Phylogeny of early tertiary swifts and hummingbirds (Aves: Apodiformes)

Auk, The, Jan 2003 by Mayr, Gerald

The Auk 120(1):145-151, 2003

ABSTRACT.-The phylogenetic relationships between early Tertiary and extant apodiform birds are only poorly understood, and this study is the first cladistic approach to this problem in which the Trochilidae are included. The analysis supports monophyly of the Lower Oligocene Jungornis and extant Trochilidae, as well as monophyly of the Middle Eocene Scaniacypselus and extant Apodidae. The "Jungornithidae" sensu Karhu (1999) are shown to be paraphyletic with the Upper Eocene Argornis being the sister taxon of the taxon (Jungornis + extant Trochilidae). The osteology of Jungornis provides a transition between that of the highly derived extant Trochilidae and that of more generalized apodiform birds. An Argornis-like apodiform bird from the Middle Eocene of Messel shows a completely unexpected combination of a greatly abbreviated, apodiform humerus with a short and broad wing, and might indicate that the Trochilidae evolved from a short-winged ancestor. Received 8 April 2002, accepted 26 October 2002.

THERE IS INCREASING consensus that tree swifts (Hemiprocnidae), true swifts (Apodidae), and hummingbirds (Trochilidae) form a monophyletic clade that is supported by derived anatomical features and most biochemical and molecular analyses (see Johansson et al. 2001, Livezey and Zusi 2001, Mayr 2002). As shown by a recent phylogenetic analysis (Mayr 2002), owlet-nightjars (Aegothelidae) are the sister group of swifts and hummingbirds. Although Apodiformes have a comparatively extensive early Tertiary fossil record, phylogenetic relationships between the fossil and the extant taxa are only insufficiently understood.

One of the earliest swift-like birds described so far is Eocypselus vincenti Harrison 1984 from the lower Eocene of England. That species, which is known from a few isolated bones of only a single individual, was classified into a monotypic taxon, Eocypselidae, by Harrison (1984) but included in the Hemiprocnidae by Mourer-Chauvire (1988).

Much better represented by numerous isolated bones are the early Tertiary Aegialornithidae Lydekker 1891, which exhibit a rather generalized overall osteology resembling both extant Hemiprocnidae and-apart from the more abbreviated humerus and tarsometatarsus-extant Aegothelidae. The Aegialornithidae are either considered to be closely related to the Hemiprocnidae (e.g. Harrison 1984; Karhu 1988, 1992; Mlikovsky 2002) or "the last representatives of an old radiation directed toward the realization of the type 'True Swifts'" (Mourer-Chauvire 1988:369).

A few isolated bones from the Upper Eocene to Lower Oligocene deposits of the Quercy were assigned to Cypselavus gallicus Gaillard 1908 (Mourer-Chauvire 1978) which is currently recognized as the earliest taxon of the Hemiprocnidae (Harrison 1984, Peters 1985, Mourer-Chauvire 1988). The earliest certain members of the Apodidae belong to Scaniacypselus Harrison 1984, which includes two species from the Middle Eocene of Denmark and Germany (Harrison 1984, Peters 1985, Mayr and Peters 1999).

Crown group Trochilidae have no early Tertiary fossil record. However, Karhu (1988) described a new apodiform taxon, Jungornis tesselatus, from the Lower Oligocene of the Northern Caucasus which agrees with extant Trochilidae in highly characteristic derived features of the humerus (Karhu 1988, 1992). Karhu (1988) classified J. tesselatus into a new taxon, Jungornithidae, to which he later (Karhu 1999) assigned the Upper Eocene species Argornis caucasicus Karhu 1999. Both Jungornis and Argornis are known from wing elements of a single individual only, and A. caucasicus clearly exhibits a less specialized wing morphology than J. tesselatus.

The only phylogenetic analysis of fossil apodiform birds in which the proposed relationships are depicted in some sort of phylogenetic tree is by Harrison (1984) who assumed a major split between a hemiprocnid and an apodid lineage (a recent cladistic analysis of apodiform birds by Dyke [2001] is based on a largely incorrect character matrix [see Mayr 2001] and is thus not discussed in the following). However, whereas Harrison (1984) listed some derived characters to support the apodid lineage, assignment of the taxa Eocypselus, Aegialornis, and Cypselavus to the hemiprocnid lineage was based on plesiomorphic characters ("more generalized humeral structure," "long slender ulna" with "more generalized proximal end;" see Harrison 1984:172).

Except for Karhu (1988, 1992, 1999), most authors further omitted the Trochilidae from their comparisons, and the present study is the first cladistic approach to the phylogeny of early Tertiary Apodiformes in which hummingbirds are included.

MATERIAL AND METHODS

Anatomical terminology follows Baumel and Witmer (1993) and Vanden Berge and Zweers (1993), if not indicated otherwise. Comparisons with extant taxa are based on skeletons in the collection of Forschungsinstitut Senckenberg; concerning extant Apodiformes the following species were studied: Aegothelidae: Aegotheles cristatus; Hemiprocnidae: Hemiprocne comata; Apodidae: Chaetura vauxi, Apus apus, Collocalia vanikorensis, Co. salangana; Trochilidae: Phaethornis pretrei, Glaucis hirsuta, Amazilia versicolor, Archilochus colubris, Calypte anna, Anthracothorax sp., and Chrysolampis mosquitus. Information on osteology of the Cypseloidinae (Apodidae) is based on illustrations and descriptions in Cohn (1968), Ballmann (1976), and Collins (1976a).