||Habitat destruction and fragmentation of remaining habitat are major threats to global biodiversity. In this paper, we drew upon data from grassland butterflies, legume-feeding herbivores and their parasitoids, and the interactions between rape pollen beetles and their paraitoids in the agricultural landscapes of Germany to explore the following issues: (1) the relative importance of small habitat fragments for the conservation of biodiversity (in contrast to the prevailing arguments in favor of large fragments); (2) the disruption of interspecific interactions in fragmented habitats and (3) the relative importance of the spatial arrangement of habitat fragments in landscapes of different complexity. The percentage of polyphagous butterfly species and their abundance were higher in small than in large calcareous grassland fragments, showing the relative importance of the landscape surrounding habitat fragments for less specialized species. A landscape perspective is also needed to explain why several small fragments supported more butterfly species (even when only endangered species were considered) than the same area composed of only one or two fragments. Analyses of insects on legumes showed trophic-level differences, in that species numbers of parasitoids, but not of herbivores, benefited from habitat subdivision in landscapes. As percentage of parasitism (i.e., the strength of ecological interactions) increased with fragment area, both the "several small" and "single large" strategies appeared to have merit. An intermediate-fragmentation strategy of habitat conservation in human-dominated landscapes may combine the advantages. Small habitat fragments should be scattered enough to cover a range of geographical area wide enough to maximize beta diversity and the spreading of but with large habitat fragments close enough to enable dispersal among fragments, to reduce the extinction probability of area-sensitive species, and to stabilize predator-prey interactions. Parasitism of rape pollen beetles exhibited a distinct edge effect: it was higher near the crop field edge, i.e., near the parasitoids’ overwintering sites (such as grassy strips). However, this was only true in landscape, dominated by annual crops in landscapes with a high percentage of permanent noncrop area (>20%), such edge effects disappeared, presumably because of the high overall density or these parasitoids. These data indicate that spatial configuration is important to mitigate extinction risks when habitat availability in a landscape is low, whereas no effect will be observed when overall area of habitat is high.