Body size is intrinsically linked to metabolic rate and life-history traits, and is a crucial determinant of food webs and community dynamics^1,2. The increased temperatures associated with the urban-heat-island effect result in increased metabolic costs and are expected to drive shifts to smaller body sizes³. Urban environments are, however, also characterized by substantial habitat fragmentation⁴, which favours mobile species. Here, using a replicated, spatially nested sampling design across ten animal taxonomic groups, we show that urban communities generally consist of smaller species. In addition, although we show urban warming for three habitat types and associated reduced community-weighted mean body sizes for four taxa, three taxa display a shift to larger species along the urbanization gradients. Our results show that the general trend towards smaller-sized species is overruled by filtering for larger species when there is positive covariation between size and dispersal, a process that can mitigate the low connectivity of ecological resources in urban settings⁵. We thus demonstrate that the urban-heat-island effect and urban habitat fragmentation are associated with contrasting community-level shifts in body size that critically depend on the association between body size and dispersal. Because body size determines the structure and dynamics of ecological networks¹, such shifts may affect urban ecosystem function.

How best to manage forest patches, mitigate the consequences of forest fragmentation, and enable landscape permeability are key questions facing conservation scientists and managers. Many temperate forests have become increasingly fragmented, resulting in reduced interior forest habitat, increased edge habitats, and reduced connectivity. Using a citizen science landscape‐scale mark–release–recapture study on 87 macro‐moth species, we investigated how both life‐history traits and landscape characteristics predicted macro‐moth responses to forest fragmentation. Wingspan, wing shape, adult feeding, and larval feeding guild predicted macro‐moth mobility, although the predictive power of wingspan and wing shape depended on the species' affinity to the forest. Solitary trees and small fragments functioned as “stepping stones,” especially when their landscape connectivity was increased, by being positioned within hedgerows or within a favorable matrix. Mobile forest specialists were most affected by forest fragmentation: despite their high intrinsic dispersal capability, these species were confined mostly to the largest of the forest patches due to their strong affinity for the forest habitat, and were also heavily dependent on forest connectivity in order to cross the agricultural matrix. Forest fragments need to be larger than five hectares and to have interior forest more than 100 m from the edge in order to sustain populations of forest specialists. Our study provides new insights into the movement patterns of a functionally important insect group, with implications for the landscape‐scale management of forest patches within agricultural landscapes.