Diet composition and food selectivity of 0-group herring (Clupea harengus L.) and sprat (Sprattus sprattus (L.)) in the northern Baltic Sea

Transcription

1 ICES Journal of Marine Science, 53: Diet composition and food selectivity of 0-group herring (Clupea harengus L.) and sprat (Sprattus sprattus (L.)) in the northern Baltic Sea Fredrik Arrhenius Arrhenius, F Diet composition and food selectivity of 0-group herring (Clupea harengus L.) and sprat (Sprattus sprattus (L.)) in the northern Baltic Sea. ICES Journal of Marine Science, 53: Diet composition and prey selectivity of the major Baltic Sea zooplanktivores, 0-group herring (Clupea harengus) and sprat (Sprattus sprattus), were studied in a coastal area. Diets were dominated by zooplankton, mainly cladocerans (Bosmina longispina maritima and Pleopsis polyphemoides) and copepods (Acartia spp. and Eurytemora affinis hirundoides/temora longicornis). There were differences in order of the prey selected by the two different 0-group predators (fish hatched during the sampling year) when analysing each species separately. The general order of prey preference for herring was Acartia>Eurytemora>cladocerans>Pseudocalanus minutus>copepod nauplii and for sprat Eurytemora>cladocerans>Acartia>Pseudocalanus>copepod nauplii. As they grew 0-group clupeoids selected larger prey; the greatest preference was shown for intermediate sized prey ( mm in cephalothorax length) International Council for the Exploration of the Sea Key words: herring, sprat, Baltic Sea, selectivity, diet. Received 18 May 1995; accepted 30 August F. Arrhenius: Department of Systems Ecology, Stockholm University, S Stockholm, Sweden. Introduction Factors affecting fish stock recruitment often operate in the early life stages that have not yet entered the fishery (Cushing, 1975; Smith, 1985). Mortality induced directly by starvation, or indirectly by retarding growth rate and thus increasing vulnerability to predation, can strongly influence recruitment and population variability of major fish stocks (Crowder et al., 1987; Miller et al., 1988). A knowledge of feeding ecology of young fish is therefore important in understanding year-class variation. Herring (Clupea harengus L.) and sprat (Sprattus sprattus (L.)) are dominant species both in the commercial fishery (Anon., 1994) and as zooplanktivores in the Baltic Sea. In particular, the young stages have been suggested as having a major impact on the zooplankton community structure (Hansson et al., 1990; Rudstam et al., 1992, 1994; Arrhenius and Hansson, 1993). Studies on the feeding of herring and sprat (Sandström, 1980; Raid, 1985; Hansson et al., 1990; Flinkman et al., 1992; Rudstam et al., 1992) show that they are selective, which may influence the zooplankton species composition (cf. Brooks and Dodson, 1965). In the open sea, Sandström (1980) and Flinkman et al. (1992) showed that herring prefer the largest zooplankton prey and late copepodite stages of calanoid copepods. In the coastal area, the most preferred prey of 0-group (fish hatched during the sampling year) herring and sprat were cladocerans, mainly Bosmina and Pleopsis in the northern part (Rudstam et al., 1992) and Eurytemora affinis in the southern Baltic (Mehner and Heerkloss, 1994). Particle size has been the primary variable considered in studies of selectivity by young fish, when prey size has been related to the gape size of the fish (Blaxter and Hunter, 1982; Checkley, 1982; Cohan and Lough, 1983; Schael et al., 1991; Graham and Sprules, 1992). For young fish that are larger than larvae, non-random consumption of available prey might result from differences in prey visibility or capture rates and differences in preference by the fish e.g. as a result of optimal foraging (Werner and Hall, 1974; Pyke et al., 1977). From earlier studies (Arrhenius and Hansson, 1993, 1994a,b), we know that 0-group herring and sprat occur together in the coastal area of the northern Baltic proper, and, since they appear to be able to influence their common zooplankton prey resource, they probably compete for food. The objective of this study is to describe in more detail the diet and feeding preferences of 0-group herring and sprat /96/ $18.00/ International Council for the Exploration of the Sea

2 702 F. Arrhenius Materials and methods Fish were sampled in JulyNovember , using small charges of explosives (15120 g of Primex 17 mm, Nitro Nobel AB). A subsample of the 0-group fish caught was immediately preserved in 70% ethanol in 1992 and deep-frozen ( 18 C) in Fish were sampled in the evening, when the stomach content is highest (Raid, 1985; Arrhenius and Hansson, 1994a,b). All sampling was carried out in the coastal area at about 58 N, 17 E in the northern Baltic proper. On each sampling occasion and from the same water column where the fish were caught, a zooplankton sample was taken with a 90 μm WP-2 net. It was assumed that the zooplankton used by the fish was found in the depth interval from where the fish were caught up to the surface, i.e. the water column sampled with the zooplankton net. The plankton net was towed vertically at a speed of 0.5 m s 1. Samples were preserved in 4% buffered formaldehyde solution. Before counting under an inverted microscope, the zooplankton samples were subsampled (Kott, 1953) and at least 500 specimens from each sample were identified to the lowest possible taxonomic level. Zooplankton biomasses were estimated from values on individual wet weight (WWT) (Hernroth, 1985), of which 13% was assumed to be dry weight (DWT) (Mullin, 1969). In the laboratory, the total length of each fish was measured to the nearest mm and WWT was determined to the nearest 1 mg. Length and weight data were corrected for effects of ethanol and deep-freeze preservation, using correction equations from Arrhenius and Hansson (1995). Diet analyses were made on 10 fish per sampling day and length interval (width 5 mm, Tables 1, 2). Each individual stomach was cut open, and the complete contents collected and analysed using a stereo microscope and an inverted microscope. Each prey was determined to the lowest possible taxonomic level. If a stomach contained a large number of prey, a subsample of about 100 identifiable items was analysed. The contents of a stomach were expressed as the percentages of different taxa, calculated from the number of identified items. Selection of prey types were estimated using a selectivity index (Chesson, 1983): where r i is the proportion of food item i in a stomach, e i is the proportion of prey belonging to this category in the environment, and k is the number of prey categories. α i ranges from 0 to 1, corresponding to complete avoidance and full selection. A value of 1/k (in this dataset 1/k=0.167 as there were six prey taxa) indicates neutral selection and positive and negative selection refer to α-values higher and lower than 1/k. The prey selectivity for fish may change with its size, which has to be considered when analysing for differences in selectivity between sprat and herring and between years. This was checked using the following procedure: for each length group (lg, width 5 mm) of fish, the average α-value was calculated for each prey species, and observations from each fish were classified as above or below this average. Frequencies of observations above and below the average were then analysed statistically (see below). The mean α-values (α lg -used in these analyses were calculated as moving averages from a 15 mm wide interval, using the equation: Three different sets of α lg -values were calculated. Two sets were derived by calculating α lg -values separately for herring and sprat and the third was a set of α lg -values in which both fish species were amalgamated. To test for differences in prey selectivities between years α-values for each fish were classified as above or below the α lg -values. Here, herring and sprat were analysed separately, using the two species-specific sets of α lg. For each species, the number of α values below and above the α lg -value were counted separately for 1992 and 1993 and frequencies were analysed with a χ 2 -technique (Sokal and Rohlf, 1981, 2*2 table with the classes above/ below as rows and 1992/1993 as columns (Figs 2, 3)). To determine if there were differences in prey selectivity between herring and sprat, an analysis on the α lg - values, calculated for the species combined was used. Again, α-values for each species were classified as above or below the α lg -value. Resulting counts were used to create a three dimensional frequency table (2*2*6), with the following categories: position in relation to α lg (above/below), species (herring/sprat) and sampling date (six occasions when both herring and sprat were caught in sufficient numbers). When generating this frequency table, only length groups in which both sprat and herring occurred were included. This frequency table was then analysed with a log-linear model (Upton, 1978; SPSS, 1993), with the position in relation to α lg as a response variable. The first step in this log-linear analysis was to generate a saturated model, in which all possible interactions between classification categories were included. This model has 0 degrees of freedom and explains all variation in counts in the frequency table (χ 2 -value=0). In the following steps, interactions between categories were removed, resulting in models that were less efficient in explaining the observed frequencies. The goal is to find interactions that significantly explain observed frequencies, i.e. the exclusion of

3 Diet composition and prey selection in herring and sprat 703 Table 1. Clupea harengus. The only prey found in the stomachs were zooplankton; therefore, only the proportions of identified zooplankton taxa (% by numbers) are shown. Proportions were calculated as the average of proportions in individual fish. Fish length interval, number of stomachs analysed (n), and the total number of identified prey items are also given. Prop.: proportion; zoopl.: zooplankton; nau.; copepod nauplii, ad&cop.: adult and copepodites; E/T: Eurytemora/Temora; A: Acartia; Bos: Bosmina; Ple: Pleopsis; Ps: Pseudocalanus. Capture Date Depth Time Fish length group Fish wet weight No. of identified Copepoda (ad. & cop.) Cladocera No. (mm) (g) s.d. n prey Nau E/T A Ps Bos Ple Other Zoopl Jul 3 m m a,b a,b a,b 5 Aug 10 m a a a Aug 15 m a Sep 20 m Oct 35 m Jul 10 m f b,c,d f b,c a,b 2 Aug 10 m f f c,e c,e 24 Aug 10 m Oct 10 m Nov 30 m a f a f a f a a Primarily zooplankton eggs. b Primarily Keratella spp. c Primarily Podon spp. d Primarily Lamellibranchiata. e Primarily Gastropoda. f One (1) stomach empty. Total no. of fish (n)=374.

4 704 F. Arrhenius Table 2. Sprattus sprattus. As for Table 1. Capture Date Depth Time Fish length group Fish wet weight No. of identified Copepoda (ad. & cop.) Cladocera No. (mm) (g) s.d. n prey Nau E/T A Ps Bos Ple Other Zoopl Aug 15 m Sep 20 m Oct 35 m Aug 10 m Oct 10 m Nov 30 m Total no. of fish (n)=203. which has significant negative effects on the remaining model (Upton, 1978). As the category above/below average was treated as a response variable, it was only the independent variable, i.e. species and sampling date, that could be excluded from interactions. Results Food availability The six most common taxa in stomachs and zooplankton samples were used in this investigation (Table 3). Zooplankton samples were dominated by copepods, among these Eurytemora affinis hirundoides/temora longicornis (E/T) and Acartia spp. (Acartia bifilosa, and/or A. longiremis) constituted between 2968% by numbers (8197% of copepodite and adult copepod abundance). Copepod nauplii and E/T were not divided into taxa due to difficulties in distinguishing different species in stomach analyses. The cladocerans Bosmina longispina maritima and Pleopsis polyphemoides constituted 024% of the abundance. Food composition A total of 374 herring and 203 sprat stomachs were analysed (Tables 1, 2). Herring were caught at all sampling occasions, while sprat did not appear until the end of August in either year. The only prey found in the stomachs were zooplankton, and these were totally dominated by crustaceans (Tables 1, 2). Among cladocerans, Bosmina and Pleopsis were most common and among copepods Acartia and Eurytemora/Temora (E/T) and Pseudocalanus minutus dominated. Copepod nauplii were abundant in zooplankton samples but less common in the stomachs. Other food items, mostly rotifers, eggs, Balanus nauplii, cladoceran Podon spp. and lamellibranchs were only present in the stomachs in small numbers. The size of the prey found in the stomachs increased with sampling dates and fish size in both years for herring (Fig. 1). For larger fish a size preference was found for intermediate sizes of copepodite stage IVV (size range mm, in Table 4). For sprat, zooplankton mm in cephalothorax length were

5 Diet composition and prey selection in herring and sprat 705 Table 3. Abundance (ind m 2 ) of dominant cladocerans and copepods in zooplankton samples in JulyNovember A: Acartia spp., E/T: Eurytemora affinis hirundoides/temora longicornis, Ps: Pseudocalanus minutus elongatus, Bos: Bosmina longispina maritima, Ple: Pleopsis polyphemoides, Nau: Copepod nauplii. Eurytemora and Temora are merged in this table, as they are difficult to separate in stomach contents from fish. Year Date E/T a A a Ps a Bos Ple Nau Jul Aug Aug Sep Oct Jul Aug Aug Oct Nov a Adults and copepodite stages. Table 4. Mean sizes (mm) of different stages of zooplankton found in stomachs of 0-group herring and sprat. Copepods were measured from the tip of the head to the end of the abdomen. For copepods, each copepodite stage was not treated separately but grouped into IIII, IVV, and adult. Among cladocerans, only the adult stage were measured. Data are average values from own measurements and values from the northern Baltic proper (Hernroth, 1985). Taxa Species Nauplii IVI (mm) Copepodite stage IIII (mm) IVV (mm) Adult (mm) Copepods Acartia sp Eurytemora affinis Pseudocalanus minutes elongatus Cladocerans Bosmina longispina maritima Pleopsis polyphemoides the most abundant (6082%) prey size in the stomachs (Fig. 2). Selectivity Herring generally had a positive selection for Acartia, Eurytemora/Temora (E/T) and Bosmina, with Acartia the most preferred. Pseudocalanus, Pleopsis and copepod nauplii were usually below average (i.e. selected against) (Fig. 3). There were, however, differences between the years. Herring had a higher selectivity for Acartia, and copepod nauplii in 1992 compared with 1993, and for E/T and Pleopsis it was the opposite. Sprat showed a similar selectivity to herring, preferring copepods, but with a higher selectivity for E/T than for Acartia (Fig. 4). As in herring, sprat had low selectivity for copepod nauplii, but no consistent selection trends for Bosmina, Pseudocalanus, and Pleopsis. Sprat showed a higher selectivity for Acartia in 1992 than in 1993, while the opposite was true for E/T and Pleopsis. There was a shift in preference for several prey taxa as the fish increased in size (Figs 3, 4). In addition to the taxa diet changes, the size (length) category of zooplankton most highly represented in fish diets also changed with fish development (Figs 1, 2). The smallest herring (<49 mm in length) selected cladocerans (preferably Bosmina, mm in length) and whereas larger fish selected copepods (mostly 0.7 mm in length). There was also a tendency for a shift between years, with an increased preference for Acartia and decreased preference for Bosmina with body size in 1992 and vice versa in 1993 for herring. For sprat, there was a trend to even lower negative selectivity for copepod nauplii ( mm in length) with increasing body size. There was a significant difference between both years in selectivity

6 706 F. Arrhenius Percentage Fish length Percentage Fish length Jul 16 Aug 5 Aug 26 Sep17 Oct Aug 26 Sep 17 Oct Percentage Fish length Percentage Fish length Jul 15 Aug 2 Aug 24 Oct 6 Nov Aug 24 Oct 6 Nov 2 30 > 1.0 mm mm mm mm Figure 1. Clupea harengus. Mean percentage (by numbers) of zooplankton prey found in 0-group fish in 1992 and 1993 divided into four different size-classes. Data are compiled from 10 fishes from each sampling period. The average fish size (mm) each sampling period is also shown. for prey species, except Pseudocalanus for both fish species and Bosmina for herring (Figs 3, 4). For all prey species except Acartia, significant threeway interactions were present between the studied variables in log-linear analysis, indicating that the model without the three-way term does not fit well (Table 5). However, log-linear statistics were used for a description of the relationship between the data and served as a good starting point for exploring the data. Therefore, by partial testing it is still useful to fit models differing only in the presence of the effect to be tested (SPSS, 1993). According to the models, by removing the effect of sampling day (pos*sampspec*samp in Table 5) there were significant intraspecific differences in selectivity between herring and sprat for the copepods Acartia, E/T, and Pseudocalanus, but not for cladocerans and copepod nauplii. However, for all prey types, intraspecific differences between fish species in selectivity were significant between sampling dates (pos*spec spec*samp in Table 5). > 1.0 mm mm mm mm Figure 2. Sprattus sprattus. See Figure 1. Discussion Diet composition 0-group herring and sprat fed exclusively on zooplankton. Prey species preferences were in agreement with other studies on young clupeoids in the Baltic Sea (van Khanh et al., 1974; Hudd, 1982; Parmanne and Sjöblom, 1984; Raid, 1985; Lankov, 1986; Franek, 1988; Rudstam et al., 1992). No benthic or epibenthic species, such as mysids and amphipoids were found in the stomachs in this study, although such prey have been observed in herring over 65 mm elsewhere (Raid, 1985; Lankov, 1986). Prey selectivity by fish species The three-factor interaction term (sampling date, fish species and position e.g. if the predator have lower selectivity than the other, selectivity is below the combined moving average, and resulted in less selectivity for the prey) were significant for all prey species, except Acartia, in log-linear statistics. This indicates that the model without the third-order term does not fit well. Therefore, it is difficult to draw any definite conclusions,

7 Diet composition and prey selection in herring and sprat Eurytemora/Temora Acartia Chesson index Pseudocalanus NS Bosmina NS Pleopsis Copepod nauplii Length interval (mm) Figure 3. Clupea harengus. The average selectivity indices (Chesson, 1983) for different length groups (Table 1) and years for six different prey taxa. The curve shows the mean α-values (α lg ) calculated from indices of all fishes and both years and the dotted line indicate the Chesson index value for no selection. The lowest and highest length interval were excluded. The 2*2 tables are the numbers of fish above and below the moving average line divided by 1992 and 1993 respectively. The χ 2 tests were performed and the significances are indicated above the tables (, p<0.001; **, p<0.01; NS=not significant p>0.05). The cladoceran Bosmina had fewer datapoints (n=291), as indices were not considered when prey were absent from both stomachs and zooplankton samples. =1992; V=1993. even when there were significant difference between the two 0-group herring and sprat of the same size for the copepods Acartia, Eurytemora/Temora and Pseudocalanus, shown by two-factor interactions in log-linear statistics (Table 5). However, there may still be a difference in selectivity, but this could either be masked by the higher order complexity and/or be explained by sampling error of zooplankton. The causes of the latter problem may be two-fold; (1) Were the zooplankton samples representative for the zooplankton fauna? On each sampling occasion, only one sample was taken, and it has been shown that the coefficient of variation is about 30% between replicate zooplankton net hauls in the area (Johansson et al., 1993). (2) Were the samples taken where and when the fish had actually been feeding? Selectivity esimates could be biased if the

8 708 F. Arrhenius Eurytemora/Temora Acartia Chesson index Pseudocalanus NS Bosmina Pleopsis ** Copepod nauplii Length interval (mm) Figure 4. Sprattus sprattus. See Figure 3. For the cladocerans, Bosmina and Pleopsis, there were only 142 and 161 datapoints, as indices were not calculated when species was absent both from stomachs and zooplankton samples. =1992; V=1993. zooplankton samples did not reflect the zooplankton abundances where the fish had been feeding. Effects of fish size on prey selectivity Both herring and sprat clearly preferred calanoid copepods (Acartia spp. and Eurytemora sp./temora sp., respectively) over cladocerans. This agrees with studies in other areas in the Baltic proper (Lankov, 1986; Mehner and Heerkloss, 1994), but is contrary to findings in our coastal area by Rudstam et al. (1992). However, there were changes in preference with fish size, from cladocerans and nauplii (size range, mm) among the smallest fish size classes, to copepods (size >0.4 mm) among larger fish. This variation in length of fish and prey influences the relationship between foraging selectivity and relative prey size as seen in other studies on clupeoids (Checkley, 1982; Cohan and Lough, 1983; Raid, 1985; Lankov, 1986; Wespestad and Moksness, 1989; Conway et al., 1991; Munk, 1992). In other planktivores prey size is also positively correlated with fish size (Mills et al., 1984; Bence and Murdoch, 1986; Confer and O Bryan, 1989; Shael et al., 1991). When fish increase in size, the size spectrum of ingested prey

9 Diet composition and prey selection in herring and sprat 709 Table 5. Result of the log-linear model analysis (SPSS, 1993) on the effect of species (herring/sprat) and sampling date (6 occasions) on prey preferences (preference index, above/below average). The six sampling dates were 26 Aug, 17 Sep, and 27 Oct 1992 and 24 Aug, 6 Oct, and 2 Nov In the highest order model (saturated), all the variables interact and the value of the χ 2 statistic is always 0. The next model includes all possible two-way interactions. This model provides a test of the hypothesis that the third order interaction terms are 0. Differences in χ 2 values between models are given in the partial testing columns, with the corresponding p-value. As seen from the table, three way interactions (combined and complex effects of fish species and sampling date on the prey selectivity of individual fish) were significant for all prey species but Acartia (tests statistics indicated with in the table). An example of interpretation of the table: when removing the three way interaction (pos*spec*samp) from the Acartia data set, the resulting model with all three two-way interactions did not produce a prediction significantly different from the saturated model (χ 2 =5.0, df=5, p>0.4). However, when removing the two-way interaction between species and position (pos*spec), i.e. the effects that fish species have on the selectivity, the resulting model produced predictions significantly different from the saturated model (χ 2 =99.5, df=6). The difference between the model including all two-way interaction was 94.5 χ 2 -units, with one degree of freedom lost. This χ 2 -value is highly significant and shows the strength of the pos*spec interaction. Prey species Log-linear models df a χ 2 value Partial testing χ 2 df a value p-value All b pos*spec*samp E/T pos*specpos*sampspec*samp pos*specspec*samp pos*sampspec*samp Acartia pos*specpos*sampspec*samp pos*specspec*samp pos*sampspec*samp Pseudocalanus pos*specpos*sampspec*samp pos*specspec*samp pos*sampspec*samp Bosmina c pos*specpos*sampspec*samp pos*specspec*samp pos*sampspec*samp Pleopsis d pos*specpos*sampspec*samp pos*specspec*samp pos*sampspec*samp Copepod nauplii pos*specpos*sampspec*samp pos*specspec*samp pos*sampspec*samp a df=degrees of freedom. b The saturated model is structurally the same for all prey and has, per definition 0 degrees of freedom and χ 2 =0. c Sampling date exclude 27 Oct 1992 and 2 Nov d Sampling date exclude 27 Oct increases, probably an effect of less gape-size limitation (Miller et al., 1988; Schael et al., 1991). The present data show preferences for intermediate sized prey (mostly copepodite stage IVV, mm), which agrees with others studies on herring (Munk, 1992) and other planktivores (Bence and Murdoch, 1986; Confer and O Bryan, 1989; Graham and Sprules, 1992). However, for adult herring, one of the largest zooplankton items in the Baltic, Limnocalanus spp. ( mm, Hernroth, 1985), but also adult stages of other copepods, has been shown to be selected (Sandström, 1980; Lankov, 1986; Flinkman et al., 1992). For small 0-group fish it is important to consider that the size of the mouth may restrict their possible diet, but for fish of the size ranges included in this paper (metamorphosed fish>35 mm), gape limitation is probably less significant. The size selectivity of fish may, however, be influenced by behavioural changes. It has been shown that herring may switch between particulate-feeding at low predator densities and filterfeeding at higher prey densities (Batty et al., 1986; Gibson and Ezzi, 1992). Such shifts could possibly explain much higher selectivity for calanoid copepods at lower zooplankton abundance for herring ( 60 mm in length in Fig. 3). Effects of prey characteristics on preferences Difference in selectivity for copepod species may be explained by differences in their vertical distribution and the day-time feeding of herring. 0-group herring appears near the surface at twilight, when stomach fullness is highest (Raid, 1985; Lankov, 1986; Arrhenius & Hansson, 1994a,b). Both Eurytemora/Temora and

10 710 F. Arrhenius Acartia shows diel vertical migration from deeper water during the day, and closer to the surface at night. This migration is, however, less pronounced for Acartia which generally occurs closer to the surface than E/T (Hansson et al., 1990). Such shifts could possibly explain differences in prey selectivity between 0-group herring and sprat. 0-group herring are closer to the surface than sprat during the main feeding period (pers. comm.; Ojaveer et al., 1981). Pseudocalanus also moves upwards at night, but have both night and day abundance maxima much closer to the bottom than other species (Hansson et al., 1990). The most preferred prey by 0-group sprat was Eurytemora/Temora, which is more active (personal observations in microscope) than the other copepods in the area. Contrast of prey against the background and escaping ability have been shown to influence feeding behaviour (Drenner et al., 1978; Checkley, 1982). The two species of Eurytemora and Temora were not separated in this study, but other studies on sprat have shown higher selectivity for Temora than Eurytemora (Shvetsov et al., 1983; Shvetsov and Rudneva, 1994). Visible egg sacs, carried by Eurytemora but not by Acartia, may also explain differences in selectivity for the two major calanoid copepods. For herring, visibility may explain the high selectivity for Bosmina (most prey were carrying eggs) during the summer, and switching to Eurytemora later in the season when cladocerans are absent (Fig. 3). In other parts of the Baltic, selective predation on larger individuals and females of copepods and cladocerans carrying eggs has also been shown for adult herring (Sandström, 1980; Flinkman et al., 1992). Ecological implications 0-group sprat and herring in the coastal waters preferred copepods and both species had similar diets, as have been shown in the North Sea (De Silva, 1973; Last, 1987). Therefore, competition could occur if the food supply was limiting. The possibilities of competition in the Baltic are lessened by differences in spawning and hatching times and location of the two predators. The spawning of sprat occurs in the open water and herring in the coastal part of the Baltic. Temora is of the most abundant calanoid copepods in the northern Baltic proper (Hernroth and Ackefors, 1979), and also the most preferred prey species by sprat in the open water (Shvetsov et al., 1983; Shvetsov and Rudneva, 1994). In late summer 0-group sprat immigrate to the coastal areas and mix with 0-group herring (Ojaveer et al., 1981), and Temora is partly replaced by Eurytemora in the upper 25 m in the water column closer to coast (Hernroth and Ackefors, 1979). However, 0-group herring have a much higher selectivity for Acartia, and this prey is the most abundant calanoid copepods in our investigation area (Johansson, 1992). Selection of certain prey may therefore reflect learning and earlier success in capture in different habitats by both predators. 0-group herring and sprat constitute a considerable proportion of the pelagic fish biomass and have a major impact on zooplankton production and community structure in late summer in this study area (Hansson et al., 1990; Rudstam et al., 1992; Arrhenius and Hansson, 1993). Direction of predation towards reproductive individuals and different primary prey species would effectively prevent a zooplankton species from achieving its maximum possible population growth rate. In a study by Arrhenius and Hansson (1995), 0-group herring were food-limited in the area and, consequently, peaks in biomass of different prey taxa could be of great importance for their growth and indirectly for their survival. The match-mismatch hypothesis by Cushing (1975) for early larval stages must also be applicable to later life stages although they have more resistance to starvation (Munk, 1992, 1993). Since at least 0-group herring are food-limited, and since predation probably influences zooplankton abundance, inter- and intraspecific competition (including also older fish) should influence growth rates of the young stages (juveniles invest most of their energy in somatic growth and less in fat storage than older fish (Blaxter and Hunter, 1982)). Predictions of growth rate from variables like zooplankton abundance and temperature are thus required for further understanding of recruitment variability in populations such as herring and sprat. As concluded by Anderson (1988), starvation and predation might be the two main factors in regulating fish recruitment. Acknowledgements I am grateful to Sture Hansson for statistical advice and valuable criticisms on earlier versions of the manuscript. I thank Dr Lars G. Rudstam, Prof. R. Elmgren, Prof. J. H. S. Blaxter and an anonymous referee for helpful comments on an earlier version of the manuscript. Björn Klinga helped in the field. Financial support was provided by the Swedish Natural Science Research Council, Hierta-Retzius foundation, Alice and Lars Siléns fund and Stockholm Centre for Marine Research. References Anderson, J. T A review of size dependent survival during prerecruit stages of fishes in relation to recruitment. Journal of Northwest Atlantic Fisheries and Sciences, 8: Anon Report of the working group on the assessment of Baltic fish. ICES CM 1994/Assess: 18, 149 pp. Arrhenius, F. and Hansson, S Food consumption of larval, young and adult herring and sprat in the Baltic Sea. Marine Ecology Progress Series, 96: Arrhenius, F. and Hansson, S. 1994a. In situ consumption by young-of-the-year Baltic Sea herring Clupea harengus: a test of predictions from a bioenergetics model. Marine Ecology Progress Series, 110:

11 Diet composition and prey selection in herring and sprat 711 Arrhenius, F. and Hansson, S. 1994b. Erratum re: Arrhenius, F., and Hansson, S Marine Ecology Progress Series 110: Marine Ecology Progress Series, 114: 314. Arrhenius, F. and Hansson, S Functional response by young-of-the-year Baltic Sea herring (Clupea harengus L.) Paper III. In Feeding ecology of Baltic herring (Clupea harengus L.) field and model studies of a dominant zooplanktivore. PhD thesis, Department of Systems Ecology, Stockholm University, Sweden. 120 pp. Batty, R. S., Blaxter, J. H. S., and Libby, D. A Herring (Clupea harengus) filter-feeding in the dark. Marine Biology, 91: Bence, J. R. and Murdoch, W. W Prey size selection by the mosquitofish: relation to optimal diet theory. Ecology, 67: Blaxter, J. H. S. and Hunter, J. R The biology of clupeoid fishes. Advances in Marine Biology, 20: Brooks, J. L. and Dodson, S. I Predation, body size and composition of plankton. Science, 150: Checkley, D. M. Jr Selective feeding by Atlantic herring (Clupea harengus) larvae on zooplankton in natural assemblages. Marine Ecology Progress Series, 9: Chesson, J The estimation and analysis of preference and its relationship to foraging models. Ecology, 64: Cohen, R. E. and Lough, R. G Prey field of larval herring Clupea harengus on a continental shelf spawning area. Marine Ecology Progress Series, 10: Confer, J. L. and O Bryan, L Changes in prey rank and preference by young planktivores for short term and long term ingestion periods. Canadian Journal of Fisheries and Aquatic Sciences, 46: Conway, D. V. P., Tranter, P. R. G., Fernandez de Puelles, M. L., and Coombs, S. H Feeding of larval sprat (Sprattus sprattus L.) and sardine (Sardina pilchardus Walbaum). ICES CM 1991/L: 76, 15 pp. Crowder, L. B., McDonald, M. E., and Rice, J. A Understanding recruitment of Lake Michigan fishes: The importance of size-based interactions between fish and zooplankton. Canadian Journal of Fisheries and Aquatic Sciences, 44: Cushing, D. H Marine Ecology and Fisheries. Cambridge University Press, London, 278 pp. De Silva, S. S Food and feeding habits of the herring Clupea harengus and the sprat C. sprattus in inshore waters of the west coast of Scotland. Marine Biology, 20: Drenner, R. W., Strickler, J. R., and O Brian, W. J Capture probability: the role of zooplankter escape in the selective feeding of planktivorous fish. Journal of Fisheries Research Board of Canada, 35: Flinkman, J., Vourinen I., and Aro, E Planktivorous Baltic herring (Clupea harengus) prey selectively on reproducing copepods and cladocerans. Canadian Journal of Fisheries and Aquatic Sciences, 49: Franek, D smelt (Osmerus eperlanus L.) and herring (Clupea harengus L.) in the food chain of the Barter Bodden. ICES CM 1988/J: 13, 16 pp. Gibson, R. N. and Ezzi, I. A The relative profitability of particulate- and filter-feeding in the herring, Clupea harengus L. Journal of Fish Biology, 40: Graham, D. M. and Sprules, W. G Size and species selection of zooplankton by larval and juvenile walleye (Stizostedion vitreum vitreum) in Oneida Lake, New York. Canadian Journal of Zoology, 70: Hansson, S., Larsson, U., and Johansson, S Selective predation by herring and mysids, and zooplankton community structure in a Baltic Sea coastal area. Journal of Plankton Research, 12: Hernroth, L Recommendations on methods for marine biological studies in the Baltic Sea. Mesozooplankton biomass assessment. Baltic Marine Biologists, Publ. no. 10, 32 pp. Henroth, L. and Ackefors, H The zooplankton of the Baltic proper. A long-term investigation of the fauna, its biology and ecology. Fisheries Board of Sweden, Institute of Marine Research. Rep. no. 2, 59 pp. Hudd, R Feeding of Baltic herring larvae (Clupea harengus L.) in the Gulf of Finland. Finnish Fisheries Research, 4: Johansson, S., Hansson, S., and Araya-Nunez, O Temporal and spatial variation of coastal zooplankton in the Baltic Sea. Ecography, 16: Kott, P Modified whirling apparatuses for the subsampling of plankton. Australian Journal of Marine and Freshwater Research, 4: Lankov, A The food of young Baltic herring Clupea harengus membras L. in the Gulf of Finland. ICES CM 1986/J: 4, 15 pp. Last, J. M The food of inmature sprat (Sprattus sprattus (L.)) and herring (Clupea harengus L.) in coastal waters of the North Sea. Journal du Conseil International pour l Exploration de la Mer, 44: Mehner, T. and Heerkloss, R Direct estimation of food consumption of juvenile fish in a shallow inlet of the southern Baltic. Int. Revue ges. Hydrobiol. 79: Miller, T. L., Crowder, L. B., Rice, L. A., and Marschall, E. A Larval size and recruitment mechanisms in fishes: toward a conceptual framework. Canadian Journal of Fisheries and Aquatic Sciences, 45: Mills, E. L., Confer, J. L., and Reeady, R. C Prey selectivity by young yellow perch: the influence of capture success, visual acuity, and prey choice. Transactions of the American Fisheries Society, 113: Mullin, M. M Production of zooplankton in the ocean: the present status and problems. Oceanography and Marine Biology Annual Reviews, 7: Munk, P Foraging behaviour and prey size spectra of larval herring Clupea harengus. Marine Ecology Progress Series, 80: Munk, P Differential growth of larval sprat Sprattus sprattus across a tidal front in the eastern North Sea. Marine Ecology Progress Series, 99: Ojaveer, E., Lindroth, A., Bagge, O., Lehtonen, H., and Toivonen, J Fishes and fisheries. In The Baltic Sea, pp Ed. by A. Voipio. Elsevier Scientific Publishing Co., Amsterdam. Parmanne, R. and Sjöblom, V The abundance of spring spawning Baltic herring larvae in the Seas around Finland in 1982 and 1983 and the correlation between the zooplankton abundance and the herring year class strength. ICES CM 1984/J: 18, 9 pp. Pyke, G. H., Pulliam, H. R., and Charnov, E. L Optimal foraging: a selective review of theory and tests. Quarterly Reviews in Biology, 52: Raid, T The reproduction areas and ecology of Baltic herring in the early stages of development found in the Soviet zone of the Gulf of Finland. Finnish Fisheries Research, 6: Rudstam, L. G., Hansson, S., Johansson, S., and Larsson, U Dynamics of planktivory in a coastal area of the northern Baltic Sea. Marine Ecology Progress Series, 80: Rudstam, L. G., Aneer, G., and Hildén, M Top-down control in the pelagic Baltic ecosystems. Dana, 10: Sandström, O Selective feeding by Baltic herring. Hydrobiologia, 69:

12 712 F. Arrhenius Schael, D. M., Rudstam, L. G., and Post, J. R Gape limitation and prey selection in larval yellow perch (Perca flavescens), freshwater drum (Aplodinotus grunniens), and black crappie (Pomoxis nigromaculatus). Canadian Journal of Fisheries and Aquatic Sciences, 48: Shvetsov, F. G., Starodub, M. L., and Sidrevits, L. L The daily feeding rhythm of the Baltic sprat, Sprattus sprattus balticus (Clupeidae). Journal of Ichthyology, 23: Shvetsov, F. G. and Rudneva, A The growth and feeding of sprat in eastern Baltic. ICES CM 1994/J: 10, 8 pp. Smith, P. E Year-class strength and survival of 0-group clupeoids. Canadian Journal of Fisheries and Aquatic Sciences, 42: Sokal, R. R. and Rohlf, F. J Biometry. 2nd edition. W. H. Freeman and Company, San Francisco, 859 pp. SPSS (Ed. by M. J. Norusis) SPSS: Advanced statistics. Release 6.0. SPSS Inc. 578 pp. Upton, G. J. G The Analysis of Cross-tabulated Data. John Wiley & Sons, London. 148 pp. van Khanh, N., Drycimski, I., and Chojnacki, J Feeding and food composition of sprat from Bornholm depth. Acta Ichthyologica et Piscatoria, 2: Werner, E. E. and Hall, D. J Optimal foraging and the size selection of prey by the bluegill sunfish (Lepomis macrochirus). Ecology, 55: Wespestad, V. G. and Moksness, E Observations on growth and survival during the early life history of Pacific herring Clupea pallasii from Bristol Bay, Alaska, in a marine mesocosm. Fishery Bulletin, 88: