ARTICLE

Angling selects against active and stress-resilient phenotypesin rainbow troutBarbara Koeck, Libor Závorka, David Aldvén, Joacim Näslund, Robert Arlinghaus, Per-Ove Thörnqvist,Svante Winberg, Björn Thrandur Björnsson, and Jörgen I. Johnsson

Abstract: Selection induced by human harvest can lead to different patterns of phenotypic change than selection induced bynatural predation and could be a major driving force of evolution of wild populations. The vulnerability of individuals to anglingdepends on the individual decision to ingest the bait, possibly mediated by their neuroendocrine response towards the associ-ated stimulus. To investigate the mechanisms behind individual vulnerability to angling, we conducted angling experiments inreplicated ponds and quantified individual behavioral traits and neuroendocrine stress responsiveness in two salmonid species,rainbow trout (Oncorhynchus mykiss) and brown trout (Salmo trutta). We discovered a phenotypic syndrome in rainbow trout, butnot in brown trout, where lower serotonergic and dopaminergic brain activity and cortisol levels (i.e., lower stress responsiveness) inresponse to a standardized experimental stressor were associated with higher activity, forming a proactive phenotype thatshowed increased vulnerability to angling. Our results show that angling targets the most stress-resilient and active phenotypesof rainbow trout, supporting the suggestion that fishing-induced phenotypic selection may lead to an increased representationof stress-responsive and low-activity phenotypes in harvested populations.

Résumé : La sélection découlant de la pêche par les humains peut mener à des motifs de changement phénotypique différentsde ceux qui découlent de la prédation naturelle et pourrait être un important facteur d’évolution de populations sauvages. Lavulnérabilité d’individus à la pêche à la ligne dépend de la décision individuelle d’ingérer l’appât, décision éventuellementmodulée par la réaction neuroendocrine de l’individu au stimulus associé. Pour étudier les mécanismes qui sous-tendent lavulnérabilité individuelle à la pêche à la ligne, nous avons mené des expériences de pêche à la ligne dans des bassins en replicaset quantifié des caractères comportementaux et la réactivité neuroendocrine au stress d’individus de deux espèces de salmoni-dés, la truite arc-en-ciel (Oncorhynchus mykiss) et la truite fario (Salmo trutta). Nous avons ainsi découvert un syndrome phéno-typique chez la truite arc-en-ciel, mais non chez la truite fario, dans lequel l’activité sérotonergique et dopaminergique ducerveau et les concentrations de cortisol (c.-à-d. une réactivité réduite au stress) en réaction à un stress expérimental normalisésont associées à une plus grande activité, formant un phénotype proactif qui présente une vulnérabilité accrue à la pêche à laligne. Nos résultats montrent que la pêche à la ligne cible les phénotypes de truite arc-en-ciel les plus actifs et résilients au stress,ce qui appuie le postulat voulant que la sélection phénotypique induite par la pêche pourrait mener à une représentation accruede phénotypes réactifs au stress et de faible activité dans les populations exploitées. [Traduit par la Rédaction]

IntroductionNatural predation and human predation often select for differ-

ent traits, which can lead to divergent or even opposing selectionpatterns (Carlson et al. 2007). For example, while natural preda-tors are usually gape limited (Godin 1997), fisheries tend to focuson the harvest of the largest individuals of the population becauseof their value and physical gear constraints and facilitated byharvest regulations (Allendorf and Hard 2009; Jørgensen and Holt2013). A large body of literature has revealed that intense andsize-selective harvesting favors evolution of fast life history (e.g.,Jørgensen et al. 2009; Heino et al. 2015; Uusi-Heikkilä et al. 2015).The pace-of-life syndrome hypothesis predicts that individual lifehistory, behavioral, and physiological traits are correlated along a

gradient of slow to fast life history strategies (Réale et al. 2010).Accordingly, individuals with, for example, a fast pace-of-life arepredicted to be bold and active and to have a low hypothalamic–pituitary–adrenal axis reactivity, i.e., low stress responsiveness,traits associated with rapid growth (Koolhaas et al. 1999; Coppenset al. 2010). Following the pace-of-life syndrome, fishing selectionmight in turn also lead to corresponding changes in the behaviorand stress responsiveness of surviving phenotypes (Biro and Post2008; Uusi-Heikkilä et al. 2008). Evidence for fisheries-inducedevolution of behavior and physiology is, however, scarce (Heinoet al. 2015; Arlinghaus et al. 2017; Díaz Pauli and Sih 2017; Hollinset al. 2018).

Furthermore, with passive fishing gear, the harvest-induced se-lection depends on the ultimate decision of an individual fish to

Received 6 March 2018. Accepted 15 May 2018.

B. Koeck,* L. Závorka,† D. Aldvén, J. Näslund,‡ B.T. Björnsson, and J.I. Johnsson. Department of Biological and Environmental Sciences, Universityof Gothenburg, Sweden.R. Arlinghaus. Department of Biology and Ecology of Fishes, Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Berlin, Germany; Division ofIntegrative Fisheries Management, Faculty of Life Sciences, Humboldt Universität zu Berlin, Germany.P.-O. Thörnqvist and S. Winberg. Department of Neuroscience, Physiology Unit, Biomedical Center BMC, Uppsala University, Sweden.Corresponding author: Barbara Koeck (emails: koeckbarbara@hotmail.com; barbara.koeck@glasgow.ac.uk).*Present address: Institute of Biodiversity, Animal Health and Comparative Medicine, College of Medical, Veterinary and Life Sciences, Graham KerrBuilding, University of Glasgow, G12 8QQ, Glasgow, United Kingdom.

†Present address: Laboratoire Évolution and Diversité Biologique (EDB UMR 5174), CNRS, Université de Toulouse, Toulouse, France.‡Present address: Department of Zoology, University of Stockholm, Sweden.Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.

320

Can. J. Fish. Aquat. Sci. 76: 320–333 (2019) dx.doi.org/10.1139/cjfas-2018-0085 Published at www.nrcresearchpress.com/cjfas on 22 May 2018.

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

ingest a bait, attack a lure, or approach a trap or gill net based onthe perception and processing of different sensory cues (Andersenet al. 2016; Arlinghaus et al. 2017; Hollins et al. 2018). A number ofstudies have shown that vulnerability to harvest by passive fishinggear, such as angling, is positively related to behavioral traits suchas boldness, aggression, exploration, or activity (Klefoth et al.2012, 2017; Sutter et al. 2012; Härkönen et al. 2015; Wilson et al.2015; Alós et al. 2016; Monk and Arlinghaus 2018). However, theevidence is far from conclusive, as several empirical studies reportno or limited correlations between angling vulnerability and cer-tain behavioral traits (Kekäläinen et al. 2014; Vainikka et al. 2016;Monk and Arlinghaus 2017).

The pathways to a decision (i.e., to a behavioral response) varybetween individuals based on their global bodily state, definedby a combination of sensory information, physiological and de-velopmental state, and motivations (e.g., hunger or threat levels:LeDoux 2012). The behavioral response can therefore differ be-tween individuals, mediated by underlying differences in the ac-tivation of the hypothalamic–pituitary–interrenal axis (HPI axis)that controls the release of corticosteroids to the blood circulation(e.g., cortisol) and brain neurochemistry (e.g., serotonin and dopa-mine monoamine neurotransmitters and noradrenalin: WendelaarBonga 1997; Coppens et al. 2010; Andersen et al. 2016). While evi-dence has been provided revealing a link between stress respon-siveness (i.e., activation of the HPI axis) and cognitive appraisal ofaversive stimuli (Moltesen et al. 2016), this link has rarely beeninvestigated in the context of vulnerability to fishing (Louisonet al. 2017) and remains not well understood (Hollins et al. 2018).

The divergent results found in studies investigating the behav-ioral drivers for vulnerability to fishing might arise from gear- andspecies-specific differences and be affected by the context of stud-ies. For instance, compared to studies in controlled laboratoryenvironments, in natural systems, the spatial component mightbe a more important driver for the encounter between gear andfish overriding the individual variability in the decision pathwaysto attack a lure or ingest a bait (Monk and Arlinghaus 2017, 2018).To improve our understanding of the mechanisms that drive thecapture process and identify which traits might be under selec-tion, studies on populations with a known phenotypic distribu-tion conducted at relevant spatiotemporal scales that allow fish toexpress their natural response to fishing gear (Klefoth et al. 2012;Härkönen et al. 2014) are required. Such studies can then be scaledto population-level processes through whole-system studies ormodeling (Andersen et al. 2018).

In the present study, we investigated, using a replicated pondsystem, the selection mechanisms of angling by looking at thebehavioral and neurobiological basis of the variability in vulner-ability to angling in two salmonid species with different domesti-cation histories, rainbow trout (Oncorhynchus mykiss) and browntrout (Salmo trutta). Specifically, we addressed the following ques-tions. (i) Do capture patterns differ between rainbow trout andbrown trout? (ii) Are behavioral type and stress responsivenesscorrelated phenotypic traits? (iii) Is vulnerability to angling linkedto individual phenotypes of fish?

Material and methods

Experimental setup and activity monitoringTo answer our questions, we used four semicontrolled meso-

cosm ponds to carry out two replicated angling experiments onstocked fish of known size distribution and for which we recordedcapture rates, order of capture, individual activity, and stress re-sponsiveness. Each mesocosm pond (area = 720 m2, depth = 2 m),belonging to the Swedish Anglers’ Association (Sportfiskarna) inGothenburg, Sweden (57.693°N, 12.037°E) (Appendix A), was stockedwith size-matched rainbow (298 ± 19 mm) and brown trout (281 ±15 mm) in equal densities (25 individuals per species and pond).Both trout species originated from the same hatchery and were

reared under comparable conditions (Källefalls Fiskodling). Thebrown trout were F1 offspring from wild parents captured in thenearby Lake Vättern. The rainbow trout were of a domesticatedstrain, bred since 1997 within the hatchery and used exclusivelyfor stocking in Swedish waters.

Prior to release, the fish were anesthetized (10% benzocaine at0.3 mL·L−1), measured for body wet mass and length and a passiveintegrated transponder (PIT) tag (HDX ISO 11784/11785, 23 mm,0.6 g in air, Texas Instruments Inc.) was inserted into the coelomiccavity for individual identification and activity monitoring in theponds (thereafter, referred to as pond activity). A custom-maderadio frequency identification detection (RFID) system was de-ployed in each pond to monitor pond activity of tagged fish, whichwas recorded over 35 days, covering a period before, during, andafter the angling experiments (between 2 October and 4 November2014). Each RFID system consisted of a set of four detection anten-nas (2.2 m × 2.3 m) deployed vertically and connected to a RFIDhalf-duplex reader (Oregon RFID), which recorded time and date,antenna ID, and the PIT tag code for each detected fish. The detec-tion range of antennas covered the entire antenna surface withina distance of 50 cm around the antenna frame. Raw detection datafrom the RFID antenna monitoring system was filtered based onfish ID, detection interval, and antenna to account for the biasinduced by repeated detections of a single fish when remaining inthe vicinity of the detection array of an antenna. The filterednumber of detections per individual was highly correlated withthe number of relocations between two antennas (i.e., number oftimes a fish moves between two antenna frames at distinct posi-tions; Spearman rank correlation: r = 0.88, p < 0.001). This indi-cated that the filtered number of detections of our RFID systemcan be used as a reliable proxy for individual activity, telling apartinactive fish from moving fish, and is not inflated by the detec-tions of stationary fish nearby an antenna.

Angling treatmentsTwo common types of angling techniques were used: natural

shrimp baits on a single hook and spinners as example of anartificial lure. Natural bait angling was carried out using theshrimp on a barbless hook attached beneath a float. Anglers werefishing passively with the shrimp bait by casting it and retrievingit slowly, while lure angling was fished actively using a spinner(Myran WIPP Yellow/White, 10 g), which was cast and retrieved ata faster speed. With the exception of these two types of terminaltackles, identical angling equipment was used for the experi-ments (Appendix A). Angling was carried out by instructed andexperienced volunteer anglers. Hooked fish were landed with aknotless net, unhooked, and kept in a holding tank until returnedto their pond at the end of the fishing event (i.e., day). A fish couldthus be captured only once per fishing event (i.e., day) but recap-tured at each new event in following days.

Experiment 1After 26 days of acclimation in the experimental ponds, includ-

ing 10 days of pre-angling activity monitoring, the first fishingexperiment was conducted. Fishing experiments were designed toensure a certain level of fish captures despite the small-scale studysystem and limited experimental period. It consisted of a total of10 days of angling, with 5 days (i.e., events) of natural bait anglingfollowed by 5 days of lure angling with two angling-free days inbetween. Each angling day, one angling event of 1 h of effectiveangling time took place at dawn during which three anglers wererotating every 10th min within or between ponds according to arandomization schedule to control for bias in fishing skills ofanglers and site preference (Appendix A). Overall, each pond re-ceived a total angling effort of 30 h (3 anglers × 1 h × 5 days ×2 weeks) or 42 angler hours per hectare. In experiment 1, anglingwas practiced in only two of the four experimental ponds (ponds 1and 3; Appendix A, Fig. A1) and was designed to identify (i) the

Koeck et al. 321

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

individual vulnerability to angling over time and its phenotypiccorrelates in rainbow trout and in brown trout and (ii) the species-specific selectivity of angling technique.

Experiment 2The second angling experiment consisted of three angling

events with simultaneous bait and lure angling and a total anglingeffort of 6 h per pond (2 anglers × 1 h × 3 days, 8.5 angler hours perhectare) or following the same procedure as in experiment 1 (seealso Appendix A). Angling was this time performed in all fourponds (see Fig. A1) to compare hook avoidance response in fishthat have either been previously exposed or were naïve to angling.Both angling techniques were used simultaneously to verify theresults from the species-specific selectivity of angling techniquefrom experiment 1 while excluding the temporal effect of a suc-cessive use of angling technique.

Stress responsiveness to a standard stressorAfter the angling experiments, we measured individual stress

responsiveness of fish by quantifying the products of the twomain hormonal systems involved in stress response in fish (i.e.,corticosteroids and catecholamines and their monoamine precur-sors: Schreck and Tort 2016) in response to a standardized expo-sure to an experimental stressor (i.e., transfer to holding tanks fora minimum of 12 h plus a 30 min open-field test). To do so, wedrained the ponds and transferred fish to flow-through holdingtanks where we kept them for a minimum of 12 h before exposingthem to a 30 min open-field test (see Appendix A for the detailedprotocol of the open-field test). Following the behavioral scoring,we anaesthetized fish (10% benzocaine at 0.5 mL·L−1) and mea-sured their body mass and standard length to get a measure ofgrowth before we assessed their individual physiological stressresponsiveness by measuring circulating plasma cortisol andbrain neurotransmitter activity (serotonin and dopamine system,noradrenaline).

Blood samples were taken from the caudal vein into a 1 mLheparinized syringe and kept on ice until sampling of a group offour fish was completed. Blood was then centrifuged for 7 min andplasma obtained, frozen on dry ice, and stored at –80 °C untilanalysis. Plasma cortisol levels were measured by radioimmuno-assay (Young 1986; Sundh et al. 2011). Anaesthetized fish wherethen euthanized by cutting the spinal cord. Brains were dissectedout and separated into hindbrain (optic tectum, cerebellum, andbrain stem) and forebrain (telencephalon and diencephalon), fro-zen in tinfoil on dry ice, and stored at –80 °C until analysis. Mono-amine neurotransmitters and metabolites were extracted fromthe hindbrain and measured using high-performance liquid chro-matography with electrochemical detection. Levels of noradrena-line (NA), serotonin (5-hydroxytryptamine, 5-HT), dopamine (DA),DA and 5-HT major metabolites (5-hydroxyindoleacetic acid (5-HIAA),and 3,4-dihydroxyphenylacetic acid (DOPAC) were measured afterbeing normalized for wet mass of brain tissue and expressed innanograms per gram of brain. The ratios between metabolite andmonoamine, 5-HIAA/5-HT and DOPAC/DA, were used as indicatorsfor serotonergic and dopaminergic activity, respectively.

Data analysis

Overall and species-specific catchabilityWe modeled the effects of species, angling gear, and previous

exposure to angling on number of captures collected from the twoangling experiments using generalized linear models (using aPoisson distribution with log link-function). Using angling exper-iment 1, we compared the number of captures using bait anglingbetween species (i.e., rainbow trout, brown trout) and the numberof captures using artificial lure between species. To get an idea ofgeneral hook avoidance, we used the capture data from experi-ment 2 to compare the number of captures based on previousexposure to angling, i.e., from ponds 2 and 4, where fish were

naïve to angling, with those from ponds 1 and 3, where fish hadalready been exposed to angling. We then used the data from onlyponds 2 and 4, with fish that were naïve to angling, to compare thenumber of captures by species and fishing gear (i.e., natural bait,artificial lure). No significant pond or size effects were found andwere thus removed from the final models.

Association between phenotypic traitsCorrelations between activity measured in the ponds and the

physiological variables involved in a stress response were testedusing Pearson’s correlation test, estimating Pearson’s r and its95% confidence intervals. Adjusted p values were computed usingthe false discovery rate method to account for multiple compari-sons (Benjamini and Hochberg 1995). Data analysis was done sep-arately for rainbow and brown trout. A further distinction wasmade for fish from ponds 1 and 3 and fish from ponds 2 and 4, asthey were exposed to different total angling efforts, which mayhave influenced the physiological stress response of fish mea-sured after the experimental angling. Ponds 1 and 3 are hereafterreferred to as high angling intensity treatment (36 h of anglingeffort or 42 angler hours per hectare) and ponds 2 and 4 as lowangling intensity treatment (6 h of angling effort or 8.5 anglerhours per hectare). Angling intensities from both treatments canbe compared to angling efforts from small lakes or put-and-takeponds over the course of a few days to a couple of weeks. Correla-tion analysis showed that pond activity and physiological mea-surements were not related to body size or to specific growth rate(SGR). In general, fish presented a slight loss in wet body massbetween the beginning and end of the experiment and a lowvariability of SGR between individual fish (mean ± SD rainbowtrout: SGR = –0.18 ± 0.07, brown trout: SGR = –0.16 ± 0.06). Ac-counting for individual fish size or growth was therefore deemedunnecessary for subsequent analysis.

Vulnerability of fish to angling and phenotypic correlatesThe vulnerability of individual fish to capture by angling was

modeled using semiparametric Cox proportional hazard regres-sions, with the probability until first capture as response variable.Comparisons were first made between species, and then for eachspecies, a separate model was constructed to investigate the rela-tionship with individual plasma cortisol levels (in response to astandard experimental stressor), activity monitored in the ponds,and activity monitored in laboratory conditions. Activity recordedin the laboratory open-field test was, however, unrelated to vul-nerability to angling in both species and could thus be removedfrom the survival models. Such survival analysis enables incorpo-ration of the information of captured and uncaptured individuals(i.e., right censored data) as well as of the time to event (e.g., thecapture order of individual fish). For the single species model, theused hazard function is of the form

h(t|z) � h0(t) exp(�z)

where h0 is the baseline hazard, z is a time-independent predictor(i.e., activity in the ponds or in laboratory conditions or plasmacortisol), and � is the hazard coefficient, estimated using a partiallikelihood function.

At the end of the experimental period, some brown trout maleswere maturing and were excluded from data analysis to avoid anyconfounding effects related to maturation. This explains theslightly lower sample sizes for the analyses of brown trout com-pared to rainbow trout data (see Results section for details). Datahandling and analysis were computed using the packages “car”and “survival” for the R statistical environment.

322 Can. J. Fish. Aquat. Sci. Vol. 76, 2019

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

Results

Technique- and species-specific vulnerability to anglingIn experiment 1, 72 out of 100 fish (72%) were captured at least

once. Overall, more rainbow trout were captured (45 out of 50 fish,i.e., 90%) than brown trout (27 out of 50 fish, i.e., 54%) (Fig. 1). Thetime to capture and the probability of remaining uncaptured weresignificantly lower in rainbow trout than in brown trout (see re-sults from the survival analysis) (Table 1; Fig. 2), indicating a lowerintrinsic vulnerability to angling and (or) stronger hook avoidancein brown trout. More recaptured individuals and higher individ-ual recapture rates were observed in rainbow trout (25 fish recap-tured up to three times and one fish up to four times over the twoexperiments) compared to brown trout (nine fish recaptured onlyonce). Additionally, in experiment 2, the number of fish capturedin ponds previously exposed to angling was significantly lowerthan in ponds with fish naïve to angling (effect of angling treat-ment: F[1,25] = 5.12, p = 0.032) (Fig. 3), indicating an avoidanceresponse in both fish species when previously exposed to angling.

During the first week of experiment 1 using natural bait, signif-icantly more rainbow trout than brown trout were captured(F[1,16] = 11.08, p = 0.004) (Fig. 1), whereas no significant differenceswere observed between rainbow trout and brown trout capturedduring the second week with spinner lures (F[1,16] = 1.61, p = 0.223)(Fig. 1). This result was confirmed in experiment 2, with simulta-neous bait and lure angling (interaction of species and anglingtechnique effect: F[1,22] = 5.71, p = 0.026): more rainbow trout werecaptured than brown trout with natural bait (Tukey post hoc test:Z = –2.75, p = 0.020) (Fig. 3), but there was no species differencein relation to captures with lures (Tukey post hoc test: Z = 0.27,p = 0.989) (Fig. 3).

Context dependency of phenotypic associations betweenbehavior and neuroendocrine stress response

For rainbow trout from the high-intensity angling treatment(n = 45) (Figs. A2 and Fig. 4), cortisol levels, 5-HIAA, and DA turn-over (DOPAC/DA) were all positively correlated to each other andnegatively correlated with pond activity. Additionally, cortisol,5-HIAA, and 5-HT were positively correlated with DOPAC. Forbrown trout from the high-intensity angling treatment (n = 35),5-HIAA and 5-HT were positively correlated with DOPAC and DAturnover (Fig. A4) and pond activity was correlated with 5-HIAA,5-HT, and 5-HT turnover. Associations between levels of mono-amine neurotransmitters in rainbow trout and in brown trout, inresponse to a standard experimental stressor, indicated the pres-ence of interindividual differences in the general activation of theHPI axis in both species. Our results further indicate an associa-tion between swimming activity in the ponds and activation ofthe HPI axis in both species.

For rainbow trout from the low-intensity angling treatment (n =43), however, no correlations between cortisol, monoaminergicactivity, or pond activity were visible (Fig. A3). For brown troutfrom the low-intensity angling treatment (n = 35), 5-HT was posi-

tively correlated with DA and activity (Fig. A5). Overall, the resultsshowed that phenotypic associations were mainly revealed undera context of high and repeated angling intensity and were lessevident for fish from the low-intensity angling treatment.

Vulnerability of fish to angling and phenotypic correlatesResults from angling experiment 1 (ponds 1 and 3) showed that

rainbow trout that were caught first and less likely to remainuncaptured over the course of the experiment had a higher pondactivity and a weaker cortisol response (Table 2; Fig. 5a), indicat-ing that the most active and stress-responsive rainbow trout wererelatively more vulnerable to angling. In brown trout, for whichcatch rates were much lower than for rainbow trout, no suchrelationship was found (Table 3; Fig. 5b).

Discussion

Species selection: conditioned hook avoidance andpotential effect of domestication

At the species level, our study confirmed that rainbow trout aremore vulnerable to angling than brown trout. This finding is con-sistent with previous angling experiments conducted in both riv-ers and ponds (Anderson and Nehring 1984; Pawson 1991; Mezzeraand Largiadèr 2001) and thus seems to be a general among-speciespattern. But we also found that the relative vulnerability to an-gling between both species was technique specific: whereas rain-bow trout was more vulnerable to natural bait angling thanbrown trout, both species were equally vulnerable to lure angling.Studies on a range of predatory species suggest that catchabilityis often higher for natural bait compared to artificial lures(Beukema 1970; Härkönen et al. 2015; Moraga et al. 2015), indicat-ing that when associated with a reward (i.e., natural bait), theconditioned avoidance in response to the aversive stimulus re-lated to angling is reduced. These findings were mirrored in ourwork, where catch rates were overall and specifically for rainbowtrout higher for natural bait than for artificial lures. Hook avoid-ance of artificial lures was generally extremely rapid within just 1or 2 fishing days, similar to findings in other studies (e.g., Askeyet al. 2006). These species-specific findings indicate that both spe-cies respond differently to the cues associated with the passivelyfished natural bait, shaping the relative vulnerability of both spe-cies. It is likely that the mechanisms underlying the effects relate

Fig. 1. Number of captured fish per angling day for successive bait and lure angling (experiment 1) in ponds 1 and 3.

Table 1. Cox-proportional hazards regression model comparing thetime until first capture and the probability of capture of rainbow trout(Oncorhynchus mykiss) with brown trout (Salmo trutta) used as baseline inthe model, with � the corresponding hazard coefficient and e� thehazard probability.

� e� SE (�) z p

Rainbow trout 0.8728 2.3935 0.252 3.454 0.0005

Note: The number of events refers to the total number of captured individuals.n = 93, number of events = 73; likelihood ratio test = 12.76 on 1 df, p < 0.001.

Koeck et al. 323

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

to interspecific behavioral and physiological differences and mayadditionally be shaped by domestication effects.

Numerous studies have highlighted behavioral and physiologicalintraspecific differences between wild and domesticated strains,the latter ones being recognized to be bolder and more risk prone(Johnsson et al. 2001; Sundström et al. 2004; Huntingford andAdams 2005) and presenting weaker neuroendocrine stress re-sponsiveness than wild genotypes (i.e., lower dopaminergic turn-over DOPAC/DA and serotonergic turnover 5-HIAA/5-HT due tolower levels of serotonin 5-HT and dopamine DA: Lepage et al.2000). Both species used in this experiment were hatchery reared,but rainbow trout came from a strongly domesticated strainshaped by decades of selective breeding, whereas the brown troutstemmed from wild parents. Although in our experiment, speciesand degree of domestication are confounded, it is possible thatthe long domestication history that rainbow trout were exposedto contributed to their higher intrinsic vulnerability, as shownbefore for common carp (Cyprinus carpio) genotypes differing bydegree of domestication (Klefoth et al. 2012). Whereas brown troutand rainbow trout differ in terms of stress physiology (e.g., rain-bow trout has a faster recovery in terms of primary and secondarystress response compared to brown trout: Ruane et al. 1999) andbehavior (e.g., foraging and habitat use: Blanchet et al. 2007), theobserved increased angling vulnerability of rainbow trout relativeto brown trout can thus also be partly attributed to the effects ofdomestication. Existing studies show indeed that wild strains ofrainbow trout have higher survival rates and are less likely to be

harvested by angling than domesticated strains (Brauhn andKincaid 1982; Dwyer and Piper 1984).

Activity and stress responsiveness explain variability invulnerability to angling in rainbow trout

As shown in previous studies, the capture data from our anglingexperiments have highlighted the presence of individual differ-ences in vulnerability of fish to angling shaping overall population-level catchability (e.g., Beukema 1970; Askey et al. 2006). Suchdifferences were particularly marked in rainbow trout for whichcertain individuals had been recaptured up to four times, whileothers remained uncaptured despite repeated angling effort. Inbrown trout, capture rates declined rapidly after 1 day of fishingand remained overall much lower than for rainbow trout. Previ-ous studies using selection lines of largemouth bass (Micropterussalmoides) of high versus low vulnerability to angling have high-lighted that differences in susceptibility to capture can be linkedto individual variability of physiological traits such as metabolismand anaerobic activity (Redpath et al. 2010) but also to levels ofcirculating cortisol as an indicator for stress responsivness of fish(Louison et al. 2017). In our experiment, after the standardizedexposure to an experimental stressor, both species showed gener-ally elevated cortisol levels, which were about 10-fold higher thanbasal levels reported for fish (Pankhurst 2011), indicating that allfish were in a stressed state when brain tissues were sampled tomeasure neurotransmitter activity. However, individual variabilityin levels of corticosteroids, catecholamines, and their monoamine

Fig. 2. Survival curves (i.e., probability of remaining uncaptured over time) of (a) rainbow trout (Oncorhynchus mykiss) and (b) brown trout(Salmo trutta) during the first angling experiment in ponds 1 and 3 (i.e., under the high-intensity angling treatment). Broken lines correspondto a 95% confidence envelop around the survival function. Days 1–5: natural bait angling, days 8–12: artificial lure angling.

Fig. 3. Number of captured fish per angling day for simultaneous bait and lure angling (experiment 2) in previously unfished control ponds(ponds 2 and 4: low angling intensity) and previously fished ponds (ponds 1 and 3: high angling intensity).

324 Can. J. Fish. Aquat. Sci. Vol. 76, 2019

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

precursors was also evident in our results, indicating differences instress responsiveness of individual fish within species.

In rainbow trout that had been repeatedly exposed to angling(i.e., high-intensity angling treatment), we discovered a pheno-typic syndrome, i.e., covariation among functionally related traitsthat defines how the organism interacts with its environment andsustains itself (Pigliucci and Hayden 2001; Závorka et al. 2017).Individual pond activity and neuroendocrine activity (i.e., dopa-mine turnover and 5-HIAA, a major metabolite of serotonin) andcortisol response (after exposure to a standardized experimentalstressor) were associated and correlated with individual vulnera-bility to angling such that active rainbow trout had a relativeweaker neuroendocrine stress responsiveness than less activerainbow trout. The role of neurotransmitters in stress physiologyand behavioral plasticity in vertebrates (Coppens et al. 2010), in-cluding fish (Winberg and Thörnqvist 2016), has recently beenhighlighted. The chronic activation of the brain serotonergic sys-

Fig. 4. Correlation plots for rainbow trout from the high-intensity angling treatment between pond activity (log transformed number ofdetections) and plasma cortisol, dopamine turnover (DOPAC/DA), and serotonin metabolite 5-HIAA levels after exposure to a standardizedexperimental stressor. Pearson’s r correlation coefficient, confidence intervals, and p values are displayed. A linear model was fitted (solidline) and a 95% confidence interval estimated (grey area).

Table 2. Cox-proportional hazards regression model examining thetime until first capture and the probability of capture of rainbow trout(Oncorhynchus mykiss) in relationship to individual pond activity (i.e.,number of detections) and plasma cortisol levels (after exposure to astandardized experimental stressor), with � the corresponding hazardcoefficient and e� the hazard probability.

� e� SE (�) z p

Detections 0.1827 1.2005 0.074 2.346 0.019Cortisol −0.0048 0.9951 0.003 −2.083 0.037

Note: The number of events refers to the number of individuals captured byangling during angling experiment 1 in ponds 1 and 3 (i.e., under the high-intensity angling treatment). n = 47, number of events = 44; likelihood ratiotest = 11.99 on 2 df, p = 0.002.

Koeck et al. 325

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

tem can cause behavioral inhibition of feeding, locomotion, andaggression (Winberg and Thörnqvist 2016). Furthermore, the do-pamine system is known to have a central role in conditionedbehavioral responses, influencing the motivational control overbehavior (Wise 2004) affecting for instance risk taking or avoid-ance behavior (Höglund et al. 2005; Arias-Carrion and Poeppel2007). Dopamine is also the precursor of noradrenaline, known toincreases arousal and general alertness (Singh et al. 2015). In rain-bow trout, individual differences in the activation of the HPI axiscould therefore explain the observed differences in their behaviortowards fishing gear, via a possible mechanism of sensory modu-lation leading to behavioral habituation or inhibition, shapingthe hook avoidance response under repeated angling stimuli.

The phenotypic syndrome associating pond activity and physi-ological stress responsiveness (association with the activation ofthe serotonergic system for brown trout and additionally withthe activation of the dopamine system and levels of circulatingplasma cortisol in rainbow trout) was, however, only visible underhigh angling intensity. In particular, in rainbow trout previouslyexposed to a low angling intensity, there was a total absence of

associations of physiological stress responsiveness and behavioraltraits. Similarly, Killen and colleagues (2013) have shown that therelationships between behavioral and physiological traits aresometimes only revealed by the presence of an environmentalstressor. We suggest that the phenotypic syndrome linking behav-ioral and physiological traits (i.e., stress response) in rainbowtrout was revealed only under repeated angling causing a chronicstress response. Furthermore, it has been shown in rainbow troutthat only the chronic activation of the brain serotonergic systemcauses a general behavioral inhibition (Winberg et al. 2001;Winberg and Thörnqvist 2016).

For brown trout, while we also noted an association betweenpond activity and serotonergic activity, we found no link betweenvulnerability to angling and the measured phenotypic traits. Thiscould be simply due to the overall lower catch rates found inbrown trout compared to rainbow trout and explained by the factthat brown trout recovers slower in terms of primary and second-ary stress response in comparison to rainbow trout (Ruane et al.1999), therefore increasing hook avoidance in that species. Addi-tionally, sexual maturity was observed in some brown trout at theend of the experiment and might have affected their stress re-sponsiveness (Pottinger et al. 1995). Further experiments, allow-ing longer recovery periods between angling events or applyinglower fishing effort that is more realistic of natural settings, arerequired to investigate the behavioral, physiological, and neuro-biological mechanisms of angling selection in brown trout andwild fish in general.

Management implicationsThe results that we found provide empirical evidence that har-

vest by angling can selectively target a nonrandom subset of har-vested populations characterized by active and stress-resilientphenotypes and that this result is not necessarily applying acrosscontexts and species. However, for some species based on thedocumented assumption that the variation of phenotypic traits

Fig. 5. Average pond activity and cortisol level (after exposure to a standardized experimental stressor) of the remaining uncaptured(a) rainbow trout (Oncorhynchus mykiss) and (b) brown trout (Salmo trutta) over angling days of experiment 1 and high-intensity anglingtreatment. Box plots of the first angling day (12 October) represent the average pond activity and cortisol level of rainbow and brown troutfrom the high-intensity angling treatment (ponds 1 and 3). After the first capture of an individual, its corresponding activity or cortisol levelvalue is removed from the data set and the average value recalculated for the remaining uncaptured fish.

Table 3. Cox-proportional hazards regression model examining thetime until first capture and the probability of capture of brown trout(Salmo trutta) in relationship to individual pond activity (i.e., numberof detections) and plasma cortisol levels (after exposure to a standard-ized experimental stressor), with � the corresponding hazard coeffi-cient and e� the hazard probability.

� e� SE (�) z p

Detections −0.0064 0.9936 0.103 −0.069 0.945Cortisol 0.0041 1.0041 0.004 0.959 0.338

Note: The number of events refers to the number of individuals captured byangling during angling experiment 1 in ponds 1 and 3 (i.e. under the high-intensity angling treatment). n = 37, number of events = 24; likelihood ratiotest = 0.82 on 2 df, p = 0.663.

326 Can. J. Fish. Aquat. Sci. Vol. 76, 2019

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

presents some degree of heritability (Philipp et al. 2009; Danchinet al. 2011), the selective removal of the most stress-resilient andactive phenotypes by intensive angling may cause a shift towardsan increase of shy phenotypes in exploited compared to unex-ploited populations (Arlinghaus et al. 2017; Louison et al. 2017).

The progressive removal of the fraction of highly vulnerableindividuals can then generate a decrease in catchability (Alós et al.2016) and affect the size selectivity of harvesting (Tsuboi et al.2016), ultimately affecting the satisfaction of anglers (Arlinghauset al. 2017). In catch-and-release fisheries, even if fish populationsmight remain unaffected by evolutionary effects linked to selec-tive removal, behavioral plasticity or adaptation in response todirect and indirect exposure to fishing gear can further decreasethe fishing quality and catchability (Askey et al. 2006). In fact, incomparison to consumptive fisheries, under catch-and-releaseregulations, fish have the potential to learn to identify and avoidthe threat associated with angling via mechanisms of private(Beukema 1970; Klefoth et al. 2013) but also public information use(Danchin et al. 2004). This suggests that under intensive angling,fishing success will increasingly become density independent. Ro-tational management systems might then be required in small-scale fisheries with high fishing pressure to enable fish to returnto vulnerable stages after direct or indirect exposure to angling(Camp et al. 2015).

Domesticated fish appear generally to be more vulnerable tocapture than wild fish (Brauhn and Kincaid 1982; Dwyer and Piper1984; García-Marín et al. 1998; Mezzera and Largiadèr 2001; Biroand Post 2008). Our study adds to this literature showing thatangling within a domesticated strain targets the most stress-resilient and active phenotypes, considered to be the fraction of apopulation with the highest invasion potential (Phillips andSuarez 2012; Juette et al. 2014). This indicates that angling could betailored to reduce the invasion potential of escaped or uninten-tionally released fish by reducing the propagule pressure of themost invasive phenotypes, at least when applied shortly after therelease and in the vicinity of the release site. In practice, cautionneeds, however, to be taken, as the efficiency of angling will bediluted in large fishery systems and with time after release due tobehavioral plasticity and adaptation. Surviving domesticated fishcan learn to avoid areas exposed to angling pressure and adapttheir behavioral response to environmental and human-inducedstressors rendering them less vulnerable to harvest over time,which may have further potential cascading effects on ecosystemfunctioning (Evangelista et al. 2017; Závorka et al. 2018). There-fore, our study shall not be misread to suggest that angling alonewill be an efficient removal method of invasive phenotypes orspecies (e.g., Paul et al. 2003; Evangelista et al. 2015).

Results from our study show that angling targets the moststress-resilient and active phenotypes of rainbow trout, support-ing the suggestion that fishing-induced phenotypic selection maylead to an increased representation of stress-responsive and low-activity phenotypes in intensively harvested populations. Suchselective removal may have further potential spillover effects athigher levels of ecological organization (Palkovacs et al. 2018)as well as for stock assessments and fisheries management(Arlinghaus et al. 2017). However, as we did not find similar resultsin brown trout, we consider that further experiments on wild fishand across environmental and fishing contexts are needed to im-prove our understanding of fishing selection and predict possibleeffects of fisheries-induced evolution. Such experiments shouldparticularly focus on further investigating the mechanistic linkbetween behavioral and physiological traits that can possibly in-fluence the fishing process (Hollins et al. 2018).

AcknowledgementsB.K., R.A., and J.I.J. developed the experimental design, B.K., L.Z.,

D.A., and J.N. carried out the experiment, P.-O.T., S.W., and B.T.B.analyzed the samples, and B.K. performed data analysis and wrote

the initial draft. All authors contributed to the interpretation ofresults and the writing and editing of the paper. We warmly thankthe volunteer anglers and Per-Erik Jacobsen, Hans Lindqwist, andMarkus Lundgren from the Swedish Anglers Association Sport-fiskarna. Many thanks to Jes Dolby for his assistance with the RFIDantennas and Johan Höjesjö and Julien Cucherousset for theirinput and help at an earlier stage of this study. We also thank twoanonymous reviewers for their excellent feedback. This study waspart of the SalmoInvade project “Causes and consequences ofinvasions of aquatic ecosystems by non-native salmonids” fundedby the BiodivERsA grant for research and innovation 2012–2013from the European Union and the Swedish Research CouncilFORMAS. J.I.J. was also financed by the Interreg-project MarGen.These experiments were conducted under license 15-2014 issuedby the Ethical Committee for Animal Research in Gothenburg andcomply with Swedish and European law. The data supporting thisarticle have been posted to figshare: http://dx.doi.org/10.6084/m9.figshare.c.3467493.

ReferencesAllendorf, F.W., and Hard, J.J. 2009. Human-induced evolution caused by unnat-

ural selection through harvest of wild animals. Proc. Natl. Acad. Sci. U.S.A.106: 9987–9994. doi:10.1073/pnas.0901069106. PMID:19528656.

Alós, J., Palmer, M., Rosselló, R., and Arlinghaus, R. 2016. Fast and behavior-selective exploitation of a marine fish targeted by anglers. Sci. Rep. 6: 38093.doi:10.1038/srep38093. PMID:27922022.

Andersen, B.S., Jørgensen, C., Eliassen, S., and Giske, J. 2016. The proximatearchitecture for decision-making in fish. Fish Fish. 17: 680–695. doi:10.1111/faf.12139.

Andersen, K.H., Marty, L., and Arlinghaus, R. 2018. Evolution of boldness and lifehistory in response to selective harvesting. Can. J. Fish. Aquat. Sci. 75(2):271–281. doi:10.1139/cjfas-2016-0350.

Anderson, R.M., and Nehring, R.B. 1984. Effects of a catch-and-release regulationon a wild trout population in Colorado and its acceptance by anglers. N. Am.J. Fish. Manage. 4: 257–265. doi:10.1577/1548-8659(1984)4<257:EOACRO>2.0.CO;2.

Arias-Carrion, O., and Poeppel, E. 2007. Dopamine, learning, and reward-seekingbehavior. Acta Neurobiol. Exp. 67: 481–488. PMID:18320725.

Arlinghaus, R., Laskowski, K.L., Alós, J., Klefoth, T., Monk, C.T., Nakayama, S.,and Schröder, A. 2017. Passive gear-induced timidity syndrome in wild fishpopulations and its potential ecological and managerial implications. FishFish. 18: 360–373. doi:10.1111/faf.12176.

Askey, P.J., Richards, S.A., Post, J.R., and Parkinson, E.A. 2006. Linking anglingcatch rates and fish learning under catch-and-release regulations. N. Am. J.Fish. Manage. 26: 1020–1029. doi:10.1577/M06-035.1.

Benjamini, Y., and Hochberg, Y. 1995. Controlling the false discovery rate: apractical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. BMethodol. 57: 289–300.

Beukema, J.J. 1970. Acquired hook-avoidance in the pike Esox lucius L. fished withartificial and natural baits. J. Fish Biol. 2: 155–160. doi:10.1111/j.1095-8649.1970.tb03268.x.

Biro, P.A., and Post, J.R. 2008. Rapid depletion of genotypes with fast growth andbold personality traits from harvested fish populations. Proc. Natl. Acad. Sci.U.S.A. 105: 2919–2922. doi:10.1073/pnas.0708159105. PMID:18299567.

Blanchet, S., Loot, G., Grenouillet, G., and Brosse, S. 2007. Competitive interac-tions between native and exotic salmonids: a combined field and laboratorydemonstration. Ecol. Freshw. Fish, 16: 133–143. doi:10.1111/j.1600-0633.2006.00205.x.

Brauhn, J.L., and Kincaid, H. 1982. Survival, growth, and catchability of rainbowtrout of four strains. N. Am. J. Fish. Manage. 2: 1–10. doi:10.1577/1548-8659(1982)2<1:SGACOR>2.0.CO;2.

Camp, E.V., van Poorten, B.T., and Walters, C.J. 2015. Evaluating short openingsas a management tool to maximize catch-related utility in catch-and-releasefisheries. N. Am. J. Fish. Manage. 35: 1106–1120. doi:10.1080/02755947.2015.1083495.

Carlson, S.M., Edeline, E., Vollestad, L.A., Haugen, T.O., Winfield, I.J.,Fletcher, J.M., James, J.B., and Stenseth, N.C. 2007. Four decades of opposingnatural and human-induced artificial selection acting on Windermere pike(Esox lucius). Ecol. Lett. 10: 512–521. doi:10.1111/j.1461-0248.2007.01046.x. PMID:17498150.

Coppens, C.M., de Boer, S.F., and Koolhaas, J.M. 2010. Coping styles and behav-ioural flexibility: towards underlying mechanisms. Philos. Trans. R. Soc. BBiol. Sci. 365: 4021–4028. doi:10.1098/rstb.2010.0217.

Danchin, É., Giraldeau, L.-A., Valone, T.J., and Wagner, R.H. 2004. Public infor-mation: from nosy neighbors to cultural evolution. Science, 305: 487–491.doi:10.1126/science.1098254. PMID:15273386.

Danchin, É., Charmantier, A., Champagne, F.A., Mesoudi, A., Pujol, B., andBlanchet, S. 2011. Beyond DNA: integrating inclusive inheritance into an

Koeck et al. 327

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

extended theory of evolution. Nat. Rev. Genet. 12: 475–486. doi:10.1038/nrg3028. PMID:21681209.

Díaz Pauli, B., and Sih, A. 2017. Behavioral responses to human-induced change:why fishing should not be ignored. Evol. Appl. 10: 231–240. doi:10.1111/eva.12456. PMID:28250808.

Dwyer, W.P., and Piper, R.G. 1984. Three-year hatchery and field evaluation offour strains of rainbow trout. N. Am. J. Fish. Manage. 4: 216–221. doi:10.1577/1548-8659(1984)4<216:THAFEO>2.0.CO;2.

Evangelista, C., Britton, R.J., and Cucherousset, J. 2015. Impacts of invasive fishremoval through angling on population characteristics and juvenile growthrate. Ecol. Evol. 5: 2193–2202. doi:10.1002/ece3.1471. PMID:26078856.

Evangelista, C., Lecerf, A., Britton, R., and Cucherousset, J. 2017. Resource com-position mediates the effects of intraspecific variability in nutrient recyclingon ecosystem processes. Oikos, 126: 1439–1450. doi:10.1111/oik.03787.

García-Marín, J.L., Sanz, N., and Pla, C. 1998. Proportions of native and intro-duced brown trout in adjacent fished and unfished Spanish rivers. Conserv.Biol. 12: 313–319. doi:10.1046/j.1523-1739.1998.96133.x.

Godin, J.-G.J. (Editor). 1997. Evading predators. In Behavioural ecology of teleostfishes. Oxford University Press, Oxford, UK. pp. 191–236.

Härkönen, L., Hyvärinen, P., Paappanen, J., and Vainikka, A. 2014. Explorativebehavior increases vulnerability to angling in hatchery-reared brown trout(Salmo trutta). Can. J. Fish. Aquat. Sci. 71(12): 1900–1909. doi:10.1139/cjfas-2014-0221.

Härkönen, L., Hyvärinen, P., Niemelä, P., and Vainikka, A. 2015. Behaviouralvariation in Eurasian perch populations with respect to relative catchability.Acta Ethol. 19: 21–31. doi:10.1007/s10211-015-0219-7.

Heino, M., Díaz Pauli, B., and Dieckmann, U. 2015. Fisheries-induced evolution.Annu. Rev. Ecol. Evol. Syst. 46: 461–480. doi:10.1146/annurev-ecolsys-112414-054339.

Höglund, E., Weltzien, F.-A., Schjolden, J., Winberg, S., Ursin, H., andDøving, K.B. 2005. Avoidance behavior and brain monoamines in fish. BrainRes. 1032: 104–110. doi:10.1016/j.brainres.2004.10.050. PMID:15680947.

Hollins, J., Thambithurai, D., Koeck, B., Crespel, A., Bailey, D.M., Cooke, S.J.,Lindström, J., Parsons, K.J., and Killen, S.S. 2018. A physiological perspectiveon fisheries-induced evolution. Evol. Appl. 11(5): 561–576. doi:10.1111/eva.12597.

Huntingford, F., and Adams, C. 2005. Behavioural syndromes in farmed fish:implications for production and welfare. Behaviour, 142: 1207–1221. doi:10.1163/156853905774539382.

Johnsson, J.I., Höjesjö, J., and Fleming, I.A. 2001. Behavioural and heart rateresponses to predation risk in wild and domesticated Atlantic salmon. Can. J.Fish. Aquat. Sci. 58(4): 788–794. doi:10.1139/f01-025.

Jørgensen, C., and Holt, R.E. 2013. Natural mortality: its ecology, how it shapesfish life histories, and why it may be increased by fishing. J. Sea Res. 75: 8–18.doi:10.1016/j.seares.2012.04.003.

Jørgensen, C., Ernande, B., and Fiksen, O. 2009. Size-selective fishing gear andlife history evolution in the Northeast Arctic cod. Evol. Appl. 2: 356–370.doi:10.1111/j.1752-4571.2009.00075.x. PMID:25567886.

Juette, T., Cucherousset, J., and Cote, J. 2014. Animal personality and the ecolog-ical impacts of freshwater non-native species. Curr. Zool. 60: 417–427. doi:10.1093/czoolo/60.3.417.

Kekäläinen, J., Podgorniak, T., Puolakka, T., Hyvärinen, P., and Vainikka, A. 2014.Individually assessed boldness predicts Perca fluviatilis behaviour in shoals,but is not associated with the capture order or angling method. J. Fish Biol.85: 1603–1616. doi:10.1111/jfb.12516. PMID:25270290.

Killen, S.S., Marras, S., Metcalfe, N.B., McKenzie, D.J., and Domenici, P. 2013.Environmental stressors alter relationships between physiology and behav-iour. Trends Ecol. Evol. 28: 651–658. doi:10.1016/j.tree.2013.05.005. PMID:23756106.

Klefoth, T., Skov, C., Krause, J., and Arlinghaus, R. 2012. The role of ecologicalcontext and predation risk-stimuli in revealing the true picture about thegenetic basis of boldness evolution in fish. Behav. Ecol. Sociol. 66: 547–559.doi:10.1007/s00265-011-1303-2.

Klefoth, T., Pieterek, T., and Arlinghaus, R. 2013. Impacts of domestication onangling vulnerability of common carp, Cyprinus carpio: the role of learning,foraging behaviour and food preferences. Fish. Manag. Ecol. 20: 174–186.doi:10.1111/j.1365-2400.2012.00865.x.

Klefoth, T., Skov, C., Kuparinen, A., and Arlinghaus, R. 2017. Toward a mecha-nistic understanding of vulnerability to hook-and-line fishing: boldness asthe basic target of angling-induced selection. Evol. Appl. 10: 994–1006. doi:10.1111/eva.12504. PMID:29151855.

Koolhaas, J.M., Korte, S.M., De Boer, S.F., Van Der Vegt, B.J., Van Reenen, C.G.,Hopster, H., De Jong, I.C., Ruis, M.A.W., and Blokhuis, H.J. 1999. Coping stylesin animals: current status in behavior and stress-physiology. Neurosci. Biobe-hav. Rev. 23: 925–935. doi:10.1016/S0149-7634(99)00026-3. PMID:10580307.

LeDoux, J. 2012. Rethinking the emotional brain. Neuron, 73: 653–676. doi:10.1016/j.neuron.2012.02.004. PMID:22365542.

Lepage, O., Øverli, Ø., Petersson, E., Järvi, T., and Winberg, S. 2000. Differentialstress coping in wild and domesticated sea trout. Brain Behav. Evol. 56:259–268. doi:10.1159/000047209. PMID:11251318.

Louison, M.J., Adhikari, S., Stein, J.A., and Suski, C.D. 2017. Hormonal respon-siveness to stress is negatively associated with vulnerability to angling cap-

ture in fish. J. Exp. Biol. 220: 2529–2535. doi:10.1242/jeb.150730. PMID:28724703.

Mezzera, M., and Largiadèr, C.R. 2001. Evidence for selective angling of intro-duced trout and their hybrids in a stocked brown trout population. J. FishBiol. 59: 287–301. doi:10.1111/j.1095-8649.2001.tb00130.x.

Moltesen, M., Vindas, M.A., Winberg, S., Ebbesson, L., de Lourdes Ruiz-Gomez, M.,Skov, P.V., Dabelsteen, T., Øverli, Ø., and Höglund, E. 2016. Cognitive appraisal ofaversive stimulus differs between individuals with contrasting stress copingstyles; evidences from selected rainbow trout (Oncorhynchus mykiss) strains.Behaviour, 153: 1567–1587. doi:10.1163/1568539X-00003405.

Monk, C.T., and Arlinghaus, R. 2017. Encountering a bait is necessary but insuf-ficient to explain individual variability in vulnerability to angling in twofreshwater benthivorous fish in the wild. PLoS ONE, 12: e0173989. doi:10.1371/journal.pone.0173989. PMID:28301558.

Monk, C.T., and Arlinghaus, R. 2018. Eurasian perch, Perca fluviatilis, spatial be-haviour determines vulnerability independent of angler skill in a whole-lakereality mining experiment. Can. J. Fish. Aquat. Sci. 75(3): 417–428. doi:10.1139/cjfas-2017-0029.

Moraga, A.D., Wilson, A.D.M., and Cooke, S.J. 2015. Does lure colour influencecatch per unit effort, fish capture size and hooking injury in angled large-mouth bass? Fish. Res. 172: 1–6. doi:10.1016/j.fishres.2015.06.010.

Palkovacs, E.P., Moritsch, M.M., Contolini, G.M., and Pelletier, F. 2018. Ecology ofharvest-driven trait changes and implications for ecosystem management.Front. Ecol. Environ. 16: 20–28. doi:10.1002/fee.1743.

Pankhurst, N.W. 2011. The endocrinology of stress in fish: an environmentalperspective. Gen. Comp. Endocrinol. 170: 265–275. doi:10.1016/j.ygcen.2010.07.017. PMID:20688064.

Paul, A.J., Post, J.R., and Stelfox, J.D. 2003. Can anglers influence the abundanceof native and nonnative salmonids in a stream from the Canadian RockyMountains? N. Am. J. Fish. Manage. 23: 109–119. doi:10.1577/1548-8675(2003)023<0109:CAITAO>2.0.CO;2.

Pawson, M.G. 1991. Comparison of the performance of brown trout, Salmo truttaL., and rainbow trout, Oncorhynchus mykiss (Walbaum), in a put-and-take fish-ery. Aquacult. Res. 22: 247–257. doi:10.1111/j.1365-2109.1991.tb00514.x.

Philipp, D.P., Cooke, S.J., Claussen, J.E., Koppelman, J.B., Suski, C.D., andBurkett, D.P. 2009. Selection for vulnerability to angling in largemouth bass.Trans. Am. Fish. Soc. 138: 189–199. doi:10.1577/T06-243.1.

Phillips, B.L., and Suarez, A.V. 2012. The role of behavioural variation in theinvasion of new areas. In Behavioural responses to a changing world: mech-anisms & consequences. Edited by U. Candolin and B.B.M. Wong. Oxford Uni-versity Press, Oxford, UK. pp. 190–200.

Pigliucci, M., and Hayden, K. 2001. Phenotypic plasticity is the major determi-nant of changes in phenotypic integration in Arabidopsis. New Phytol. 152:419–430. doi:10.1046/j.0028-646X.2001.00275.x.

Pottinger, T.G., Balm, P.H.M., and Pickering, A.D. 1995. Sexual maturity modifiesthe responsiveness of the pituitary–interrenal axis to stress in male rainbowtrout. Gen. Comp. Endocrinol. 98: 311–320. doi:10.1006/gcen.1995.1073. PMID:7628690.

Réale, D., Garant, D., Humphries, M.M., Bergeron, P., Careau, V., andMontiglio, P.-O. 2010. Personality and the emergence of the pace-of-life syn-drome concept at the population level. Philos. Trans. R. Soc. B Biol. Sci. 365:4051–4063. doi:10.1098/rstb.2010.0208.

Redpath, T.D., Cooke, S.J., Suski, C.D., Arlinghaus, R., Couture, P., Wahl, D.H.,and Philipp, D.P. 2010. The metabolic and biochemical basis of vulnerabilityto recreational angling after three generations of angling-induced selectionin a teleost fish. Can. J. Fish. Aquat. Sci. 67(12): 1983–1992. doi:10.1139/F10-120.

Ruane, N.M., Wendelaar Bonga, S.E., and Balm, P.H.M. 1999. Differences betweenrainbow trout and brown trout in the regulation of the pituitary–interrenalaxis and physiological performance during confinement. Gen. Comp. Endo-crinol. 115: 210–219. doi:10.1006/gcen.1999.7292. PMID:10417234.

Schreck, C.B., and Tort, L. 2016. The concept of stress in fish. In Fish physiology.Edited by C.B. Schreck, L. Tort, A.P. Farrell, and C.J. Brauner. Academic Press.pp. 1–34.

Singh, C., Oikonomou, G., and Prober, D.A. 2015. Norepinephrine is required topromote wakefulness and for hypocretin-induced arousal in zebrafish. eLife,4: e07000. PMID:26374985.

Sundh, H., Calabrese, S., Jutfelt, F., Niklasson, L., Olsen, R.E., and Sundell, K. 2011.Translocation of infectious pancreatic necrosis virus across the intestinalepithelium of Atlantic salmon (Salmo salar L.). Aquaculture, 321: 85–92. doi:10.1016/j.aquaculture.2011.08.011.

Sundström, L.F., Petersson, E., Höjesjö, J., Johnsson, J.I., and Järvi, T. 2004. Hatch-ery selection promotes boldness in newly hatched brown trout (Salmo trutta):implications for dominance. Behav. Ecol. 15: 192–198. doi:10.1093/beheco/arg089.

Sutter, D.A.H., Suski, C.D., Philipp, D.P., Klefoth, T., Wahl, D.H., Kersten, P.,Cooke, S.J., and Arlinghaus, R. 2012. Recreational fishing selectively capturesindividuals with the highest fitness potential. Proc. Natl. Acad. Sci. U.S.A. 109:20960–20965. doi:10.1073/pnas.1212536109. PMID:23213220.

Tsuboi, J.-I., Morita, K., Klefoth, T., Endou, S., and Arlinghaus, R. 2016. Behaviour-mediated alteration of positively size-dependent vulnerability to angling inresponse to historical fishing pressure in a freshwater salmonid. Can. J. Fish.Aquat. Sci. 73(3): 461–468. doi:10.1139/cjfas-2014-0571.

Uusi-Heikkilä, S., Wolter, C., Klefoth, T., and Arlinghaus, R. 2008. A behavioral

328 Can. J. Fish. Aquat. Sci. Vol. 76, 2019

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

perspective on fishing-induced evolution. Trends Ecol. Evol. 23: 419–421.doi:10.1016/j.tree.2008.04.006. PMID:18582988.

Uusi-Heikkilä, S., Whiteley, A.R., Kuparinen, A., Matsumura, S., Venturelli, P.A.,Wolter, C., Slate, J., Primmer, C.R., Meinelt, T., Killen, S.S., Bierbach, D.,Polverino, G., Ludwig, A., and Arlinghaus, R. 2015. The evolutionary legacy ofsize-selective harvesting extends from genes to populations. Evol. Appl. 8:597–620. doi:10.1111/eva.12268. PMID:26136825.

Vainikka, A., Tammela, I., and Hyvärinen, P. 2016. Does boldness explain vulner-ability to angling in Eurasian perch Perca fluviatilis? Curr. Zool. 62: 109–115.doi:10.1093/cz/zow003. PMID:29491897.

Wendelaar Bonga, S. 1997. The stress response in fish. Physiol. Rev. 77: 591–625.doi:10.1152/physrev.1997.77.3.591. PMID:9234959.

Wilson, A.D.M., Brownscombe, J.W., Sullivan, B., Jain-Schlaepfer, S., andCooke, S.J. 2015. Does angling technique selectively target fishes based ontheir behavioural type? PLoS ONE, 10: e0135848. doi:10.1371/journal.pone.0135848. PMID:26284779.

Winberg, S., and Thörnqvist, P.-O. 2016. Role of brain serotonin in modulatingfish behavior. Curr. Zool. 62: 317–323. doi:10.1093/cz/zow037. PMID:29491919.

Winberg, S., Øverli, Ø., and Lepage, O. 2001. Suppression of aggression in rain-bow trout (Oncorhynchus mykiss) by dietary L-tryptophan. J. Exp. Biol. 204:3867–3876. PMID:11807104.

Wise, R.A. 2004. Dopamine, learning and motivation. Nat. Rev. Neurosci. 5:483–494. doi:10.1038/nrn1406. PMID:15152198.

Young, G. 1986. Cortisol secretion in vitro by the interrenal of coho salmon(Oncorhynchus kisutch) during smoltification: relationship with plasma thyrox-ine and plasma cortisol. Gen. Comp. Endocrinol. 63: 191–200. doi:10.1016/0016-6480(86)90156-5. PMID:3023179.

Závorka, L., Koeck, B., Cucherousset, J., Brijs, J., Näslund, J., Aldvén, D.,Höjesjö, J., Fleming, I.A., and Johnsson, J.I. 2017. Co-existence with non-nativebrook trout breaks down the integration of phenotypic traits in brown troutparr. Funct. Ecol. 31: 1582–1591. doi:10.1111/1365-2435.12862.

Závorka, L., Lang, I., Raffard, A., Evangelista, C., Britton, J.R., Olden, J.D., andCucherousset, J. 2018. Importance of harvest-driven trait changes for invasivespecies management. Front. Ecol. Environ. 16(6): 317–318. doi:10.1002/fee.1922.

Appendix A. Angling selects against stress-resilientand active phenotypes

Experimental pond setupTwo full-scale replicated angling experiments were successively

carried out in four semicontrolled mesocosm ponds (area = 30m ×24m = 720 m2, depth = 2 m) belonging to the Swedish anglersassociation Sportfiskarna in Gothenburg, Sweden (57.693°N,12.037°E). Prior to the experiments, vegetation and fish were re-moved from all ponds to create comparable conditions. Pondswere supplied with unfiltered lake water (Lake Delsjön) contain-ing a natural food supply for trout. Each pond was stocked withsize-matched rainbow trout (Oncorhynchus mykiss) (298 ± 19 mm)and brown trout (Salmo trutta) (281 ± 15 mm) in equal densities(25 individuals per species and pond). Both trout species originatedfrom the Källefalls hatchery (http://www.hokensas.se/en/fish-farming/kallefall/) and were reared under comparable conditions.The brown trout were F1 offspring from wild parents captured inthe nearby Lake Vättern. The rainbow trout were of a domesti-cated strain, now bred since 1997 within the hatchery and usedexclusively for stocking for angling in Swedish waters.

Angling gear and designTwo commonly used angling techniques were used for the an-

gling experiments, natural shrimp baits and spinnerbait lures, tobe representative of angler habits. Natural bait angling was car-ried out using the shrimp on a barbless hook (size 6) tackledbeneath a floater. Anglers were fishing passively with the shrimpbait by casting it and retrieving it slowly, while lure angling wasfished actively using a spinnerbait (Myrans WIPP Yellow/White,10 g; Sportsystem i Jönköping AB, Huskvarna), which was cast andretrieved at a fast speed. Despite the fact that angling is consid-ered a generally passive fishing method, to discriminate both an-gling methods in the present study, we refer to lure angling as anactive method and to bait angling as a passive method. With theexception of these two types of terminal tackles, identical anglingequipment was used for all angling experiments (Abu Garcia rod:

7 ft, power 10–40 g; braided line: resistance 10 lb; “Impact” haspelspinning reel: size 3000).

Angling experiments were carried out by experienced volun-teer anglers, who were instructed in the protocol of the experi-ments to reduce bias associated with fish handling after capture.Hooked fish were landed with a knotless net and unhooked byfishers using a plier designed for that specific purpose. Fish werethen kept in a holding tank 90 cm long × 90 cm wide × 40 cm deep)at the side of the pond until returned to their pond at the end ofthe fishing event. A fish could thus be captured only once perfishing event but recaptured at each new event. Unless conditionsrequired so, handling of fish with bare hands was avoided to limitmucus abrasion. Date, time, pond, and PIT tag number (hand-heldPIT tag reader; BTS-ID, Helsingborg, Sweden) were recorded.

In angling experiment 1, angling was practiced only in ponds 1and 3 of the four experimental ponds and consisted of 5 days ofnatural bait angling followed by 5 days of lure angling, with 2 dayintervals without fishing in between (Fig. A1a). In angling experi-ment 2, angling was performed in all four ponds with ponds 1 and3 previously exposed to angling and ponds 2 and 4 containing fishnaïve to angling (Fig. A1b). It consisted of three angling events withsimultaneous bait and lure angling to verify the results from thespecies-specific selectivity of angling technique from experiment1 while excluding the temporal effect of a successive use of an-gling technique. To spread out the angling effort evenly, anglerswere assigned to defined fishing zones within in each pond. Every10th min, anglers changed fishing zone and (or) angling gear (i.e.,artificial lure or natural bait) according to a randomization sched-ule to control for bias in fishing skills of anglers, site preference,and gear effect. At the end of the two angling experiments, ponds 1and 3 had each received a total fishing effort of 36 h (high anglingintensity treatment) and ponds 2 and 4 had received only 6 h ofangling (low angling intensity treatment).

Behavioral scoringThe scoring consisted of recording activity of individual fish in

an open-field test in white barren tanks (90 cm long × 90 cm wide ×40 cm deep) using a video camera positioned above the tanksduring 20 min following 10 min of acclimation. The automatedimage analysis software LoliTrack (4.0, Loligo Systems ApS) wasused to analyze the video records and extract an activity proxybased on the cumulated distance moved during the 20 min trial byeach individual. Fish from the same holding tank were scored atthe same time to standardize fish handling and to reduce any biaslinked to handling stress.

Correlations between individual activity and physiological stressresponses.

Figures A1–A5 appear on the following pages.

Fig. A1. Experimental treatments in the four mesocosm pondsfor (a) the first angling experiment and (b) the second anglingexperiment, where fish in ponds 1 and 3 had already experiencedangling and fish in ponds 2 and 4 were naïve to angling.

Koeck et al. 329

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

Fig. A2. Pearson’s correlation matrix for rainbow trout (Oncorhynchus mykiss) from the high-intensity angling treatment between pond activity(NbDet.Log: log-transformed number of total detections, nb.reloc: number of relocations between antennas), plasma cortisol (ng·mL−1),neurotransmitters (ng·g−1), and neurotransmitter activity levels (Dopac.DA: dopaminergic activity, 5-HIAA.5-HT: serotonergic activity) afterexposure to a standardized experimental stressor. Pearson’s r correlation coefficient and adjusted p values are given in the lower half of thematrix and correlation plots in the upper part of the matrix.

330 Can. J. Fish. Aquat. Sci. Vol. 76, 2019

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

Fig. A3. Pearson’s correlation matrix for rainbow trout (Oncorhynchus mykiss) from the low-intensity angling treatment between pond activity(NbDet.Log: log-transformed number of total detections, nb.reloc: number of relocations between antennas), plasma cortisol (ng·mL−1),neurotransmitters (ng·g−1), and neurotransmitter activity levels (Dopac.DA: dopaminergic activity, 5-HIAA.5-HT: serotonergic activity) afterexposure to a standardized experimental stressor.

Koeck et al. 331

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

Fig. A4. Pearson’s correlation matrix for brown trout (Salmo trutta) from the highiintensity angling treatment between pond activity(NbDet.Log: log-transformed number of total detections, nb.reloc: number of relocations between antennas), plasma cortisol (ng·mL−1),neurotransmitters (ng·g−1), and neurotransmitter activity levels (Dopac.DA: dopaminergic activity, 5-HIAA.5-HT: serotonergic activity) afterexposure to a standardized experimental stressor.

332 Can. J. Fish. Aquat. Sci. Vol. 76, 2019

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.

Fig. A5. Pearson’s correlation matrix for brown trout (Salmo trutta) from the low-intensity angling treatment between pond activity(NbDet.Log: log-transformed number of total detections, nb.reloc: number of relocations between antennas), plasma cortisol (ng·mL−1),neurotransmitters (ng·g−1), and neurotransmitter activity levels (Dopac.DA: dopaminergic activity, 5-HIAA.5-HT: serotonergic activity) afterexposure to a standardized experimental stressor.

Koeck et al. 333

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y L

EIB

NIT

Z-I

NST

ITU

T F

UR

on

01/2

9/19

For

pers

onal

use

onl

y.