Challenges and opportunities for data integration to improve estimation of migratory connectivity

Understanding migratory connectivity, or the linkage of populations between seasons, is critical for effective conservation and management of migratory wildlife. A growing number of tools are available for understanding where migratory individuals and populations occur throughout the annual cycle. Integration of the diverse measures of migratory movements can help elucidate migratory connectivity patterns with methodology that accounts for differences in sampling design, directionality, effort, precision and bias inherent to each data type.

The R package MigConnectivity was developed to estimate population‐specific connectivity and the range‐wide strength of those connections. New functions allow users to integrate intrinsic markers, tracking and long‐distance reencounter data, collected from the same or different individuals, to estimate population‐specific transition probabilities (estTransition) and the range‐wide strength of those transition probabilities (estStrength). We used simulation and real‐world case studies to explore the challenges and limitations of data integration based on data from three migratory bird species, Painted Bunting (Passerina ciris), Yellow Warbler (Setophaga petechia) and Bald Eagle (Haliaeetus leucocephalus), two of which had bidirectional data.

We found data integration is useful for quantifying migratory connectivity, as single data sources are less likely to be available across the species range. Furthermore, accurate strength estimates can be obtained from either breeding‐to‐nonbreeding or nonbreeding‐to‐breeding data. For bidirectional data, integration can lead to more accurate estimates when data are available from all regions in at least one season.

The ability to conduct combined analyses that account for the unique limitations and biases of each data type is a promising possibility for overcoming the challenge of range‐wide coverage that has been hard to achieve using single data types. The best‐case scenario for data integration is to have data from all regions, especially if the question is range‐wide or data are bidirectional. Multiple data types on animal movements are becoming increasingly available and integration of these growing datasets will lead to a better understanding of the full annual cycle of migratory animals.