Multi‐scale responses of plant species diversity in semi‐natural buffer strips to agricultural landscapes

Question: How does agricultural land usage affect plant species diversity in semi‐natural buffer strips at multiple scales? Location: Lepsämä River watershed, Nurmijärvi, Southern Finland. Methods: Species diversity indicators included both richness and evenness. Plant communities in buffer strips were surveyed in 29 sampling sites. Using ArcGIS Desktop 9.0 (ArcInfo) and Fragstats 3.3 for GIS analysis, the landscape composition around each sampling site was characterized by seven parameters in square sectors at five scales: 4, 36, 100, 196, and 324ha. For each scale, Principle Component Analysis was used to examine the importance of each structural metric to diversity indicators using multiple regression and other simple analyses. Results: For all but the smallest scales (4 ha), two structural metrics including the diversity of land cover types and percentage of arable land were positively and negatively correlated with species richness, respectively. Both metrics had the highest correlation coefficients for species richness at the second largest scale (196 ha). The density of arable field edges between the fields was the only metric that correlated with species evenness for all scales, which had highest predictive power at the second smallest scale (36 ha). Conclusions: Species richness and evenness of buffer strips had scale‐dependent relationships to land use in agricultural ecosystems. The results of this study indicated that species richness depends on the pattern of arable land use at large scales, which may relate to the regional species pool. Meanwhile, species evenness depended on the level of field edge density at small scales, which relates to how the nearby farmland was divided by the edges (e.g. many small‐scale fields with high edge density or a few big‐scale fields with low edge density). This implies that it is important to manage the biodiversity of buffer strips within a landscape context at multiple scales.