Species assemblages across the deforestation gradient: winners and losers
Ordinations based on Bray-Curtis dissimilarity of sandfly species relative abundance (k = 2; stress = 0.11) and vertebrate species relative abundance (k = 2; stress=0.18) show that there is no clear separation of the vector and the non-vector communities nor the host and the non-host communities in ordination space (Fig. 4 left side panels). However, these ordinations reveal that the relative abundance of individual species differed across the environmental gradients with most of the sandfly species ordinated closer to increased forest intactness and away from the most heavily deforested sites with minimum forest cover (Fig. 4 top left panel) and that many important host species align along the percentage pasture vector (Fig. 4 bottom left panel). Regression models more clearly show the responses of individual species to deforestation (Fig. 4 right side panels; Appendix S1: Table S2). Higher pasture land cover was a significant negative predictor of the probability of encountering a vector (p = 0.04) and the vector genusNyssomyia spp. (p = 0.03), while higher forest cover was an important predictor of the probability of encountering the vector species Psychodopygus davisi (p = 0.10). The most important finding from sandflies feeding on vertebrate species showed that decreasing forest cover was a significant predictor of the probability of finding a sylvatic host in a sandfly pool (p < 0.05), as well as the probability of finding the most competent host species (nine-banded armadillo) in a sandfly pool (p < 0.05) (Fig. 4 bottom right panel; Appendix S1: Table S2).