Predation by brown trout (Salmo trutta) along adiversifying prey community gradient

Hallvard Jensen, Kimmo K. Kahilainen, Per-Arne Amundsen, Karl Øystein Gjelland,Antti Tuomaala, Tommi Malinen, and Thomas Bøhn

Abstract: Predation has a fundamental role in aquatic ecosystems, but the relative importance of factors governing preyselection by predators remains controversial. In this study, we contrast five lakes of a subarctic watershed to explore howprey community characteristics affect prey selection and growth rate of the common top predator, brown trout (Salmotrutta). The lakes constitute a distinct gradient of different coregonid prey fish, ranging from monomorphic common white-fish (Coregonus lavaretus) to polymorphic whitefish co-occurring with vendace (Coregonus albula). The brown trout wasa morph–species- and size-specific pelagic predator, selecting the small-sized, pelagic whitefish morph or vendace overthe benthic whitefish morphs. In all lakes, the average prey size increased with predator size, but small-sized prey werealso included in the diet of large predators. The selection of small-sized, pelagic prey fish appeared to be a favourable for-aging strategy for the brown trout, yielding higher growth rates and an earlier ontogenetic shift to piscivory. The findingsemphasize that piscivory appear to be shaped by the diversity, size-structure, and abundance of available prey in a givencommunity.

Resume : La predation joue un role fondamental dans les ecosystemes aquatiques, mais l’importance relative des facteursqui determinent la selection des proies par les predateurs reste un sujet controverse. Dans notre etude, nous comparonscinq lacs d’un bassin versant subarctique afin de determiner comment les caracteristiques de la communaute de proies af-fectent la selection des proies et le taux de croissance du predateur superieur commun, la truite brune (Salmo trutta). Leslacs forment un gradient marque de proies constituees de divers poissons coregonides, allant de coregones lavarets (Core-gonus lavaretus) monomorphes a des coregones lavarets polymorphes cohabitant avec des coregones blancs (Coregonus al-bula). La truite brune est un predateur pelagique qui choisit ses proies en fonction de la forme–espece et de la taille et quiprefere la forme pelagique de petite taille du lazaret et le coregone blanc au lazaret benthique. Dans tous les lacs, la taillemoyenne des proies augmente en fonction de la taille des predateurs, mais les proies de petite taille font aussi partie du re-gime des grands predateurs. La selection de poissons pelagiques de petite taille comme proies semble etre une strategiebenefique chez la truite brune, car elle permet des taux de croissance plus eleves et un passage ontogenique a l’ichtyo-phagie plus precoce. Ces resultats soulignent le fait que l’ichtyophagie semble etre determinee par la diversite, la structureen taille et l’abondance des proies disponibles dans une communaute donnee.

[Traduit par la Redaction]

Introduction

Predator–prey interactions are essential for the populationdynamics of both predator and prey (e.g., Sinclair et al.2003; Myers et al. 2007; Persson et al. 2007). Most preda-tors are selective feeders, having a diet that constitutes onlya subset of the available prey (reviewed by Stephens andKrebs 1986). The feeding selectivity of the predator is influ-enced by the relative species composition of the prey com-

munity, and predators are anticipated to select prey thatprovide an optimum energy gain relative to the time spenton foraging (Stephens and Krebs 1986; Sih and Christensen2001). Diet selection depends on characteristics of both thepredator (e.g., morphological adaptations) and the prey(e.g., energetic value, predation susceptibility), and predatorsmay furthermore exhibit functional responses where theirconsumption rate is influenced by the prey abundance(Abrams and Ginzburg 2000). Hence, several factors may

Received 2 August 2007. Accepted 25 February 2008. Published on the NRC Research Press Web site at cjfas.nrc.ca on 7 August 2008.Reposted on the Web site with corrections on 28 August 2008.J20121

H. Jensen.1 Norwegian Institute for Agricultural and Environmental Research, Arctic Agriculture and Land Use Division, N-9296 Tromsø, Norway; Department of Aquatic BioSciences, Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway.K.K. Kahilainen, A. Tuomaala, and T. Malinen. Department of Biological and Environmental Sciences, University of Helsinki, FIN-00014, Finland.P.-A. Amundsen and K.Ø. Gjelland. Department of Aquatic BioSciences, Norwegian College of Fishery Science, University ofTromsø, N-9037 Tromsø, Norway.T. Bøhn. Norwegian Institute of Gene Ecology, The Science Park, N-9294 Tromsø, Norway.

1Corresponding author (e-mail: hallvard.jensen@bioforsk.no).

1831

Can. J. Fish. Aquat. Sci. 65: 1831–1841 (2008) doi:10.1139/F08-096 # 2008 NRC Canada

contribute to the feeding patterns of a predator, and potentialoutcomes may range from predators feeding randomly topredators that feed highly selectively on specific prey (Eg-gers 1977; Sih and Moore 1990). Although these basicmechanisms of predator–prey interactions are of major inter-est in aquatic ecology (reviewed in Werner and Gilliam1984; Sih and Christensen 2001), they are often difficult tostudy comprehensively in natural communities (Rough-garden 1986; Kramer et al. 1997).

In lakes, predation by piscivorous fish may affect the preyfish communities directly (i.e., by reducing prey fish densityor altering size structure; Persson et al. 1996) or indirectlyby determining prey fish behaviour with respect to foraging,habitat use, shoaling, diurnal activity patterns, or growth(Turner and Mittelbach 1990; Lima 1998). Habitat utiliza-tion is an important factor in predator–prey interactions(Persson and Greenberg 1990; Bystrom et al. 2003) and, incontrast to the open-water pelagic habitat, the littoral andprofundal zones of lakes may provide refuges for prey fishin terms of structural complexity and darkness, respectively.Predator–prey relationships are also highly size dependent(Juanes 1994; Persson et al. 1996; Scharf et al. 2000). Be-cause of gape-size limitations, predatory fish usuallyexperience an ontogenetic diet shift from invertebrates tofish at a certain size depending on availability, size, andspecies-specific characteristics of the prey (Werner andGilliam 1984; Mittelbach and Persson 1998). Some preyfish may, on the other hand, be able to grow out of the ‘‘pre-dation window’’ (i.e., the size range where they are vulner-able to predation; Claessen et al. 2002). Different preyspecies may furthermore have different growth trajectories,leading to different time frames of predation susceptibility.However, studies combining field data on prey diversity,size structure, and abundance with prey utilization of toppredators are scarce.

Subarctic lakes in northern Europe usually represent rela-tively pristine, low-diversity ecosystems often dominated bysalmonid fish. Brown trout (Salmo trutta) is a common top-predator in many of these lakes, particularly in systems do-minated by two closely related coregonid fish (commonwhitefish (Coregonus lavaretus) and (or) vendace (Core-gonus albula)), where the brown trout may reach a largebody size through piscivory (Næsje et al. 1998; Vehanen etal. 1998). The coregonids in these northern lakes exhibit ex-tensive polymorphism and highly variable life histories (Ka-hilainen et al. 2003; Kahilainen et al. 2005; Amundsen et al.2004a) and likely constitute heterogeneous prey assemblagesfor the piscivorous brown trout, potentially resulting in con-trasting predator–prey interactions.

This study compares the predator–prey relationships ofbrown trout and coregonids between five lakes of the sub-arctic Paatsjoki–Pasvik watershed in northern Europe. Threewhitefish morphs have been identified in this watershed(Amundsen et al. 2004b; Kahilainen and Østbye 2006, Øst-bye et al. 2006), and vendace is also present as an intro-duced, non-native species in the lower reaches (Amundsenet al. 1999). The five lakes therefore comprise a gradient ofdifferent coregonid prey communities for the brown trout,including one lake with a single whitefish morph present,two lakes inhabited by three sympatric morphs, and twolakes where the whitefish morphs co-occur with vendace.

These study lakes in the same watershed thus enable com-parisons of predator–prey interactions along a gradient of in-creasing diversity of coregonid prey in otherwise similarlakes. The following hypotheses were tested: (i) the predator(i.e., brown trout) preys selectively on coregonids and pre-fers the habitat with the most profitable prey community;(ii) the predator switches to piscivory at smaller sizes if theprey community includes abundant, small-sized prey; (iii)prey size selection is a positive function of predator size;and (iv) predator growth rate is positively correlated withthe density of potential prey in the environment.

Materials and methods

Study sites and fish communitiesThe headwaters of the Paatsjoki–Pasvik watershed (688–

698N, 268–308E) discharge into Lake Inari (surface area1102 km2) in northern Finland, run throughout Russia for30 km, and finally form the border between Norway andRussia for a distance of approximately 120 km before enter-ing the Arctic Ocean (Fig. 1). The total catchment area ofthe watershed is 18 403 km2. The five lakes in the presentstudy are Lake Vuontisjarvi (referred to as Lake 1), LakeMuddusjarvi (Lake 2), and Lake Paadar (Lake 3) on the Fin-nish side and Lake Skrukkebukta (Lake 4) and Lake Vagga-tem (Lake 5) on the Norwegian side. The surface area of thelakes varies from 7 to 48 km2, with maximum and meandepths of 30–73 m and 6.5–14 m, respectively (Table 1).All the lakes are oligotrophic (total phosphorus (totP) = 6–9 mg�L–1, total nitrogen (totN) = 145–170 mg�L–1), neutral(pH of 6.8–7.2), and with some humic impacts (Secchidepths between 3 and 8 m). The ice-free season in the lakeslasts from May–June to October–November.

All study lakes have a coregonid-dominated fish fauna.Beside coregonids and brown trout, the most common fishspecies include European perch (hereafter referred to asperch; Perca fluviatilis), northern pike (Esox lucius), burbot(Lota lota), grayling (Thymallus thymallus), ninespinestickleback (Pungitius pungitius), and Eurasian minnow(Phoxinus phoxinus) (Table 1). Brown trout is the mostabundant salmonid predator in all lakes, consisting of bothstocked (>60% of the catches in all lakes) and native fish(Kahilainen and Lehtonen 2002; Jensen et al. 2004). Stockedand native fish are approximately equal in size (±20 cm) atthe start of the first growing season in the lake during earlysummer (Kahilainen & Lehtonen 2001). In the presentstudy, the native and stocked brown trout have been pooledin the analyses, as they have similar diet preferences andgrowth patterns (Kahilainen and Lehtonen 2001; Jensen etal. 2004).

The whitefish occur as three morphologically and ecolog-ically distinct morphs. Adaptive radiation and incipient eco-logical speciation is a likely explanation for the adaptiveevolution of these whitefish morphs (Østbye et al. 2006),and in this study the morphs are considered as different bio-logical species. In allopatry, the large sparsely rakered(LSR) whitefish has a wide niche utilizing all principal hab-itat types and prey (Amundsen et al. 2004a; Kahilainen etal. 2007). In sympatry, each whitefish morph is specializedto one principal niche, including a profundal-dwelling ben-thivore (i.e., the small sparsely rakered whitefish (SSR

1832 Can. J. Fish. Aquat. Sci. Vol. 65, 2008

# 2008 NRC Canada

whitefish)), a littoral-dwelling benthivore (i.e., LSR white-fish), and a pelagic planktivore (i.e., the densely rakeredwhitefish morph (DR whitefish)) (Amundsen et al. 1999,2004a; Kahilainen et al. 2004). Lake 1 is inhabited by a sin-gle LSR morph, whereas all three whitefish morphs (SSR,LSR, and DR) occur in sympatry in Lakes 2 and 3 (Kahi-lainen et al. 2004). In Lakes 4 and 5, the sympatric white-fish morphs co-occur with an introduced non-nativecoregonid species, the vendace, which has become a highlyabundant pelagic species after invading these lakes in theearly 1990s (Amundsen et al. 1999; Bøhn and Amundsen2001, Bøhn et al. 2004). The somatic growth rates differ be-tween the whitefish morphs, being highest in LSR whitefish,intermediate to low in the DR whitefish, and lowest in the

SSR whitefish (Kahilainen and Lehtonen 2003; Bøhn andAmundsen 2004). The vendace in Lakes 4 and 5 have veryslow growth rates, attaining maximum sizes well below15 cm (Bøhn et al. 2004).

Field samplingA total of 2430 brown trout were sampled throughout

June to October in 1998–2004 by gill nets and fishing rods(sampling details in Kahilainen and Lehtonen 2003; Jensenet al. 2004, 2006). The total lengths and weights of the fishwere measured to an accuracy of 1 mm and 1 g, respec-tively, and stomachs were removed and frozen (–20 8C) forfurther analysis. In addition, scales and otoliths were takenfor age determination.

Table 1. Abiotic and biotic background data from the five study lakes.

Parameter Lake 1 Lake 2 Lake 3 Lake 4 Lake 5Latitude (8N) 69801’ 69800’ 68852’ 69833’ 69814’Longitude (8E) 27804’ 26850’ 26835’ 30807’ 29815’Country Finland Finland Finland Norway NorwaySurface area (km2) 11 48 21 7 15Altitude (m above sea level) 151 146 144 21 52Maximum depth (m) 31 73 56 38 30Mean depth (m) 6.5 8.5* 11.7 14 4Secchi depth (m) 8 3 6* 4–5.5 3–4.5Color (mg Pt�L–1) 8* 15* 15* 16 17pH 7.2* 7.2* 7.1* 6.9 6.8Total P (mg�L–1) 7* 5* 6* 7 9Total N (mg�L–1) 170* 160* 160* 156 145Coregonid species and morphs LSR LSR, SSR, DR LSR, SSR, DR LSR, DR, VEN LSR, DR, VENOther fish species b, c, d, e, f, g, h a, b, c, d, e, f, g, h b, c, d, e, f, g, h b, c, e, f, g, h b, c, e, f, g, h

Note: Other fish species present in the study lakes in addition to brown trout are indicated with abbreviations: LSR, large sparsely rakered whitefish(Coregonus lavaretus); SSR, small sparsely rakered whitefish, DR, densely rakered whitefish; VEN, vendace (Coregonus albula); a, Arctic char (Salve-linus alpinus); b, grayling (Thymallus thymallus); c, Eurasian minnow (Phoxinus phoxinus); d, threespine stickleback (Gasterosteus aculeatus);e, ninespine stickleback (Pungitius pungitius); f, European perch (Perca fluviatilis); g, northern pike (Esox lucius); h, burbot (Lota lota).*Data from Lapland Regional Environment Centre, P.O. Box 8060, FIN-96101 Rovaniemi, Finland.

Fig. 1. (a) Map of Northern Europe showing the location of the Paatsjoki–Pasvik watershed. (b) Map of the Paatsjoki–Pasvik watershedindicating the location of study Lakes 1–5.

Jensen et al. 1833

# 2008 NRC Canada

Prey fish samples were collected from the study lakesduring September in 1998–2004. For prey fish sampling inLakes 1–3, gillnet series with mesh sizes (knot to knot) of12, 15, 20, 25, 30, 35, 45, and 60 mm were used in the lit-toral and profundal habitats, and a small pair trawl (5 mhigh, 8 m wide, and cod-end mesh size 3 mm) was used inthe pelagic habitat (details in Kahilainen et al. 2004). InLakes 4–5, prey fish were caught in the littoral and profun-dal habitats using bottom gillnet series with mesh sizes of10, 12.5, 15, 18.5, 22, 26, 35, and 45 mm; in the pelagichabitat, floating gillnet series with mesh sizes of 8, 10,12.5, 15, 18.5, 22, 26, and 35 mm were used.

Prey fish were identified to species and whitefish tomorph according to head and gillraker morphology(Amundsen et al. 2004a; Kahilainen and Østbye 2006). Thetotal lengths and weights of the prey fish were determined toan accuracy of 1 mm and 1 g, respectively. Habitat use ofcoregonids was evaluated using catch per unit effort (CPUEbased on number of caught fish) of different morphs–speciesin the littoral, profundal, and pelagic habitats (details in Ka-hilainen et al. 2004; Amundsen et al. 2004a). Coregonidswere the dominant fish species in all the study lakes (pro-portion in catches > 80% in every lake), and the subsequentanalyses were concentrated only on this prey resource.

To assess the density of pelagic coregonids, areas deeperthan 6 m were sampled using a SIMRAD EY-500echosounder equipped with downfacing split-beam trans-ducers operating at 120 kHz frequency (ES120-7F in Lakes1–3 and ES120-4 � 10 in Lakes 4 and 5). Transects wereplaced equidistantly in Lakes 1–3 (Kahilainen et al. 2004),whereas a combination of zigzag and parallel transects wasapplied in Lakes 4 and 5. Details of equipment and settingsare described in Kahilainen et al. (2004).

Laboratory and data analysesThe prey items in brown trout stomachs were identified as

far as possible. Prey fishes were identified to species by theremaining external features, and whitefish were identified tomorph by gillraker examination (Kahilainen and Lehtonen2002; Amundsen et al. 2004b). Fish prey other than corego-nids (e.g., ninespine sticklebacks, perch, Arctic char, andburbot) were pooled as other fish, whereas aquatic insects(including Ephemeroptera, Trichoptera, Plecoptera, Odonata,Chironomidae, and Coleoptera) were pooled as inverte-brates. Prey abundance (Ai, percent volume), or the percent-age of each diet category in the stomachs (sum of allcategories = 100%), was calculated as follows:

Ai ¼X

Si=X

Stot � 100

where Si is fullness for diet category i, and Stot is the totalstomach fullness (Amundsen et al. 1996).

The similarity in coregonid composition between thebrown trout stomach contents and the proportional corego-nid CPUEs from different habitats (littoral, profundal, pela-gic) was calculated using Schoener’s (1970) index:

� ¼ 1� 0:5Xni¼1

jPxi � Pyij !

where Pxi is the proportion of coregonid species–morph i

consumed by brown trout population x, Pyi is the proportionof coregonid species–morph i dwelling in habitat y, and n isthe number of coregonid taxa in the lake. The index wasalso used to compare the similarity in brown trout diets be-tween the lakes. The index gives a values from 0 to 1,where 0.00 and 1.00 indicates no overlap and completeoverlap, respectively. An index value ‡ 0.60 is consideredto represent a biologically significant similarity (Wallace1981).

The total length of undigested prey fish in the brown troutstomachs were measured to an accuracy of 1 mm, and thelength of partially or entirely digested coregonid prey wereestimated from the otolith length (Kahilainen and Lehtonen2001). The average coregonid prey length (log-transformed)in brown trout stomachs was compared statistically betweenthe lakes using analysis of covariance (ANCOVA) withpredator length as covariate. Pairwise comparisons weremade with Tukey’s honestly significant difference (HSD)test. The relationship between coregonid prey and browntrout length was furthermore estimated using linear regres-sion analysis for each lake separately. To develop a generalmodel of the predator–prey length relationship betweenbrown trout and coregonids in the studied lakes, data werepooled from all lakes and we performed a quantile regres-sion analysis, estimating the median and upper (99th quan-tile) and lower (1st quantile) bounds of the relationship(Cade et al. 1999).

Brown trout age was determined from both scales andotoliths in Lakes 1–3 and from otoliths in Lakes 4–5. Thesomatic growth rate of brown trout was measured by the an-nual specific growth rate (G, year–1):

G ¼ 100ðln W2 � ln W1Þðt2 � t1Þ

where W2 and W1 are average body weights in age classes t2and t1, respectively. G was estimated between different ageclasses for wild fish and between years after stocking forstocked fish. G values of the same age class in differentlakes were compared with analysis of variance (ANOVA),and pairwise comparisons between age classes were madewith Tukey’s HSD test.

Hydroacoustic data were analyzed using EP500 (Lakes 1–3) and Sonar5 (Lakes 4 and 5) postprocessing software. Theanalysis was started at a depth of 3 m and stopped 0.5 mabove the bottom. Integration threshold was –60 dB for alllakes, and all targets were assumed to be coregonids, withthe exception of very small fish targets, which were likelyninespine or threespine sticklebacks. These were excludedfrom the coregonid density estimates by setting the targetstrength (TS) thresholds between –54 and –59 dB based onTS distributions. The coregonid density of each transect wascomputed using observed TS distributions (for details seeKahilainen et al. 2004). The mean density in each lake andthe variance of the means were computed by taking variabletransects length into account (Shotton and Bazigos 1984),and 95% confidence limits were estimated assuming a Pois-son distribution (Jolly and Hampton 1990).

Results

Predator diet selectionThe brown trout were mainly piscivorous in all lakes. In

1834 Can. J. Fish. Aquat. Sci. Vol. 65, 2008

# 2008 NRC Canada

Lakes 1 and 2, an ontogenetic diet shift from invertebratesto fish was observed with increasing brown trout length, butin Lakes 3–5 all examined length groups were almost exclu-sively piscivorous (Fig. 2). In all lakes, coregonids were thedominant fish prey for the brown trout, and other fish (nine-spine stickleback, perch, burbot, and Arctic char) were onlyoccasionally eaten. In Lake 1, brown trout fed on the onlyavailable whitefish morph (LSR), whereas the diet shiftedto the pelagic DR whitefish in Lakes 2 and 3 with the threesympatric morphs present. In Lake 4, brown trout preyed onDR whitefish and vendace, whereas the predation shifted al-most exclusively to vendace in Lake 5 (Fig. 2). The profun-dal SSR whitefish was not found in any of the brown troutstomachs.

According to Schoener’s index (a), the similarity betweenthe coregonid proportions in the predator diet and the fishcatches differed between habitats (except in Lake 1, whereLSR whitefish was the only coregonid present). In Lakes 2–

5, the highest similarity (a = 0.80–0.96) was consistentlyfound for the pelagic habitat (Table 2). The similarity wassomewhat lower for the profundal habitat in Lakes 2–4 (a =0.51–0.76) and was generally the lowest for the littoral hab-itat of the lakes (a = 0.20–0.60). The brown trout diet simi-larity between lakes was biologically significant in twocases: between Lakes 2 and 3 (a = 0.99) and between Lakes4 and 5 (a = 0.69) (Table 2, Fig. 3). In the other compari-sons, the brown trout diets were less similar (a < 0.60).

Predator diet versus prey diversity, size distribution, andabundance

The average coregonid prey length in the brown troutstomachs was significantly different between the lakes andshowed a decreasing trend from Lake 1 towards Lake 5(Fig. 3a), irrespective of the predator length (ANCOVA,F[4,1793] = 129.4, P < 0.0001). The prey length further dif-fered significantly in all pairwise comparisons between thelakes (Tukey’s HSD tests, P < 0.001), except between Lakes1 and 2 (P > 0.05). In Lake 1, the brown trout stomachs in-cluded prey fish larger than those observed in the pelagichabitat, suggesting that feeding to a large extent occurred inthe benthic habitats. Accordingly, the pelagic fish density inLake 1 was the lowest of all the study lakes. In Lakes 2–5,the brown trout tended to select similarly sized (Lakes 2–3)or slightly smaller (Lakes 4–5) coregonid prey than thoseobserved in the catches (Fig. 3). This difference in prey sizeis likely due to selectivity differences between gill net andpelagic trawl.

There was furthermore a high similarity between the

Fig. 2. Diet composition of different total length groups of brown trout (Salmo trutta) from the studied Lakes 1–5. The number of examinedstomachs containing prey items is indicated above the bars. LSR, large sparsely rakered whitefish (Coregonus lavaretus); DR, densely ra-kered whitefish.

Table 2. Similarity (Schoener’s a) between pooled brown troutstomach contents and proportional coregonid catch per unit ef-fort (CPUE) in different habitats of the study lakes.

Habitat Lake 1 Lake 2 Lake 3 Lake 4 Lake 5Littoral 1 0.60 0.53 0.30 0.20Profundal 1 0.76 0.75 0.62 0.51Pelagic 1 0.95 0.96 0.80 0.93

Note: Values ‡ 0.60 (shown in bold) are considered to represent bio-logically significant similarities (except Lake 1 with only one prey typepresent).

Jensen et al. 1835

# 2008 NRC Canada

Fig. 3. (a) Coregonid length distribution in the pelagic catches and in brown trout (Salmo trutta) diet in study lakes. Number of samples (n)and the average total length (TL) are also indicated. (b) Proportion of different coregonids in pelagic catch per unit effort (CPUE; uppercircle) and brown trout diet (lower circle) for each lake. (c) Coregonid density with 95% confidence limits (upper and lower bound valuesindicated) in pelagic areas (depth > 6 m) estimated with echosounding.

1836 Can. J. Fish. Aquat. Sci. Vol. 65, 2008

# 2008 NRC Canada

brown trout diet and the available coregonid prey in the pe-lagic habitat in all the lakes (Fig. 3b, Table 2), and the dietsgenerally reflected the differences in prey communities be-tween lakes. In Lake 1, only LSR whitefish was available;in Lakes 2 and 3, DR whitefish dominated strongly; and inLakes 4 and 5, DR whitefish co-occurred with vendace. Ac-cordingly, the lakes divided into three distinct groups interms of the brown trout diet: (1) allopatric LSR was thesingle prey species in Lake 1, (2) DR whitefish dominatedthe diet in Lakes 2–3, and (3) vendace and DR whitefishwere the most important prey in Lakes 4–5. The meandensity of pelagic coregonid prey increased from group 1(10 individuals�ha–1) to group 2 (640–1180 individuals�ha–1)and peaked in group 3 (2640–2690 individuals�ha–1)(Fig. 3c). Differences between the three groups were statisti-cally significant according to the confidence limits of thedensity estimates.

Predator–prey size relationshipWithin all lakes, the length of coregonid fish prey in-

creased significantly with predator size, but the slope of theregression was subject to large between-lake variations(Fig. 4, Table 3). Prey length increased most rapidly withpredator length in Lake 1, where the abundance and diver-

sity of coregonid prey was the lowest (Figs. 3 and 4). Withthe exception of Lake 4, the slopes generally decreased withincreasing abundance of small pelagic coregonid prey, andthe lowest slope was observed in Lake 5 (Figs. 3 and 4, Ta-ble 3).

In the quantile regressions of the predator–prey length re-lationship, the median and lower and upper bound slopes allincreased significantly with increasing size of the predator(Fig. 4b). The range of prey sizes consumed expanded withincreasing predator size, as the upper bound exhibited asteeper slope than the lower one. The maximum, median,and minimum prey sizes of a 40 cm predator were estimatedto be approximately 16, 10, and 5 cm (i.e., 40%, 25%, and12%, respectively) of the predator size. For all three param-eters, the relative prey length decreased with increasingpredator size. The slope of the lower bound (i.e., the 1%quantile) was very low, suggesting a continuum of small-sized coregonid prey in the diet even when the predator at-tained considerably large sizes.

Predator growth rateThe annual specific growth rate (G) of brown trout during

the first lake-year (age class 1) differed significantly be-tween the lakes (ANOVA, F[4,528] = 9.7, P < 0.0001) andwas significantly lower in Lake 1 (G = 48.9 g�year–1) com-pared with the other lakes (range 80.9–93.9 g�year–1; Tu-key’s HSD tests, P < 0.0001) (Fig. 5). Furthermore, therewere no significant differences in growth rates between thethree age classes of brown trout in Lake 1 (ANOVA,F[2,59] = 1.3, P = 0.274), whereas there was a significant de-crease in G values of age classes 2 and 3 compared with ageclass 1 in Lakes 2–5 (Tukey’s HSD tests, P < 0.0001).When comparing the growth rates of age classes 2 and 3,no significant differences were found between the lakes(ANOVA: F[4,216] = 2.1, P = 0.088 and F[4,112] = 0.3, P =0.902, respectively).

Discussion

Individuals of a variety of predator taxa are known toshift habitat and diet to increase foraging gain, and theseshifts may be a result of different availability of prey resour-ces and ontogenetic changes in resource use (Werner andGilliam 1984; Stephens and Krebs 1986). The present studyexamined different size groups of a predator feeding in fivedifferent prey communities. Brown trout larger than 20 cmwere highly piscivorous in all the lakes, feeding almost ex-clusively on coregonid prey fish. A minor exception was

Fig. 4. (a-1 to a-5) Predator–prey total length relationships in Lakes1–5 estimated with linear regression analysis. Solid line indicatesregression line and broken lines indicate 95% confidence limits.(b) Quantile regression of the pooled predator–prey size relation-ships with median (solid line) and 1% and 99% quantiles (brokenlines). The estimated slopes from the quantile regression were 0.12(median), 0.17 (upper bound, 99% quantile), and 0.07 (lowerbound, 1% quantile) (P < 0.001).

Table 3. Estimated parameters from linear regressions of thepredator–prey total length relationships in the studied lakes.

Lake nSlope(±95% CL)

Intercept(±95% CL) r2 P

1 119 0.23±0.11 2.52±4.55 0.12 <0.0012 677 0.13±0.02 6.38±1.02 0.14 <0.0013 202 0.06±0.03 5.68±1.28 0.06 <0.0014 364 0.13±0.05 4.11±1.94 0.07 <0.0015 437 0.03±0.02 8.66±0.96 0.02 <0.011–5 1799 0.11±0.02 5.84±0.72 0.08 <0.001

Note: CL, confidence limit.

Jensen et al. 1837

# 2008 NRC Canada

represented by the inclusion of invertebrates in the smallestlength groups in Lakes 1 and 2. The switch to piscivory ingape-limited predators depends on the availability of small-sized prey fish (e.g., Juanes 1994; Mittelbach and Persson1998), and the delayed switch to complete piscivory ob-served in Lakes 1 and 2 was most likely related to a lowabundance of small-sized pelagic prey, particularly in Lake1. In contrast, the brown trout switched earlier to piscivoryin Lakes 3–5 with high abundances of pelagic prey avail-able. Our findings are equivalent to the general observationsof an ontogenetic niche shift to piscivory at lengths of 20–30 cm in salmonids (Keeley and Grant 2001) and further-more support the interpretation of pelagic fish as highly pre-ferred prey for piscivorous brown trout.

In general, the foraging patterns were changing from theconsumption of a monomorphic and mainly benthic-dwellingcoregonid (LSR whitefish in Lake 1) towards the pelagicspecialist vendace dominating the diet in Lake 5. This re-flects the selective predation by the brown trout. We usedSchoener’s similarity index to explore the main feedinghabitat of the brown trout, and as the highest similarity be-tween the coregonid composition in the environment and inthe pooled brown trout diets were found for the pelagichabitat, our data support the conclusion that the browntrout predominantly feeds in the open waters of the lakes.This is also supported by the fact that ninespine stickle-backs and minnows, which are common prey for benthicpredators like perch, northern pike, and burbot in thewatershed (Amundsen et al. 2003), rarely were observedin the brown trout stomachs. Furthermore, despite the pres-ence of LSR whitefish in all lakes, the piscivorous browntrout in Lakes 2–5 were feeding almost exclusively on DRwhitefish and vendace (i.e., the species that predominantlyoccupied the pelagic habitat of these lakes). In Lake 1 incontrast, the brown trout utilized prey fish larger thanthose observed in the pelagic habitat, suggesting that thepredator chose to forage on larger prey in the benthic hab-itat when the abundance of small, pelagic whitefish waslow. Foraging of piscivorous brown trout has been reportedelsewhere to occur in both littoral (e.g., Næsje et al. 1998)and pelagic habitats (e.g., Hyvarinen and Huusko 2006).Our results suggest that the brown trout is a piscivore thatmay utilize different coregonid prey species in differenthabitats, but selectively switches to feed on relativelysmall-sized, pelagic DR whitefish or vendace when theseare abundant.

For potential prey species, habitat selection is a trade-off

between optimal foraging and predator avoidance behavior(Lima and Dill 1990). In many systems, the most profitablehabitat for feeding is also considered to have the highest riskof predation (Werner and Gilliam 1984; Bystrom et al.2003). This is obviously the case for DR whitefish and ven-dace feeding on zooplankton in the pelagic habitat (Bøhnand Amundsen 2001; Kahilainen et al. 2007), which is alsothe prime feeding habitat for the piscivorous brown trout(Kahilainen and Lehtonen 2002). The profundal zone is ap-parently the least profitable foraging habitat in these lakesand is mainly inhabited by the slow-growing SSR whitefishmorph. The low light levels in this habitat provide a goodprey refuge from brown trout, and SSR whitefish were neverfound in the brown trout stomachs. This is in accordancewith the general description of brown trout as a visual pred-ator, relying upon good light conditions for efficient preda-tion (Langeland et al. 1991; Schulz and Berg 1992).

A positive relationship between body size of predator andprey has been recognized in a large number of animal taxa(Werner and Gilliam 1984; Wootton 1998; Sinclair et al.2003). Whereas piscivorous fish are able to ingest prey fishup to approximately 50% of their own length (Juanes 1994),the prey–predator size ratio of 20%–30% found in thepresent study was close to the average ratios found in otherstudies of piscivores (Hoyle and Keast 1987; Hambright etal. 1991; Mittelbach and Persson 1998). Our results show apositive relationship between predator size and prey size inall lakes, but the strength of the correlation was dependenton species composition, size structure, and abundance of theprey. The steepest incline in prey size with increasing pred-ator size was seen in Lake 1 (slope = 0.23) where allopatricLSR whitefish was the only coregonid prey fish present,whereas the slope decreased to almost zero in Lake 5, whichwas dominated by a small-sized, high-abundant vendacepopulation (slope = 0.03). Lake 4 represented an exceptionin this general trend of predator–prey length slopes, obvi-ously related to a more pronounced availability of larger-sized (>13 cm) pelagic prey than in Lakes 3 and 5.

In Lake 1, the growth rate of brown trout during the firstyear in the lake appeared to be limited by the low availabil-ity of small-sized prey as compared with the other lakes.The growth rates of larger age classes of the predatorseemed in contrast not to be limited by the prey size distri-bution in any of the lakes, as the growth rates of these ageclasses were similar between all lakes. Thus, although preysize increased with increasing predator size, large predatorswere not growth-limited by the absence of large prey. First

Fig. 5. Annual specific growth rate (G, year–1) of brown trout (Salmo trutta) with 95% confidence limits in Lakes 1–5.

1838 Can. J. Fish. Aquat. Sci. Vol. 65, 2008

# 2008 NRC Canada

year growth and the ontogenetic switch to piscivory were onthe other hand clearly enhanced by high abundance ofsmall-sized prey fish. Energetically, a positive correlationbetween predator and prey size is expected to occur, sincelarger prey contain more energy than smaller prey, but thismay be counteracted by increased energetic costs of search-ing, pursuing, and handling and also lower capture rates oflarger-sized prey (Townsend and Winfield 1985; Crawleyand Krebs 1992; Sih and Christensen 2001). When combin-ing all lakes, the quantile regression showed a moderate,positive correlation between brown trout and coregonid preysize. Although this shows that larger predators include largerprey in their diet, the emerging pattern is that the size rangeof utilized prey broadens with predator size and that rela-tively small prey are continuously included in the diet. Fur-thermore, the size of the coregonid prey wasgenerally <15 cm, suggesting that coregonids up to this sizeare most vulnerable to predation and thus are inside the pre-dation window of brown trout (Bøhn et al. 2002; Kahilainenand Lehtonen 2003), which apparently is quite narrow.

To maximize foraging efficiency and growth, a predatorshould select the most abundant and available prey resources(i.e., Elliott and Hurley 2000). The prey fish density in thepelagic habitat was lowest in the single-morph whitefishlake (Lake 1) and increased markedly towards the vendace-dominated lake (Lake 5). In Lakes 2 and 3, DR whitefishdominated the pelagic habitat at medium densities (see alsoKahilainen et al. 2004, 2005), whereas the highest pelagicprey densities were observed in the vendace-invaded Lakes4 and 5. Our study showed that the ontogenetic shift to pis-civory and the growth rate of brown trout during the firstyear in the lake was positively related to the abundance ofpelagic prey fish and also demonstrated the energetic profit-ability of pelagic foraging of the piscivorous trout. The ap-parent profitability of selecting small-sized fish prey furthersuggests that the impact of such a piscivore feeding strategymay be greater than earlier assumed (Miller et al. 1988) andcould have major effects on prey morphology (Bronmarkand Miner 1992) and community structure (Hambright et al.1991; Tonn and Paskowski 1992; Bystrom et al. 2003).Thus, high densities of small-sized prey, in particular ven-dace and DR whitefish, resulted in early shifts to piscivory,whereas low densities and higher growth rates of the avail-able prey fish apparently delayed the ontogenetic shift topiscivory and reduced the predator growth.

In conclusion, our study demonstrates that brown trout isa habitat-, species-, and size-specific piscivore. The small-sized pelagic prey (DR whitefish and vendace) were consis-tently selected over the larger, benthic, coregonid morph. Aweak, positive correlation was observed between predatorand prey length, but the continuous inclusion of small preysuggests that this may represent favourable prey even forlarge predators. Brown trout furthermore switched to pisci-vory earlier and had a higher growth rate in lakes withsmall-sized pelagic prey present. In the absence of small-sized pelagic prey species like DR whitefish and vendace,the brown trout in contrast shifted to larger fish prey and ap-parently also to feeding in the littoral habitat. Hence, thebrown trout is apparently also a flexible fish predator, beingable to change feeding habitat and prey selection in accord-ance with the availability of suitable prey.

AcknowledgementsThe authors thank J. Malcolm Elliott for valuable com-

ments on the manuscript and P.E. Aspholm, B. Danielsen,L. Dalsbø, J. Evjen, K. Maenpaa, O. Aikio, J. Pohtila, J.Marttila, S. Vatanen, K. Walden, V. Leppaniemi, T. Linnan-saari, M. Salonen, J. Saren, P. Jaaskelainen, J. Niemisto, andK. Johansson for helpful assistance during the field and lab-oratory work. Financial support was kindly received fromJenny and Antti Wihuri Foundation, Ministry of Agricultureand Forestry, Municipality of Inari, Otto A. Malm Founda-tion, Lapland and Finnish Cultural Foundation, The Norwe-gian Research Council, Norwegian Directorate for NatureManagement, The County Governor of Finnmark, PasvikKraft AS, and the Norwegian Institute for Agricultural andEnvironmental Research.

ReferencesAbrams, P.A., and Ginzburg, L.R. 2000. The nature of predation:

prey dependent, ratio dependent or neither? Trends Ecol. Evol.15: 337–341. doi:10.1016/S0169-5347(00)01908-X. PMID:10884706.

Amundsen, P.-A., Gabler, H.-M., and Staldvik, F.J. 1996. A newapproach to graphical analysis of feeding strategy from stomachcontents data—modification of the Costello (1990) method. J.Fish Biol. 48: 607–614.

Amundsen, P.-A., Staldvik, F.J., Reshetnikov, Y.S., Kashulin, N., Lu-kin, A., Bøhn, T., Sandlund, O.T., and Popova, O.A. 1999. Invasionof vendace Coregonus albula in a subarctic watercourse. Biol. Con-serv. 88: 405–413. doi:10.1016/S0006-3207(98)00110-4.

Amundsen, P.-A., Bøhn, T., Popova, O.A., Staldvik, F.J., Reshetni-kov, Y.S., Kashulin, N.A., and Lukin, A.A. 2003. Ontogeneticniche shifts and resource partitioning in a subarctic piscivorefish guild. Hydrobiologia, 497: 109–119. doi:10.1023/A:1025465705717.

Amundsen, P.-A., Knudsen, R., Klemetsen, A., and Kristoffersen,R. 2004a. Resource competition and interactive segregation be-tween sympatric whitefish morphs. Ann. Zool. Fenn. 41: 301–307.

Amundsen, P.-A., Bøhn, T., and Vaga, G.H. 2004b. Gill raker mor-phology and feeding ecology of two sympatric morphs of Eur-opean whitefish (Coregonus lavaretus). Ann. Zool. Fenn. 41:291–300.

Bøhn, T., and Amundsen, P.-A. 2001. The competitive edge of aninvading specialist. Ecology, 82: 2150–2163.

Bøhn, T., Amundsen, P.-A., Popova, O., Reshetnikov, Y.S., andStaldvik, F.J. 2002. Predator avoidance by coregonids: Can habi-tat choice be explained by size-related prey vulnerability? Arch.Hydrobiol. Spec. Issues Adv. Limnol. 57: 183–197.

Bøhn, T., and Amundsen, P.-A. 2004. Invasion-mediated life his-tory changes in a dimorph whitefish Coregonus lavaretus popu-lation. Ann. Zool. Fenn. 41: 125–136.

Bøhn, T., Sandlund, O.T., Amundsen, P.-A., and Primicerio, R.2004. Rapidly changing life history during a fish invasion. Oi-kos, 106: 138–150. doi:10.1111/j.0030-1299.2004.13022.x.

Bronmark, C., and Miner, J.G. 1992. Predator-induced phenotypicalchange in body morphology in crucian carp. Science (Washington,D.C.), 258: 1348–1350. doi:10.1126/science.258.5086.1348.PMID:17778362.

Bystrom, P., Persson, L., Wahlstrom, E., and Westman, E. 2003.Size- and density-dependent habitat use in predators: conse-quences for habitat shifts in young fish. J. Anim. Ecol. 72: 156–168. doi:10.1046/j.1365-2656.2003.00681.x.

Jensen et al. 1839

# 2008 NRC Canada

Cade, B.S., Terrel, J.W., and Schroeder, R.L. 1999. Estimating ef-fects of limiting factors with regression quantiles. Ecology, 80:311–323.

Claessen, D., Van Oss, C., De Roos, A.M., and Persson, L. 2002.The impact of size-dependent predation on population dynamicsand individual life history. Ecology, 83: 1660–1675.

Crawley, M.J., and Krebs, J.R. 1992. Foraging theory. In Naturalenemies. The population biology of predators. Parasites and dis-eases. Edited by M.J. Crawley. Oxford University Press, Oxford,UK. pp. 90–114.

Eggers, D.M. 1977. The nature of prey selection by planktivorousfish. Ecology, 58: 46–59. doi:10.2307/1935107.

Elliott, J.M., and Hurley, M.A. 2000. Daily energy intake andgrowth of piscivorous brown trout, Salmo trutta. Freshw. Biol.44: 237–245. doi:10.1046/j.1365-2427.2000.00560.x.

Hambright, K.D., Drenner, R.W., McComas, S.R., and Hairston,N.G.J. 1991. Gape-limited piscivores, planktivores size refuges,and the trophic cascade hypothesis. Arch. Hydrobiol. 141: 389–404.

Hoyle, J.A., and Keast, A. 1987. The effect of prey morphologyand size on handling time in a piscivore, the largemouth bass(Micropterus salmoides). Can. J. Zool. 65: 1972–1977. doi:10.1139/z87-300.

Hyvarinen, P., and Huusko, A. 2006. Diet of brown trout in rela-tion to variation in abundance and size of pelagic fish prey. J.Fish Biol. 68: 87–98. doi:10.1111/j.0022-1112.2006.00879.x.

Jensen, H., Amundsen, P.-A., Bøhn, T., and Aspholm, P.E. 2004.Feeding ecology of piscivorous brown trout (Salmo trutta L.) ina subarctic watercourse. Ann. Zool. Fenn. 41: 319–328.

Jensen, H., Amundsen, P.-A., Elliott, J.M., Bøhn, T., and Aspholm,P.E. 2006. Prey consumption rates and growth of piscivorousbrown trout in a subarctic watercourse. J. Fish Biol. 68: 838–848. doi:10.1111/j.0022-1112.2006.00972.x.

Jolly, G.M., and Hampton, I. 1990. Some problems in the statis-tical design and analysis of acoustic surveys to assess fishbiomass. Rapp. P.-V. Reun. Cons. Int. Explor. Mer, 189:415–420.

Juanes, F. 1994. What determines prey size selectivity in piscivor-ous fishes? In Theory and application in fish feeding ecology.Edited by D.J. Stouder, K.L. Fresh, and R.J. Feller. South Caro-lina University Press, Columbia, S.C. pp. 79–100.

Kahilainen, K., and Lehtonen, H. 2001. Resource use of native andstocked brown trout Salmo trutta L., in a subarctic lake. Fish.Manage. Ecol. 8: 83–94. doi:10.1046/j.1365-2400.2001.00240.x.

Kahilainen, K., and Lehtonen, H. 2002. Brown trout (Salmo truttaL.) and Arctic charr (Salvelinus alpinus (L.)) as predators onthree sympatric whitefish (Coregonus lavaretus (L.)) forms inthe subarctic Lake Muddusjarvi. Ecol. Freshw. Fish, 11: 158–167. doi:10.1034/j.1600-0633.2002.t01-2-00001.x.

Kahilainen, K., and Lehtonen, H. 2003. Piscivory and prey selec-tion of four predator species in a whitefish dominated subarcticlake. J. Fish Biol. 63: 659–672. doi:10.1046/j.1095-8649.2003.00179.x.

Kahilainen, K., and Østbye, K. 2006. Morphological differentiationand resource polymorphism in three sympatric whitefish Core-gonus lavaretus (L.) forms in a subarctic lake. J. Fish Biol. 68:63–79. doi:10.1111/j.0022-1112.2006.00876.x.

Kahilainen, K., Lehtonen, H., and Kononen, K. 2003. Consequenceof habitat segregation to growth rate of two sparsely rakeredwhitefish (Coregonus lavaretus (L.)) forms in a subarctic lake.Ecol. Freshw. Fish, 12: 275–285. doi:10.1046/j.1600-0633.2003.00029.x.

Kahilainen, K., Malinen, T., Tuomaala, A., and Lehtonen, H. 2004.Diel and seasonal habitat and food segregation of three sympa-

tric Coregonus lavaretus forms in a subarctic lake. J. Fish Biol.64: 418–434. doi:10.1111/j.0022-1112.2004.00307.x.

Kahilainen, K., Alajarvi, E., and Lehtonen, H. 2005. Planktivoryand diet-overlap of densely rakered whitefish (Coregonus lavar-etus (L.)) in a subarctic lake. Ecol. Freshw. Fish, 14: 50–58.doi:10.1111/j.1600-0633.2004.00075.x.

Kahilainen, K., Malinen, T., Tuomaala, A., Alajarvi, E., Tolonen,A., and Lehtonen, H. 2007. Empirical evaluation of phenotype–environment correlation and trait utility with allopatric and sym-patric whitefish (Coregonus lavaretus (L.)) populations in sub-arctic lakes. Biol. J. Linn. Soc. 92: 561–572. doi:10.1111/j.1095-8312.2007.00856.x.

Keeley, E.R., and Grant, W.A. 2001. Prey size of salmonid fishesin streams, lakes, and oceans. Can. J. Fish. Aquat. Sci. 58:1122–1132. doi:10.1139/cjfas-58-6-1122.

Kramer, D.L., Rangeley, R.W., and Chapman, L.J. 1997. Habi-tat selection: patterns of spatial distribution from behaviouraldecisions. In Behavioural ecology of teleost fishes. Edited byJ.-G.L. Godin. Oxford University Press, Oxford, UK. pp. 37–80.

Langeland, A., L’Abee-Lund, J.H., Jonsson, B., and Jonsson, N.1991. Resource partitioning and niche shift in Arctic charr Sal-velinus alpinus and brown trout Salmo trutta. J. Anim. Ecol. 60:895–912. doi:10.2307/5420.

Lima, S.L. 1998. Nonlethal effects in the ecology of predator preyinteractions. Bioscience, 48: 25–34. doi:10.2307/1313225.

Lima, S.L., and Dill, L.M. 1990. Behavioral decisions made underthe risk of predation: a review and prospectus. Can. J. Zool. 68:619–640. doi:10.1139/z90-092.

Miller, T.J., Crowder, L.B., Rice, J.A., and Marschall, E.A. 1988.Larva size and recruitment mechanism in fishes: towards a con-ceptual framework. Can. J. Fish. Aquat. Sci. 45: 1657–1670.doi:10.1139/f88-197.

Mittelbach, G.G., and Persson, L. 1998. The ontogeny of piscivoryand its ecological consequences. Can. J. Fish. Aquat. Sci. 55:1454–1465. doi:10.1139/cjfas-55-6-1454.

Myers, R.A., Baum, J.K., Shepherd, T.D., Powers, S.P., and Peter-son, C.H. 2007. Cascading effects of the loss of apex predatorysharks from a coastal ocean. Science (Washington, D.C.), 315:1846–1850. doi:10.1126/science.1138657. PMID:17395829.

Næsje, T.F., Sandlund, O.T., and Saksgaard, R. 1998. Selectivepredation of piscivorous brown trout (Salmo trutta L.) on poly-morphic whitefish (Coregonus lavaretus L.). Arch. Hydrobiol.Spec. Issues Adv. Limnol. 50: 283–294.

Østbye, K., Amundsen, P.-A., Bernatchez, L., Klemetsen, A.,Knudsen, R., Kristoffersen, R., Næsje, T.F., and Hindar, K.2006. Parallel evolution of ecomorphological traits in the Eur-opean whitefish Coregonus lavaretus (L.) species complex dur-ing postglacial times. Mol. Ecol. 15: 3983–4001. doi:10.1111/j.1365-294X.2006.03062.x. PMID:17054498.

Persson, L., and Greenberg, L.A. 1990. Optimal foraging and habi-tat shift in perch (Perca fluviatilis) in a resource gradient. Ecol-ogy, 71: 1699–1713. doi:10.2307/1937579.

Persson, L., Andersson, J., Wahlstrom, E., and Eklov, P. 1996.Size-specific interactions in lake systems: predator gape limita-tions and prey growth rate and mortality. Ecology, 77: 900–911.doi:10.2307/2265510.

Persson, L., Amundsen, P.-A., De Roos, A.M., Klemetsen, A.,Knudsen, R., and Primicerio, R. 2007. Culling prey promotespredator recovery — alternative states in a whole-lake experi-ment. Science (Washington, D.C.), 316: 1743–1746. doi:10.1126/science.1141412. PMID:17588929.

Roughgarden, J. 1986. A comparison of food-limited and space-limited animal competition communities. In Community ecol-

1840 Can. J. Fish. Aquat. Sci. Vol. 65, 2008

# 2008 NRC Canada

ogy. Edited by J.M. Diamond and T.J. Case. Harper & Row,New York. pp. 492–516.

Scharf, F.S., Juanes, F., and Rountree, R.A. 2000. Predator size –prey size relationships of marine fish predators: interspecificvariation of ontogeny and body size on trophic niche breadth.Mar. Ecol. Prog. Ser. 208: 229–248. doi:10.3354/meps208229.

Schoener, T.W. 1970. Nonsynchronous spatial overlap of lizards inpatchy habitat. Ecology, 51: 408–418. doi:10.2307/1935376.

Schulz, U., and Berg, R. 1992. Movements of ultrasonically taggedbrown trout (Salmo trutta L.) in Lake Constance. J. Fish Biol.40: 909–917. doi:10.1111/j.1095-8649.1992.tb02636.x.

Shotton, R., and Bazigos, G.P. 1984. Techniques and considera-tions in the design of acoustic surveys. Rapp. P.-V. Reun. Cons.Int. Explor. Mer, 184: 34–57.

Sih, A., and Christensen, B. 2001. Optimal diet theory: when doesit work and when does it fail? Anim. Behav. 61: 379–390.doi:10.1006/anbe.2000.1592.

Sih, A., and Moore, R.D. 1990. Interacting factor of predator andprey behavior in determining diets. In Behavioural mechanismof food selection. Edited by R.N. Hughes. Springer-Verlag, Ber-lin, Germany. pp. 771–796.

Sinclair, A.R.E., Mduma, S., and Brashares, J.S. 2003. Patterns ofpredation in a diverse predator–prey system. Nature (London),425: 288–290. doi:10.1038/nature01934. PMID:13679915.

Stephens, D.W., and Krebs, C.J. 1986. Foraging theory. PrincetonUniversity Press, Princeton, N.J.

Tonn, W.M., and Paskowski, C.A. 1992. Piscivory and recruitment:mechanisms structuring prey populations in small lakes. Ecol-ogy, 73: 951–958. doi:10.2307/1940171.

Townsend, C.R., and Winfield, I.J. 1985. The application of opti-mal foraging theory to feeding behaviour in fish. In Fish ener-getics. Edited by P. Tytler and P. Calow. Croom Helm, London,UK. pp. 67–98.

Turner, A.M., and Mittelbach, G.G. 1990. Predator avoidance andcommunity structure: interactions among piscivores, planktivores,and plankton. Ecology, 71: 2241–2254. doi:10.2307/1938636.

Vehanen, T., Hyvarinen, P., and Huusko, A. 1998. Food consump-tion and prey orientation of piscivorous brown trout (Salmotrutta) and pikeperch (Stizostedion lucioperca) in a large regu-lated lake. J. Appl. Ichthyol. 14: 15–22. doi:10.1111/j.1439-0426.1998.tb00608.x.

Wallace, R.K. 1981. An assessment of diet-overlap indexes. Trans.Am. Fish. Soc. 110: 72–76. doi:10.1577/1548-8659(1981)110<72:AAODI>2.0.CO;2.

Werner, E.E., and Gilliam, J.F. 1984. The ontogenetic niche andspecies interactions in size-structured population. Annu. Rev.Ecol. Syst. 15: 393–425. doi:10.1146/annurev.es.15.110184.002141.

Wootton, R.J. 1998. Ecology of teleost fishes. Chapman & Hall,London, UK.

Jensen et al. 1841

# 2008 NRC Canada