Whooping Crane Migrational Habitat Use Draft (2)

Home Browse Search

Trends in the Index of Use The change in the frequency of use of the study area by whooping cranes throughout the CA period was estimated with the observed trend in an index of use. The index of crane use was defined as the number of crane groups observed per number of flights (PRESP 2005) and was calculated for each monitoring season. Survey detection rates were incorporated into this analysis through adjustment of the observed number of cranes by the estimated probability of detection during the flight, based on the predictive detectability model (Thompson 1992) and the recalculation of the index of use with the adjusted number of crane groups. Simple linear regression with normal errors was used to estimate the change in the index through time. Time was represented by the year and the proportion of year at the midpoint of the survey. The winter peak count of the Aransas-Wood Buffalo population was included as a covariate. Activity Summary The percentage of time a crane group was engaged in each activity was calculated for an observation session, where an observation session is defined as all the time a crane group was observed during a day. These percentages were averaged across all observation sessions recorded and the variance of this average was used to calculate confidence intervals. The summaries were made by migration season (spring and fall) and for both migration seasons combined and represent diurnal activities. Habitat Selection Characteristics of whooping crane habitat selection in the central Platte River was modeled through the development of resource selection functions (RSF). The basic premise of resource selection modeling (Manly et al. 2002) is that resources (which may be food items, land cover types, or any quantifiable habitat characteristic) that are important to animals will be "used" disproportionately to the availability of those resources in the environment. The RSF model uses an exponential function to model and predict the relative probability of use from the landscape characteristics. In this analysis, the characteristics at the used locations were contrasted to characteristics at randomly selected "available" locations in the same region. There were two sets of used locations for the RSF analyses: all locations (in and out of channel) for summarizing study area selection, and the subset of in-channel locations for summarizing in- channel selection. A location is defined as a point in space used by a crane group at one point in time. These locations represent multiple re-locations of the same group of birds but since individual cranes were not identifiable in the field and observations of whooping cranes are rare, we have maintained all the observations in the dataset we analyzed. For the models relating landscape level characteristics to crane group selection in the entire study area, we defined the available set of points with a 500x500m grid randomly placed over the entire study area. The intersection of the gridlines results in 6,420 available points. A subset of 189 points were in the 'wetted channel' or "barren beach/bar" designation of the 1998 land cover/use GIS layer and comprised the available points for the in-channel analysis. 4