WaterSubject

We investigated lethal and sublethal predation effects of the benthivorous gudgeon (Gobio gobio) on the grazing mayfly Rhithrogena semicolorata and the shredding amphipod Gammarus pulex over 21 months in a largely detritus-based small stream. We hypothesised that shredders are generally less vulnerable to fish predation and therefore less likely to be predation-controlled than grazers, because the latter are visible to the predators during their feeding on stone surfaces, while shredders may hide between leaves during foraging.

The hypothesis was tested in two 400 m experimental reaches of a natural stream, which were manipulated in order to contain and to lack fish, respectively. Biomass of G. pulex was significantly reduced in the fish section while that of R. semicolorata was not. Since approximately 91% of the annual production of G. pulex but only 12% of R. semicolorata production was consumed by gudgeon, the observed biomass difference of G. pulex is likely due to a lethal predation effect. However, no sublethal predation effects such as reduced concentration of storage components (triglycerides, glycogen) or reduced reproductive success were observed for both species. Lower mean body length of the R. semicolorata larvae in the fish section did not result in a lower number of eggs in the abdomen of the last instar larvae. Hence, in contrast to our initial hypothesis, in the studied stream the shredder was top-down-controlled, while the grazer was not. It is concluded that top-down control depends on the ecological characteristics of a specific predator–prey pair rather than on trophic guild of the prey.

Various experiments have been conducted in order to investigate the importance of predator–prey interactions in stream food webs, which resulted in different findings. Some studies showed strong and cascading top-down effects of fish predation, while others found no such interactions. One of the possible factors responsible for this variability may consist in differing vulnerability of the invertebrate prey, depending on their functional feeding group. Predation effects in detritus-based food chains are expected to be lower since invertebrate shredders are considered to be generally less vulnerable to predators. This assumption originates from mesocosm experiments with artificial substrates and the observation that grazing invertebrates moving on the stone surface to scrap periphyton are more visible to predators than shredders which hide and feed between leaves or in interstitial spaces. In fact, most field experiments investigating predator impact on stream food webs dealt with grazers (mostly mayflies) and their impact on periphyton biomass. Trophic cascades were shown regularly in mesocosm experiments and in large-scale approaches In contrast, only few field studies dealt with shredders abundances and their effect on leaf litter processing. Evidence for trophic cascades could only recently be provided in mesocosm experiments. Different strength of fish predation in detritus-based and primary production-based food chains comes to be important for the understanding of stream food webs. Thus, we hypothesised in our study that invertebrate grazers are more vulnerable to lethal and sublethal effects of fish predation than invertebrate shredders. This hypothesis was tested by comparing the population dynamics and physiological parameters of the grazer Rhithrogena semicolorata (Ephemeroptera: Heptageniidae) and the shredder Gammarus pulex (Crustacea: Amphipoda), which are two common and abundant representatives of their functional feeding groups in Central European streams.

In order to estimate lethal effects, we monitored the biomass and production of the two prey organisms and their consumption by the fish predator. To detect sublethal effects on the physiological level, we estimated the status of biochemical compounds serving to the storage of energy (glycogen and triglycerides), and the reproductive success of the two prey species. Hereby, we assumed that predator avoidance behaviour would lead to a decrease in food intake or an increase of energy consumption, which should be detectable as a decrease in the energy storage components. This effect should be observed much earlier than changes in population abundance caused by the combination of lethal and sublethal predation effects.The hypothesis that detritus-based food chains are less likely to be top down-controlled could not be supported. We found that shredders inhabiting leaf litter such as G. pulex are not necessarily less affected by fish predation than mayfly grazers foraging on stone surfaces. On the contrary, predation pressure on grazing R. semicolorata was not detectable, while gudgeon affected G. pulex biomass. Additionally, prey vulnerability seemed to be independent of the functional feeding group. In spite of the relatively intensive predation of gudgeon on G. pulex, the shredder Nemoura sp. showed no predation effect at all. Further, no trophic cascade could be observed in the stream since no differences in the amount of algae or organic matter were detectable between the reference and the fish reaches.

From our initial hypotheses, only the lethal effects of fish on shredders were supported, as observed in G. pulex. Gudgeon used a relatively large proportion of the annual production of G. pulex (<90%) and therefore probably accounts for the lower mean biomass of G. pulex observed in the fish reach compared to the reference reach.

For R. semicolorata, on the other hand, no predation effect was detectable in this experiment. Biomass was even higher in the fish reach. The relatively low proportion of annual production that was consumed by gudgeon supports this result. There are two possible explanations for the low consumption of R. semicolorata which could occur simultaneously: (1) low predator–prey encounter rates because of low habitat overlap, or (2) efficient predator avoidance strategies of R. semicolorata such as synchronised larval development, timing of feeding activity or reduced visibility to the predator.

We did not find any evidence in the present study for the hypothesis that specific feeding guilds have a different risk to be top-down controlled by benthivorous fish as assumed by. Based on our results, it is more likely that vulnerability of the prey is specific for every predator–prey pair. Predator impact should depend on factors determining the feeding rate of the predator such as habitat overlap and encounter rate. Those species suffering high mortality per unit production by having a high rate of encounter with predators and/or lacking effective avoidance strategies are likely to be predation controlled. This seems to be the case for the pair G. pulex–gudgeon, since both species live in pool habitats and large prey individuals are available through all seasons. For R. semicolorata on the other hand, habitat overlap is much smaller because it prefers riffle habitats. Further, large individuals are available only during April and May since individual growth is highly synchronised. Thus, in general, detritus-based food chains are not a priori less likely to be top-down controlled since the biomass of common and important detritivores as amphipods can be significantly influenced by predation.

Besides the lethal effects of predation, we hypothesised a predator-induced reduction of the energy storages in the prey organisms due to reduced activity and increased use of refuges in the presence of predators. This effect may lead to lower food intake. A lower energy amount available for growth and reproduction should reduce size at maturity and fecundity through an energy allocation trade-off. This should be detectable at least in R. semicolorata, because of the synchronic larval development and the well-defined point of maturity (emergence). However, no reduction of energy storages could be shown as the unchanged contents of triglyceride and glycogen in the two species revealed for both the reference and the fish reach. Nevertheless, individuals of R. semicolorata reached a higher mean body length in the reference section especially prior to emergence. Demographic models suggest that sublethal predation effects can have an even larger impact on population dynamics than predator consumption for species in which body size at maturity influences fecundity, as in mayflies. A general relationship between female body size and fecundity is assumed. Further, reduced size at emergence in the presence of fish has been shown for mayflies both in small-scale experiments and whole system approaches. Thus, a demographic effect such as reduced abundances in the second generation seems to be likely. However, the reduced body length of R. semicolorata larvae in the fish reach did not translate into a lower individual egg numbers, as frequently assumed in former studies. We conclude that reduced size at maturity of mayflies does not always mean reduced reproduction success and therefore cannot generally be interpreted as a sublethal predation effect.

One may criticise that two experimental conditions of our approach could have constrained the response of the prey organisms to predation and/or the detection of this response: (1) a too low fish density, and (2) the lacking replication. Average gudgeon biomass in our experimental reach (3.5–3.8 g wet weight m^?2) and even maximum biomass immediately after restocking (17 g wet weight m^?2) were lower than in most enclosure experiments with other benthivorous fish species. Although there are no data available on natural densities of gudgeon, the experimental densities chosen in the presented study (mean 0.5, maximum 1.4 fish m^?2) were well within the range observed for other benthivorous fish. Thus, regarding or argument (1) we conclude that the fish biomass in this study was closer to natural densities than to enclosure studies cited above. With respect to argument (2) it is obvious that this study was not replicated on the reach scale. However, we decided that work on a larger spatial and temporal scale was a more important task than replication, when approaching the question of possible top-down control of benthic communities.

The overall conclusion can be drawn that detritus-based food chains can be controlled by fish predation equally strong as primary production-based food chains. The intensity of the predation effects on single prey species is likely to depend on the species-specific fish predation rate rather than on a particular feeding mode of the prey.