Get access



  • William Gould

    1. Institute of Arctic and Alpine Research (INSTAAR), University of Colorado, Campus Box 450, Boulder, Colorado 80309-0450 USA
    Search for more papers by this author
    • Present address: Northern Ecosystem Analysis and Mapping Lab, Institute of Arctic Biology, University of Alaska, 311 Irving I, PO Box 757000, Fairbanks, Alaska 99775-7000 USA. E-mail:


Understanding mesoscale patterns of ecosystem properties is important if we are to effectively monitor ecosystem change due to land use and climate change. Remote sensing provides the best tool for looking at large areas of the earth's surface to analyze, map, and monitor ecosystem patterns and processes. Patterns of vegetation and variation in biodiversity are important ecosystem properties, with strong relationships to important ecosystem functions. Species richness is the most widely used measure of biodiversity, and mapping patterns of species richness within a landscape can provide a basis for future monitoring and an ecological basis for land management and conservation decisions.

This study presents (1) a map of the vegetation of the Hood River region of the Central Canadian Arctic derived from a supervised classification of Landsat Thematic Mapper (TM) satellite imagery, (2) estimations and maps of regional variation in plant species richness, and (3) a comparison of three species richness estimation techniques. The three vascular species richness estimates are derived from measures of variation in normalized difference of vegetation index (NDVI), abundance of mapped vegetation types weighted by relative potential species richness, and a multiple regression of both these variables for 17 sampling sites of 500 pixels each.

Ground-based measures of species richness range from 69 to 109 vascular plant species per 0.5-km2 sample area. Variation in NDVI is positively correlated with measured species richness and a weighted abundance of mapped vegetation types. Multiple regression indicates that variation in NDVI and weighted abundance of mapped vegetation types explain 79% of the variance in ground-based measures of species richness. Three methods for remotely estimating species richness agree to within ±15 species over 60% of the area and ±30 species over 93% of the area.