Analyses of Selected Factors Affecting the Accuracy of Aerial Surveys of Mottled Ducks.

Two experiments were conducted to test the effects of observer ability, ambient illumination, glare, sun angle, time of day, bird reaction, wetland type, vegetative cover, and non-flushing birds on results of aerial surveys of mottled ducks (Anas fulvigula). In Experiment I, aerial surveys for mottled ducks were conducted in Louisiana from January 1985 through April 1986 to determine the effects of observer, weather, glare, time of day, and bird reaction on results. An experienced, but non-current, waterfowl observer initially detected a greater number of birds than an experienced and current observer, observers did not differ in 17 subsequent surveys. Time of day, ambient illumination, and glare did not significantly affect survey results. Variation of observers\u27 estimates was greater during midday and late afternoon survey periods and after recruitment of the annual cohort. Variation of estimates between clear and cloudy weather exhibited no discernible pattern. Standard deviations of the observers\u27 estimates were greater on glare-free transects in 1 of 2 experiments and during 5 of the 7 months surveyed in both experiments. Mean sighting probability of mottled ducks in the detected population was 0.50 for two paired observers. Numbers of flushing and non-flushing mottled duck responses to the survey aircraft did not differ nor was the number of responses different during clear or cloudy weather. Variation in the number of observations of flushing mottled ducks was 19% greater on clear days. In Experiment II, decoys were used to assess effects of wetland type, vegetative density, ambient light, glare, sun angle, and observer on aerial surveys of non-flushing mottled ducks. Percentages of decoys detected did not differ between experienced observers, emergent wetland types, ambient illumination conditions, or glare and sun angle conditions. More decoys were detected in late winter-early spring surveys when vegetative cover was sparse. Variation of observers\u27 estimates was greater on glare-affected transects. Paired observers detected only 20% of 2,519 decoys available along transect lines. Amount of observer experience did not affect survey results but more experienced observers detected decoys more precisely