We used themodel to simulate the lake’s biogeochemical dynamics
for two data sets: a time series collected in 2013 and a single-timepoint
survey collected in 2008. In both cases, the model predicted
chemical dynamics (Fig. 2 and Supplementary Fig. 6) that were consistent
with those expected from a eutrophic, dimictic lake19. In
2013,we had a time series covering the fivemonths before the bacterial
survey, so we initialized the model using the observed chemical parameters
fromthe first survey in the timeseries.The chemical dynamics
predicted by the model (Fig. 2) accorded with our measured time
series, and the predicted distribution of chemical species agreed
with the final survey (Supplementary Fig. 2).