Tall-grass and mixed-grass prairie biomes have experienced extensive anthropogenic land cover change inducing habitat loss/fragmentation, changes to landscape structure and degraded patch quality. Such changes could impact prairie insect communities via a number of mechanisms. 1) Patch area: via species-area relationships. 2) Patch quality: vegetation changes can reduce habitat suitability. 3) Landscape composition: novel land cover elements could supplement habitat or act as population sinks. 4) Landscape edges: edges between land cover elements may impact movement. 5) Landscape diversity: more land cover types could increase species richness via complementation. 6) Landscape configuration: the shape of patches in the landscape influences edge effects. The relative importance and, for the landscape mechanisms, the scale at which these mechanisms function is unclear. To investigate the relative roles of these mechanisms on insect communities we trapped insects on 20 native prairie patches in western Minnesota and eastern North Dakota. Vegetation composition and density were measured at each patch. Land cover in the surrounding landscape was determined from digitized aerial photographs. Various measures of landscape composition, edges, diversity and configuration were calculated in GIS at multiple 100 m buffers out to a total distance of 1000 m.

Results/Conclusions

For each landscape variable we used multi-model inference to identify the scale where that variable had the strongest effect. For both Coleopterans and Orthopterans most landscape variables had the strongest impact at relatively large scales (700 to 1000 m). For each of the six potential mechanisms we constructed separate multiple linear regression models using appropriate patch level variables or the landscape variables measured at the best scale. For mechanisms with multiple variables we used AICc scores to simplify the full regression model to the best regression model. Finally, the best multiple regression models for each of the six mechanisms were compared using multi-model inference. For Coleopterans all landscape mechanisms influenced diversity but only landscape configuration and composition influenced richness, based on ΔAICc. Similarly, for Orthopterans landscape configuration alone influenced diversity while landscape edge, configuration and diversity, and patch quality and area all influenced richness. For prairie insect communities the nature of the surrounding landscape out to a distance between 700 and 1000 m had as much or more statistical support than patch variables. Our findings suggest that communities on remnant habitat patches are impacted as much, if not more so, by landscape context than the quality of the patch itself.