Disease emergence and dynamics on biologically motivated contact networks

Infectious disease transmission requires that epidemiologically relevant contact occurs between infectious and susceptible individuals. Thus, for mathematical models to accurately predict disease emergence and dynamics they must incorporate the contact patterns responsible for transmission. In this context, this thesis investigates how the level of contact detail included in an infectious disease model influences its predictions. Three models are considered. The first investigates infections spreading through territorial populations, with potential canine rabies spread in Australian wild dogs a case study. Two factors governing wild dog contacts are considered: geographic distance and heterogeneous wild dog behaviour. Not including spatial constraints results in a model that overestimates the probability of an epidemic and that fails to generate the outcome 'rate of spread'. Conversely, not incorporating heterogeneous dog behaviour results in a model that underestimates the probability an epidemic will occur. The second model investigates tick-borne pathogen spread between ticks and vertebrate hosts. Key features of tick feeding behaviour include: tick aggregation on hosts, co-aggregation of larval and nymphal ticks on the same hosts, and co-feeding. Co-aggregation increases R0. Models failing to incorporate tick co-aggregation will therefore underestimate the likelihood of pathogen emergence, especially in geographic regions and seasons where larval burden is high and for pathogens mainly transmitted during co-feeding. The third model investigates the effect of clustering (triangle and square contact patterns) on the spread of infection through social networks. Clustering reduces R0 and the magnitude of the reduction increases with higher transmission probabilities. Models that fail to incorporate clustering will overestimate the likelihood of disease establishment, especially for highly transmissible diseases. In conclusion, the three disease models collectively reveal model predictions are improved and additional outcomes are generated by the inclusion of realistic host contact patterns. These findings reinforce the value of incorporating biologically-faithful contact patterns into infectious disease models