Climate, host and landscape determinants of bovine babesiosis in Great Britain: analysis of Imidocarb treatment notifications, 2016–2024

Abstract Background Bovine babesiosis is a clinically important tick-borne disease of cattle in Great Britain, yet its spatial distribution and key geographical risk factors remain poorly characterised due to limited routine surveillance. Imidocarb dipropionate (Imizol), the only licensed treatment for bovine babesiosis, must be notified to the Animal and Plant Health Agency whenever it is prescribed or administered to cattle. These compulsory notifications therefore provide a proxy dataset for examining spatial patterns in clinically recognised bovine babesiosis. Methods Imizol notification records from 2016 to 2024 were used as a proxy for disease occurrence to model the spatial suitability for bovine babesiosis in Great Britain. Presence was defined at a 2 km grid resolution and modelled using machine learning. Predictor variables captured seasonal climate, vegetation greenness, livestock densities and predicted habitat suitability for red, roe and fallow deer as key risk factors. Models were trained using southwest England data from 2016 to 2022, temporally validated using southwest England data from 2023 to 2024, and externally evaluated using independent data from outside the training region. Results Models showed high performance in internal cross-validation (mean AUC = 0.81; mean correlation = 0.55) and temporal validation (continuous Boyce Index = 0.81), and external validation outside southwest England indicated moderate transferability (Boyce Index = 0.58). Climatic predictors accounted for the greatest relative influence in the models, particularly autumn cooling (mean relative influence = 37.3%) and atmospheric moisture (8.5%), consistent with seasonal climatic constraints on tick activity. Among non-climatic predictors, roe deer habitat suitability and sheep density contributed moderate influence, while vegetation greenness had a smaller effect. Proportional land-cover variables showed consistently low individual influence. Conclusions Mandatory veterinary medicines notification data can be used as a proxy for disease occurrence to model spatial suitability of bovine babesiosis at national scale. Seasonal climatic constraints on tick activity and host-associated factors were the strongest determinants of predicted disease patterns. Resulting risk maps may help anticipate areas of elevated risk and guide disease surveillance at regional scales, including highlighting locations where environmental conditions appear suitable for disease occurrence but treated cases have not yet been reported.