Aim To compare the geographical distributions of two tick-borne pathogens vectored by different tick species, to examine the relative importance of climate, land cover and host density in structuring these distributions, and to assess the spatial variability of these environmental constraints across the species ranges.
Location South-central and south-eastern North America.
Methods Presence/absence data for two tick-borne pathogens, Ehrlichia chaffeensis and Anaplasma phagocytophilum, were obtained for 567 counties from a regional data base based on white-tailed deer (Odocoileus virginianus) serology. Environmental variables describing climate, land cover and deer density were calculated for these counties. Global logistic regression analysis was used to screen the environmental variables and select a parsimonious subset of predictors. Local analysis was carried out using geographically weighted regression (GWR) to explore spatial variability in the parameters of the regression models. Cluster analysis was applied to the GWR output to identify zones with distinctive species–habitat relationships.
Results Global habitat models for E. chaffeensis and A. phagocytophilum included temperature, humidity, precipitation and forest cover as explanatory variables. The E. chaffeensis model also included forest fragmentation, whereas the A. phagocytophilum model included deer density. Local analyses revealed that climate was the primary correlate of pathogen presence in the eastern portion of the study area, whereas forest cover and fragmentation constrained the western range boundaries. Habitat relationships for all variables were weak in and around the Mississippi Delta.
Main conclusions Efforts to model pathogen and disease ranges, and to predict shifts in response to global change should consider future scenarios of land-cover change as well as climate change, and should address the possibility of spatial heterogeneity in species–habitat relationships. The methods presented here outline an approach for objectively delineating geographical zones with similar species–environment relationships, which can then be used to stratify landscapes for the purposes of further explanatory and predictive modelling.