Cessation of hunting shifts transmission networks and increases R0
We found that reducing hunting mortality had profound effects on FIVpco transmission dynamics. Even though the populations in the treatment and stable region were of comparable size (Table S1), our estimates of R0 for the same virus were two-fold higher in the treatment region compared to the stable region (with non-overlapping 95% high probability density intervals indicating that the difference is significant, Fig. 1). Other model parameters, such as generation time (time between initial FIVpcoinfection and onward transmission, Fig. S2) and the proportion of missing cases (Fig. S3) yielded similar estimates in both regions. This burst of transmission in the treatment population was likely a product of transmission between males as they were dominant in the network (Fig. 1a). In the treatment population, males had an overall mean weighted degree (i.e., the number of likely transmission events per individual, weighted by probability of transmission occurring) double that of females (0.14 compared to 0.37), only one putative transmission event occurring between sexes, and no detected female-female transmission events. When we assessed weighted degree homophily of male-male transmission events (i.e., the number of edges only between males), and simulations revealed that the dominance of male-male transmission events in the network was not random (1000 simulated annealing network iterations, p < 0.001, Fig. S4a). Putative transmission events largely occurred when hunting mortality was eliminated (Fig. 1a), during which time the survival of adults and subadult males was high, age structure increased, and the abundance of independent pumas increased (Logan & Runge 2020b). Male survival rates in the hunting period were also lower than for either sex in the treatment region (Logan & Runge 2020a). Females were, however, much less connected in the transmission network in the treatment region compared to the stable region, where they were more central (Fig. 1b). In contrast to the treatment region, the stable region showed evidence of transmission from females to both females and males. Average weighted degree was higher overall for males than females in the stable region (0.46 vs 0.29). Even though weighted female-female degree homophily was higher between regions (0 vs 0.05), our simulations show that we could not reject the null hypothesis that this difference was by chance (p = 0.692, Fig. S4b). Female-to-female transmission events occurred between highly related females supporting our previous findings of a significant role of host relatedness in FIV spread for puma living in this region (Fountain-Jones et al. 2019). Taken together, our results indicate that lower hunting mortality was associated with an increase in the number of transmission events and a transmission shift towards males.
After hunting was prohibited, the greater survival and increasing abundance of males probably resulted in greater competition between males for mates. As the dominant transmission mode for FIVpco is considered to be via aggressive contacts (VandeWoude & Apetrei 2006), increased male competition for mates appears a probable explanation for the change in transmission dynamics. Further interrogation of our transmission network supports this theory, as in all but two instances, male-to-male transmission occurred between individuals with overlapping territories in the treatment region (Fig. 2/S5/S6). One transmission pair was unusual in having less spatial proximity, yet one puma of this pair was a likely immigrant to the region (M133) and could have passed through M73’s territory at some point (Fig. 2). With the exception of M73 (~6 y.o. at time of infection), all individuals involved in these transmission events were between 1-3 y.o., which is a period when males are establishing new territories and are starting to compete for access to females (Logan & Sweanor 2001; Hornocker & Negri 2010). Our results suggest it is unlikely that these males transmitted to each other prior to dispersal or via maternal or paternal contacts—since these individuals were not related based on genomic data (Trumbo et al.2019). While our estimates suggest that we were able to sample approximately 40% of the FIVpco infections in both regions (Fig. S3)—arguably good coverage for secretive, free-ranging wildlife—our models account for this type of missing data (Didelotet al. 2017). For example, nearly all putative transmission events we identified from our transmission networks were between individuals on the landscape at the same time and in most cases were captured in close spatial proximity to each other. The biological plausibility of these transmission events demonstrates the power of adapting transmission network models to trace transmission and gain epidemiological insights in systems that are difficult to observe.