Model validation
We validated the model by comparing simulated (= predicted) timings of TDM and OAB to observations from lakes for which relevant data on phytoplankton (15 lakes) and Daphnia (18 lakes) could be extracted from the literature (Supplement S3 Model validation). Linear regressions of observed vs. predicted timings of both OAB and TDM give good statistical fits (R2 ≥ 0.6) and indicate that the predictions are unbiased, i.e. regression slopes are not significantly different from 1 and intercepts not different from zero (Supplement S3 Model validation). A similar analysis based on observations of PLD was not meaningful, because we were unable to find data with sufficient temporal resolution on both OAB and TDM for more than five lakes.