Strong context dependence in the relative importance of climate and habitat on nation-wide macro-moth community changes.

Habitat modification and climate change are major threats to biodiversity. Understanding the magnitude of their impact and their relative contributions across large spatial scales is important but challenging, given potential context dependence and biases arising from data gaps. Here, we apply a novel predictive framework to tackle these challenges in a study of community data across multiple climatic zones. We used joint species distribution modelling to analyse moth species' occurrence and abundance variation in response to the climate and habitat characteristics within a landscape of 109 sites across Finland spanning 23 years. To correct for spatiotemporal gaps in survey data, we used predictions from the fitted models to evaluate the relative importance of individual drivers on species' occurrence and abundance variation, as well as their effects on community diversity, across the entire country (i.e. beyond the sampled sites). To characterise potential context dependence, we extended model predictions with conditional variance partitioning analysis across species grouped by their functional similarity (functional context) and across sites grouped by the most dominant habitat types (environmental context). Finally, to assess the magnitude of each driver's effect alone on community-level change, we implemented separate scenario predictions for changes in climate only vs. changes in habitat only. Our results show that the dynamics of communities depended on both functional and environmental contexts. Across Finland, variation in species occurrence was mostly explained by habitat characteristics, but the relative importance of climate and habitat varied according to species functional characteristics and to the dominant habitat types within a landscape. Consequently, temporal trends in community diversity varied across space, and temporal predictions based on changes in climate only vs. changes in habitat characteristics only missed important features of the realised community dynamics. Our results underpin the importance of habitat composition as a key driver of community change-even among temperature-sensitive ectotherms. Climate and habitat contributed unequally to species occurrence and abundance, and consequently, predictions based on a single driver of environmental change misrepresented realised community dynamics. To understand and predict community change, we thus need to account for the imprints of both climate and land use.