Additional file 1: Table S1. The number of females collected by trap and site. The mosquito column (mosq) indicates the number of females collected; the traps column of trapping nights represented. Table S2. The number of females collected of each species by region of the park. Table S3. Summary of weather conditions sampled within each region of the park. Numbers are displayed as median and range in parentheses.
Additional file 2: Text S1.
Additional methods and results for the regression analysis. Table S4. Model parameters, estimates, standard error (SE) and hypothesis tests for the Poisson regression analyses in Fig. 3.
Additional file 3: Table S5. Data on the number of species collected.
Additional file 4: Table S6. Descriptive results comparing species-specific shifts in mosquito communities collected in the net and CDC trap. Species more commonly collected in a trap are listed if 5 more were collected in that trap after all sampling days at the site. Figure S1. The apparent richness (number of unique species) and diversity for sites within each region. Figure S2. Non-metric multidimensional scaling ordinations of trap differences in mosquito communities in Malelane, Satara, Shingwedzi and Punda Maria. Figure S3. Species-specific trap preferences for the net vs CDC trap difference based on rare species not displayed in Fig. 3. Dots represent the difference in the number of mosquitoes collected in the net vs the CDC trap based on the total number of mosquitoes sampled across nights at each site. Figure S4. The net trap and the CDC trap caught higher numbers of mosquitoes (Fig. 3) and this pattern was not driven by any species or genus-specific trap bias (left figures) but by variation in the total number of the species collected (right figures). Figure S5. Dendrogram of species composition based on Bray-Curtis dissimilarity and the hierarchical clustering algorithm.