Date of this Version
Caven AJ, Varner DM, and Drahota J. (2020) Sandhill Crane abundance in Nebraska during spring migration: making sense of multiple data points. Transactions of the Nebraska Academy of Sciences and Affiliated Societies 40 (2020), pp 6-18.
The USFWS conducts an annual one-day aerial survey of the North and Central Platte River Valleys, generally on the fourth Tuesday in March, to estimate the abundance of the midcontinent Sandhill Crane population. However, these abundance indices demonstrate unrealistic inter-annual variation as a result of deviations in migration chronology and other factors. Additional research efforts have been undertaken within the region to estimate Sandhill Crane abundance over time and space but these projects generally seek to answer unique questions, employ differing survey methods, and cover overlapping yet distinct survey areas. Despite the wealth of information there remains significant uncertainty regarding the actual abundance of Sandhill Cranes in Nebraska during the peak of migration. We conducted a model-based metadata analysis relying on the distinctive strengths of three databases to assess USFWS data, identify annual abundance estimates that may not be robust, and developed parameter-based and factorbased corrections to USFWS indices from 2000 to 2019. Our analyses suggest that at the peak of spring migration there is likely between 1.1 and 1.4 million Sandhill Cranes in the North and Central Platte River Valleys of Nebraska. Our best performing models indicated the most likely peak estimate was 1.27 million Sandhill Cranes with approximately 220,000 in the North Platte River Valley and 1,050,000 in the Central Platte River Valley in both 2018 and 2019. Our assessment suggests that 25% of USFWS aerial estimates are robust, with the rest representing underestimates as both exogenous and endogenous factors such as migration chronology and survey methodology serve to bias indices downward. Given this downward bias, the three-year running average used by the USFWS actually provides a robust estimate for only 5–15% of the years analyzed. By contrast, we found that a five-year rolling maximum provides a robust estimate for 70–75% of the years analyzed.