Multiscale effects for landscape metrics and species diversity under the different disturbance: A case study of Gongyi city

Abstract

Heterogeneity and stability of the agricultural landscape is the premise to maintain the stable and high efficient agricultural ecosystem. Currently, habitat loss and the changes of landscape structure caused by human activities have become the important reason for the loss of biodiversity of the agricultural landscape. Although there are numerous metrics available to investigate the spatial structure of landscape, only little empirical research has examined these metrics to indicate the plant species diversity at several scales. The study area is in Gongyi city which is distributed in the mountainous and hilly areas of MiddleLower Yellow River. We select three types of habitat, such as shrub, plantation and farmland edge to investigate plant species by the typical sample method. These three types of habitat also represent three different disturbance levels. Shrub corresponds to the natural habitat, plantation is under semiNatural habitat, and the farmlandedge habitat is under the strongest human disturbance level. Natural and semiNatural habitats in agricultural landscapes are of vital importance for preserving biodiversity in agroecosystems. Based on the principles of Landscape Ecology, under the technical support of Geographic Information System (GIS), each sample plot is considered as the center to set different radius (150, 250, 500, 750, 1000, 1250, 1500m) of buffers. In each buffer, we extract six classes of landscape metrics, which include patch shape, edge contrast, similarity and proximity, landscape diversity, texture, patch size and patch density, a total of 52 landscape metrics. By applying the redundancy analysis (RDA), the landscape metrics that affect plant species diversity significantly in agriculture landscape are extracted from different scales. Through the analysis of the RDA, the result shows that different extent, the relationship between landscape metrics and species diversity changes significantly. In terms of shrub habitat, the landscape metrics of SHAPE_AM and PARA_AM can well support the species diversity at the extent of 500-750m, and the cumulative percentage variance can get to 33.6%. For plantation habitat, the landscape metrics of SHAPE_AM, AREA_CV, SIMI and PAFRAC can well explain the species diversity at the extent of 1000-1250m, and the cumulative percentage variance is 48.1%. For farmland edge habitat, the GYRATE_CV, ENN_ CV, PARA_AM and FRAC_AM are major metrics have strong relationship with species diversity, and the cumulative percentage variance is 32%. At the extent of 750-1250m, Simpsons diversity index (SIDI) shows significant relation with shrub plants; while the metric of ENN_CV only has correlation with farmland plants at the extent of 1000-1250m. This study indicates that the effects of landscape metrics on plant species diversity are strongly depended on the spatial scales, since no metric has constant relation with plant diversity across all extents. This study implies that some landscape metrics can be good indicators for revealing diversity characteristics of different habitats. To get a better relationship of the landscape heterogeneity and species diversity, we recommend further research about scale effects of the landscape metrics and much more application studies of these metrics in Landscape Ecology.