Ethics Statement

Ethical approval was given by the ethical committee at Institute of Hydrobiology, Chinese Academy of Science. In the current study, the use of tissue material from animals killed is a part of routine commercial fishery production. The commercial fishery complies with the local fishery law in Jiangsu Province. No specific permits were required for the described field studies. The location studied is not privately-owned or protected in any way and the field studies did not involve endangered or protected species.

Sampling

The study was conducted on the eastern (predominantly macrophyte-covered) side of Gonghu Bay (31°25′N; 120°18′E), Lake Taihu China, between April and November 2005. Three sites were randomly selected as replicates and were separated from each other by at least 3000 m. The dominant predatory fish is yellow catfish (Pelteobagrus fulvidraco, Richardson, 1846), whose prey organisms are zooplankton, benthic macroinvertebrates, freshwater shrimps and small fishes [11], [12]. Yellow catfish and its prey were censused at these sites once every month during the investigation period ( April to November 2005). Fish were sampled using trawlnets (each 2 m×2 m, mesh size = 7.0 mm, warp length = 3 m) towed at each sampling site with a net speed of approximately 4000 m per hour. All fish were sorted aboard the vessel and were counted and weighted. The freshwater shrimps (Exopalaemon modestus and Macrobrachium nipponensis), and prey fish, such as Pseudorasbora spp., Rhinogobius spp., and Odontobutis spp., were also counted and weighted. This size and type of fish are the main prey items of yellow catfish in this lake [11], [12]. Benthic macroinvertebrates, including the snail Bellamya aeruginosa, and the filter-feeding mussel Cristaria plicata and Anodonta woodiana woodiana, were collected.

Abundance of zooplankton and benthic invertebrates were also assessed in each site. Zooplankton (cladoceran and copepods) were collected using a plankton net (mesh size 112 µm) by filtering 10 L lake water into acid washed PVC bottles (50 ml) and preserved with Lugol solution prior to laboratory identification. Benthic macroinvertebrates were sampled by collecting surface sediments using an Ekman grab (1/40 m²), washed with a hand net (mesh size 1 mm) and preserved with 95% ethanol for laboratory identification.

All individuals of yellow catfish, at least five individuals of prey fish and freshwater shrimps of yellow catfish, grazing snail (B. aeruginosa) and filter-feeding mussels (C. plicata and A. woodiana woodiana) were collected per site for stable isotope analysis. Samples were placed on ice and transported to laboratory. For fishes and shrimps, white muscle tissues were dissected in the laboratory. These tissues are representative of the overall stable isotope signature in fish (Hesslein et al. 1993). A sample of foot muscle tissue of molluscs was in the laboratory taken for stable isotope analysis since the shell material is enriched in ¹³C and does not reflect what is actually assimilated by the snail (Mitchell et al., 1996). Muscle samples were oven-dried to a constant weight at 60°C and then ground to a fine homogeneous powder with a mortar and a pestle. The mortar and pestle were acid-washed and dried to prevent cross-contamination between samples. The powdered samples were kept in acid-washed glass tubes and sealed in desiccators with silica gel for future analysis [13].

Analysis of stable isotopes

The δ¹³C and δ¹⁵N values were generated after analysis with a Delta Plus (Finnigan, Bremen, Germany) continuous-flow isotope ratio mass spectrometer coupled to a Carlo Erba NA2500 elemental analyzer (Carlo Erba Reagenti, Milan, Italy). Stable isotope ratios were expressed as parts per thousand (‰) deviation from the international standards according to the equation: δX = [(Rsample/Rstandard)−1]×1000, where X is ¹⁵N or ¹³C and R is the corresponding ratio ¹⁵N/¹⁴N or ¹³C/¹²C. δ is the measure of heavy to light isotope in the sample, whereby higher δ values denote a greater proportion of the heavy isotope. The standard for nitrogen was atmospheric nitrogen and that for carbon was Vienna Pee Dee belemnite. The reference material for δ¹⁵N was ammonium sulfate (IAEA-USGS25), and that for δ¹³C was carbonatite (IAEA–NBS18), supplied by the U.S. Geological Service (Denver, Colombia, USA) and certified by International Atomic Energy Agency (Vienna, Austria). On a daily basis, an internal working standard, urea (δ¹⁵N = −1.53‰; δ¹³C = −49.44‰) was employed. Twenty percent of the samples were run in duplicate; the average standard errors of replicate measurements for δ¹³C and δ¹⁵N were both less than 0.3‰.

Trophic niches calculation

According to the previous gut content analysis [11], [12] (See also Table S1 and Table S2), dominant food items of yellow catfish were classified into three guilds: zooplankton, benthic primary invertebrates, and shrimps and fishes, in order to effectively calculate the feasible contribution to each individual of yellow catfish. Benthic primary consumers included animals such as Oligochaeta, Chironomidae, and snails. We estimated planktivory, benthivory, and piscivory, defined as the estimated reliance on planktonic and benthic primary consumer resources and carnivorous resources of intermediate consumers in the food web, using δ¹³C and δ¹⁵N data and the SIAR (Stable Isotope Analysis in R) software (Version 2.13.1). SIAR is based on a Bayesian approach that estimates possible distributions of resource contributions to a consumer diet by accounting for all uncertainties of the input data [14]. The effects of differing isotopic compositions over time were eliminated by this calculation from the matrices of stable isotopes of end members. To account for trophic fractionation we used fractionation factors and uncertainties of 0.4±1.3‰ for δ¹³C and 3.4±1.0‰ for δ¹⁵N [15]. For running the mixing models, the appropriate number of iterations (up to 5×10⁵) was chosen according to SIAR's convergence diagnostic. The procedure then provided information about the range and distribution of possible source contributions. From the resulting up to 30 000 dietary proportions, we selected the maximum feasible contribution of food sources to each individual from the top distribution in histograms. The sum of combinations of each source contribution was 100%.

With respect to this stable isotope methodology, the relatively slow stable isotope turnover in yellow catfish compared to their planktonic and benthic primary consumer resources is a potential concern. We thus used stable isotope data of the grazing snail, B. aeruginosa, and the filter-feeding mussels, C. plicata and A. woodiana woodiana, as the end members for planktivory and benthivory calculations in SIAR. These data were used as estimates of niche shifts between planktonic and benthic primary consumer resources [15], and the results in the mixing model could then reduce the bias caused by stable isotope values of zooplankton and small benthic invertebrates reflecting their feeding on a shorter time period than yellow catfish. For the end members of piscivory calculations in SIAR, we pooled the stable isotope values of the potential carnivorous resources of intermediated consumers in the current food web, including freshwater shrimps, Exopalaemon modestus and Macrobrachium nipponensis, and prey fish (Pseudorasbora spp., Rhinogobius spp., and Odontobutis spp.).

We calculated the predator/prey ratios (PPR) between the predator, yellow catfish, and its prey, zooplankton (PPR_z), benthic invertebrates (PPR_b), prey fishes and shrimps (PPR_f), respectively.

Statistical analysis

We used analysis of variance (ANOVA) for the comparisons on δ¹³C (and δ¹⁵N) of end-members of prey sources to test the utility of stable isotope analysis for determining the energy base of the predator fish in time. Type of end-members is used as a fixed factor and time as a random factor. If δ¹³C and δ¹⁵N of end-members of prey sources are not distinct, the mixing model (SIAR) is not appropriate. Significance levels were α = 0.05. Bonferroni post-hoc test were used to compare among the prey sources.

We tested whether planktivory, benthivory, and piscivory of yellow catfish varied over time using one way ANOVA, with date as a fixed factor and site as a random factor. A non-parametric smoothed curve (LOWESS) was used to search for functional relationships to see if there were any functional change in planktivory, benthivory, and piscivory with the size of yellow catfish.

The relationships between size of yellow catfish and their foraging niches (including planktivory, benthivory, and piscivory) was analyzed with one-way analysis of covariance (ANCOVA) with date as factor, planktivory, benthivory, and piscivory as variables, respectively, and body size as covariate.

The predator-prey ratio was calculated by dividing the total biomass of predator individuals with the total biomass of zooplankton, benthic invertebrates, and forage shrimps and fishes, respectively. All the data were log(1+x) transformed to improve homogeny in the dataset. To further test whether each type of predator-prey ratios differed over time, we used ANOVA with date as a fixed factor. We tested the relationships between biomass of the predator and its prey using linear regression analysis. We also tested the relationship between monthly mean foraging niche indices (including planktivory, benthivory, and piscivory) and each corresponding predator-prey ratio using linear regression analysis, with foraging niches as the explanatory variables and predator-prey ratios as response variables. Seasonal variation in PPR between yellow catfish and its prey was tested using ANOVA.

To separate the combining effects of growth and prey availability, we carried out a progress of statistics. We first extracted trophic niche residuals after the LOWESS analysis (to remove the effects of the predator growth), including planktivory, benthivory, piscivory residuals. We then used Pearson correlation analysis between the residuals and seasonal variations of the prey biomass. This two step is similar to partial correlation analysis. In probability theory and statistics, partial correlation measures the degree of association between two random variables, with the effect of a set of controlling random variables removed. All Statistical analyses were performed with R 2.13.0 [16].

Resource availability and seasonal trophic niche shift

Seasonal heterogeneity of prey abundance and availability plays a role in determining relative prey profitability, because the distribution of prey has a strong effect on energetic gains and costs of foraging [24], foraging success, and overall predator performance [25]. In communities where predators and prey have coexisted for long time, such as those in the current study, predators often respond to prey by developing specific foraging strategies to track the rapidly changing prey patterns in time and space [26], [27]. During the initial period of the summer season, densities of zooplankton and benthic invertebrates decreased due to the intense predation pressure. In such situations, predators are in need for improved hunting efficiency which leads to greater suppressive effect on the large size preys. Our results show that the changes in the foraging niches of yellow catfish imposed intense population suppression on prey fishes late in the season, mirrored in a high predator/prey fishes ratio.

Size-related foraging niche adaptation

Trends in foraging niches of yellow catfish along a size gradient, including planktivory, benthivory and piscivory were found in this study. On a percentage basis, planktonivory of yellow catfish was high and constant before the size of approximately 120 mm, after which it decreased to a low level. Similarly, benthivory decreased with size and reached a steady low level at sizes above 120 mm, while piscivory increased steadily with body length. Most aquatic food web studies associated with body size reveal that larger fish have higher trophic positions because they are able to consume larger prey [28], [29]. Changes in morphology, behavior, habitat, and distribution during the ontogeny have been suggested to be associated with changes in feeding strategies of organisms [30], [31]. For yellow catfish, one potential explanation for the size-related foraging niche adaptation is the efficient use of prey with different body size, such as zooplankton, zoobenthos and secondary consumers, for example decapods and small fishes. In the current study area, zoobenthos weights were, on average, more than two orders of magnitude larger than zooplankton weight [32]. Furthermore, secondary consumers prey, such as decapods and small fishes, had average weights more than two to three orders of magnitude larger than zoobenthos. The existence of large-sized prey translates into greater foraging profitability for P. fulvidraco feeding on secondary consumers relative to zooplankton and zoobenthos, although prey abundances and availability undoubtedly also play a role in determining relative prey profitability as discussed above. This energetic benefit of feeding on large prey is consistent with studies based on optimal foraging theory of of other fish species [5]. Combining the analysis of fish stomach contents and the stable isotope analysis has demonstrated that body size of predators and their prey are significantly correlated with each other, but not with trophic level [33].

Changes of top-down effects associated with foraging niche shift

Empirical studies from a variety of systems indicate that trophic cascades are widespread, although many factors regulate their occurrence [34]. In natural food webs, trophic interactions are complex, which could reduce the strength and penetrance of trophic cascades [35]. However, strong trophic interactions can lead to trophic cascades in complex systems and further influence the properties of the systems [34], [36]. Moreover, also large-scale environmental changes, such as climate change and brownification, may affect the strength of the feeding links in aquatic ecosytesms [37].

Yellow catfish is a functionally dominant species because it is the only top predator species present in the study area during our sampling period. Prey of yellow catfish, including small fishes and shrimps, commonly feeding on algae, zooplankton and benthic invertebrates, and therefore overlap in food choice with yellow catfish [11], [12], [38]. Foraging niche shift by the top predator thus affect the functional response of the prey and its food. As the predator/prey ratios indicate, the foraging niche of yellow catfish shifted from zooplankton and zoobenthos in spring to secondary consumer preys in summer and resulted in a compensatory increase in abundances of zooplankton and zoobenthos in summer. The consequence was that total planktivory and benthivory actually decreased as piscivory of yellow catfish increased over the study period, which was mirrored in a predation-driven change in the abundance of fish prey. The increased predatory pressure on secondary consumers from yellow catfish thus facilitated for primary consumers both in planktonic and benthic food webs through top-down control. However, seasonal zooplankton dynamics, and the mechanisms driving their variability, are highly susceptible to changes in environmental conditions. For example, other filter-feeding and highly efficient zooplanktivorous fish species may also generate high predation pressure on zooplankton, and seasonal dynamics in nutrent dynamics may cause seasonal changes in phytoplankton productivity and standing stock, which affects zooplankton foraging and subsequent biomass development [39], [40]. Although our results cannot exclude alternative bottom-up and top-down processes affecting the dynamics of primary consumers, the statistical results suggested that planktivory and piscivory of yellow catfish were significantly negatively and positively correlated with abundance of prey fishes and shrimps, respectively. In addition, the marginally significant correlation of planktivory of yellow catfish with zooplankton biomass suggests that zooplankton abundance was governed partially by the predation from yellow catfish. However, these correlation does not necessarily implies cause-effect. For example, no relationship was found between invertebrate prey abundance and benthivory of yellow catfish, and thus the low number of yellow catfish in this lake might not act alone to control the changes in prey abundance.

In freshwater lakes, top-down effects from piscivorous fish can reduce the predation pressure on zooplankton by preying upon planktivorous fish, with cascading effects on the phytoplankton [41]. In addition, consumption of benthic prey by fish is thought to subsidize top-down control in the planktonic food web in some systems [42]. In the present study, the increase in piscivory by yellow catfish leads to another important consequence, the uncoupling of the predator-prey between the yellow catfish and the primary consumers (zooplankton and zoobenthos). It has been suggested that the top-down effect is generally weak in lakes situated in warmer regions around the world [43], because the high abundance, small size and multiple reproductive events of several native omnivorous fishes result in a high predation pressure on primary consumers year around [44], [45]. However, we show here that despite Lake Taihu is a sub-tropical lakepredatory catfish control the abundance of the intermediate consumers, such as prey fishes and shrimps, which allows the development of primary consumers through seasonal niche shifts.

Yellow catfish is an important commercial fish species in Lake Taihu and the preferred catchable size is typically larger than 120 mm, i.e. in the piscivorous stage. Continued fishing on yellow catfish from September and the following months reduces the portion of large size classes resulting in less predation on intermediate consumers, such as prey fish and shrimps, which reach maximum abundances at the end of the growing season. The consequence is a recovery of intense predation on zooplankton and zoobenthos, leading to a the rapid decrease of their abundances and, consequently, intesified algal blooms.

In summary, yellow catfish, a generalist predator, is capable of seasonally responding to local changes in prey abundance and availability. These response patterns were consistent with foraging niche shifts of yellow catfish as indicated by stable isotope signatures. The foraging niche shift of yellow catfish can be regarded as an adaptation to match prey abundance and availability at a temporal scale. Size-related foraging niche shifts of this species are also an important adaptation to resource use during its ontogeny. The season- and size-specific nature of these interactions is of great importance for the local structure and function of the ecosystem, and results in the seasonal dynamics of trophic cascades. Further studies of the impact of niche shift on resource and community dynamics in this eutrophic lake are necessary to better understand this complicated issue and to provide a better framework for management and conservation.