When scale matters to disentangle the effect of habitat and temperature on ground-dwelling spider communities in urban environments

Urbanisation is an ongoing process associated with multiple environmental changes affecting ecosystems worldwide. Temperature and habitat are main drivers of animal communities within cities, but quantifying their relative weights remains a challenge, as urban heat islands (UHI) often co-vary with land-cover. This study aims to disentangle the effects and relevant scale of temperature and habitat on ground-dwelling spider communities. Based on an original sampling design, we collected 20,761 spider individuals belonging to 137 species at 36 sampling sites in the city of Rennes (northwest France). We characterised communities by assessing the number of trapped individuals at each site, as well as calculating several metrics to estimate taxonomic and functional diversities. Temperature metrics were obtained from two sensor networks monitoring UHI (100-m resolution) and near-ground temperature (1-m resolution) independently. Land-cover and isolation were used to describe landscapes, and vegetation structure to describe local habitats. We used generalized linear mixed models to disentangle the effects of temperature from those of habitat at the landscape and local scales, and identified relationships between community descriptors and predictors. We show that temperature-related metrics are important predictors of spider communities, and that the landscape and local scales have independent effects. Near-ground temperature alone explained 24% of the number of trapped individuals, whereas UHI explained 20% of taxonomic diversity. Local vegetation height and cover were significant predictors of functional diversity, and explained 22% and 25% of variance, respectively. We conclude that locally applied planning measures could mitigate the loss of taxonomic diversity induced by the atmospheric UHI and promote the establishment of more diverse communities.