Evolution and Historical Biogeography of a Song Sparrow Ring in Western North America

Abstract

The Song Sparrow, Melospiza melodia (Aves: Emberizidae), exhibits a greater degree of geographic variation than does any other North American bird species. Detailed morphological work has demonstrated that a subset of the 25 diagnosable subspecies forms a classic ring species in the western United States. The ring’s center is the Sierra Nevada and Mojave Desert in California and adjacent Nevada, and its connecting point is in southeastern California, where an olive and black subspecies of the coastal slope interbreeds sporadically with a gray and rufous subspecies of the arid interior. However, song differences associated with habitat segregation lead to assortative mating between the two subspecies that meet in the Coachella Valley at the southern base of San Gorgonio Pass. Moving clockwise around the ring from the connecting point one finds a gradation of subspecies that become paler, rustier, and grayer. Standard models of ring species evolution imply the connecting point is the region occupied most recently, in this case after sparrows would have spread southward down either side of the mountains and desert. This scenario is plausible given molecular evidence of a glacial refugium on the Queen Charlotte Islands, British Columbia, suggesting that ancestral birds could have moved south in this pattern. By contrast, another postulated refugium is what is now the arid desert of southeastern California or northeastern Baja California, Mexico. This refugium’s location – coupled with a recent meta-analysis of North American hybrid zones that identifies the San Gorgonio Pass region as an ancestral contact zone of coastal and desert fauna – implies that the connecting point is the region occupied earliest, an alternative that would mean the Song Sparrow ring differs fundamentally from one that would have evolved via the standard model. Biogeographical and morphological data support the latter, more radical interpretation, but genetic, vocal, ecological, and behavioral data are needed around the ring to determine conclusively which model is best supported.