•  
  •  
 

Abstract

Many rangelands in the southwestern United States provide quality habitat for Northern Bobwhites (Colinus virginianus). These same habitats are frequently managed for livestock production and thus are subjected to various brush management practices that are meant to enhance forage production. Bobwhites rely on woody cover for food, thermal and loafing cover, and protection from avian predators. Implementing brush management practices that reflect bobwhite requirements is important for managing usable space and viable populations. We described the structural vegetative characteristics associated with bobwhite locations and random locations on native rangelands in the upper Rolling Plains of Texas that are dominated by honey mesquite (Prosopis glandulosa) and managed with aerial herbicide and prescribed fire. We also used binary logistic regression to model habitat selection; the initial model was built using 67% of the data (n = 179 used-random paired points) and then validated using the remaining 33% of the data (n = 88 used-random paired points). Locations used by bobwhites had significantly larger mean values of percent brush canopy cover, visual obstruction, and angle of obstruction than did random locations; random locations had a greater mean value of percent bare ground than locations used by bobwhites. The resulting logistic regression model contained only the angle of obstruction; the model had an 80% probability of correctly classifying used and random locations based on the area under the receiver operating curve (ROC). The model maintained a high classification probability when applied to the smaller validation data set, with an area under the ROC of 0.78.

Share

COinS