Appendix E. A supplementary discussion on differences in predictability between arthropod assemblages.
Although the main interest in our study lies in the comparison between predictor factors, not in comparing arthropod groups, it is interesting to note the remarkable differences in over-all predictability between arthropod groups. This variation should be interpreted with care, however, since different arthropod groups were sampled using different techniques. Each method has its own characteristics, leading to intrinsic differences in sampling intensity between the groups.
Another obvious difference between the arthropod groups sampled is that they differ widely in species richness. The predictability of an arthropod group can be expected to depend on this. Groups with few species might be easier to predict than larger groups, simply because there is less variation that needs to be accounted for. The high over-all predictability for grasshoppers and hoppers (20 and 23 species only) fits this methodological expectation.
The relatively weak relationships for hoverflies and bees, however, can not be ascribed to such effects as the two groups do not contain exceptionally high, but rather intermediate species numbers (61 and 91, respectively). For comparison: the spider data set is similar in size (73 species) and the number of weevil and carabid species is even higher (115 and 126), while each of these three groups shows much higher prediction levels.
Since both the bees and hoverflies are flower visitors, we might have expected to see more distinct relationships with flower density and diversity (Thomson 1981, Salveter 1998, Steffan-Dewenter and Tscharntke 2001) but also the performance ofthis factors was poor. Hoverflies and bees forage using flight, and movement over larger distances is normal all-day behavior. Their relatively large home range could be the cause of their low predictability by local factors. This may lead us to expect better predictions by landscape composition for these insect groups, but prediction levels nevertheless remain below those of the local plant species composition.
Similar patterns can be seen from other studies as well. Jeanneret et al. (2003), in a study covering local as well as landscape factors, also showed that the species composition of flying foragers (butterflies) could not be explained as well as that of other groups (carabids and spiders). We suggest that the connectivity of habitat fragments (Steffan-Dewenter 2003, Webb and Hopkins 1984) or the availability of miniature landscape features ("partial habitats") may be more important for these flying specialist groups than over-all landscape composition.
LITERATURE CITED
Jeanneret, Ph., B. Schüpbach, L. Pfiffner, F. Herzog, and T. Walter. 2003. Arthropod reaction to landscape and habitat features in agricultural landscapes. Landscape Ecology 18:253–263.
Salveter, R. 1998. Habitat use of adult syrphid flies (Diptera: Syrphidae) in a highly diversified agricultural landscape. Mitteilungen der Schweizerischen Entomologischen Gesellschaft 71:49–71.
Steffan-Dewenter, I. 2003.Importance of habitat area and landscape context for species richness of bees and wasps in fragmented orchard meadows. Conservation Biology 17:1036–1044.
Steffan-Dewenter, I., and T. Tscharntke. 2001. Succession of bee communities on fallows. Ecography 24:83–93.
Thomson, J. D. 1981. Spatial and temporal components of recourse assessment by flower-feeding insects. Journal of Animal Ecology 50:49–59.
Webb, N. R., and P. J. Hopkins. 1984. Invertebrate diversity on fragmented Calluna heathland. Journal of Applied Ecology 21:921–933.