• Coevolutionary arms race;
  • divergent selection;
  • geographic mosaic of coevolution;
  • Loxia megaplaga;
  • Pinus occidentalis;
  • predator–prey;
  • speciation;
  • species interactions

Crossbills (Aves: Loxia) and several conifers have coevolved in predator–prey arms races over the last 10,000 years. However, the extent to which coevolutionary arms races have contributed to the adaptive radiation of crossbills or to any other adaptive radiation is largely unknown. Here we extend our previous studies of geographically structured coevolution by considering a crossbill–conifer interaction that has persisted for a much longer time period and involves a conifer with more variable annual seed production. We examined geographic variation in the cone and seed traits of two sister species of pines, Pinus occidentalis and P. cubensis, on the islands of Hispaniola and Cuba, respectively. We also compared the Hispaniolan crossbill (Loxia megaplaga) to its sister taxa the North American white-winged crossbill (Loxia leucoptera leucoptera). The Hispaniolan crossbill is endemic to Hispaniola whereas Cuba lacks crossbills. In addition and in contrast to previous studies, the variation in selection experienced by these pines due to crossbills is not confounded by the occurrence of selection by tree squirrels (Tamiasciurus and Sciurus). As predicted if P. occidentalis has evolved defenses in response to selection exerted by crossbills, cones of P. occidentalis have scales that are 53% thicker than those of P. cubensis. Cones of P. occidentalis, but not P. cubensis, also have well-developed spines, a known defense against vertebrate seed predators. Consistent with patterns of divergence seen in crossbills coevolving locally with other conifers, the Hispaniolan crossbill has evolved a bill that is 25% deeper than the white-winged crossbill. Together with phylogenetic analyses, our results suggest that predator–prey coevolution between Hispaniolan crossbills and P. occidentalis over approximately 600,000 years has caused substantial morphological evolution in both the crossbill and pine. This also indicates that cone crop fluctuations do not prevent crossbills and conifers from coevolving. Furthermore, because the traits at the phenotypic interface of the interaction apparently remain the same over at least several hundred thousand years, divergence as a result of coevolution is greater at lower latitude where crossbill–conifer interactions have been less interrupted by Pleistocene events.