Characterizing urban landscapes using very-high resolution satellite imagery to predict Ae. albopictus larval presence probability in public spaces.

The global spread of Aedes albopictus raises growing public health concerns due to its role in transmitting dengue, chikungunya, and Zika. In southern France, the increase in imported dengue cases and local transmission underlines the urgent need for effective vector control. While efforts primarily target private breeding sites, public spaces also contribute notably to larvae presence. Understanding the impact of urban landscapes on the distribution of breeding sites is crucial for optimizing vector control strategies, identifying high-risk areas, and reducing mosquito populations. This study aims to investigate how urban landscapes impact the distribution of Ae. albopictus larvae in public spaces, with a focus on storm drains and telecom cable chambers in Montpellier, France. Very high-resolution satellite imagery was used to characterize urban landscapes through textural analyses of spectral indices. Environmental bias was assessed by analyzing the representativity of sampled breeding sites within the diverse urban landscapes. Species distribution models (SDMs) were built, their predictive accuracy was evaluated, and an ensemble model was created to predict larval presence across the study area. SDMs predicted a high probability of larval presence in the western and northeastern parts of Montpellier, with low uncertainty. The most influential variables for predicting larval presence were the mean of Normalized Difference Vegetation Index (NDVI), texture indices from both NDVI, brightness index (BI), and the panchromatic image. Urban vegetation significantly influences larval presence, although higher vegetation index values correlate with a decreased probability of larval occurrence. Additionally, the combination of vegetation and urban structures plays a crucial role in determining the presence of Ae. albopictus larvae in public spaces, where small, organized urban objects and large patches of vegetation increase the likelihood of larval presence. This study highlights the potential of very high-resolution remote sensing and species distribution modeling for enhancing urban mosquito control strategies, ultimately contributing to improved public health policies outcomes in the face of vector-borne disease threats.