J. Michael Reed

Quantifying phenological landmarks of migration shows nonuniform use of the Caribbean by shorebirds


Of   the   boreal-  and   Arctic-    breeding North     American shorebirds that   migrate south through the   Caribbean, most    individuals continue farther south.     However, for   many species, some    individuals remain beyond the   southbound migration period (i.e., throughout the   temperate winter and/or summer). This   variation among individuals adds   complexity to  observation data,    obscures migration patterns, and   could    preventthe   examination of  the   use   of  different Caribbean regions by  various shorebird species during     migration and   in  the   nonmigratory seasons. Here,    we   present a  novel    method that   leverages a  well-   established statistical approach (generalized additive models) to systematically identify migration phenology even    for   complex passage migrant spe -cies   with    individuals that   remain beyond migration. Our   method identifies the   active migration period    using    derivatives of  a  fitted    GAM   and   then   calculates phenology metrics   based    on  quantiles of  that   migration period. We   also   developed indices to  quantify oversummering and   overwintering patterns with    respect to  migration. We   analyzed eBird    data    for   16  North    American shorebird species as  they    traveled South    through the   insular Caribbean, identifying separate migratory patterns for   Cuba,    Puerto Rico, Guadeloupe, Aruba, Bonaire, Curaçao, and   Trinidad and   Tobago. Our   results confirm past   reports and   provide additional detail    on   shorebird migration in  the   Caribbean, and   identify several previously unpublished regional patterns. Despite Puerto Rico being    farther north    and   closer    to  continental North    America, most  species reached Puerto Rico   later    than    other    regions, supporting a  long-   standing hypothesis that   mi -gration strategy (transcontinental vs.   transoceanic) leads    to  geographic differences in  migration timing. We   also   found  distinct patterns of  migration curves, with    some regions and   species consistently having either    symmetrical or  skewed curves; these differences in  migration curve    shape    reflect     different migration processes. Our   novel  method proved reliable and   adaptable for   most    species and   serves     as  a  valuable tool for   identifying phenological patterns in  complex migration data,    potentially unlocking previously intractable data.

Data type
Scientific article
Research and monitoring
Geographic location