Laboratory oviposition assays
Among all two-choice oviposition experiments, we found three significant preferences (Figure 4 and S11): Rabai Kwa Bendegwa village colony preferred forest water samples over village water samples, and preferred forest mosquito larval density over village larval density; La Lopé forest colony preferred the bacterial culture started with peridomestic water samples over that started with forest water samples. However, there is substantial within-colony variation in most experiments. The beta-binomial models did not find any significant difference between colonies or between the habitat types in any assays (Table S8). We applied a negative-binomial model to analyze the oviposition assays testing multiple bacterial densities (Figure 5 and S12). Neither colonies nor the habitats of the colonies had a significant effect on the mosquito’s preference for the five bacterial densities (Table S8). La Lopé village colonies showed a weak preference for lower bacterial densities, but the trend was not statistically significant (ANOVA for the effect of oviposition choices: F = 1.56, df = 4, p = 0.200).