsenstein 1995). We rooted the N–J tree at the Table 2) were used to genotype individual Owens Pupfish. The midpoint. PCR conditions were identical to those in Burg et al. (2002), Population bottlenecks and effective population size.— with forward primers labeled with VIC, 6FAM, NED or PET flu- Effective population size (Ne) is a theoretical property of a orophores (Applied Biosystems, Carlsbad, California). One µL population that is a function of the rate of genetic drift. Estimat- of PCR product (diluted with water at a 1:5 ratio) was added to ing effective population size is of great interest to endangered 0.2 µL LIZ600 size standard and 8.8 µL HiDi formamide (ABI) species managers since Ne, rather than census size, describes in individual wells on a 96-well reaction plate. The DNA was how a population responds to evolutionary forces. A smaller ◦ denatured at 95 C for 3 min before being run on an AB3130XL Ne will result in faster loss of genetic variation due to drift, Genetic Analyzer. The resulting electropherograms were an- possibly reducing the adaptive potential of a population. Values alyzed using GENEMAPPER 4.0 software (Applied Biosys- of Ne for each population were calculated using two methods: tems), with allele sizes being confirmed by visual inspection of (1) the linkage disequilibrium method (Ne(LD); Waples 2006) the electropherograms. and (2) the sibship assignment method (Ne(SA); Wang 2009). Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 MICRO-CHECKER (Van Oosterhout et al. 2004) was used The Ne(LD) method operates on the theory that genetic drift to detect the presence of null alleles or genotyping errors. CON- drives the amount of linkage disequilibrium in isolated, ran- VERT software (Glaubitz 2004) was used to create input files domly mating populations (Hill 1981). Therefore, linkage dis- for Genetic Data Analysis software (GDA; Lewis and Zaykin equilibrium can be measured and used to retrospectively ap- 2001), GENEPOP version 4.0 software (updated from Raymond proximate the Ne of recent generations. The assumptions of this and Rousset 1995; Rousset 2008), and ARLEQUIN (Excoffier method include random mating, isolation, selective neutrality et al. 2005). We used GDA to calculate observed and expected of the markers used, no genetic structure within the population, heterozygosity (Ho and He), FIS values, and private alleles. De- and discrete generations. Waples (2006) notes that when the viations from Hardy–Weinberg equilibrium were calculated in assumption of discrete generations is violated, the estimate re- GENEPOP using the exact test with a Markov chain–Monte flects the effective number of breeders (NB) that produced the Carlo (MCMC) estimator of the probability that the observed sampled cohort. The program LDNe was used to estimate Ne(LD) sample was taken from a population in Hardy–Weinberg equi- (Waples and Do 2008). Alleles with frequencies less than 0.02 librium (Guo and Thompson 1992). Significant departures from were excluded from the analysis (as recommended by Waples linkage equilibrium were calculated in GENEPOP, followed and Do 2008), and 95% parametric confidence intervals were by a sequential Bonferroni correction to correct for multiple calculated. 1434 FINGER ET AL.

The Ne(SA) method estimates the frequencies of full and half to determine the sensitivity of bottleneck detection to changes siblings in a cohort, and in turn uses this analysis to derive con- in these parameters. We inferred a bottleneck when the M-ratio temporary Ne. The assumptions of the Ne(SA) method include a of the population was lower than the M-ratio expected at equi- random sample of individuals in the population from the same librium in 95% of 10,000 simulations (P < 0.05). Two loci that cohort (rather than parent–offspring relationships), but not ran- did not conform to the mutation models were dropped to cal- dom mating. The Ne(SA) method can accept samples in which culate the M-ratios: AC35 (which does not have a dinucleotide- multiple cohorts are included (such as with Owens Pupfish), but or tetranucleotide-repeat motif) and AC17 (which had a single power is reduced (Wang 2009). We used the software program large allele [270] in one population [Marvin’s Marsh in 2010]). COLONY (Jones and Wang 2010) to estimate Ne(SA) using the options of female and male polygamy, nonrandom mating, and the option of full-likelihood, medium-length runs. RESULTS Directly measuring a population bottleneck is difficult with- out knowledge of historical population sizes. However, bottle- Genetic Diversity and Divergence necks can be inferred using microsatellite data with assump- Nine microsatellite loci (Table 2) were used to genotype 454 tions regarding microsatellite mutational models (Cornuet and Owens Pupfish individuals. The loci had from three (AC9 and Luikart 1996; Garza and Williamson 2001). We used two dif- AC29)to13(GATA39) alleles per locus. Locus AC25 had the ferent tests to detect population bottlenecks: (1) the Wilcoxon largest size range (166–270 bp), and AC35 had the smallest size range (176–184 bp) (Table 3). See Table A.1 in the appendix signed-rank test for excess heterozygosity (Hk; Cornuet and Luikart 1996) implemented in the software BOTTLENECK for the allele frequencies and He and Ho values for each locus (Piry et al. 1999) and (2) the M-ratio test (Garza and Williamson in each sample group of Owens Pupfish. 2001) implemented in the software M P Val (http://swfsc. No significant departures from Hardy–Weinberg equilibrium noaa.gov/textblock.aspx?Division=FED&id=3298). Relative were detected. In the linkage disequilibrium tests, 25 out of 360 to the M-ratio test, the H test performed in BOTTLENECK of these tests were significant prior to Bonferroni correction k < detects bottlenecks that are more recent, of lower severity, or for (nominal P 0.05), but none were significant after correction. None of the 25 tests that were significant before the Bonfer- which the prebottleneck value of θ (θ = 4Neµ, where µ is the mutation rate) was small. In contrast, the M-ratio test is preferred roni correction involved consistent locus pairs across individual when detecting bottlenecks that are more severe (lasting several populations or all populations. Marvin’s Marsh (2007) had the = = generations), the prebottleneck θ is large, or the population has fewest alleles (NA 32) and Pond D (2011) had the most (NA made a demographic recovery (Williamson-Natesan 2005). 50). Expected heterozygosity values ranged from 0.514 in BLM Spring (2007) to 0.597 in Pond G–H (2011; Table 3). The Hk test operates on the theory that during a bottleneck rare alleles are more likely to be lost while common ones are re- The neighbor-joining tree (Figure 2) had strong bootstrap tained and that the latter have proportionately stronger influence support for four general groupings representing locations. Ad- on heterozygosity (Cornuet and Luikart 1996). BOTTLENECK ditionally, Lower Mule Spring Pond (2010) grouped with BLM creates a null distribution of alleles under mutation drift equilib- Spring (2007 and 2010), with support for a closer relationship rium using a chosen mutation model. A Wilcoxon signed-rank with BLM Spring (2007) than BLM Spring (2010). test is used to test for significant heterozygosity excess in com- parison with the null distribution. We used a two-phased model TABLE 3. Population-genetic parameters for sample groups of Owens Pup- (TPM) with the parameters recommended in Piry et al. (1999) fish at nine microsatellite loci; N = sample size and Rs = mean allelic richness. (variance = 12, proportion of stepwise mutations = 0.95). In Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 addition, we altered the variance (12, 24, or 36) and the propor- Mean Private tion of stepwise mutations (0.80, 0.90, or 0.95) to determine the Location Year NHe Ho NA alleles Rs sensitivity of the results to changes in these parameters. Each MM 2007 41 0.556 0.594 3.56 2 3.41 run consisted of 5,000 iterations. MM 2010 28 0.539 0.532 3.78 3.75 The M-ratio (M) is the ratio of the number of alleles at a PD 2007 35 0.568 0.591 4.56 2 4.44 locus (k) over the observed range of allele fragment sizes at PD 2011 47 0.595 0.596 5.56 4.91 that locus (r). The value of M will decline after a bottleneck BLM 2007 49 0.514 0.531 4.78 3 4.42 when alleles are randomly lost, opening “gaps” in the expected BLM 2010 50 0.537 0.503 5.22 4.71 series of alleles faster than the size range declines. To reduce W368 2007 57 0.561 0.513 5.11 0 4.54 the likelihood of type I error (a false detection of a bottleneck), W368 2010 48 0.541 0.565 4.78 4.42 we calculated M using conservative values of the proportion of PG–H 2010 49 0.597 0.621 4.44 1 4.24 stepwise mutations (p = 0.90), the average size of non–one- s LMS 2010 50 0.524 0.524 4.78 1 4.30 step mutations (delta = 3.5), and θ (θ = 10), as recommended g Total N 454 by Garza and Williamson (2001). In addition, we varied the Mean 0.553 0.557 4.66 1.5 4.31 values of θ (1, 5, or 10), deltag (2.8 or 3.5), and ps (0.80 or 0.90) CONSERVATION GENETICS OF THE OWENS PUPFISH 1435

FIGURE 2. Consensus neighbor-joining tree for visualizing the genetic distances between sample years and locations. The tree was created using PHYLIP Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 (Felsenstein 1995); the genetic distances are Cavalli-Sforza and Edwards (1967) chord distances; the tree is unrooted; and boostrap values are based on 1,000 replicates. See Table 1 for the abbreviations of the locations.

Allelic richness (Rs) was calculated with 54 genomic copies, within-year pairwise FST values between populations increased the least number of gene copies found for any locus (both AC9 over time except for that between BLM Spring (2007) and Pond and GATA73) in any one population (Marvin’s Marsh, 2010). D (2007) (FST = 0.095) and that between BLM Spring (2010) The Rs values ranged from 3.41 (Marvin’s Marsh, 2007) to 4.91 and Pond D (2011) (FST = 0.059). (Pond D, 2011) (Table 3). Except for Pond D (FST = 0.034), no significant pairwise FST The within-year pairwise FST values between the locations changes were found within populations with repeat sampling sampled in 2007 were all significant (P < 0.05; Table 4). The (Table 4). within-year pairwise FST values for 2010–2011 ranged from 0.005 to 0.100, and all but two were significant—that between Population Bottlenecks and Effective Population Size Pond D (2011) and Pond G–H, and that between BLM Spring The Ne(LD) estimates ranged from 34.2 (Marvin’s Marsh, (2010) and Lower Mule Spring Pond (2010) (Table 4). The 2007) to 347.8 (Well 368, 2010), and some of the 95% CIs were 1436 FINGER ET AL.

TABLE 4. Pairwise FST values as calculated in ARLEQUIN (Excoffier et al. 2005). The upper left quadrant shows pairwise FST values between all locations sampled in 2007. The lower right quadrant shows pairwise FST values between all locations sampled in 2010–2011. The bottom left quadrant shows pairwise FST values between locations sampled in 2007 and those locations plus two additional locations sampled in 2010–2011. Values in bold italics are significant (P < 0.05) after 1,000 permutations.

Location and MM, PD, BLM, W368, MM, PD, BLM, W368, PG–H, LMS, year 2007 2007 2007 2007 2010 2011 2010 2010 2010 2010 MM, 2007 PD, 2007 0.060 BLM, 2007 0.059 0.095 W368, 2007 0.067 0.051 0.067 MM, 2010 <0.001 0.065 0.060 0.070 PD, 2011 0.043 0.034 0.074 0.064 0.064 BLM, 2010 0.061 0.080 <0.001 0.052 0.070 0.059 W368, 2010 0.079 0.067 0.089 0.001 0.084 0.077 0.069 PG–H, 2010 0.045 0.031 0.070 0.064 0.061 0.008 0.066 0.087 LMS, 2010 0.061 0.094 <0.001 0.076 0.066 0.067 0.006 0.100 0.061

quite wide and included infinity (Table 5). The model returned DISCUSSION a negative value for BLM Spring (2010), which we interpreted Genetic Diversity and Divergence as infinity (Waples and Do 2008). The Ne(SA) estimates ranged from 19 to 29, with 95% CIs much narrower than those for Each sample of Owens Pupfish is genetically distinct, and Ne(LD) (Table 5). we detected private alleles in all populations except Well 368. The TPM model as implemented in the Hk test suggests that Owens Pupfish have fewer alleles per locus but higher mean only Marvin’s Marsh (2007) was declining or had undergone expected heterozygosity (4.66 alleles/locus, mean He = 0.553; a recent bottleneck (Table 5). This result was not sensitive to Table 3) than other freshwater fishes that have been studied (7.5 changes in variance or the proportion of non-stepwise mutations. alleles/locus, mean He = 0.46; DeWoody and Avise 2000). The The significance of the comparisons between M-values was allelic richness values for Owens Pupfish populations (3.41– not sensitive to changes in θ,ps, or deltag. There is generally 4.91) are lower than those for wild populations of Ash Mead- evidence of a bottleneck if M < 0.68 (Garza and Williamson ows Amargosa Pupfish C. nevadensis mionectes (mean = 11.6; 2001). Marvin’s Marsh (2007) was the only population not to Martin and Wilcox 2004) and refuge populations of the Desert show evidence of a bottleneck by this method (M = 0.688, P Pupfish C. macularius (6.8) and Sonoyta Pupfish C. eremus (9.1; = 0.076); the M-values in all other populations and years were Koike et al. 2008). Based on these comparisons, Owens Pupfish smaller, showing evidence of bottlenecks. are less diverse genetically than most other pupfishes.

TABLE 5. Summary of bottleneck and effective population size (Ne) tests. For bottlenecks, M-andP-values are reported for simulations implemented in M P Val; significant values are given in bold italics. For the Wilcoxon sign-rank test (Hk) of bottlenecks, the table includes P-values from BOTTLENECK; the Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 number in bold italics indicates a significant excess of heterozygotes. For effective population size, the Ne(LD) values were calculated in LDNe with 95% parametric confidence intervals and the Ne(SA) values were calculated in COLONY with 95% confidence intervals.

Location and year MP-value Hk P-value Ne(LD) 95% Parametric CI Ne(SA) 95% CI MM, 2007 0.688 0.076 0.014 34.2 16.3–123.0 21 12–38 MM, 2010 0.572 0.002 0.213 56.1 18.0–∞ 19 10–38 PD, 2007 0.571 0.001 0.248 40.1 21.2–118.7 23 13–40 PD, 2011 0.542 0.002 0.590 44.8 26.5–95.6 28 17–48 BLM, 2007 0.509 <0.001 0.545 122.1 48.3–∞ 23 14–42 BLM, 2010 0.582 0.001 0.850 –221.1 315.9–∞ 29 17–48 W368, 2007 0.593 0.001 0.633 60.4 34.5–143.9 24 14–43 W368, 2010 0.567 0.004 0.590 347.8 73.1–∞ 24 14–41 PG–H, 2010 0.534 <0.001 0.180 51.8 29.2–126.7 29 18–48 LMS, 2010 0.589 0.009 0.674 42.3 25.3–85.9 29 18–48 CONSERVATION GENETICS OF THE OWENS PUPFISH 1437

Our analyses suggest that Owens Pupfish populations are los- C. tularosa in refuges had diverged from that of the source ing genetic diversity due to population fluctuation, low effective population over a period of 30 years and that these changes were population size, and bottlenecks. However, most populations are heritable. However, Martin and Wainwright (2013) show how currently demographically robust, partly due to intensive habi- stabilizing selection for generalist phenotypes could inhibit the tat management. In 2007, exhaustive trapping found only 38 evolution of trophic specialization in Cyprinodon. Further, pup- individuals in Pond D and a bottleneck was detected. However, fish exhibit extreme phenotypic plasticity (Lema 2008), which in 2011 hundreds of fish were unexpectedly captured in Pond may itself be under selection. Further research using different D, and the genetic data revealed previously undetected alleles markers is needed to determine whether Owens Pupfish are un- and increases in other measures of genetic diversity. In addi- dergoing adaptive divergence due to selection in their respective tion, there was a significant pairwise FST value between Pond D habitats. (2007) and Pond D (2011) and a closer relationship on the N–J tree between Pond G–H (2010) and Pond D (2011) than between Population Bottlenecks and Effective Population Size Pond D (2007) and Pond D (2011) (Figure 2). We attribute this Population bottlenecks occur when there is a drastic reduc- change to immigration from areas of the Letter Pond cienega tion in population size and often result in a loss of genetic during the 4-year interval between samples. Pond D is a small variation. Bottlenecks are of conservation concern because they semi-isolated fragment on the fringe of the larger cienega and increase genetic drift and the chance of inbreeding, which can appears to be dependent on periodic immigration to maintain reduce diversity, fitness, adaptive potential, population viability, genetic variation and, perhaps, demographic persistence. Pond and, by extension, increase the risk of extinction in small pop- G–H apparently hosts more robust pupfish populations. Taken ulations (e.g., Quattro and Vrijenhoek 1989; Frankham et al. as a whole, the Letter Ponds cienega represents an important ge- 2002). netic and demographic reservoir for Owens Pupfish. The Letter Owens Pupfish refuge populations are known to have under- Ponds were established by 65 individuals in 1987 and have re- gone past demographic bottlenecks (e.g., the observed popula- ceived no further population management. Perhaps the pupfish tion low in 1964 of approximately 200 individuals), and genetic sampled in Pond G–H have undergone fewer bottlenecks and effects of these bottlenecks were detected by this study. These are more representative of past levels of genetic variation in the results may reflect any number of scenarios, including decline species. Ironically, BLM Spring (2010) and Pond D (2011) now prior to the initial observed bottleneck in 1964, bottlenecks in have the greatest pairwise FST value despite being the closest individual populations subsequent to their founding events, or a geographically. combination of the two. Interestingly the M-ratio tests detected Further evidence of genetic drift is the observed temporal a historic bottleneck in every population–year combination ex- increase in pairwise FST value in every case except between cept for Marvin’s Marsh (2007), which was also close to being BLM Spring and Pond D (which was probably influenced by significant (P = 0.076; Table 5). Conversely, the Hk test did not immigration from Pond G–H). Though only 3–4 years elapsed detect recent bottlenecks in any population–year combination between repeat sampling and the results could be influenced except for Marvin’s Marsh (2007). Taken together, the M-ratio by the sample effect, our analyses suggest that Owens Pupfish tests provide evidence that most populations underwent severe populations are diverging within 3 years (3–9 generations), in- genetic bottlenecks several generations ago, while the Hk test dicating that genetic drift may have led to rapid divergence of suggests that these populations are now at mutation-drift equi- populations within a short period of time. librium (or the test may lack power, given our markers). Though one cannot measure adaptive divergence with these The linkage disequilibrium estimates have broad confidence microsatellites, it is possible that selection is contributing to the intervals and, except for BLM Spring (2010), fall below the Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 observed divergence in Owens Pupfish populations, and man- thresholds recommended for population maintenance, e.g., 500 agers should be aware of the possibility of adaptation to different (Franklin 1980) to 5,000 (Lande 1995), implying recent and habitats while managing separate populations. For example, the ongoing attrition of genetic variation. The Ne(SA) point esti- habitats of the refuge populations are different: Marvin’s Marsh mates from the sibship model are consistently smaller than is a shallow marshy seep; the Letter Ponds are an interconnected those of Ne(LD), falling between 19 and 29 individuals. Dur- network of ponds and marsh; Well 368 has rapid current, is ing the study period, population sizes were estimated for Pond shallow (<0.3 m), and fluctuates seasonally in area; and BLM D (2007) by census (N = 38), Lower Mule Spring Pond (2010) Spring is a larger, stable, 1-m deep channel. In addition, the by mark–recapture (386; 0.95 CI, 297–403), and BLM Spring refuges variously host nonnative competitors such as the West- (2010) by visual estimation (>3,000). BLM Spring is also ern mosquitofish Gambusia affinis and red swamp crayfish Pro- much larger than the other refuge habitats (Table 1), so it is cambarus clarki, which subject some populations to different reasonable to expect a larger census size. However, when we stressors than others. Though rapid evolution has been shown collected BLM Spring (2007), we found the population to be in some taxa, including Darwin’s finches (Grant and Grant critically reduced following intrusion by Largemouth Bass Mi- 2006), copepods (Hairston and Dillon 1990), and salmonids cropterus salmoides. The predatory fish were removed, but we (Heath et al. 2003), selective forces must be very strong. Collyer were obliged to wait 4 months for the population to rebound et al. (2011) found that the body shape of White Sands Pupfish until sufficient numbers of pupfish entered our traps to sample. 1438 FINGER ET AL.

This observed demographic bottleneck may account for the low replication of variation (i.e., create “backup” copies of alle- Ne(LD) observed in BLM Spring (2007) compared with BLM les) and reduce its depletion through genetic drift by increasing Spring (2010). Strangely, the Ne(SA) estimates for BLM Spring the global Ne of the species (Rieman and Allendorf 2001), al- are similar to those for the other refuges despite large differ- though these would come at the expense of reduced population ences in habitat area and census size, and no recent bottleneck structure. was detected with the Hk test despite the observed demographic To develop effective strategies to preserve genetic diversity bottleneck. We suspect that the small number of loci used, low in fragmented populations such as the Owens Pupfish, we rec- polymorphisms per locus, and both known and unknown vio- ommend collaboration between geneticists and managers based lations of the models’ assumptions (e.g., discreet generations) on a deep understanding of the natural and population histo- reduced the power of the Ne estimation methods. ries of the target species. For Owens Pupfish, for which habitat Given that the methods varied considerably with respect to has been maximized and migration corridors are not viable, we both the estimates of Ne and the confidence intervals and that suggest that managers now have three major options: (1) active some model assumptions may have been violated, we think that translocations of individuals, (2) no translocations except in the managers should use Ne as a guide for Owens Pupfish and similar case of extirpation, and (3) conditional supplementation into cases and use caution when interpreting these results. However, declining populations. For all of these options, regular genetic by both methods all Owens Pupfish populations have estimated monitoring will ensure that sufficient genetic diversity is main- Ne values smaller than those suggested for a healthy population. tained or reveal whether continued translocations are necessary. Maintaining the species as a fragmented set of subpopulations We now outline the pros and cons of each. can only continue this phenomenon, despite intensified and im- proved management of the habitats. Option 1: Active Maintenance of Gene Flow Translocation refers to the movement of one or more in- dividuals from one population to another (IUCN 1987). For CONSERVATION IMPLICATIONS single-species conservation, the goals for translocation range Early actions taken to rescue Owens Pupfish were not guided from maintaining overall population resilience and genetic di- by genetic considerations, which led to inadvertent losses of versity to genetic rescue (e.g., Hedrick 1995; Westemeier et al. genetic variation through founder effects and demographic fluc- 1998; Tallmon et al. 2004; Weeks et al. 2011). Managers must tuation. Genetic considerations are similarly not a major focus identify the purpose of translocation, weigh the risks, and deter- of the federal recovery plan (USFWS 1998, 2009), which aims mine measures of success (e.g., a target Ne or census size) before to establish self-sustaining populations that meet certain demo- acting. Some urge caution before conducting translocations to graphic criteria. When implementing recovery plan objectives, avoid outbreeding depression (Huff et al. 2011), and in general managers should put into practice measures to minimize the managers should avoid translocating individuals from dissimi- loss of genetic diversity specieswide, preserving as much evo- lar populations or environments (Edmands 2007). For example, lutionary potential as possible so that populations can persist Goldberg et al. (2005) found that outbreeding depression in and adapt to their environments (Moritz 1999). Two factors Largemouth Bass led to increased susceptibility to disease, and complicate this: (1) refuge populations can be unpredictably Huff et al. (2011) found that outbreeding depression resulted extirpated or go through major bottlenecks (e.g., BLM Spring in reduced fitness surrogates, such as the size of young of the [2007]), where no adults were observed at all yet the popula- year. As discussed, in the case of the Owens Pupfish, popula- tion recovered in 1 year) and (2) individual refuge populations tion divergence is most likely due to genetic drift. Given that continue to diverge, most likely through genetic drift, and each the subpopulations were recently a single global population, Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 (except Well 368) has a portion of the genetic diversity that has outbreeding depression is of less concern and translocations been lost by all others (represented by private alleles). Every may have the added benefit of increasing the Ne of each pop- extant population has been recently and artificially subdivided, ulation, fostering long-term maintenance of genetic variation. and all are subject to potential catastrophic failure, as witnessed However, large-scale mixing could swamp local adaptation and at Warm Springs during the period of this investigation. The ex- reduce global Ne, and this diversity should be preserved. For tensive distribution of private alleles among the existing refuges Owens Pupfish, we adopt the recommendation of Mills and demonstrates that the system lacks the redundancy necessary to Allendorf (1996) of 1–10 migrants per generation from each safeguard genetic diversity from predictable but unpreventable stable refuge population into one or more other refuge pop- stochastic losses. The loss of or a significant bottleneck in any ulation, with the aim of minimizing the loss of polymorphism population now would result in the irretrievable loss of a por- and heterozygosity while allowing diverged allele frequencies to tion of the species’ microsatellite variation, taken here as an persist. indicator of potential adaptive variation. The lack of gene flow among populations contributes to low population values for Ne Option 2: No Translocations and the accumulation of private alleles. Restoring a level of The decision not to translocate except in the case of reestab- gene flow among populations would improve the geographic lishment has the advantage of preventing potential selective CONSERVATION GENETICS OF THE OWENS PUPFISH 1439

sweep and/or outbreeding depression caused by mixing popula- CONCLUSION tions adapted to different environments. However, small isolated This study showed that the Owens Pupfish that survived the populations may run the risks of inbreeding depression, low Ne, 1964 bottleneck was fragmented into isolated refuge popula- reduced genetic variation, and decreased fitness and long-term tions that have lost variation and are diverging through genetic viability. Evidence for inbreeding depression tends to appear drift. If this trend continues, Owens Pupfish may manifest re- in stressful environments (Armbruster and Reed 2005; but see ductions in individual fitness and evolutionary potential, jeop- Marr et al. 2006). Some of the Owens Pupfish refuge popu- ardizing their continued existence, as has been seen in closely lations are in stable springs (e.g., BLM Spring, Lower Mule related species. For example, the Devils Hole pupfish is an em- Spring Pond), so those populations may be less susceptible blematic western pupfish recognized as being on the brink of to inbreeding depression because stable environments are less extinction, with persuasive evidence that genetic load is play- stressful. In contrast, Well 368, Marvin’s Marsh, and the Let- ing a decisive role in its fate (Martin et al. 2012). Continentally, ter Ponds are spatially and thermally variable (and therefore seven species and one subspecies of Cyprinodon became extinct potentially more stressful), so inbreeding depression may be a as early as 1930 and as recently as 1994 (Burkhead 2012). We greater concern or these populations may be more subject to believe it is important to arrest the erosion of genetic diversity outbreeding depression if individuals from other populations in Owens Pupfish to help prevent this species from joining the are added (McClelland and Naish 2007). On the other hand, list of extinctions. Given the uncertainties and technical diffi- small Owens Pupfish populations may have gone through purg- culties of monitoring fitness and the complicated interactions ing, a reduction of deleterious alleles resulting in increased fit- between selection, drift, inbreeding, and bottlenecks (Bouzat ness and resistance to inbreeding depression. Theory suggests 2010), managers must make decisions to translocate individu- that populations that have gone through multiple generations als based on logistical constraints, risk assessment, and explicit of inbreeding are more likely to show purging, but experimen- goals for genetic management. Active management of habitats tal evidence provides mixed results (see review by Crnokrak and gene flow and the establishment of additional refuge pop- and Barrett 2002). Given the evidence of strong drift in Owens ulations will be crucial to the long-term maintenance of the Pupfish populations, their demographic histories, and their re- species’ residual diversity. Genetic data will inform the selec- duced genetic diversity, we believe that loss of genetic diversity tion of donor populations and the numbers of individuals for due to genetic drift and consequent inbreeding depression is a future translocations and provide baseline data for monitoring greater risk than outbreeding depression and do not recommend their effects. option 2.

ACKNOWLEDGMENTS Option 3: Supplement Declining Populations The authors are grateful to Phil Pister for personally pre- If managers deem the risk of outbreeding depression to out- venting the extinction of Owens Pupfish in 1969 and to the weigh the risk of inbreeding depression and choose not to start CDFW for encouraging and supporting this study. We thank a full-scale program of regular translocations, we recommend Karrigan Bork, Molly Stephens, Churchill Grimes, and two a cautious approach, i.e., if a refuge population is in decline anonymous reviewers for insightful comments and advice on and nongenetic factors such as poor habitat quality have been the manuscript. This study was cooperatively funded by the ruled out, to supplement that population with individuals from U.S. Fish and Wildlife Service and the California Department other populations. In this case, we again recommend staying of Fish and Wildlife through Endangered Species Act Section 6 within the 1–10 migrants per generation guidelines of Mills recovery grant E-2-F-43. Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 and Allendorf (1996); however, sharply declining populations may benefit from more input than stable populations until the recipient population stabilizes. REFERENCES With all options, catastrophic failure of some Owens Pupfish Allendorf, F. W. 1986. Genetic drift and the loss of alleles versus heterozygosity. populations is still likely, if history is any guide. If managers Zoo Biology 5:181–190. choose to reestablish populations, we recommend using roughly Armbruster, P., and D. H. Reed. 2005. Inbreeding depression in benign and equal proportions of 30–50 individuals from each of the extant stressful environments. Heredity 95:235–242. Barlow, G. W. 1961. Social behavior of the Desert Pupfish, Cyprinodon macu- populations, limited to no more than 10% of the smallest donor larius, in the field and in the aquarium. American Midland Naturalist 65:339– population to minimize genetic or demographic impacts. Thirty 359. or more fish could capture >98% of the global standing genetic Bouzat, J. L. 2010. Conservation genetics of population bottlenecks: the role of variation (Frankel and Soule´ 1981) if all founders contribute chance, selection, and history. Conservation Genetics 11:463–478. equally to the next generation (Weeks et al. 2011). After the Burg, T. M., J. L. Wilcox, and A. Martin. 2002. Isolation and characterization of polymorphic microsatellite loci in pupfish (genus Cyprinodon). Conservation population is established, continued supplementation of 1–10 Genetics 3:197–204. individuals per generation from each available source should Burkhead, N. M. 2012. Extinction rates in North American freshwater fishes, ensue, along with monitoring. 1900–2010. BioScience 62:798–808. 1440 FINGER ET AL.

Castleberry, D. T., J. E. Williams, G. M. Sato, T. E. Hopkins, A. M. Brasher, and Heath, D. D., J. W. Heath, C. A. Bryden, R. M. Johnson, and C. W. Fox. 2003. M. S. Parker. 1990. Status and management of Shoshone Pupfish, Cyprin- Rapid evolution of egg size in captive salmon. Science 299:1738–1740. odon nevadensis shoshone (Cyprinodontidae), at Shoshone Spring, Inyo Hedrick, P. W. 1995. Gene flow and genetic restoration: the Florida panther as County, California. Bulletin Southern California Academy of Sciences 89: a case study. Conservation Biology 9:996–1007. 19–25. Hill, W. G. 1981. Estimation of effective population size from data on linkage Cavalli-Sforza, L. L., and A. W.F. Edwards. 1967. Phylogenetic analysis: models disequilibrium. Genetical Research 38:209–216. and estimation procedures. Evolution 21:550–570. Huff, D. D., L. M. Miller, C. J. Chizinski, and B. Vondracek. 2011. Mixed- Chen, Y.,S. Parmenter, and B. May. 2013. Genetic characterization and manage- source reintroductions lead to outbreeding depression in second-generation ment of the endangered Mohave Tui Chub. Conservation Genetics 14:11–20. descendents of a native North American fish. Molecular Ecology 20:4246– Collyer, M. L., J. S. Heilveil, and C. A. Stockwell. 2011. Contemporary evo- 4258. lutionary divergence for a protected species following assisted colonization. Hurt, C., and P. Hedrick. 2004. Conservation genetics in aquatic species: general PLoS (Public Library of Science) ONE [online serial] 6(8):e22310. approaches and case studies in fishes and springsnails of arid lands. Aquatic Cornuet, J. M., and G. Luikart. 1996. Description and power analysis of two Sciences 66:402–413. tests for detecting recent population bottlenecks from allele frequency data. IUCN (International Union for Conservation of Nature). 1987. IUCN position Genetics 144:2001–2014. statement on translocation of living organisms: introductions, re-introductions Crnokrak, P., and S. C. H. Barrett. 2002. Purging the genetic load: a review of and re-stocking. IUCN, Species Survival Commission, Gland, Switzer- the experimental evidence. Evolution 56:2347–2358. land. Available: www.iucnsscrsg.org/download/IUCNPositionStatement.pdf. DeWoody, J. A., and J. C. Avise. 2000. Microsatellite variation in marine, (April 2012). freshwater and anadromous fishes compared with other animals. Journal of Jones, O. R., and J. L. Wang. 2010. COLONY: a program for parentage and sib- Fish Biology 56:461–473. ship inference from multilocus genotype data. Molecular Ecology Resources Echelle, A. A. 2008. The western North American pupfish clade (Cyprinodon- 10:551–555. tidae: Cyprinodon): mitochondrial DNA divergence and drainage history. Kalinowski, S. T. 2004. Counting alleles with rarefaction: private alleles and Pages 27–38 in M. C. Reheis, R. Hershler, and D. M. Miller, editors. Late hierarchical sampling designs. Conservation Genetics 5:539–543. Cenozoic drainage history of the southwestern Great Basin and lower Col- Kalinowski, S. T. 2005. HP-RARE 1.0: a computer program for performing rar- orado River region: geologic and biotic perspectives. Geological Society of efaction on measures of allelic richness. Molecular Ecology Notes 5:187–189. America, Special Paper 439, Boulder, Colorado. Koike, H., A. A. Echelle, D. Loftis, and R. A. Van Den Bussche. 2008. Mi- Edmands, S. 2007. Between a rock and a hard place: evaluating the relative risks crosatellite DNA analysis of success in conserving genetic diversity after of inbreeding and outbreeding for conservation and management. Molecular 33 years of refuge management for the Desert Pupfish complex. Animal Ecology 16:463–475. Conservation 11:321–329. Excoffier, L., G. Laval, and S. Schneider. 2005. Arlequin (version 3.0): an inte- Lande, R. 1995. Mutation and conservation. Conservation Biology 9:782–791. grated software package for population genetics data analysis. Evolutionary Le Corre, V., and A. Kremer. 1998. Cumulative effects of founding events Bioinformatics Online 1:47–50. during colonisation on genetic diversity and differentiation in an island and Felsenstein, J. 1995. PHYLIP: phylogeny inference package (version 3.69). stepping-stone model. Journal of Evolutionary Biology 11:495–512. University of Washington, Seattle. Available: http://evolution.genetics. Lema, S. C. 2008. The phenotypic plasticity of Death Valley’s pupfish. American washington.edu/phylip.html. (April 2012). Scientist 96:28–36. Frankel, O. H., and M. E. Soule.´ 1981. Conservation and evolution. Cambridge Lewis, P. O., and D. Zaykin. 2001. GDA (genetic data analysis): a computer University Press, Cambridge, UK. program for the analysis of allelic data, version 1.0 (d16c). University of Con- Frankham, R., J. D. Ballou, and D. A. Briscoe. 2002. Introduction to conserva- necticut, Storrs. Available: lewis.eeb.uconn.edu/lewishome/software.html. tion genetics. Cambridge University Press, Cambridge, UK. (April 2012). Franklin, I. R. 1980. Evolutionary change in small populations. Pages 135– Marr, A. B., P. Arcese, W. M. Hochachka, J. M. Reid, and L. F. Keller. 2006. 149 in M. E. Soule´ and B. A. Wilcox, editors. Conservation biology: an Interactive effects of environmental stress and inbreeding on reproductive evolutionary–ecological perspective. Sinauer, Sunderland, Massachusetts. traits in a wild bird population. Journal of Animal Ecology 75:1406–1415. Garza, J. C., and E. G. Williamson. 2001. Detection of reduction in popula- Martin, A. P., A. A. Echelle, G. Zegers, S. Baker, and C. L. Keeler-Foster. 2012. tion size using data from microsatellite loci. Molecular Ecology 10:305– Dramatic shifts in the gene pool of a managed population of an endangered 318. species may be exacerbated by high genetic load. Conservation Genetics Gilpin, M. E., and M. E. Soule.´ 1986. Minimum viable populations: processes of 13:349–358.

Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 species extinction. Pages 19–34 in M. E. Soule,´ editor. Conservation biology: Martin, A. P., and J. L. Wilcox. 2004. Evolutionary history of Ash Meadows the science of scarcity and diversity. Sinauer, Sunderland, Massachusetts. pupfish (genus Cyprinodon) populations inferred using microsatellite mark- Glaubitz, J. C. 2004. CONVERT: a user-friendly program to reformat diploid ers. Conservation Genetics 5:769–782. genotypic data for commonly used population genetic software packages. Martin, C. H., and P.C. Wainwright. 2013. Multiple fitness peaks on the adaptive Molecular Ecology Notes 4:309–310. landscape drive adaptive radiation in the wild. Science 339:208–211. Goldberg, T. L., E. C. Grant, K. R. Inendino, T. W. Kassler, J. E. Claussen, and McClelland, E. K., and K. A. Naish. 2007. What is the fitness outcome of D. P. Philipp. 2005. Increased infectious disease susceptibility resulting from crossing unrelated fish populations? a meta-analysis and an evaluation of outbreeding depression. Conservation Biology 19:455–462. future research directions. Conservation Genetics 8:397–416. Grant, P. R., and B. R. Grant. 2006. Evolution of character displacement in Meffe, G. K. 1990. Genetic approaches to conservation of rare fishes: examples Darwin’s finches. Science 313:224–226. from North American desert species. Journal of Fish Biology 37(Supplement Gumm, J. M., J. L. Snekser, J. M. Leese, K. P. Little, J. K. Leiser, V. E. A):105–112. Imhoff, B. Westrick, and M. Itzkowitz. 2011. Management of interactions Meffe, G. K., and R. C. Vrijenhoek. 1988. Conservation genetics in the man- between endangered species using habitat restoration. Biological Conserva- agement of desert fishes. Conservation Biology 2:157–169. tion 144:2171–2176. Miller, R. R. 1948. The cyprinodont fishes of the Death Valley system of eastern Guo, S. W., and E. A. Thompson. 1992. Performing the exact test of Hardy– California and southwestern Nevada. University of Michigan, Museum of Weinberg proportion for multiple alleles. Biometrics 48:361–372. Zoology, Miscellaneous Publication 68, Ann Arbor. Hairston, N. G., Jr., and T. A. Dillon. 1990. Fluctuating selection and response Miller, R. R. 1969. Conservation of fishes of the Death Valley system in Cali- in a population of freshwater copepods. Evolution 44:1796–1805. fornia and Nevada. California–Nevada Wildlife Transactions 1969:107–122. CONSERVATION GENETICS OF THE OWENS PUPFISH 1441

Miller, R. R., and E. P. Pister. 1971. Management of the Owens Pupfish, Cyprin- Slatkin, M. 1995. A measure of population subdivision based on microsatellite odon radiosus, in Mono County, California. Transactions of the American allele frequencies. Genetics 139:457–462. Fisheries Society 100:502–509. Soltz, D. L., and R. J. Naiman. 1978. The natural history of native fishes in the Mills, L. S., and F. W. Allendorf. 1996. The one-migrant-per-generation rule in Death Valley system. Natural History Museum of Los Angeles County, Los conservation and management. Conservation Biology 10:1509–1518. Angeles. Minckley, W. L. 1995. Translocation as a tool for conserving imperiled Stockwell, C. A., M. Mulvey, and G. L. Vinyard. 1996. Translocations and the fishes: experiences in western United States. Biological Conservation 72: preservation of allelic diversity. Conservation Biology 10:1133–1141. 297–309. Tallmon, D. A., G. Luikart, and R. S. Waples. 2004. The alluring simplicity and Moritz, C. 1994. Defining “evolutionarily significant units” for conservation. complex reality of genetic rescue. Trends in Ecology and Evolution 19:489– Trends in Ecology and Evolution 9:373–375. 496. Moritz, C. 1999. Conservation units and translocations: strategies for conserving USFWS (U.S. Fish and Wildlife Service). 1998. Owens Basin wetland and evolutionary processes. Hereditas 130:217–228. aquatic species recovery plan: Inyo and Mono counties, California. US- Moyle, P. B., and J. E. Williams. 1990. Biodiversity loss in the temperate zone: FWS, Region 1, Portland, Oregon. Available: ecos.fws.gov/docs/recovery decline of the native fish fauna of California. Conservation Biology 4:275– plan/980930b.pdf. (April 2012). 284. USFWS (U.S. Fish and Wildlife Service). 2007. Safe harbor agreement with Naiman, R. J. 1976. Productivity of a herbivorous pupfish population (Cyprin- the Duckwater Shoshone tribe to recover the Railroad Valley Springfish odon nevadensis) in a warm desert stream. Journal of Fish Biology 9: (Crenichthys nevadae) at Big Warm Spring. USFWS, Reno, Nevada. 125–137. USFWS (U.S. Fish and Wildlife Service). 2009. Owens Pupfish (Cyprinodon Nei, M., T. Maruyama, and R. Chakraborty. 1975. The bottleneck effect and radiosus) 5-year review: summary and evaluation. USFWS, Ventura Fish genetic variability in populations. Evolution 29:1–10. and Wildlife Office, Ventura, California. Available: inyo-monowater.org/ Piry, S., G. Luikart, and J. M. Cornuet. 1999. BOTTLENECK: a computer wp-content/uploads/2011/09/Owens Pupfish 5yrReview 2009.pdf. (April program for detecting recent reductions in the effective population size using 2012). allele frequency data. Journal of Heredity 90:502–503. Van Oosterhout, C., W. F. Hutchinson, D. P. M. Wills, and P. Shipley. 2004. Pister, E. P. 1993. Species in a bucket. Natural History 102(1):14–19. MICRO-CHECKER: software for identifying and correcting genotyping errors in Pister, E. P. 2001. Threatened fishes of the world: Cyprinodon radiosus Miller, microsatellite data. Molecular Ecology Notes 4:535–538. 1948 (Cyprinodontidae). Environmental Biology of Fishes 61:370. Vrijenhoek, R. C. 1998. Conservation genetics of freshwater fish. Journal of Quattro, J. M., and R. C. Vrijenhoek. 1989. Fitness differences among remnant Fish Biology 53(Supplement A):394–412. populations of the endangered Sonoran topminnow. Science 245:976–978. Wang, J. L. 2009. A new method for estimating effective population sizes Raymond, M., and F. Rousset. 1995. GENEPOP (version 1.2): population genet- from a single sample of multilocus genotypes. Molecular Ecology 18:2148– ics software for exact tests and ecumenicism. Journal of Heredity 86:248–249. 2164. Reed, D. H. 2010. Albatrosses, eagles and newts, oh my!: exceptions to the Waples, R. S. 2006. A bias correction for estimates of effective population size prevailing paradigm concerning genetic diversity and population viability? based on linkage disequilibrium at unlinked gene loci. Conservation Genetics Animal Conservation 13:448–457. 7:167–184. Reed, D. H., and R. Frankham. 2003. Correlation between fitness and genetic Waples, R. S., and C. Do. 2008. LDNE: a program for estimating effective diversity. Conservation Biology 17:230–237. population size from data on linkage disequilibrium. Molecular Ecology Reheis, M. C., A. M. Sarna-Wojcicki, R. L. Reynolds, C. A. Repenning, and Resources 8:753–756. M. D. Mifflin. 2002. Pliocene to middle Pleistocene lakes in the western Great Weeks, A. R., C. M. Sgro, A. G. Young, R. Frankham, N. J. Mitchell, K. A. Basin: ages and connections. Smithsonian Contributions to Earth Sciences Miller, M. Byrne, D. J. Coates, M. D. B. Eldridge, P. Sunnucks, M. F. Breed, 33:53–108. E. A. James, and A. A. Hoffmann. 2011. Assessing the benefits and risks of Reynolds, J., B. S. Weir, and C. C. Cockerham. 1983. Estimation of the coances- translocations in changing environments: a genetic perspective. Evolutionary try coefficient: basis for a short-term genetic distance. Genetics 105:767–779. Applications 4:709–725. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223–225. Westemeier, R. L., J. D. Brawn, S. A. Simpson, T. L. Esker, R. W. Jansen, Rieman, B. E., and F. W. Allendorf. 2001. Effective population size and genetic J. W. Walk, E. L. Kershner, J. L. Bouzat, and K. N. Paige. 1998. Tracking the conservation criteria for Bull Trout. North American Journal of Fisheries long-term decline and recovery of an isolated population. Science 282:1695– Management 21:756–764. 1698. Rousset, F. 2008. GENEPOP ’007: a complete re-implementation of the GENEPOP Wilke, P. J., and H. W. Lawton, editors. 1976. The expedition of Capt. J. W.

Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 software for Windows and Linux. Molecular Ecology Resources 8:103–106. Davidson from Fort Tejon to the Owens Valley in 1859. Ballena Press, So- Sada, D. W., and J. E. Deacon. 1994. Spatial and temporal variability of pupfish corro, New Mexico. (genus: Cyprinodon) habitat, populations, and microhabitat use at Salt Creek Williamson-Natesan, E. G. 2005. Comparison of methods for detecting bottle- and Cottonball Marsh, Death Valley National Monument, California. Report necks from microsatellite loci. Conservation Genetics 6:551–562. to the U.S. National Park Service, Death Valley National Monument, Death Young, D. 1976. Repopulation of a desert spring by Owens Pupfish, Cyprinodon Valley, California. radiosus. Master’s thesis. California State University, Fresno. 1442 FINGER ET AL.

Appendix: Detailed Allelic Results

TABLE A1. Allele frequencies and expected heterozygosity (He) and observed heterozygosity (Ho) values for nine microsatellite loci in each sample group of Owens Pupfish. Allele sizes are given in number of base pairs.

MM, MM, BLM, BLM, LMS, W368, W368, PD, PD, PG–H, Locus Allele size 2007 2010 2007 2010 2010 2007 2010 2007 2011 2010 Overall AC9 105 0.793 0.796 0.898 0.890 0.890 0.728 0.750 0.829 0.830 0.847 0.825 109 0.010 0.044 0.033 0.086 0.117 0.102 0.039 115 0.207 0.204 0.102 0.100 0.110 0.228 0.217 0.086 0.053 0.051 0.136 He 0.333 0.331 0.185 0.200 0.200 0.420 0.393 0.303 0.298 0.272 0.293 Ho 0.317 0.333 0.204 0.100 0.140 0.333 0.413 0.343 0.300 0.306 0.279 AC17 130 0.329 0.357 0.214 0.210 0.240 0.316 0.406 0.243 0.120 0.224 0.266 144 0.079 0.457 0.250 0.357 0.114 146 0.020 0.020 0.004 148 0.415 0.429 0.673 0.660 0.690 0.500 0.500 0.257 0.543 0.347 0.501 150 0.010 0.001 154 0.256 0.214 0.112 0.100 0.050 0.105 0.094 0.043 0.087 0.071 0.113 He 0.662 0.655 0.493 0.515 0.468 0.639 0.582 0.674 0.627 0.704 0.602 Ho 0.780 0.571 0.592 0.500 0.560 0.596 0.625 0.771 0.652 0.714 0.636 AC25 152 0.598 0.607 0.663 0.730 0.590 0.675 0.802 0.671 0.553 0.459 0.635 160 0.402 0.357 0.337 0.270 0.410 0.325 0.198 0.329 0.447 0.541 0.361 166 0.018 0.002 270 0.018 0.002 He 0.487 0.512 0.451 0.398 0.489 0.442 0.321 0.448 0.500 0.502 0.455 Ho 0.415 0.536 0.510 0.380 0.460 0.404 0.354 0.486 0.511 0.469 0.452 AC29 149 0.329 0.482 0.429 0.290 0.440 0.188 0.167 0.271 0.128 0.235 0.296 155 0.305 0.161 0.061 0.100 0.070 0.161 0.198 0.100 0.372 0.378 0.191 159 0.366 0.357 0.510 0.610 0.490 0.652 0.635 0.629 0.500 0.388 0.514 He 0.673 0.625 0.558 0.539 0.567 0.519 0.535 0.529 0.601 0.659 0.581 Ho 0.732 0.679 0.633 0.420 0.640 0.482 0.583 0.514 0.617 0.653 0.595 AC35 176 0.010 0.001 180 0.793 0.839 0.520 0.469 0.490 0.728 0.740 0.800 0.649 0.612 0.664 182 0.018 0.021 0.021 0.006 184 0.207 0.161 0.480 0.520 0.510 0.254 0.240 0.200 0.330 0.388 0.329 He 0.333 0.275 0.504 0.514 0.505 0.408 0.399 0.325 0.475 0.480 0.422 Ho 0.415 0.250 0.551 0.469 0.580 0.368 0.458 0.286 0.532 0.612 0.452 GATA2 167 0.032 0.020 0.030 0.029 0.106 0.041 0.026 171 0.053 0.040 0.040 0.009 0.014 Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 175 0.027 0.018 0.085 0.080 0.080 0.421 0.542 0.186 0.181 0.184 0.180 179 0.054 0.149 0.180 0.200 0.035 0.010 0.171 0.202 0.163 0.117 183 0.311 0.250 0.213 0.180 0.320 0.018 0.021 0.243 0.298 0.296 0.215 187 0.018 0.021 0.030 0.040 0.009 0.010 0.014 0.014 191 0.189 0.268 0.181 0.150 0.130 0.307 0.281 0.257 0.096 0.184 0.204 195 0.064 0.090 0.010 0.044 0.021 0.014 0.024 199 0.081 0.107 0.032 0.040 0.010 0.009 0.031 0.074 0.092 0.048 203 0.338 0.339 0.170 0.190 0.140 0.149 0.083 0.086 0.043 0.041 0.158 He 0.753 0.752 0.863 0.866 0.819 0.709 0.625 0.814 0.819 0.815 0.783 Ho 0.730 0.714 0.915 0.860 0.760 0.684 0.542 0.857 0.872 0.878 0.781 GATA26 196 0.010 0.001 200 0.031 0.040 0.050 0.026 0.031 0.021 0.020 204 0.792 0.839 0.898 0.820 0.890 0.789 0.729 0.686 0.670 0.765 0.788 208 0.194 0.089 0.010 0.080 0.040 0.132 0.167 0.257 0.160 0.102 0.123 CONSERVATION GENETICS OF THE OWENS PUPFISH 1443

TABLE A1. Continued.

MM, MM, BLM, BLM, LMS, W368, W368, PD, PD, PG–H, Locus Allele size 2007 2010 2007 2010 2010 2007 2010 2007 2011 2010 Overall 212 0.014 0.018 0.011 0.010 0.005 224 0.011 0.001 281 0.009 0.031 0.011 0.005 285 0.054 0.051 0.050 0.010 0.026 0.031 0.014 0.096 0.082 0.041 289 0.010 0.010 0.010 0.011 0.004 297 0.018 0.010 0.003 312 0.043 0.011 0.031 0.008 He 0.340 0.290 0.192 0.320 0.206 0.361 0.442 0.468 0.521 0.400 0.354 Ho 0.417 0.321 0.204 0.360 0.220 0.316 0.458 0.457 0.404 0.429 0.359 GATA39 283 0.088 0.011 0.010 287 0.015 0.001 291 0.012 0.0410 0.060 0.060 0.214 0.152 0.088 0.033 0.0830 0.074 295 0.317 0.375 0.020 0.050 0.040 0.018 0.098 0.265 0.402 0.250 0.184 299 0.024 0.018 0.133 0.150 0.130 0.125 0.076 0.147 0.065 0.135 0.100 303 0.427 0.357 0.510 0.390 0.450 0.179 0.207 0.132 0.174 0.219 0.304 307 0.207 0.250 0.214 0.270 0.220 0.402 0.391 0.191 0.261 0.281 0.269 311 0.012 0.010 0.070 0.080 0.018 0.011 0.044 0.025 315 0.031 0.010 0.029 0.022 0.009 319 0.010 0.011 0.021 0.004 323 0.031 0.010 0.009 0.011 0.010 0.007 327 0.036 0.065 0.011 0.011 331 0.010 0.001 He 0.682 0.681 0.679 0.749 0.728 0.750 0.770 0.848 0.742 0.793 0.742 Ho 0.683 0.536 0.612 0.760 0.620 0.732 0.783 0.824 0.696 0.771 0.702 GATA73 264 0.023 0.010 0.010 0.028 0.044 0.063 0.067 0.031 0.028 272 0.010 0.001 276 0.157 0.111 0.128 0.120 0.140 0.046 0.033 0.031 0.011 0.082 0.086 280 0.300 0.389 0.465 0.410 0.360 0.157 0.244 0.359 0.211 0.204 0.310 284 0.357 0.296 0.267 0.290 0.310 0.278 0.278 0.406 0.378 0.388 0.325 288 0.014 0.019 0.047 0.070 0.030 0.185 0.167 0.063 0.156 0.071 0.082 292 0.171 0.167 0.058 0.060 0.140 0.250 0.189 0.078 0.133 0.224 0.147 296 0.019 0.046 0.033 0.011 0.011 300 0.009 0.011 0.002 312 0.010 0.022 0.003 316 0.010 0.011 0.002 Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 320 0.012 0.020 0.003 He 0.739 0.734 0.698 0.732 0.741 0.804 0.804 0.702 0.774 0.753 0.748 Ho 0.857 0.852 0.558 0.680 0.740 0.704 0.867 0.781 0.778 0.755 0.757 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:38 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Abundance and Size Structure of Shortnose Sturgeon in the Altamaha River, Georgia Douglas L. Peterson a & Michael S. Bednarski a b a Warnell School of Forestry and Natural Resources , University of Georgia , 180 East Green Street, Athens , Georgia , 30602 , USA b Massachusetts Division of Marine Fisheries , New Bedford Field Facility , 1213 Purchase Street, New Bedford , Massachusetts , 02740 , USA Published online: 06 Sep 2013.

To cite this article: Douglas L. Peterson & Michael S. Bednarski (2013) Abundance and Size Structure of Shortnose Sturgeon in the Altamaha River, Georgia, Transactions of the American Fisheries Society, 142:5, 1444-1452, DOI: 10.1080/00028487.2013.802254 To link to this article: http://dx.doi.org/10.1080/00028487.2013.802254

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1444–1452, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.802254

ARTICLE

Abundance and Size Structure of Shortnose Sturgeon in the Altamaha River, Georgia

Douglas L. Peterson* and Michael S. Bednarski1 Warnell School of Forestry and Natural Resources, University of Georgia, 180 East Green Street, Athens, Georgia 30602, USA

Abstract Endangered Shortnose Sturgeon Acipenser brevirostrum inhabit large tidal rivers along the Atlantic coastline of North America, ranging from the St. John River, Canada, southward to the St. Johns River, Florida. Currently, long- term assessments of the abundance and age structure of southern populations are completely lacking. To address this information gap, we assessed recent changes in Shortnose Sturgeon abundance and age structure by sampling with anchored entanglement gear in the Altamaha River, Georgia, during summer in 2004–2010. To estimate abundance, we used the Huggins closed-capture model in Program MARK. We assessed size structure by interpreting the first, second, and third quartiles of the FL of individuals captured during each year. In total, we captured 1,737 Shortnose Sturgeon (72 recaptures). Total estimated abundance was variable, ranging from 1,206 individuals (95% lognormal confidence interval [CI] = 566–2,759) in 2009 to 5,551 individuals (95% CI = 2,804–11,304) in 2006. Much of the annual variation in total abundance was attributable to wide variations in juvenile abundance, which ranged from a low of 62 individuals (95% CI = 24–181) in 2009 to a high of 3,467 individuals (95% CI = 1,744–7,095) in 2006. Annual shifts in size structure were indicative of rapid population turnover, probably resulting from a combination of mortality and permanent emigration. Although the Altamaha River Shortnose Sturgeon population shares several characteristics with northern populations (e.g., variable juvenile abundance and stable adult abundance), our results suggest that southern populations are more susceptible to decline because of their accelerated life cycle and inherently lower abundances.

The Shortnose Sturgeon Acipenser brevirostrum inhabits et al. 1984; Kynard 1997). Because the life cycle of Short- most of the large tidal rivers along the Atlantic coastline of North nose Sturgeon is characterized by several discrete developmen- America, ranging from the St. John River, New Brunswick, tal stages, each with unique habitat needs and environmental Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 Canada, southward to the St. Johns River, Florida (Vladykov tolerances, multiple interconnected habitats are required for this and Greeley 1963; Dadswell et al. 1984; Kynard 1997). Com- species to complete its life cycle (Buckley and Kynard 1981; pared with other sturgeons, Shortnose Sturgeon are relatively Dadswell et al. 1984; Richmond and Kynard 1995). Throughout small, reaching a maximum size of 1.0–1.3 m (Dadswell et al. the species’ range, habitat alteration caused by dam construc- 1984; Kynard 1997). Like other Acipenser species, Shortnose tion, nutrient enrichment, and sedimentation have adversely Sturgeon exhibit a periodic life history characterized by late affected most populations (Collins et al. 2000). Currently, Short- age at maturity, long life span, high weight-specific fecundity, nose Sturgeon are classified as endangered in the United States and protracted spawning periodicity (Dadswell 1979; Dadswell and as a species of special concern in Canada; few extant

*Corresponding author: [email protected] 1Present address: Massachusetts Division of Marine Fisheries, New Bedford Field Facility, 1213 Purchase Street, New Bedford, Massachusetts 02740, USA. Received December 14, 2012; accepted April 30, 2013 Published online September 6, 2013

1444 ABUNDANCE OF SHORTNOSE STURGEON 1445

populations contain more than 1,000 individuals (Kynard 1997; River may give managers a better understanding of the popu- NMFS 1998). lation demographics that are typical of the species within the Recent studies suggest that river systems north of the Mid- southern portion of its range. Atlantic Bight support the largest extant populations of Short- nose Sturgeon (Kynard 1997; NMFS 1998), including those found in the Saint John River, New Brunswick (estimated N = METHODS 18,000 individuals), and the Hudson River, New York(estimated Study site.—The Altamaha River system is formed by the N = 61,057 individuals; Dadswell 1979; Bain et al. 2007). Two confluence of the Ocmulgee and Oconee rivers near Lumber other northern river systems (Kennebec River, Maine; Delaware City, Georgia, and flows approximately 212 km to its outlet River, Delaware) host populations exceeding 7,000 individuals at the Atlantic Ocean, approximately 1 km south of Darien, (Squiers et al. 1982; NMFS 1998). In contrast, southern Short- Georgia (Figure 1). The Altamaha River is one of the largest nose Sturgeon populations are typically much smaller. Within sources of freshwater on the Atlantic coast, with an annual the southeastern United States, only three river systems contain mean discharge of over 375 m3/s (U.S. Geological Survey gauge populations in excess of 1,000 individuals: the Altamaha River, 02226000; 1932–2009 data). The Altamaha River is completely Georgia; the Savannah River, Georgia–South Carolina; and the free flowing, with no man-made or natural barriers throughout Pee Dee River, North Carolina (Rogers and Weber 1994; Smith its course. et al. 1995; Kynard 1997). There is limited empirical informa- Shortnose Sturgeon have access to a variety of habitats within tion linking the low abundances of Shortnose Sturgeon in the the Altamaha River system. Hard-bottom spawning habitat is southeastern United States with regional or river-specific fac- abundant above river kilometer (rkm) 80 on the Altamaha River tors; however, high summer temperatures, low dissolved oxy- and throughout the Oconee and Ocmulgee rivers (Flournoy et al. gen levels, and incidental bycatch of adult Shortnose Sturgeon 1992). The lower 40 rkm of the Altamaha River are tidally influ- in commercial fisheries have been identified as potential im- enced and represent the primary habitat of Shortnose Sturgeon pediments to recovery in many southern rivers (Kynard 1997; (Flournoy et al. 1992; Rogers and Weber 1994). During fall Collins et al. 2000; Bahn et al. 2012). through spring, juveniles and adults may use the mesohaline Assessments of southern Shortnose Sturgeon populations are and polyhaline portions of the estuary (Dadswell et al. 1984; needed to protect and ultimately restore the existing populations. Collins et al. 2000). During summer, Shortnose Sturgeon ap- Although previous studies have provided valuable insights into pear to occur primarily in deep (>5 m) main-channel areas near the overall status of Shortnose Sturgeon populations, long-term the freshwater–saltwater interface of the Altamaha River estuary sequential assessments of total abundance and population struc- (Rogers and Weber 1995; Collins et al. 2000). ture are completely lacking (NMFS 1998). Because of the late Sampling.—Shortnose Sturgeon were sampled from late age at maturity and protracted spawning periodicity of Short- May to early August in 2004–2010. We allocated the major- nose Sturgeon, short-term population assessments are of limited ity of our sampling effort between rkm 10 and 35, a reach that utility for accurately assessing population trends. represents the general location of the freshwater–saltwater in- Shortnose Sturgeon in the Altamaha River may provide the terface within the Altamaha River estuary (Figure 1). Within best remaining example of an undisturbed population within the this reach, we selected specific netting locations where prelimi- southern portion of the species’ range. Although some develop- nary sonar surveys indicated that the bottom was relatively snag ment has occurred in the Altamaha River watershed, the river free. At these sites, sampling was conducted 1–3 times/week has no impoundments downstream of the fall line, thus provid- using anchored monofilament gill nets and trammel nets that ing Shortnose Sturgeon with unimpeded access to all known were deployed in the main channel for 30–45 min, primarily Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 historic habitats within the system. Moreover, in contrast to during slack tides. Total weekly effort varied between 3 and 16 other southern river systems, essential estuarine habitats within sampling events. All nets were 91.4 m long and 3.1 m deep. Gill the Altamaha River do not exhibit summer hypoxia (dissolved nets were constructed of three 30.5-m panels composed of 7.6-, oxygen < 2.5 mg/L; Flournoy et al. 1992; Collins et. al 2000). 10.2-, or 15.3-cm monofilament mesh (stretch measure); panels Therefore, the Altamaha River provides a unique opportunity to were sewn together in random order. Trammel nets consisted better understand the ecology and population dynamics of a rel- of an inner panel with 7.6-cm mesh and two outer panels with atively undisturbed southern population of Shortnose Sturgeon. 30.5-cm mesh. As nets were retrieved, the captured Shortnose Information on the population dynamics and demographics of Sturgeon were removed and immediately placed in a floating net Altamaha River Shortnose Sturgeon may assist fisheries man- pen. Once all of the nets had been pulled, Shortnose Sturgeon agers in devising realistic, regionally specific recovery goals for were measured (FL) and scanned for an existing PIT tag. If no the species. The specific objective of our study was to char- PIT tag was detected, one was inserted into the musculature be- acterize the Altamaha River Shortnose Sturgeon population by neath the fourth dorsal scute. All fish were released immediately quantifying annual abundance and size structure over a con- after processing. secutive 7-year period. Our results describing how Shortnose Data analyses.—We classified Shortnose Sturgeon as either Sturgeon abundance and size structure vary in the Altamaha juveniles or adults based on FL only: 499-mm FL and smaller 1446 PETERSON AND BEDNARSKI

FIGURE 1. Map of the study site within the estuary of the Altamaha River, Georgia. Black diamonds represent individual sampling locations. The single solid line represents the upstream incidence of Shortnose Sturgeon capture between 2004 and 2010; the double line represents the downstream incidence of Shortnose Sturgeon capture. Interstate 95 and Georgia State Route 17 are shown for spatial reference.

individuals were assigned to the juvenile life stage, and 500- function of individual FL, and we used Akaike’s information mm FL and larger individuals were considered adults (Figure 2; criterion corrected for small sample size (AICc; Akaike 1973; Dadswell et al. 1984; Kynard 1997). We then used the Hug- Burnham and Anderson 2002) to compare those model results gins closed-capture model in Program MARK to estimate juve- with the results of an alternative model in which capture prob- nile and adult abundances (Huggins 1989; White and Burnham ability was held constant. Model selection was conducted by 1999). In contrast to the traditional closed-capture model of Otis identifying the model with the lowest AICc. Because weekly et al. (1978), the Huggins model allows the incorporation of indi- sampling effort was not equal, we modeled capture probability Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 vidual covariates, such as length, into the modeling framework as time varying among individual capture periods. We then fixed (Huggins 1989). We sampled only during the late-spring and the capture probability for marked and unmarked individuals as summer months, when Shortnose Sturgeon are known to occur equal in all models. in low-salinity (<4.2 psu) areas near the freshwater–saltwater The precision of each mark–recapture estimate was assessed interface (Flournoy et al. 1992; Collins et al. 2000). We as- by calculating a 95% lognormal confidence interval (95% CI) sumed that the population was closed; to ensure that the clo- within Program MARK, as described by White and Burnham sure assumption was met, we regularly sampled upstream and (1999). Inferences regarding abundance trends were based only downstream of the study area, allocating at least 10 sampling on estimates that did not have overlapping 95% CIs. Relative events/season to the 10 rkm immediately upstream and down- weights were also calculated within Program MARK. Any an- stream of the study reach. nual estimates that did not have overlapping 95% CIs were Population models were constructed using each sampling considered significantly different from each other. Year-to-year week as a unique capture period. During each year of the study, variation in abundance was quantified by calculating the co- we considered the capture probability of juveniles and adults efficient of variation (CV = 100 × [SD/mean]) of the point to be equal. To test the validity of this assumption, we mod- estimates for total, adult, and juvenile abundances (Rothschild eled capture probability as either a linear function or a quadratic and DiNardo 1987; Woodland and Secor 2007). ABUNDANCE OF SHORTNOSE STURGEON 1447 Downloaded by [Department Of Fisheries] at 21:38 27 October 2013

FIGURE 2. Annual length frequency histograms of Shortnose Sturgeon captured in the Altamaha River during spring and summer in 2004–2010. The vertical line represents the FL used to separate juveniles (≤499 mm) and adults (≥500 mm). 1448 PETERSON AND BEDNARSKI

Because Shortnose Sturgeon are listed as endangered under 2004. Adult catch was less variable, ranging from 56 individuals the federal Endangered Species Act, traditional methods for (1 recapture) in 2006 to 211 individuals (14 recaptures) in 2010. estimating ages of wild fish (e.g., analysis of otoliths or pec- The results of our Huggins closed-capture models demon- toral fin spines) were not used in this study. Instead, annual strated that in 4 of 7 years, models that excluded a linear or length frequency histograms constructed from individual FLs quadratic relationship between FL and capture probability had of captured fish were used as a noninvasive, indirect measure the lowest AICc, suggesting that no size-selective gear bias oc- of the population’s age structure. Previously collected length- curred during most of the study years (Table 2). Among the at-age data from the Altamaha River population (DeVries and years for which the model with the lowest AICc did include an Peterson 2008) were used to confirm the age estimates for spe- effect of FL on capture probability, 2 years were best described cific cohorts that were identified in our length frequency his- by models that included linear relationships, and 1 year was tograms. Although we initially intended to examine differences best described by a model that displayed a quadratic relation- in size structure through the use of hypothesis testing, a pre- ship; these results suggested that yearly relationships between liminary analysis indicated that variability and distributional FL and capture probability were inconsistent. Consequently, all parameters were inconsistent among years, thereby precluding of our subsequent inferences regarding abundance were based this approach. Consequently, we based all inferences regarding only on the yearly model that (1) considered capture probability changes in size structure on descriptive summary statistics, in- to be equal for juveniles and adults and (2) excluded an effect cluding the minimum, maximum, and mean FLs and the 25th, of FL on capture probability. Because repeated sampling above 50th, and 75th percentiles (first, second, and third quartiles) of and below the freshwater–saltwater interface yielded no Short- FL for a given year. nose Sturgeon, the requirement for population closure appeared To interpret the magnitude of year-to-year differences in size to have been met during the sampling period in each year of the structure, we calculated the 25th, 50th, and 75th percentiles of study. FL for both the total Shortnose Sturgeon population and the adult Total annual abundance of Shortnose Sturgeon was variable portion of the population. For the adult population, decreases in over the 7 years of the study, ranging from a low of 1,206 the 25th percentile between years were interpreted as the entry individuals (95% CI = 566–2,759) in 2009 to a high of 5,551 of juveniles into the adult population. Increases in the median individuals (95% CI = 2,804–11,304) in 2006, with a CV of FL (50th percentile) for the total population or adult population 56.6% (Figure 3). Juvenile abundance was even more variable, were interpreted as increases in the average size of individuals ranging from a low of 62 individuals (95% CI = 24–181) in within that population. 2009 to a high of 3,467 individuals (95% CI = 1,744–7,095) in 2006, with a CV of 88.9% (Figure 3). Adult abundance was comparatively constant, ranging from a low of 707 individuals RESULTS (95% CI = 367–1,421) in 2005 to a high of 2,122 individuals From 2004 to 2010, we set a total of 1,173 nets, yielding (95% CI = 1,059–4,377) in 2006, with a CV of 32.0%. The a total catch of 1,737 Shortnose Sturgeon (Table 1). Total an- 95% CI of the adult population estimate overlapped among all nual effort varied from a minimum of 89.1 net-hours in 2004 to years of the study. a maximum of 255.2 net-hours in 2010. Likewise, total catch Over the 7 years, we detected the presence of age-1 cohorts in varied annually from 116 individuals (4 recaptures) in 2009 only 3 years: 2004, 2006, and 2010. As these juvenile cohorts re- to a high of 412 individuals (15 recaptures) in 2004. Annual cruited to the adult population, a marked shift in adult size (age) catch of juveniles was highly variable, ranging from 6 individ- structure was evident (Figures 2, 4). Annual 25th-percentile FL uals (1 recapture) in 2009 to 265 individuals (10 recaptures) in of the total population ranged from a low of 311 mm in 2004 to a Downloaded by [Department Of Fisheries] at 21:38 27 October 2013

TABLE 1. Annual sampling effort (net-hours, using gill nets and trammel nets) and catch of Shortnose Sturgeon juveniles (≤499 mm FL) and adults (≥500 mm FL) in the Altamaha River, Georgia, 2004–2010.

Juveniles Adults Year Sampling period Effort (net-hours) Marked Recaptured Marked Recaptured 2004 May 30–Aug 7 89.1 265 10 147 5 2005 Jun 5–Aug 13 144.8 163 7 56 1 2006 Jun 4–Aug 5 104.0 182 6 112 1 2007 Jun 3–Aug 12 123.6 88 3 89 2 2008 Jun 8–Aug 9 140.7 23 1 184 10 2009 May 31–Aug 8 160.1 6 1 110 3 2010 May 31–Aug 1 255.2 101 8 211 14 ABUNDANCE OF SHORTNOSE STURGEON 1449

TABLE 2. Preliminary Huggins closed-capture models with capture probability as a linear or quadratic function of FL for Shortnose Sturgeon in the Altamaha River, Georgia, 2004–2010 (AICc = Akaike’s information criterion corrected for small sample size; AICc = difference in AICc between the given model and the best-performing model). Models with no effect of FL are designated as FL(none). All models included time-varying capture probability (M[t]).

Model AICc AICc Relative weight Likelihood Parameters 2004 M(t), FL(none) 1,931.00 0.00 0.63 1.00 10 M(t), FL(linear) 1,933.98 1.98 0.23 0.37 11 M(t), FL(quadratic) 1,934.06 3.06 0.14 0.22 12 2005 M(t), FL(none) 1,064.91 0.00 0.63 1.00 10 M(t), FL(linear) 1,066.58 1.68 0.27 0.43 11 M(t), FL(quadratic) 1,068.60 3.69 0.10 0.16 12 2006 M(t), FL(linear) 1,338.55 0.00 0.45 1.00 10 M(t), FL(none) 1,338.95 0.40 0.37 0.82 9 M(t), FL(quadratic) 1,340.33 1.78 0.18 0.41 11 2007 M(t), FL(none) 815.42 0.00 0.39 1.00 10 M(t), FL(quadratic) 815.66 0.24 0.35 0.89 12 M(t), FL(linear) 816.28 0.86 0.26 0.65 11 2008 M(t), FL(linear) 913.43 0.00 0.45 1.00 9 M(t), FL(none) 914.33 0.90 0.29 0.64 8 M(t), FL(quadratic) 914.50 1.07 0.26 0.59 11 2009 M(t), FL(quadratic) 533.30 0.00 0.69 1.00 12 M(t), FL(linear) 535.08 1.78 0.28 0.41 11 M(t), FL(none) 540.23 6.93 0.02 0.03 10 2010 M(t), FL(none) 1,607.60 0.00 0.66 1.00 9 M(t), FL(linear) 1,609.61 2.01 0.24 0.37 10 M(t), FL(quadratic) 1,611.62 4.02 0.09 0.14 11

high of 561 mm in 2009 (Figure 4). The annual 25th-percentile pled both upstream and downstream of the freshwater–saltwater FL of the adult population ranged from a low of 521 mm in interface; however, we detected no immigration or emigration Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 2007 to a high of 604 mm in 2010 (Figure 4). Decreases in during the period for which population estimates were gener- 25th-percentile size indicated the entry of new individuals into ated. Therefore, even though we cannot prove that immigration the adult population during 2005, 2006, and 2007. or emigration did not occur, the lack of captures outside of the study area was consistent with the assumption that the popu- lation was closed during the sampling period upon which the DISCUSSION estimates were based. Likewise, the Huggins closed-capture As with any quantified population assessment, the population model also assumed that all individuals within the population parameters presented in this study are estimates of true popula- had an equal capture probability regardless of size. Although the tion demographics. Although we included error terms on these model with the lowest AICc for 3 years of the study did include parameter estimates when appropriate, the validity of some of length as a covariate, the relationship between FL and capture the underlying model assumptions cannot be proven. For exam- probability was inconsistent among these years. ple, use of the Huggins closed-capture model to estimate annual Regardless of some inherent uncertainties in the modeling abundances required us to assume that the Shortnose Sturgeon methods we used, our results constitute a quantified, multiyear population was closed to births, deaths, immigration, and emi- assessment of the abundance and size structure of Shortnose gration during each of our sampling periods. We routinely sam- Sturgeon within a southern river system. Our highest estimate 1450 PETERSON AND BEDNARSKI

FIGURE 3. Total, juvenile, and adult Shortnose Sturgeon abundances in the Altamaha River, 2004–2010. Error bars represent 95% lognormal confidence FIGURE 4. Box plots illustrating the minimum and maximum FLs (lower and intervals. upper ends of whiskers) and the first, second, and third quartiles of FL (lower end of box, line within box, and upper end of box, respectively) for the total of total abundance (5,551 individuals in 2006) was more than population and the adult population of Shortnose Sturgeon in the Altamaha double those reported for the Pee Dee River (N = 1,000 in- River, 2004–2010. dividuals; Kynard 1997) and the Savannah River (N = 1,676 individuals; Smith et al. 1995), confirming that the Altamaha stocks. This may help to explain why southern Shortnose Stur- River population is the largest known Shortnose Sturgeon pop- geon populations are typically smaller than populations found ulation south of the Delaware River. Our study also showed in the northern half of the species’ range, where slower growth Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 that although the Altamaha River population was character- and greater longevity result in larger, more stable populations. ized by stable adult abundance, even our highest abundance Like other Shortnose Sturgeon populations throughout the estimate was an order of magnitude lower than estimates re- species’ range, the Altamaha River population exhibited wide ported for several well-studied northern populations, including variation in annual recruitment. We observed approximately the St. John River (N = 18,000 individuals; Dadswell 1979), 50-fold variation in annual juvenile abundance. Based on the the Hudson River (N = 56,708 individuals; Bain 2001, cited annual length frequency histograms and the rapid shifts in an- by Woodland and Secor 2007), and the Delaware River (N = nual size structure that occurred as juvenile cohorts recruited to 14,000 individuals; O’Herron et al. 1993). The results of our size the adult population, we concluded that strong year-classes were structure analyses indicated that the Altamaha River population produced during only 3 of the 7 years of our study. Although displayed rapid juvenile growth, early onset of maturity, and corroborative studies of other southern populations are needed, rapid downward shifts in median adult size as juveniles entered Woodland and Secor (2007) also observed variable year-class the adult population. Given that adult abundance remained rel- strength in the Hudson River (CV = 66.0%) and St. John River ative constant during our study, these findings suggest that the (CV = 44.0%) populations. Because Shortnose Sturgeon are Altamaha River population displays a much higher rate of pop- iteroparous, highly fecund, and long lived, their populations are ulation turnover than has been previously inferred for northern naturally buffered against annual recruitment variation caused ABUNDANCE OF SHORTNOSE STURGEON 1451

by stochastic events, such as drought and floods. Hence, vari- enough juveniles survived to produce a shift in the adult size able juvenile abundance is probably typical of healthy Short- structure (Figure 4). Because adult abundance did not increase, nose Sturgeon populations in both northern and southern river the recruitment of these juveniles must have been offset by an systems. increase in adult mortality, emigration, or both. Assuming that Adult abundance was less variable than juvenile abundance the Altamaha River population is typical of Shortnose Sturgeon and did not appear to be significantly different among study populations in other southern river systems, our results suggest years. Interestingly, previous studies have documented similar that recruitment dynamics in southern rivers are quite differ- patterns in other Shortnose Sturgeon populations. A recent study ent from those in the northern part of the species’ range, where of Shortnose Sturgeon in the Ogeechee River, Georgia (Peterson greater longevity allows the adult population to accumulate over and Farrae 2011), documented a comparable level of population several decades. Although additional corroborative studies are variability from 2007 to 2009 (CV = 33.0%). Although Bain needed, we recommend that recovery goals for southern popu- et al. (2007) did not provide a quantitative estimate of variability lation segments be reevaluated within this context. in annual adult abundance, their data showed that adult abun- The final recovery plan for Shortnose Sturgeon (NMFS 1998) dance in the Hudson River Shortnose Sturgeon population was identified the need for quantitative population assessments as steady over the 4 years of their study. one of the highest research priorities for recovery of the species. Rapid and substantial shifts in adult size structure have not Despite the fact that Shortnose Sturgeon have been classified been reported for other Shortnose Sturgeon populations, yet as endangered under the U.S. Endangered Species Act for more marked shifts in the size structure of the adult population were than four decades, valid population estimates have been obtained documented in several of our study years. Some of these fluctu- for only a few river systems, most of which are in the northern ations could have been attributable to our use of pooled length portion of the range. The results of this study demonstrate that data from both gill nets and trammel nets; however, the length intensive mark–recapture sampling of summer Shortnose Stur- frequency histograms we constructed (Figure 2) clearly illus- geon aggregations can yield quantitative abundance estimates trate the emergence and recruitment of large juvenile cohorts in for southern populations. Similar methods have also been used 2004, 2006, and 2010. Based on the annual length frequency for studies of juvenile Atlantic Sturgeon Acipenser oxyrinchus histograms and on the decreases in 25th-percentile FL of adults oxyrinchus (Schueller and Peterson 2010) and likely have utility from 2005 to 2006 and from 2006 to 2007, we inferred that the for other sturgeon species. 2004 and 2006 age-1 cohorts grew rapidly and recruited to the Although the threshold for minimum viable population size adult population in subsequent years. These observations are is uncertain, there is little doubt that the probability of long- consistent with those of Dadswell et al. (1984), who concluded term persistence of a population is positively related to total that the mean age at maturity for the Altamaha River population abundance (Shaffer 1981). As such, abiotic constraints on habi- was 2–4 years. In contrast, the more stable size structures typi- tat availability may be particularly important for the viability cally reported for northern populations are probably the result of of Shortnose Sturgeon populations in southern rivers like the the slower growth rates, later maturation, and greater longevity Altamaha River, where total abundance is variable but typically typical of those populations (Dadswell 1979; Dadswell et al. less than 6,000 individuals. Moreover, because southern popula- 1984; Kynard 1997; Bain et al. 2007). The results of our study tions exhibit high rates of turnover and low adult longevity, they suggest that that the faster growth and accelerated life cycle may be particularly sensitive to any type of human disturbance of southern Shortnose Sturgeon populations cause them to be that increases variability in annual recruitment or adult survival. inherently smaller and more dynamic than populations in the Reduced longevity of southern Shortnose Sturgeon (Dadswell northern part of the species’ range. et al. 1984) may limit the ability of southern populations to per- Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 The combination of rapid juvenile growth and relatively early sist when human disturbances exacerbate or prolong the interval age at maturity, as observed in this study, suggested that adult of recruitment failure, especially if those disturbances also im- abundance in the Altamaha River Shortnose Sturgeon popula- pede adult survival. Undoubtedly, further studies are needed to tion should have increased sharply during the latter years of quantify interactions among habitat availability, survival, envi- our study, yet annual abundance estimates indicated that the ronmental variation, recruitment success, and long-term viabil- adult population was remarkably stable. Although the impreci- ity of southern populations. Once these linkages are identified, sion of these estimates may have limited our ability to detect managers will be better able to construct river-specific recov- slight changes in adult abundance, the recruitment of more than ery plans that can help to ensure the long-term persistence of 3,000 juveniles to an adult population of approximately 1,500 Shortnose Sturgeon in southern river systems. should have been obvious. We attribute the stability of the adult population to a relatively high rate of population turnover. Sev- eral mechanisms may foster such rapid turnover. First, density- ACKNOWLEDGMENTS dependent factors may have increased juvenile mortality as We thank the National Marine Fisheries Service and the strong year-classes approached adulthood; however, decreases Georgia Department of Natural Resources for financial sup- in the 25th-percentile FL of the adult population indicated that port. David Higginbotham provided invaluable logistical and 1452 PETERSON AND BEDNARSKI

editorial support. Reviews by Colin Shea and Vanessa Lane O’Herron, J. C., II, K. W. Able, and R. W. Hastings. 1993. Movements of improved the quality and scope of this manuscript. Shortnose Sturgeon (Acipenser brevirostrum) in the Delaware River. Estuaries 16:235–240. Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Sta- REFERENCES tistical inference from capture data on closed animal populations. Wildlife Akaike, H. 1973. Information theory and an extension of the maximum likeli- Monographs 62. hood principle. Pages 267–281 in B. N. Petrov and F. Csaki,´ editors. Proceed- Peterson, D. L., and D. J. Farrae. 2011. Evidence of metapopulation dynamics ings of the second international symposium on information theory. Akademiai´ in Shortnose Sturgeon in the southern part of their range. Transactions of the Kiado,´ Budapest. American Fisheries Society 140:1540–1546. Bahn, R. A., J. E. Fleming, and D. L. Peterson. 2012. Bycatch of Shortnose Richmond, A. M., and B. Kynard. 1995. Ontogenetic behavior of Shortnose Sturgeon in the commercial American shad fishery of the Altamaha River, Sturgeon, Acipenser brevirostrum. Copeia 1995:172–182. Georgia. North American Journal of Fisheries Management 32:557–562. Rogers, S. G., and W. Weber. 1994. Occurrence of Shortnose Sturgeon Bain, M. B., N. Haley, D. L. Peterson, K. K. Arend, K. E. Mills, and P. J. (Acipenser brevirostrum) in the Ogeechee-Canoochee River system, Geor- Sullivan. 2007. Recovery of a US endangered fish. PLoS (Public Library of gia during the summer of 1993. Final Report to the Nature Conservancy of Science) ONE [online serial] 2(1):e168. Georgia, U.S. Army, Atlanta. Buckley, J., and B. Kynard. 1981. Spawning and rearing of Shortnose Sturgeon Rogers, S. G., and W. Weber. 1995. Status and restoration of Atlantic and from the Connecticut River. Progressive Fish-Culturist 43:74–76. Shortnose sturgeon in Georgia. Final Report to the National Marine Fisheries Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel Service, Silver Spring, Maryland. inference: a practical information-theoretic approach, 2nd edition. Springer- Rothschild, B. J., and G. T. DiNardo. 1987. Comparison of recruitment vari- Verlag, New York. ability and life history data among marine and anadromous fishes. Pages Collins, M. R., S. G. Rogers, T. I. J. Smith, and M. L. Moser. 2000. Primary fac- 531–546 in M. J. Dadswell, R. J. Klauda, C. M. Moffitt, R. L. Saunders, R. A. tors affecting sturgeon populations in the southeastern United States: fishing Rulifson, and J. E. Cooper, editors. Common strategies of anadromous and mortality and degradation of essential habitats. Bulletin of Marine Science catadromous fishes. American Fisheries Society, Symposium 1, Bethesda, 66:917–928. Maryland. Dadswell, M. J. 1979. Biology and population characteristics of the Shortnose Schueller, P., and D. L. Peterson. 2010. Abundance and recruitment of juve- Sturgeon, Acipenser brevirostrum LeSueur 1818 (Osteichthyes: Acipenseri- nile Atlantic Sturgeon in the Altamaha River, Georgia. Transactions of the dae), in the Saint John River estuary, New Brunswick, Canada. Canadian American Fisheries Society 139:1526–1535. Journal of Zoology 57:2186–2210. Shaffer, M. L. 1981. Minimum population sizes for species conservation. Bio- Dadswell, M. J., B. D. Taubert, T. S. Squiers, D. Marchette, and J. Buckley. 1984. Science 31:131–134. Synopsis of biological data on Shortnose Sturgeon, Acipenser brevirostrum Smith, T. I. J., L. D. Heyward, W. E. Jenkins, and M. R. Collins. 1995. Culture LeSueur 1818. NOAA (National Oceanic and Atmospheric Administration) and stock enhancement of Shortnose Sturgeon, Acipenser brevirostrum, in Technical Report NMFS (National Marine Fisheries Service) 14. the southern United States. Pages 204–214 in A. D. Gershanovich and T. I. J. DeVries, R., and D. L. Peterson. 2008. Population dynamics and critical habitats Smith, editors. Proceedings of the international symposium on sturgeons, of the Shortnose Sturgeon, Acipenser brevirostrum, in the Altamaha River, 6–11 September 1993. VNIRO Publishing, Moscow. Georgia. Final Project Report to the National Marine Fisheries Service, Silver Squiers, T. S., M. Smith, and L. Flagg. 1982. American Shad enhancement Spring, Maryland. and status of sturgeon stocks in selected Maine waters. Final Report to the Flournoy, P.H., S. G. Rogers, and P.S. Crawford. 1992. Restoration of Shortnose National Marine Fisheries Service, Gloucester, Massachusetts. Sturgeon in the Altamaha River. Final Report to the U.S. Fish and Wildlife Vladykov, V. D., and J. R. Greeley. 1963. Order Acipenseroidei. Pages 24– Service, Georgia Department of Natural Resources, Atlanta. 60 in V. H. Olsen, editor. Fishes of the western North Atlantic, part Huggins, R. M. 1989. On the statistical analysis of capture experiments. III. Sears Foundation for Marine Research, Yale University, New Haven, Biometrika 76:133–140. Connecticut. Kynard, B. 1997. Life history, latitudinal patterns, and status of the Shortnose White, G. C., and K. P. Burnham. 1999. Program MARK: survival estima- Sturgeon, Acipenser brevirostrum. Environmental Biology of Fishes 48:319– tion from populations of marked animals. Bird Study 46(Supplement 1): 334. 120–139. NMFS (National Marine Fisheries Service). 1998. Final recovery plan for the Woodland, R. J., and D. H. Secor. 2007. Year-class strength and recovery of Shortnose Sturgeon (Acipenser brevirostrum). NMFS, Shortnose Sturgeon endangered Shortnose Sturgeon in the Hudson River, New York. Transactions

Downloaded by [Department Of Fisheries] at 21:38 27 October 2013 Recovery Team, Silver Spring, Maryland. of the American Fisheries Society 136:72–81. This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:39 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 Evidence for Density-Dependent Changes in Growth, Downstream Movement, and Size of Chinook Salmon Subyearlings in a Large-River Landscape William P. Connor a , Kenneth F. Tiffan b , John M. Plumb b & Christine M. Moffitt c 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 c U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit , University of Idaho , Moscow , Idaho , 83844 , USA Published online: 06 Sep 2013.

To cite this article: William P. Connor , Kenneth F. Tiffan , John M. Plumb & Christine M. Moffitt (2013) Evidence for Density- Dependent Changes in Growth, Downstream Movement, and Size of Chinook Salmon Subyearlings in a Large-River Landscape, Transactions of the American Fisheries Society, 142:5, 1453-1468, DOI: 10.1080/00028487.2013.806953 To link to this article: http://dx.doi.org/10.1080/00028487.2013.806953

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1453–1468, 2013 American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.806953

ARTICLE

Evidence for Density-Dependent Changes in Growth, Downstream Movement, and Size of Chinook Salmon Subyearlings in a Large-River Landscape

William P. Connor* U.S. Fish and Wildlife Service, Idaho Fishery Resource Office, 276 Dworshak Complex Drive, Orofino, Idaho 83544, USA Kenneth F. Tiffan and John M. Plumb U.S. Geological Survey, Western Fisheries Research Center, 5501A Cook-Underwood Road, Cook, Washington 98605, USA Christine M. Moffitt U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, Moscow, Idaho 83844, USA

Abstract We studied the growth rate, downstream movement, and size of naturally produced fall Chinook Salmon On- corhynchus tshawytscha subyearlings (age 0) for 20 years in an 8th-order river landscape with regulated riverine upstream rearing areas and an impounded downstream migration corridor. The population transitioned from low to high abundance in association with U.S. Endangered Species Act and other federally mandated recovery efforts. The mean growth rate of parr in the river did not decline with increasing abundance, but during the period of higher abundance the timing of dispersal from riverine habitat into the reservoir averaged 17 d earlier and the average size at the time of downstream dispersal was smaller by 10 mm and 1.8 g. Changes in apparent abundance, measured by catch per unit effort, largely explained the time of dispersal, measured by median day of capture, in riverine habitat. The growth rate of smolts in the reservoir declined from an average of 0.6 to 0.2 g/d between the abundance periods because the reduction in size at reservoir entry was accompanied by a tendency to migrate rather than linger and by increasing concentrations of smolts in the reservoir. The median date of passage through the reservoir was 14 d earlier on average, and average smolt size was smaller by 38 mm and 22.0 g, in accordance with density-dependent behavioral changes reflected by decreased smolt growth. Unexpectedly, smolts during the high-abundance period had begun to reexpress the migration timing and size phenotypes observed before the river was impounded, when

Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 abundance was relatively high. Our findings provide evidence for density-dependent phenotypic change in a large river that was influenced by the expansion of a recovery program. Thus, this study shows that efforts to recover native fishes can have detectable effects in large-river landscapes. The outcome of such phenotypic change, which will be an important area of future research, can only be fully judged by examining the effect of the change on population viability and productivity.

Juvenile growth patterns, the timing of downstream move- dance over time is one factor that can affect long-term changes ment, and corresponding body size are important phenotypic in growth, movement, and body size. Much of the understand- traits that vary widely among native salmonids across the water- ing of such density-dependent phenotypic responses comes sheds of northern latitudes. A steady increase in juvenile abun- from localized field studies in low-order streams coupled with

*Corresponding author: william [email protected] Received November 26, 2012; accepted May 13, 2013 Published online September 6, 2013 1453 1454 CONNOR ET AL.

laboratory work (e.g., Keeley 2001; Weber and Fausch 2003; In the decades before the completion of Lower Granite Dam Grant and Imre 2005). A finding for density-dependent growth and Reservoir, the fall Chinook Salmon population was rela- is a not a novel outcome in small-stream and laboratory stud- tively abundant and natural subyearling smolts passed down- ies, and it is well established that density-dependent interactions stream and entered the Columbia River (Figure 1) well before brought about by limited food and space in small streams can spring runoff was complete (Mains and Smith 1964). During alter the timing of downstream dispersal and the size at dis- the 1990s, Connor et al. (2002) reported that emergence be- persal (Chapman 1962; Stein et al. 1972; Keeley 2001). In the gan in early spring and that the natural parr reared in riverine absence of density-dependent pressures in a small stream, plots habitat prior to dispersing downstream into the reservoir before of the number of individuals present over time could have a midsummer. The natural parr then lingered in the low-velocity negatively skewed distribution because at low population lev- water of the reservoir as they grew and became actively migrat- els fish might gradually accumulate in rearing areas until they ing subyearling smolts that passed Lower Granite Dam (Connor eventually disperse downstream in response to environmental et al. 2003b). Large portions of the natural smolt population conditions and physiological maturation. Growth of the popu- passed the dam after spring runoff was complete, and some fish lation over time could result in a positively skewed distribution wintered in the reservoir and passed the dam as yearlings (age if fish accumulate rapidly, leading to increased competition that 1) the following spring (Connor et al. 2002). The large change causes many fish to disperse downstream early, leaving some in downstream passage timing caused by dam construction sub- fish to rear under lower densities. For the time-of-presence dis- jected the natural smolts to low flows and warm temperatures tributions to strictly adhere to the patterns described above, that were not favorable for survival. density-independent factors for movement would have to be Management efforts have been implemented in response constant as juvenile abundance increased. to ESA listing and other federally mandated mitigation pro- Cyclical variation in density-independent factors can have a grams to increase the size of the population and the survival large effect on phenotypic change in native juvenile salmonids. of Snake River basin fall Chinook Salmon; such efforts have Temperature is one such factor, as it influences fry emergence included reduced harvest (Peters et al. 2001), stable mini- timing as well as growth opportunity (Thorpe 1989) and hence mum spawning flows (Groves and Chandler 1999), summer size and size-related movement behaviors. A long-term warm- flow augmentation (Connor et al. 2003a), predator control ing trend in the aquatic environment could also change move- (Beamesderfer et al. 1996), increased hatchery production and ment timing by affecting behavioral thermoregulation (Kaya supplementation and improved dam passage structures (Rainey et al. 1977; Nielsen et al. 1994; Tiffan et al. 2009a). Variation et al. 2006), and court-ordered summer spill operations (CBR in stream flow could change the amount and quality of rearing 2012). The number of natural- and hatchery-origin adult fall habitat (Bjornn and Reiser 1991; Tiffan et al. 2002), so that a Chinook Salmon that passed upstream of Lower Granite Dam transition from a period of dry years to one of wet years (or vice and Reservoir to spawn increased from several hundred in 1992 versa) could in turn influence the growth, movement, and body to over 40,000 in 2010 (CBR 2012). The number of redds (i.e., size of juvenile salmonids. Likewise, changes in water veloc- spawning nests) counted within our Lower Snake River study ity related to climate-induced runoff patterns or anthropogenic area increased from 67 in 1994 to a high of 2,944 in 2010 activity could affect the variation in downstream movement tim- (Groves et al. 2013). The supplementation program that began ing by affecting the rate of downstream movement of both pas- in 1995 made off-station releases of up to 5 million subyearling sive and active migrants (McCormick et al. 1998; Tiffan et al. smolts (hereafter, hatchery smolts) throughout the drainage. Of 2009b). the management actions taken to increase abundance, the hatch- Unlike with small streams, relatively little is known about ery program had the greatest potential to increase the annual Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 density-dependent phenotypic change among native anadro- passage abundance of the basinwide population of subyearling mous salmonids inhabiting large-river landscapes. Over the smolts in the reservoir because hatchery smolts disperse rapidly 20-year period 1992–2011, we studied age-0 Chinook Salmon through riverine habitat and then concentrate in the reservoir, Oncorhynchus tshawytscha (hereafter, natural parr and smolts) where velocities and turbulence decline abruptly and form a that were produced by adults that entered freshwater during passage bottleneck (Smith et al. 2003; Tiffan et al. 2009a). We late summer and spawned through fall (i.e., fall-run fish) along focused on density-dependent phenotypic change in our analy- the lower Snake River. The lower Snake River is an 8th-order ses because the inevitable increase in juvenile abundance—to river regulated by Hells Canyon Dam that supports two of rely on a metaphor—was “the elephant in the room.” the four spawning aggregates of the fall-run Chinook Salmon The objectives of our study were to document and understand population upstream of Lower Granite Dam and Reservoir changes in (1) abundance during rearing in riverine habitat and (Figure 1) listed for protection under the U.S. Endangered throughout early seaward migration in the reservoir; (2) the Species Act (ESA; NMFS 1992). Adults from these two pop- growth rate of parr in riverine habitat (parr growth), the tim- ulation aggregates spawn from late October to early Decem- ing of downstream dispersal of parr into the reservoir (time of ber (Groves and Chandler 1999), with little annual variation in dispersal), and the size of parr at downstream dispersal (size at spawn timing. dispersal); and (3) the growth rate as the fish passed through the PHENOTYPIC CHANGE IN A LARGE RIVER 1455 Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 FIGURE 1. Map of the Snake River basin, including the free-flowing but regulated upper reach of the Snake River (Hells Canyon Dam to Salmon River mouth) and lower reach (Salmon River mouth to the upper end of Lower Granite Reservoir), where natural fall Chinook Salmon parr were captured while rearing during 1992–2011, as well as Lower Granite Reservoir, Lower Granite Dam, and Little Goose Dam, which were encountered by the fish after they had become smolts during early seaward migration. River flow was measured by gauges at Hells Canyon Dam, Anatone, Washington, and Troy, Oregon (black circles).

reservoir (smolt growth), time of passage through the reservoir METHODS (time of passage), and the size of smolts that passed through the reservoir (smolt size). We structured our analyses around two Abundance in Riverine Habitat research hypotheses concerning density dependence (Figure 2). We used a beach seine to collect fish at 11–15 permanent With respect to parr in riverine habitat, we hypothesized that stations located along 142 contiguous km of the two riverine increased abundance affected parr growth, the time of disper- reaches studied (see Connor et al. 2002). Large portions of the hatchery smolts released into the river were released without an sal, and size at dispersal (H1). Our second research hypothesis was that increased abundance affected smolt growth, the time external mark or fin clip (i.e., unmarked). After hatchery smolts were released upstream of a given seining station, the origin of passage, and smolt size in the reservoir (H2). 1456 CONNOR ET AL.

Recovery program expanded rates for fish recaptured within each year and interannual mean growth rates for each abundance period. A statistical comparison of parr growth was not conducted between abundance periods Juvenile abundance increased over time because large sample sizes and high statistical power would have made the comparison statistically significant. (H12 ) (H ) Time of dispersal.—To provide a measure of time of presence along the riverine shorelines, we assigned a day of capture to Competition increased in Reservoir concentrated each natural parr that was beach-seined (January 1 = day 1). riverine habitat migrants We then calculated annual date-of-capture distributions (10th, 25th, 50th, 75th, and 90th percentiles). We reasoned that time of Parr growth decreased Smolt growth decreased dispersal became earlier as the median day of capture became earlier and vice versa (after Connor et al. 2002). From the annual medians, we calculated interannual means for each of the two Earlier time of passage Earlier time of dispersal abundance periods. To evaluate skewness in each annual date- Low Low of-capture distribution, we calculated a skewness coefficient High High (S; SAS 2012). Date Date We used median date of capture (loge transformed to stabi- lize the variance) as the response variable in regression analyses to weigh the evidence for density-dependent change in time of Parr size decreased Smolt size decreased dispersal. We selected explanatory variables a priori. The an- nual mean CPUE of natural and hatchery fish combined was FIGURE 2. Potential density-dependent outcomes that would support research selected as a measure of relative abundance, and a set of tem- hypotheses H1 and H2 (see text), including theoretical examples (when all density-independent factors are constant) of shifts in dispersal and passage tim- perature and flow metrics was selected to index environmental ing shown by negatively skewed box plots (gray shading) under low abundance changes that could have had density-independent influences on and positively skewed box plots (white) under high abundance derived from fish movement and seining efficiency (Table 1). This and all time-of-presence and passage observations collected on individual fish. subsequent regression analyses were conducted by applying the Burnham and Anderson (2002) approach to model selection (i.e., natural or hatchery) of each collected unmarked fish was as modified by Connor and Tiffan (2012). To select the best classified based on morphology (overall accuracy, 98.7%; Tiffan candidate model, we applied a second-order bias correction to and Connor 2011). the Akaike information criterion (AICc). To rank the candidate We calculated the number of natural parr, hatchery smolts, regression models, we calculated simple differences in AICc and total subyearlings that were captured per seine haul (CPUE) values between the model with the lowest AICc (i.e., the best at each station by week and averaged across stations and weeks model) and the other models. We made partial regression plots within each year to calculate annual mean CPUE (N = station (Moya-Larano˜ and Corcobado 2008) for the explanatory vari- visits per year) as an index of apparent abundance in riverine ables in Table 1 that were included in the best model, and the r2 rearing habitat. For purposes of comparison and statistical anal- values were compared among the plots to weigh the evidence ysis, we then divided the years into a period of high abundance for density-dependent change. and a period of low abundance based on interannual trends. In Size at dispersal.—We calculated the annual mean FLs (mm) this and all subsequent analyses, data were pooled irrespective and weights (0.1 g) of all natural parr captured in the riverine Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 of riverine reach. In both text and tables, means are reported ± habitat, reasoning that decreases in size reflected decreases in SDs and means of means are reported ± SEs. size at dispersal, and vice versa. We also calculated the inter- annual mean FLs and weights for each of the two abundance H1: Increased Abundance in Riverine Habitat Affected periods. We plotted weight against FL for individual fish to de- Parr Growth, Dispersal, and Size termine whether these size metrics varied by abundance period. Parr growth.—During 1992–2007, we implanted each natu- A statistical comparison of dispersal size and length–weight re- ral parr ≥60 mm long collected during seining in riverine habi- lationships was not conducted between abundance periods for tat with an 11.5-mm passive integrated transponder (PIT) tag the reason given in the case of parr growth. (Prentice et al. 1990a). During 2008–2011, in addition to tag- ging fish longer than 59 mm with 11.5-mm tags, we tagged Abundance in the Reservoir natural parr that were 50–59 mm long with 8.5-mm tags. Natu- The basinwide population of subyearling smolts consists ral parr were given 15 min to recover from tagging in an aerated of the aggregate of natural and hatchery fish produced or re- 19-L bucket of river water before release at their capture site. leased upstream of Lower Granite Reservoir. The turbine in- We calculated the absolute growth rates (g/d) of individual takes and spillways were the primary routes of smolt passage at PIT-tagged natural parr recaptured by beach seine as (weight2 − the dam. A portion of the basinwide subyearling smolt popula- weight1)/(day2 − day1). We calculated annual mean growth tion that entered the turbine intakes was routed away from the PHENOTYPIC CHANGE IN A LARGE RIVER 1457

TABLE 1. Candidate explanatory variables analyzed to weigh the evidence for research hypotheses H1 and H2. All variables under H2 were measured at the level of individual fish.

Explanatory variable Date range Source

H1: increased abundance in riverine habitat affected parr growth, dispersal, and size Annual mean CPUE (fish/haul) Spring to summer Field notes Annual mean temperature (◦C) 27 Oct–20 Jun Thermographs Annual maximum spring temperature 20 Mar–20 Jun Thermographs Annual day of maximum spring temperature Year dependent Thermographs Annual mean spring flow (1,000 m3/s) 20 Mar–20 Jun Figure 1 Annual maximum spring flow 20 March–20 June Figure 1 Annual day of maximum spring flow Year dependent Figure 1

H2: increased abundance in the reservoir affected smolt growth, passage, and size Weight at capture (0.1 g) Day of capture PTAGIS (2012) Cumulative abundance calculated as the sum of the daily Day of capture to detection at Lower Granite Dam FPC (2012) passage abundances estimated for the basinwide population of subyearlings Days at large Day of capture to detection at Lower Granite Dam PTAGIS (2012) Mean reservoir temperature (◦C) Day of capture to detection at Lower Granite Dam CBR (2012) Mean reservoir flow (1,000 m3/s) Day of capture to detection at Lower Granite Dam CBR (2012) Rate of reservoir warming (◦C /week) Day of capture to day of maximum ◦C CBR (2012)

powerhouse and into a bypass by submersible bar screens. recorded the date any fish that we had tagged in the river was Timed daily subsamples of these bypassed fish were taken bypassed (Prentice et al. 1990b). Such “detection” data were col- and counted each year. Natural smolts were not distinguished lected 24 h/d from spring through early winter (PTAGIS 2012; from hatchery smolts in these counts. The combined timed daily PTOC 2012). We used electronic diversion devices (Marsh et al. counts of natural and hatchery smolts were divided by the daily 1999; Downing et al. 2001) located within the bypass to recap- sampling rate to estimate the number of smolts that entered the ture random subsamples of the tagged natural smolts after de- bypass (FPC 2012). tection during 1992–1995 and 2005–2011. We also recaptured We used the daily estimates of the number of subyearling tagged natural smolts detected at Little Goose Dam (Figure 1), smolts that entered the bypass to estimate the daily and an- which was also equipped with PIT-tag detection and recapture nual passage abundances at Lower Granite Dam for the basin- devices, during 1996–1998. We measured and weighed the re- wide population of subyearling smolts, as generally described captured natural smolts within 48 h of initial detection at each by Plumb et al. (2012). We bootstrapped 99.9% confidence dam. To compile the data set for subsequent analyses, we queried limits (Efron and Tibshirani 1998) on each estimate of annual the regional database to identify natural smolts detected and re- passage abundance. We then examined the results to determine captured at Lower Granite Dam and natural smolts that had been whether the low- and high-abundance periods identified in the recaptured at Little Goose Dam after detection at Lower Gran- CPUE data for natural parr in riverine habitat were evident in ite Dam. If these conditions were not met, we removed the fish Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 the annual passage abundances for the basinwide population from analyses because a detection date at Lower Granite Dam of smolts in Lower Granite Reservoir. Since origin was not was mandatory for subsequent analyses on time of passage. distinguished in the smolt counts at the dam, and it was well We calculated absolute growth rates for each PIT-tagged nat- beyond our resources to develop and apply a method for es- ural smolt we recaptured as described for natural parr in riverine timating annual passage abundance especifically for hatchery habitat. The natural smolts we recaptured were not recaptured smolts, we regressed the annual number of hatchery smolts re- in the exact proportion required to fully represent the growth leased upstream of the dam against the estimated annual passage and time of passage of the all the PIT-tagged natural smolts abundance of the basinwide population of subyearling smolts at that passed Lower Granite Dam. To best represent this popula- the dam to assess the influence of hatchery releases on annual tion of PIT-tagged fish with the recapture data during statistical passage abundance. analyses, we created a daily weighting factor for a given recap- tured fish that was equal to the estimate of passage abundance H2: Increased Abundance in the Reservoir Affected made for all tagged fish on the day that the recaptured fish was Smolt Growth, Passage, and Size detected. We used the model of Plumb et al. (2012) to esti- Smolt growth.—Lower Granite Dam was equipped with a mate daily passage abundance of PIT-tagged natural smolts as PIT-tag detection system located within the bypass flumes that described for the basinwide population of smolts, except that 1458 CONNOR ET AL.

instead of inputting daily bypass count data into the model we information on smolt size. To account for the growth of the nat- input the daily PIT-tag detection count data. ural smolts recaptured at Little Goose Dam after they passed We calculated annual mean growth rates from the absolute Lower Granite Dam, we multiplied the absolute growth rates growth rates of the individual PIT-tagged natural smolts we re- (mm/d and g/d) of each fish by the number of days the fish captured. We then calculated interannual mean growth rates for was at large between dams. The resulting lengths and weights each of the two abundance periods. We square root transformed were subtracted from the FL and weight of the fish recorded at the growth rates to stabilize the variance, and then used a mixed Little Goose Dam to assign a FL and weight to the fish when it model (Littell et al. 1996) with abundance period as the fixed was detected and passed Lower Granite Dam. We calculated the term and natal reach (i.e., upper or lower) and the period × reach annual mean FLs and weights of all recaptured natural smolts interaction as random terms to determine whether the overall and interannual means for each of the two abundance periods, weighted mean growth in the reservoir (i.e., fish pooled across and then plotted observed weights against observed FLs for in- years within each period) varied between abundance periods dividual recaptured fish to determine whether these size metrics (α = 0.05). We then weighed the evidence for density-dependent varied by abundance period. variation in growth in the reservoir by conducting weighted re- gression analyses on the square root–transformed growth rates RESULTS using the approach described for analyses of dispersal timing from the river, except that this analysis was conducted with Abundance in Riverine Habitat explanatory variables measured at the level of individual fish The observed trends in CPUE confirmed the expectation instead of groups of fish over a year. The explanatory vari- that the abundance of natural parr would increase over time in ables were selected a priori to index both the environmental and riverine habitat in response to the expanding recovery program biological conditions experienced by each recaptured fish, in- (Table 2). Annual mean CPUE was always higher for natural cluding an index of the abundance of the basinwide population parr than for hatchery smolts because the hatchery smolts spent of subyearlings that was anticipated under H2 to be inversely little time in the river. There was over a threefold increase in proportional to growth in the reservoir (Table 1). annual mean CPUE for natural parr and for natural and hatchery Time of passage.—To provide a measure of time of passage fish combined between 1999 and 2000. The annual mean CPUE through Lower Granite Reservoir and past Lower Granite Dam for the PIT-tagged natural smolts, we used the daily passage TABLE 2. Annual mean catch per unit effort (CPUE; fish/seine haul) in river- abundance estimates for these fish described above to calculate ine habitat along the lower Snake River for natural fall Chinook Salmon parr annual time of passage distributions. We reasoned that time of and hatchery fall Chinook Salmon subyearling smolts separately and combined, 1992/1995–2011. No hatchery smolts were released during 1992–1994. passage through the reservoir became earlier as the estimated median day of passage became earlier, and vice versa (after CPUE Connor et al. 2002). From the estimated annual medians, we calculated interannual means for each of the two abundance Year Hauls Natural Hatchery Combined periods. We then calculated and interpreted annual skewness 1992 175 3.9 ± 0.5 3.9 ± 0.5 coefficients as described for capture date in the river. We also 1993 247 3.6 ± 0.4 3.6 ± 0.4 estimated the daily passage abundance of PIT-tagged natural 1994 249 6.1 ± 0.9 6.1 ± 0.9 yearling migrants that had overwintered in the reservoir. 1995 269 4.7 ± 0.5 0.2 ± 0.1 4.9 ± 0.5 All subsequent analyses were conducted on the PIT-tagged 1996 199 2.1 ± 0.2 0.4 ± 0.1 2.6 ± 0.3 natural smolts we recaptured, as the absolute growth rates and ± ± ± Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 1997 238 3.0 0.4 2.9 0.6 5.9 0.8 sizes of the recaptured fish were needed to fully evaluate the 1998 222 9.2 ± 1.0 1.3 ± 0.3 10.5 ± 1.1 weight of evidence for the passage timing and smolt size portions 1999 252 6.1 ± 0.7 0.3 ± 0.1 6.3 ± 0.7 of H2. We weighted the detection date of each recaptured natural 2000 139 21.6 ± 4.0 0.1 ± 0.1 21.7 ± 4.0 smolt as described for the analyses of growth in the reservoir. A 2001 189 11.7 ± 1.6 1.7 ± 0.5 13.4 ± 1.8 weighted detection for a particular recaptured natural smolt on 2002 216 17.7 ± 3.1 0.3 ± 0.1 18.0 ± 3.1 a given day was equivalent to a passage date. We weighed the 2003 198 29.4 ± 3.4 1.4 ± 0.3 30.8 ± 3.4 evidence for density-dependent variation in time of passage by 2004 205 56.6 ± 8.2 1.0 ± 0.4 57.6 ± 8.2 conducting weighted regression analyses on the day of passage 2005 259 52.8 ± 6.6 4.0 ± 1.0 56.8 ± 7.0 of individual recaptured fish using the approach described for 2006 294 7.3 ± 0.8 5.9 ± 0.7 13.2 ± 1.4 analyses of smolt growth. We analyzed the explanatory variables 2007 212 38.2 ± 6.9 3.7 ± 1.6 41.9 ± 7.1 in Table 1, except that weight at capture was replaced with day 2008 261 20.5 ± 3.0 12.8 ± 3.8 33.2 ± 5.7 of capture and cumulative abundance was replaced with smolt 2009 234 20.0 ± 3.2 8.7 ± 2.6 28.7 ± 5.0 growth. 2010 273 27.1 ± 2.4 4.9 ± 1.5 32.1 ± 3.1 Smolt size.—The length and weight data for the PIT-tagged 2011 240 17.5 ± 2.1 3.1 ± 1.0 20.6 ± 2.6 natural smolts we recaptured at Lower Granite Dam provided PHENOTYPIC CHANGE IN A LARGE RIVER 1459

for natural and hatchery fish combined during 1992–1999 never 1992 N= 2,121 S = -0.52 M = 135 exceeded that for the combined group of fish during 2000–2011. 1993 N= 2,415 S = -0.56 M = 160 Based on these trends, we divided the years studied into a low- 1994 N= 4,789 S = -0.65 M = 151 1995 N= 2,767 S = -0.56 M = 151 abundance period (1992–1999) and a high-abundance period 1996 N= 1,156 S = 0.53 M = 142 (2000–2011). Thus, the number of years used to calculate the 1997 N= 1,172 S = -0.47 M = 161 1998 N= 4,013 S = 0.06 M = 141 interannual means for the low- and high-abundance periods with 1999 N= 3,515 S = -0.16 M = 147 data collected in riverine habitat were 8 and 12, respectively. 2000 N= 3,740 S = 0.43 M = 123 2001 N= 5,690 S = -0.12 M = 135

Year 2002 N= 9,887 S = 0.50 M = 129 H1: Increased Abundance in Riverine Habitat Affected 2003 N= 14,032 S = -0.06 M = 128 Parr Growth, Dispersal, and Size 2004 N= 28,615 S = 0.24 M = 127 2005 N= 34,170 S = 0.14 M = 131 Parr growth.—Parr growth in the river varied annually but did 2006 N= 4,672 S = -0.18 M = 138 not differ between the abundance periods (Figure 3), providing 2007 N= 20,809 S = 0.41 M = 128 2008 N= 12,826 S = 0.13 M = 136 no evidence for the growth portion of research hypothesis H1. 2009 N= 10,410 S = -0.01 M = 134 Time of dispersal.—Analyses of the time of natural parr dis- 2010 N= 17,660 S = 0.16 M = 133 persal from the river and into the reservoir provided evidence 2011 N= 8,183 S = 0.37 M = 137 for the movement portion of research hypothesis H1. With the 100 110 120 130 140 150 160 170 180 190 exception of 1992, the date of capture distributions showed that Day of year time of dispersal was later during the low-abundance period than FIGURE 4. Annual capture date distributions for natural fall Chinook Salmon during the high-abundance period (Figure 4). As a result of the parr during the low (1992–1999; gray box plots) and high (2000–2011; white differences in the date-of-capture distributions, the interannual box plots) abundance periods, including the total number of natural subyearlings mean median day of capture was 17 d later during the low- captured (N), the skewness coefficient (S), and the median day of capture (M) abundance period (149 ± 9) than during the high-abundance for each year. The vertical lines represent the medians, the box dimensions the 25th to 75th percentiles, and the whiskers the 10th to 90th percentiles. period (132 ± 5). During 6 of the 8 years of the low-abundance period, the skewness coefficient for the capture date distribu- tions was negative, providing some evidence that in most years natural subyearlings gradually accumulated along the shorelines results of regression analyses on annual median day of cap- and lingered during rearing prior to dispersing downstream into ture further supported research hypothesis H1. Of the regres- the reservoir. During 8 of the 12 years of the high-abundance sion models fitted to weigh the evidence for density-dependent period, the skewness coefficient for the annual distributions of dispersal timing, 15 had AICc values less than 10 and were week of capture was positive, providing some evidence that accepted as plausible. Four regression models were equally in- in most years natural parr accumulated more rapidly than dur- formative ( AICc values 0–1.8). Of these four models, three ing the low-abundance period and then many natural parr be- included combinations of CPUE, maximum spring river tem- ◦ gan downstream dispersal early, leaving some fish behind to perature (Max C), and mean spring river flow (Flow), and one continue rearing. was fitted from CPUE alone. For the sake of brevity, we report After environmental variation (which can influence fish be- the results from the model fitted from all three of the above 2 havior and beach seining efficiency) was accounted for, the explanatory variables (R = 0.64; N = 20; P = 0.0007):

0.5 log (annual median day of capture) Low abundance period High abundance period e 106 ◦ Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 261 Mean = 0.2 + 0.1 g/d Mean = 0.2 + 0.1 g/d = 5.164 − 0.003 · CPUE − 0.012 · Max C 0.4 238 64 N = 8 years N = 12 years 66 98 + 0.031 · Flow. 175 98 0.3 36 354 545 571 553 938 161 131 572 2 288 The r value for the annual mean CPUE partial regression model 2 0.2 386 943 was roughly five times higher than the r values for the maxi- mum spring river temperature and mean spring river flow par- 0.1

Annual mean growth rate (g/d) rate growth mean Annual tial regression models (Figure 5). The annual median day of capture was inversely proportional to annual mean CPUE, and 0.0 the partial regression plot indicates that time of dispersal gen- 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 erally became earlier as apparent abundance along the river Year shorelines increased between abundance periods (Figure 5a). The annual median day of capture was inversely proportional FIGURE 3. Annual mean growth rates of natural fall Chinook Salmon parr in riverine rearing habitat along the lower Snake River during the low (1992–1999) to annual maximum spring river temperature, indicative of and high (2000–2011) abundance periods. The whiskers represent SDs; sample temperature-related downstream dispersal that was common be- sizes of PIT-tagged fish are given above the whiskers. tween the two abundance periods (Figure 5b), as there was little 1460 CONNOR ET AL.

0.2 50 2 (a) r = 0.46 Low abundance period (N = 18,395) 40 0.1 N = 20 High abundance period (N = 153,314) P = 0.001 0.0 30 20 -0.1 Weight (g) 10 -0.2 -30.0 -20.0 -10.0 0.0 10.0 20.0 30.0 0 Adjusted mean CPUE 20 30 40 50 60 70 80 90 100 110 120 130 140 150 0.15 Fork length (mm) (b) r 2 = 0.07 0.10 FIGURE 6. Scatterplot showing the relationship between weight and FL for P = 0.3 natural fall Chinook Salmon parr captured in riverine habitat along the lower 0.05 Snake River during the low (1992–1999; gray circles) and high (2000–2011; 0.00 white circles) abundance periods. -0.05 -0.10 Size at dispersal.—The change in time of dispersal of natu- -0.15 ral parr from the river and into the reservoir associated with -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 increases in apparent abundance resulted in a shorter time spent rearing and growing in riverine habitat during the high- Adjusted maximum river temperature abundance period. Consequently, and in support of the size Adjusted median day of capture day of median Adjusted 0.15 portion of research H1, natural parr were 10 mm longer and 0.10 1.8 g heavier on average ± 95% confidence limit [CL] at the time of downstream dispersal from the river during the low- 0.05 abundance period (65 ± 1 mm; 3.8 ± 0.1 g) than during the 0.00 high-abundance period (55 ± 1 mm; 2.0 ± 0.1 g). Although -0.05 natural parr were larger on average during the low-abundance (c) r 2 = 0.10 period, we found similar fish weights at any given FL between -0.10 P = 0.2 abundance periods (Figure 6) because growth in the river did -0.15 not vary between abundance periods. Although there were pro- -0.8 -0.5 -0.3 0.0 0.3 0.5 0.8 portionately more large natural parr during the low-abundance Adjusted mean river flow period, the presumably competitive individuals during the high- abundance period that remained along the shoreline after early FIGURE 5. Partial regression plots showing (a)–(c) the relationships inherent dispersal of their smaller cohort members grew to large FLs and = · in the regression model loge (annual median day of capture) 5.164 – 0.003 weights. CPUE – 0.012 · maximum spring river temperature + 0.031 · mean spring river flow fitted to explain the variation in dispersal timing of natural fall Chinook Salmon parr from the lower Snake River during the low (1992–1998; gray Abundance in the Reservoir Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 circles) and high (2000–2011; white circles) abundance periods. The trends in estimated annual passage abundance of the bas- inwide population of subyearling smolts in the reservoir con- firmed a general increase in abundance in the reservoir during 1992–2011, while supporting the delineation of study years into difference in river temperature between abundance periods (see the low- and high-abundance periods (Figure 7). The estimated Supplementary Table 1 in the online version of this article). annual passage abundance generally increased from a low of The annual median day of capture was directly proportional to 13,777 (99.9% CI, 12,867–21,745) fish in 1992 to a high of annual mean spring river flow, suggestive of flow-related fish 3,764,636 (3,405,621–6,298,873) fish in 2005. Estimated pas- dispersal or seining efficiency. Although spring (Mar 20–Sep sage abundance increased fourfold between 1999 and 2000 and 20) flows were somewhat higher during the low-abundance pe- exceeded 1 million fish between 2000 and 2011. The number riod, the difference in median day of capture persisted under of hatchery smolts released upstream of Lower Granite Reser- the high spring flow conditions common to both abundance pe- voir (annual ranges; low-abundance period: 0–713,800; high- riods (Supplementary Table 1). Thus, the effect of flow on the abundance period: 1,516,992–5,139,320) explained 67% of the behavior of natural parr or beach seining was common between annual variation in annual passage abundance of the basinwide abundance periods (Figure 5c). population of subyearling smolts, conveying the substantial PHENOTYPIC CHANGE IN A LARGE RIVER 1461

7 1.2 150 Low abundance period Basin-wide passage abundance 105 Mean = 0.6 + 0.1 g/d 6 Hatchery fish released 1.0 N = 7 years r2 = 0.67 179 116 32 31 5 High abundance period N = 20 0.8 P < 0.0001 Mean = 0.2 + 0.1 g/d 4 22 N = 7 years 0.6 55 15 14 103 3 147 0.4 168 46 2 Number of fish (millions) fish of Number Mean growth rate (g/d) rate growth Mean 0.2 1

0 0.0 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 1992 1993 1994 1995 1996 1997 1998 2005 2006 2007 2008 2009 2010 2011 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 Year Year FIGURE 8. Annual weighted mean growth rates of PIT-tagged natural fall FIGURE 7. Estimated passage abundance of the basinwide population of fall Chinook Salmon smolts recaptured after passing Lower Granite Dam during Chinook Salmon subyearling smolts (natural and hatchery origin combined) at portions of the low (1992–1998) and high (2005–2011) abundance periods. The Lower Granite Dam during 1992–2011; the whiskers indicate the 99.9% confi- whiskers represent SDs; sample sizes of PIT-tagged fish are given above the dence intervals. The number of subyearling hatchery smolts released upstream whiskers. of the dam and the coefficient of determination and P-value for the regression model that predicted abundance from the number of hatchery smolts released are also shown. Since the Plumb et al. (2012) collection probability model was The best weighted regression model for assessing the ev- fitted with data collected after 12-m screens were installed in all six turbine idence for density-dependent smolt growth (R2 = 0.77; N = intakes at Lower Granite Dam, we used information collected by Swan et al. < (1990) to calculate correction factors that we multiplied by the daily collection 1,181; P 0.0001) was much more informative than the second- probabilities estimated for 1992–1995 to account for the decrease in the number best model (AICc value, 195). The best model was fitted of fish routed into the bypass when shorter screens were in place in all or some from weight at initial capture in the river (Weight), days at of the turbine intakes. The correction factors were 0.868 for 1992–1994 and large (DAL), mean reservoir water temperature (Reservoir◦C), 0.890 for 1995. and cumulative passage abundances of basinwide subyearling smolts (Abundance) measured between the date on which each natural smolt was captured in the river and the date on which effect of the expanding hatchery program on subyearling smolt it was detected at Lower Granite Dam. The interannual means abundance in the reservoir (Figure 7). of the above explanatory variables generally reflect the differ- ences in the environmental and biological experiences of natural H2: Increased Abundance in the Reservoir Affected Smolt smolts between the low- and high-abundance periods (Table 3). Growth, Passage, and Size During the high-abundance period, natural smolts weighed less Smolt growth.—Analyses of natural smolts strongly sup- when they were initially captured as parr rearing in the river ported the growth portion of H2. The mean weighted growth due to the density-dependent responses described for H1;were rate of natural smolts in the reservoir varied annually, and in at large for fewer days in the reservoir, which is indicative of contrast to parr growth in riverine habitat, the interannual mean a density-dependent change in migration behavior; experienced growth rate of smolts was higher during the low-abundance pe- cooler reservoir water temperatures; and experienced markedly riod than during the high-abundance period (Figure 8). The over- higher cumulative passage abundances of basinwide subyear- Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 all weighted mean growth rate of natural smolts (low-abundance ling smolts. Natural smolts experienced cooler reservoir tem- period: 0.6 ± 0.4 g/d, N = 628 fish; high-abundance period: peratures not because the temperature regime changed between 0.2 ± 0.3 g/d, N = 555 fish) varied significantly (P = 0.03) abundance periods (see Supplementary Table 2 in the online ver- between abundance periods. sion of this article), but because they entered the reservoir early

TABLE 3. Interannual means for the explanatory variables in the best weighted regression model fitted to weigh the evidence for density-dependent growth in Lower Granite Reservoir of PIT-tagged natural fall Chinook Salmon smolts recaptured after passing Lower Granite Dam during portions of the low (1992–1998) and high (2005–2011) abundance periods.

Number of Weight at Days at Reservoir Cumulative passage Period years capture (g) large temperature (◦C) abundance Low 7 6.0 ± 0.3 43 ± 3 16.4 ± 0.5 20,437 ± 5,422 High 7 3.0 ± 0.3 33 ± 1 12.9 ± 0.4 1,072,933 ± 206,785 Difference 3.0 10 3.5 –1,052,496 1462 CONNOR ET AL.

0.60 0.60 0.40 0.40 0.20 0.20 0.00 0.00 -0.20 (a) r 2 = 0.23 -0.20 (b) r 2 = 0.26 N -0.40 = 1,181 -0.40 P < 0.0001 P < 0.0001 -0.60 -0.60 -10.0 -5.0 0.0 5.0 10.0 15.0 -40.0 0.0 40.0 80.0 Adjusted weight at capture Adjusted days at large 0.60 2 0.60 (c) r = 0.19 r 2 0.40 P (d) = 0.22 < 0.0001 0.40 P < 0.0001 Adjusted growth rateAdjusted 0.20 0.20 0.00 0.00 -0.20 -0.20 -0.40 -0.40 -0.60 -0.60 -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 -2.0 -1.0 0.0 1.0 2.0 3.0 Adjusted reservoir temperature Adjusted abundance √ FIGURE 9. Partial regression plots showing (a)–(d) the relationships inherent in the weighted regression model growth rate = –0.207 + 0.030 · weight at capture + 0.004 · days at large in the reservoir + 0.033 · mean reservoir temperature − 0.00000004 · cumulative passage abundance fitted to explain the variation in growth rate (g/d) of PIT-tagged natural fall Chinook Salmon smolts in Lower Granite Reservoir during portions of the low (1992–1998; gray circles) and high (2005–2011; white circles) abundance periods.

and spent little time rearing in that environment when reservoir 1992 N^= 85 S = -1.35 M = 174 N^= 0 temperatures increased later in the summer. 1993 N^= 483 S = 0.44 M = 203 ******* N^= 31 6.03% 1994 N^= 450 S = 0.98 M = 198 *** N^= 14 3.02% ^ * ^ The equation for the best weighted regression model for 1995 N= 889 S = 0.98 M = 207 N= 0 ^ ^ weighing the evidence for density-dependent smolt growth was 1996 N= 352 S = 0.79 M = 197 N= 0 1997 N^= 322 S = 0.99 M = 193 ***** N^= 0 √ 1998 N^= 1,291 S = 1.56 M = 190 **** N^= 20 1.53% 1999 N^= 1,268 S = 1.29 M = 197 N^= N/A Smolt growth rate ^ ^ 2000 N= 733 S = 1.94 M = 183 **** N= 12 1.61% =−0.207 + 0.030 · Weight + 0.004 · DAL + 0.033 2001 N^= 394 S = 1.84 M = 188 No detection N^= 0 2002 N^= 1,463 S = 1.59 M = 182 data N^= 9 0.61% ◦ Year *** · Reservoir C − 0.00000004 · Abundance. 2003 N^= 2,332 S = 1.30 M = 178 * N^= 3 0.13% 2004 N^= 3,238 S = 2.42 M = 176 N^= 4 0.12% ^ ** ^ 2005 N= 3,990 S = 0.45 M = 174 * N= 3 0.08% Smolt growth rate was directly proportional to weight at cap- 2006 N^= 755 S = 1.19 M = 175 N^= 0 2007 N^= 1,476 S = 2.38 M = 175 **** * N^= 17 1.14% Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 ture, days at large, and mean reservoir temperature and inversely 2008 N^= 2,341 S = 0.04 M = 188 * N^= 4 0.17% proportional to cumulative passage abundance (Figure 9). Af- 2009 N^= 1,425 S = 0.53 M = 182 * N^= 9 0.63% 2010 N^= 2,279 S = 1.22 M = 190 N^= 0 ^ ^ ter accounting for the variation in the other explanatory vari- 2011 N= 1,123 S = 0.91 M = 184 *** N= 6 0.53% ables, the explanatory variable days at large in the reservoir 150 180 210 240 270 300 330 360 25 55 85 115 145 175 was most highly correlated with growth rate, followed closely Day of year by weight at capture and cumulative passage abundance (Fig- ure 9). Thus, variation in smolt growth reflected and subsumed FIGURE 10. Estimated annual passage date distributions for PIT-tagged nat- ural fall Chinook Salmon smolts during the low (1992–1999; gray box plots) density-dependent mechanisms, including size at downstream and high (2000–2011; white box plots) abundance periods, including the total dispersal from the river and into the reservoir, movement behav- number of subyearling smolts estimated to have passed Lower Granite Dam ior after reservoir entry, and the concentration of both natural (N), the skewness coefficient (S), and median day of capture (M) for each year. and hatchery fish in the reservoir caused by impoundment and No detection data were collected when the bypass was not operated (see PTOC the expansion of the hatchery program. 2012 for dates). The asterisks represent the passage dates of fish that overwin- tered in the reservoir and are followed by the estimated number of such fish Time of passage.—The estimated annual time of passage and the percentage of annual total passage abundance that they represent (fish distributions at Lower Granite Dam for natural smolts provided tagged in 1999 that passed in 2000 could not be detected due to equipment evidence for the movement portion of H2 (Figure 10). With the upgrades). See Figure for additional details. PHENOTYPIC CHANGE IN A LARGE RIVER 1463

exception of natural smolts in 1992, which had the same median the flow regime varied markedly between abundance periods date of estimated passage as natural smolts in 2005, the median (Supplementary Table 2) but because they entered the reservoir date of estimated passage through the reservoir was later during early and passed through it quickly before reservoir flows fell every year of the low-abundance period than during the high- to low summer levels. abundance period. Furthermore, the interannual mean median The equation for the best weighted regression model for day of passage was 14 d later during the low-abundance period weighing the evidence for density-dependent time of passage of (195 ± 10) than during the high-abundance period (181 ± natural smolts was 6). The time-of-passage distributions were protracted during 6 √ of the 8 years of the low-abundance period, whereas the dis- Day of passage tributions were compressed during 8 of the 12 years of the = 12.415 + 0.011 · Day of capture − 0.488 · Delta◦C high-abundance period. Of the 96 possible interannual compar- − 0.187 · Flow + 1.597 · Smolt growth rate. isons between abundance periods, 67 (69.8%) showed that the time-of-passage distributions were less positively skewed dur- Time of passage through the reservoir was directly proportional ing the low-abundance period than during the high-abundance to day of capture and smolt growth rate and indirectly propor- period. Natural fish that overwintered in the reservoir and passed tional to rate of reservoir warming and mean reservoir flow the dam the following spring as yearling smolts made up small (Figure 11). After the variation in the other explanatory vari- percentages of the estimated annual passage abundances. The ables was accounted for, the smolt growth rate was most highly expression of this overwintering tactic was slightly higher on correlated with passage timing. average during the low-abundance period (1.51%) than during Smolt size.—The density-dependent changes in movement the high-abundance period (0.42%). Together, these time-of- behavior in riverine habitat and the reservoir reflected in smolt passage results reinforce the fact that during the low-abundance growth, and consequently in passage timing and the tempera- period natural fish tended to linger in the reservoir and pass tures experienced by the natural smolts in the reservoir, resulted downstream more slowly in cohorts spread out over time than in a large change in smolt length and weight between abundance did natural fish during the high-abundance period. periods that supported the size portion of H . Natural smolts The best weighted regression model for assessing the evi- 2 were 38 mm longer and 22.0 g heavier on average ( ± 95% CL) dence for density-dependent time of passage of natural smolts when passing Lower Granite Dam during the low-abundance through the reservoir (R2 = 0.55; N = 1,181; P < 0.0001) was period (N = 7 years; 132 ± 11 mm; 31.0 ± 8.5 g) than during much more informative than the second-best model (AIC c the high-abundance period (N = 7 years; 94 ± 4 mm; 9.0 ± value, 55). The best model was fitted from day of capture in ◦ 1.0 g). In contrast to the case with fish in the river, there was the river, rate of reservoir warming (Delta C), mean reservoir a marked difference in the observed FL–weight relationships flow (Flow), and smolt growth rate. The interannual means for of natural smolts between abundance periods (Figure 12). On the above explanatory variables help to illustrate the differences average, natural smolts had greater weight at a given FL dur- and similarities in the environmental and biological experiences ing the low-abundance period than during the high-abundance of natural smolts between the two abundance periods (Table 4). period. Only one natural smolt was smaller than 90 mm during Compared with natural smolts during the low-abundance period, the low-abundance period. Few natural smolts (n = 4) exceeded natural smolts during the high-abundance period were captured 125 mm during the high-abundance period. as parr in riverine habitat earlier on average due to the density- dependent responses (H1 analyses), experienced similar rates of reservoir warming but higher levels of reservoir flow, and grew DISCUSSION Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 at lower rates in the reservoir largely in response to the density- The growth rates of natural parr in riverine rearing habi- dependent mechanisms identified by the best growth rate regres- tat did not change as parr abundance increased but, consistent sion model reported above. During the high-abundance period, with H1, the timing of downstream dispersal into the reser- natural smolts experienced higher reservoir flows not because voir was earlier and fish were smaller at dispersal during the

TABLE 4. The inter-annual means for the explanatory variables in the best weighted regression model fitted to weigh the evidence for density-dependent passage timing through Lower Granite Reservoir of PIT-tagged natural fall Chinook Salmon smolts recaptured after passing Lower Granite Dam during portions of the low (1992–1998) and high (2005–2011) abundance periods.

Number of Day of Mean reservoir flow Growth in the Period years capture Delta◦C (1,000 m3/s) reservoir (g/d) Low 7 153 ± 30.8± 0.1 2.390 ± 0.397 0.6 ± 0.0 High 7 139 ± 20.9± 0.1 3.212 ± 0.367 0.2 ± 0.0 Difference 14 –0.1 –0.822 0.4 1464 CONNOR ET AL.

3 3 2 2 1 1 0 0 -1 (a) r 2 = 0.04 -1 2 -2 N = 1,181 -2 (b) r = 0.01 P < 0.0001 P = 0.0007 -3 -3 -40-30-20-100 1020304050-1.0 -0.5 0.0 0.5 1.0 Adjusted day of capture Adjusted DeltaO C 3 3 (c) r 2 = 0.10 2 P < 0.0001 2 Adjusted day of passage of day Adjusted 1 1 0 0 -1 -1 r2 -2 -2 (d) = 0.33 P < 0.0001 -3 -3 -3 -2 -1 0 1 2 3 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 Adjusted Flow Adjusted growth rate √ FIGURE 11. Partial regression plots showing (a)–(d) the relationships inherent in the weighted regression model day of passage = 12.415 + 0.011 · day of capture – 0.488 · delta◦C – 0.187 · mean reservoir flow + 1.597 · growth rate that was fitted to explain the variation in day of passage of PIT-tagged natural fall Chinook Salmon smolts recaptured after passage at Lower Granite Dam during portions of the low (1992–1998; gray circles) and high (2005–2011; white circles) abundance periods.

high-abundance period. One explanation for the lack of a dif- (Reimers 1973; Chandler and Bjornn 1988; Grant and Kramer ference in parr growth between abundance periods is that the 1990). This social hierarchy would result in an overall shift recaptured PIT-tagged natural parr were the larger, more com- to earlier dispersal of the population, positive skewness in the petitive members of the riverine population. Such winners could date-of-capture distribution, and a decrease in the mean size at maintain growth under high fish densities if smaller, less com- dispersal without an apparent effect on the growth or size of the petitive conspecific losers dispersed downstream in sufficient winners. numbers to preserve adequate growth conditions for the winners Our results on the growth, passage, and size of natural smolts in Lower Granite Reservoir provided ample evidence for H2. Smolt growth declined between abundance periods in associa- 100 tion with density-dependent changes in size at dispersal from

Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 90 Low abundance period (N = 628) riverine habitat, migrational behavior in the reservoir, and in- 80 High abundance period (N = 555) creased concentrations of subyearling smolts in the reservoir as 70 60 the hatchery program expanded. In turn, the time of passage 50 through the reservoir of natural smolts became earlier and smolt size decreased. During the low-abundance period, natural smolts Wei ght (g) 40 30 also remained in the reservoir later into the summer, which al- 20 lowed them to experience warmer temperatures and grow to 10 larger sizes. However, during the high-abundance period, nat- 0 ural smolts spent more time migrating and less time lingering 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 and feeding, resulting in passage through the reservoir before Fork length (mm) they could benefit from warmer water temperatures that favored growth. FIGURE 12. Scatterplot showing the relationship between weight and FL for PIT-tagged natural fall Chinook Salmon smolts recaptured after passage at In contrast to our findings on natural parr in riverine habitat, Lower Granite Dam during portions of the low (1992–1998; gray circles) and we found little evidence of winners in Lower Granite Reser- high (2005–2011; white circles) abundance periods. voir during the high-abundance period, as there were marked PHENOTYPIC CHANGE IN A LARGE RIVER 1465

differences between abundance periods not only in smolt growth attributes (Albert and Johnson 2012; Oliveira 2012). Clearly, but also in weight at a given length and the presence of some processes affecting fish populations can be attributed to large smolts. This finding raises the question whether density- directional selection, such as the timing of adult migrations to dependent movement behavior varies between riverine and im- reproductive sites as affected by the timing of ocean harvest pounded habitats. The riverine habitat we studied was more (Quinn et al. 2007). Low survival to adulthood of late-migrating structurally complex than the reservoir, and complex habitats smolts might result in selection over time for returning adults can encourage juvenile fish to establish territories for feeding that were early-migrating smolts, which then pass on the early- and holding in energetically profitable positions (Fausch 1984; migration trait to their offspring. However, the fall Chinook Rosenfeld and Boss 2001; Venter et al. 2008). The presence Salmon we studied spent only a small portion of their life cycle of some fish with established territories yields greater dispersal in freshwater, and the timing and success of their seawater tran- of fish without territories (Keeley 2001), as we observed for sition and subsequent adult return was influenced by spatially natural parr in riverine habitat. Natural smolts in the reservoir complex climate and oceanic events (e.g., Burke et al. 2013; resided in homogenous habitat with less complexity, lower ve- Sharma et. al. 2013) that could compensate for low freshwater locities, and increased depths, and the shoreline orientation of survival. Furthermore, we only studied 4–6 generations of fish, fish declined (Connor et al. 2003b; Tiffan et al. 2009b). The and it is not clear that directional selection caused by a change influence of a similar change in habitat complexity on fish be- in freshwater survival could become apparent so rapidly in an havior was observed in the estuary of the Sixes River, Oregon, indigenous species in their native habitat. where fall Chinook Salmon subyearlings switched from antag- Another plausible explanation for the changes in smolt onistic to schooling behavior when velocities declined during growth, passage timing, and size that we observed is that they flood tides (Reimers 1968). Moreover, schooling was the most represent functional responses to predation. The two primary common social behavior displayed by subyearlings rearing in predators in Lower Granite Reservoir are Northern Pikeminnow homogenous, low-velocity habitats in the Hanford Reach of the Ptychocheilus oregonensis and Smallmouth Bass Micropterus Columbia River, Washington (Tiffan et al. 2010). dolomieu. When smolt abundance was low during the 1990s, We hypothesize that density-dependent schooling behavior natural smolts were not important items in the diet of preda- played a role in shaping the phenotypic changes observed in tors in the reservoir (Naughton et al. 2004). However, as their natural smolts as they passed downstream in Lower Granite abundance increased in association with the expanding hatch- Reservoir. When smolt abundances were low in the reservoir, ery program, it is possible that predators identified subyearling schools of natural smolts were likely small and widely spaced, smolts as prey (Collis et al. 1995; Martinelli and Shively 1997; with little social pressure to move. As noted by Hoar (1958), Petersen 2001). Predators can become more efficient as their ex- Keenleyside’s (1955) definition of a fish school stresses the perience with prey increases, and natural smolts that might have fact that “the chief factor common to all schools is a definite lingered and grown to large sizes during the high-abundance pe- mutual attraction between individuals.” Such mutual attraction riod would have been subject to experienced predators primed was likely enhanced between schools of natural and hatchery to feed as the reservoir temperature warmed (Vigg et al. 1991). smolts in the low-velocity reservoir as abundance increased. Fur- In conclusion, our study provides evidence for density- thermore, hatchery smolts in Lower Granite Reservoir tend to dependent phenotypic change in Chinook Salmon subyearlings migrate more than linger and feed (Connor et al. 2004). School- native to an 8th-order river that was influenced by the expansion ing with actively migrating hatchery smolts could intermittently of a recovery program. In a circuitous way, natural parr and move some natural fish long distances downstream over short smolts have begun to regain the movement timing and size phe- periods of time, while sustaining downstream movement in oth- notypes expressed by their historical counterparts under sim- Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 ers (e.g., Atlantic Salmon Salmo salar; Hansen and Jonsson ilarly high abundances. Thus, this study shows that efforts to 1985). This phenomenon, referred to as the “Pied Piper effect” recover native fishes can have detectable effects in large-river (Hillman and Mullan 1989), would partly explain our reservoir landscapes. The outcome of such phenotypic change, which is findings because natural smolts that temporarily or permanently an important area of future research in large-river landscapes, joined schools of hatchery smolts would spend more time mi- can only be fully judged by examining the effect of the change grating and less time feeding, arrive at Lower Granite Dam early, on population viability and productivity. Multistage life cycle and be exposed to relatively cool seasonal temperatures. models fitted with covariates to explore an array of density- While our density-dependent explanation for the phenotypic dependent and management scenarios would provide a way to changes observed in natural smolts is speculative, we do not understand how each life stage of a population contributes to its believe that the change was solely due to the Pied Piper ef- growth rate (e.g., Moussalli and Hilborn 1986; Zabel et al. 2006; fect. In all likelihood, other factors associated with the recovery Brooks and Powers 2007). To fit such models for Snake River program and the increase in abundance had some influence on basin fall Chinook Salmon, it will be necessary to develop meth- the phenotypic changes observed. Recent reviews have explored ods to estimate both juvenile and adult abundance separately for the balance between the evolution of adaptive social and pheno- natural- and hatchery-origin fish. Once these methods are de- typic plasticity and directional selection of distinct life history veloped and tested, the phenotypic traits of natural juveniles 1466 CONNOR ET AL.

and the abundance of hatchery smolts in the reservoir will be Chandler, G. L., and T. C. Bjornn. 1988. Abundance, growth, and interactions of important covariates in analyses. juvenile steelhead relative to time of emergence. Transactions of the American Fisheries Society 117:432–443. Chapman, D. W. 1962. Aggressive behavior in juvenile Coho Salmon as a cause of emigration. Journal of the Fisheries Research Board of Canada 19:1047– ACKNOWLEDGMENTS 1080. We thank our U.S. Fish and Wildlife Service and U.S. Ge- Collis, K., R. E. Beaty, and B. R. Crain. 1995. Changes in catch rate and diet of ological Survey colleagues whose efforts contributed to this Northern Squawfish associated with the release of hatchery-reared juvenile salmonids in a Columbia River reservoir. North American Journal of Fisheries 20-year study. We are grateful for the wealth of data collected Management 15:346–357. by our coworkers from the Idaho Power Company, Fish Pas- Connor, W. P., H. L. Burge, R. Waitt, and T. C. Bjornn. 2002. Juvenile life sage Center, Nez Perce Tribe, National Oceanographic and At- history of wild fall Chinook Salmon in the Snake and Clearwater rivers. mospheric Administration, Smolt Monitoring Program, Univer- North American Journal of Fisheries Management 22:703–712. sity of Idaho, University of Washington, U.S. Army Corps of Connor, W. P., H. L. Burge, J. R. Yearsley, and T. C. Bjornn. 2003a. Influence of flow and temperature on survival of wild subyearling fall Chinook Salmon in Engineers, and Washington Department of Fish and Wildlife, the Snake River. North American Journal of Fisheries Management 23:362– including S. Downing, P. Groves, D. Marsh, F. Mensik, D. 375. Milks, C. Morrill, P. Verhey, and D. Ross. This study would Connor, W. P., S. G. Smith, T. Andersen, S. M. Bradbury, D. C. Burum, E. not have been possible without the participation of personnel E. Hockersmith, M. L. Schuck, G. W. Mendel, and R. M. Bugert. 2004. of the Pacific States Marine Fisheries Commission, including Postrelease performance of hatchery yearling and subyearling fall Chinook Salmon released into the Snake River. North American Journal of Fisheries D. Marvin (through 2010) and N. Tancreto (2011–2013), who Management 24:545–560. helped operate and maintain the Columbia Basin PIT-Tag Infor- Connor, W. P., R. K. Steinhorst, and H. L. Burge. 2003b. Migrational behavior mation System. Peer review by R. Zabel, D. Rondorf, an anony- and seaward movement of wild subyearling fall Chinook Salmon in the Snake mous reviewer, and the editorial staff improved this manuscript. River. North American Journal of Fisheries Management 23:414–430. This study was funded by the Bonneville Power Administra- Connor, W. P., and K. F. Tiffan. 2012. Evidence for parr growth as a factor af- fecting parr-to-smolt survival. Transactions of the American Fisheries Society tion and administered by D. Docherty and S. Bradbury under 141:1207–1218. project number 199102900. The U.S. Army Corps of Engineers Downing, S. L., E. F. Prentice, R. W. Frazier, J. E. Simonson, and E. P. Nun- cost-shared the project during 2005–2011, with valuable over- nallee. 2001. Technology developed for diverting passive integrated transpon- sight by S. Dunmire and D. Holecek. Any use of trade, firm, der (PIT) tagged fish at hydroelectric dams in the Columbia River basin. or product names is for descriptive purposes only and does not Aquacultural Engineering 25:149–164. Efron, B., and R. J. Tibshirani. 1998. An introduction to the bootstrap. CRC imply endorsement by the U.S. Government. The findings and Press, Boca Raton, Florida. conclusions in this article are those of the authors and do not Fausch, K. D. 1984. Profitable stream positions for salmonids: relating specific necessarily represent the views of the U.S. Fish and Wildlife growth rate to net energy gain. Canadian Journal of Zoology 62:441–451. Service. FPC (Fish Passage Center). 2012. Information on matters related to juvenile and adult salmon and steelhead passage through the mainstem hydrosystem in the Columbia River basin. FPC, Portland, Oregon. Available: www.fpc.org. (June 2012). REFERENCES Grant, J. W. A., and I. Imre. 2005. Patterns of density-dependent growth in Albert, J. S., and D. M. Johnson. 2012. Diversity and evolution of body size in juvenile stream-dwelling salmonids. Journal of Fish Biology 67(Supplement fishes. Evolutionary Biology 39:324–340. B):100–110. Beamesderfer, R. C. P., D. L. Ward, and A. A. Nigro. 1996. Evaluation of Grant, J. W. A., and D. L. Kramer. 1990. Territory size as a predictor of the the biological basis for a predator control program on Northern Squawfish upper limit to population density of juvenile salmonids in streams. Canadian (Ptychocheilus oregonensis) in the Columbia and Snake rivers. Canadian Journal of Fisheries and Aquatic Sciences 47:1724–1737.

Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 Journal of Fisheries and Aquatic Sciences 53:2898–2908. Groves, P. A., and J. A. Chandler. 1999. Spawning habitat used by fall Chinook Bjornn, T. C., and D. W. Reiser. 1991. Habitat requirements of salmonids in Salmon in the Snake River. North American Journal of Fisheries Management streams. Pages 83–138 in W. R. Meehan, editor. Influences of forest and 19:912–922. rangeland management on salmonid fishes and their habitats. American Fish- Groves, P.A., J. A. Chandler, B. Alcorn, T. J. Richter, W. P.Connor, A. P.Garcia, eries Society, Special Publication 19, Bethesda, Maryland. and S. M. Bradbury. 2013. Evaluating salmon spawning habitat capacity using Brooks, E. N., and J. E. Powers. 2007. Generalized compensation in stock– redd survey data. North American Journal of Fisheries Management 33:707– recruit functions: properties and implications for management. ICES Journal 716. of Marine Science 64:413–424. Hansen, L. P., and B. Jonsson. 1985. Downstream migration of hatchery-reared Burke, B. J., W. T. Peterson, B. R. Beckman, C. Morgan, E. A. Daly, and M. smolts of Atlantic Salmon (Salmo salar L.) in the River Imsa, Norway. Litz. 2013. Multivariate models of adult Pacific salmon returns. PLoS (Pub- Aquaculture 45:237–248. lic Library of Science) ONE [online serial] 8(1):e54134. doi: 10.1371/jour- Hillman, T. W., and J. W. Mullan. 1989. Effect of hatchery releases on the nal.pone.0054134. abundance and behavior of wild juvenile salmonids. Pages 265–285 in Don Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel Chapman Consultantsr, editors. Summer and winter ecology of juvenile Chi- inference: a practical information-theoretic approach, 2nd edition. Springer- nook Salmon and steelhead trout in the Wenatchee River, Washington. Don Verlag, New York. Chapman Consultants, Boise, Idaho. CBR (Columbia Basin Research). 2012. Columbia River DART (data ac- Hoar, W. S. 1958. The evolution of migratory behaviour among juvenile salmon cess in real time). CBR, University of Washington, Seattle. Available: of the genus Oncorhynchus. Journal of the Fisheries Research Board of www.cbr.washington.edu/dart/dart.html. (June 2012). Canada 15:391–428. PHENOTYPIC CHANGE IN A LARGE RIVER 1467

Kaya, C. M., L. R. Kaeding, and D. E. Burkhalter. 1977. Use of a cold-water D. Prince, and G. A. Winans, editors. Fish-marking techniques. American refuge by Rainbow and Brown trout in a geothermally heated stream. Pro- Fisheries Society, Symposium 7, Bethesda, Maryland. gressive Fish-Culturist 39:37–39. PTAGIS (Columbia Basin PIT Tag Information System). 2012. PTAGIS. Pa- Keeley, E. R. 2001. Demographic responses to food and space competition by cific States Marine Fisheries Commission, Portland, Oregon. Available: juvenile steelhead trout. Ecology 82:1247–1259. www.ptagis.org. (June 2012). Keenleyside, M. H. A. 1955. Some aspects of the schooling behaviour of fish. PTOC (PIT Tag Interrogation Site Operations and Maintenance). 2012. PTOC. Behaviour 8:183–248. Pacific States Marine Fisheries Commission, Portland, Oregon. Available: Littell, R. C., G. A. Milliken, W. W. Stroup, and R. D. Wolfinger. 1996. SAS www.ptoccentral.org. (January 2012). system for mixed models. SAS Institute, Cary, North Carolina. Quinn, T. P., S. Hodgson, L. Flynn, R. Hilborn, and D. E. Rogers. 2007. Direc- Mains, E. M., and J. M. Smith. 1964. The distribution, size, time, and cur- tional selection by fisheries and the timing of Sockeye Salmon (Oncorhynchus rent preferences of seaward migrant Chinook Salmon in the Columbia and nerka) migrations. Ecological Applications 17:731–739. Snake rivers. Washington Department of Fisheries, Fisheries Research Papers Rainey, S. R., L. A. Reese, and T. O. Wik. 2006. Removable spillway weir de- 2(3):5–43. velopment for fish passage, power, and water quality benefits. Hydro Review Marsh, D. M., G. M. Matthews, S. Achord, T. E. Ruehle, and B. P. Sand- 25:118–124. ford. 1999. Diversion of salmonid smolts tagged with passive integrated Reimers, P. E. 1968. Social behavior among juvenile fall Chinook Salmon. transponders from an untagged population passing through a juvenile col- Journal of the Fisheries Research Board of Canada 25:2005–2008. lection system. North American Journal of Fisheries Management 19: Reimers, P. E. 1973. The length of residence of juvenile fall Chinook Salmon 1142–1146. in Sixes River, Oregon. Oregon Fish Commission Research Report 4: Martinelli, T. L., and R. S. Shively. 1997. Seasonal distribution, move- 3–43. ments, and habitat associations of Northern Squawfish in two lower Rosenfeld, J. S., and S. Boss. 2001. Fitness consequences of habitat use for Columbia River reservoirs. Regulated Rivers: Research and Management 13: juvenile Cutthroat Trout: energetic costs and benefits in pools and riffles. 543–556. Canadian Journal of Fisheries and Aquatic Sciences 58:585–593. McCormick, S. D., L. P. Hansen, T. P. Quinn, and R. L. Saunders. 1998. SAS (Statistical Analysis Systems). 2012. SAS/STATR 9.2 user’s guide, 2nd Movement, migration, and smolting of Atlantic Salmon (Salmo salar). edition. SAS Institute, Cary, North Carolina. Available: support.sas.com. Canadian Journal of Fisheries and Aquatic Sciences 55(Supplement 1): (June 2012). 77–92. Sharma, R., L. A. Velez-Espino,´ A. C. Wertheimer, N. Mantua, and R. C. Moussalli, E., and R. Hilborn. 1986. Optimal stock size and harvest rate in Francis. 2013. Relating spatial and temporal scales of climate and ocean multistage life history models. Canadian Journal of Fisheries and Aquatic variability to survival of Pacific Northwest Chinook Salmon (Oncorhynchus Sciences 43:135–141. tshawytscha). Fisheries Oceanography 22:14–31. Moya-Larano,˜ J., and G. Corcobado. 2008. Plotting partial correlation and re- Smith, S. G., W. D. Muir, E. E. Hockersmith, R. W. Zabel, R. J. Graves, C. V. gression in ecological studies. Web Ecology 8:35–46. Ross, W. P. Connor, and B. D. Arnsberg. 2003. Influence of river conditions Naughton, G. P., D. H. Bennett, and K. B. Newman. 2004. Predation on juve- on survival and travel time of Snake River subyearling fall Chinook Salmon. nile salmonids by Smallmouth Bass in the Lower Granite Reservoir system, North American Journal of Fisheries Management 23:939–961. Snake River. North American Journal of Fisheries Management 24:534– Stein, R. A., P. E. Reimers, and J. D. Hall. 1972. Social interaction between juve- 544. nile Coho (Oncorhynchus kisutch) and fall Chinook salmon (O. tshawytscha) Nielsen, J. L., T. E. Lisle, and V. Ozaki. 1994. Thermally stratified pools and in Sixes River, Oregon. Journal of the Fisheries Research Board of Canada their use by steelhead in northern California streams. Transactions of the 29:1737–1748. American Fisheries Society 123:613–626. Swan, G. A., B. H. Monk, J. G. Williams, and B. P. Sandford. 1990. Fish NMFS (National Marine Fisheries Service). 1992. Threatened status for guidance efficiency of submersible traveling screens at Lower Granite Dam, Snake River spring/summer Chinook Salmon, threatened status for Snake 1989. Report to the U.S. Army Corps of Engineers, National Marine Fisheries River fall Chinook Salmon. Federal Register 57:78(22 April 1992):14653– Service, Northwest Fisheries Science Center, Seattle. 14663. Thorpe, J. E., C. E. Adams, M. S. Miles, and D. S. Keay. 1989. Some influences Oliveira, R. F. 2012. Social plasticity in fish: integrating mechanisms and func- of photoperiod and temperature on opportunity for growth in juvenile Atlantic tion. Journal of Fish Biology 81:2127–2150. Salmon, Salmo salar L. Aquaculture 82:119–126. Peters, C. N., D. R. Marmorek, and R. B. Deriso. 2001. Application of decision Tiffan, K. F., and W. P. Connor. 2011. Distinguishing between natural and analysis to evaluate recovery actions for threatened Snake River fall Chi- hatchery Snake River fall Chinook Salmon subyearlings in the field using

Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 nook Salmon (Oncorhynchus tshawytscha). Canadian Journal of Fisheries body morphology. Transactions of the American Fisheries Society 140:21– and Aquatic Sciences 58:2447–2458. 30. Petersen, J. H. 2001. Density, aggregation, and body size of Northern Tiffan, K. F., R. D. Garland, and D. W. Rondorf. 2002. Quantifying flow- Pikeminnow preying on juvenile salmonids in a large river. Journal of Fish dependent changes in subyearling fall Chinook Salmon rearing habitat us- Biology 58:1137–1148. ing two-dimensional spatially explicit modeling. North American Journal of Plumb, J. M., W. P. Connor, K. F. Tiffan, C. M. Moffitt, R. W. Perry, and N. S. Fisheries Management 22:713–726. Adams. 2012. Estimating and predicting collection probability of fish at dams Tiffan, K. F., T. J. Kock, W. P. Connor, R. K. Steinhorst, and D. W. Rondorf. using multistate modeling. Transactions of the American Fisheries Society 2009a. Behavioural thermoregulation by subyearling fall (autumn) Chinook 141:1364–1373. Salmon Oncorhynchus tshawytscha in a reservoir. Journal of Fish Biology Prentice, E. F., T. A. Flagg, and C. S. McCutcheon. 1990a. Feasibil- 74:1562–1579. ity of using implantable passive integrated transponder (PIT) tags in Tiffan, K. F., T. J. Kock, C. A. Haskell, W.P.Connor, and R. K. Steinhorst. 2009b. salmonids. Pages 317–322 in N. C. Parker, A. E. Giorgi, R. C. Hei- Water velocity, turbulence, and migration rate of subyearling fall Chinook dinger, D. B. Jester Jr., E. D. Prince, and G. A. Winans, editors. Fish- Salmon in the free-flowing and impounded Snake River. Transactions of the marking techniques. American Fisheries Society, Symposium 7, Bethesda, American Fisheries Society 138:373–384. Maryland. Tiffan, K. F., T. J. Kock, and J. J. Skalicky. 2010. Diel behavior of rearing fall Prentice, E. F., T. A. Flagg, C. S. McCutcheon, and D. F. Brastow. 1990b. Chinook Salmon. Northwestern Naturalist 91:342–345. PIT-tag monitoring systems for hydroelectric dams and fish hatcheries. Pages Venter, O., J. W. A. Grant, M. V. Noel,¨ and J. W. Kim. 2008. Mechanisms 323–334 in N. C. Parker, A. E. Giorgi, R. C. Heidinger, D. B. Jester Jr., E. underlying the increase in young-of-the-year Atlantic Salmon (Salmo salar) 1468 CONNOR ET AL.

density with habitat complexity. Canadian Journal of Fisheries and Aquatic Weber, E. D., and K. D. Fausch. 2003. Interactions between hatchery and wild Sciences 65:1956–1964. salmonids in streams: differences in biology and evidence for competition. Vigg, S., T. P.Poe, L. A. Prendergast, and H. C. Hansel. 1991. Rates of consump- Canadian Journal of Fisheries and Aquatic Sciences 60:1018–1036. tion of juvenile salmonids and alternative prey fish by Northern Squawfish, Zabel, R. W., M. D. Scheuerell, M. M. McClure, and J. G. Williams. Walleyes, Smallmouth Bass, and Channel Catfish in John Day Reservoir, 2006. The interplay between climate variability and density dependence Columbia River. Transactions of the American Fisheries Society 120:421– in the population viability of Chinook Salmon. Conservation Biology 20: 438. 190–200. Downloaded by [Department Of Fisheries] at 21:39 27 October 2013 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:40 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transactions of the American Fisheries Society Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/utaf20 The Effects of Ethanol Preservation on Fish Fin Stable Isotopes: Does Variation in C:N Ratio and Body Size Matter? Carmella Vizza a b , Beth L. Sanderson a , Douglas G. Burrows a & Holly J. Coe a a National Marine Fisheries Service, Northwest Fisheries Science Center , 2725 Montlake Boulevard East , Seattle , Washington , 98112 , USA b Department of Biological Sciences , University of Notre Dame , 292 Galvin Life Sciences Center, Notre Dame , Indiana , 46556 , USA Published online: 06 Sep 2013.

To cite this article: Carmella Vizza , Beth L. Sanderson , Douglas G. Burrows & Holly J. Coe (2013) The Effects of Ethanol Preservation on Fish Fin Stable Isotopes: Does Variation in C:N Ratio and Body Size Matter?, Transactions of the American Fisheries Society, 142:5, 1469-1476, DOI: 10.1080/00028487.2013.816366 To link to this article: http://dx.doi.org/10.1080/00028487.2013.816366

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1469–1476, 2013 American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.816366

NOTE

The Effects of Ethanol Preservation on Fish Fin Stable Isotopes: Does Variation in C:N Ratio and Body Size Matter?

Carmella Vizza,*1 Beth L. Sanderson, Douglas G. Burrows, and Holly J. Coe National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, Washington 98112, USA

(Hobson et al. 1997; Arrington and Winemiller 2002; Sweeting Abstract et al. 2004; but see Syvaranta¨ et al. 2011). Unfortunately, this Although chemical preservation of stable isotope samples has method can be problematic when samples are collected from been studied in a variety of species and tissue types, the effects remote sampling sites without access to freezers or dry ice. of ethanol preservation on fish fin tissue have not been exam- ined. Using caudal fin samples from juvenile Chinook Salmon On- Chemical preservation, which is widely used for specimens in corhynchus tshawytscha and Rainbow Trout O. mykiss or steelhead historical studies or archived collections, is a viable alternative (the anadromous form of Rainbow Trout), we investigated how for samples that must be transported long distances between the storage time (2, 4, and 6 months), fin composition (C:N ratio), field and laboratory. and fish body size (50–130 mm FL) influence preservation-induced Studies examining the effects of chemical preservation on changes in δ13Candδ15N. In both species, we found that treatment fins (frozen and later preserved in ethanol) exhibited higher δ13C stable isotopes have found inconsistent results. For example, than did paired reference fins (frozen). The changes in δ15N, how- Hobson et al. (1997) demonstrated that preservation media can ever, were smaller in magnitude and less consistent. Preservation- differentially affect δ13Corδ15N (ratio of 13Cto12Cor15N induced increases in fin δ13C, but not δ15N, were significantly cor- to 14N relative to an international standard; Peterson and Fry related with the change in C:N ratio (treatment–reference) in both δ13 1987). Storing quail blood and muscle in ethanol had no sig- species. In addition, these increases in C were more highly corre- δ13 δ15 lated with body size in O. mykiss than in Chinook Salmon. Storage nificant effect on either Cor N, while quail Coturnix time had a significant effect on the shift in treatment fin δ13Cand coturnix japonica blood and muscle preserved in formalin ex- a small, but insignificant, effect on δ15NinO. mykiss.However, hibited lower δ13C signatures. Preservation techniques, such as storage time was not a significant factor for explaining the iso- formalin fixation and transfer to ethanol in natural history col- topic shifts observed in Chinook Salmon fin tissue. This is the first lections (Arrington and Winemiller 2002), and the length of time study to document variation in preservation-induced changes in δ13C within a species and to link this variation to C:N ratio. Future in which a sample is stored before analysis (Kaehler and Pakho- studies using species-specific and tissue-specific models to correct mov 2001; Sarakinos et al. 2002) can both have unique effects for preservation-induced shifts in stable isotope ratios should be on stable isotope ratios. Furthermore, the effects of preservation aware that these models do not account for intraspecific variation may differ among taxonomic groups or species; Kaehler and Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 in tissue composition. Pakhomov (2001) found that the effect of preservation on δ13C varied largely between unrelated marine species (common kelp Stable isotope analysis is widely used for examining ecolog- Ecklonia radiata,KobArgyrosomus hololepidotus, and com- ical questions about nutrient and energy flow, animal migration, mon octopus Octopus vulgaris). These effects can differ within and trophic interactions (Peterson and Fry 1987; Hobson 1999; similar taxa, like ray-finned fishes (Actinopterygii; B. Kelly Post 2002). While use of this technology has grown rapidly over et al. 2006), and even among tissue types of an individual fish the past decade (Mart´ınez del Rio et al. 2009), questions linger (Sweeting et al. 2004; B. Kelly et al. 2006). about how to best collect and prepare samples for stable iso- Differences between species and tissue types with respect tope analysis. Freezing is the preferred method of preservation to how chemical preservation affects stable isotope ratios may for samples because it generally does not affect isotopic values depend on the biochemical composition of the tissues analyzed

*Corresponding author: [email protected] 1Present address: Department of Biological Sciences, University of Notre Dame, 292 Galvin Life Sciences Center, Notre Dame, Indiana 46556, USA. Received May 21, 2012; accepted June 5, 2013 Published online September 6, 2013 1469 1470 VIZZA ET AL.

(Sweeting et al. 2004; B. Kelly et al. 2006). For example, tis- replacing the ice every 24 h, until they could be frozen at −20◦C. sues with greater lipid content generally have a more negative Although our inability to immediately freeze the samples is a δ13C value because lipid synthesis discriminates against 13Cin limitation of our study, the average air temperature was 7◦C favor of the lighter 12C isotope (DeNiro and Epstein 1977). Un- during the sampling trip. Therefore, we believe that the samples derstanding the interaction between chemical preservation and were adequately preserved, especially given that the potential lipids is crucial to interpreting the shifts in δ13C, given that lipid for decay in salmonid carcasses is higher for muscle tissue than content varies widely between species, among tissue types, and for other tissues such as skin or internal organs (Chaloner et al. with body size. Because the relationship between lipid content 2002). (%) and the carbon-to-nitrogen ratio (C:N) is very strong in We began with 30 fish of each species and assigned 10 fish animals (Post et al. 2007), C:N ratios can be useful indicators from a range of sizes to each of three groups based on storage of the presence of lipids for studies examining the effects of time in ethanol (2, 4, or 6 months). Our goal was to ensure preservation on stable isotopes. While the C:N ratio has not different-sized fish were represented in each group (Chinook been directly linked to lipid content in fish fin tissue, we believe Salmon: 59–92 mm FL, O. mykiss: 50–130 mm FL). For each that it is, at the very least, an informative indicator of variation in fish, the caudal fin was removed. Half of the caudal fin was tissue composition that could influence the response to ethanol immediately frozen at −20◦C (reference tissue). The other half preservation. of the fin was placed in a 2-mL centrifuge tube with 70% pure Although studies have investigated preservation effects for a ethanol–30% deionized H2O and stored in solution for 2, 4, or species using a single individual (Kaehler and Pakhomov 2001; 6 months at 20◦C (treatment tissue). After storage, the ethanol- Sweeting et al. 2004), we chose multiple individuals in an ef- preserved fin tissue was rinsed with deionized water and imme- fort to examine intraspecific variation in preservation-induced diately frozen at −20◦C for 24 h as a prerequisite for the freeze- changes in stable isotope ratios. We were particularly interested drying process. Because we wanted to be able to attribute any in how factors such as storage time, C:N ratio, and individual change in the stable isotope values to ethanol preservation, we body size affect the stable isotope ratios of fish fin tissue pre- did not want to subject the reference and treatment fins to dif- served in ethanol. We chose fin tissue because it has become ferent drying processes (e.g., freeze-drying versus oven drying). a viable alternative to the lethal sampling of muscle tissue for At the end of the experiment, reference and treatment fin tissues many threatened and endangered fish species (M. H. Kelly et al. were freeze-dried, pulverized, and weighed (0.5–0.7 mg) into tin 2006; Sanderson et al. 2009; Hanisch et al. 2010) and because capsules. the effect of ethanol preservation on stable isotope ratios of this Stable isotope ratios were determined using the procedure tissue has not been studied. Using juvenile Chinook Salmon described in Sanderson et al. (2009), and ratios of C:N were Oncorhynchus tshawytscha and Rainbow Trout O. mykiss or determined from percent element data. Due to the limited sam- steelhead (the anadromous form of Rainbow Trout), our primary ple material available in a single juvenile caudal fin, C:N ratios objective was to examine differences in stable isotope values as a are a good alternative to analyzing tissues for lipid or protein result of ethanol preservation. Understanding the predictability content because they are obtained during routine stable isotope and consistency of isotopic shifts related to methodological and analysis and do not require a larger amount of tissue for ad- biological factors is crucial to the interpretation of dietary and ditional extractions (Post et al. 2007). The amount of sample trophic information from chemically preserved samples. Specif- material needed to meet the stable isotope analysis requirement ically, our research questions were as follows: (1) How do δ13C, (0.5–0.7 mg) can be a problem when dealing with juvenile fin δ15N, and C:N differ between ethanol-preserved and reference tissue (Sanderson et al. 2009). If we were unable to obtain data fins? (2) How does storage time affect differences in δ13C and for both the treatment and reference fin of an individual (i.e., Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 δ15N of fin tissue due to preservation? and (3) Do differences in the fin tissue weighed less than 1 mg after drying), the fish was δ13C and δ15N vary with the change in C:N or the size of the fish? excluded from analysis, which resulted in some storage time groups having less than 10 fish (Table 1). Nonetheless, there were no significant differences in FL between groups for either METHODS species (Chinook Salmon: F 2, 26 = 0.04, P = 0.96; O. mykiss: Our study was conducted in 11 tributaries of the Salmon River F 2, 20 = 0.38, P = 0.69). basin in central Idaho that support threatened populations of Analysis.—Descriptive statistics are reported as mean ± SD, spring–summer Chinook Salmon, Rainbow Trout and steelhead, and a significance level of α = 0.05 was established for all sta- and Bull Trout Salvelinus confluentus. Because we were unable tistical tests a priori. Using the data from all storage times, we to discern the particular life history of each juvenile O. mykiss, conducted paired t-tests to determine whether δ13C, δ15N, and hereafter we will refer to anadromous and resident fish jointly C:N of treatment fin tissue differed from that of the reference as O. mykiss. Juvenile Chinook Salmon and O. mykiss were tissue. For each storage time, Pearson’s product-moment cor- collected via electrofishing during September 2008. Fish were relations were used to compare the relationships between δ13C then measured to obtain FL and wet weighed. While working and δ15N in reference and treatment fins. Although FL was posi- in this remote area, we placed the fish on ice for 1 to 8 d, tively correlated with the C:N ratio of reference fin for Chinook NOTE 1471

TABLE 1. Sample size and FL minimum, mean, and maximum (mm) for ence fins was weakest for samples stored in ethanol for 6 months each treatment of Chinook Salmon and O. mykiss. in both species (Figure 1). δ13 Species Treatment n Minimum Mean Maximum When we modeled preservation-induced shifts in Cfor Chinook Salmon, storage time was not a significant factor (Fig- Chinook 2 months 9 59 75.3 92 ure 2; Table 2). The difference in δ13C varied significantly with 4 months 10 60 74.1 90 C:N but only slightly with FL (Table 2). Storage time was 6 months 10 59 74.0 92 significant in the model for O. mykiss δ13C; however, a Games- O. mykiss 2 months 8 55 80.1 128 Howell post hoc test found no significant differences between 4 months 9 50 76.6 131 the storage time groups (Figure 2; Table 2; P ≥ 0.09 for all 6 months 6 55 89.8 130 pairwise comparisons). The difference in δ13C varied signifi- cantly with C:N and slightly with FL for O. mykiss (Table 2). While the effects of FL were insignificant in the models, the Salmon (r = 0.48, P < 0.01), we chose to include both factors in preservation-induced shifts in δ13C were more highly corre- our analyses because the relationship was not significant for O. lated with FL in O. mykiss (r = 0.52, P < 0.05) than in Chi- mykiss (r = 0.36, P > 0.05). We employed analysis of covari- nook Salmon (r =−0.13, P > 0.05). Storage time, C:N, ance to examine preservation-induced shifts in δ13C and δ15N and FL were not significant factors explaining the difference as a function of storage time, change in fin tissue composition between treatment and reference fin δ15N for Chinook Salmon (C:N), and body size: (Figures 2; Table 2). For O. mykiss, we observed a slight but insignificant effect of storage time (Figure 2; Table 2), while δ15 yT − yR = β0 + β1ST + β2C:N+ β3FL + ε, the difference in N did not vary significantly with C:N or FL (Table 2).

where yT and yR are the isotopic values of the treatment and reference tissues, ST is storage time (fixed factor), C:N is the DISCUSSION difference between treatment and reference tissues (covariate), FL is fork length (covariate), and ε is the error component. If a Effects of Ethanol Preservation on Fin ␦13C, ␦15N, and C:N model detected a significant effect of storage time, we conducted Our results demonstrate that storing Chinook Salmon and a Games-Howell post hoc test to make pairwise comparisons and O. mykiss fin samples in ethanol alters their isotopic characteris- to account for unequal variances between groups. tics. Whereas some studies have found no change in either δ13C or δ15N following preservation of muscle, blood, and whole- bodied organisms in ethanol (Hobson et al. 1997; Syvaranta¨ RESULTS et al. 2008), others have observed a significant difference in Effects of Ethanol Preservation on Fin ␦13C, ␦15N, and C:N δ13C only (B. Kelly et al. 2006; Bugoni et al. 2008). The 0.2– 15 Treatment fin δ13C was higher than that of the reference fin 0.4‰ mean differences between treatment and reference δ N in both species (Figure 1). Across all storage times, δ13Cin that we observed were consistent with those of other studies treatment fins was 1.4 ± 0.3‰ higher in Chinook Salmon (t = (Hobson et al. 1997; B. Kelly et al. 2006). While we did ob- 28.46, df = 28, P < 0.001) and 1.3 ± 0.4‰ higher in O. mykiss serve a significant difference between treatment and reference 15 (t = 14.99, df = 22, P < 0.001). In comparison, the difference in fin δ N for Chinook Salmon, the mean difference of 0.4‰ was δ15N between paired treatment and reference fins was smaller, very close to the measurement error for our analysis ( ± 0.3‰). 15 at 0.4 ± 0.3‰ (t = 6.40, P < 0.001) for Chinook Salmon and In general, the way in which ethanol preservation affects δ N Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 0.2 ± 0.4‰ (t = 1.93, P = 0.07) for O. mykiss (Figure 1). is not well understood (Arrington and Winemiller 2002). How- The C:N ratios of treatment and reference fins were 2.9 ± ever, the literature does suggest that this effect is neither due to 15 0.1 and 3.7 ± 0.2, respectively, for both species. Therefore, the uptake of N from the preservative, because ethanol con- 14 treatment fins consistently had lower C:N ratios than reference tains no nitrogen, (Sarakinos et al. 2002), nor to the loss of N fins (Chinook Salmon: t =−29.56, P < 0.001; O. mykiss: t = from the sample, because ethanol has little effect on proteins −22.92, P < 0.001). (Sweeting et al. 2004). The difference between δ13C of treatment and reference Storage Time and Biological Factors fins in this study (1.3–1.4‰) was greater in magnitude than Carbon isotopic values in treatment and reference fin tissues in preservation studies of fish muscle, probably due to differ- were highly correlated for all storage times for both Chinook ences in tissue composition. For example, B. Kelly et al. (2006) Salmon and O. mykiss (P < 0.001 for all tests; Figure 1). Ni- detected a 0.8‰ increase in Arctic Char Salvelinus alpinus, trogen isotopic values in treatment and reference fins were also while Sarakinos et al. (2002) observed an increase of 0.2‰ in significantly correlated; however, the strength of these correla- Sacramento Sucker Catostomus occidentalis. This increase in tions was generally lower than those for δ13C (Figure 1). The δ13C can be the result of the differential loss or leaching of strength of the correlation in δ15N between treatment and refer- sample molecules containing the lighter isotope (12C) or the 1472 VIZZA ET AL.

Chinook O. mykiss -18

-20

-22

-24

C Treatment -26 13 δ (2 months) r = 0.98*** (2) r = 1.00*** -28 (4 months) r = 1.00*** (4) r = 0.96*** (6 months) r = 0.99*** (6) r = 0.99*** -30 -30 -28 -26 -24 -22 -20 -18 -30 -28 -26 -24 -22 -20 -18

δ13C Reference δ13C Reference

10

9

8 N Treatment 15 δ 7 (2) r = 0.96*** (2) r = 0.92*** (4) r = 0.93*** (4) r = 0.82** (6) r = 0.78** (6) r = 0.11 6 678910678910 δ15N Reference δ15N Reference

FIGURE 1. Scatter plots of treatment fin versus reference fin δ13C (top panels) and δ15N (bottom panels) of Chinook Salmon (left panels) and O. mykiss (right panels) for the three storage times. Each plot has a 1:1 line drawn for reference. Pearson’s product-moment correlation coefficients (r) are given on the bottom right of each panel followed by the significance level if P ≤ 0.05 (one asterisk for P ≤ 0.05, two asterisks for P < 0.01, and three asterisks for P < 0.001). Downloaded by [Department Of Fisheries] at 21:40 27 October 2013

incorporation of molecules containing the heavier isotope (13C) preservation could be one explanation for δ13C enrichment in from the preservative (Hobson et al. 1997; Sweeting et al. treatment fins. B. Kelly et al. (2006) found that extracting lipids 2004). Because Chinook Salmon and O. mykiss fins have a of muscle prior to preservation decreased the magnitude of the larger C:N ratio relative to that of muscle (3.7 as compared to tissue’s enrichment in δ13C, and Sweeting et al. (2004) observed 3.3; B. Sanderson, unpublished data), our results likely dif- that the magnitude of preservation shifts increased with fat con- fer from other studies due to potential differences in tissue tent of Atlantic Cod Gadus morhua tissues (muscle: + 0.5‰, composition. roe: + 0.8‰, liver: + 1.6‰). We do not know of any studies The decrease we observed in the C:N ratio of the treatment that have evaluated fin lipid content, but it is likely that lipid tissue (2.9 ± 0.1) in comparison to that of the reference tissue content in different tissues will vary by fish species based on (3.7 ± 0.2) denotes a loss of sample carbon. Because lipid con- histology and unique life history strategies. Another possible tent is strongly correlated with C:N (Sweeting et al. 2006; Post explanation for the increase in treatment fin δ13C is the loss of et al. 2007; Logan et al. 2008) and lipid synthesis discriminates some other carbon-containing compound that has a lower δ13C against 13C (DeNiro and Epstein 1977), loss of lipids during value relative to the δ13C in the remaining fin tissue. NOTE 1473

Chinook O. mykiss

2.5

2.0

1.5 C

13 1.0 Δδ

0.5

0.0

-0.5 2 months 4 months 6 months 2 months 4 months 6 months

1.4

1.0

0.6 N 15

Δδ 0.2

-0.2

-0.6 2 months 4 months 6 months 2 months 4 months 6 months

-0.2

-0.4

-0.6 C:N Δ -0.8 Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 -1.0

-1.2 2 months 4 months 6 months 2 months 4 months 6 months

FIGURE 2. Box plots of the differences between the treatment and reference tissues by storage time for δ13C (top panels), δ15N (middle panels), and C:N ratios (bottom panels) of Chinook Salmon (left panels) and O. mykiss (right panels). The lowest boundary of the box indicates the 25th percentile, a line within the box marks the median, and the highest boundary of the box indicates the 75th percentile. The error bars below and above the box indicate the 10th and 90th percentiles, and the black dots represent outlying points.

Storage Time variable for O. mykiss. In addition, the changes in δ15N due to Our data suggest that most of the preservation-induced preservation did not differ significantly among storage times for change in fin δ13C happened within the first 2 months of storage either species, despite the slight upward trend seen in O. mykiss. for both species. While changes in δ13C were fairly consistent These findings were also consistent with those of other studies across storage time groups for Chinook Salmon, they were more that have found no significant effect of storage time (ranging 1474 VIZZA ET AL.

TABLE 2. Analysis of covariance summaries by species, where the response variable is either the difference (treatment – reference) in δ13Corδ15N, the storage time is a fixed factor, and the difference in the C:N ratio (C:N) and the FL of the fish are both covariates. Statistically significant factors (P ≤ 0.05) appear in bold italics.

Response variable Species Factor df FP δ13C Chinook Salmon Storage time 2, 24 0.94 0.40 C:N 1, 24 8.70 0.007 FL 1, 24 3.33 0.08 O. mykiss Storage time 2, 18 3.88 0.04 C:N 1, 18 20.14 <0.001 FL 1, 18 3.92 0.06 δ15N Chinook Salmon Storage time 2, 24 0.57 0.57 C:N 1, 24 0.08 0.77 FL 1, 24 0.001 0.98 O. mykiss Storage time 2, 18 2.88 0.08 C:N 1, 18 0.10 0.75 FL 1, 18 0.59 0.45

from 1 d to 21 months) on δ13Corδ15N (Sweeting et al. 2004; for both species (Table 2). We believe that FL may have ex- Syvaranta¨ et al. 2008). plained some variation in isotopic shifts due to preservation, However, we did observe more variation in δ13C and δ15N perhaps because the components of individual tissue types (e.g., with increased storage time for O. mykiss. The increased vari- protein, lipid, and carbohydrate in fin tissue) may vary with de- ation in δ13C may have reflected greater variability in C:N velopment and growth. Assuming that fish of similar body sizes between storage time groups for this species (Figure 2), and it are undergoing similar histological changes due to growth, FL should be noted for O. mykiss that reference fin δ15N values at may provide additional insight into preservation-induced shifts 6 months were confined to a particularly narrow range. Though in δ13C. This assumption is supported by the observation that the results were not statistically significant, the decline in the the difference in δ13C between treatment and reference fins was strength of correlations between reference and treatment δ15N significantly correlated with body size for O. mykiss only, which suggests that storage time may have accounted for some of the is probably due to the greater range of development captured variability in O. mykiss. The increased variability in δ15Nis in this species (Table 1). Juvenile Chinook Salmon fry emerge difficult to explain; however, other studies conducting lipid ex- in early spring and rear in streams through the fall (Bjornn traction, which involves exposing the sample to methanol and 1971); the Chinook Salmon we collected in September were chloroform, have also noted that the shift in δ15N is much less all subyearlings of about the same size (59–92 mm FL), which predictable than that in δ13C (Pinnegar and Polunin 1999; Sweet- explains why the variation in δ13C after preservation is bet- ing et al. 2006). In the absence of a strong effect of storage time ter explained by C:N of their fin tissues. The O. mykiss we on preservation-induced isotopic shifts and because our sam- sampled were either fry that emerged in midsummer (Bjornn ple size was relatively limited per storage group (n = 10), it is 1971) or resident yearlings. Because the size range of the O. difficult to make general recommendations about storage time mykiss (50–131 mm FL) was more than double that of the Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 beyond cautioning others that some fin tissue components may Chinook Salmon, it is not surprising that FL was more useful leach at different rates. for explaining δ13C differences between treatment and refer- ence fin tissues. In addition, we detected visual color changes Biological Factors from gray to white in the fins of larger O. mykiss that we did The difference in δ13C between treatment and reference fins not notice in Chinook Salmon or the smaller O. mykiss.This varied significantly with the C:N for both species, which anecdotal observation suggests that a chemical “extraction” of means that fin tissues exhibiting a greater loss of lipid or some colored compounds in fin samples from larger fish may have other carbon-containing compound after preservation demon- occurred. strated higher δ13C values. While Sweeting et al. (2004) found In contrast to δ13C, differences in δ15N were not correlated a similar result within different tissues of an individual fish, with C:N or fish size in either species. Because lipids do not this is the first study to document a link between shifts in contain protein, we would not have expected changes in δ15Nfol- δ13C following preservation and intraspecific variations in tissue lowing preservation to be related to lipid content (Arrington and composition. Winemiller 2002). The data therefore suggest that intraspecific In contrast, body size had a slight effect on preservation- variation in the C:N ratio and in body size can be largely elim- induced changes in δ13C, but it was an insignificant predictor inated as sources of variability in the change of δ15N following NOTE 1475

preservation in ethanol. However, the removal of lipoproteins REFERENCES by ethanol could have caused a small amount of variation in Arrington, D. A., and K. O. Winemiller. 2002. Preservation effects on stable iso- δ15N. tope analysis of fish muscle. Transactions of the American Fisheries Society 131:337–342. Conclusion Bjornn, T. C. 1971. Trout and salmon movements in two Idaho streams as related to temperature, food, stream flow, cover, and population density. Transactions In light of these results, we suggest that isotopic ratios of of the American Fisheries Society 100:423–438. both carbon and nitrogen in ethanol-preserved tissues be inter- Bugoni, L., R. A. R. McGill, and R. W. Furness. 2008. Effects of preservation preted with caution, and we recommend that future studies be methods on stable isotope signatures in bird tissues. Rapid Communications conducted with hatchery fish or laboratory organisms in order in Mass Spectrometry 22:2457–2462. to avoid limited sample sizes. The changes in fin δ13C observed Chaloner, D. T., M. S. Wipfli, and J. P. Caouette. 2002. Mass loss and macroin- vertebrate colonisation of Pacific salmon carcasses in south-eastern Alaskan for all groups following ethanol preservation were greater in streams. Freshwater Biology 47:263–273. magnitude than those for δ15N, and they were relatively more DeNiro, M. J., and S. Epstein. 1977. Mechanism of carbon isotope fractionation consistent. It should be noted that the biological meaningfulness associated with lipid synthesis. Science 197:261–263. of these shifts is entirely context dependent. In many ecosys- Finlay, J. C. 2001. Stable-carbon-isotope ratios of river biota: implications for tems, δ13C of carbon sources differs by more than by 5‰ (Finlay energy flow in lotic food webs. Ecology 82:1052–1064. Hanisch, J. R., W. M. Tonn, C. A. Paszkowski, and G. J. Scrimgeour. 2010. δ13C 2001). Therefore, if samples from other taxa in the food web and δ15N signatures in muscle and fin tissues: nonlethal sampling methods were subject to the same preservation bias, preserving samples for stable isotope analysis of salmonids. North American Journal of Fisheries in ethanol could be appropriate for determining carbon sources Management 30:1–11. in some instances. Hobson, K. A. 1999. Tracing origins and migration of wildlife using stable Alternatively, the typical shift between trophic levels for δ15N isotopes: a review. Oecologia 120:314–326. Hobson, K. A., M. L. Gloutney, and H. L. Gibbs. 1997. Preservation of blood is 3–4‰ (Vander Zanden and Rasmussen 2001) so ethanol- and tissue samples for stable-carbon and stable-nitrogen isotope analysis. preserved tissues could probably be used to document stream Canadian Journal of Zoology 75:1720–1723. food web structure, assuming species-specific preservation ef- Kaehler, S., and E. A. Pakhomov. 2001. Effects of storage and preservation on 13 15 fects are understood and their intraspecific variation is low. Like the δ Candδ N signatures of selected marine organisms. Marine Ecology B. Kelly et al. (2006), we believe that future studies should de- Progress Series 219:299–304. Kelly, B., J. B. Dempson, and M. Power. 2006. The effects of preservation on velop species-specific and tissue-specific calibrations for stable fish tissue stable isotope signatures. Journal of Fish Biology 69:1595–1611. isotope samples undergoing chemical preservation, but we wish Kelly, M. H., W. G. Hagar, T. D. Jardine, and R. A. Cunjak. 2006. Nonlethal to caution that such models do not currently account for intraspe- sampling of sunfish and Slimy Sculpin for stable isotope analysis: how scale cific variation in tissue composition. Future research should and fin tissue compare with muscle tissue. North American Journal of Fish- be done to determine the mechanism by which preservation- eries Management 26:921–925. Logan, J. M., T. D. Jardine, T. J. Miller, S. E. Bunn, R. A. Cunjak, and M. δ13 induced increases in C correspond with decreases in tissue E. Lutcavage. 2008. Lipid corrections in carbon and nitrogen stable isotope C:N ratio and whether these isotopic shifts can be linked to analyses: comparison of chemical extraction and modelling methods. Journal the loss of lipids or to some other carbon-containing compound of Animal Ecology 77:838–846. such as bone collagen. Increasing the number of laboratory ex- Mart´ınez del Rio, C., N. Wolf, S. A. Carleton, and L. Z. Gannes. 2009. Isotopic periments aimed at determining the mechanisms that explain ecology ten years after a call for more laboratory experiments. Biological Reviews 84:91–111. variability in stable isotope patterns will undoubtedly enhance Peterson, B. J., and B. Fry. 1987. Stable isotopes in ecosystem studies. Annual the ecological application of these analytical tools (Mart´ınez del Review of Ecology and Systematics 18:293–320. Rio et al. 2009). Pinnegar, J. K., and N. V. C. Polunin. 1999. Differential fractionation of δ13C and δ15N among fish tissues: implications for the study of trophic interactions.

Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 Functional Ecology 13:225–231. ACKNOWLEDGMENTS Post, D. M. 2002. Using stable isotopes to estimate trophic position: models, We thank participants of the University of Washington In- methods, and assumptions. Ecology 83:703–718. Post, D. M., C. A. Layman, D. A. Arrington, G. Takimoto, J. Quattrochi, and ternship Program and the Watershed Program at the Northwest C. G. Montana.˜ 2007. Getting to the fat of the matter: models, methods and Fisheries Science Center for providing field support as well as assumptions for dealing with lipids in stable isotope analyses. Oecologia Vija Pelekis for assisting in the field and laboratory. William 152:179–189. Reichert, Jennie Bolton, and other members of the Environ- Sanderson, B. L., C. D. Tran, H. J. Coe, V. Pelekis, E. A. Steel, and W. L. mental Assessment Program were instrumental in analyzing the Reichert. 2009. Nonlethal sampling of fish caudal fins yields valuable sta- ble isotope data for threatened and endangered fishes. Transactions of the stable isotope samples. Mark Scheuerell and Jim Faulkner of- American Fisheries Society 138:1166–1177. fered invaluable suggestions on the statistical analyses. Dave Sarakinos, H. C., M. L. Johnson, and M. J. Vander Zanden. 2002. A synthesis of Herman and Rich Zabel also contributed insightful comments tissue-preservation effects on carbon and nitrogen stable isotope signatures. on the manuscript. This project was funded by the National Canadian Journal of Zoology 80:381–387. Oceanic and Atmospheric Administration, Fisheries. Reference Sweeting, C. J., N. V. C. Polunin, and S. Jennings. 2004. Tissue and fixative dependent shifts of δ13Candδ15N in preserved ecological material. Rapid to trade names does not imply endorsement by the National Communications in Mass Spectrometry 18:2587–2592. Marine Fisheries Service, National Oceanic and Atmospheric Sweeting, C. J., N. V. C. Polunin, and S. Jennings. 2006. Effects of chem- Administration. ical lipid extraction and arithmetic lipid correction on stable isotope 1476 VIZZA ET AL.

ratios of fish tissues. Rapid Communications in Mass Spectrometry 20:595– Syvaranta,¨ J., S. Vesala,M. Rask, J. Ruuhijarvi,¨ and R. I. Jones. 2008. Evaluating 601. the utility of stable isotope analyses of archived freshwater sample materials. Syvaranta,¨ J., A. Martino, D. Kopp, R. Cer´ eghino,´ and F. Santoul. 2011. Freezing Hydrobiologia 600:121–130. and chemical preservatives alter the stable isotope values of carbon and Vander Zanden, M. J., and J. B. Rasmussen. 2001. Variation in δ15Nandδ13C nitrogen of the Asiatic clam (Corbicula fluminea). Hydrobiologia 658:383– trophic fractionation: implications for aquatic food web studies. Limnology 388. and Oceanography 46:2061–2066. Downloaded by [Department Of Fisheries] at 21:40 27 October 2013 This article was downloaded by: [Department Of Fisheries] On: 27 October 2013, At: 21:44 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 A Review of “Ecology of Australian Freshwater Fishes” Tim M. Berra a b a Charles Darwin University , Darwin , Northern Territory , Australia b Department of Evolution, Ecology, and Organismal Biology , Ohio State University , Mansfield , Ohio , 44906 , USA Published online: 06 Sep 2013.

To cite this article: Tim M. Berra (2013) A Review of “Ecology of Australian Freshwater Fishes”, Transactions of the American Fisheries Society, 142:5, 1477-1480, DOI: 10.1080/00028487.2013.836029 To link to this article: http://dx.doi.org/10.1080/00028487.2013.836029

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. 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. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Transactions of the American Fisheries Society 142:1477–1480, 2013 C American Fisheries Society 2013 ISSN: 0002-8487 print / 1548-8659 online DOI: 10.1080/00028487.2013.836029

BOOK REVIEW

Ecology of Australian Freshwater Fishes. Edited by Paul Humphries southwestern province is a hotbed of endemism for fishes as and Keith Walker. CSIRO Publishing, Collingwood, Victoria, Australia. well as plants and other biota due to its prolonged isolation 2013. 423 pages. $A130.00 (hardcover). by arid regions, yet it shares some relatives with southeastern Australia. I wish, however, that the chapter had included the This book is a synthesis of fisheries biology and ecology. It vicariance versus dispersal arguments of the 1970s that cen- applies Northern Hemisphere experiences to the unique Aus- tered on Common Galaxias Galaxias maculatus (also known as tralian context of a fish fauna evolved in isolation on the driest Inanga) and so consumed graduate students of that era (Berra and flattest continent on the Earth. The authorship includes 23 et al. 1996). well-known aquatic biologists. Each of the 13 chapters is stand- Chapter 3 is a primer on genetics for the fish ecologist, a alone, with approximately 2,088 references at the end of the topic that informs taxonomy, dispersal, management, and con- book, the most recent of which are from 2012. servation. One of the most interesting and useful aspects of After paying homage to aquatic ecology pioneers Stephen this chapter by Hammer, Adams, and Hughes is Table 3.2. It Forbes, G. E. Hutchison, and Noel Hynes, the editors point out gives the relative expense and technical expertise required for that only two Antipodean ecologists, R. M. McDowall (New obtaining data such as morphology, karyotypes, allozymes, am- Zealand) and J. Allen Keast (Australia) achieved much interna- plified fragmented length polymorphisms, mtDNA, introns, ex- tional recognition in the 20th century. This is offered as a motive ons, microsatellites, and single-nucleotide polymorphisms. A for the book. brief summary of each technique (including DNA bar coding Fish ecology received a boost from the 1949 studies of Shelby and GENETAG) shows how and for what purpose each method Gerking, and the principles were codified by George Nikolsky. can be used. Important journals like the Transactions of the American Fish- The chapter on habitats by Koehn and Kennard describes eries Society prospered, and classic works such as Matthew’s the diversity of Australia’s freshwater environments from large 1998 Patterns in Freshwater Fish Ecology emerged. I recom- rivers such as the Murray–Darling system to ephemeral streams. mend that Saunders (2013) be consulted for additional historical The demands of droughts and floods are embedded in the DNA presentations. of Australia’s biota. This unpredictability has consequences for After the stage is set, the origin of the Australian freshwa- fish faunas throughout much of the continent. Tropical northern ter fish fauna is discussed. This fauna consists of about 256 Australia, on the other hand, has predictable wet and dry seasons, species that are listed alphabetically by family in the appendix. which present a different set of circumstances. Natural lentic Only three species of Gondwanan origins are present: one lung- habitats are few and far between, except in Tasmania. There are fish (Neoceratodus) and two osteoglossids (Scleropages). The no “Great Lakes” of Australia as in North America or Africa, scientific name of the Spotted Bonytongue, S. leichardti, is con- but oxbow lakes (billabongs) are common. Reservoirs have been sistently misspelled throughout the book with “hh” as in Leich- formed by impounding rivers with a loss of native fish habitat. hardt’s name (Berra 1989), although the correct spelling is used Lake Eyre, a gigantic basin in arid northern South Australia, fills

Downloaded by [Department Of Fisheries] at 21:44 27 October 2013 in the index. A review of Australian fish biology highlights the with water and fishes about every 50 years, but oddly it is not work of Gilbert Whitley, John Lake, and Gerry Allen. mentioned in this chapter or in the index. Hydrology, hydraulics, Peter Unmack’s review of Australian ichthyogeography water quality, and physical structure are discussed, with many points out that the fauna is dominated by acanthopterygians, examples of how various species cope. For example, the iconic not the ostariophysians that are so common on other continents. Murray Cod Maccullochella peeli and two of its congeners are Only two native ostariophysian families (Ariidae and Plotosi- consistently associated with woody debris. dae) and four endemic families (Neoceratodontidae, Lepido- Koehn and Crook present a master class on movements and galaxiidae, Melanotaenidae, and Pseudomugilidae) occur in the migration in Chapter 5. We learn that Golden Perch Macquaria freshwaters of Australia and New Guinea (A–NG). Limited ge- ambigua move up to 2,300 km through inland rivers. Such large- ological relief has allowed dispersal across low barriers. The scale movement wholly within freshwater is termed potamod- fauna is about 91% endemic at the species level. More species romy. Aboriginal people had knowledge of fish migrations and remain to be described in the 36 families. Most of the freshwater used stone fish traps to catch Golden Perch, Murray Cod, and fish species of A–NG have marine affinities, but not necessar- the catadromous Southern Shortfin Eel Anguilla australis (also ily recent ones. The Spangled Perch Leiopotherapon unicolor known as just the Shortfin Eel) for millennia. The authors pro- is Australia’s most widespread freshwater fish. Australia’s vide a table of 16 terms to explain various fish movements.

1477 1478 BOOK REVIEW

The mechanics of swimming is discussed in detail. Home range more and reach 1.4 m TL. In terms of approach, the authors and homing as well as diadromy, catadromy, and anadromy are round up the usual suspects, such as length-frequency analysis, illustrated by many Australian examples. Amphidromy (a type scales, otoliths, and the venerable von Bertalanffy equation. of diadromy between freshwater and the sea not for spawning) Data on the validation of aging methods for many Australian is illustrated by the Australian Grayling Prototroctes maraena. species is lacking. Otoliths are especially useful since they grow Both high- and low-tech methods of studying fish movements continuously and their chemical structure is not remetabolized. are summarized, including passive integrated transponder tags, As a general rule, small species mature in their first year and telemetry, and otolith chemistry (whereby variations in the ratio live 1–5 years, whereas large-bodied fishes mature later and of trace elements between the cores and edges of otoliths are live longer. As ectotherms, fish have very plastic growth rates thought to show movements between marine and freshwater en- and continue to grow throughout life. Environmental conditions vironments). There is much to recommend this chapter to fish play a large role in fish growth rates. Larger fishes tend to have ecologists who are not experts in this subspecialty. great reproductive potential, and females usually are larger than Stoffel’s chapter on trophic ecology covers the ways in which males of the same species. To its credit, this chapter is strongly fish sense and ingest prey and the bioenergetics and nutrition of based on the fish physiology that explains the methods of the various diets. Trophic guilds are discussed, and food webs are fisheries ecologist. spun in an Australian context. This involves knowledge of the Population dynamics is the foundation upon which manage- physics and chemistry of fish vision, mechanoreception, and ment decisions are made, and this is the subject of Chapter 9 chemoreception. A major contribution of this chapter is Fig- by Harris, Bond, Closs, Gehrke, Nicol and Ye. Metapopula- ure 6.17, which shows 92 fish species arranged in six clades tions (populations of populations), catch per unit effort, recruit- of piscivores, algivores–detritivores, surface carnivores, micro- ment, and other subjects are explained. Table 9.1 is a compila- crustacivores, aquatic insectivores, and omnivores. At least one tion of demographic data (longevity, size, age at maturity, and of the papers cited for this figure (Berra et al. 1987) is missing fecundity) for 39 species. Such summary sources are reason from the references, and another that could have been included enough to own this book. Desiccating Lake Eyre and the strand- is not (Berra and Wedd 2001). ing of 40 million Bony Herring Nematalosa erebi and hardy- Reproduction and early life history are expertly handled heads Craterocephalus spp. get a mention in this chapter. A by King, Humphries, and McCasher in Chapter 7. K- and r- boxed presentation of coldwater pollution from reservoirs shows selection theory and semelparous versus iteroparous breeding how the warmwater Murray Cod has been adversely affected. strategies are explained with Australian examples. Of the 92 Mitigation ideas are offered. Today, most commercial harvest in Australian freshwater fish species studied, about 35% show Australian freshwaters is for alien species. Recreational fishing some form of parental care, which is almost always provided is the dominant form of exploitation of native fishes. Australia by the male. Australia’s premier game fish, Barramundi Lates ranked only 55th among nations in total fisheries production calcarifer (also known as Barramundi Perch), is a protandrous in 1989. Four native freshwater fishes are commercially ex- hermaphrodite that transitions from male to female at about ploited: Barramundi, Golden Perch, Longfin Eel A. reinhardti, 80–100 cm after spawning in brackish waters at 6–8 years old. and Southern Shortfin Eel. Redfin Perch Perca fluviatilis (also Protogyny (female-to-male conversion) is the more usual pattern known as Eurasian Perch) and Common Carp Cyprinus carpio among hermaphroditic fishes, but this is rare in Australia. Fig- are the two alien species that support commercial fishing. Three ure 7.10 is an informative graphic representation of the spawn- salmonids (Atlantic Salmon Salmo salar, Brown Trout S. trutta, ing time, rainfall, and temperature patterns in five regions of and Rainbow Trout Oncorhynchus mykiss) are raised in aqua- Australia. One can see that many species spawn in the trop- culture. Many other native species (including Murray Cod) were Downloaded by [Department Of Fisheries] at 21:44 27 October 2013 ical Alligator Rivers of the Northern Territory from Septem- previously exploited commercially, but overfishing, habitat de- ber to December on the buildup to the wet season. This can struction, alien species, and other anthropogenic factors have be compared with southwestern Australia or Tasmania, where had a deleterious impact on stocks. Australians are mad-keen rainfall and temperature patterns are very different. Table 7.3 anglers, and about 20% of the population over 5 years old fish shows the reproductive guilds of Australian freshwater fishes. annually and spend $A1.8 billion doing so. About 20% of this Early life history stages are reviewed, along with anthropogenic effort is in freshwaters. Freshwater fishing is very important to disturbances. the lives and cultural traditions of Aboriginal people, especially Crook and Gillander do their best to make age and growth in the Northern Territory. The pros and cons of stocking pro- interesting in their comprehensive chapter. How old is a 30- grams are debated at the end of the chapter. There is a strong kg Murray Cod? It could be 20–48 years old or more. This conservation ethic among most anglers and indigenous people. demonstrates the difficulty of pinning down a specific answer This chapter has much food for thought for the angler as well in animals with highly variable growth rates. Yet management as the professional fisheries biologist. and conservation decisions depend on having such answers. A group of species that occur together in a single locality One of the geologically oldest fish species on the planet, the is considered an assemblage, and this is the topic of Chap- Australian Lungfish Neoceratodus forsteri, can live 65 years or ter 10 by Arthington, Kennard, Pusey, and Balcombe. Abiotic BOOK REVIEW 1479

and biotic factors interact to structure assemblages, and it and agricultural influences are greatest. Fish populations in the is necessary to understand these interactions to positively sparsely populated north are in better shape. Table 12.1 lists influence physical restoration efforts. Both geological and hu- 49 taxa and their various categories, such as critically endan- man history affect which species and how many will be avail- gered, endangered, vulnerable, etc. Members of the Galaxiidae able in the regional species pool, and then various abiotic and and Percichthyidae have the most problems. The Pedder Galax- biotic filters determine the composition of the local assemblage. ias Galaxias pedderensis, a Tasmanian endemic, is extinct in For example, Australia’s aridity is a major filter and may yield the wild but survives in two translocated populations. All four depauperate assemblages. Lake Eyre, the world’s 18th largest species of cod (Maccullochella) are considered endangered or lake (when it is actually full), is discussed in more detail here, vulnerable and management plans are in place, but these plans along with its rarely connecting stream (more often a string may vary from state to state. Some threatened species may be of water holes), Cooper Creek. Various assemblages, such as captured and retained, whereas others must be released if caught. those of springs, coastal lakes, and rivers, and their filters and There are also some unknowns. Do hooked fish survive release? species are reviewed. Both species richness and piscivory in- Can anglers differentiate closely related species? The principal crease downstream. A complex web of interacting abiotic fac- threats to fish include habitat modification, altered flow, wa- tors is diagramed in Figure 10.3. Predation and competition ter quality, barriers, alien species, translocation and stocking, are two important biotic filters. Local assemblages that in- and overfishing. These threats are explored with specific ex- clude some dispersal and interaction between them constitute amples and management responses. The Murray–Darling basin metacommunities. provides world-class examples of fishway solutions along the John Harris is the author of the very important chapter enti- 2,225 km from the mouth of the Murray River to Hume Dam. tled “Fishes from Elsewhere.” This title applies to exotic, alien, Since 2000, more than 79 million fish of 15 species have been and translocated native species–and even to different genetic stocked in New South Wales and Victoria. About half of these stocks of native species. In this context, “exotic” species are were salmonids and half were natives, with Golden Perch and defined as ones that do not exist in the wild but rather are held Murray Cod predominant among the natives. The freshwater captive in aquaculture or aquariums. They are potential “alien” fish harvest consists mainly of Common Carp, Redfin Perch, species, which are defined as “imported and established.” Ta- Golden Perch, salmonids, Australian Bass Percalates novemac- ble 11.2 lists 43 such alien fishes. There are 18 cichlids, 8 ulata, Barramundi, and Murray Cod. cyprinids, 6 poeciliids, and 5 salmonids on the list. All are from The final chapter by the editors reflects on where Australian west of Wallace’s Line. The International Union for the Con- freshwater fish ecology has been and where it needs to go. servation of Nature considers five of these taxa to be among the A scenario for 2050 is proposed in which “we must prepare 100 most invasive species in the world: Brown Trout, Common for change, rather than resist it.” After all, isn’t this what the Carp, Redfin Perch, Eastern Gambusia Gambusia holbrooki, Australian freshwater fishes have done through their life history and Mozambique Tilapia Oreochromis mossambicus. Acclima- strategies? It is called evolution. tization societies brought in many European species for food My quibbles with the book, already mentioned and other- and recreation in the 19th century. Some species escaped from wise, are minor. For example, in Figure 1.1 (an Aboriginal rock the aquarium trade. Australia’s northern waters are ideal for in- painting) an eeltail catfish (Plotosidae) is clearly shown, but it vasive tropical aliens. Since Australia has no major freshwater is identified as a fork-tailed catfish (Ariidae). On page 5, the shipping ports, it has avoided ballast-water alien introductions American Society of Ichthyologists and Herpetologists is mis- as experienced in the North American Great Lakes. Murray named, and the correct date of its founding is 1913 rather than Cod are highly prized, and there has been pressure to translo- 1914 (Berra 1984). Downloaded by [Department Of Fisheries] at 21:44 27 October 2013 cate them widely. This brings up the issue of hybridization of I really enjoyed this book. It made me homesick for all the different genetic stocks of the same species. A total of 76 native people, places, and species I’ve encountered in a 45-year career species in 28 families have been translocated, mostly in eastern devoted to studying weird Australian fishes. Anyone interested Australia. Salmonids are generally thought by most Australians enough to read this review and the journal it is published in will to be a “good” introduction, but ecologists consider them to benefit from owning this book. be detrimental to native coolwater species such as blackfish (Gadopsidae), Australian Grayling, Trout-Cod Maccullochella REFERENCES macquariensis, some galaxiids, and other fishes. Various eradi- Berra, T. M. 1984. A chronology of the American Society of Ichthyologists cation plans are in place, including the “daughterless carp” pro- and Herpetologists through 1982. American Society of Ichthyologists and gram that attempts to manipulate sex ratios to eliminate females. Herpetologists, Special Publication 2, Gainesville, Florida. The characteristics of a successful invader and their impact on Berra, T. M. 1989. Scleropages leichardti Gunther¨ (Osteoglossiformes): the case native species are discussed with real examples. of the missing H. Bulletin of the Australian Society for Limnology 12:15–19. Berra, T. M., A. Campbell, and P. D. Jackson. 1987. Diet of the Australian Conservation and management is tackled by Lintermans. grayling, Prototroctes maraena Gunther¨ (Salmoniformes: Prototroctidae), Many Australian freshwater fish species are in decline, espe- with notes on the occurrence of a trematode parasite and black peritoneum. cially in the southern part of the continent where the population Australian Journal of Marine and Freshwater Research 38:661–669. 1480 BOOK REVIEW

Berra, T. M., L. E. L. M. Crowley, W. Ivantsoff, and P. A. Fuerst. 1996. Galax- TIM M. BERRA ias maculatus: an explanation of its biogeography. Marine and Freshwater Research 47:845–849. Berra, T. M., and D. Wedd. 2001. Alimentary canal anatomy and diet of the Charles Darwin University, nurseryfish, Kurtus gulliveri (Perciformes: Kurtidae), from the Northern Ter- Darwin, Northern Territory, Australia; ritory of Australia. The Beagle, Records of the Museums and Art Galleries of the Northern Territory 17:21–25. and Department of Evolution, Ecology, Saunders, B. 2012. Discovery of Australia’s fishes: a history of Australian and Organismal Biology, Ohio State University, ichthyology. CSIRO Publishing, Collingwood, Victoria, Australia. Mansfield, Ohio 44906, USA Downloaded by [Department Of Fisheries] at 21:44 27 October 2013