Skeletons and the genera of Sparrows (Emberizinae)

ABSTRACT.-Based on comparisons of skeletons from 228 species of sparrows (Emberizinae), we detected 24 characters that were useful for generic separations. Of these, 19 quantitative characters were used in a detrended correspondence analysis (DCA) ordination. Presence or absence of a lacrimal bone, degree of inflation of the squamosal area, length of the lateral process of the laterosphenoid, ratio of ulna length to femur length, length of the skull, and premaxilla length-to-width ratio were the most informative characters. The first two axes of the DCA clearly grouped most genera. Based on our analysis of these skeletal characters, we recommend the following changes to the classification of Paynter (1970): Sporophila obscura belongs in Tiaris; Emberizoides ypiranganus is a valid species; Pselliophorus and Pezopetes should be lumped with Atlapetes; and Torreornis and Oriturus should be lumped with Aimophila. In addition, a different sequence of genera should be adopted. Two groups of genera are distinct: seedeaters and North American sparrows. A third group, South American grassland finches, overlaps with the first two groups. Received 17 February 1998, accepted 15 March 1999.

THE SUBFAMILY EMBERIZINAE constitutes a large (279 species in 65 genera) group of songbirds adapted for eating seeds during at least part of the year. Most species occur only in the Western Hemisphere, but two are Holarctic, and four genera are confined to the Eastern Hemisphere. The last comprehensive classification of the Emberizidae was by Sharpe (1888), although Ridgway (1901) characterized the North and Middle American forms and some of the South American genera. Hellmayr (1938) listed all of the Western Hemisphere forms, noting the characters of some genera. Vaurie (1959) and Cramp and Perrins (1994) reviewed the Eurasian species, and Meyer de Schauensee (1970a) and Ridgely and Tudor (1989) listed the South American species and provided characterizations. Hall and Moreau (1970) reviewed the African species. Paynter (1970) listed all of the forms and their geographic ranges in the most recent complete classification based on specimens, but he did not include characterizations; we follow his classification in this paper except where noted otherwise. Sibley and Monroe (1990) listed the genera and species with their ranges; they did not include characterizations, and their classification was based on DNA-DNA hybridization studies (Sibley and Ahlquist 1990) of 22 (34%) genera and 24 (9%) species. Sibley and Monroe's most striking innovation was transferring 52 of the 65 genera (a group they called the "tanager finches") to the Thraupini (=Thraupinae of most authors) based on DNA evidence from only 9 of the 52 genera involved.

Comparative studies of the skeletal elements of emberizines were conducted by Parker (1878), Shufeldt (1888), Sushkin (1924), Linsdale (1928), Beecher (1953), Tordoff (1954), Berger (1957), Bock (1960, 1962), Bowman (1961), George (1962, 1968), Robins and Schnell (1971), Wolf (1977), Webster and Goff (1979), Steadman (1982), Zink (1982), Moreno (1984), Rising (1988), Webster (1993), and Patten and Fugate (1998). Of these studies, Tordoff's is based on the largest number of species, Bowman's (1961) study of the 14 Galapagos species is the most intensive skeletal study but does not include other species.

Overall, classification of the emberizine genera and species has been rather thoroughly studied for North American and European taxa. Knowledge of the systematics of Asian, South American, and some of the Middle American forms, however, is modest and poorly coordinated with that of the better-known groups. Here, we provide additional information on the skeletons of nearly all species of emberizines and suggest changes in current classification of emberizines where skeletal evidence is clear. For convenience and clarity, we have followed Paynter's classification (1970) throughout most of the paper.

MATERIALS AND METHODS

Our strategy was to be extensive rather than intensive. We tried to examine skeletons of all species of Emberizinae and their close relatives and were able to study 794 specimens of 228 species in 62 genera (Table 1). Thirty-nine of the 50 species and all three of the genera (i.e. Latoucheornis, Oreothraupis, Charitospiza) that we did not examine apparently do not exist in the world's museums (Wood and Schnell 1986). The species recognized by Paynter (1970) that we did not examine are listed in the Appendix 1. On each specimen we made 15 measurements (two of angles and 13 of distances) and 27 other observations. Thirty-nine characters were tabulated, 12 of these as ratios of two measurements. A few specimens were broken or incomplete, so all measurements could not be performed on some specimens.

We also examined the following skeletons that do not belong to Paynter's (1970) Emberizinae: Parulidae, 26 genera and 109 species; Thraupinae, 57 genera and 195 species; Tersininae, 1 species; Catamblyrhynchinae, 1 species; Cardinalinae, 8 genera and 22 species; Icteridae, 22 genera and 79 species; Fringillidae, 15 genera and 24 species.