COS 69-1
A centroid model of species distribution with applications to the Carolina Wren (Thryothorus ludovicianus) and House Finch (Haemorhous mexicanus) in the United States

Wednesday, August 12, 2015: 8:00 AM
338, Baltimore Convention Center
Qiongyu Huang, Geographical Sciences, University of Maryland, College Park, MD
Ralph Dubayah, Department of Geography, University of Maryland, College Park, MD
John Sauer, USGS Patuxent Wildlife Research Center, Laurel, MD
Anu Swatantran, Geographical Sciences, University of Maryland, College Park, MD

Drastic shifts in species distributions are a cause of concern for ecologists. Such shifts pose great threat to biodiversity especially under unprecedented anthropogenic and natural disturbances. Many studies have documented recent shifts of species distribution. However, most of these studies are limited to regional scales, and do not consider the abundance structure within species range. Developing methods to detect systematic changes of species' distributions over their full ranges is critical for understanding the impact of changing environments and for successful conservation planning. Here, we demonstrate a centroid model for range-wide analysis of distribution shifts using the North American Breeding Bird Survey data. The centroid model is based on a hierarchical Bayesian framework which models population change within physiographic strata while accounting for several factors affecting species detectability. Yearly abundance-weighted range centroids are estimated. As case studies, we derive annual centroids for the Carolina Wren and House Finch in their ranges in the U.S. as well as in two sub-ranges. We further evaluate the first-difference correlation between species’ centroid movement and changes in winter severity, total population abundance. We also examined associations of change in centroids from sub-ranges. 


Change in full-range centroid movements of Carolina Wren significantly correlate with snow cover days (r=-0.58). For both species, the full-range centroid shifts also have strong correlation with total abundance (r=0.65, and 0.51 respectively). The movements of the full-range centroids of the two species are correlated strongly (up to r=0.76) with that of the sub-ranges with more drastic population changes. Our study demonstrates the usefulness of centroids analyzing distribution changes in a two-dimensional spatial context. Particularly it highlights applications that associate the centroids with factors such as environmental stressors, population characteristics, and progression of invasive species. Routine monitoring of changes in centroid will provide useful insights into long-term avian responses to environmental changes.