Jonas Wallin

West Nile virus (WNV) is a growing risk to public and veterinary health in Europe, with intensifying outbreaks over recent decades coinciding with a rapidly warming climate. Thermal biology models parameterized through laboratory experiments on the mosquito vectors and the pathogen play a central role in our causal understanding of the effects of temperature on WNV transmission. We evaluated the ability of a thermal biology model of WNV's relative basic reproduction number (R₀) to explain patterns of real-world transmission risk using monthly records of human West Nile virus neuroinvasive disease (WNND) in Europe between 2010 and 2023. We assessed spatial and temporal alignments of R₀ estimates and WNND observations and calculated R₀ estimates from temperature data of varying resolution, assessing the value of these estimates for WNND risk ranking compared to temperature alone. Moreover, we used generalized additive models to investigate if the effect of temperature on the WNND incidence across Europe mirrors the R₀ temperature response, including the optimal temperature for transmission. We found that R₀ accurately captured the seasonality of WNND, the temperatures associated with peak risk, and marginally improved WNND risk ranking compared to temperature alone. However, R₀ poorly explained the irregular interannual incidence pattern and more limited geographical range of reported WNND cases. Additionally, nonlinear averaging of R₀ calculations from high temporal resolution temperature data slightly improved risk ranking at elevated temperatures, whereas estimates based on average temperatures were better when lower temperatures were also included. Sensitivity analyses suggested that the validation of the optimal transmission temperature was largely informed by observations from Greece, as other countries contributed little information at such high temperatures. These findings demonstrate that thermal biology models capture important aspects of WNND risk in Europe. However, incorporation of additional ecological and epidemiological drivers is needed to develop more accurate risk predictions.