Rearing environment influences boldness and prey acquisitionbehavior, and brain and lens development of bull trout

William R. Brignon & Martin M. Pike &

Lars O. E. Ebbesson & Howard A. Schaller &

James T. Peterson & Carl B. Schreck

Received: 28 June 2017 /Accepted: 4 December 2017 /Published online: 16 December 2017# Springer Science+Business Media B.V., part of Springer Nature 2017

Abstract Animals reared in barren captive environ-ments exhibit different developmental trajectories andbehaviors than wild counterparts. Hence, the captivephenotypes may influence the success of reintroductionand recovery programs for threatened and endangeredspecies. We collected wild bull trout embryos from theMetolius River Basin, Oregon and reared them in dif-fering environments to better understand how captivityaffects the bull trout Salvelinus confluentus phenotype.We compared the boldness and prey acquisition behav-iors and development of the brain and eye lens of bull

trout reared in conventional barren and more structurallycomplex captive environments with that of wild fish.Wild fish and captive reared fish from complex habitatsexhibited a greater level of boldness and prey acquisi-tion ability, than fish reared in conventional captiveenvironments. In addition, the eye lens of convention-ally reared bull trout was larger than complex rearedcaptive fish or same age wild fish. Interestingly, wedetected wild fish had a smaller relative cerebellum thaneither captive reared treatment. Our results suggest thatrearing fish in more complex captive environments cancreate a more wild-like phenotype than conventionalrearing practices. A better understanding of the effectsof captivity on the development and behavior of bulltrout can inform rearing and reintroduction programsthough prediction of the performance of releasedindividuals.

Keywords Boldness behavior . Behavioraldevelopment . Behavioral plasticity . Braindevelopment . Eye development . Species conservation .

Species recovery . Bull trout

Introduction

Phenotypic plasticity results in varying forms of physi-ology, morphology, and behavior and is a function ofgenetics and the environment in which the animal lives(West-Eberhard 1989). Homogeneous captive environ-ments can result in animals that are phenotypically anddevelopmentally different than those reared in more

Environ Biol Fish (2018) 101:383–401https://doi.org/10.1007/s10641-017-0705-z

W. R. Brignon (*)Columbia River Fisheries Program Office, United States Fish andWildlife Service, 1211 SE Cardinal Court, Suite 100, Vancouver,WA 98683, USAe-mail: bill_brignon@fws.gov

W. R. Brignon : J. T. Peterson :C. B. SchreckU.S. Geological Survey, Oregon Cooperative Fish and WildlifeResearch Unit, U.S.G.S., Oregon State University, 104 Nash Hall,Corvallis, OR 97331-3803, USA

M. M. PikeAdvanced Imaging Research Center, Oregon Health and ScienceUniversity, 3181 SW Sam Jackson Park Rd, L452, Portland, OR97239, USA

L. O. E. EbbessonUni Research Environment, University of Bergen,Thormøhlensgate 53 A/B, Bergen, Norway

H. A. SchallerFish and Aquatic Conservation Program, United States Fish andWildlife Service, Pacific Regional Office, 911 NE 1th Avenue,Portland, OR 97232, USA

heterogeneous or wild environments. Wild phenotypesare established out of necessity to survive in stochasticand uncertain natural habitats whereas captive pheno-types arise from a relaxation of, or shift in, the selectivepressures experienced in man-made environments(Lorenzen et al. 2012; Teletchea 2017). Phenotypicdifferences between hatchery and wild fish can be seenin morphology (Taylor 1986; Currens et al. 1989), braindevelopment (Marchetti and Nevitt 2003; Kihslingeret al. 2006; Kihslinger and Nevitt 2006), physiology(Woodward and Strange 1987; Dickens et al. 2010),lateral line development (Brown et al. 2013), and be-havior (Fernö et al. 2011; Salvanes et al. 2013), and mayinfluence post-release survival in captive-reared salmo-nids. Low post-release survival results in limited num-bers of adults available to support fishery programs.This can be counteracted to some extent with increasedrelease numbers thereby providing sufficient animals forharvest. However, the consequences of reduced survivalmay limit the efficacy and success of conservation andrecovery programs if captive-reared fish are releasedinto wild habitats with the expectation that they spawnnaturally and create self-sustaining populations(Lorenzen 2014). In these types of programs the goalshould be to raise and release fish that are developmen-tally and behaviorally similar to wild fish (Brown andDay 2002).

The effect of rearing environment on salmonid braindevelopment and cognition has received increased at-tention in the last few years (Ebbesson and Braithwaite2012). Fish reared in conventional hatchery and labora-tory environments that lack structural complexity havebeen shown to develop significantly smaller brains thantheir wild counterparts (Marchetti and Nevitt 2003;Kihslinger et al. 2006; Burns et al. 2009). It is possibleto produce fish with brains more similar in size to wildfish by adding complexity to the captive rearing envi-ronment (Kihslinger and Nevitt 2006). However, re-gardless of whether a fish is raised in a conventional orcomplex captive environment their brains are still sig-nificantly smaller than size matched wild fish collectedfrom the same river basin as the captive stock(Kihslinger et al. 2006).

The teleost brain includes four primary regions. Thelateral part of the dorsal telencephalon has been impli-cated in complex spatial learning and memory (Lópezet al. 2000; Rodríguez et al. 2002; Broglio et al. 2003;Wilson and McLaughlin 2010); the cerebellum coordi-nates movements and controls associative learning,

classical conditioning, and basic spatial cognition(Rodríguez et al. 2005; Broglio et al. 2011); the optictectum controls the processing of sensory information(Broglio et al. 2011), and the olfactory bulb processesodors from the environment (Gonda et al. 2012). Un-derstanding the function of a particular brain region canbe used to correlate development caused by rearingenvironment with the cognitive ability and behavior ofthe animal.

The influence of telencephalon size on fish behaviorcan be seen as early as emergence with individuals thatpossess a large telencephalon using more space whenforaging relative to individuals with a smaller telenceph-alon (Wilson and McLaughlin 2010). These differencesmay give rise to the development of diverse resourcepolymorphs as fish mature. For example, benthicthreespine stickleback Gasterosteus aculeatus have alarger telencephalon than pelagic conspecifics whichcould allow for enhanced spatial learning in a morestructurally complex habitat (Park et al. 2012). Thesedevelopmental differences can alter cognitive abilityand result in fish that are more likely to make quickand potentially inaccurate decisions while moving inspace (Burns and Rodd 2008). This suggests it is equallyimportant to evaluate difference in brain developmentbetween treatments as well as determining the conse-quences to the cognitive abilities and behavioral ecologyof the individual (Kihslinger et al. 2006; Healy andRowe 2007; Ebbesson and Braithwaite 2012).

A fish uses its sensory organs to communicate infor-mation about its surroundings to the brain. Abioticcharacteristics of these surroundings can influence thedevelopment of sensory organs, and ultimately behav-ior. Hatchery fall Chinook salmon Oncorhynchustshawytscha reared in clear water developed larger pu-pils than wild reared fish from the more turbid SnakeRiver (Tiffan and Connor 2011). Similarly, graylingThymallus thymallus and Arctic char Salvelinus alpinusthat inhabit dark lake habitats exhibit smaller eye diam-eter than brown trout Salmo trutta and Atlantic salmonSalmo salar from clear riverine habitats (Pakkasmaaet al. 1998). It is difficult to determine if these differ-ences in eye size are a function of the rearing environ-ment, inheritance, species or a combination of factors.McPhail (1984) provides evidence that this is a heritabletrait and found that limnetic three-spine stickleback hadlarger eye diameters than benthic individuals in bothlaboratory and wild reared individuals. A fish focusesits vision by moving the eye lenses towards or away

384 Environ Biol Fish (2018) 101:383–401

from the pupil thereby changing the field of view(Kröger 2011). Given these mechanics, a larger eyemay allow for greater visual acuity and prey detection(Hairston et al. 1982).

When rearing captive fish for reintroduction pro-grams it is important to consider the developmentaltrajectory and the behavioral phenotypes it produces.For each developmental trajectory there can be multiplebehavioral phenotypes, suggesting a wide scope of be-havioral plasticity (West-Eberhard 1989). The behavior-al phenotype of captive reared animals intended forreintroduction and recovery programs should includethe ability to avoid predators, locate food, interact withconspecifics, locate or build shelter and nests, and nav-igate and move in novel and complex environments(Kleiman 1996; Brown and Day 2002). An environmen-tal effect on these behaviors can begin as soon as em-bryogenesis and the influence can last a lifetime(Jonsson and Jonsson 2014).

Behavioral effects of adding complexity to captiveenvironments is well documented in salmonids andother fishes (see review by Näslund and Johnsson2016). Captive fish, reared in structurally complex hab-itats, exhibit greater exploratory behavior, social domi-nance (Berejikian et al. 2000; Berejikian et al. 2001),and propensity to seek shelter in a novel environment(Näslund et al. 2013) and forage on novel live prey(Brown et al. 2003) than fish reared in conventionalcaptive environments. Exactly what type of, or amountof, habitat complexity that is necessary to raise a wild-like fish will depend on the species, life stage, naturalhistory, and preferences of the animal (Näslund andJohnsson 2016). Animals that are less fit for survivalin the wild may partly explain why reintroductionsemploying hatchery reared individuals failed 71% ofthe time due to poor recruitment and 77% as definedby authors of the various studies (Cochran-Biedermanet al. 2015). These high instances of failure suggest thereis room for improvement when using captive rearingstrategies to recover and reintroduce threatened andendangered fishes and managers employing such strat-egies should proceed with caution.

Our goal was to better understand the effects ofcaptive rearing environments on bull trout Salvelinusconfluentus and how these captive effects influence theefficacy and ultimate success of reintroduction and re-covery programs. The study objectives were to evaluatethe development and behavior of juvenile bull troutreared in three environments; conventional and complex

captive environments, and wild environments. To thisend, we evaluated boldness and prey acquisition; twobehaviors that are important for post-release survival.Boldness is defined as the propensity to take risks andexplore novel habitats (Brown and Braithwaite 2004;Brown et al. 2007). In addition, we compared the de-velopment of the whole brain, telencephalon, cerebel-lum, and optic tectum as well as the development of theeye lens to better understand the effects of rearing envi-ronment on cognitive ability to perceive environment.

Materials and methods

Ethics statement

Federal and State collection and transport permits wereobtained for these animals and the original researchreported herein was performed under guidelinesestablished by the Institutional Animal Care and UseCommittee at Oregon State University. The AnimalCare and Use Program at Oregon State University isAssociation for Assessment and Accreditation of Labo-ratory Animal Care International (AAALAC)accredited.

Collection, incubation, and rearing

In November 2012, eyed bull trout embryos werecollected from three redds in Canyon Creek, Oregon,a tributary to theMetolius River. Due to the threatenedstatus of the fish, hydraulic sampling as described byMcNeil (1964) and Berejikian et al. (2011) was usedto collect embryos. Hydraulic sampling of embryosallows for collection of a portion of the redd andfocuses on a life stage that exhibits relatively highmortality rates as compared to older life stages.Timing of sampling was based on known thermalhistory (~ 5 °C) of the embryos and bull trout devel-opmental rates identified from the literature. Embryoswere transferred to the Fish Performance and GeneticsLaboratory at Oregon State University for incubationand rearing. Eggs from a redd were divided in half andincubated in Heath trays with either no substrate (i.e.,conventional habitat) or with pea gravel as substrate(i.e., complex habitat). Heath trays were supplied with~6 °C chilled well water. In late January 2013, afterhatching and before yolk sac absorption, alevins weretransferred by redd and rearing treatment to small light

Environ Biol Fish (2018) 101:383–401 385

green rearing tanks (0.5 m long × 0.33 m wide ×0.33 m deep) supplied with 8 °C to 10 °C chilled wellwater at ~0.2 m deep. The conventional rearing tankswere barren with no added structure and a base of riverrock (2.5 cm to 5 cm diameter) substrate was added tothe complex rearing tanks. The tanks were coveredwith shade cloth to minimize stress from outside ac-tivity and fry were hand fed a combination of frozenbrine shrimp, brine shrimp flakes, and commercialstarter feed (BioClarks Starter) in the morning, eve-ning, and at night. Once active feeding, fish were fedmorning and evening on commercial diets at the man-ufacturer (BioOregon) recommended feed size andamounts. In early June 2013 all family groups werecombined and transferred to large light blue troughsfor extended rearing. Troughs measured 4.9 m long ×1.0 m wide × 0.8 m deep and were supplied with 8 °Cto 13 °C well water that was ~20 cm deep. The con-ventional rearing treatment was void of any habitatenhancement and the complex rearing trough had abase of river rock similar to the complex tank rearingenvironment and both treatment tanks were coveredwith shade cloth. Fish were reared under these condi-tions until behavioral trials began. Each fish wastagged with a 12 mm passive integrated transponder(PIT) in August 2013 when their minimum size was65 mm.

Wild bull trout were collected in late May, 2014from screw traps in Canyon Creek, Jack Creek, orCandle creek; all tributaries to the Metolius River.The fish were transferred to the Fish Performanceand Genetics Laboratory at Oregon State University,PIT tagged, and held in a trough outside of the facility.The trough was 4.9 m long × 1.0 m wide × 0.8 m deep,dark green in color, covered with shade cloth, andcontained river rock and larger rocks for direct over-head cover. Wild fish were provided live spring chi-nook and commercial feed for two days prior to be-ginning trials.

Spring Chinook salmon used in the prey acquisi-tion trials were of hatchery origin obtained as eggsfrom the South Santiam River (Oregon Department ofFish and Wildlife) stock. Chinook salmon hatched inlate December 2013 and were incubated in heath traysand reared in circular tanks measuring 1.0 m diameter× 0.6 m deep. Heath trays and tanks consisted ofsimilar habitat treatments as the captive reared bulltrout (i.e., conventional and complex). Tanks weresupplied with well water ranging from 8 °C to 13 °C

and fish were fed daily with commercial diets at themanufacturer (BioOregon) recommended feed sizeand quantity.

Boldness trials

To test boldness we used four replicate behavioralarenas at the Fish Performance and Genetics Laboratory(Fig. 1a and b), each consisting of a 25 cm PVC bafflesheeting placed in the middle of a fiberglass tank (1 mdeep × 1mwide × 2m long). The baffle divided the tankinto two chambers which were connected via a PVCpassage tube that measured 6.4 cm in diameter. Thepassage tube was flush with the baffle and located6 cm off the bottom of the tank and in the middle ofthe baffle width (Fig. 1b). Access to the passage tubewas blocked until the trial began and then opened via apulley system to minimize impact on fish behavior.Flow was held constant at ~5 L/min of 11 to 13 °C wellwater added directly below the water surface and held ata depth of ~60 cm by a stand pipe at the outflow.Preliminary trials suggested that fish would not navigatethe passage tube if structure was present. Therefore, topromote movement the upstream chamber of the arenawas void of structure similar to the conventional rearingenvironment and the downstream chamber consisted ofa river rock base similar to the complex rearing envi-ronment (Fig. 1a).

A video recording system and PIT tag detectionsystem were installed at each arena to monitor fishmovement and behavior. The video system consistedof infrared cameras mounted above each arena andconnected to a digital video recorder and monitor. Theinfrared technology has limited capability penetratingwater so while the infrared cameras were useful formonitoring daytime and nighttime behavior, the night-time video is of lower clarity and quality. The PITdetection system was installed to provide the requiredmonitoring during the nighttime hours. The PIT systemconsisted of detection antennae that were installed undereach arena and directly below the baffle (Fig. 1a and b).Each antenna was constructed and tuned to only readPIT tags that were present inside the passage tube. Thefour antennae were connected to a Biomark FS1001Mmultiplexing receiver and computer to log detectioninformation (i.e., tag code, time of day).

The boldness trials took place during two time pe-riods. In September and October, 2013, conventionaland complex treatment fish were tested and in

386 Environ Biol Fish (2018) 101:383–401

May 2014 wild fish were tested. Wild fish were allowedtwo days after collection to recovery from capture andtransport to the Fish Performance and Genetics Labora-tory. Seven trials were conducted with conventional andcomplex captive reared fish and 6 trials were conductedwith wild reared fish for a total of 20 trials. Five bulltrout were netted from the rearing trough and placed inthe conventional habitat chamber of each arena. Eachtrial began with four to six bull trout. Two fish from thecomplex rearing treatment jumped out of the tank duringtrials and were removed from the analysis. All fish werestarted in the conventional treatment chamber only dueto a limited number of individuals. Our assumption wasthat barren habitat was a less desirable and would stim-ulate movement and activity regardless of treatment.Each fish was only used once in the boldness trials tolimit any potential bias associated with arena experi-ence. The average size of wild fish was larger and morevariable than captive-reared fish (Table 1).

Fish were allowed a 1 h recovery period in the arenaafter which time access to the passage tube was allowedand the trial began. All trials started at mid-day andcontinued for 24 h. At the end of a trial any fish thatpassed from the conventional rearing chamber to the

complex rearing chamber was considered bold and eachfish was enumerated, scanned for PIT tag identification,measured for length and weight and returned to therearing trough. Overhead video and PIT detection datafor each trial was archived and reviewed to determinethe time of day and minutes after the start of a trial that abold fish navigated the passage tube. Fish were able topass back and forth between the chambers.

Prey acquisition trials

To test how rearing environment affects the propensityof bull trout to feed on live prey we reconfigured thefour behavioral arenas used in the boldness trials. Allbaffles and river rock were removed from the tanks withthe exception of small pile of rocks in one corner of thearena (Fig. 1c). The small pile of rocks provided thepredator (i.e., Bull Trout) with cover to minimize stressand produce a more natural foraging behavior. Flowwasheld constant at ~5 L/min of 11 to 13 °C well wateradded directly below the water surface and held at adepth of ~60 cm by a stand pipe at the outflow.

The predator acquisition trials took place betweenApril 2014 and May 2014 for all three rearing

Fig. 1 Diagram depicting the arena (0.5 m long × 0.33 m wide ×0.33 m deep) used for behavioral trials. An overhead view (a) andview of the baffle from inside a behavioral arena (b) for boldness

trials and an overhead view (c) of the arena for the prey acquisitiontrials. Substrate is indicated by checkered patterns

Environ Biol Fish (2018) 101:383–401 387

treatments. Prior to a trial, four bull trout (i.e., thepredator; two from each treatment) were netted fromtheir rearing trough, anesthetized with 100 mg/L ofbuffered MS-222, PIT tag scanned, and length andweight taken. Bull trout were then held in a smallertrough (2 m long × 0.5 m wide × 0.3 m deep), withsimilar treatment, near the behavioral arenas for 48 hwithout feed. Twenty trials were conducted with con-ventional and complex captive reared fish and 11 trialswere conducted with wild reared fish for a total of 51trials. On the morning of a trial, spring Chinook salmon(i.e., the prey) were removed from their circular rearingtanks, anesthetized with 100 mg/L of buffered MS-222,individually measured for length, and 10 fish pooledtogether for weight. Ten spring Chinook were thenplaced in each of the four behavioral arenas and allowedto recover from anesthesia and acclimate for one hour.After one hour the trial began when a bull trout wasadded to each of the arenas. This study was conductedwith a factorial design with two spring Chinook rearingtreatments (i.e., complex and conventional) and three

bull trout rearing treatments. All trials began in themorning and were ended after 24 h the following morn-ing. At the end of a trial all unpredated spring Chinooksalmon were removed, enumerated, and euthanized. Allbull trout were returned to their respective rearingtroughs. Bull trout were only used once in these trials.The average size of wild bull trout was smaller and lessvariable than those of the captive rearing treatments(Table 1).

Brain and eye development

Brain dimensions for complex and conventional rearedbull trout were assessed in November 2013 and wildreared bull trout were assessed in June 2014 after allbehavioral trials were completed. Twelve bull trout fromeach rearing treatment were euthanized, PIT tagscanned, measured for length and weight, placed in anindividual plastic bag, and transported on ice to theAdvanced Imaging Research Center at Oregon Healthand Science University (OHSU) for magnetic resonance

Table 1 Length and weight of bull trout and Chinook salmon sampled in boldness and prey acquisition trials, and brain and eye lensdevelopment comparisons

Rearing habitat Length (mm) Weight (g)

N Max Min Mean SD N Max Min Mean SD

Boldness

Bull Trout

Conventional 36 117 72 87.8 10.4 36 14.4 3.8 7.5 2.6

Complex 33 136 79 103.1 15.3 33 27.3 5.4 12.6 5.7

Wild 29 162 80 110.6 26.3 29 48.0 4.5 17.0 12.7

Prey acquisition

Bull Trout

Conventional 20 205 154 183.6 14.0 20 99 39 70.0 15.6

Complex 20 215 138 184.2 21.5 20 119 30 73.3 25.1

Wild 11 157 120 138.2 11.3 11 36 15 23.6 6.0

Chinook salmon

Conventional 200 60 48 52.8 2.5 200 18.1 11.0 15.9 2.1

Complex 200 60 48 52.7 2.6 200 19.1 11.3 16.0 2.0

Wild 110 55 40 49.0 3.5 110 13.8 5.2 8.9 3.0

Brain and eye lens development

Bull Trout

Conventional 12 133 85 108.3 15.2 12 26.1 6.2 13.7 6.3

Complex 12 136 87 109.8 13.9 12 27.8 7.0 15.0 6.0

Wild 8 112 85 94.1 9.9 8 15.5 5.5 8.8 3.5

388 Environ Biol Fish (2018) 101:383–401

imaging (MRI). Images were acquired on a Bruker-Biospin 11.75 T small animal MR system with aParavision 5.1 software platform, 9-cm inner diametergradient set (750 mT/m), and a 20 mm I.D. quadratureradiofrequency transceiver coil (m2 m Imaging Corp).The fish were left in the plastic bags and positionedright-side up for scanning. A coronal 40 slice T2-weight-ed image set was obtained from the brain region,employing a Paravision RARE sequence (TR6200 ms, TEeff 32 ms, RARE factor 8, FOV 2.5 cm,256 × 256 matrix, 0.35 mm slice width, 98 × 98 μm in-plane resolution, two averages, acquisition time 6 m36 s). Images were processed with Osyrix software(Version 6.0.2; Rosset et al. 2004) by delineating fiveregions of interest (ROIs); the whole brain, telencepha-lon, optic tectum, cerebellum, and both eye lenses. Wefollowed similar methods as Kihslinger and Nevitt(2006) to determine the limits of the whole brain andbrain regions. Delineation of the whole brain started onthe image where telencephalon was first observed andcontinued through the final image containing cerebel-lum. Published fish brain atlases (Ullmann et al. 2010;Simões et al. 2012) and rainbow trout brain atlas (Davisand Northcutt 1983) were used as examples for delin-eation of the telencephalon, optic tectum, andcerebellum.

We determined the width of the optic tectum bytaking six linear measurements of optic tectum widthfrom an image at the approximate middle of the optictectum and averaging the measurements. The lens is acrystalline component of the eye (Wall 1963; Hargis1991) and left and right eye lenses were easily delineat-ed as the dark sphere in the MRI images (Fig. 2). Weused the BROI volume^ feature in Osyrix to determinethe volume of each region of interest (ROI). Imagestacks for all fish were randomized to minimize poten-tial bias with delineating ROIs.

Data analysis

We used logistic regression to examine the relationshipbetween boldness and rearing environment, and preyacquisition and rearing environment (Hosmer andLemeshow 2000). Before fitting logistic regressionmodels, we fit random effects ANOVAs to account fordependence among observations within individual be-havioral arenas or trials (Conroy and Peterson 2013). Ifthe variance was greater than 0.00 then a random effectfor arena and/or trial was included in a mixed effect

model to account for dependence, otherwise all modelsincluded only fixed effects. Models were determined apriori to test biologically relative hypotheses involvingrearing environment and fish size, as well as, preyrearing environment and average size for the prey ac-quisition trials.

The initial analysis suggested boldness was indepen-dent of tank or trial; as such the boldness analysisconsisted of five candidate models including boldnessas a binary response variable and a combination ofrearing habitat and length or weight as fixed explanatoryvariables. The initial analysis of the prey acquisitiondata suggested a random effect of trial. Therefore weexamined prey acquisition by modeling proportion ofprey eaten as a function of fixed effects for predatorrearing habitat, length, and weight, and prey rearinghabitat and average length, with a random effect for trial.

The fit of each candidate model was assessed usingAkaike’s Information Criteria corrected for smallsample sizes (AICc) and the model with the lowestAICc was considered the best-fitting model. The rela-tive plausibility of each model was assessed withAkaike weights (wi), with the most plausible candi-date model having the highest wi (Burnham andAnderson 2002). Odds ratios were calculated for thethree best fitting models for ease of interpretation andthe 95% confidence interval of the odds ratio wascalculated to assess precision of the explanatory var-iable. Confidence intervals that include 1.00 wereconsidered inconclusive of a positive or negative re-lationship between the parameter and boldness or preyacquisition (Thompson and Lee 2000; Weigel et al.2003). We assessed goodness-of-fit for the three bestfitting models by examining the deviance chi-squarestatistic for fixed effect models and a Hosmer-Lemeshow chi-square statistic for mixed effectmodels. The goodness-of-fit for the prey acquisitionmodels suggested a poor fit due to overdispersion. Toaccount for overdispersion an additional random ef-fect for each observation was added to the prey acqui-sition models (Williams 1982).

Growth of the brain and eye are more closely relatedto age than length (Pankhurst and Montgomery 1994).The dataset consisted of equal samples of fish per rear-ing treatment (N = 12) however subsequent aging ofwild fish otoliths identified four Age-2 individuals.Therefore, the brain analysis consists of four rearinggroups: conventional, complex, Age 1 wild, and Age 2wild. All captive reared fish were known to be Age 1.

Environ Biol Fish (2018) 101:383–401 389

Fig. 2 Example of a stack of 40 magnetic resonance imaging(MRI) slices taken from the transverse plain of a bull trout. Slide 1depicts the most caudal section and slide 40 depicts the mostrostral section. Regions of interests used to calculate volumes of

the cerebellum (slide 5 to 13), optic tectum (slide 14 to 23),telencephalon (slide 24 to 29), and right and left eye lens (slide24 to 30) are outlined in white. Slide 19 shows the six linearmeasurements collected to determine average optic tectum width

390 Environ Biol Fish (2018) 101:383–401

The body, brain, and eye of a fish exhibit allometricgrowth and will continue to grow throughout life(Zupanc 2006; Devlin et al. 2012). We accounted forallometric growth of the eye lens, telencephalon, optictectum, and cerebellum by calculating the ratio of thebrain region volume to whole brain volume and used aone-way analysis of variance (ANOVA) to test for dif-ferences as a function of rearing treatment. The optictectum is a layered structure that receives informationfrom all regions of the brain. The superficial layers areused in processing sensory information and deep layerscontrol motor function (Northmore 2011). We calculat-ed the ratio of the optic tectum width to whole brain,optic tectum, cerebellum, and telencephalon volume tobetter understand how these brain regions develop inrelation to the thickness of the optic tectum. One-wayanalysis of variance (ANOVA) was used to test fordifferences in the ratio of a brain region as a functionof rearing treatment. When ANOVA identified signifi-cant differences between rearing treatments we usedTukey’s honest significant difference (HSD) test to iden-tify pair-wise differences. All statistical conclusionswere conducted at the α = 0.05 significance level andwe used residual and QQ plots to assess the assumptionsof ANOVA.

Results

Embryo collections

A total of 1068 bull trout embryos were delivered to therearing facilities at OSU. Survival from embryo collec-tion in November 2012 to PIT tagging in August 2013was 0.48 and 0.18 for conventionally and complexreared fish, respectively. As fish absorbed the yolk sacand began exogenous feeding a higher level of canni-balism was observed in the complex reared treatmentthan the conventional rearing treatment.

Boldness trials

We conducted the initial boldness analysis with all fish.However, after otolith aging associated with the brain andeye development study determined that wild fish greaterthan 120 mm were Age-2, we excluded these older fishfrom the boldness analysis and compared the results withonly Age-1 fish. The results from both analyses weresimilar and we opted to include all fish in the analysis

presented here. Of the 98 fish in the boldness trials, 21fish expressed boldness (i.e., 4/36 or 11% of convention-al; 12/33 or 36% of complex; 5/29 or 17% of wild) andonly two of those fish, both wild, traveled back and forthin passage tube multiple times.

Odds ratios from the most plausible model suggestedthat bull trout reared in complex environments are 4.6times (1.4 to 18.1 95% CI) more likely and wild fish are1.7 times (0.4 to 7.37 95% CI) more likely to exhibitboldness behavior than conventional reared fish. Rear-ing habitat was included as a significant predictor ofboldness in the top three ranked models with weight andlength included in the second and third rank models,respectively (Table 2). Confidence intervals for the oddsratios of length and weight include 1.00 and are consid-ered inconclusive of a positive or negative effect. Wildfish appeared to exhibit boldness more than convention-al reared fish in all models. However, the confidenceinterval of the odds ratio includes 1.00 and is consideredinconclusive of a positive or negative effect. The onlytwo instance of fish traveling back and forth in thepassage tube occurred with wild fish; suggesting anadditional level of boldness that could not be accountedfor with our analysis. The deviance chi-square test sug-gested adequate model fit for the top ranked models(Table 3). Four conventionally reared fish and 12 com-plex reared fish navigated the passage tube. These cap-tive reared fish navigate the passage tube throughout thetrials and throughout the day. Of the five wild fish thatexhibited boldness, four navigated the passage tube inthe early morning, near the end of the trial (Fig. 3).

Prey acquisition trials

All wild fish in the prey acquisition trials were greaterthan 120 mm and therefore Age-2 or older. As such weconducted the analysis by including all wild fish. Thebest fitting model suggested that bull trout reared incomplex environments are 4.8 times (1.4 to 18.1 95%CI) more likely and wild fish are 70.6 times (12.5 to400) more likely to acquire live prey than fish rearedin conventional captive environments. Similar to theboldness trials, rearing habitat was included as a sig-nificant predictor of prey acquisition in the top threeranked models (Table 4). Length was included in thebest fitting model and weight was included in thesecond best fitting model as predictors for prey acqui-sition, however the confidence intervals include 1.00and are considered inconclusive of an effect. Prey

Environ Biol Fish (2018) 101:383–401 391

rearing habitat and average length were not includedin any of the top ranked models. The HosmerLemeshow chi-square test suggested adequate modelfit for all top ranked models (Table 3).

Brain and eye development

Analysis of variance identified a difference in the ratioof cerebellum volume to brain volume (ANOVA:

Table 2 Results of top ranked models describing boldness of bull trout reared in conventional and complex captive environments and wildenvironments. Akaike weights (w) of each model are reported

Model parameter Parameter estimate Standard error P-value

Odds ratio

Estimate Lower 95% CI Upper 95% CI

Best-fitting model (w = 0.327)

Intercept −2.08 0.53 0.000 – – –

Rearing habitat (complex) 1.52 0.64 0.018 4.57 1.39 18.13

Rearing habitat (wild) 0.51 0.72 0.480 1.67 0.40 7.37

Second-best-fitting model (w = 0.320)

Intercept −1.67 0.61 0.006 – – –

Rearing habitat (complex) 1.80 0.68 0.008 6.03 1.71 25.38

Rearing habitat (wild) 0.90 0.76 0.235 2.46 0.55 11.57

Weight (g) −0.06 0.04 0.182 0.95 0.86 1.02

Third-best-fitting model (w = 0.271)

Intercept −0.31 1.46 0.834 – – –

Rearing habitat (complex) 1.83 0.69 0.008 6.26 1.73 27.15

Rearing habitat (wild) 0.90 0.77 0.242 2.47 0.54 11.90

Length (mm) −0.02 0.02 0.198 0.98 0.95 1.01

Table 3 Model selection statistics for candidate models describing the boldness and prey acquisition behavior of captive reared and wildbull trout. Goodness-of-fit statistics for the top three ranked models are reported

Candidate Model K −2 ln L AICc ΔAICc wi x2 Df P

Boldness trials

Rearing habitat 3 95.04 101.3 0.00 0.327 95.0 95 0.48

Rearing habitat, Weight 4 92.91 101.3 0.04 0.320 92.9 94 0.51

Rearing habitat, Length 4 93.24 101.7 0.37 0.271 93.2 94 0.50

Weight 2 101.15 105.3 3.98 0.045 – – –

Length 2 101.54 105.7 4.38 0.037 – – –

Prey acquisition trials

Predator rearing habitat, Predator length 5 107.61 121.5 0.00 0.510 0.45 45 0.99

Predator rearing habitat, Predator weight 5 108.45 122.4 0.84 0.335 0.69 45 0.98

Predator rearing habitat 4 113.12 124.5 2.94 0.117 0.47 46 0.99

Predator rearing habitat, Prey rearing habitat 5 112.87 126.8 5.27 0.037 – – –

Average prey length 3 126.96 135.8 14.32 0.000 – – –

Prey rearing habitat 3 131.62 140.5 18.98 0.000 – – –

Predator length 3 131.84 140.7 19.20 0.000 – – –

Predator weight 3 132.10 141.0 19.46 0.000 – – –

392 Environ Biol Fish (2018) 101:383–401

F3,32 = 3.03, P = 0.044). Contrasts identified differencesbetween the conventional rearing group and Age-1 wildfish where complex reared and Age-2 wild fish exhibit-ed similar relative cerebellum volume to conventionaland Age-2 wild fish. Age-1 wild fish had a smallerrelative cerebellum than the conventional treatmentgroup and the conventional rearing group exhibited thelargest relative cerebellum. There was no effect of rear-ing group on the ratio of telencephalon volume to brainvolume (ANOVA: F3,32 = 0 .58, P = 0.64) or the ratio ofoptic tectum volume to brain volume (ANOVA: F3,32 =0.68, P = 0.57). After accounting for optic tectum vol-ume, Age-2 wild fish had the largest relative eye lensvolume, followed by conventionally reared fish. Age-1wild and complex reared fish had similar average eyelens volumes that were smaller than conventionallyreared fish and Age-2 wild fish (ANOVA: F3,32 =16.04, P < 0.0001; Fig. 4).

Age-1 wild fish exhibited larger relative optic tectumwidths than the other three rearing treatments after ac-counting for whole brain (ANOVA: F3,32 = 8.75,P < 0.001) and cerebellum volumes (ANOVA: F3,32 =8.58, P < 0.001). Optic tectum (ANOVA: F3,32 = 7.10,P < 0.001)and telencephalon volumes (ANOVA:F3,32 = 5.68, P = 0.003) were also different between

rearing with Age-1 wild fish having the largest relativeoptic tectum width on average yet not statistically dif-ferent than all the other rearing treatments (Fig. 5).

Discussion

We found that adding a simple form of structural com-plexity to captive rearing environments can result in bulltrout that are more similar to wild fish by exhibiting agreater propensity to take risks, explore novel habitat(i.e., boldness), and acquire live prey than convention-ally reared fish. In addition, we found that the captiverearing habitat results in differences in the relative de-velopment of various brain regions and the eye lens, andthat researchers should account for age when makingsuch comparisons. Producing fish in captivity that ex-hibit behaviors that are thought to be essential for sur-vival in the wild and are developmentally similar to wildfish should improve the success of reintroduction pro-grams (Brown and Day 2002).

The expression of boldness is a heritable trait (Brownet al. 2007) that can differ depending on the origin of theanimal (Sundström et al. 2004). Repeated spawning ofhatchery stocks may result in animals that are less fitthan those of wild origin and this can occur in as few asone or two generations (Araki et al. 2008). The captivebull trout in this study were collected as wild embryosfrom the same family groups, and from a populationwith zero history of captive influence suggesting theresults were a function of the early rearing environmentand not domestication. An early rearing environmentwith little to no complexity does not provide sufficientopportunity to learn essential life skills. Whereas, struc-turally complex habitats have a gradient of resourcesthat are available in space and provide experience indefense of territories (Metcalfe et al. 2003), navigating,and locating cover (Näslund and Johnsson 2016). Weobserved fish using the interstitial spaces of the riverrock from the time of hatching until behavioral trialsbegan over a year later suggesting these behaviors startdeveloping immediately and are lasting (Jonsson andJonsson 2014).

Boldness is variable and differentially expressed de-pending on the situation and can change throughout ananimal’s lifetime (Sinn et al. 2008). For any behavior tobe advantageous it should be expressed in situations thatresult in improved growth or survival of the individualand population. Bold individuals may improve growth

Fig. 3 Times of day (a) and hours after the start of a trial (b) thatbull trout navigate the passage tube and exhibited boldness. Pointsare jittered along the x-axis for presentation purposes

Environ Biol Fish (2018) 101:383–401 393

and survival by outcompeting shy individuals for foragelocations, food resources, and mates. However, theserewards are associated with increased metabolic costs(Metcalfe et al. 1995) and predation risk (Lima and Dill1990). In most captive rearing environments salmonidsreared with a constant and continual food source andlittle to no predation results in fish that are less apt toselect favorable foraging locations (Brown et al. 2003)and avoid predators (Olla and Davis 1989). Bull troutare different than other salmonids in that they becomepiscivorous and cannibalistic at a young age and there-fore are constantly living with risk of predation while incaptivity. We observed bull trout fry from both rearing

treatments (~5 cm in length) prey upon similarly sizedfish from the same family group. It is possible that byadding structural complexity to captive environmentswe created foraging and cover habitats worth defendingwhich resulted in increased agonistic interactions andcannibalism in the complex rearing treatment. Theseexperiences may have resulted in learned boldness be-havior or selection towards bold individuals.

Salmonids and other hatchery reared species can read-ily shift from commercial feeds to live prey while incaptivity (Olla et al. 1998; Sundström and Johnsson2001). This is similar to our findings in that fish fromboth captive rearing groups foraged on live spring

Table 4 Results of the top ranked models from logistic regression models describing prey acquisition behavior of bull trout reared inconventional and complex captive environments and wild environments. Akaike weights (w) of each model are reported

Modelparameter

Parameterestimate

Standarderror

P-value

Odds ratio

Estimate Lower 95%CI Upper 95%CI

Best-fitting model (w = 0.510)

Fixed effects

Intercept −9.64 2.54 0.000 – – –

Rearing habitat(complex)

1.56 0.59 0.008 4.77 1.51 15.05

Rearing habitat (wild) 4.26 0.88 0.000 70.61 12.47 400.00

Length (mm) 0.03 0.01 0.020 1.03 −0.95 3.02

Random effects

Trial 0.00 0.00 – – – –

Overdispersion 0.01 0.41 – – – –

Second-best-fitting model (w = 0.335)

Fixed effects

Intercept −5.84 1.05 0.000

Rearing habitat(complex)

1.50 0.59 0.011 4.48 1.42 14.15

Rearing habitat (wild) 4.06 0.87 0.000 57.98 10.54 318.89

Weight (g) 0.00 0.00 0.031 1.00 −0.96 2.96

Random effects

Trial 0.01 0.29 – – – –

Overdispersion 0.00 0.35 – – – –

Third-best-fitting model (w = 0.117)

Fixed effects

Intercept −4.13 0.59 0.000

Rearing habitat(complex)

1.67 0.59 0.004 5.30 1.68 16.71

Rearing habitat (wild) 2.91 0.69 0.000 18.29 4.75 70.38

Random effects

Trial 0.35 0.41 – – – –

Overdispersion 0.01 0.40 – – – –

394 Environ Biol Fish (2018) 101:383–401

Chinook in the prey acquisition trials. However, thesesame animals will need to locate and capture live preywhen released into the wild and natural forage bases arerarely as abundant and consistently available as feed incaptivity. Upon release into the wild, hatchery fish shift tolive prey later and forage less than wild fish, resulting inreduced growth and survival (Olla et al. 1998). If ourlaboratory results translate to the wild, then bull troutreared in complex habitats may be better suited to forageon live prey after release, improving growth and survival,and ultimately the probability of a successful reintroduc-tion program.

While bull trout reared in complex environmentswere successful in capturing live prey, they did not

perform as well as wild fish. This is most likely afunction of wild fish having past experiences capturingprey (Brown et al. 2003), exhibiting hyperphagia(Armstrong and Bond 2013) or the added complexityof wild habitats. Providing captive reared bull trout withlive prey may improve their capture and forage abilityand lower the instances of cannibalism by appealing totheir natural predatory instincts. Whether fish are pro-vided commercial diets or live prey the offering shouldbe presented at uncertain time frames and locations tosimulate a more wild-like forage regime. Variability infood resources results in hyperphagia, a foraging strate-gy where an animal consumes prey while in abundanceand survives off fat reserves during times of low

Fig. 4 Plots of the eye lensvolume relative to optic tectumvolume and telencephalon, optictectum, and cerebellum volumerelative to whole brain volume ofbull trout reared in complex andconventional captiveenvironments and Age-1 andAge-2 wild fish. The boxrepresents the mean and the errorbars represent the 95%confidence interval. Differentlower case letters representstatistically significant differences(α = 0.05) between points inpairwise comparisons

Environ Biol Fish (2018) 101:383–401 395

abundance (Armstrong and Bond 2013). This maybe abeneficial forage strategy for bull trout, which pri-marily feed on juvenile salmon and steelhead, andwill need to survive through periods when little or noforage is available. If using a variable feeding strat-egy when captive rearing fish, it will be important toprovide suitable amount of nutrition for growth andmetabolism while still simulating the uncertainty offeed availability in the wild. It is possible that thecaptive reared fish did not perform as well as wildfish because wild fish were a year older or the com-plex habitats were over-simplified and lacked vari-able flows, overhead structure, vegetation, and othercharacteristics that are present in the wild.

Given the growing number of publications that doc-ument decreased developmental trajectories in brains offish reared in barren captive environments (Marchettiand Nevitt 2003; Kihslinger et al. 2006; Burns et al.2009) we expected similar results from our study. Whileunable to detect any differences in the morphologicaldevelopment of the telencephalon and optic tectum aftercorrecting for optic tectum width, our results do suggestthat the cerebellum of Age-two wild fish and complexreared fish develop similarly to that of conventional fishand Age-2 wild fish. Whereas, the cerebellum of Age-1wild fish is significantly smaller than that of conven-tional reared fish. This result contrasts our expectationthat both wild fish groups would have larger relative

Fig. 5 Plots of the optic tectumwidth relative to the whole brain,optic tectum, cerebellum, andtelencephalon volume of bulltrout reared in complex andconventional captiveenvironments and Age-1 andAge-2 wild fish. The boxrepresents the mean and the errorbars represent the 95%confidence interval Differentlower case letters representstatistically significant differences(α = 0.05) between points inpairwise comparisons

396 Environ Biol Fish (2018) 101:383–401

cerebellum volume than fish reared in captivity.Kihslinger and Nevitt (2006) compared cerebellumgrowth of swim-up steelheadO. mykiss reared in similarcaptive treatments to our study and compared them tosize matched wild fish. They found that wild and com-plex reared fry had similar relative cerebellum sizes andboth were larger than conventionally reared fry andhypothesized that these differences may be due to theeffects of increased rearing density in captivity, warmerand more variable water temperatures experienced byin-river fish, or increased cell proliferation caused bycomplex environmental stimuli. All captive fish in thisstudy were reared in low densities relative to typicalproduction facilities and if the water temperature hy-pothesis was accurate thenwewould expect to see largercerebellum in both the captive reared treatments giventhat they were reared at temperatures between 10 to13 °C as comparted to the wild fish that experienced 6to 8 °C. Our results lend little support to the cell prolif-eration hypothesis because we would expect Age-1 wildfish to have larger cerebellum then the captive rearinggroups, not smaller, as our results suggest. In conjunc-tion with cerebellum size differences, Kihslinger andNevitt (2006) showed that fish with a smaller relativecerebellum were more likely to actively move. The wildfish were collected using screw traps which are designedto capture actively migrating fishes and the results maybe an artifact of sampling bias and life history. We mayhave only captured Age-1 wild fish that were activelymoving as a function of a smaller cerebellum, whereasthe Age-2 fish we captured exhibited a larger cerebellumand spent an extra year in the headwaters before migrat-ing as a function of life history strategy and not braindevelopment. Although the gross morphological differ-ences we evaluated are not always present there is thepotential that the development of complex neuronalconnections is effected by rearing environment and willinfluence behavioral and survival (Ebbesson andBraithwaite 2012).

Similar to Kihslinger and Nevitt’s (2006) workwith steelhead alevin, Näslund et al. (2012) docu-mented larger brains and brain regions in Atlanticsalmon alevin hatched and reared in complexenvironments. However, any differences in brainsize were undetectable only a month after pondingall fish in barren tanks. This suggests that the rearingenvironment can rapidly influence the developmentaltrajectory of the brain and that added complexityshould be maintained until release. In our study, a

little over two weeks passed between capturing wildbull trout, running the behavioral trials, andconducting brain MRIs. It is difficult to know if thisshort time in a complex captive environment wasenough to institute a change in brain size and if soto what degree. If brain volume was altered it mightexplain the lack of differences in the telencephalon,and optic tectum, as well as, the cerebellum results.In addition, Näslund et al. (2012) released a group offish into local rivers five months after sampling fry.After six months of wild rearing, some of thoseindividuals were recaptured in a screw trap and theirdorsal brain area was compared to that of fish held inthe hatchery for the same time period. They foundthat wild fish had larger bodies and smaller total brainareas than hatchery reared fish suggesting that energywas allocated to somatic rather than neural growth.When living with a predictable food supply (e.g.,captivity) it may be advantageous to more evenlyallocate energy to neural and somatic growth. Weoffer a similar hypothesis to explain why the relativeoptic tectum width was larger in Age-1 wild fish. Theoptic tectum is a layered structure that receives inputfrom the retina and many other brain regions, includ-ing the telencephalon and hindbrain (Northmore2011). Allocating energy to this brain region earlyin development may be more important for a youngwild fish and once the layers of the optic tectum aredeveloped energy allocation can be shifted to devel-oping other brain regions. Given that food availabil-ity is more certain for captive reared fish they allocateenergy for development more evenly across the brainregions.

We found that after accounting for the volume of theoptic tectum, Age-1wild reared and complex reared fishwere more likely to have smaller eye lens volumes thanconventional reared fish and all groups were more likelyto be smaller than Age-2 wild fish. These findingssuggest that eye development may be a function of thecomplexity or color of rearing environments and chang-es with age. Fish that rear in murky or benthic environ-ments exhibit smaller eye diameter than fish rearing inclearer or pelagic environments (McPhail 1984;Baumgartner et al. 1988; Pakkasmaa et al. 1998;Tiffan and Connor 2011). In an extreme case, the plas-ticity of eye size is evident in cave fishes Astyanax spp.where surface oriented fish that experience some lightwill develop a functional eye whereas the eye willalmost completely degenerate in fish that reside deeper

Environ Biol Fish (2018) 101:383–401 397

in a cave (Jeffery 2001) and the optic tectum of cavedwelling fish is reduced in volume as compared to thesurface oriented fish (Soares et al. 2004). All the fish inthis study were reared in clear well water or in naturalspring feed streams. Therefore, the difference we ob-served in eye development may be a result of life inblue-green fiberglass tanks without structure on whichto focus. The river rock in the complex habitats provideda darker benthic environment on which to focus whilemaintaining the blue-green walls that provided an evenbrighter rearing environment than the wild fish experi-enced. The differences we detected in eye developmentof Age-1 and Age-2 wild fish suggests that as fishtransition from Age-1 to Age-2 they shift energy allo-cation from optic tectum development to optic tectummaintenance and eye lens development. The teleost eyeis a complex organ and it is difficult to determine exactlywhat effect different eye lens volumes, pupil sizes, ortotal eye diameters will have on the visual acuity of afish. The lens is used to refract light that enters the pupiland focuses an image by moving towards and awayfrom the retina where photoreceptor cells, containinglight-sensitive pigments, collect light for vision process-ing in the optic tectum (Bowmaker 2011; Kröger 2011;Northmore 2011). It is clear that future work focused oneye development and how it may influence a fish’svisual acuity and the resulting behavior should collec-tively consider all components of the fish eye.

Understanding the correlation between cognition,behavior and brain development in fish has advancedsignificantly in the past few years (Ebbesson andBraithwaite 2012; Fernald 2015). The growing bodyof literature suggests that rearing environment plays animportant role in behavior and development and provid-ing a level of complexity to captive habitats may resultin animals that are better fit to accomplish managementgoals. The ability to rear fish in captivity is an importanttool that can be used to accomplish a suite of manage-ment objectives including providing put-and-take fish-ery opportunities, upholding mitigation responsibilities,and conservation oriented objectives like providing fishfor research and reintroduction programs, or in worstcase scenarios maintaining refugia populations. Rearingsmall groups of wild-like fish for research projects islogistically more feasible than rearing larger groups offish for reintroduction and recovery programs. This is afunction of available donor stocks, and animal husband-ry considerations including rearing space and density,amount and type of feed (i.e., commercial, natural, or

both), and maintaining a clean captive environment.Future research investigating how best to scale-up nat-ural rearing approaches while maintaining wild-like be-haviors and development would be beneficial to recov-ery programs. In 1999, the bull trout were listed underthe Endangered Species Act as threatened in the UnitedStates and various captive rearing strategies have beenconsidered to provide individuals for reintroductionprograms (MBTSG 1996; Shively et al. 2007; USFWS2015). Ultimately, producing a behaviorally and devel-opmentally wild-like fish in captivity should improvethe post-release survival of individuals and improve thesuccess rate of reintroduction programs aimed at recov-ering threatened and endangered species.

Acknowledgments The authors declare no conflict of interest.We would like to thank Dr. Jeffrey Jolley, Greg Silver, Dr. KariDammerman, Dr. Robert Mason, Dr. Jason Dunham, and Dr.Jacob Raber for thoughtful reviews of the manuscript. Dr. MattMesa provided productive discussion on study design. RobChitwood, Olivia Hakanson, and Rachel Palmer provided animalhusbandry. References to trade names do not imply endorsementby the U.S. Government. The findings and conclusions in thismanuscript are those of the author and do not necessarily representthe views of the U.S. Fish and Wildlife Service.

References

Araki H, Berejikian BA, Ford MJ, Blouin MS (2008) Fitness ofhatchery-reared salmonids in the wild. Evol Appl 1(2):342–355. https://doi.org/10.1111/j.1752-4571.2008.00026.x

Armstrong JB, Bond MH (2013) Phenotype flexibility in wildfish: Dolly Varden regulate assimilative capacity to capitalizeon annual pulsed subsidies. J Anim Ecol 82(5):966–975.https://doi.org/10.1111/1365-2656.12066

Baumgartner JV, Bell MA, Weinberg PH (1988) Body form dif-ferences between the Enos Lake species pair of threespinesticklebacks (Gasterosteus aculeatus complex). Can J Zool66(2):467–474. https://doi.org/10.1139/z88-066

Berejikian BA, Tezak EP, Flagg TA, LaRae AL, Kummerow E,Mahnken CV (2000) Social dominance, growth, and habitatuse of age-0 steelhead (Oncorhynchus mykiss) grown inenriched and conventional hatchery rearing environments.Can J Fish Aquat Sci 57(3):628–636. https://doi.org/10.1139/f99-288

Berejikian BA, Tezak EP, Riley SC, LaRae AL (2001)Competitive ability and social behaviour of juvenile steel-head reared in enriched and conventional hatchery tanks anda stream environment. J Fish Biol 59(6):1600–1613.https://doi.org/10.1111/j.1095-8649.2001.tb00224.x

Berejikian BA, Gable J, Vidergar D (2011) Effectiveness andtrade-offs associated with hydraulic egg collections fromnatural salmon and steelhead redds for conservation hatchery

398 Environ Biol Fish (2018) 101:383–401

programs. Trans Am Fish Soc 140(3):549–556. https://doi.org/10.1080/00028487.2011.583540

Bowmaker JK (2011) Photoreceptors and visual pigments. In:Farrell AP, Stevens ED, Cech JJ, Richards JG (eds)Encyclopedia of fish physiology: from genome to environ-ment. Academic Press, Maryland Heights, pp 110–130.https://doi.org/10.1016/B978-0-12-374553-8.00091-5

Broglio C, Rodríguez F, Salas C (2003) Spatial cognition and itsneural basis in teleost fishes. Fish Fish 4(3):247–255.https://doi.org/10.1046/j.1467-2979.2003.00128.x

Broglio C, Gómez A, Durán E, Salas C, Rodríguez F (2011) Brainand cognition in teleost fish. In: Brown C, LalandK,Krause J(eds) Fish cognition and behavior. Wiley-Blackwell, NewYork, pp 325–358. https://doi.org/10.1002/9781444342536.ch15

Brown C, Braithwaite VA (2004) Size matters: a test of boldness ineight populations of the poeciliid Brachyraphis episcopi.Anim Behav 68(6):1325–1329. https://doi.org/10.1016/j.anbehav.2004.04.004

Brown C, Day RL (2002) The future of stock enhancements:lessons for hatchery practice from conservation biology.Fish Fish 3(2):79–94. https://doi.org/10.1046/j.1467-2979.2002.00077.x

Brown C, Davidson T, Laland K (2003) Environmental enrich-ment and prior experience of live prey improve foragingbehaviour in hatchery-reared Atlantic salmon. J Fish Biol63(s1):187–196. ht tps: / /doi .org/10.1111/ j .1095-8649.2003.00208.x

Brown C, Burgess F, Braithwaite VA (2007) Heritable and expe-riential effects on boldness in a tropical poeciliid. Behav EcolSociobiol 62(2):237–243. https://doi.org/10.1007/s00265-007-0458-3

Brown AD, Sisneros JA, Jurasin T, Nguyen C, Coffin AB (2013)Differences in lateral line morphology between hatchery-andwild-origin steelhead. PLoS One 8(3):e59162. https://doi.org/10.1371/journal.pone.0059162

Burnham KP, Anderson DR (2002) Model selection andmultimodel inference: a practical information-theoretic ap-proach. Springer-Verlag, New York

Burns JG, Rodd FH (2008) Hastiness, brain size and predationregime affect the performance of wild guppies in a spatialmemory task. Anim Behav 76(3):911–922. https://doi.org/10.1016/j.anbehav.2008.02.017

Burns G, Saravanan A, Rodd FH (2009) Rearing environmentaffects the brain size of guppies: lab-reared guppies havesmaller brains than wild-caught guppies. Ethology 115(2):122–133. https://doi.org/10.1111/j.1439-0310.2008.01585.x

Cochran-Biederman JL, Wyman KE, French WE, Loppnow GL(2015) Identifying correlates of success and failure of nativefreshwater fish reintroductions. Conserv Biol 29(1):175–186.https://doi.org/10.1111/cobi.12374

Conroy MJ, Peterson JT (2013) Decision making in natural re-source management: a structured, adaptive approach. Wiley-Blackwel l , New York . h t tps : / /do i .o rg /10 .1002/9781118506196

Currens KP, Sharpe CS, Hjort R, Schreck CB, Li HW (1989)Effects of different feeding regimes on the morphometricsof Chinook salmon (Oncorhynchus tshawytscha) and rain-bow trout (O. mykiss). Copeia 3:689–695

Davis RE, Northcutt RG (1983) Fish Neurobiology, Vol 1 and 2.University of Michigan Press, Ann Arbor

Devlin RH, Vandersteen WE, Uh M, Stevens ED (2012)Genetically modified growth affects allometry of eye andbrain in salmonids. Can J Zool 90(2):193–202. https://doi.org/10.1139/z11-126

Dickens MJ, Delehanty DJ, Romero LM (2010) Stress: an inevi-table component of animal translocation. Biol Conserv143 (6 ) :1329–1341 . h t t p s : / / do i . o rg / 10 .1016 / j .biocon.2010.02.032

Ebbesson LOE, Braithwaite VA (2012) Environmental effects onfish neural plasticity and cognition. J Fish Biol 81(7):2151–2174. https://doi.org/10.1111/j.1095-8649.2012.03486.x

Fernald RD (2015) Social behaviour: can it change the brain?Anim Behav 103:259–265. https://doi.org/10.1016/j.anbehav.2015.01.019

Fernö A, Huse G, Jakobsen PJ, Kristiansen TS, Nilsson J (2011)Fish behaviour, learning, aquaculture and fisheries. In:Brown C, Laland K, Krause J (eds) Fish cognition andbehavior. Wiley-Blackwell, New York, pp 359–404.https://doi.org/10.1002/9781444342536.ch16

Gonda A, Välimäki K, Herczeg G, Merilä J (2012) Brain devel-opment and predation: plastic responses depend on evolu-tionary history. Biol Lett 8(2):249–252. https://doi.org/10.1098/rsbl.2011.0837

Hairston NG, Li KT, Easter SS (1982) Fish vision and the detec-tion of planktonic prey. Science 218(4578):1240–1242.https://doi.org/10.1126/science.7146908

Hargis WJ (1991) Disorders of the eye in finfish. Annu Rev FishDis 1:95–117. https://doi.org/10.1016/0959-8030(91)90025-F

Healy SD, Rowe C (2007) A critique of comparative studies ofbrain size. Proc R Soc B Biol Sci 274(1609):453–464.https://doi.org/10.1098/rspb.2006.3748

Hosmer DW, Lemeshow S (2000) Applied logistic regression.Wiley Interscience, New York. https://doi.org/10.1002/0471722146

Jeffery WR (2001) Cavefish as a model system in evolutionarydevelopmental biology. Dev Biol 231(1):1–12. https://doi.org/10.1006/dbio.2000.0121

Jonsson B, Jonsson N (2014) Early environment influences laterperformance in fishes. J Fish Biol 85(2):151–188. https://doi.org/10.1111/jfb.12432

Kihslinger RL, Nevitt GA (2006) Early rearing environment im-pacts cerebellar growth in juvenile salmon. J Exp Biol209(3):504–509. https://doi.org/10.1242/jeb.02019

Kihslinger RL, Lema SC, Nevitt GA (2006) Environmental rear-ing conditions produce forebrain differences in wild Chinooksalmon Oncorhynchus tshawytscha. Comp BiochemPhysiol-Part A: Mol Integr Physiol 145(2):145–151.https://doi.org/10.1016/j.cbpa.2006.06.041

Kleiman DG (1996) Reintroduction programs. In: Kleiman DG,Allen M, Thompson K, Lumpkin S, Harris H (eds) Wildmammalian captivity: principles and techniques. Universityof Chicago Press, Chicago, pp 297–305

Kröger RHH (2011) Physiological optics in fishes. In: Farrell AP,Stevens ED, Cech JJ, Richards JG (eds) Encyclopedia of fishphysiology: from genome to environment. Academic Press,Maryland Heights, pp 102–109. https://doi.org/10.1016/B978-0-12-374553-8.00090-3

Lima SL, Dill LM (1990) Behavioral decisions made under therisk of predation: a review and prospectus. Can J Zool 68(4):619–640. https://doi.org/10.1139/z90-092

Environ Biol Fish (2018) 101:383–401 399

López JC, Broglio C, Rodríguez F, Thinus-Blanc C, Salas C(2000) Reversal learning deficit in a spatial task but not in acued one after telencephalic ablation in goldfish. Behav BrainRes 109(1):91–98. https://doi.org/10.1016/S0166-4328(99)00167-9

Lorenzen K (2014) Understanding and managing enhancements:why fisheries scientists should care. J Fish Biol 85(6):1807–1829. https://doi.org/10.1111/jfb.12573

Lorenzen K, Beveridge M, Mangel M (2012) Cultured fish: inte-grative biology and management of domestication and inter-actions with wild fish. Biol Rev 87(3):639–660. https://doi.org/10.1111/j.1469-185X.2011.00215.x

Marchetti MP, Nevitt GA (2003) Effects of hatchery rearing onbrain structures of rainbow trout, Oncorhynchus mykiss.Environ Biol Fish 66(1):9–14. https://doi.org/10.1023/A:1023269221678

MBTSG (Montana Bull Trout Scientific Group) (1996) The role ofstocking in bull trout recovery. Montana Bull TroutRestoration Team, Helena

McNeil WJ (1964) A method of measuring mortality of pinksalmon eggs and larvae. Fish Bull 63:575–588

McPhail JD (1984) Ecology and evolution of sympatric stickle-backs (Gasterosteus): morphological and genetic evidencefor a species pair in Enos Lake, British Columbia. Can JZool 62(7):1402–1408. https://doi.org/10.1139/z84-201

Metcalfe NB, Taylor AC, Thorpe JE (1995) Metabolic rate, socialstatus and life-history strategies in Atlantic salmon. AnimBehav 49(2) :431–436 . h t tps : / / do i .o rg /10 .1006/anbe.1995.0056

Metcalfe NB, Valdimarsson SK, Morgan IJ (2003) The relativeroles of domestication, rearing environment, prior residenceand body size in deciding territorial contests between hatch-ery and wild juvenile salmon. J Appl Ecol 40(3):535–544.https://doi.org/10.1046/j.1365-2664.2003.00815.x

Näslund J, Johnsson JI (2016) Environmental enrichment for fishin captive environments: effects of physical structures andsubstrates. Fish Fish 17(1):1–30. https://doi.org/10.1111/faf.12088

Näslund J, Aarestrup K, Thomassen ST, Johnsson JI (2012) Earlyenrichment effects on brain development in hatchery-rearedAtlantic salmon (Salmo salar): no evidence for a criticalperiod. Can J Fish Aquat Sci 69(9):1481–1490. https://doi.org/10.1139/f2012-074

Näslund J, Rosengren M, Del Villar D, Gansel L, Norrgård JR,Persson L,Winkowski JJ, Kvingedal E (2013) Hatchery tankenrichment affects cortisol and shelter-seeking in Atlanticsalmon (Salmo salar). Can J Fish Aquat Sci 70(4):585–590. https://doi.org/10.1139/cjfas-2012-0302

Northmore D (2011) Optic tectum. In: Farrell AP, Stevens ED,Cech JJ, Richards JG (eds) Encyclopedia of fish physiology:from genome to environment. Academic Press, MarylandHeights, pp 131–142. https://doi.org/10.1016/B978-0-12-374553-8.00093-9

Olla BL, Davis MW (1989) The role of learning and stress inpredator avoidance of hatchery-reared Coho salmon(Oncorhynchus kisutch) juveniles. Aquaculture 76(3-4):209–214. https://doi.org/10.1016/0044-8486(89)90075-6

Olla BL, Davis MW, Ryer CH (1998) Understanding how thehatchery environment represses or promotes the developmentof behavioural survival skills. Bull Mar Sci 62:531–550

Pakkasmaa S, Ranta E, Piironen J (1998) A morphometric studyon four land-locked salmonid species. Ann Zool Fenn 35:131–140

Pankhurst NW, Montgomery JC (1994) Uncoupling of visual andsomatic growth in the rainbow trout Oncorhynchus mykiss.Brain Behav Evol 44(3):149–155. https://doi.org/10.1159/000113586

Park PJ, Chase I, Bell MA (2012) Phenotypic plasticity of thethreespine stickleback Gasterosteus aculeatus telencephalonin response to experience in captivity. Current Zoology 58(1):189–210. https://doi.org/10.1093/czoolo/58.1.189

Rodríguez F, López CJ, Vargas JP, Gómez Y, Broglio C, Salas C(2002) Conservation of spatial memory function in the pallialforebrain of reptiles and ray-finned fishes. J Neurosci 22(7):2894–2903

Rodríguez F, Durán E, Gómez A, Ocaña FM, Álvarez E, Jiménez-Moya F, Broglio C, Salas C (2005) Cognitive and emotionalfunctions of the teleost fish cerebellum. Brain Res Bull 66(4-6):365–370. https://doi.org/10.1016/j.brainresbull.2004.11.026

Rosset A, Spadola L, Ratib O (2004) OsiriX: an open-sourcesoftware for navigating in multidimensional DICOM images.J Digit Imaging 17(3):205–216. https://doi.org/10.1007/s10278-004-1014-6

Salvanes AG, Moberg O, Ebbesson LO, Nilsen TO, Jensen KH,Braithwaite VA (2013) Environmental enrichment promotesneural plasticity and cognitive ability in fish. Proc R Soc B280:3113

Shively D, Allen C, Alsbury T, Bergamini B, Goehring B,Horning T, and Strobel B. 2007. Clackamas River BullTrout Reintroduction Feasibility Assessment. Published byUSDA Forest Service, Mt. Hood National Forest; U.S. Fishand Wildlife Service, Oregon State Office; and OregonDepartment of Fish and Wildlife, North Willamette Region

Simões JM, Teles MC, Oliveira RF, Van der Linden A, VerhoyeM(2012) A three-dimensional stereotaxicMRI brain atlas of thecichlid fish Oreochromis mossambicus. PLoS One 7(9):e44086. https://doi.org/10.1371/journal.pone.0044086

Sinn DL, Gosling SD, Moltschaniwskyj NA (2008) Developmentof shy/bold behaviour in squid: context-specific phenotypesassociated with developmental plasticity. Anim Behav 75(2):433–442. https://doi.org/10.1016/j.anbehav.2007.05.008

Soares D, Yamamoto Y, Strickler AG, JefferyWR (2004) The lenshas a specific influence on optic nerve and tectum develop-ment in the blind cavefish Astyanax. Dev Neurosci 26(5-6):308–317. https://doi.org/10.1159/000082272

Sundström LF, Johnsson JI (2001) Experience and social environ-ment influence the ability of young brown trout to forage onlive novel prey. Anim Behav 61(1):249–255. https://doi.org/10.1006/anbe.2000.1593

Sundström LF, Petersson E, Höjesjö J, Johnsson JI, Järvi T (2004)Hatchery selection promotes boldness in newly hatchedbrown trout (Salmo trutta): implications for dominance.Behav Ecol 15(2):192–198. https://doi.org/10.1093/beheco/arg089

Taylor EB (1986) Differences in morphology between wild andhatchery populations of juvenile Coho salmon. Prog FishCult 48(3):171–176. https://doi.org/10.1577/1548-8640(1986)48<171:DIMBWA>2.0.CO;2

Teletchea F (2017) Wildlife conservation: is domestication a solu-tion?. Global exposition of wildlife management. InTech

400 Environ Biol Fish (2018) 101:383–401

Thompson WL, Lee DC (2000) Modeling relationships betweenlandscape-level attributes and snorkel counts of Chinooksalmon and steelhead parr in Idaho. Can J Fish Aquat Sci57(9):1834–1842. https://doi.org/10.1139/f00-135

Tiffan KF, Connor WP (2011) Distinguishing between natural andhatchery Snake River fall Chinook Salmon subyearlings in thefield using body morphology. Trans Am Fish Soc 140:21–30

Ullmann JF, Cowin G, Kurniawan ND, Collin SP (2010) A three-dimensional digital atlas of the zebrafish brain. NeuroImage51(1):76–82. https://doi.org/10.1016/j.neuroimage.2010.01.086

USFWS (U.S. Fish and Wildlife Service). (2015) Recovery planfor the coterminous United States population of bull trout(Salvelinus confluentus). Portland

Wall GL (1963) The vertebrate eye and its adaptive radiation.Haffner Publishing, New York

Weigel DE, Peterson JT, Spruell P (2003) Introgressive hybridiza-tion between native cutthroat trout and introduced rainbowtrout. Ecol Appl 13(1):38–50. https://doi.org/10.1890/1051-0761(2003)013[0038:IHBNCT]2.0.CO;2

West-Eberhard MJ (1989) Phenotypic plasticity and the origins ofdiversity. Annu Rev Ecol Syst 20(1):249–278. https://doi.org/10.1146/annurev.es.20.110189.001341

Williams DA (1982) Extra-binomial variation in logistic linearmodels. Appl Stat 31(2):144–148. https://doi.org/10.2307/2347977

Wilson ADM, McLaughlin RL (2010) Foraging behaviour andbrain morphology in recently emerged brook charr,Salvelinus fontinalis. Behav Ecol Sociobiol 64(11):1905–1914. https://doi.org/10.1007/s00265-010-1002-4

Woodward CC, Strange RL (1987) Physiological stress responsesin wild and hatchery-reared rainbow trout. Trans Am FishSoc 116(4):574–579. https://doi.org/10.1577/1548-8659(1987)116<574:PSRIWA>2.0.CO;2

Zupanc G (2006) Neurogenesis and neuronal regeneration in theadult fish brain. J Comp Physiol A 192(6):649–670.https://doi.org/10.1007/s00359-006-0104-y

Environ Biol Fish (2018) 101:383–401 401