Using image texture analysis to predict birds’ response to fine-scale habitat heterogeneity in semi-arid ecosystems

Contact: Véronique St-Louis

Collaborators: Anna M. Pidgeon and Volker C. Radeloff

Quantifying birds' broad-scale response to habitat heterogeneity in semi-arid ecosystems is not trivial. The spatial analysis tools most commonly used to characterize species response to landscape pattern, often based on the calculation of landscape indices from distinct cover classes, represent inherent limitations in semi-arid ecosystems due to the non-discrete nature of different cover types. Additionally, accuracy levels and costs associated with classified maps may limit their usefulness. In this research we are developing non-traditional image analysis tools to be able to better understand bird species responses to habitat heterogeneity in grassland and shrubland dominated habitat in a semi-arid landscape. We are using image texture analysis to predict breeding bird species richness and abundance. The first part of this project, i.e., building predictive models based on existing data, is conducted on the McGregor Range of Fort Bliss Military Reserve, New Mexico. We are relating bird data, collected at 42 sites between 1996 and 1998, to texture measures calculated from a series of digital orthophotos (DOQ's) and an unclassified Landsat scene taken in June 1996. Preliminary results suggest a strong, positive relationship between species richness and image texture. Texture measures derived from the Landsat image are better predictors of species richness than those derived from the DOQ's. There is a lot of variability among the models predicting species abundance from texture measures. Some species do not respond at all to texture, while others show very strong relationships. This is probably due to differences in life history, which we will investigate further in this research. The second part of the project will be to test the generality of these models within arid ecosystems by applying them to the avian community of the dry prairie at Fort McCoy, Wisconsin. Our approach proposes a non-traditional use of image texture to address specific ecological questions. We believe that this will improve the field of landscape ecology by providing techniques of quantifying heterogeneity with continuous variables.

Acknowledgements:
Financial support is provided by the Strategic Environmental Research and Development Program (SERDP) ,US Department of Defense.

Presentations:
St-Louis, V., A.M. Pidgeon, V.C. Radeloff & T.J. Hawbaker. 2005. Using image texture analysis to predict breeding birds' response to fine scale habitat variability in semi-arid environments 90th Annual meeting Ecological Society of America (ESA), Montréal, QC, Canada.