1 Introduction
The spatial distribution of direct contact among animals (co-location of animals at the same time) greatly influences the dynamics of various ecological processes, such as disease transmission, social organization, and human-wildlife conflict (Meijaard et al. 2011, Kurvers et al. 2014, Craft 2015). The spatial distributions of contacts across landscapes are often heterogeneous due to various factors, including animal movement patterns (e.g., migration, speed), social behaviors (e.g., mating, territoriality, fission-fusion dynamics), densities of animals, and external factors such as food availability and predators (Kareiva and Odell 1987, Spiegel et al. 2017). Assessing drivers of such heterogeneity is key for mechanistic understanding of contact-driven ecological processes.
While social factors like group membership can influence contact structure and heterogeneity (i.e., animals within the same social groups tend to stay together, resulting in a higher contact rate) (Silk et al. 2019), landscape features like the distribution of resources also impact contact patterns (McClure et al. 2022). Areas with abundant resources or preferred habitats may attract more animals, including from different social groups, structuring contacts. Exploring the role of the landscape in structuring contacts can facilitate understanding of the spatial distributions and extents of contact based ecological processes and enable the prediction of contacts or contact-driven ecological processes on the landscape.
Frameworks to investigate environmental drivers of contacts and probabilistically predict their distributions are limited. Koen et al. (2017) adopted social network analysis to estimate the effects of landscape connectivity on contact rates by comparing contact rates based on broad-scale environmental conditions. In addition, some spatial models (e.g., conditional autoregressive model) have also been applied to quantify the effects of different factors on the spatial distribution of contact rates (Yang et al. 2021a). However, these approaches have not explicitly compared the spatial conditions where contacts occurred to those where contacts could have occurred. As such, these approaches did not estimate and disentangle how different landscape features drive probability of contact occurrence across the landscape. Building off the assumption that contacts differentially occur at the places with abundant resources or preferred habitat, previous studies have used the spatial overlap of individual or population-level habitat selection as a proxy for contact probability (Habib et al. 2011). However, it remains unclear whether patterns of resource selection can accurately predict hotspots of contact.
In this study, we developed a method to quantify landscape factors influential to contact locations. We applied the method to movement data from two wild pig (Sus scrofa ) populations in dissimilar landscapes, exemplifying how our approach can identify geographical attributes associated with contact events. We use our framework to test the hypothesis that the landscape variables that drive contacts are the same as those that drive individual-level space use, as has been assumed in previous applications of resource selection analysis to contact behavior. Consequently, we assess the utility of aggregating individual resource selection functions to predict the spatial properties of contact behavior. We discuss the implications of this work for understanding contact dynamics in wild pig systems.