Foraging capacity and resource synchronization in an ontogenetic diet switcher, pikeperch (Stizostedion lucioperca)

Transcription

1 Foraging capacity an resource synchronization in an ontogenetic iet switcher, pikeperch (Stizosteion lucioperca) Persson, Aners; Brönmark, Christer Publishe in: Ecology Publishe: Link to publication Citation for publishe version (APA): Persson, A., & Brönmark, C. (2002). Foraging capacity an resource synchronization in an ontogenetic iet switcher, pikeperch (Stizosteion lucioperca). Ecology, 83(11), General rights Copyright an moral rights for the publications mae accessible in the public portal are retaine by the authors an/or other copyright owners an it is a conition of accessing publications that users recognise an abie by the legal requirements associate with these rights. Users may ownloa an print one copy of any publication from the public portal for the purpose of private stuy or research. You may not further istribute the material or use it for any profit-making activity or commercial gain You may freely istribute the URL ientifying the publication in the public portal Take own policy If you believe that this ocument breaches copyright please contact us proviing etails, an we will remove access to the work immeiately an investigate your claim. L UNDUNI VERS I TY PO Box L un

2 Ecology, 83(11), 2002, pp by the Ecological Society of America FORAGING CAPACITY AND RESOURCE SYNCHRONIZATION IN AN ONTOGENETIC DIET SWITCHER, PIKEPERCH (STIZOSTEDION LUCIOPERCA) ANDERS PERSSON 1 AND CHRISTER BRÖNMARK Department of Ecology, Limnology, Lun University, Ecology Builing, S Lun, Sween Abstract. Species unergoing ontogenetic iet shifts face a risk of resource competition that elays transitions between feeing stages. Such ontogenetic bottlenecks are common in piscivorous fish because competition with future prey may retar growth an prevent a size avantage. In pikeperch (Stizosteion lucioperca), year class strength of 0 cohorts is highly variable an positively relate to early onset of piscivory. To ientify the causes of this pattern, we experimentally quantifie size-epenent planktivorous an piscivorous foraging capacity an incorporate the ata into a growth moel. For any given prey type an size, foraging capacity escribe a hump-shape relationship with preator size. Foraging capacity on aphnis peake at a pikeperch length of 66 mm, suggesting a narrow scope of planktivory. The highest capacity in the piscivorous niche was reache at a preator-to-prey length ratio of 5, where the ratio was an integrate measure of preator size over several prey sizes. With the growth moel, we erive size istributions of 0 cohorts as functions of resource levels. Simulations reveale two major eterminants for the year class strength of pikeperch. First, iscontinuous availability of prey sizes counteracte switching to piscivory within the first growing season. This was accentuate by prey fish growth, which cause the planktivory an piscivory niches to separate over time an limite the time winow when iet shift was possible. Secon, the hump-shape relationship between size an foraging capacity resulte in growth reuction when growing out of the planktivorous niche. Switching to piscivory in our moel occurre along a perpenicular relationship between fish prey an zooplankton ensity. Zooplankton ensity etermine whether pikeperch reache a size avantage over prey, an fish prey ensity affecte whether the foraging return of piscivory was higher than planktivory. Iniviuals not reaching a size avantage over prey an failing to become piscivorous were stunte at a size when consumption balance metabolic requirements. Piscivorous iniviuals, however, continue to grow fast throughout the season by feeing on the wave of the prey cohort. Our results highlight the importance for preators that shift iet to be synchronize with fluctuations in resource availability, such as the pulses of new cohorts of prey fish. Key wors: growth moel; ontogenetic iet shift; pikeperch; piscivory; planktivory; size-base moel; size-structure population; Stizosteion lucioperca. INTRODUCTION Piscivorous fish are top preators in aquatic environments an may profounly affect community structure an ynamics (Carpenter an Kitchell 1993, Brönmark an Weisner 1996). Size an structure of piscivorous fish populations are strongly influence by the success of the juvenile stage (Mittelbach an Persson 1998). To obtain a mechanistic unerstaning of community patterns it is therefore necessary to etermine what constraints act on piscivore recruitment to the ault phase. Most species of piscivorous fish pass through a phase feeing on smaller foo items such as zooplankton an benthic invertebrates before switching to a fish iet (Mittelbach an Persson 1998). If resource ensity in this stage is low, it may constitute an ontogenetic bot- Manuscript receive 22 October 2001; revise 1 March 2002; accepte 15 March aners.persson@limnol.lu.se 3014 tleneck by slowing growth an elaying the iet shift (Persson an Greenberg 1990). Thus, to reach the piscivore niche, iniviual fish must possess enough capacity to grow an survive in the juvenile niche in competition with, e.g., future prey fish species (Byström et al. 1998). This constitutes an ontogenetic traeoff, as any increase in juvenile competitive ability is likely to be at the expense of ault competitive ability in the piscivore niche. Typically, specialize piscivores switch to a fish iet at an early stage; whereas less specialize species may wait several years (Keast 1985). Early switching requires high growth rate, which implies high ensities of proper resources an safe foraging opportunities. If unfavorable conitions prevail, growth is reuce an this may seriously affect recruitment into the ault stage because mortality rate ue to starvation an preation is a size-epenent process (Juanes 1994, see also Mangel an Abrahams 2001). The specific requirements of specialize piscivores in their early ontogeny often result in high year-

3 November 2002 ONTOGENETIC DIET SHIFT IN PIKEPERCH 3015 to-year variation in year class strength an size istribution after the first growing season (Olson 1996, van Densen et al. 1996). Thus, constraints relate to the onset of piscivory, such as the size relation between iniviual preators an prey, will have consequences at both population an community levels. Here, we investigate the mechanisms behin, an the consequences of, an ontogenetic trae-off in a piscivorous specialist, the pikeperch (Stizosteion lucioperca; see Plate 1). Fiel stuies have shown that recruitment success of ault, piscivorous pikeperch is variable both between an within systems, an is positively correlate with year class strength, i.e., the number of surviving iniviuals after the first growing season (Buijse an Houthuijzen 1992). Large iniviual size an an early shift to piscivorous feeing characterize strong year classes. As outline above, this suggests significant costs, i.e., low survival, associate with failure to reach the piscivorous stage uring the first season. To analyze the consequences of size-base foraging performance, we first experimentally quantifie the foraging rate of pikeperch as a function of preator an prey size. The parameterize foraging abilities were then use in a size-base growth moel to erive a mechanistic unerstaning of the causes of observe variability in year class strength an size istribution of cohorts. FORAGING EXPERIMENTS PLATE 1. Pikeperch (Stizosteion lucioperca) is esigne for piscivory, with a large mouth an sharp teeth. Foraging capacity, in terms of attack rates, in pikeperch was etermine in laboratory (planktivory) an fiel enclosure experiments (piscivory). Daphnia an fish were selecte as representative prey for the planktivore an the piscivore niche, respectively. The experiments were esigne as functional response experiments to erive accurate parameter estimates of the iniviual growth moel (see Moel escription). Pikeperch of 200 mm total length were obtaine from a fishery in Lake Ringsjön, southern Sween, an pikeperch 200 mm were elivere from a pikeperch fish farm that uses pons an natural live prey. In total more than 100 pikeperch varying in size from 38 to 380 mm were use. Each preator was use only once for each prey size an prey ensity combination. Experiments with pikeperch foraging on zooplankton were performe in 60 L tanks at a constant temperature of 18C. One iniviual was introuce per tank at least 1 wk before the experiment starte an was allowe to fee on zooplankton prey until the ay before initiating experiments. As zooplankton prey we use Daphnia sp. (mean length 0.93 mm) collecte from nearby pons. An experiment starte when a known number of zooplankton iniviuals were mixe into the water of a test tank containing pikeperch. The time from the onset of searching for the first prey item until the fifth prey item was completely ingeste was recore. We performe replicate (n 5) functional response experiments on nine ifferent sizes of pikeperch ( mm) foraging at six prey ensities ( iniviuals/l). Piscivory experiments were performe in two outoor an one inoor pon/pool systems; experimental scales were 100 an 28 m 2 in outoor an 5 m 2 in inoor units. The outoor systems were use for larger preators (135 mm) an the inoor system for the smallest preator size class (130 mm). Outoor experiments were performe at a water temperature ranging between 16 an 20C. The water temperature of the inoor experiments was kept at 18C an the light ark cycle at 14:10 to simulate summer conitions. Prey fish (roach Rutilus rutilus an bream Abramis brama) were capture with electrofishing an seining in nearby lakes. All preators were allowe to fee on fish prey for at least 1 wk before the experiments. Each experimental unit containe one ( mm length) or two preators ( mm length). After preators were confine to cages, prey fish (size limits mm) were introuce an allowe to accommoate for 1 h. The experiment starte by releasing the preators, an it was terminate after 3 when remaining fish were capture an counte. In pon experiments an for some prey sizes in pool experiments (90, 110, an 150 mm), the preator consumption rate was only etermine at three prey ensities. In those cases, the attack rate was etermine by fitting the consumption rates to a linear functional response. In cases where more prey ensities were use, attack rates were erive from a type II functional response equation (Eq. 2) using least squares nonlinear regression. Experiments were performe by either varying prey size (pikeperch sizes 130 or 350 mm) or by varying pikeperch size (prey sizes 35 or 50 mm). Consequently, estimates on attack rates were specific to each preator size an prey size combination. The estimate attack rates were fitte to the size-epenent attack rate function (Eq. 3), again using least squares nonlinear regression. This attack rate function (see Persson et al.

4 3016 ANDERS PERSSON AND CHRISTER BRÖNMARK Ecology, Vol. 83, No. 11 TABLE 1. Moel variables an parameters for pikeperch (Stizosteion lucioperca) an its prey. Category for variables Symbol Value Unit Description Source Zooplankton Z varie g/l ensity (40 g/zooplankter) a Fish prey Preator Planktivory Piscivory Hanling Metabolism F l F g F 1F 2F w l A l R varie 0.5, varie P 56 2P 0.31 A Z l OZ Z A F l OR F e g/l mm mm/ mm/g 2F g mm mm/mm mm/g 2P m 3 / mm m 3 / mm/mm g1 mm allometric g (1) / ensity length growth rate scalar for l w conversion exponent for l w conversion preator mass preator length relative preator length (Prey fish are available when l R 2) scalar of l w conversion exponent of l w conversion maximum attack rate optimal preator length allometric exponent maximum attack rate optimal relative length allometric exponent scalar allometric exponent allometric scalar allometric exponent assimilation efficiency Sources are: a, A. Persson, unpublishe ata; b, Romare an Bergman (1999); c, Turesson et al. (2002);, this stuy (see Results); e, Persson et al. (1998), Claessen et al. (2000). a, b a a c, a a e e e e 1998 for a thorough escription) escribes the attack rate on a specific prey type as a hump-shape relationship with preator size. We chose this particular function for two reasons. First, a hump-shape relationship has been foun in other species to aequately escribe the attack rate on specific prey (Werner 1988, 1994, Byström an Garcìa-Berthou 1999, Hjelm 2000, Wahlström et al. 2000, Persson an Brönmark 2002). Secon, parameter estimates were available for three species normally coexisting with pikeperch, namely perch (Perca fluviatilis), bream (Abramis brama), an roach. Hence, we coul assess the competitive ability in pikeperch by comparing the parameter estimates of these species. MODEL DESCRIPTION The size-base growth moel is a simplifie version of the consumer-resource moel escribe by Persson et al. (1998), with the extension of a secon resource (fish). Parameter values use when moeling consumption were generate in the experimental stuy, whereas metabolic requirements were estimate using publishe relationships (Table 1). Iniviual growth rate is moele as a mass balance between intake rate (c) an metabolism (m), where intake rate is a function of the length relation between preator an prey (l) an resource ensity (R), an metabolic requirement is a function of preator mass (w) accoring to the following: w ec(l, R) m(w) (1) t where e is a coefficient etermining conversion efficiency (incluing specific ynamic action). Hence, the basic unit of the moel is preator wet mass, but mass is converte to preator length for use in consumption equations, an to prouce output ata that is comparable with fiel ata. Iniviual length is escribe as a function of boy mass (l 1 w 2 ) with the reservation that an iniviual is allowe to ecrease in mass, but not in length. Starvation mortality occurs if iniviual mass rops below 70% of the mass preicte by the length mass equation. If an iniviual temporarily ecreases in mass ue to starvation to thereafter increase in mass, it will not increase in length until it has reache the mass preicte by the length mass equation. This latter case is introuce for logical reasons but it is never realize in the simulations performe in this stuy. To etermine the size-epenent consumption rates, we assume that iniviual fish capture prey accoring to a Holling type II functional response. The size specific consumption rate (c(l, R)) as a function of prey ensity is given by a(l)r c(l, R) (2) 1 a(l)h(l)r where a(l) is attack rate, h(l) is the hanling time in-

5 November 2002 ONTOGENETIC DIET SHIFT IN PIKEPERCH 3017 cluing igestion, an R is the prey ensity. One major objective was to compare the competitive ability of pikeperch in relation to other species in the planktivorous an piscivorous niches. The attack rate for a given prey size has been shown to escribe a humpshape relationship with preator size (Werner 1988, Persson et al an references therein). The increase in attack rate with preator size is relate to improve visual acuity an locomotion ability (Schoener 1969, Peters 1983). The ecreasing part is relate to ecrease maneuverability with preator size compare to the prey (Domenici 2001), an ecrease ability to iscern small prey ue to ecrease cone ensity (Breck an Gitter 1983). Following Persson et al. (1998), the attack rate was escribe as a function of boy length accoring to [ ] 0 0 l l a(l) A exp 1 (3) l l where l 0 is the boy length at which the maximum rate, A, is achieve, an is a size-scaling exponent, which affects the rate by which the attack rate increases below an ecreases above l 0. The size-epenent attack rate function (Eq. 3) is applicable to one specific prey type as in the planktivorous feeing here. However, for fishfeeing pikeperch we aime at escribing consumption rates for several combinations of preator an prey sizes. To solve this, we use the ratio of preator length an prey length (l R ) as a measure of relative preator size (Scharf et al. 2000), instea of absolute length (l A ), when fitting attack rates of fish feeing pikeperch to the size-epenent attack rate function (Eq. 3). The relative preator length is frequently use when escribing preator an prey relations in size-structure populations (e.g., Scharf et al. 1998, Hartman 2000). Our ata suggest it is legitimate to assume that the parameters are inepenent on preator size if correcte for prey size for the size combinations use in our experiments. At all times, a zooplankton (Z ) an a fish resource (F ) is available. Because each resource type requires a completely ifferent foraging moe, an iniviual preator is not allowe to aopt a strategy combining the two resources (i.e., R Z or F ). Thus, an iniviual is assume to choose the foraging strategy generating the highest foraging return accoring to: c(l A, Z) c(l i, R) max (4) c(l, F). The ensities of Z an F are fixe on a wet mass basis in each simulation. However, iniviual fish prey grow over the season, which results in relatively large numbers of small prey in the beginning an few large prey at the en of the season (e.g., when F 4 mg/l, fish prey numbers are 1 an iniviuals/l at t 0 an 100, respectively). Iniviual fish prey length, l F, is moele as a fixe aily increment in length, g F, R (l F /t g F ). The initial prey length, i.e., the length of prey fish when they appear in the pelagic part of the system, was assume to be 10 mm base on empirical observations (Hamrin et al. 1998). The gape size of the preator sets the upper limit of prey available for consumption (Hambright 1994). Base on experiments (Turesson et al. 2002; see Results) an fiel observations (Popova an Sytina 1977), this limit occurs for prey that are 0.5 times the pikeperch length. Hence, in the moel pikeperch were not allowe to fee on fish prey until l R 2. Metabolic eman an igestive capacity in Stizosteion species are largely base on the closely relate an more intensively stuie Perca species (e.g., Hewett an Johnson 1992). We use the relationships evelope by Claessen et al. (2000) for perch, where metabolic eman an hanling costs (incluing igestion) are epenent on preator mass accoring to allometric functions (metabolic eman: m(s) s, where an are constants; hanling time h(s) s, where an are constants). Iniviual growth simulations were performe using STELLA software (High Performance Systems, Hanover, New Hampshire). Each simulation was run for one season, which was assume to last for 100, using a time step of 1. First, we teste the effects of resource ensities on iet an growth in pikeperch. We use 100 pikeperch with an unimoal size istribution ranging from 9 mm (size at first feeing; McElman an Balon 1979) to 13 mm (mean 11 mm; CV 10%), to inclue variations relate to, e.g., hatching size an time. Data were sample at the en of each simulation an size istributions constructe using 5-mm size classes. The initial size variation use in the moel is a conservative estimate in comparison with fiel observations, which show a coefficient of variation in length early in the season in the range of 15 20% (Hamrin et al. 1998, Romare an Bergman 1999). We also teste for interaction effects of resource ensities an prey fish growth, an synchronization between pikeperch an prey fish hatching. In these simulations we use a fixe initial pikeperch size of 9 mm. RESULTS AND DISCUSSION Foraging capacity Species with ontogenetic iet shifts face a trae-off between being a specialist forager on one resource or a generalist forager on both. Specialists trae risks associate with low competitive ability in one niche with being superior in another niche. Consequently, population ensities of specialists are generally a result of their success in the feeing stage where they experience intense competition, because competition reuces growth an prevents transition between feeing stages. Specialization in one niche shoul be associate with costs in competitive ability in other niches (Persson 1988, Werner 1988). Base on fiel ata showing that

6 3018 ANDERS PERSSON AND CHRISTER BRÖNMARK Ecology, Vol. 83, No. 11 pikeperch has a short planktivorous phase an early switching to piscivory (reviewe in Mittelbach an Persson 1998), pikeperch was consiere a specialist piscivore. The aequacy of this statement coul be evaluate by examining the quantitative estimates of foraging capacity erive from the functional response experiments, where we quantifie size-epenent attack rates in pikeperch when foraging on zooplankton an fish. The attack rate first increase with pikeperch size to a maximum, an then ecrease asymptotically towar zero. The shape of this relationship correspons with earlier stuies. The parameters of the size-epenent attack rate function summarize species-specific characteristics an are useful for interspecific comparisons of competitive ability between coexisting species (Byström an Garcìa-Berthou 1999, Persson et al. 2000, Persson an Brönmark 2002). Pikeperch showe a smaller optimum size (66 mm), an a lower maximum attack rate (13 m 3 /; Fig. 1a) when feeing on zooplankton compare with foraging abilities estimate for both a specialist planktivore (roach Rutilus rutilus; Hjelm 2000) an an ontogenetic generalist (perch Perca fluviatilis; Wahlström et al. 2000). When comparing zooplankton feeing pikeperch an the specialist benthivore bream (Abramis brama), optimum size was similar, whereas maximum attack rate was twice as high in bream (Persson an Brönmark 2002). This suggests a narrow scope for planktivory in pikeperch compare with roach, perch, an bream, renering further support to the view that piscivore specialization is relate to costs in competitive ability in the planktivorous niche. Quantification of foraging capacity over ontogeny is rare, an our ata on piscivorous foraging capacity is unique. Pikeperch in the piscivory experiments were able to ingest all prey sizes use, i.e., from a preator: prey length ratio of 2 to 14, an maximum attack rate (9.6 m 3 /) was achieve at a preator: prey length ratio of 5.3 (Fig. 2b). The estimate maximum attack rate for pikeperch is in the upper range of attack rates use by Claessen et al. (2000), an higher than the value suggeste for perch. Interestingly, perch usually nee several years to become piscivorous an rely on other resources (benthic invertebrates) until proper conitions appear. Thus, our ata support the general view of pikeperch being a specialist piscivore. FIG. 1. Size-epenent attack rates ( 1 SE) of (a) planktivorous pikeperch foraging on 1-mm Daphnia an (b) piscivorous pikeperch foraging on ifferent size prey fish. The attack rates escribe a hump-shape relationship with (a) pikeperch boy length an (b) relative pikeperch length. Optimum size, maximum attack rate, an the size-scaling exponent of the size-specific attack rate function (base on leastsquares nonlinear regression; Eq. 3) are, respectively, (a) l ZO, mm, A Z 13 3m 3 /, Z 3.5 3; an (b) l FO mm/mm, A F 9.6 2m 3 /, F Note that the ata for planktivorous pikeperch assumes 12 h of feeing per ay. Growth simulations Simulations showe that resource synchronization might be crucial for the recruitment success of a preator population where iniviuals unergo ontogenetic iet shifts. Any mismatch in time or size availability of the secon resource cause a stunte population. This mismatch was ue to preators growing out of the first niche, without being large enough to fit the secon niche. Thus, the origin of the stunte population was primarily iscontinuous availability of prey sizes, which, in turn, was cause by the growth of prey fish that separate the two niches over time. Resource ensity inirectly affecte recruitment success by enabling a higher growth rate of the preator, an thus making a iet shift possible early in the season when the size ifference between the two resources was still small. The hump-shape relationship (Fig. 1) between attack rate an preator size ha funamental consequences for the growth rate of the preator. When it grew into the niche there was a positive feeback as growth increase the attack rate, which in turn increase growth, an so on. This proceee until the optimum prey-to-preator size relationship was passe (i.e., the maximum of the attack rate function), after which the opposite was true, i.e., any growth increment mae the preator less efficient in this feeing niche. In the planktivorous niche this resulte in iniviuals attaining a size for which consumption balance met-

7 November 2002 ONTOGENETIC DIET SHIFT IN PIKEPERCH 3019 FIG. 2. Simulate pikeperch iet at ifferent combinations of zooplankton an fish prey ensity (left panels), an examples of size frequency istributions corresponing to case a c (right panels). Open bars enote initial size istribution, an fille bars enote final size istribution. When zooplankton ensity 0.05, all pikeperch survive, which correspons to minimum resource requirements of 9-mm pikeperch. Shae areas enote resource combinations where (a) pikeperch remain planktivorous the whole season, (b) the pikeperch cohort consists of both planktivorous an piscivorous iniviuals (the winow of bimoality ), an (c) all iniviuals are piscivorous. Initial fish prey size 10 mm; g F 0.5 mm/. Seasonal length was 100. abolic requirements (Fig. 2a). In the piscivorous niche, however, pikeperch continue to grow as prey fish grew (Fig. 2c). Metaphorically speaking, pikeperch were feeing on the prey cohort wave, synchronizing their own growth in relation to prey growth. We ientifie three regions in the fish an zooplankton ensity phase plane, representing ifferent foraging moes (Fig. 2). Below a critical zooplankton ensity (0.04 iniviuals/l), no pikeperch survive. When resource ensity is just beyon this critical level, surviving pikeperch stay in the planktivore niche throughout the season (Fig. 2a). If cohorts remain planktivorous, iniviuals accumulate at the maximum attainable size set by zooplankton ensity. The size istribution is therefore narrow, with a low level of ispersion. If resource ensity is sufficient, iet shift to piscivory is promote. This shift is affecte by both zooplankton an prey fish ensity, the threshol level being a perpenicular relationship between zooplankton an fish prey ensity (Fig. 2). In the winow of bimoality (Fig. 2b), ifferent iniviuals of the cohort apply ifferent foraging moes. The size of this region is fully epenent on the ispersion of the initial size istribution. Hence, if the ispersion of the initial istribution is 0, the winow of bimoality isappears. The level of ispersion of the final size istribution epens on both the initial istribution an on the ifference in profitability between the planktivorous an piscivorous stage, which results in ifferent growth trajectories. Consequently, the largest size ifference is attaine when zooplankton ensity is low an fish prey ensity is high, i.e., in the upper left region of the shae area in Fig. 2b. When all iniviuals remaining in the cohort become piscivorous (Fig. 2c), the size istribution becomes unimoal with an intermeiate level of ispersion compare to the stunte unimoal an the bimoal size istribution. It is notable that the simulate size istributions (Fig. 2) resemble patterns

8 3020 ANDERS PERSSON AND CHRISTER BRÖNMARK Ecology, Vol. 83, No. 11 shown by natural populations of pikeperch (Buijse an Houthuijzen 1992, Frankiewicz et al. 1996, van Densen et al. 1996), walleye (Maenjian an Carpenter 1991), largemouth bass (Aams an DeAngelis 1987), an Arctic char (Hammar 1998). Switching to piscivory in our moel occurre along a perpenicular relation between fish prey ensity an zooplankton ensity (Fig. 2b). Left of the vertical part of the shae area, an below the horizontal part, pikeperch remain planktivorous. This pattern is prouce by the two criteria for iet switching. First, prey size must be within the winow of susceptibility to preation. Hence, left of the vertical part, pikeperch grow too slow to reach a size avantage over prey (i.e., preator : prey length ratio, l R 2). The secon criterion for iet switching was a higher foraging return of piscivory compare to planktivory (Eq. 4). Because the hanling time is the same for both prey types, the highest intake rate coul be etermine by comparing the prouct of attack rate an resource ensity. Below the horizontal part, fish prey ensity is too low for foraging return on zooplankton to rop below that on fish (whatever the value of zooplankton ensity). The most likely outcome in nature is that at least one resource, fish prey or zooplankton, is in low abunance. A large population of planktivorous fish prey woul reuce the zooplankton resource. This results in ifferent profitability in the two niches, which shoul reinforce any initial ifference in size between pikeperch iniviuals. It shoul also be note that if the zooplankton resource is heavily exploite, it shoul affect the growth rate of both pikeperch an fish prey. Consequently, slow growing prey is more likely to be foun an increase the probability of iet switching at low zooplankton availability. Conversely, if zooplankton ensity is high (ue to low abunance of planktivorous fish) there is relatively little to be gaine by switching iet early because the growth opportunities in both niches are more similar. If keeping fish prey ensity constant, the simulations show that switching is possible above a certain zooplankton ensity because pikeperch reach a size avantage (l R 2). Increasing zooplankton ensity above this threshol will initially allow switching earlier an at a smaller preator size. However, if increasing the ensity further, iet switching will occur later in the season an at a larger preator size. In the first case, when switching time an size are negatively relate to zooplankton ensity, switching occurs immeiately when prey fall within the winow of susceptibility to preation (l R 2). In the secon case, when the relationship between switching time an zooplankton ensity is positive, foraging capacity on zooplankton is larger than on fish when pikeperch reach a size avantage over prey. Hence, switching oes not occur until the foraging return on zooplankton rops below that on fish, which will occur later an at a larger size with increasing zooplankton ensity. In fact, pikeperch may reach a size of 166 mm the first season FIG. 3. Effects of initial prey fish length (lower x-axis) an prey fish growth rate, g F, (upper x-axis) on simulate final pikeperch length. Cases a c correspon to the combinations of zooplankton an fish prey ensities marke in Fig. 2, which are: (a) 1 iniviual/l an 0.25 mg/l, (b) 1.5 iniviuals/l an 1.5 mg/l, an (c) 5 iniviuals/l an 1.5 mg/l, respectively. Final length is etermine by zooplankton ensity if pikeperch fail in switching to piscivory. When the preator : prey length ratio is synchronize just beyon l OR, maximum growth rate is obtaine. Initial pikeperch length 9 mm. feeing only on zooplankton (initial length 9 mm; zooplankton ensity 100 iniviuals/l), which excees the length obtaine in the main part of the shae region in Fig. 2c. In the moel, iscontinuous prey size availability was cause by the growth of prey fish, which separate the two niches over time. A iscontinuous availability of prey sizes may also be generate if some preferre prey sizes are remove or reuce to low levels (van Densen et al. 1996). This possibility is not specifically taken into account here as incluing a feeback between resource consumption an resource ynamics was beyon the scope of the stuy. Excluing a feeback to the zooplankton resource probably overestimates the probability of a bimoal size istribution. However, pikeperch has been shown in this stuy to have a low foraging capacity on zooplankton an woul probably only have a minor influence on zooplankton ynamics in comparison to zooplanktivore specialists. Conversely, it is reasonable to believe that pikeperch in fact rive the ynamics of the fish resource. If we inclue a ynamical interaction between pikeperch an the fish resource, we woul expect the smallest prey sizes to become remove first, as these iniviuals are the first to fall within the prey size winow of pikeperch. That narrows the size range of the prey cohort, which woul further increase the iscontinuity in resource availability, an hence the winow of bimo-

9 November 2002 ONTOGENETIC DIET SHIFT IN PIKEPERCH 3021 ality. In that sense, this moel provies a conservative preiction of the probability of a bimoal size istribution. We also teste how pikeperch growth is affecte by prey growth rate an initial prey size (a surrogate for spawning ate). Fig. 3 shows that both these factors have similar effects on final pikeperch length. Prey fish growth rate affects if an when pikeperch grow into the winow of piscivory, but also the growth opportunities once in the piscivorous niche. If prey fish grow slowly, pikeperch also experience slow growth in the piscivorous niche because pikeperch growth is synchronize with prey growth at a preator: prey length ratio well beyon optimum (Fig. 3). The prey fish growth rate yieling highest pikeperch growth rate is 0.2 mm/ (Fig. 3), an occurs when pikeperch are able to maintain preator: prey length ratio just beyon the optimum, l OR. It is notable that the simulations show pikeperch being sensitive to prey fish growth rates within the range measure in fiel situations, i.e., when prey fish reach mm total length the first season. Initial prey size influence whether or not pikeperch reache a size avantage over the prey population. If prey fish are given a hea start, pikeperch only become piscivorous at high resource ensities. In lakes of Sween, pikeperch spawn simultaneously or later than most prey fish, suggesting that timing in recruitment inee is one critical factor for recruitment success. Whether or not pikeperch reach the piscivorous stage within the first season may have significant consequences at the population an community levels. Moel simulations reveale that iscontinuous prey availability an the growth rate of pikeperch in relation to prey fish etermine recruitment success. In particular, the fact that the level of iscontinuity increases over the season exerts further constraint on recruitment success. This points to the importance of being synchronize with preictable fluctuations in resource availability, such as the pulses of new cohorts of prey fish. ACKNOWLEDGMENTS We thank Michael Neubert, Pia Romare, Håkan Turesson, an three anonymous referees for valuable comments on the manuscript. We are particularly grateful for the constructive comments an clarifications provie by Anré M. e Roos. This stuy was supporte by grants from the European Community contract FAIR CT an the Sweish Council for Forestry an Agriculture. LITERATURE CITED Aams, S. M., an D. L. DeAngelis Inirect effects of early bass sha interactions on preator population structure an foo web ynamics. Pages in W. C. Kerfoot an A. Sih, eitors. Preation: irect an inirect impacts on aquatic communities. University Press of New Englan, Hanover, New Hampshire, USA. Breck, J. E., an M. J. Gitter Effects of fish size on the reactive istance of bluegill (Lepomis macrohirus) sunfish. Canaian Journal of Fisheries an Aquatic Sciences 40: Brönmark, C., an S. E. B. Weisner Decoupling of cascaing trophic interactions in a freshwater, benthic foo chain. Oecologia 108: Buijse, A. D., an R. P. Houthuijzen Piscivory, growth an size-selective mortality of age 0 pikeperch (Stizosteion lucioperca). Canaian Journal of Fisheries an Aquatic Sciences 49: Byström, P., an E. Garcìa-Berthou Density epenent growth an size specific competitive interactions in young fish. Oikos 86: Byström, P., L. Persson, an E. Wahlström Competing preators an prey: juvenile bottlenecks in whole-lake experiments. Ecology 79: Carpenter, S. R., an J. F. Kitchell The trophic cascae in lakes. Cambrige University Press, Cambrige, UK. Claessen, D., A. M. e Roos, an L. Persson Dwarfs an giants: cannibalism an competition in size-structure populations. American Naturalist 155: Domenici, P The scaling of locomotor performance in preator prey encounters: from fish to killer whales. Comparative Biochemistry an Physiology Part A 131: Frankiewicz, P., K. Dabrowski, an M. Zalewski Mechanisms of establishing bimoality in a size istribution of age-0 pikeperch, Stizosteion lucioperca (L.) in the Sulejow Reservoir, Central Polan. Annales Zoologica Fennici 33: Hambright, K. D Morphological constraints in the piscivore planktivore interaction: implications for the trophic cascae. Limnology an Oceanography 39: Hammar, J Evolutionary ecology of Arctic char (Salvelinus alpinus (L.)). Intra- an interspecific interactions in circumpolar populations. Ph.D. thesis. Uppsala University, Uppsala, Sween. Hamrin, S. F., T. Soler, M. Eriksson, J. Svensson, H. Linge, G. Cronberg, an P. Romare From seiment to fish an outline of the Lake Vombsjön ecosystem [In Sweish with English summary.] Fiskeriverket Rapport 1: Hartman, K. J The influence of size on stripe bass foraging. Marine Ecology Progress Series 194: Hewett, S. W., an B. L. Johnson Fish bioenergetics 2. University of Wisconsin Sea Grant Institute, Maison, Wisconsin, USA. Hjelm, J Size-epenent performance an ontogenetic niche shifts. Ph.D. thesis. Umeå University, Umeå, Sween. Juanes, F What etermines prey-size selectivity in piscivorous fishes? Pages in D. J. Stouer, K. L. Fresh, an R. J. Feller, eitors. Theory an application in fish feeing ecology. University of South Carolina Press, Columbia, South Carolina, USA. Keast, A The piscivore feeing guil of the fishes in small freshwater ecosystems. Environmental Biology of Fishes 12: Maenjian, C. P., an S. R. Carpenter Iniviual-base moel for growth of young-of-the-year walleye: a piece of the recruitment puzzle. Ecological Applications 1: Mangel, M., an M. V. Abrahams Age an longevity in fish, with consieration of the ferox trout. Experimental Gerontology 36: McElman, J. F., an E. K. Balon Early ontogeny of walleye, Stizosteion vitreum vitreum, with saltatory evelopment. Environmental Biology of Fishes 4: Mittelbach, G. G., an L. Persson The ontogeny of piscivory an its ecological consequences. Canaian Journal of Fisheries an Aquatic Sciences 55: Olson, M Ontogenetic niche shifts in largemouth bass: variability an consequences for first-year growth. Ecology 77: Persson, A., an C. Brönmark Foraging capacities an

10 3022 ANDERS PERSSON AND CHRISTER BRÖNMARK Ecology, Vol. 83, No. 11 effects of competitive release on ontogenetic iet shift in bream (Abramis brama). Oikos 96, in press. Persson, L Asymmetries in competitive an preatory interactions in fish populations. Pages in B. Ebenman an L. Persson, eitors. Size-structure populations: ecology an evolution. Springer-Verlag, Berlin, Germany. Persson, L., P. Byström, E. Wahlström, A. Nijlunsing, an S. Rosema Resource limitation uring early ontogeny: constraints inuce by growth capacity in larval an juvenile fish. Oecologia 122: Persson, L., an L. A. Greenberg Juvenile competitive bottlenecks: the perch (Perca fluviatilis) roach (Rutilus rutilus) interaction. Ecology 71: Persson, L., K. Leonarsson, A. M. e Roos, M. Gyllenberg, an B. Christensen Ontogenetic scaling of foraging rates an the ynamics of a size-structure consumer-resource moel. Theoretical Population Biology 54: Peters, R. H The ecological implication of boy size. Cambrige University Press, Cambrige, UK. Popova, O. A., an L. A. Sytina Foo an feeing relations of Eurasian perch (Perca fluviatilis) an pikeperch (Stizosteion lucioperca) in various waters of the USSR. Journal of the Fisheries Research Boar of Canaa 34: Romare, P., an E. Bergman Juvenile fish expansion following biomanipulation an the effect on the preation pressure on zooplankton. Hyrobiologia 404: Scharf, F. S., J. A. Buckel, F. Juanes, an D. O. Conover Preation by juvenile piscivorous bluefish (Pomatomus saltatrix): the influence of prey to preator size ratio an the prey type on preator capture success an prey profitability. Canaian Journal of Fisheries an Aquatic Sciences 55: Schoener, T. W Moels of optimal size for solitary preators. American Naturalist 103: Turesson, H., A. Persson, an C. Brönmark Prey size selection in piscivorous pikeperch (Stizosteion lucioperca) inclues active prey choice. Ecology of Freshwater Fish, in press. Van Densen, W. L. T., W. Ligtvoet, an R. W. M. Roozen Intra-cohort variation in the iniviual size of juvenile pikeperch, Stizosteion lucioperca, an perch, Perca fluviatilis, in relation to the size spectrum of their foo items. Annales Zoologica Fennici 33: Wahlström, E., L. Persson, S. Diehl, an P. Byström Size-epenent foraging efficiency, cannibalism an zooplankton community structure. Oecologia 123: Werner, E. E Size, scaling, an the evolution of complex life cycles. Pages in B. Ebenman an L. Persson, eitors. Size-structure populations: ecology an evolution. Springer-Verlag, Berlin, Germany. Werner, E. E Ontogenetic scaling of competitive relations: size-epenent effects an responses in two anuran larvae. Ecology 75: