Phylogenetic Patterns in the Trochilidae

Abstract

-Although many aspects of hummingbird biology have been studied, few recent analyses of higher-level systematic relationships exist. Based on morphology, it has been hypothesized that the Trochilidae includes six major clades. We used starch-gel electrophoresis to construct and test phylogenetic hypotheses for representatives of the six clades, using two species of swifts (Apodidae) as outgroups. Of 45 loci scored, 38 were polymorphic. The average Nei's genetic distance (D) among 14 hummingbird taxa was 0.625; D averaged 1.61 between the swifts and hummingbirds. These distances are large, and are consistent with other nonpasserine groups, suggesting that hummingbird taxa are relatively old. Phylogenetic analyses generally were consistent with the hypothesis that hermits are a sister group to all other trochilines. Within the Trochilinae, two broad groups are recognized, here called trochiline-A and B, which correspond to the morphologically determined and trochiline groups of Zusi and Bentz (1982). Androdon aequatorialis is genetically distinct but generally aligns with the trochiline-A group. Within the trochiline-B group, four radiations hypothesized by Zusi (pers. comm.), here called Bee, Amazilia (Emeralds), Andean, and High Andean, were corroborated by our analyses. Our distance analysis suggests a phylogenetic pattern consistent with that derived from Sibley and Ahlquists' (1990) and Bleiweiss et al.'s (1997) DNA-DNA hybridization studies. Received 31 October 1996, accepted 20 June 1997. THE HUMMINGBIRDS (TROCHILIDAE) form one of the largest bird families, with approximately 325 species. Although many aspects of hummingbird biology have been studied, few higher-level studies of systematic relationships exist. Early taxonomists (Gould 1861, Boucard 1895, Simon 1921, Peters 1945, Zimmer 19501953) provided species descriptions and general details of geographic variation in most species, but they did not explicitly identify systematic relationships. Recently, several approaches have been used to identify phylogenetic patterns within parts of this family: comparative analysis of vocalizations and mating behavior (Schuchmann unpubl. data), external morphology and ecology (Graves 1980, 1986; Stiles 1983, 1996; Hinklemann 1989), comparative myology (Zusi and Bentz 1982), protein electrophoresis (Gerwin and Zink 1989, Gill and Gerwin 1989 ), mitochondrial DNA sequences (Hernandez-Banos et al. unpubl. data), and 2 Present address: North Carolina State Museum of Natural Sciences, P.O. Box 29555, Raleigh, North Carolina 27626, USA. E-mail: ncs1416@interpath.com 3Address correspondence to this author. Present address: James Ford Bell Museum, University of Minnesota, 100 Ecology Building, St. Paul, Minnesota 55108, USA. E-mail: rzink@biosci.umn.edu DNA-DNA hybridization (Sibley and Ahlquist 1990; Bleiweiss et al. 1994, 1997). Most of these studies were not comprehensive at the level of the family. However, the study by Bleiweiss et al. (1997), which involved 26 species, provided a molecular phylogenetic hypothesis for the major clades in the family. We report an allozymic survey of major groups of hummingbirds designed to test phylogenetic hypotheses derived from previous morphological and DNA-DNA hybridization analyses. Our study was based in part on Zusi and Bentz's (1982) hypothesis of higher-level groups in the Trochilidae. They identified four major groupings: hermits (Phaethorninae), trochilines, and two groups of trochilines. The terms primitive and advanced are reserved for discussion of characters and are misleading without an in-depth phylogenetic analysis; we refer to the two groups as trochiline-A (Zusi and Bentz's [1982] primitive group) and trochiline-B (their advanced group). The widespread notion that the hermits represent the hummingbirds is incorrect in one sense: if these are sister clades, each is by definition the same age. Phylogenetic analysis is required to show that the hermit clade retains a disproportionate number