Testing the spectral variation hypothesis by using satellite multispectral images

In the present paper, a test of the spectral variation hypothesis (SVH) was performed using multispectral high resolution satellite data. The SVH was tested by comparing the relationship between the spectral heterogeneity and species richness in plots of different size (10010000 m(2)) in a complex wetland ecosystem, the "Montepulciano Lake", Central Italy. The nature reserve of the Montepulciano Lake is centered on a 100 ha shallow lake surrounded on three sides by a Phragmites australis and Carex sp. pl. marsh of about 280 ha. The monitoring pro-ram for the reserve vegetation started in 2002 and is based on the analysis of 1,100 m(2) and 1 ha (10000 m(2)) Plots, organized in such a manner that four of the smaller size plots are nested, following a random design, within a larger one. Data on species composition and community structure were collected in the plots and stored in a GIS-linked archive. A multispectral Quickbird satellite image (3 m spatial resolution) acquired of the wetland and lake ecosystem during the same period was radiometrically and geometrically corrected. We performed an analysis to examine the use of spectral heterogeneity using the four visible and infrared wavebands of the satellite image to predict species richness at the different spatial scales. The spectral heterogeneity was found to explain about 20% of the variance of species richness at the 100 m(2) scale and about 50% at the 1 ha scale. It was concluded that multispectral high resolution satellite data can contribute to the biodiversity assessment of complex wetland ecosystems. (C) 2004 Elsevier SAS. All rights reserved.