Measuring and predicting abundance and dynamics of habitat for piping plovers on a large reservoir

•Remotely-sensed breeding habitat for a federally-listed species.•Habitat varied markedly among years (1986–2009) ranging from 9 to 5195ha.•Predicted habitat (ha) with water levels, modeled revegetation rate, and topography.•Predictive model could inform conservation through what-if scenarios.•Approach adaptable to other systems to improve habitat and population estimation. Measuring habitat and understanding habitat dynamics have become increasingly important for wildlife conservation. Using remotely-sensed data, we developed procedures to measure breeding habitat abundance for the federally listed piping plover (Charadrius melodus) at Lake Sakakawea, North Dakota, USA. We also developed a model to predict habitat abundance based on past and projected water levels, vegetation colonization rates, and topography. Previous studies define plover habitat as flat areas (30% bare-substrate obstruction) were 76% correct and omission and commission errors were equal. Due to water level fluctuations, habitat abundance varied markedly among years (1986–2009) ranging from 9 to 5195ha. The proportion bare substrate declined with the number of years since a contour was inundated until 5 years (βˆ=−0.65, SE=0.05), then it stabilized near zero, and the decline varied by shoreline segment (5, 50, and 95 percentile were βˆ=−0.19, SE=0.05, βˆ=−0.63, SE=0.05, and βˆ=−0.91, SE=0.05, respectively). Years since inundated predicted habitat abundance well at shoreline segments (R2=0.77), but it predicted better for the whole lake (R2=0.86). The vastness and dynamics of plover habitat on Lake Sakakawea suggest that this is a key area for conservation of this species. Model-based habitat predictions can benefit resource conservation because they can (1) form the basis for a sampling stratification, (2) help allocate monitoring efforts among areas, and (3) help inform management through simulations or what-if scenarios.