Whooping Crane Migrational Habitat Use Draft (2)

Home Browse Search

The standard errors for the coefficients were 0.77, 0.03, < 0.01, and <0.01. This model indicated significant selection for local in-channel areas with large proportions of open water, and wetted channel. The relative probability of selection was low for small distances to transportation and increased with large distances to transportation (Figure 21). Study area selection of characteristics measured on the ground The final resource selection model, w(x), contained the linear and quadratic effects of obstruction to obstruction width (o_to_owidth). The form of the final model was: W(x) = exp[(0.014*o to_owidth) - (0.001* o_to_owidth2)] The standard errors for the coefficients were <0.01 and <0.01 for the linear and quadratic effects respectively. This model indicated significant selection for large obstruction to obstruction widths up to a point, and then the relative probability of selection decreased with increased obstruction to obstruction width (Figure 22). The quadratic function was maximized at 343 meters (80% CI: 324, 363). The models estimated from the combined systematic and opportunistic samples are in Appendix D. Study area selection of flow dependent characteristics The final resource selection model, w(x), contained the linear and quadratic effects of width and the linear effect of depth. The form of the final model was: W(x) = exp [(0. 020 *width) - (0.001 *width2) + (5.629*depth)] The standard errors for the coefficients were <0.01, <0.01, and 1.55 for the linear effect of width, the quadratic effect of width, and depth respectively. This model indicated significant selection for greater widths up to a point, and then the relative probability of selection decreased with increased width (Figure 23). The quadratic function was maximized at 300 meters (80% CI: 237, 460). The model also indicated significant selection for greater depths. We also predicted the relative probability of selection for the in-channel area at low, medium, and high flows (Figures 24-26). In general, the abundance of areas with higher probability use increased as flows increased throughout the study area, but particularly in areas west of Kearney, Nebraska. Local area selection of flow dependent characteristics The final resource selection model, w(x), contained the linear and quadratic effects of depth and the linear effect of width. The form of the final model was: W(x) = exp[(2.344* depth) +(5.382* depth2) +(0.004* width)] The standard errors for the coefficients were 5.36, 8.39, and <0.01 for the linear effect of depth, the quadratic effect of depth, and width respectively. This model indicated significant selection for greater depths (Figure 27) and widths. 14