Use of sightability models and resource selection functions to enhance aerial population surveys of elk (Cervus elaphus) in Alberta
As part of the Central East Slopes Elk Study (CESES), this project was meant to provide meaningful elk population estimates to enhance current wildlife management. Using radio collared elk, a sightability model was developed to correct for elk missed during aerial surveys. During trials, if a radio collared elk was observed, 11 factors were recorded: light intensity, aspect, activity, topography, percent vegetation screening, vegetation class, percent snow cover, elk group size precipitation, temperature and observer experience. If the elk was not observed, the survey crew used telemetry receivers to locate the elk and record the same factors. A logistic regression approach was used to develop a correction based on environmental factors that affected sightability. Significant variables affecting sightability were, elk group size, percent vegetation screening, elk activity, percent snow cover and light intensity. Survey design can also increase precision of population estimates. When there is high spatial variation in animal numbers, spatial stratification is one approach by which the precision of estimates can be increased. This study compared a typical stratified random sample design using tree canopy for stratification to a stratification approach with strata using GIS-based covariates. This approach assumes that sample units with similar environmental covariates will have similar elk densities. GIS- based covariates were used to develop a winter elk resource selection function (RSF). The mean RSF value in each survey cell was used to post-stratify the survey cells for improved precision of the population estimate.