Design considerations for breeding waterfowl surveys

Abstract Each year, surveys are conducted across North America to provide crucial information that guides sustainable waterfowl population management. The design of surveys requires careful evaluation to produce accurate and precise waterfowl population estimates. We assessed the impact of different survey design considerations on bias and precision of breeding waterfowl population estimates in the highly‐modified landscape of Iowa's Prairie Pothole Region (PPR). We used breeding pair counts of wood duck ( Aix sponsa ), blue‐winged teal ( Spatula discors ), and mallard ( Anas platyrhynchos ) collected during aerial surveys in the Iowa PPR from 2016 to 2018 to evaluate the influence of survey replication (1, 2, and 4 visits within a year) on precision of breeding pair estimates. Next, we simulated breeding abundance estimates for each species for a single year using 100 sample draws from 3 wetland sampling strategies and compared simulated estimates from the different sampling strategies to predicted abundance estimates. Breeding pair estimates for all species were less biased and more precise using data from 4 surveys ( = 0.06) than estimates using data from a single survey ( = 0.09) and 2 surveys ( = 0.07). For all species, breeding abundance estimates from the Neyman (i.e., optimal) sampling strategy were similar to simulated abundances but were less precise ( = 0.53) than estimates from both the equal (i.e., stratified uniform; = 0.31) and proportional (i.e., probability proportional to size; = 0.27) sampling strategies. We recommend that future waterfowl surveys include at least 2 visits to wetlands per season and employ the Neyman wetland sampling strategy to minimize bias in breeding pair abundance estimates.