Figure 2 . The total number of individuals in each microcosm changes over time, from the beginning of drought treatment (week 4) to the end of experiment (week 16). Data was log (x+1) transformed; number of replicates for each treatment N=6. Stable lines in two constant treatments indicate that populations were not stressed and increased constantly throughout the experiment. The potential confounding effect mentioned in the methods section can be seen in the differing starting populations across the treatments.