Why deterministic trait-based models may misrepresent the expected spatiotemporal dynamics of Hyalomma marginatum under uncertainty

Trait-based mechanistic models are widely used to quantify climate-driven reproduction numbers (R₀) for vectors and pathogens. These models are commonly evaluated using deterministic parameter values, with uncertainty reported only as confidence intervals around point estimates. However, whether uncertainty propagation in constant variables merely inflates uncertainty bounds or fundamentally alters expected model outcomes remains poorly examined. Here, we develop a stage-structured, climate-dependent R₀ framework parameterized using empirically derived temperature and moisture response functions for Hyalomma marginatum, the primary vector of several tick-borne diseases affecting humans and livestock. Using Monte Carlo simulations, we systematically propagate uncertainty through constant model components that are often treated as fixed due to a lack of weather-related relations and compare stochastic outcomes with their deterministic counterparts. Propagating uncertainty in fecundity and host-finding behaviour increased the expected reproduction number across temperature–vapour pressure deficit (0.5–3.5 kPa) gradients, with the stochastic mean R₀ exceeding deterministic estimates in a spatially coherent and largely uniform manner. The distribution of log-scale differences between stochastic and deterministic reproduction numbers was consistently positive, tightly clustered (mean = 0.12, SD = 0.015), and narrowly concentrated across space and time. This indicates that incorporating uncertainty in fecundity and host-finding affects the expected intensity of population build-up rather than the geographic extent of potential risk. We show that parameters commonly treated as constants or fixed thresholds can meaningfully influence inference when uncertainty is accounted for. Our findings highlight the importance of explicit uncertainty propagation in trait-based R₀ models and caution against relying solely on deterministic estimates when producing climate-driven risk decision support tools.