Roads affect the spatial structure of butterfly communities in grassland patches.

Roads may have an important negative effect on animal dispersal rate and mortality and thus the functioning of local populations. However, road verges may be surrogate habitats for invertebrates. This creates a conservation dilemma around the impact of roads on invertebrates. Further, the effect of roads on invertebrates is much less understood than that on vertebrates. We studied the effect of roads on butterflies by surveying abundance, species richness and composition, and mortality in ten grassland patches along high-traffic roads (∼50-100 vehicles per hour) and ten reference grassland patches next to unpaved roads with very little traffic (<1 vehicle per day) in southern Poland. Five 200-m transects parallel to the road were established in every grassland patch: at a road verge, 25 m from the verge, in the patch interior, and 25 m from the boundary between the grassland and field and at the grassland-arable field boundary. Moreover, one 200-m transect located on a road was established to collect roadkilled butterflies. The butterfly species richness but not abundance was slightly higher in grassland patches adjacent to roads than in reference grassland patches. Butterfly species composition in grasslands adjacent to roads differed from that in the reference patches. Proximity of a road increased variability in butterfly abundances within grassland patches. Grassland patches bordering roads had higher butterfly abundance and variation in species composition in some parts of the grassland patch than in other parts. These effects were not found in reference grassland patches, where butterfly species and abundance were more homogenously distributed in a patch. Plant species composition did not explain butterfly species. However, variance partitioning revealed that the presence of a road explained the highest proportion of variation in butterfly species composition, followed by plant species richness and abundance in grassland patches. Road mortality was low, and the number of roadkilled butterflies was less than 5% of that of all live butterflies. Nevertheless, the number and species composition of roadkilled butterflies were well explained by the butterfly communities living in road verges but not by total butterfly community structure in grassland patches. This study is the first to show that butterfly assemblages are altered by roads. These results indicate that: (1) grassland patches located near roads are at least as good habitats for butterflies as reference grassland patches are, (2) roads create a gradient of local environmental conditions that increases variation in the abundance of certain species and perhaps increases total species richness in grassland patches located along roads, and (3) the impact of roads on butterflies is at least partially independent of the effect of plants on butterflies. Furthermore, (4) the direct impact of road mortality is probably spatially limited to butterflies living in close proximity to roads.