Accuracy and Precision of Aerial Moose Censusing

Abstract

Forty-nine observers flew 72 replicate counts over four fenced 1-square-mile (2.6 km'2) moose (Alces alces) enclosures. Under the most ideal conditions, experienced observers flying 15 minutes over each square mile in PA-18-150 aircraft saw only 68 percent of the moose present. Inexperienced observers saw 43 percent. Accuracy of counts was significantly affected by observer experience and currency, by number of observers, by snow conditions, by habitat and terrain, and by time of day. Moose density did not affect accuracy of counts at densities of seven or more per square mile. Counts were relatively precise within individual experienced observers (C.V. = 7-18 percent) and within some experience-current groups (C.V. = 8.5-9.9 percent). Experienced observers without current experience were both imprecise (C.V. = 24 percent) and inaccurate (46 percent). Aerial counts are not considered valid estimates of absolute moose numbers, but may be used as trend indicators in carefully controlled situations. J. WILDL. MANAGE. 38(2):175-182 Aerial censusing is presently the only practical way to estimate moose numbers in most of North America (Rausch and Bratlie 1965, Rausch and Bishop 1968, Bishop 1969, Bergerud and Manuel 1969). Determination of the extent to which such methods underestimate numbers has remained a problem when absolute numbers are sought. Benson (1966), Goddard (1966), and Bergerud (1968) have reviewed aerial censusing techniques and have suggested that substantial errors are inherent in most. Caughley and Goddard (1972) have suggested statistical methods for estimating these errors. Siniff and Skoog's (1964) random stratified, quadrat sampling method, modified for moose by Evans et al. (1966), made population estimations more plausible, but these techniques were still predicated on knowing what proportion of animals flown over were actually observed. More recently Mantle (1972) in Ontario and Jordan on Isle Royale (Personal communication) have xperimented with aggregate-related counting methods and new flight-path techniques. Many people participated and assisted in this study. We thank all counters and pilots involved. Special thanks go to A. Anderson of Andy's Flying Service, Inc. (Kenai) for his skillful piloting and other assistance, to R. A. Richey of the Kenai National Moose Range, and to J. L. Davis and L. Nichols, Alaska Department of Fish and Game. J. L. Oldemeyer, A. W. Franzmann, D. E. McKnight, and R. H. Bishop read the manuscript and offered helpful suggestions. J. L. Oldemeyer and S. J. Harbo offered statistical assistance during study design and analysis.