Abstract
An explicit consideration of spatial structure in ecological studies plays an increasingly important role in attempts to better understand and manage ecological processes, such as deforestation, forest homogenization, and escalating landscape heterogeneity. The goal of this chapter is to quantify the relationship between forest cover data and ovenbird (Seiurus aurocapilla) abundance¿a ground nesting passerine that breeds in contiguous forests¿in southern Ontario (Canada). To quantify this relationship, we use the Ontario Breeding Bird Atlas 2001¿2005 and compare two spatially explicit modeling methods: geographically weighted regression (GWR) and regression kriging (RK). We show how GWR and RK account for residual spatial autocorrelation in models of forest cover and ovenbird abundance, and we examine the insights they provide. Based on regression kriging, we found that 68 % (adjusted R 2 ) of ovenbird abundance was explained by forest cover, which was an improvement over ordinary least-square regression (adjusted R 2 = 43%), but was not uniformly better than variance explained by GWR in different subregions. These results emphasize the importance of both performing spatial data exploration prior to statistical analyses and accounting for spatial structure during the analysis
Original language | English |
---|---|
Title of host publication | Real World Ecology: large-scale and long-term case studies and methods |
Editors | S. Miao, S. Carsten, M. Nungesser |
Place of Publication | New York |
Publisher | Springer |
Pages | 137-160 |
ISBN (Print) | 9780387779416 |
DOIs | |
Publication status | Published - 2009 |