As a consequence of the increasing agricultural intensification, the area of natural and semi-natural grasslands declined worldwide (Reidsma et al., 2006). This decline created a network of isolated and fragmented grassland habitats. Habitat fragmentation has two main components. First, the total area of the habitat sustaining populations decreases. Secondly, these habitats tend to be more isolated (Saunders et al., 1991). Less mobile arthropod species, like ground beetles (Coleoptera: Carabidae), are especially sensitive to habitat loss and isolation (Samways, 2005; Lôvei et al., 2006). A central issue in conservation biology and nature management is whether characteristic species of a given habitat type could be preserved by fragmented habitat patches or not. The classical theory of island biogeography predicts that the number of species supported by an island increases with the area of the island (MacArthur and Wilson, 1967). Recently, however, several papers refined the oversimplified original assumptions of the island biogeography theory concerning habitat island, which emphasizes the effects of surrounding habitats on the species richness (Kupfer et al., 2006; Lôvei et al., 2006, Magura and Kôdôbôcz, 2007).
In the present case study evaluating ground beetles in sandy grasslands, we demonstrated that depending on the ratio of specialist and generalist species in an assemblage, the species-area relationship may be positive or negative.
Eight patches of formerly contagious sandy grasslands located in the Nyirség region (the Great Hungarian Plain, Eastern Hungary) were studied. In the 19th century, this region was covered by natural habitats (marshes, fen meadows, mires, sandy grasslands and sandy oak woods). During the 20th century, as a consequence of the intensification of farming and forestry, these habitats were abolished or became highly fragmented. Today, the fragmented sandy grassland patches are surrounded by arable lands and non-native tree plantations. Recently, the studied grassland patches (prevalent vegetation association was Potentillo arenariae-Festucetum pseudovinae) have been lightly grazed with cows and sheep (cattle density was less than 0.25 heads/ha). The matrix habitats surrounded by these patches were similar: non-native deciduous tree plantations (black locust and ennobled poplar species) and croplands (maize and corn). The area of the studied patches varied between 2.3-353.5 ha and the distance between the patches was at least 2 km.
Ground beetles of the eight sandy grasslands were collected during three years (20012003) using unbaited pitfall traps, consisting of plastic cups (diameter 100mm, volume 500ml) with 70% ethylene glycol as a killing and preserving solution. There were 10 traps, scattered randomly within the individual patches (at least 100m from the grassland edges).
Traps were checked fortnightly from the end of March to the end of October in every year. Ground beetles were identified to species using the keys of Hurka (1996). To ensure a more complete species inventory in the studied grassland patches, beetles caught from the three trapping years were pooled.
Based on the literature data (Hurka, 1996), considering local conditions, collected ground beetle species were divided into two ecological groups according to their habitat preference: habitat specialist species (open-habitat species associated with sandy soils) and generalist species (species occurring likewise in both the closed and open canopy habitats and are not associated with sandy soils). The relationship between the patch area and the number of ground beetle species was examined by linear regression analysis (Kutner et al., 1996). We analyzed separately the total number of ground beetle species collected in the fragment, as well as the ratio of habitat specialists and generalists to the total number of species. Analyzing the ratio (or relative frequency) instead of the actual numbers removes the inherent differences in species richness among the studied patches. The distribution of data used in the linear regression analyses was normal (tested by the Kolmogorov-Smirnov test, Sokal and Rohlf, 1995).
Was this article helpful?