An experiment was conducted during April and June 1978 to estimate aerial-visibility bias in searches for bald eagle (Haliaeetus leucocephalus) nests in northwestern Ontario, Canada. Randomly selected quadrats were searched independently by 2 separate teams of pilots and observers. Using either joint binomial probabilities or Petersen Index calculations, we estimated that approximately 76% of the nests or 85% of the breeding areas were found on a single search of shoreline. There were no differences (P > 0.05) in visibility due to season or observer, although undetected differences among observers may exist (i.e., type-II error). Aerial surveys of this type can be hazardous; 1 aircraft crashed during the project. Recommendations for standardized searches, and suggestions to improve safety are given. J. WILDL. MANAGE. 45(1):83-92 In spite of the large, bulky appearance of most bald eagle nests and the large size of the birds, they may be overlooked easily during surveys. The nests are usually below the crowns of the trees and dispersed over vast expanses of forested shoreline. Birds or nests that are present but not seen are difficult to count. They create a visibility bias in population estimates, a common, if not universal, problem in aerial surveys (Caughley 1974). Attempts to estimate the bias include regression of visibility against several survey factors for African mammals (Caughley 1974, and others cited by him), the use of visibility indices for waterfowl (Diem and Lu 1960) and ospreys (Pandion haliaetus) (Henny et al. 1977), an extensive experiment performed with moose (Alces alces) (LeResche and Rausch 1974), and the use of group-size information for white-tailed deer (Odocoil us virginianus) and moose (Cook and Jacobson 1979). Earlier aerial surv ys of bald eagles in the Canadian shield region indicated that a real visibility bias exists for both eagles and their nests (Whitfield et al. 1974, Grier 1977). The only estimate of the size of this bias (29% of all breeding areas and 19% of active breeding areas not seen) (Leighton et al. 1979) was based on a small sample of known nests counted from the ground. Because factors such as type and speed of aircraft, pilot, observer, season, and habitat type might influence such visibility bias, we conducted an experiment using 2 independent searches of randomly selected quadrats to estimate the bias during spring and summer 1978 in northwestern Ontario. We thank J. C. Williamson, Fish and Wildlife Coordinator for the West Patricia Land Use Plan, for his advice, encouragement, and administrative assistance. T. Hanaway and R. Rienhold, owners of Lac Seul Airways and Northern Waters Air Service, respectively, provided the aircraft. We are especially indebted to R. Buckler, of Loon Air, for leasing an additional airplane to North'This study was conducted as a part of the Ontario Ministry of Natural Resources West Patricia Land Use Plan. 2 Present address: Wildlife Service, Government of the Northwest Territories, Yellowknife, N.W.T., Canada. J. Wildl. Manage. 45(1):1981 83 This content downloaded from 157.55.39.158 on Fri, 18 Nov 2016 04:13:46 UTC All use subject to http://about.jstor.org/terms 84 BIAS IN SURVEYS OF NESTING EAGLES * Grier et al.