Production and implementation of a habitat suitability model for breeding Bald Eagles in the lower Chesapeake Bay (Phase II: Model construction through habitat mapping)
Since its elevation to endangered status in 1978, protection of the Bald Eagle and its habitat is governed by the Endangered Species Act. Under this designation, critical habitat is defined as any area essential to the survival and recovery of the species. Current habitat management strategies for nesting Bald Eagles are centered around the protection of active nest trees. Although this practice is essential, it does not address potential nesting habitat. Much habitat remains unoccupied that is both critical to the continued recovery and maintenance of the population and is under imminent risk of development. We quantified 61 topographic, land-use, and disturbance variables within 127 active eagle territories and around 127 randomly chosen points to evaluate their potential as predictors of habitat quality for breeding Bald Eagles. Fifty-four of 61 variables were significantly different between the two samples. Compared to random sites, eagles prefer to nest in areas situated close to large water bodies, away from extensive human disturbance, and having considerable forest cover. A discriminant function analysis was used to determine the linear combination of variables that best differentiate between active and random sites. Sixteen variables conformed to parametric assumptions and were entered into a step-wise discriminant function procedure. The final 4-variable model constructed produced a classification accuracy of 81.5%. In addition to the model variables, 4 distribution constraints were identified within the data set. A combination of these constraints and the 4-variable model were used in the final land classification model. The final model was used to classify lands along a 100 mi. reach of the James River and a 75 mi. reach of the Rappahannock River. All lands along these drainages or their tributaries that fell within 3 km of a channel at least 250 m wide were classified. This land mass included over 2,300 square kilometers. Classification of the area was accomplished by establishing a network of over 15,000 registration points, parameterizing the model variables for each point independently, and employing the classification model. A substantial portion of the James and Rappahannock River drainages (458 and 274 square kilometers respectively) was classified as unsuitable due to high housing density and/or the lack of adequate nesting substrate. However, a comparable portion of both drainages was also found to contain either good or very good habitat for breeding (344 and 349 square kilometers for the James and Rappahannock Rivers respectively). Relationships between habitat quality and the model variables were consistent with those expected based on the univariate results.