Node Calibrated. Divergence between Apodidae and Hemiprocnidae.

Fossil Taxon. Scaniacypselus wardi (Harrison, 1984).

Specimen. NHMUKA5430 (Natural History Museum, London, UK): humerus, ulna, carpometacarpus, and alular phalanx.

Phylogenetic Justification. Phylogenetic analysis of morphological data (Mayr, 2003, 2005c) and

combined morphological and molecular data (Ksepka et al., 2013).

Minimum Age. 51 Ma.

Age Justification. NHMUKA5430 was collected from Bed R6 of the RØsnaes Clay Formation of Ølst, Denmark (Harrison, 1984). Bed R6 is the uppermost division of the RØsnaes Clay Formation, and is overlain by Bed L1 of the Lillebaelt Clay Formation (Heilmann-Clausen et al., 1985). Magnetostratigraphic work (Heilmann-Clausen et al., 2010) demonstrates that Bed L2 of the Lillebaelt Clay spans part of Chron 22r, which constrains the age of the base of the Lillebaelt Clay to>49 Ma. Thiede et al. (1980) assigned the upper calcareous beds of the RØsnaes Clay Formation, including R5 and R6 to nannoplankton biozones NP11 and NP12. Biostratigraphy supports correlation of the RØsnaes Clay Formation to the European mammal reference biozone MP8 (Mlíkovsky, 1996), which suggests an age>50 Ma (Gradstein et al., 2004). A conservative minimum age of 51 Ma is proposed, based specifically on the estimated age of the upper boundary of NP12, which is dated to 51 Ma (Gradstein et al., 2012).

Phylogenetic Position of Apodidae. Nearly all phylogenetic analyses have strongly supported a sister group relationship between Apodidae and Hemiprocnidae, with Trochilidae forming the sister group to the Apodidae + Hemiprocnidae clade (Mayr, 2003, 2005c, 2010a; Cracraft et al., 2004; Barrowclough et al., 2006; Ericson et al., 2006; Hackett et al., 2008; Braun and Huddleston, 2009; Nesbitt et al., 2011; Ksepka et al., 2013; though see Livezey and Zusi 2006, 2007) (Figure 7). Given the broad consensus that Apodidae and Hemiprocnidae are sister taxa, there is no conflict between morphological and molecular date to reconcile as pertains to the placement of Scaniacypselus wardi along the stem lineage leading to Apodidae.

Fossil Record of Pan-Apodidae. Swifts are known from a few good European specimens subsequent to their first appearance, though their fossil record is otherwise sparse (Figure 8). Scaniacypselus szarskii is known from multiple complete skeletons, some with intact feathering, from the Middle Eocene Messel Formation (Peters, 1985; Mayr and Peters, 1999). Additional fossils belonging to the true swift lineage include an isolated ulna of cf. Scaniacypselus from the Late Eocene of France (Mourer-Chauviré and Sigé, 2006), Procypseloides from the Eocene/Oligocene of France (Milne-Edwards, 1871; Mourer-Chauviré et al., 2004), and specimens from the Late Oligocene/ Early Miocene of Australia (Boles, 2001) and the Early and Middle Miocene of France (Milne-Edwards, 1871; Ennouchi, 1930; Collins, 1976; Harrison, 1984; Mlíkovsky, 2002).

Fossil Record of Related Clades. Hemiprocnidae lack a fossil record. The classification of the Eocene fossil Eocypselus vincenti within Hemiprocnidae (Harrison, 1984; Dyke, 2001b) has been shown to be erroneous, and this taxon is now considered a basal representative of Pan-Apodiformes (Mayr, 2001a, 2010b; Ksepka et al., 2013). Pan-Trochilidae (stem hummingbirds) is represented in Middle Eocene, Late Eocene, and Early Oligocene deposits (Karhu, 1988, 1999; Mayr, 2004 a, 2007; Bocheński and Bocheński, 2008; Louchart et al., 2008; Mayr and Micklich, 2008). Basal Pan-Apodiformes (taxa that diverged prior to the swift-hummingbird split) are also present in Eocene deposits in Europe and North America (Collins, 1976; Mourer-Chauviré, 1978, 1988a; Harrison and Walker, 1975; Mayr, 2010b; Ksepka et al., 2013). Because these stem swifts, stem hummingbirds, and basal pan-apodiforms were all small birds, it is plausible that the absence of Hemiprocnidae in the fossil record is due to geographical collecting biases. Hemiprocnidae occur today in Asia and Australasia, where relatively minimal fossil collecting effort has been expended compared to Europe and North America. Thus, the absence of Hemiprocnidae is not cause for extreme concern over the reliability of the calibration outlined here, though the small size of swifts in general implies a greater probability of a long gap at the base of the Apodidae-Hemiprocnidae divergence than might be expected for larger birds.