Integrated models typically couple spatially-explicit, individual-level data with marginal data that aggregate across many individuals to infer a single derived quantity, most frequently, population size or growth rate. While these quantities are both of critical interest in population ecology, they sometimes overshadows the potential for integrated approaches to add value to other avenues of ecological inquiry. For example, recasting an integrated population model to estimate metrics describing system connectivity would be useful for ecologists studying the contact-related processes that underlie pathogen transmission, or gene flow. In fact, integrated approaches may hold even higher inferential value for these metrics than they do for marginal quantities like population growth rate. Here, I retool a conventional integrated model to draw inferences about contact network structure. I first review a few metrics that describe network structure and population connectivity, and then embed those metrics as derived quantities in an integrated model, estimating their posteriors via conventional methods from integrated population modeling. Finally, I demonstrate the real-world utility of this approach with a case-study comparing contact dynamics of bighorn sheep populations located along a latitudinal gradient from southern Nevada to northern Idaho and Montana.
Results/Conclusions
This exploration shows the potential utility that integrated modeling approaches provide for ecological inference beyond population growth rate or size. Simulations demonstrated how constraints on marginal data like patch quality or group size can dramatically contact network topology. Feeding those simulated data forward into an integrated modeling framework allowed me to estimate several underlying graph parameters (Von Neumann entropy and dispersion of the degree distribution), albeit with varying precision. In the case study, I was able to successfully apply marginal datasets to refine inferences about population mixing processes. That analysis, which was foremostly a proof-of-concept, compared spatial and social drivers of mixing dominate contact network structure under different environmental context for one species, bighorn sheep. Posterior estimates of the network metrics suggested that contact processes in bighorn sheep are dominated by social interactions in herds with relatively contiguous habitat and resources, but are tightly tied to spatial structure when patch quality is overdispersed.