Employing Digital Still Imaging And Observer Counts To Estimate Bias In Aerial Surveys Of Wintering Sea Ducks

To better understand the bias in aerial survey abundance estimation that results from aircraft and observer effects, we conducted a study to estimate availability bias, detection, and misidentification by collecting images and observer counts of wintering sea ducks. We employed a fixed-wing aircraft with two digital still-image cameras and two observers to survey multiple 50m-wide strip-transects. A forward-facing (FF) camera attached to the wing strut captured images of the transect 250-300m ahead of the aircraft, while a point-of-view (POV) camera mounted to the rear window captured images of the transect abeam of the aircraft. Comparing the FF to POV camera counts allowed estimation of availability bias, or aircraft effect (the proportion of birds that flew out of the transect or dove due to the aircraft). Analyses suggested that 5-30% of sea ducks dove or flew off-transect as the aircraft approached: 5% for goldeneye, 20% for long-tailed duck and surf scoter, 25% for white-winged scoter, and 30% for bufflehead. Comparing the POV camera to observer counts allowed estimation of detection (the percentage of birds within the transect that were detected by observers) and misidentification. Estimates of birds detected on transects ranged from 50% to 95%, varying by species and observer, with long-tailed duck detection being slightly higher than other species. Misidentification of surf and white-winged scoters was about 1% and 4-6%, respectively. Our results suggest that, due to a combination of aircraft and observer effects, current survey estimates could be increased by a factor of 1.3 to 2.2 depending on species. We are developing methods for additional studies to address these biases, as well as effects of transect width and aircraft type