Assessment of fine-scale plant species beta diversity using WorldView-2 satellite spectral dissimilarity

Plant species beta diversity is influenced by spatial heterogeneity in the environment. This heterogeneity can potentially be characterised with the help of remote sensing. We used WorldView-2 satellite data acquired over semi-natural grasslands on The Baltic island of Oland (Sweden) to examine whether dissimilarities in remote sensing response were related to fine-scale, between-plot dissimilarity (beta diversity) in non-woody vascular plant species composition within the grasslands. Fieldwork, including the on-site description of a set of 30 2 m x 2 m plots and a set of 30 4 m x 4 m plots, was performed to record the species dissimilarity between pairs of same-sized plots. Spectral data were extracted by associating each plot with a suite of differently sized pixel windows, and spectral dissimilarity was calculated between pairs of same-sized pixel windows. Relationships between spectral dissimilarity and beta diversity were analysed using univariate regression and partial least squares regression. The study revealed significant positive relationships between spectral dissimilarity and fine-scale (2 m x 2 m and 4 m x 4 m) between-plot species dissimilarity. The correlation between the predicted and the observed species dissimilarity was stronger for the set of large plots (4 m x 4 m) than for the set of small plots (2 m x 2 m), and the association between spectral and species data at both plot scales decreased when pixel windows larger than 3 x 3 pixels were used. We suggest that the significant relationship between spectral dissimilarity and species dissimilarity is a reflection of between-plot environmental heterogeneity caused by differences in grazing intensity (which result in between-plot differences in field-layer height, and amounts of biomass and litter). This heterogeneity is reflected in dissimilarities in both the species composition and the spectral response of the grassland plots. Between-plot dissimilarities in both spectral response and species composition may also be caused by between-plot variations in edaphic conditions. Our results indicate that high spatial resolution satellite data may potentially be able to complement field-based recording in surveys of fine-scale species diversity in semi-natural grasslands. (C) 2013 Elsevier B.V. All rights reserved.