Using Worldview-2 satellite imagery to detect indicators of high species diversity in grasslands

Abstract: The high small-scale diversity of plant species in semi-natural grasslands can be seen as a function of environmental conditions and land use history. This study explores the potential of using Worldview-2 spectral imagery and accessible GIS data to identify a set of vegetation characteristics known to influence biodiversity in semi-natural grasslands. Field sampling was done in 52 grassland sites, with presence and frequency of plant species and vegetation structural composition recorded in 4 m x 4 m plots. Plant species data were used to calculate overall species richness, grassland specialist richness, grassland generalist richness and Ellenberg indicator values for reaction (R), nutrients (N), soil moisture (M) and light (L). Generalized Additive models (GAM) were constructed to explain observed vegetation variables, predicted by mean values and standard deviations of WordView-2 satellite spectral reflectance and GIS data of grassland habitat area, soil type and land use history. The study was carried out on two spatial scales: 4m x 4m plots and grassland sites (0.25 ha - 14 ha). The results show that high resolution satellite imagery has potential of characterizing species diversity indirectly by the habitat productivity and heterogeneity. Grassland habitats with high small-scale species diversity had relatively low spectral heterogeneity. It was difficult to measure species diversity on a fine spatial scale using only remote sensing variables. Grassland management history is a very good predictor of species composition and diversity, especially for specialized grassland species. Ellenberg values for soil moisture (M) and nutrients (N) were successfully modelled using remote sensing data. In grasslands where the species diversity is largely driven by environmental gradients like nutrients or soil moisture, ecological indicators can be used as an alternative to species diversity to assess habitat quality.