Estimating true density in large, alpine herbivores using Google Earth imagery

Density is a fundamental ecological parameter that is often difficult to estimate in large herbivores due to the wide spatial extent of their populations and imperfect detectability of individuals. We searched Google Earth imagery using an independent double-observer protocol to estimate elk Cervus elaphus density in Yellowstone National Park, WY, USA, on summer range that had never been systematically surveyed. Detectability of elk was high at 0.836 (95% CRI: 0.769, 0.909) producing high precision in our density estimates. Elk density estimates ranged from a minimum of 8.9 elk km(-2) over a 214.8 km(2) search area (95% CRI: 7.9, 10.9) to a maximum of 14.0 elk km(-2) over 98.5 km(2) of nonforested, alpine terrain (95% CRI: 13.5, 14.9). These estimates of elk density are comparable to the highest elk densities ever recorded in Yellowstone which primarily derive from winter, at low elevations, prior to wolf Canis lupus reintroduction. A log-linear positive effect of open landcover on density suggested preference for open terrain. Multiple survey designs may be possible with satellite imagery that would allow further refinement of noninvasive abundance estimation over large scales.