Whooping Crane Migrational Habitat Use Draft (2)

Home Browse Search

Habitat Selection - On the Ground Characteristics Habitat selection models of characteristics measured on the ground documented the use of channels with large unobstructed views. The model, which predicts the probability of use as a function of obstruction to obstruction width, estimated the maximum probability occurs when obstruction to obstruction width was 343 meters or 1,125 feet, and then declines beyond this point (Figure 22). This is not wetted width or channel width as distances to obstruction frequently extend outside of the channel across agriculture or hay fields, large sand bars, etc. This result supports the assertion that whooping crane groups use at a disproportionate rate, areas with wide unobstructed views, as measured across the channel. On the other hand, the probability of whooping crane group use actually began to decline once the obstruction to obstruction width exceeded 343 meters. Interestingly, distance to closest visual obstruction, whether across the channel or upstream/downstream, was not in the best model (with AIC selection criterion) for whooping crane habitat selection. Substrate abundance (e.g., sand) was also not in the best model (with AIC selection criterion) for whooping crane habitat selection. There was no difference in the biological interpretation between the model developed for the systematic observations and the combined systematic and opporiunistic observations. Model selection for the combined dataset led to the same final models, and there was no biological difference in model parameterization. The two models had highly correlated predictions and very similar maximum points in the quadratic relationship (339 meters). Habitat Selection - Flow Dependent Characteristics Habitat selection models relating to flow dependent characteristics indicated significant crane selection for greater channel widths up to a point, then the relative probability of selection decreased with increased width. Again, the predictive ability of these models is limited, as specific values are estimated with error. According to this model, predicted the maximum probability of selection occurred when wetted width was 300 meters or 984 feet. The model also indicated significant selection for large depths, but it is unclear how a crane could detect these deeper water areas. It is possible that depth is a secondary factor that comes in to play as more water enters the system to make wider channels. Local area selection of flow dependent characteristics also indicated significant selection for large depths and widths. There was no difference in the biological interpretation of flow dependent models developed for the systematic observations and the combined systematic and opportunistic observations. Model selection for the combined dataset led to slightly different final models, though there was no biological difference between the model parameterization and the interpretation of the common effects in the models. In summary, when considering all models together the habitat selection analysis suggested that cranes prefer wider channels, wider unobstructed widths and agricultural lands in the absence of transportation corridors and forest and shrub cover. These preferences appear to indicate an interaction of channel characteristics and upland area characteristics. For example, cranes 17