5^' PROMEROPS AS A THRUSH, AND ITS IMPLICATIONS FOR THE EVOLUTION OF NECTARIVORY IN BIRDS STORRS L, OLSON' & PETER L. AMES^ 'National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560, U.S.A. ^Harza Engineering Company, 150 South Wacker Drive, Chicago, Illinois 60606, U.S.A. Received September 1984 SUMMARY OLSON, S. L. & AMES, P. L. 1984. Promerops as a thrush and its implications for the evolution of nectar- ivory in birds. Ostrich 56:213-218 The familial relationships of the South African sugarbirds {Promerops) have never been firmly established, although an alliance with the Meliphagidae or the Sturnidae has been suggested on rather equivocal grounds. The syrinx in Promerops is now shown to have the derived "turdine thumb" configuration that is characteristic of the true Old Worid flycatchers and the thrushes (Muscicapini auct. and Turdinae), to which group Promerops should now be moved. Promerops shows that the highly specialized morphology and behaviour associated with nectarivory can evolve in a "saltational" manner, leaving no intermediate forms and few clues to the identity of the ancestral taxon. Nectarivorous feeding adaptations are a very poor indication of relationships. The closest relatives of other nectarivorous groups of birds are probably best sought among taxa that lack morphological specializations for nectar feeding. INTRODUCTION To those familiar with the South African Cape and Gurney's Sugarbirds Promerops cafer and P. gurneyi, either in the field or the museum, the title of this paper at first may seem unlikely, for it is difficult to reconcile the appearance and behav- iour of Pomerops with that of the thrushes, chats, and flycatchers of the family Muscicapidae. Yet we have discovered that the morphology of the syrinx makes this the most reasonable hypothesis of re- lationship of Promerops; indeed, it is now seen to be the only one for which any really sound evi- dence exists. The sugarbirds are adapted for feeding on the nectar of Proteaceae and consequently have a long, slightly decurved bill and a nectar-adapted tongue with a fringed tip. Both species have greenish- yellow rumps and conspicuous yellow undertail coverts and sides of the rump. The tail is elongate in both sexes of Promerops gurneyi and is spectac- ularly so in adult males of P. cafer, which, with the elaborate displays of the latter, make it a highly distinctive bird. As Skead (1967: 232) has re- marked, the sugarbirds "are not close relatives of the sunbirds [Nectariniidae, which are also nectar- adapted], and nothing else in Africa is really like them." Sibley & Ahlquist (1974) have outlined the taxo- nomic history of Promerops in detail. In brief, ex- cept for Sibley and Ahlquist themselves, virtually all previous authors have allied Promerops with other nectar-feeding birds, either the Nectarinii- dae or the Australasian honeyeaters of the family Meliphagidae. There has never been much sus- tained support for a nectariniid relationship for Promerops, but the genus has frequently been placed in or near the Meliphagidae. The more re- cent systematic studies of Promerops have dwelt on the differences between this genus and the Meli- phagidae, whereas no cogent morphological simi- larities between Promerops and any other group have been identified. It is now generally conceded that the feeding adaptations of Promerops and the Meliphagidae are only convergently similar. This, coupled with the zoogeographical difficulties in- herent in a relationship with the Mehphagidae, has caused some authors to place Promerops in its own family, Promeropidae. Sibley & Ahlquist (1974) on the other hand used the electrophoretic pattern of egg white proteins to show that Promerops was a nectar-adapted deriva- 213 214 OLSON & AMES: SUGARBIRD SYSTEMATICS OSTRICH 55 tive of the starlings (Sturnidae), and they sug- gested several external and behavioural characters supporting this. The unreliability of the egg white protein techniques (see Brush 1979) has at least in some instances resulted in what Sibley & Ahlquist (1980) have noted may have been erroneous con- clusions that may be viewed as compromising their data concerning Promerops. More importantly, their paper (Sibley & Ahlquist 1974) makes no comparisons between Promerops and the Musci- capidae, so their data cannot be evaluated in light of our findings. THE SYRINX AND RELATIONSHIPS OF PROMEROPS The syrinx in the Muscicapidae (including Tur- didae) is highly distinctive in having the dorsolat- eral musculature protruding as an elongated thumb-like structure, the "turdine thumb", first recognized by Ames (1975). This is characteristic of the true muscicapid flycatchers and the thrushes and has not yet been found in any genus thought to be unrelated to this group, with the exception of the Bornean genus Chlamydochaera, previously placed in the Campephagidae but now removed to the Muscicapidae {e.g. Ames 1975; Ahlquist etal. 1984; Olson MS). Certain genera thought to be re- lated to the Muscicapidae have been found to lack the "turdine thumb" and either have been, or should be, removed to other famihes (Ames 1975; Harrison 1976; Olson etal. 1983; Olson 1984). The relationships of most of these had been disputed in any case. The "turdine thumb" of the syrinx de- fines a natural, monophyletic group the inter- relationships of which have long been recognized on the basis of more traditional, though not always consistent, external systematic characters. On 11 July 1984, while examining the syringes of various Meliphagidae and some of their presumed relatives, we discovered that the syrinx in Prome- rops has the "turdine thumb" and is in every re- spect typically thrush-like (Fig. 1). We examined the syrinx in one specimen of Promerops gurneyi (AMNH 2765), and three specimens oiPromerops cafer (USNM 558752-53, AMNH no number), and found the "turdine thumb" extremely pro- nounced in all of them. Olson had collected two of the specimens of P. cafer in 1983 and prepared them as skeletons, at which time he saved the syringes and tongues in fluid. Although he routinely examined the syrinx * in the various muscicapids and presumed muscica- pids that he collected (see Olson 1984), the sy- ringes of Promerops were kept mainly because the taxonomic position of the genus had so long been in doubt, but Olson did not scrutinize these speci- mens for muscicapid affinities. Thus the "turdine thumb" was overlooked. Because Ames (1971) had previously examined the syrinx in Promerops and briefly reported on its condition in Sibley & Ahlquist (1974), a word of explanation for his also having overlooked the "turdine thumb" is in order. The syringeal charac- ter by which the Meliphagidae may be recognized occurs in the ventral musculature and consists of a distinctive posterior gap between the two halves of the ventral muscle mass. It is not necessary to ex- pose and examine the dorsal musculature that makes up the "turdine thumb" to check the meli- phagid condition, which is lacking in Promerops. Indeed, when we made our discovery, each of us had been looking at a syrinx of Promerops under a dissecting microscope for some time before Olson chanced to clear away the tissue overlying the dor- sal musculature to expose the obvious "turdine thumb". When Ames (1975) conducted his orig- inal survey of the distribution of this character, he looked for it among the various groups of "Old World Insect Eaters" with which the Muscicapidae are usually associated, but there was no reason at that time to examine Promerops in this connec- tion. In the present study, Ames re-examined the specimen that he described originally and found the "turdine thumb" to be present and well devel- oped. The tarsus is not booted nor are the young spotted in Promerops. These features are charac- teristic of many of the Muscicapidae but are nei- ther restricted to them nor invariably present. Par- ticularly in some of the chats, the tarsus may be scutellate and the young unspotted (Ripley 1952). The yellow in the plumage and the elongated tail of Promerops that Sibley & Ahlquist (1974) cite as being similar to those of some starlings, also occur in the Muscicapidae {e.g. Pogonocichla, the adults of which are mostly yellow, and Enicurus, in which the tail is greatly elongated). The peculiarly shaped primaries, particularly the fifth, of adult males of Promerops, which are not so much notched as they are greatly expanded in the basal ? 1984 OLSON & AMES: SUGARBIRD SYSIEMAUCS 215 FIGURE 1 Syringes in lateral view. A, Greater Blue-eared Starling, Lamprotornis chalybeus (Sturnidae), show- ing the generalized oscine configuration; B, Whitecrowned Cliffchat Thamnolaea coronara showing the typical "turdine" configuration; C, Gurney's Sugarbird Promeropsgurneyi; D, Cape Sugarbird P. cafer. The "turdine thumb" that defines the Muscicapidae, including Promerops, is indicated by arrows. two-thirds of the inner web, are not really similar to the notched condition in the starlings of the ge- nus Lamprotornis. The primaries may be modified in the Muscicapidae as well (e.g. Cercomela si- nuata), but again are not similar to the condition in Promerops. Because the primaries are much less modified in P. gurneyi and its displays correspon- dingly simpler (Skead 1967), this species appears to be the more primitive. The extreme specializa- tion of the primaries very likely arose within Pro- merops, as probably did the yellow in the plumage and the elongated tail, in which case these charac- ters are not likely to be useful indicators of re- lationship. The open, cup-shaped nest of Promerops is more like that in the Muscicapidae than in the Sturnidae, which typically, but not always, nest in cavities or crevices or build domed nests. Many other birds build cup-like nests, however, so this has little significance other than slightly favouring a muscicapid relationship over a sturnid one. The same holds for the eggs, which have a buffy ground colour and are heavily spotted and scrawled in Promerops, whereas in starlings the eggs are usually blue (rarely white) and usually spotted with red. The eggs in the Muscicapidae are vari- able in colour, sometimes being buffy and often being spotted, although to judge from the descrip- 216 OLSON & AMES: SUGARBIRD SYSTEMATICS OSTRICH 55 tions of South African species in McLachlan & Liversidge (1978), none seems to be as heavily marked and scrawled as in Promerops. One factor favouring a muscicapid rather than a meliphagid relationship for Promerops is geog- raphy, for the Muscicapidae have speciated and di- versified extensively in southern Africa. In the area covered by McLachlan & Liversidge (1978) there are about 43 resident species of Muscicapi- dae in 15 genera {Erithacus, Pogonocichla, Eryth- ropygia, Cossypha, Sax?cola, Myrmecocichla, Thamnolaea, C?rcamela, Oenanthe, Mont?cola, Zoothera, Turdus, Melaenornis, Namibornis, and Musc?capa). Some of these are among the most characteristic birds of drier, unforested areas such as frequented by Promerops. In all fairness, how- ever, it should be noted that none of the above genera bears the slightest resemblance to Prome- rops. There is no evidence to suggest that the distinc- tive "turdine" syrinx has evolved more than once. Apart from Promerops, this condition is present only in birds that had already been considered al- lied on traditional morphological characters and that can reasonably be assumed to constitute a monophyletic group. With the protein data mooted, the only evidence to be advanced in the attempt to determine the relationships of Prome- rops is in the morphology of the feeding apparatus and the syrinx. Nectarivorous specializations have evidently evolved independently in "at least 12 os- cine groups" (Sibley & Ahlquist 1974: 24), so con- vergence in feeding adaptations seems much more likely than in syringeal morphology. Furthermore, it had already been largely agreed upon that the feeding adaptations of Promerops probably evolved independently of those in other nectar- ivorous birds. Any argument that the similarities in syringeal morphology between Promerops and the Muscicapidae are the result of convergence would be unconvincing for the simple reason that there are no other characters of Promerops that strongly suggest a closer association between it and some other group. On the basis of the available evidence, we would place Promerops in the Muscicapidae, as defined by the syrinx. Within this group the relationships O? Promerops are otherwise obscure, but because it is clearly one of the most highly derived mem- bers of this radiation, Promerops can conveniently be placed at the end of any listing of the species of * Muscicapidae. THE IMPLICATIONS OF PROMEROPS FOR THE EVOL- UTION AND SYSTEMATICS OF NECTARIVOROUS BIRDS Many birds feed on nectar, even though not es- pecially adapted for it. In South Africa alone. Oat- ley & Skead (1972) have reported 73 species that are not primarily nectarivorous as feeding on nec- tar, including seven non-passerine families: doves (Columbidae), parrots (Psittacidae), buries (Mu- sophagidae), colies (Coliidae), woodhoopoes (Phoeniculidae), barbets (Capitonidae), and honeyguides (Indicatoridae). Among passerines they record drongos (Dicruridae), orioles (Orioli- dae), crows (Corvidae), tits (Paridae), babblers (Timaliidae), bulbuls (Pycnonotidae), warblers (Sylviidae), starlings (Sturnidae), and various finch-Hke birds (Ploceidae, Estrildidae, Cardueh- nae), among others, as feeding on nectar. Of inter- est in connection with Promerops, they list five genera and seven species of Muscicapidae as feed- ing on nectar (Cape Rock Trush Mont?cola rupes- tris. Mocking Chat Thamnolaea cinnamomeiven- tris. Buff streaked Chat Oenanthe bifasciata, Whitebrowed Robin Erythropygia leucophrys. Pallid Flycatcher Melaenornis pallidus, Black Fly- catcher M. pammelaina, and Fiscal Flycatcher M. silens). To these may be added Sentinel Rock Thrush Mont?cola explorator (T. B. Oatley, pers. comm.). Selection pressure to evolve morphological specializations that enhance the ability to feed on nectar must be fairly intense, with some such adap- tations having evolved more than a dozen times within the oscine passerines (Sibley & Ahlquist 1974). The case of Promerops is particularly in- structive in that it illustrates the fact that a high de- gree of specialization for nectarivory can appear in a "saltational" manner, leaving no intermediate forms and obliterating most traces of the ancestry of the newly evolved taxon. Non-nectar-adapted relatives have been identi- fied for some of the specialized nectarivores such as the Drepanidini, which are cardueline finches (Raikow 1976). Many of the Drepanidini, how- ever, are still finchlike and not nectarivorous. The various members of the "Coerebidae" are derived from the New World nine-primaried oscines 1984 OLSON & AMES: SUGARBIRD SYSTEMATICS 217 (Fringillidae sensu lato), mostly from the Thraupi- nae. Chloropsis, sometimes placed in its own fam- ily or in the "Irenidae" is almost certainly a nec- tarivorous bulbul (Pycnonotidae) (Olson, in prep.). The closest relatives of the Nectariniidae have not been identified, but if plumage similari- ties may be taken as suggestive, then it may be the sunbirds rather than Promerops that are derived from starlings (Sturnidae). Once specialized adaptations for nectarivory have evolved, the transformed species may then give rise to a radiation of numerous similarly adapted species. This has not taken place with Pro- merops, but it certainly has in such groups as the Nectariniidae, Mehphagidae, and the Drepani- dini. The many instances and apparent ease with which specializations for nectarivory have ap- peared suggest abundant opportunities for parallel evolution. Thus a single family or even a single ge- nus might give rise to specialized nectarivores on more than one occasion. This appears to have hap- pened in the Drepanidini, where the tubular tongue has almost certainly evolved more than once (Olson & James, in prep.). The "Coerebi- dae" is without doubt an aritifical collection in- cluding several independently evolved nectar- adapted derivatives of tanagers (Olson in Wet- more et al. 1984). Given the differences in tongue structure within the Nectariniidae (see Gill 1971, and references therein), it might be questioned whether this family is strictly monophyletic, that is, whether the closest common ancestor of all spe- cies of Nectariniidae was also specialized for nec- tar feeding. In summary, we should recognize (a) that ex- treme specialization for nectarivory may appear essentially de novo in a group that otherwise shows no such morphological tendencies, (b) that these specializations may effectively mask the affinities of specialized nectarivores, but (c) should not be used to exclude nectarivorous taxa from the fam- ily-level group from which they were derived once their relationships can be established and (d) that the strong possibility of parallel evolution may call into doubt the strict monophyly even of closely similar groups of species of nectarivores. As a thrush, Promerops demonstrates with unparal- leled force and clarity the fact that nectarivorous feeding adaptations are extremely unreliable for determining affinities within the Passeriformes. ACKNOWLEDGMENTS We are very much indebted to the Percy FitzPatrick Institute of African Ornithology for stimulating Olson's interest in South African birds and providing the oppor- tunity for him to observe and collect Promerops, in which he was guided by Anthony Rebelo. We are grate- ful to the American Museum of Natural History (AMNH), New York, for lending additional spirit speci- mens of Promerops, to Richard L. Zusi for transporting the specimen of P. gurneyi, to Terry B. Oatley for sup- plying references and information, and Richard K. Brooke for comments on the manuscript. The photo- graphs are by Victor E. Krantz, Smithsonian Institution. REFERENCES AHLQUIST, J. E., SHELDON, F. H. & SIBLEY, C. G. 1984. The relationships of the Bornean Bristlehead {Pityria- sis gymnocephala) and the Black-collared Thrush (Chlainydochaerajefferyi).J. Orn. 125: 129-140. AMES, P. L. 1971. Themorphology of the syrinx in pass- erine birds. Peabody Mus. Mat. Hist. Yale Univ. Bull. 37: 1-194. AMES, P. L. 1975. The application of syringeal mor- phology to the classification of the Old World Insect Eaters (Muscicapidae). Bonn. zool. Beitr. 26: 107-134. BRUSH, A. H. 1979. Comparison of egg-white proteins: effect of electrophoretic conditions. Biochem. Syst. Ecol. 1: 155-165. Gu.L.F. B. 1971. Tongue structure of the sunbird/iypo- gramma hypogrammica. Condorl3: 485-486. HARRISON, C. J. O. 1976. The syrinx of the Southern Scrub-robin Drymodes brunneipygia [sic]. Emu 76: 154. MCLACHLAN, G. R-, & LivERSiDGE, R. 1978. Roberts birds of South Africa. 3rd ed. Cape Town: Trustees of the John Voelcker Bird Book Fund. OATLEY, T. B. , & SKEAD, D. M, 1972. Nectar feeding by South African birds. Lammergeyer 15: 65-74. OLSON, S. L. 1984. Syringeal morphology and re- lationships of Chaetops (Timaliidae) and certain South African Muscicapidae. Ostrich 55: 30-32. OLSON, S. L. , PARKES, K. C-, CLENCH, M. H. & BORECKY, S. R. 1983. The affinities of the New Zealand pass- erine genus Turnagra. Notornis 30: 319-336. RAIKOW, R. J. 1976. The origin and evolution of the Ha- waiian honeycreepers (Drepanididae). Living Bird 15:95-117. RIPLEY, S. D. 1952. The thrushes. Postilla 13:1-48. SKEAD, C. J. 1967. The sunbirds of southern Africa, also the sugarbirds, the white-eyes and the spotted creeper. Cape Town: A.A. Baikema. SIBLEY, C. G. , & AHLQUIST, J, E. 1974. The relationships of the African sugarbirds {Promerops). Ostrich 45: 22-30. SIBLEY, C. G. , & AHLQUIST, J. E. 1980. The relationships 218 OLSON& AMES: SUGARBIRDSYSTEMATICS OSTRICH 55 of the "Primitive Insect Eaters" (Aves: Passed- WETMORE, A., PASQUIER, R, F., & OLSON, S, L. 1984. formes) as indicated by DNA x DNA hybridization. Birds of the Republic of Panama. Part 4. Smithson. Proc. 17th Int. Orn. Congr. 1215-1220. Misc. Coll. 150 (4).