PARENT SESSION
PS1 - Wildlife Toxicology
Sunday, 17 November 2002
8:00 AM to 6:30 PM
Exhibit Hall
(P032) Using CART to understand habitat and contaminant factors on bald eagle productivity in Michigan.
Eisenreich, Karen*,1, Steen, Carrie2, Best, David3, Postupalsky, Sergej4, Bowerman, William1, Johnson, Alan1, 1 Clemson University, Pendleton, South Carolina, United States2 Michigan State University, East Lansing, Michigan, United States3 U.S. Fish and Wildlife Service, East Lansing, Michigan, United States4 Raptor Biologist, Madison, Wisconsin
ABSTRACT- The bald eagle is one of the most studied birds in North America. A significant amount of information exists on the reproductive effects of various stressors. There is a forty-year database that documents reproductive success and contaminants including mercury, polychlorinated biphenyls (PCBs), dieldrin, and DDT. There is a continuing statewide monitoring project in place to monitor bald eagle productivity and contamination levels. Analysis of these data can identify trends that may not be observed within a smaller subset of data. When interfaced with GIS, the organization, synthesis, and analysis of these data can determine spatial and temporal relationships over large geographic areas. Classification and regression trees (CART) and simple linear regression were used to elucidate the relationships between eagle recovery and contaminants, distance of each nest to the nearest major road, distance of each nest to the nearest body of water, and land cover within 500m of each nest. Models were constructed for each five-year interval from 1981 to 2000. The explanatory variables best able to predict bald eagle productivity were area of water within 500m of each nest for 1981-1985, p,p′-DDE in eaglet plasma for 1986-1990, p,p′-DDE in eggs for 1991-1995, and the area of agriculture within 500m of each nest for 1996-2000. Regression analysis similarly found significant correlations with productivity for most of the explanatory variables identified by CART analysis. No single variable explains a large portion of the variance in productivity of Michigan bald eagles, however, several contaminant levels and features of the physical habitat are clearly important. This study, while preliminary, has established methods to better understand patterns in bald eagle productivity, and in addition has created many future opportunities to build on previously completed work.
Key words: bald eagle, classification and regression trees, productivity, michigan