Subsidy between ecosystems has been considered in many natural ecosystems, and should alter food webs and communities in human-impacted ones. We estimated how drifting plankton from a reservoir contribute to downstream food webs and showed that they alter community structures over a 10-km reach below the dam. To estimate the contribution of the drifting plankton to macroinvertebrates, we used C and N isotopes and an IsoSource mixing model. In spring and autumn, contributions of plankton to collector-filterer species were highest 0.2 km downstream of the dam, and clearly decreased from 0.2 to 10 km. At 0.2 km, the contribution of plankton to a predator stonefly was remarkably high. These results indicated that drifting plankton from a dam reservoir could subsidize downstream food webs and alter their energy base, but the importance of this subsidy decreased as distance from the reservoir increased. The general linear models indicated that the abundance of collector-filterers and predators was related positively to zooplankton density in stream water. Thus, food source alteration by drifting plankton also influenced the community structures downstream of the dam.

Food‐chain length (FCL) in ecosystems has been studied extensively, and numerous hypotheses to predict FCL, productivity, ecosystem size, and productive space have been proposed. For example, the productivity hypothesis suggests that resource availability limits FCL, whereas the productive‐space hypothesis predicts that per‐unit‐size resource availability and ecosystem size equally limit FCL. However, previous studies have only measured total productivity to test FCL and have never tested the impact of resource availability within a system on FCL. Therefore, we estimated FCL in 15 ponds using stable isotope techniques to test the most common hypotheses for predicting FCL. We also measured total productivity, pond volume (ecosystem size), and edible microalgal carbon (resource availability) in each pond. We found that productive‐space (edible carbon+pond volume) was the best model to predict FCL, and FCL was significantly correlated with edible carbon and pond volume. However, factors such as total productivity did not directly predict FCL of the ponds. Therefore, our results suggest that both resource availability and ecosystem size predict FCL in pond ecosystems and play significant roles in maintaining longer FCLs. Furthermore, the productive‐space hypothesis appears to be particularly important for determining the FCL of ponds.

No abstract