Detection Probability and Disturbance of Waterfowl During Aerial Surveys

Aerial surveys of waterfowl have been conducted in the Illinois and Mississippi River floodplains since 1948. These traditional surveys provide an index of waterfowl population size and are used to track migration events, set harvest regulations, and for research purposes. New methods are being evaluated to estimate population size by randomizing survey locations and estimating detection probabilities. We used double sampling to determine a correction factor for waterfowl estimates during fall aerial surveys. Immediately before an aerial survey, a ground observer conducted an intensive survey of waterfowl in predetermined locations from an elevated, unobstructed location where probability of detection was assumed to be 100%. Aerial counts were divided by ground counts for all common species and foraging guilds to determine detection probability. Preliminary results indicate that mean detection rate for all waterfowl was 100.2% (SE = 17.9). Mean detection rate was 115.1% (SE = 21.2) for ducks, 115.2% (SE = 21.5) for dabbling ducks, 95.6% (SE=36.9) for diving ducks, 50.7% (SE = 24.9) for mergansers, and 93.0% (SE = 15.4) for geese. While conducting ground surveys, observers also documented any disturbance to waterfowl caused by aerial surveys. Our preliminary findings indicated that on average 13.1% (SE = 4.1) of waterfowl, 7.5% (SE = 2.9) of ducks, 9.7% (SE = 3.9) of dabbling ducks, 3.7% (SE = 2.2) of diving ducks, 4.5% (SE = 4.5) of mergansers, and 11.1% (SE = 5.8) of geese exhibited a negative response (i.e., flew short distances, swam away, changed behavior significantly) to aerial surveys. Our preliminary findings indicated that on average 5.6% (SE = 3.2) of waterfowl, 2.7% (SE = 1.6) of ducks, 1.2% (SE = 0.8) of dabbling ducks, 2.8% (SE = 2.0) of diving ducks, 4.5% (SE = 4.5) of mergansers, and 4.8% (SE = 4.6) of geese abandoned the survey site and did not return following aerial surveys. With our findings, traditional aerial surveys conducted in the Mississippi and Illinois River floodplains can be used to minimize bias in population estimates associated with aerial survey techniques. In our presentation, we will model detection probability as a function of habitat type, month, taxa, and other factors and present data from fall 2014–2015.