Plant-insect trophic systems should be particularly sensitive to processes altering species spatial co-occurrences, as impacts on one level can cascade effectively through the strong trophic reliance to the other level. Here, we predicted the biogeography of Lepidoptera-plant communities under global-change scenarios, exploiting spatially resolved data on 423 Lepidoptera species and their 848 food plants across the German state of Baden-W & uuml;rttemberg (ca. 36,000 km2). We performed simulations of plant extinction and Lepidoptera expansion, and respectively assessed their cascading consequences-namely secondary extinction of Lepidoptera and change in functional distance of plants-on the interaction networks. Importantly, the simulations were spatially explicit, as we accounted for realistic landscape contexts of both processes: Plant extinctions were simulated as "regional" (a species goes extinct in the whole region at once) vs. "isolation-driven" (a species gradually goes extinct from the peripheral or isolated localities according to its real regional distribution); Lepidoptera expansions were simulated with random, northward, and upward directions according to real topography. The consequences were assessed based on empirical community composition and trophic relationships. When evaluated by regional richness, the robustness of Lepidoptera assemblages against secondary extinctions was higher under isolation-driven plant extinctions than regional plant extinction; however, this relationship was reversed when evaluated by averaged local richness. Also, with isolation-driven plant extinctions, Lepidoptera at the central sub-region of Baden-W & uuml;rttemberg appeared to be especially vulnerable. With Lepidoptera expansions, plants' functional distances in local communities dropped, indicating a possible increase of competition among plants, yet to a lesser extent particularly with upward movements. Together, our results suggested that the communities' composition context at the landscape scale (i.e., how communities, with respective species composition, are arranged within the landscape) matters when assessing global-change influences on interaction systems; spatially explicit consideration of such context can reveal localised consequences that are not necessarily captured via a spatially implicit, regional perspective.
