An aerial grid-based sampling method to estimate nonbreeding duck abundance

Probability-based sampling designs for aerial surveys are useful for estimating wintering waterfowl abundances in large areas with contiguous habitat (e.g., Mississippi Alluvial Valley). The effectiveness of these approaches for estimating abundance of nonbreeding waterfowl in small areas with discontinuous habitat has rarely been assessed. Surveys conducted within riverine areas introduce sampling design difficulties because of discontinuous bottomland habitat and irregularity of flooding events. Therefore, we implemented and evaluated a generalizable grid-based, stratified random sampling design to estimate weekly duck abundance within the Wabash River floodplain in southeastern Illinois and southwestern Indiana, USA, during midwinter and early spring 2012 and 2013. We used aerial and ground-based counts to generate abundance estimates and evaluate within-week variation of counts. Peak duck abundance totaled 279,717 (SE = 69,101) in 2012 and 742,027 (SE = 296,563) in 2013. Dabbling ducks were the most abundant duck guild detected and abundance estimates for this group met our precision goal (CV ≤ 0.25) in 4 of 18 surveys = 0.33, range = 0.20–0.51). We compared the grid-based approach with a traditional aerial inventory and found that estimates from the grid-based approach were on average 75,175 (SE = 46,768) greater than inventory counts. Desired levels of precision were often not met, yet design improvements in 2013 notably increased precision, suggesting that further modifications may allow this method to be useful in riverine areas with discontinuous and ephemeral habitat.