M. LeCroy & W. S. Peckover 236 Bull. B.O.C. 1998 118(4) The general size and description point to Rallina tricolor, but the species has never been collected in the Louisiades. A medium-sized kingfisher was seen perched in bright sunlight on a low branch over the river. It was a solid, almost royal, blue with no indication ofscaling on the head and the underparts were a rich rufous. When it flew, the only turquoise visible was in the rump. The bill was long, black and sturdy with a conspicuous red strip at the base of the lower mandible, and a white marking on the neck was very obvious. It resembled Alcedo azurea, which has not been reported from the Louisiades, except that azurea has no turquoise in the rump and neither the literature nor specimens indicate the presence of red at the base of the lower mandible. Acknowledgements We were able to visit Misima through the courtesy ofthe Misima Mines Pty. Ltd. and Placer Pacific Ltd., especially with the assistance of Ron Hiatt. The following mine personnelwereextremelyhelpfultouswhilewewereonMisima, allowingustousemine facilitiesandpersonneltransport("manhauls"): Poate Edoni, Ellis Illaia, Ian Lewis, Don Reid, Allan Storck, Alan Stevens, and Isreal Isreal, as was Len Selly ofPoon Catering. We are especially grateful to Alan Stevens for his reports ofa pair ofPeregrine Falcons and a White-bellied Sea Eagle on the island, as well as other birds. In Gulewa, Artu Sodias was very helpful in providing accommodation and assistance. Both he and Raynold Joshua helped us daily in the field and Darcy Siguia was of assistance in Gulewa. Our thanks also to Beresford and Anne Love ofPort Moresby for their ongoing help and friendship, and to Navu Kwapena of the Department of Conservation for his continued interest and help. Michael Quinnell, Anthropology Department, and Heather Janetski, Bird Depart- ment, Queensland Museum, tracked down relevant papers and specimens. Mark Peck and BradMillenatthe Royal Ontario Museum, Toronto, providedcompleteinformation on the Fleming Collection. At The Natural History Museum, London, J. C. Thackray, Archivist, and his assistants were very helpful during research on the Meek letters and Phil Ackery, Department of Entomology, replied to our queries concerning butterfly specimens; RobertPrys-JonesandMichaelWaltersatTringsearchedoutinformationon the MeekMisimaspecimenshoused inThe Natural History Museum, andansweredour queries concerning the specimens collected by Richards and Thomson. In addition, Prys-Jones read drafts ofthis manuscript and made many very helpful suggestions, and Walters informed us ofthe published Tristram type list. Clemency Fisher and Anthony Parker, LiverpoolMuseum, kindlyprovideduswithinformationon Meek, Richards, and Thomson specimens in their care; and Chris Milensky found Meek specimens in the National Museum of Natural History in Washington. David W. Snow and Chris Feare read the manuscript and suggested useful changes. To all ofthese we are most grateful. The Library StaffatAMNH, Donald ClydeandRoscoeThompsoninparticular, were alwayshelpful, aswere Diane Treonand MariaRios, Ornithology Department, and Dan Rutter, Graphics, who prepared the maps. Thanks also to Lauren May and Ramsey Togo, whose computer expertise is most appreciated. Brian Coates' two fine volumes (1985, 1990) have greatly facilitated our research. References: BrassA,rcL.hbJo.ld19E5x9.pedReistuilotnstoofNtehwe AGrucihnbeoald(1E9x5p6e-d1i9t5i7o)n.s.BulNlo..A7m9e.r.SuMmums.arNyato.fHtihset.F1i1f8t:h 1-69. Coates, B. J. 1985. The Birds ofPapua New Guinea, including the Bismarck Archipelago and Bougainville. Vol. 1. Non-passerines. Dove Publications Pty. Ltd., Alderley, Australia. Coates, B. J. 1990. The Birds ofPapua New Guinea, including the Bismarck Archipelago and Bougainville. Vol.2. Passerines. Dove Publications Pty. Ltd., Alderley, Australia. M. LeCroy & W. S. Peckover 237 Bull. B.O.C. 1998 118(4) DeVis,.C. W. 1890. Reports on birds from British New Guinea. Appendix G. InBritish Netv Guinea. AnnualReport bv HerMajesty'sAdministrator ofthe Governmentfrotn 4th September, 1888, to 30thJune, 1899. Victoria. DeVis, C. W. 1892. Report on the zoological gleanings ofthe administration during the years 1890-1891. Appendix CC. In Annual Report on British Nezv Guineafrotn 1st July, 1890, to 30thJune, 1891; with Appendices. Victoria. Diamond,J. M. & LeCrov, M. 1979. BirdsofKarkarand Bagabag Islands, New Guinea. Bull. Amer. Mus. Nat. Hist. 164: 467-531. Finch, B. S. [Compiler] 1985. Noteworthy observations in Papua New Guinea and Solomons. Papua Nezv Guinea BirdSoc. Nezvsletter, no. 215: 6-12. Galbraith, I. C. J. 1956. Variation, relationships and evolution in the Pachycephala pectoralis superspecies (Aves, Muscicapidae). Bull. Brit. Mus. (Nat. Hist.), Zoology 4: 133-222. Hamlin, H. MS. [Tvpescript Journal of 21-30 Julv 1930.] In Journal and Notes of Hannibal Hamlin, Nov. 8, 1928-Aug. 8, 1930. Vol. T, Whitney South Sea Expedition Journals in Dept. ofOrnithology, Amer. Mus. Nat. Hist., New York. Hartert, E. 1898a. [Mr. Ernst Hartert Characterized Three New Species of Birds from the Louisiade Archipelago.] Bull. Brit. Orn. CI. 8(58): 21. Hartert, E. 1898b. On the birds collected on Sudest Island inthe LouisiadeArchipelago by Albert S. Meek. Novit. Zool. 5: 521-532. Hartert, E. 1899. On the birds collected by Mr. Meek on St. Aignan Island in the Louisiade Archipelago. Novit. Zool. 6: 206-217. Hartert, E. 1919. Types of birds in the Tring Museum. B. Types in the General Collection. 1. Corvidae to Meliphagidae. Novit. Zool. 26: 123-178. Hartert, E. 1920. Types of birds in the Tring Museum. B. Types in the General Collection [Nectariniidae to Troglodytidae]. Novit. Zool. 27: 425-505. Hilder, B. 1980. The Voyage ofTorres. Univ. Queensland Press, St. Lucia. Maitland, A. G. 1893. GeologicalobservationsinBritishNewGuineain 1891. Appendix M. InAnnualReport onBritishNezu Guineafrom 1stJuly, 1891, to30thJune, 1892; zcith Appendices. Victoria. Mayr,E. 1940. BirdscollectedduringtheWhitneySouth SeaExpedition. XLI. Noteson New Guinea Birds. VI. Amer. Mus. Novit. 1056: 12pp. Mavr, E. 1941. List ofNew Guinea Birds. Amer. Mus. Nat. Hist., New York. Mavr, E. 1944. The birds of Timor and Sumba. Bull. Amer. Mus. Nat. Hist. 83: 123-194. Mavr, E. 1955. Notes on birds of Northern Melanesia. 3. Passeres. Amer. Mus. Novit. '1707: 46pp. Meek, A. S. 1913. A Naturalist in CannibalLand. T. Fisher Unwin, London. Mees, G. 1961. A systematic review ofthe Indo-Australian Zosteropidae (Part II). Zool. Verhandel. no. 50: 168pp. Mees, G. 1982. Birds from the Lowlands of Southern New Guinea (Merauke & Koembe). Zool. Verhandel. no. 191: 188pp. Paynter, R. A. Jr. (Ed.) 1967. Check-list of Birds of the World, Vol.12. Museum of Comparative Zoologv, Cambridge, Alass. Rothschild, M. 1983. DearLordRothschild. Balaban Publishers, Philadelphia. Rothschild, Hon. W. & Hartert, E. 1901. Notes on Papuan birds. III. Columbae. Novit. Zool. 8: 102-135. Rothschild, Lord & Hartert, E. 1918. A few additional notes on the birds of Rossel Island, Louisiade Group. Novit. Zool. 25: 311-312. Rothschild, Hon. W. & Jordan, K. 1898. On some new Lepidoptera from the East. Papilionidae. Novit. Zool. 6: 429-444. Salomonsen, F. 1983. Revision of the Melanesian Swiftlets (Apodes, Aves) and their conspecific forms in the Indo-Australian and Polvnesian region. Noona Dan Papers No. 141. Biol. Skr. Dan. Vid. Selsk. vol. 23, no. 5: pp. 1-112. Stresemann, E. 1913. Ornithologische Miszellen aus dem Indo-australischen Gebiet. XXVI. Die Formen von Alcedo ispida im Ostlichen und Sudlichen Asien und dem Indo-australischen Archipel. Novit. Zool. 20: 312-316. Thomson, B. H. 1889. New Guinea: narrative of an exploring expedition to the Louisiade and D'Entrecasteaux Islands. Proc. Roy. Geogr. Soc. 11: 525-542. Tolhurst, L. 1996. Two new species for Misima Island, Milne Bay Province. Muruk 8: 34-35. y. Holmgren 238 Bull. B.O.C. 1998 118(4) Tristram, H. B. 1882. Description ofa new Fruit-pigeon ofthe genus Carpoghaga from the Louisiade Archipelago. Proc. Zool. Soc. London (1881): 996. Tristram, H. B. 1889a. Some stray ornithological notes. Ibis, Ser. 6, 1: 227-229. Tristram, H. B. 1889b. On a small collection of birds from the Louisiade and D'Entrecasteaux Islands. Ibis, Ser. 6, 1: 553-558. Tristram, H. B. 1889c. Catalogue ofa Collection ofBirds. Printed at the "Advertiser" Office, Durham. Wagstaffe, R. 1978. Type Specimens of Birds in the Merseyside County Museums. Merseyside County Council, Liverpool. Warren, R. L. M. 1966. Type-specimens of Birds in The British Museum (Natural History). Vol. 1. Non-passerines. British Museum (Natural History), London. Warren, R. L. M. & Harrison, C. J. O. 1971. Type-specimens ofBirds in The British Museum (Natural History). Vol.2. Passerines. British Museum (Natural History), London. Warren, W. 1899. New species and genera of the families Drepanulidae, Thyrididae, Uraniidae, Epiplemidae, and Geometridae from the old-world regions. Novit. Zool. 6: 1-66. Wichmann, A. 1909. Nova Guinea, Entdeckungsgeschichte von Neu-Guinea bis 1828. Vol. 1. E. J. Brill, Leiden. Wichmann, A. 1910. Nova Guinea, Entdeckungsgeschichte von Neu-Guinea (1828 bis 1885). Vol.2, Pt. 1. E. J. Brill, Leiden. Wichmann, A. 1912. Nova Guinea, Entdeckungsgeschichte von Neu-Guinea (1885 bis 1902). Vol.2, Pt.2. E. J. Brill, Leiden. Addresses: Mary LeCroy, Department of Ornithology, American Museum of Natural History, New York, NY 10024, U.S.A. William S. Peckover, 14 Balanda Street, Jindalee, Queensland 4074, Australia. © British Ornithologists' Club 1998 A parsimonious phylogenetic tree for the swifts, Apodi, compared with DNA-analysis phylogenies hy Jan Holmgren Received 12 February 1997 In recent years, new and independent data for evaluating phylogenies in birds have been produced through biochemical and statistical analyses of genetic structures. The techniques used are complex, and room for errors of methodology and ambiguities of interpretation should be admitted (Sarich et al. 1989, Lanyon 1992). However, to a substantial extent the new results are consistent with one another, but are sometimes in conflict with traditional views (Bleiweiss et al. 1994, Lee et al. 1996). The swifts as a group generally seem to exhibit slow, continuous adaptations for their aerial life, giving the impressions of a morpho- logical conservatism caused by the uniformity of the food niche. The great variability in body size may perhaps be linked with interspecific competition. A great deal of other morphological variation may probably be linked with the intermittent contacts with solid matter in breeding and roosting. Here I present a parsimonious phylogenetic tree y. Holmgren 239 Bull.B.O.C. 1998 118(4) Afix)tropical origin 12 3 6 11 12 13 Apodini +4tH 4 Apus Charactersfound Tachymarptis inCypeeloidini: Cypaiums Twocarotid artories Aeronautea, Panyptila,Tachoniis- Nouseof salivainnest Schoutedenapus building 9 10 Chaeturini Broadsternum, Hinmdapua slenderhiunerus Chaetura Diastataxic wing Meamsia,Zoorxauena^ Telacanthura, Nakedhindlimb, Rhaphidura,Neafrapua- anisodactylfoot 7 8 Collocaliini -hrf Aervdramua CoUocalia Hydrochous Hemiprocnini Hemiprocne Cypseloidini Cypseloides Streptoprocne Figure 1. Hypothetical phylogenetic tree for the swifts. No time scale. Numbers denote change ofcharacters as follows: 1. One carotid artery. 2. Use ofsaliva in nestbuilding. 3. Gradual changetowardsnarrowsternumand stouthumerusas adaptations foraerial life. 4. Nestingandroostingonbranches, perching. 5. Nearstasis(perhapsevenregression) in change ofsternumandhumerus owingto less aerial life. 6. Eutaxicwing. 7. Nesting and roosting in caves, some have feathered hindlimbs. 8. Echolocation. 9. Nesting and roosting inside hollow trees, square and spined tails. 10. No nestbuilding. 11. Feathered hindlimbs. 12. Nesting and roosting in foliage of trees, reduced numbers of phalangi, reversed hindtoe, toes in opposite-pairs. 13. Change to more solid nest and roost sites, tendency towards "all four toes forwards". for the swifts, which is consistent with a hypothetical historical narrative (Mayr 1988) and zoogeographical facts. This simple tree may serve as a null hypothesis: suggested instances of further convergent or parallel evolution in the specified characters should be explicitly analysed. Methods The suggested phylogenetic tree in Figure 1 is derived from key characters (two carotid arteries or one carotid artery (Glenny 1955, Sibley & Ahlquist 1990), no use ofsaliva or use ofsaliva in nestbuilding y. Holmgren 240 Bull. B.O.C. 1998 118(4) (Marin & Stiles 1992, Chantler & Driessens 1995), gradual change towards narrow sternum and stout humerus as adaptations for aerial life (Sclater 1865, Harrison 1984), diastataxic (i.e. with apparent absence of 5th secondary remex) or eutaxic wing (Clark 1906, Sibley & Ahlquist 1990)) linked with a general understanding ofswift evolution, taking into account the role of various behaviours and substrata in breeding and roosting. A similar approach has proved successful in the swallows (Winkler & Sheldon 1993). The clarification of nest building in the Cypseloidine swifts (Marin & Stiles 1992) is an important recent achievement. The first methodical survey of all swift species (Chantler & Driessens 1995) facilitated an overview. In any phylogenetic tree, any lineage from the ancestral species to a recent species may be envisaged as a straight continuous line, with all other lineages branching off from that arbitrarily chosen trunk of the tree. In order to standardise trees for the swifts, I suggest that Apus apus should be chosen as this favoured recent species, simply because this will produce trees that are most similar to the traditionally accepted order between the main groups of swifts, and so will make direct comparisons easier. Since only the Cypseloidine swifts show primitive states in all the key characters, they are placed at the base of the hypothetical phylogenetic tree. Results an—d discussion The hypotheticalphylogenetic tree Figure 1 Brooke's (1970a) division of the Apodidae into the subfamilies Cypseloidinae and Apodinae (instead of the then traditional division between the Chaeturinae and the Apodinae, based on differences in the hindlimbs, derived from Sclater's (1865) Chaeturinae and Cypselinae) was based on differences in a number of characters. One of Brooke's Cypseloidine characters (diastataxic wing) is also present in the treeswifts, which I now suggest should be placed in Apodinae. Two characters (anisodactylous feet, downlike semiplumes on young nestlings) are present also in some Apodine species. For three characters (egg white profiles with four peaks, little or no transpalatine processes, simple Musculus splenius capitis) there is uncertainty about their exact validity, and one character (that there is a sexual dimorphism of plumage in some Cypseloidine species) is questionable since it may be an age character (females acquiring adult plumage later, M. Marin, pers. comm., Chantler & Driessens 1995). However, Brooke's division into subfamilies still seems warranted by two of his characters: the presence of two carotid arteries (Glenny 1955) and the now confirmed (Marin & Stiles 1992) lack of saliva in nest building in the Cypseloidine swifts. Sibley & Ahlquist (1990) critically reviewed morphological charac- ters used in phylogenetic work. Generally, conditions of the carotid arteries and the 5th secondary were found to be useful to some degree, but not consistent. However, within groups, two carotid arteries and diastataxy were consistently found to represent the ancestral condition. Use of saliva in nest building is special for swifts, and thus represents the derived condition. y. Holmgren 241 Bull. B.O.C. 1998 118(4) The treeswifts have one carotid artery, and they use saUva in nest building. Hypothetically, I suggest that the treeswifts should be treated as a tribe Heniiprocnini within Apodinae, so all extant swifts belong to the familyApodidae. Differences in anatomical details (Lowe 1939) and plumage in the treeswifts may well be explained by their different lifestyle. Fossil and osteological evidence (Sclater 1865, Harrison 1984) show a gradual change of sterna and humeri in the swifts, caused by A selection forces linked with their extremely specialised aerial life. broad sternum and slender humerus in the treeswifts may be explained by near stasis (perhaps even some regression) in this change, following the adoption of a less aerial life. Analysis ofchanges in humeral structures led Karkhu (1992) to quite different conclusions, suggesting long independent evolution of the Hemiprocnide and the Apodide swifts. He even suggested that the treeswifts should be placed in a suborder Hemiprocni, while the sub- order Apodi would include the other swifts and the hummingbirds, Trochilidae. He did not consider, however, the possibility ofnear stasis in the treeswifts, coinciding with convergent changes in other lineages leading to recent taxa. His interesting analysis of functional causes for the observed changes may be interpreted in support for convergent change. The Cypseloidine and Hemiprocnine swifts have diastataxic wings, while all other swifts have eutaxic wings (Clark 1906). In Collocaliini the Giant Swiftlet Hydrochous gigas is remarkably similar to the Cypseloidine swifts, and its choice of nest site is similar (M. Marin, pers. comm., Somadikarta 1968, King 1987). It cannot echolocate, and its nest is similar to Cypseloidine nests, except that some saliva is used (Becking 1971). When erecting the genus Schoutedenapus, De Roo (1968) pointed out a close similarity to Apus, but he found it impossible to place the taxon within Apus owing to the unmodified (anisodactyl) feet. He discussed, however, the possibility that the feathering of the legs may have been a first step in the evolution of the Apodine foot. Brooke (1970a) temporarily placed Schoutedenapus in Collocaliini, and he is followed by Chantler & Driessens (1995). In agreement with De Roo, I contend that the feathering of the hindlimbs in Schoutedenapus may be a first step in the evolution of the Apodine foot. Important evidence is still missing; for example no skeleton has been examined (C. T. Collins, pers. comm.). As it is, I suggest that Schoutedenapus should temporarily be included, as an early diverged taxon, in Apodini. The historical narrative Ifthe birds, as seems likely (Futuyma 1986, Chiappe 1995; for a good overview, though discordant in conclusions, see Feduccia 1996), evolved from a bipedal running and jumping insectivore catching prey in the air with the mouth (Caple et al. 1983), the swifts in principle might be the living representatives of a continuous trajectory, increasingly refining the faculty of flight, all the time using similar DNA food: flying insects. analyses suggest that the hummingbirds diverged from the swifts about 95 million years ago (Sibley & Ahlquist y. Holmgren 242 Bull. B.O.C. 1998 118(4) 1990), obviously changing to a different food niche. The earhest records in a revision of fossil swifts (Harrison 1984, see also Feduccia 1996) are from Europe and about 50 million years old. The swallows have had a shorter time than the swifts for the adaptations to their aerial life. According to DNA analyses (Sibley & Ahlquist 1990) the passerines diverged about 100 million years ago, with the swallows diverging after about half that time. Both swallows and swifts seem to have evolved in and radiated from the Afrotropical region (Brooke 1970b, Turner & Rose 1989). One might have expected a more complex pattern for the swifts, considering their more ancient origin. Perhaps earlier radiations ofswifts were largely wiped out in the great extinction about 65 million years ago (not unrealistic, if it was caused by an extraterrestrial impact, Alvarez et al. 1980), while one or a few species survived in the Afrotropics, and so had a chance to become ancestral to all later forms. Winkler & Sheldon (1993) superimposed the nest-building behav- iours on a DNA-hybridization phylogeny for 17 species of swallows. A remarkable evolutionary conservatism was revealed. Three main groups are characterised by distinct methods ofnest construction. Each method seems to have evolved just once, and the three groups have largely distinct geographical distributions. Those that excavate burrows for breeding are in the Afrotropics, except Cheramoeca in Australia, and Riparia, which has spread to the northern continents. Those that typically adopt cavities are in the New World, mainly tropical America, except Psalidoprocne fuliginosa and Phedina borhonica in the Afro- tropics. In those that typically build mud nests, Hirundo species are spread almost globally, while Delichon is restricted to Eurasia. The Tree Martin Hirundo nigricans in Australia may seem a cavity adopter, but uses mud on occasions, sometimes building a full mud nest. In contrast to earlier hypotheses (Mayr & Bond 1943, Turner & Rose 1989), burrow excavating was found to be a more primitive state than cavity adoption. A similar evolutionary conservatism is evident in the five main groups of swifts, which use five distinct substrata for breeding and roosting: (1) the ground, (2) branches of trees, (3) caves, (4) the inside of hollow trees, (5) foliage of trees, holes in trees, crevices in cliffs, etc. Similarly, the five groups have largely distinct geographical distributions. The first group, the Cypseloidinae, are restricted to the New World, but fossil finds in Europe indicate that they represent an earlywestward radiation from the Old World. They have an ancient method of nest building, "rooting" living vegetable matter, and so are restricted to humid sites with some daylight near waterfalls (Marin & Stiles 1992, Marin 1997). In the other groups, probably uniquely in birds, the use of saliva in nest building gave new evolutionary options: to place nests in various dry and/or dark sites. The treeswifts, in using "the outside" of trees, may seem to have a behaviour with some similarity to that ofthe Apodine swifts. However, several characters indicate that they diverged much earlier than the y. Holmgre?! 243 Bull. B.O.C. 1998 118(4) Apodine swifts, even earlier than Hydrochous, using branches and not fohage or holes. Their present distribution is in India and eastwards to the Solomon Islands. In the Collocaliini, Hydrochous gigas is similar to the Cypseloidine swifts, nests in daylight near waterfalls (and accordingly has no capability of echolocation), and uses much the same vegetable matter as the Cypseloidine swifts for nest building, but with some saliva mixed in (Somadikarta 1968, Becking 1971). A few Cypseloidine species sometimes nest in dark caves, where they place the eggs on ledges without building a nest (Whitacre 1989). The presence of the tendency towards using caves in some Cypseloidine swifts may strengthen the idea that Hydrochous represents an early eastward radiation by Cypseloidine-like swifts, however in an early stage ofusing saliva in nest building, in the course of time becoming ancestral to the swiftlets. Cypseloidine-like features can be found in some other swiftlets. The Glossy Swiftlet Collocalia esculenta cannot echolocate, nests in fairly well lit caves, and uses much vegetable material in its nests (Francis 1987, Chantler & Driessens 1995). Tarburton (pers. comm.) in Western Samoa observed a small colony of the White- rumped Swiftlet Aerodramus spodiopygius nesting in the spray of a waterfall in a very light situation on a cliff. Feathering on hindlimbs is present in some Collocaliine species and subspecies (Chantler & Driessens 1995). The Chaeturine swifts, the spinetails and needletails, all typically use the inside of hollow trees or similar structures, such as chimneys. Their use of the tail as a prop is generally accepted as an explanation for their stiff and square tails w4th spines. Several genera are present in the Afrotropics; a few of their species are also found far eastwards. The genera Hirundapus and Mearnsia clearly represent eastward radiations, while Chaetura, in the New World, represents a westward radiation. Most Apodine species are known to use solid sites, like crevices and holes in cliffs or trees, for nesting and roosting. However, I suggest that the original causes for the changes in their hindlimbs (feathering, reduced numbers of phalangi in toes III and IV from 4 and 5 respectively to 3, reorientation of toe I (hindtoe) inwards-forwards, the toes forming a pincer-like grasp, toes I and II opposing toes III and IV, Collins 1983) are linked w4th use of foliage of trees for nesting (palm swifts. Bock & Miller 1959) and roosting (a behaviour still present, for example, in the Common Swift, Holmgren 1993), and not primarily with clinging to vertical rough surfaces, as has often been claimed. Schoutedenapus (with feathered hindlimbs) and the palm swifts {Aeronautes, Panyptila, Tachornis, Cypsiurus, with the toes more clearly oriented in opposed pairs than inApus and Tachymarptis, which tend towards "all four toes forwards") then may be thought to represent early stages in these changes. I suggest that the Apodine swifts should be characterised by a gradual change in the hindlimbs, in the form of a transformation series, rather than by just the most advanced stages of that change, a redefinition that opens up the possibility to include Schoutedenapus. They are present in the y. Holmgren 244 Bull. B.O.C. 1998 118(4) Afrotropics and the Palearctic, but some species spread eastwards, and others westwards to tropical America (Aeronautes, Panyptila, Tachornis). The hypothetical phylogenetic tree in Figure 1, following this historical narrative, gives a parsimonious solution, and it suggests few changes in the current taxonomy of the swifts (cf. Brooke 1970a, Chantler & Driessens 1995). At present, this only means that it is a simple model for the understanding of swift evolution, and it cannot claim to be faultless. In the phylogenetic tree there are two cases of convergent evolution. The first one is caused by my insistence on placing the Hemiprocnine swifts together with all other swifts that have one carotid artery and use saliva in nest building. This forces me to admit gradual change towards a narrow sternum and stout humerus in two different lineages. However, that tendency is generally present in the swifts owing to their aerial life style, so this convergence should be expected to be omnipresent in the group, except in the treeswifts. The second case is the feathering on the legs in some swiftlets, which may be convergent with the earliest stage in the evolution of the Apodine foot. DNA Comparisons with -analysis phylogenies In Figure 2 the phylogeny for the swifts presented by Sibley & Ahlquist (1990) is adapted to the earlier proposed standard, so it can be directly compared with the tree in Figure 1. Agreement with Sibley & Ahlquist's results, with Hemiprocne in a separate family being the DNA-DNA earliest branching, was found in a repeated study using hybridisation (Bleiweiss et al. 1994). The trees in Figures 3 and 4 are adapted from Lee et al. (1996), who used DNA-sequencing. They derived the first tree by maximum likelihood estimate based on cytochrome b sequences, the second is a bootstrap consensus tree for 1,000 replicates. Both trees have Hemiprocne in the same position as Figure 2. Nevertheless I suggest that this question should be regarded as unresolved, until the new molecular techniques are better understood. Whether the Cypseloidinae are monophyletic seems to be a question that requires more precision than is attained at present (different results appear in Figures 3 and 4). The next problem in Figure 2 is the place of Tachornis. Convergent change in the hindlimbs in different lineages is certainly possible (cf. the feathering on the legs in some swiftlets). Aeronautes, Panyptila and Tachornis might form a separate tribe, representing an earlier branching than the Apodini. However, that the (reduced) numbers of phalangi in the toes are exactly the same seems to be such a specific character that it strongly suggests monophyly. Also worrying is the place of Tachornis between the needletails and the spinetails. It is possible, but seems unlikely. The place of Collocalia nearest to Apus then seems very unlikely, since Hydrochous so obviously has several characters in common with the Cypseloidine swifts. J. Holmgren 245 Bull. B.O.C. 1998 118(4) 50 40 30 20 10 Millionyearsago 3 126: 11 12 13 -h-h Apusaffinis Collocalia Chaeturapelagica Rhaphidura 11 12 +H- Tachomis 9 10 Hirundapus Streptoprocne 2 4 5 Hemiprocne Figure 2. Phylogenetic tree for the swifts, adapted from Sibley & Ahlquist 1990 figure 361, derived by DNA-DNA hybridisation. A T5oH:^4.5 million years. Numbers have same meanings as in Figure 1. Figures 3 and 4 also have Hydrochous near Apus. This would force us to infer a very remarkable regression in Hydrochous: loss ofusing caves, loss ofecholocation (since it is placed withAerodramus), minimising the use of saliva in nest building, and a renewed use of a behaviour similar to that of the Cypseloidine swifts, nesting near waterfalls and using similar vegetable matter. A remarkable result in Figures 3 and 4 is the polyphyly of the CoUocaliine swifts. In Figure 3, allowing for only a little uncertainty in the applied technique, an arrangement similar to Figure 1, lessening the problem with Hydrochous, would appear by connecting the Collocalia and the Aerodramus branches, leaving Chaetura on its own. This also would fit the ectoparasite evidence mentioned by Lee et al. (1996). In Figure 4 the situation is far from clear, and seems to reflect difficulties with the applied technique.