Using time series data to assess recent population dynamics of Bald Eagles in the northeast United States

Abstract

Abstract We explored the use of time series data to study the recent demographics of Bald Eagles in the northeast United States. We merged a stage‐structured population matrix model (PMM) into a combinatorial optimization algorithm (COA) to estimate stage‐wise abundances, survival rates, and other dynamics of the eagle population from 1990 to 2018. The PMM includes biannual periods (breeding and non‐breeding), density dependence, and stage‐wise migration and dispersal. We demonstrated theoretically that the information contained in 17 yr of time series data of the reproductive stage would have contained all the information for the entire population. We then tested the performance of the COA in estimating transient and asymptotic population dynamics using simulated data. Finally, we assessed the COA under differing modeling conditions (without stage‐wise dispersal and in the absence of biological assumptions). We found that time series data of the reproductive class, the skeleton of the PMM, and COA are sole ingredients to reliably estimate stage‐wise abundances and survival, but additional biological assumptions are helpful in honing precision of the estimates. The algorithm is a tool to assess population dynamics when demographic data are limited to counts of reproductive adults.

Keywords

• algorithm
• Bald Eagle
• conservation
• data science
• difference equation
• machine learning