Appendix A. Model sensitivity to parameter(recovery rate from worm infection).
Parameteris the rate at which worm-infected hosts, IW and ICO recover from their worm infections and revert to S and ITB status, respectively. We ran our model with values ofbetween 0 and 1, at intervals of 0.1, to assess how our choice ofaffects model output, for models including higher mortality of coinfected individuals (dCO) only, and for models including both elevated dCO and transmission heterogeneity as mechanisms generating observed patterns of worm and TB infection. We examined two components of model output, (i) the relationship between worm prevalence and TB prevalence in the population, at low (0.15), intermediate (0.3), and high (0.6) annual mortality of coinfected individuals (dCO), as shown in the main panels of Fig. 5; and (ii) the difference in TB prevalence between hosts with and without worms, at low, intermediate and high dCO, as shown in the inset panels of Fig. 5. (For this sensitivity analysis we set= 3 as in the model output presented in the main text).
(i) The relationship between herd worm prevalence and TB prevalence changes very little with: At medium dCO the negative slope of the regression of TB prevalence versus worm prevalence gets slightly less steep asis increased, while changes inhardly affect the regression slope at all when dCO is at the high or low setting. Across the whole range of, both models produce negative slopes for intermediate and high dCO, and slopes close to 0 for low dCO (Fig. A1).
(ii) In the model including only dCO, TB prevalence in hosts with worms is higher than in worm-free hosts for all levels of dCO (Fig. A2a). In the model including dCO and transmission heterogeneity, TB prevalence in worm-free hosts exceeds TB prevalence in worm-infected hosts when dCO is high, but not when dCO is at the medium or low setting (Fig. A2b). These relationships hold across all levels of, although the difference in TB prevalence between worm-free and worm-infected hosts is always larger for lower values of.
|FIG. A1. Model sensitivity to parameter γ – relationship between herd TB prevalence and herd worm prevalence, (a) in the model including only dCO and (b) in the model including both dCO and transmission heterogeneity. The y-axis shows the negative slope of the regression of herd TB prevalence on herd worm prevalence. Variation inhas little effect on the regression slopes for either model.|
|FIG. A2. Model sensitivity to parameter– TB prevalence according to worm status, (a) in the model including only dCO and (b) in the model including both dCO and transmission heterogeneity. The y-axis shows the difference in TB prevalence (PTB) between worm-infected and worm-free buffalo (PTB(W+) – PTB(W-)). The difference in TB prevalence between worm-infected and worm-free hosts is larger at low values. TB prevalence in worm-free hosts only exceeds TB prevalence in worm-infected hosts (yielding negative y) in the model including transmission heterogeneity (b) at high dCO.|