Whooping Crane Migrational Habitat Use Draft (2)

Home Browse Search

Predictive Maps We used the model coefficients to predict the relative probability of selection for the habitat parameter values at the available sample of points. Land cover maps were based on the land use/cover parameters from the 1998 BOR GIS. Flow dependent models were mapped at three flow levels, based on the quartiles of observed daily flow at the Kearney, NE gage during all crane use days. Flow parameters were output from HECRAS for the 165 base transects during the three flow levels, 250, 350, and 786 cfs. RESULTS There were 353 observations of crane groups recorded in the study area during the 11 monitoring seasons of the Cooperative Agreement time period (Figure 1; Table 5). Of the 320 locations within the land use/land cover GIS layer (BOR 2000), 150 were in the wetted channel, 143 in agriculture, 21 in grass, 3 in shrub/forest, and 3 in wet grass (Table 6). There were 72 observations of crane groups detected with the systematic aerial survey flights (Figure 2). There were 58 observations of crane groups in the wetted channel detected with the systematic aerial survey flights, 10 were in agriculture, 2 in grass, 1 in shrub/forest, and 1 in wet grass. Aerial Survey Detection Rates A predictive model was developed for use in analyses conducted with the monitoring data. The final predictive model contained parameters for strata, contractor, and altitude. The form of the final model was: P(det) = exp[ -1.10 + 2.74(Strata) + 1.42(ContractorAIM) + 0.97(Contractor GREYSTOrrE) + 3.34(Contractor oTTERTAIL) - 1.89(Altitude)] where Strata was 1 for the river and 0 for the upland, Contractor coefficients take on the value of 1 for a given contractor and 0 for WEST, and Altitude was 1 for 750 feet and 0 for 1000 feet. The standard errors of the estimates were 0.56, 0.54, 0.49, 0.59, 0.79, and 0.62 respectively. The model was used to estimate the probability of detection for each flight. The sum and average predicted values across the flights estimates the overall detectability for a survey (Figure 3; Table 7). The average predicted probability of detection for each survey ranged from 0.34 to 0.78. Imperfect detection probability of whooping cranes within the study area will bias the frequency of use estimates and any conclusions based on the observed spatial distribution. The aerial survey detection rate study has documented differences in the estimated detection probability, which will be accounted for in the analyses of these data. The estimation of the detection bias with the final predictive model enabled the calculation of an index of use, trend, and resource selection parameters which are unbiased with respect to crane group detection in the study area. 10