Using (modular) random graphs to explore the effect of modular contact networks on wildlife disease spread
Although epidemic dynamics in spatially structured metapopulations have been well studied, there is still a lack of comprehensive understanding on disease spread in heterogeneous wildlife populations that show modularity (i.e. sub grouping) in their contact structure. This is because disentangling the effect of modularity from other network properties such as clustering coefficient and assortativity can be a challenge in empirical networks. Here, we first discuss a generative model to produce graphs with a specified degree distribution and community structure while maintaining a graph structure that is as random as possible. This null model allows for the systematic study of the presence of modularity in contact network and its impact on disease spread through populations. Our model also allows for a systematic investigation of several network types (Poisson, geometric and power law) which represent the spectrum of empirical contact network structures and contact heterogeneity. Using Susceptible-infected-recovered (SIR) model of disease spread, we next demonstrate the impact of modularity in heterogeneous contact networks on infection spread through wildlife populations.
We show that modular random graphs can be used as powerful tools to disentangle the effect of modularity on disease dynamics in heterogeneous contact network across different wildlife populations and species. Our numerical simulations of disease spread reveal high modularity in contact network to alleviate burden of infection in populations. This is because highly modular contact networks are more efficient in preventing disease spillover from initial infected groups as compared to low modularity contact networks. However, the effects of modularity in containing disease spread differs across various contact network types (Poisson, geometric and power law). Contact network modularity has a greater impact in networks where a small fraction of individuals are highly gregarious (such as geometric and power law networks) as compared to networks where most individuals have similar levels of sociality (such as Poisson networks). Overall, we demonstrate that application of modular random graphs offers unique strengths of modeling disease dynamics within and between wildlife subpopulations, and thereby, designing effective conservation efforts by targeting groups most vulnerable to receiving and spreading future infections.