Bridging the divide: Movement and disease ecology of free-roaming domestic dogs (Canis lupus familiaris) across ecological and anthropogenic interfaces

Abstract

Animal movement is a process which can connect otherwise distinct communities. When moving across ecosystem boundaries, animals have distinct behavioral responses, and decisions along these interfaces can affect ecosystem patterns such as the distribution of species, pathogens, or populations across space. However, hypothesis testing of individual movement decisions is difficult due to high inter-individual variability, high computational overhead, and lack of precision of movement predictions. Free-roaming cats and dogs are species which live in both anthropogenic and natural ecosystems. In this dissertation, we ask: 1) how does parasite prevalence of free-roaming cats and dogs change across socioecological gradients; 2) how do fine-scale landscape features interact with individual animal movement decisions to affect landscape-scale cross-ecosystem land-use patterns; 3) does movement preference drive free-roaming dog spatial heterogeneity in forest reserves. We found that for every 10% increase in the proportion of the human population with improved sanitation access, free-roaming domestic cat (Felis catus) and dog (Canis lupus familiaris) parasite and pathogen prevalence decreased by 12% (5-19%, 95% C.L.; p = 0.002). Then, we tested the influence of land cover permeability on domestic dog ingress into forest reserves and found that greater edge permeability increased site-use intensity and heterogeneity. We further developed the model framework to include validation of simulated presence at sites throughout each reserve with observed camera trap data and used this framework to test the hypothesis that directional preference drives site-use of domestic dogs in forest reserves in Andasibe, Madagascar. We found that the observed numbers of captures fell within the distribution of simulated captures for all but two camera trap sites, suggesting that the movement model is realistic at the site-level for most locations. We found that directional preference was not supported, where simulations using calibrated input from observed forest space-use data did not improve predictions (AICc < 2). Top models of predictive accuracy for all simulations were exponential with distance to the largest village in the region as the only covariate. In light of these three studies, pathogen prevalence and fine-scale movement decisions and parasite prevalence of free-roaming dogs may be driven more strongly by anthropogenic than natural ecosystems. However, additional studies using this validation framework are needed to understand the magnitude and relative contributions of ecological vs anthropogenic factors on fine-scale movement drivers.