• Leaf warblers;
  • developmental constraints;
  • reaction-diffusion models;
  • pattern formation;
  • melanin patterns;
  • phylogeny


Patterns of melanin pigmentation in birds are extremely varied. Nevertheless it is easy to think of many patterns that are never observed, and others that frequently recur in diverse and distantly related species. Using as our model the avian genus Phylloscopus we ask how the restricted range of observed patterns might be attributable to a restricted range of variants produced by developmental perturbations. The patterns we consider consist of unmelanized patches on the wings, crown and rump on otherwise pigmented upperparts. We use reaction-diffusion models to show that gross features of the pattern can be simply predicted from considerations of embryo shape. We suggest that birds are expected to have more patterned heads, because the head region is relatively larger than other regions in the developing embryo. A comparative analysis across many species of birds and a phylogenetic analysis within the genus Phylloscopus show that the component elements of the pattern have repeatedly been lost and gained during evolution. A shift in a threshold reading could explain the appearance and disappearance of the unmelanized patches, perhaps through changes in the sensitivity of melanocytes to epidermal signals. Such threshold shifts would make the transition between patterned and unpatterned forms particularly easy once the patterns have been exposed to selection in some distant ancestor. This partitioning of the roles of selection and development implies that many features of the patterns reflect developmental mechanisms in both immediate and more distant ancestors.