Estimating sightability of greater sage-grouse at leks using an aerial infrared system and N-mixture models
Counts of grouse present at leks (breeding grounds) during spring are widely used to monitor population numbers and assess trends. However, only a proportion of birds available to count are detected resulting in a biased population index. We designed a study using an aerial integrated infrared imaging system (AIRIS) and experimental pseudo-leks to quantify sightability (proportion of birds detected) of conventional ground-based visual (GBV) surveys for greater sage-grouse Centrocercus urophasianus . Specifically, we calibrated AIRIS at pseudo-leks composed of known numbers of captively-raised birds, primarily ring-necked pheasant Phasianus colchicus . We then carried out AIRIS and GBV surveys, simultaneously, on nearby sage-grouse leks, allowing us to model AIRIS and GBV sightability. AIRIS detected ∼93% of birds on pseudo-leks while GBV detected ∼86% of sage-grouse on leks. Thus, the ground count observation error was –14% from the ‘true' number of male sage-grouse attending the leks. We also found sagebrush cover decreased sightability for GBV counts but did not influence sightability by AIRIS. Because standard GBV protocols typically make repeated counts of sage-grouse in a single morning, we also modeled repeated GBV counts using N -mixture models and found an 88% sightability, which was nearly the same as GBV sightability from the AIRIS analysis. This suggests that the use of repeated morning counts can potentially account for imperfect detection in the standard GBV surveys currently implemented. We also provide generalized correction values that could be employed by resource managers using either GBV or AIRIS to better estimate ‘true’ numbers of sage-grouse attending leks within similar environments to this study. The findings and interpretation presented can help guide effective monitoring protocols that account for observation error and improve accuracy of data used for population trend and abundance estimation.