Investigating the mechanisms which underlie the biomass fluctuations of populations and communities is important to better understand the processes which buffer community biomass in a variable environment. Based on long-term data of plankton biomass in Lake Constance (Bodensee), this study aims at explaining the different degree of synchrony among populations observed within two freshwater plankton groups, phytoplankton and ciliates. Established measures of temporal variability such as the variance ratio and cross-correlation coefficients were combined with first-order autoregressive models that allow estimating species interactions from time-series data. We found that predation was an important driver of the observed seasonal variability patterns in phytoplankton and ciliates, and that competitive interactions only played a subordinate role. In Lake Constance copepods and cladocerans, two major invertebrate predator groups, focus their grazing pressure at different times of the season. Model results suggested that compensatory dynamics detected in phytoplankton originate from the differential vulnerability of species to either one of these two predator groups. For ciliates model results advocated that synchrony among species occurs because ciliates tend to be vulnerable to both predator groups. Our findings underline the necessity of extending studies of community variability to multiple trophic levels because accounting for predator-prey interactions may often be more important than accounting for competitive interactions at one trophic level.