Larval diet breadth and wingspan mediate landscape–richness relationship in butterfly communities

Landscape structure and species traits both shape butterfly assemblages, but their joint effects, and how landscapes restructure trait space independently of richness, remain less understood. We surveyed butterflies at 50 semi-natural grasslands (Serbia) and modelled species richness (SR) with GAMs using two sets of landscape predictors within 2-km buffers: composition (% cover of grassland, forest, complex agriculture) and configuration (distance to the nearest natural patch, nearest natural-patch size, edge density). We included community-weighted mean (CWM) of two traits: wingspan (WS) and host-plant specificity (HPS) and fit landscape-only models, trait-additive models and pre-specified trait–landscape interactions per set (HPS and complex agriculture; WS and distance to the nearest natural patch). To test whether these gradients restructure trait space, we modelled functional divergence (FDiv) as a complementary response. The only landscape predictor of SR was complex agriculture, which increased richness but reduced FDiv. Beyond landscape composition, HPS consistently improved SR models in both sets, with richness peaking at intermediate-to-narrow diet breadth and declining towards extremes. Isolation increased SR only in large-winged communities (WS × distance to the nearest natural patch), and FDiv increased with isolation, evidencing greater representation of trait extremes. Crucially, landscape effects on FDiv persisted after conditioning on SR, showing that landscapes filter which traits persist, not just how many species occur. Butterfly assemblages are driven by resource-based and movement-based filtering rather than landscape structure alone. Our key novelty shows that the same landscape gradients decouple taxonomic from functional diversity; agricultural complexity adds species while compressing trait breadth, whereas isolation benefits large-winged communities and expands trait dispersion. Conservation should therefore track functional structure as well as counts, maintaining diverse larval host-plants and stepping-stone connectivity to sustain both species richness and the functional breadth that underpins resilience. Read the free Plain Language Summary for this article on the Journal blog.