Satellite image texture and a vegetation index predict avian biodiversity in the Chihuahuan Desert of New Mexico

Abstract

Predicting broad‐scale patterns of biodiversity is challenging, particularly in ecosystems where traditional methods of quantifying habitat structure fail to capture subtle but potentially important variation within habitat types. With the unprecedented rate at which global biodiversity is declining, there is a strong need for improvement in methods for discerning broad‐scale differences in habitat quality. Here, we test the importance of habitat structure (i.e. fine‐scale spatial variability in plant growth forms) and plant productivity (i.e. amount of green biomass) for predicting avian biodiversity. We used image texture (i.e. a surrogate for habitat structure) and vegetation indices (i.e., surrogates for plant productivity) derived from Landsat Thematic Mapper (TM) data for predicting bird species richness patterns in the northern Chihuahuan Desert of New Mexico. Bird species richness was summarized for forty‐two 108 ha plots in the McGregor Range of Fort Bliss Military Reserve between 1996 and 1998. Six Landsat TM bands and the normalized difference vegetation index (NDVI) were used to calculate first‐order and second‐order image textures measures. The relationship between bird species richness versus image texture and productivity (mean NDVI) was assessed using Bayesian model averaging. The predictive ability of the models was evaluated using leave‐one‐out cross‐validation. Texture of NDVI predicted bird species richness better than texture of individual Landsat TM bands and accounted for up to 82.3% of the variability in species richness. Combining habitat structure and productivity measures accounted for up to 87.4% of the variability in bird species richness. Our results highlight that texture measures from Landsat TM imagery were useful for predicting patterns of bird species richness in semi‐arid ecosystems and that image texture is a promising tool when assessing broad‐scale patterns of biodiversity using remotely sensed data.