SUMMARY An attempt is made on the basis of anatomical material, to show that the neotropical Honey Creepers (Coerebidae) are an artificial group composed of nectar-adapted warblers and nectar-adapted tanagers that have become convergently similar due to similar diet. Considerable care is taken to select for comparison anatomical characters likely to have been modified but little by such convergence. This precaution is necessitated by the close relationship of the warbler and tanager families due to their apparent common origin from ancestral vireos. Evidence from jaw muscle-pattern and horny palate relief is given special weight because of the constancy of these features in other passerine families studied anatomically. Tanagers and warblers have distinctly different patterns for both characters and it has been found that a clean separation of the Coerebidae into nectar-adapted warblers and nectar-adapted tanagers is possible. This dichotomy is in good agreement with plumage differences, the latter group having tanager plumage, the former, warbler plumage. A functional difference in the manner of feeding on nectar is also in agreement with this division. Such characters as bill shape and tongue specialization overlap convergently and are unreliable. On the basis of this analysis it is concluded that the family Coerebidae should be submerged as an unnatural group, and its genera distributed among the warblers and tanagers. The genera Coereba, Conirostrum and Ateleodacnis could be regarded as a nectar-specialized tribe of the Parulidae--the Coerebini. The genera Diglossa, Cyanerpes, Chlorophanes, Iridophanes, Hemidacnis, Dacnis nis and Euneornis, and probably also Xenodacnis and Oreomanes, could be regarded as a nectar-specialized tribe of the Thraupidae--the Dacnini.

XTENSIVE efforts of the writer to find a sound anatomical basis for determining the phylogenetic relationships of passerine families leave it clear that the hazard of adaptive convergence in bird systematics has been underestimated. The present analysis of convergence in the neotropical Honey Creepers (family Coerebidae) offers evidence that this is an artificial group. It appears to be composed, in fact, of nectar-adapted warblers (Parulidae) and nectar-adapted tanagers (Thraupidae) that have evolved convergently because of similarity of diet. The Convergence Hazard in Taxonomy Sound systematic work in the higher categories demands sound criteria for clearly distinguishing between adaptation and phylogeny. The investigator at this level sees phylogeny ttrough a screen of food and niche adaptations which often obscure true relationships. Such classic cases of convergence between Old and New World groups as were recently reviewed by Friedmann (1946) are obvious and constitute no hazard. But convergence between members of closely related groups occupying the same range may be such that the most expert taxonomists are unable to decide the true affinities on the basis of external characters alone (Beecher, 1950). This is no reflection on the taxonomists, who have generally been the first to recognize the problem, referring such moot groups as the Coerebidae to the comparative anatomist for additional evidence. But internal characters are not necessarily more reliable than external ones for indicating phylogeny. They are merely additional clues, often of a very conservative sort, but sometimes capable of adaptive changes as rapid as those known for any external features. Sclater (1886: 1) long ago observed that it was "in some instances difficult to distinguish" the Coerebidae from warblers on the one hand and tanagers on the other. Lucas (1894: 299-309) made an anatomical survey of several of the most important coerebid genera; though he considered his findings confusing and inconclusive, they nevertheless confirmed an opinion many times expressed that the group needed study and was probably heterogeneous. Ridgway (1902: 377) obviously regarded the Coerebidae as close to the Parulidae and Thraupidae. He even removed the The writer is greatly indebted to the United States National Museum, the American Museum of Natural History, and the Chicago Natural History Museum for lending material. For use of specimens in their care or for information or advice, he wishes to thank Alexander Wetmore, Herbert Friedmann, John T. Zimmer, Ernst Mayr, Dean Areadon, Charles O'Brien, Josselyn Van Tyne, D. Dwight Davis, Emmet R. Blake and Melvyn A. Traylor, Jr. 274 William J. Beecher CONVERGENCE IN THE COEREBIDAE 275 genera Conirostrum and Ateleodacnis to the Parulidae, whence Lucas (1894: 309-310) had earlier removed Certhidea. Both Sclater (1886: 47) and Ridgway (loc. cit.) placed Glossiptila (Euneornis) in a monotypic subfamily under the Coerebidae. Hellmayr (1935: 218-331), however, placed all these genera, except Certhidea, in the Coerebidae without subfamilial distinction. Thus there has been a difference of opinion as to the status of certain genera, but a consensus that the family belongs close to the warblers and tanagers. Convergence in the Coerebidae It is axiomatic that a family must be erected on a basis of characters com- mon to the included genera, and the assumption is that these characters will be fundamental, expressing a common phylogenetic origin. As pointed out by Mayr (1942: 276) and Simpson (1945: 17), any higher category must be mono- phyletic. The Coerebidae do not appear to satisfy these requirements. Examining the characters of the group we find none that really distinguish it. Some species are brilliant in plumage, others dull. The bill is extremely variable--long, short, straight, curved, notched, even hooked--and this di- versity of form extends to the tongue and to the head as a whole. The only feature common to all members is the adaptive trait of nectar-feeding. This trait is naturally suspect as a character if the birds involved are to be classified according to common origin. Lucas (1894), in attempting to determine the origins of the family, seized principally upon those features most strongly modified by the nectar-feeding adaptation--bony palate, tongue, and intestine. Ridgway (1902: 375) erred, however, when he stated that Lucas considered the Coerebidae "apparently more nearly related to the Australasian family Meliphagidae (Honey-eaters) than to the American families Mniotiltidae and Tanagridae .... "What Lucas did say (1894: 309) was that "the Coerebidae do not form a homogeneous group .... "His principal reference to the Meli- phagidae (loc. cit.) concerned the tongue of Certhiola, which he believed to resemble "that of some of the Meliphagidae .... " Evidence of convergence is to be obtained through a careful study of morph- ology, especially through study of parts of the animal not likely to be modified by the selection pressure bringing about convergence. It is well known that dif- ferent parts of an animal evolve at different rates (Simpson, 1944). The hind limb is very similar in muscle pattern throughout the order Passeriformes and seems to have become relatively static after attaining a high leve! of general efficiency. The syringeal musculature, slightly less conservative, has served to divide this large order into suborders. But there is evidence (Beecher, 1951) that the bill and skull are still rapidly evolving in passerines leading the way in evolution. Having dissected to date nearly a thousand specimens of oscinine birds in all families (over six hundred species), the writer finds that the more rapidly-evolving jaw muscle-patterns are clearly different for each Dec. 195l 276 THE WILSON BULLETIN vo. 63, No. 4 family. When used with the several additional lines of evidence indicated later in this paper they seem to delineate an evolutionary sequence based on adap- tation to various food types. Evidence of Convergence from Jaw Muscle-Patterns To recognize the family differences in jaw muscle-pattern it is necessary to distinguish between two basic types of skeletal muscle--pinnate and parallel (Pfuhl, 1936). In the parallel type all fibers run the full length of the muscle and exert pull, upon contraction, directly between the points of attachment. The vireos (V[reonidae) typify this pattern (Fig. 1) with nearly the entire V1RO ~fi, l.t;!.o9uus'" VIRtONIDAEr This famity FIG. 1. Jaw muscle-pattern of Vireo altiloquus broken down into functional groups. In gaping, M. depressor mandibulae (1) depresses the lower mandible, while M. protractor qua& rati (2) elevates the upper mandible. The combined action of the palatine retractors 3, 4, and 5 draws the upper mandible down- ward. The palatine retractors are: 3. M. pterygoideus dorsalis (a) anterior; (b) posterior 4. M. pterygoideus ventralis (a) anterior; (b) posterior 5. M. pseudotemporalis profundus The combined action of the mandibular adductors 6 and 7 draws the lower mandible upward. The mandibular adductors are: 6. M. pseudotemporalis supeFScialis 7. M. adductor mandibulae (a) externus superficialis; (b) externus mediaIls; (c) externus pro- fundus; (d) posterior jaw musculature parallel. The pinnate type is one in which the tendon or raphe runs the length of the muscle and the short fibers originate from it as the barbs originate from the shaft of a feather. The warblers show this pattern in M. pseudotemporalis superficialis (6) and M. adductor mandibulae externus superficialis (7a) as seen in Figure 2 (Oporornis). Grant (1942: 384) has pointed out that such a muscle is more efficient, having "a much greater functional cross section for its bulk." It has appeared to the writer, on the basis of both skull structure and parallel Fro. 2. Comparison of jaw muscle-pattern in typical warblers (Parulidae) and tanagers (Thraupidae) with that of their nectar-adapted forms. 277 Dec. 1951 278 THE WILSON BULLETIN Vol. 63, No. 4 jaw muscle-pattern, that the Vireonidae, like the Old World Monarch fly- catchers from which it stems, is composed of phylogenetically primitive insect- eaters (Beecher, unpublished). Distribution patterns suggest that such a type may have existed even prior to the origin of flowering plants in the Upper Cretaceous and that all of the nine-primaried American families stem from ancestral vireos. The lighter, more agile Parulidae appear to have arisen with pinnate adductors of less mass and greater efficiency--a more adaptable group of insect-eaters. The heavier Thraupidae--a group adapted to feed on the abundant fruit of flowering plants--appear to have arisen with somewhat less pinnate adductors. The Parulidae and Thraupidae are, in fact, to be regarded as real adaptive branches on the phylogenetic tree, each having arisen in re- sponse to intense selection pressure on the primitive insect-eaters. The impor- tance of the huge food supply furnished by the origin of flowering plants in producing this pressure cannot be over-emphasized. Logically it should come to bear similarly on warblers and tanagers, independently producing nectar- adapted and seed-adapted groups with ever-increasing tendency toward pinnate adductors. TACI"IYPHONU5 PIRANGA CARPODACU5 TANAG!RP Fc,. 3. An apparenl shift of muscle fibers from M7b to 316 (from left to right) with creased adducting power in the cardueline line of tanagers. To argue the question whether the increase in pinnate musculature is phylo- genetic or adaptive is academic. It is both, but there is every indication that such advances in jaw muscle-pattern are conservative enough to have phylo- genetic value for taxonomic diagnosis at the family level of category. The black- bird family (Icteridae) which appear to have arisen from the emberizine finches via the finch-like Cowbird, Molothrus (Beecher, 1951), retains pinnate adductors despite all the food adaptations and accompanying bill changes for which its genera are noted. The muscle-patterns of tanagers and warblers differ from each other and offer a sound basis for taxonomic separation; but, before presenting evidence of convergence in the Coerebidae, it is necessary to em- phasize some unusual features of the Thraupidae. This family, as outlined below, is very large and diverse. Two main lines are distinguishable, leading almost without disjunction to the finch subfamilies William J. Beecher CONVERGENCE IN THE COEREBIDAE 279 Carduelinae and Richmondeninae. On the basis of jaw muscle-pattern these lines stem from a relatively primitive tanager group comprising Calospiza; Thraupis and Ramphocelus, of which the large genus Calospiza is considered by the writer to be the simplest, anatomically, and Ramphocelus the most advanced. The latter continues the series into the Richmondeninae in the se- quence Ramphocelus--Hemispingus--Spindalis Saltator--Pitylus--Pheucticus --Richmondena. This is not necessarily a direct evolutionary sequence, although it moves in the direction of increasing finch character, especially in the bill and in the heightened pinnate character of M. adductor mandibulae externus superficialis (M7a), the latter beginning in Hemispingus and increasing in the series. There is also good plumage agreement, especially in females, and the horny palate shows an increased tendency to develop a subsidiary lateral ridge, additional to the one so prominent in all tanagers (see Calospiza, Fig. 4). CA LO , PIZA CYAN LAT'AL CO_RE-BA OPORORNIS Fro. 4. Comparison of relief pattern seen in the horny palate of typical warblers (Parulldae) and tanagers (Thraupidae) with that of their nectar-adapted forms. Ramphocelus might, because of its apparent close relationship with Tachy- phonus, be considered a point of origin for the Carduelinae also--a group quite different in its muscular emphasis. Tachyphonus lies at the base of this line. The muscle-pattern of the weak-billed T. surinamus is like that of Ramphocelus, hence also of Calospiza (Fig. 2); but in the heavier-billed T. rufus the orbital slip of M. adductor mandibulae (M7b) is pinnate (Fig. 3). This appears to be correlated with increased adducting power, but with still further increase in the series Tachyphonus--PirangHabia--Tanagra--Chlorophonia--Stephano- phorus--Carpodacus (Fig. 3) there appears to be a shift in emphasis from M7b to M6. The series terminates in the tremendous forward advance of the latter in Carpodacus, which is typical for the Carduelinae as a whole. That this shift involves a transfer of fibers to M6 is suggested in the sketch of these two mus- Dec. 1951 280 THE WILSON BULLETIN Vo. 63, No. 4 cles for Tanagra chlorotica (Fig. 3), where transfer seems to be taking place. The horny palate in this series agrees closely with that of Calospiza (Fig. 4), and plumage generally supports the thesis of close relationship. This digression on the complexities of the Thraupidae is necessary. In no other of the sixty-odd oscinine families do such variations in the muscle-pat- tern occur. It points up the peculiar fact that the richmondenine finches arise so uninterruptedly out of the tanagers that ornithologists have had to draw the dividing line between the two groups arbitrarily. Now we find the cardueline ("Old World") finches arising also without disjunction from another line of tanagers. Chlorophonia cyanea, with a muscle-pattern like that of Carpodacus, has a flattened tanager bill. The horny palate of Stephanophorus diadematus, a form internally almost identical with Carpodacus, has crenulated lateral ridges like those of the plant cutter Phytotoma--a further adaptation for bud- eating, so common in the Carduelinae. Apparently it will be necessary to declare these two groups of tanager-finches subfamilies of the Thraupidae, maintaining another subfamily for the more typical tanagers. It would hardly do to extend the Richmondeninae and Carduelinae to include the two generic series cited above as stemming from Ramphocelus. Such a grouping would be as arbitrary as the one currently fol- lowed and the writer contemplates no change other than the including of the finch groups in the Thraupidae. But it is necessary to recognize that the ana- tomical modifications terminating in these finch groups have their roots deep in the typical tanagers, the Thraupinae. For this reason the basic muscle- pattern for tanagers is regarded by the writer as that found only in Calospiza, Thraupis and Ramphocelus, though genera not yet dissected by him may be included later. It is this pattern that the writer uses in his present re-examina- tion of the Coerebidae. A survey of the Vireonidae, here considered as ancestral to both the Parulidae and Thraupidae, reveals a muscle-pattern that is notably non-pinnate. In Vireo altiloquus and V. olivaceus a suggestion of pinnate character appears in M. adductor mandibulae externus medialis (M7b in Fig. 1), and this muscle, along with M7a, c, and M6, becomes fully pinnate in the shrike-like Cyclarhis and Smaragdolanius. But Vireo griseus and V. fiavifrons lack this suggestion, as do Hylophilus olivaceus, H. ypoxanthus and H. decurtatus. This muscle is short and non-pinnate in the basic muscle-pattern for tanagers such as Calospiza but is always long and pinnate in such warblers as Oporornis (Fig. 2), Setophaga, Basileuterus, Myioborus, Granatellus, Dendroica, Protonotaria, Geothlypis, Parula, Seiurus, Mniotilta and Icteria. Thus we see a segregation on the basis of M7b that is diagnostic of warblers and tanagers. Warblers differ from the basic tanagers in having a far more pinnate muscle- pattern, this in spite of the fact that they have much more slender bills and much lighter muscle mass. The pattern seen in Oporornis is typical. Pinnate muscles are M. depressor mandibulae (M1), M. pseudotemporalis superficialis William J.2Be1/2cher CONVERGENCE IN THE COEREBIDAE 281 (M6), and three slips of M. adductor mandibulae, viz., externus superficialis (M7a), externus medialis (M7b), and externus profundus (M7c). In the basic tanagers only M6 and M7c are pinnate. M. depressor mandibulae (M1) is always more highly developed in warblers than in tanagers (Fig. 2). It was at first puzzling to find both tanager and warbler muscle-patterns appearing in different members of the family Coerebidae. The jaw muscle- pattern of Coereba fiaveola (Fig. 2) corresponds perfectly, in fact, with that of Oporornis formosus, a typical warbler, while that of Cyanerpes cyaneus corre- sponds perfectly with that of Calospiza arthus, a typical tanager. It is true that Coereba and Cyanerpes, in adaptation to nectar-feeding, have an enlarged palatine salivary gland which causes the insertion of M. plerygoideus ventralis posterior (M4b) to split into two slips. Even here, however, the manner of the musde's division is different: in Coereba it is split laterally and posteriorly-- in Cyanerpes, laterally and medially. That this is not simply coincidental is borne out by the fact that, wherever this muscle is split by an enlarged palatine gland in coerebid genera, the disposal of the divided slips is either as in Coereba or as in Cyanerpes. Intermediate arrangements do not occur. On the basis of muscle-pattern, then, it would appear that Coereba is nothing but a nectar-adapted warbler, and Cyanerpes a nectar-adapted tanager. The differences in muscle-pattern and -arrangement may be interpreted as the slightly different ways of achieving the same adaptation by two slightly different, converging stocks. Confidence in this interpretation is encouraged by the fact that a complete and sharp dichotomy of the Coerebidae is possible on the basis of the above differences. Those members which appear to be warblers are Coereba, Conirostrum and Ateleodacnis. Those which appear to be tanagers are Cyanerpes, Chlorophanes, Iridophanes, Diglossa, ttemidacnis, Dacnis and Euneornis. Evidence of Convergence from Plumage Pattern The writer has not examined Xenodacnis or Oreomanes, but he presumes them to be nectar-adapted tanagers. Such a presumption is plausible because the suggested dichotomy of coerebid genera is supported by plumage differences. Coereba, Coniroslrum and Aleleodacnis have a warbler type of plumage, the other genera a tanager type of plumage, most of them with the same iridescent blues and greens observable in Calospiza, a few tending toward the plumbeous blues of Thraupis. The bizarre bill of Diglossa and unique tongue of Euneornis might cast doubt on this simple picture but the division is supported by muscle- pattern and the relief pattern of the horny palate. Moreover, there is no reason to think either that a single warbler genus and a single tanager genus gave rise to nectar-adapted warblers and tanagers, or that the nectar-adapted forms are all equally adapted to the same flowers. As will be shown beyond, the adaptation has apparently occurred in far-removed cases (ttylophilus poici- lotis is a nectar-adapted vireo that escaped inclusion in the Coerebidae) and Dec. 195i 282 THE WILSON BULLETIN Vo. 63, No. 4 it is partially achieved in warblers and tanagers never included in the Coerebidae. Regarding the reliability of plumage as an indicator of relationship, studies of the American orioles (Beecher, 1950) show that species are not generally under strong selection.pressure to change plumage pattern once this pattern has reached a state of adaptive adjustment to the environment. If the environ- ment changes greatly, plumage tends to evolve rapidly in re-adjustment. Applying this to the present group, there appears to be no reason why tanagers taking up nectar-feeding should change the plumage facies drastically. Since they have not, despite structural changes in the head region accompanying diet specialization, changed their environmental niche, there is no reason for expecting this plumage to change much. The same applies to nectar-special- ized warblers: they may logically be expected to retain warbler plumage patterns. Only when change in feeding habit requires change in habitat, as in the derivation of buntings (Emberizinae) from the Parulidae, may rapid and drastic plumage changes be expected. Evidence of Convergence from the Horny Palate Sushkin (1927: 3) was impressed with the value of the relief pattern of the horny palate in diagnosing large groups. This character, not to be confused with the bony palate as seen in cleaned skulls, is apparent in the roof of the upper mandible of bird skins prepared with the bill open or in alcoholic speci- mens. As shown in Figure 4, there are three distinct ridges on the horny palate of both warblers and tanagers, and the differential emphasis on these in the two groups permits separation of their convergent representatives in the Coe- rebidae. In tanagers and their nectar-adapted forms the lateral ridges are strongly emphasized and continuous to the back of the tomium, whereas in warblers and their nectar-adapted forms they are little emphasized and die out posteriorly. The palate, posterior to the central ri:tge, is flat and virtually without relief in tanagers, vaulted and sculptured in warblers. These features, especially the condition of the lateral ridges, are diagnostic of warblers and tanagers and support the same dichotomy of coerebid genera as has been indicated above by muscle-pattern and plumage. Evidence of Convergence from the Tongue If more than one warbler genus and more than one tanager genus evolve nectar-adapted forms, and if these in turn are specialized for particular flower types rather than for all flowers, convergent overlap of bills and tongues is to be anticipated. It is clear that diagnostic distinctions in external bill structure do not occur among the warblers and tanagers feeding on nectar, even though the horny palate and muscle-pattern have in each case passed through this adaptation without loss of their basic characters. In the tongue, which varies considerably even among closely related species of warblers, we should expect William j. Beether CONVERGENCE IN THE COEREBIDAE 283 least of all to find conservatism, particularly if there are special adaptations for special kinds of flowers. Here, in fact, we find the least satisfactory means for distinguishing between nectar-adapted warblers and nectar-adapted tan- agers because basic distinctions are apparently obliterated by convergent "overlap" of these too-plastic characters. In general the Hawaiian Drepanididae, 2 which are almost certainly of tan- ager origin (Beecher, unpublished), have progressed farther in tongue modifi- cation than any coerebid genus. As Lucas (1894: 306) has observed, the tongues of Himatione, Hemignathus and Vestiaria are perfectly tubular, the fimbriated edges actually overlapping. If the bifid tongues of Cyanerpes and Dacnis con- tinued to curl, instead of simply forming a highly fimbriated half-curl, the result would be the twin tubes of the Diglossa tongue, and this tends to draw all nectar-adapted tanagers together. But there is apparently no sound basis for completely separating the nectar-adapted warblers with their bifid tongues from this broad and variable pattern. It is, however, noteworthy that the whipped out tongue of Coereba is remarkably similar to that of the oriole I-I A3/4IAI L.N I I W'I"' Vcs\a coms ~ DRPANIDID/ ' Mike Cus bull emized for zm Fro. 5. Jaw muscle-pattern of Vestiaria coccinea of the Drepani(lidae, showing similarity to Calospiza and Cyanerpes despite development of M1 and M2 for gaping. Bananivorus cucullalus (Beecher, 1950). Both are believed to arise from the warbler phyletic line, the emberizine finches being ancestral to the blackbirds (Icteridae), so this is apparently a case of common stock accomplishing the same adaptation twice in about the same way. The tongue of Euneornis is unique and, without histological sections, it is not even quite clear how it functions. But the bird is a tanager on the basis of jaw muscles and horny palate, and probably closely related to Loxigilla, which it resembles in these features and even in plumage pattern. Both genera occur in Jamaica, to which Euneornis is restricted, and it is possible that the finch Loxigilla arose from this tanager. Plumage change might not be expected since Gosse (1847) states that both inhabit the same niche (forest) there. 2 Spelled also Drepaniidae and Drepanidae. See review of Amadon's "The Hawaiian Honeycreepers (Aves, Drepaniidae)" in this issue of the Bulletin, and also the comments of Ernst Mayr (1943. Condor, 45: 46) on the matter.--Editors. Dec. 1951 284 THE WILSON BULLETIN Vol. 63. No. 4 Evidence of Convergence from the Bill 3/4otch The bill notch is so widespread in passefines that one is forced to conclude that this is a general adaptation for grasping slippery prey and, therefore, not likely to persist in flower probers. In vireos the bill is strongly notched, its shrike-like cast reaching full expression in the tribe Cyclarhini. In tanagers, which may well have arisen from ancestral vireos, the bill is also strongly notched, becoming shrike-like in Lanio. But in the warblers, also apparently evolved from vireos, it is weakly notched or un-notched. In Coereba, Coni4 rostrum and A teleodacnis, here regarded as nectar-adapted warblers, the bill is warbler-like and in the nectar-adapted tanagers--Dacnis, Chlorophanes, Cyanerpes, Hemidacnis and Euneornis--the bill is tanager-like, generally. But in some of the latter it is doubtfully notched and there is so much individual variation that one cannot safely say a particular genus is notched or not. Again we find adaptive overlap of too-plastic characters, susceptible to obliteration by the suspected convergence, and it is plain that the bill-notch cannot be used to segregate warblers and tanagers in the Coerebidae. In sup- port of the view that the notch easily disappears in nectar-adapted tanagers we have the evidence that it has done so in some species of the nectar-adapted vireo, tJylophilus. I)ISCIJSSION Is submergence of the family Coerebidae and the assignment of its genera among the warblers and tanagers justified? Here it should be appreciated that --if both warblers and tanagers arose from vireos--the two families would no sooner have established distinctions through evolutionary divergence than they would, through adaptive convergence, have begun to lose them. The differences to be looked for will not, therefore, be very dramatic and convergent obliteration may be expected especially in the very plastic bill and tongue which fall strongly under selection pressure in nectar-feeding. The considerable within-the-genus differences of tongues in both the Paru- lidae and Thraupidae (Gardner, 1925) and of tongues and bills in the Icteridae (Beecher, 1950; 1951) suggest the unreliability of these characters. On the other hand, the constancy of the jaw muscle-pattern in the Icteridae, despite great modification of bill and tongue among the genera, and the conservatism of the pattern of the horny palate, inspire confidence. The latter characters have not only permitted the linking of tanagers to richmondenine and car- dueline finches but also of warblers to emberizine finches and the latter to the Icteridae. This is not the place to detail these relationships, but characters capable of retaining their essential features through such adaptive trans- formations of skull and bill can be used with assurance in the present problem. It is these highly dependable characters of jaw muscle-pattern and horny palate relief that accomplish the division of the genera of the Coerebidae among William J. Beechef CONVERGENCE IN THE COEREBIDAE 285 the warblers and tanagers--and this in good agreement with plumage pattern. The variable bill and tongue characters overlap adaptively. It is therefore concluded that, in view of the relatively meager diagnostic differences expected in convergent genera arising from groups as closely related as warblers and tanagers, the evidence is sufficient to justify submerging the family Coere- bidae. This seems entirely reasonable. Nectar-feeding has been acquired inde- pendently in many American passerine groups: vireos, warblers, tanagers, blackbirds and even finches--not to mention derived groups like the geospizids of the Galpagos and the drepanidids of the Hawaiian Islands. Various de- grees of adaptive modification for the trait are seen, not only in the bills and tongues but also in relative development of jaw muscles. In fact two distinct mechanical adaptations may be distinguished in the nectar-feeders of the Thraupidae and Parulidae and this may be regarded as an additional diagnostic difference. The nectar-adapted warblers are "gapers" (Beecher, 1950; 1951). As noted for Coereba in Figure 2, M. depressor mandibulae (M1) is highly developed in both Conirostrum and Ateleodacnis. The bill is apparently plunged into a flower closed, then opened forcefully to spread the flower parts and permit the tongue to reach the nectar--or possibly a hole is pecked in the side of a flower and enlarged in this manner. Functionally it is the same adaptation noted in the Icteridae and in the Old World Sturnidae, with a lever-like posterior extension of the mandible. In Figure 2 Coereba is compared with Opcrornis, which shows the same adaptation, though not all warblers are gapers in this degree. The Neotropical nectar-adapted tanagers are non-gapers. In none of them is M1 more highly developed than in Cyanerpes (Fig. 2), and it is apparent that these species simply insert the bill into flowers and sip nectar. Diglossa has been observed by Moller (1931: 292) to tear a hole in the side of a tubular corolla with its short, hooked bill to get at the nectar it could not otherwise reach. However, the tanager stock apparently can produce gapers. In the nectar- adapted drepanidids, Vestiaria coccinea (Fig. 5), ttimatione virens, and ttemig- natbus lucidus, the origins of protractors M1 and M2 are greatly expanded for gaping in tubular corollas, and the adductors M6 and M7 are greatly reduced. Otherwise these birds are very similar to Calospiza or Cyanerpes. That the drepanidids have a tanager origin is supported also by the identity of muscle- patterns in the Hawaiian finch, Psitlirostra cantans, and the cardueline finch, Carpodacus mexicanus. This may be interpreted as two independent origins of nearly identical finches from a common tanager ancestor. As for Spodiornis rusticus, Wetmore and Phelps (19a,9: 378) observed that "the form of the tail and of the tongue suggest affinities with the Coerebidae, ... the family in which P. L. Sclater originally described Spodiornis." The palate of Spodiornis suggests relationships with primitive emberizine finches Dec. 1951 286 THE WILSON BULLETIN Vol. 63, o. 4 like Phrygilus, however, and this may be one of those rare compromises--an intermediate between nectar-adapted warbler and emberizine finch. The highly angulated commissure of nectar-feeders, correlated with forward vision, is thought to pre-adapt them for seed-cracking before the bill becomes too long. This paper has concerned itself with the submergence of the family Coere- bidae as an unnatural category and the re-assignment of its genera among the warblers and tanagers. Some ornithologists will wish to retain this family name for some part of the group. There appears to be no valid reason for so doing, though the nectar-adapted warblers might be known as the tribe Coere- bini and the nectar-adapted tanagers as the tribe Dacnini. But nectar-feeding in the American nine-primaried assemblage should not be over-emphasized. Numerous references to the sucking of fruit and flower juices by warblers and tanagers never included in the Coerebidae could be cited. In the present state of our knowledge it would be difficult to know which ones to include in these tribes--unless we settle upon the genera dealt with above for the time being. Nectar-feeding is a general trait of the Thraupidae, Parulidae, Vireonidae, Icteridae, and even the polyphletic "Fringillidae." No useful purpose is served by giving it too much recognition in the classification of these higher categories. LITERATURE CITED BEECHEll, WILLIAM J. 1950 Convergent evolution in the American orioles. Wilson Bulletin, 62: 51-86. 1951 Adaptations for food-getting in American blackbirds. Auk, 68:411 440. FRIEDMANN, HERBERT 1946 Ecological counterparts in birds. S1/2i. Monthly, 63: 395-398. GARDNER, LEON L. 1925 The adaptive modifications and the taxonomic value of the tongue in birds. Pro1/2. U.S. Natl. Mug. 67: 1-49. GOSSE PHILIP HENRY 1847 The birds of Jamaica. John Van Voorst, London. GRANT, J. C. B. 1942 The musculature. Morris' human anatomy. Tenth edition. Blakislon, Phil- adelphia. HELLMAYR CtIARLES E. 1935 Catalogue of birds of the Americas. Field Mug. Nat. Hist., ZoOl. Set., 13, pt. 8. LUCAS, FREDERIC A. 1894 Notes on the anatomy and affinities of the Coerebidae and other American birds. Proc. U.S. Natl. Mug., 17: 299-312. MAYR, ERNST 1942 Systematics and the origin of species. Columbia Univ. Press, New York. MOLLER, WALTER 1931 Vorl[tufige Mitteilung fiber die Ergebnisse einer Forschungsreisc nach Costa Rica zu Studien fiber die Biologie bliltenbesuchender V6gel. Biologia Generalis, 7: 287- 312. PFUHL, WILIIELM 1937 Die gefiederten Muskeln, ihre Form und ihre Wirkungsweise. Zeitschrift f'tr A natomie und Entwickhmgsgeschichte, 106:749 769. RIDGWAY ROBERT 1902 The birds of North and Middle America. U.S. Natl. Mug. Bull. 50, pt. 2. SCLATER PttlLIP LUTLEY 1886 Catalogue of the birds in the British Museum. Vol. 11. London. SIMPSON, GEORGE GAYLORD 1944 Tempo and mode in evolution. Columbia Univ. Press, New York. 1945 The principles of classification and a classification of mammals. Bull. A met. Mug. Nat. Hist., 85: 1-350. gUSHKIN, PETER P. 1927 On the anatomy and classification of the weaver-birds. Bull. A mer. Ms. Nat. Itrist., 57: 1-32. WETMORE, ALEXANDER, AND WILLIAM H. PHELPS, JR. 1949 A new race of bird of the genus Spodiornis from Venezuela. Jour. Washington Acad. gel., 39: 377-378. CHICAGO NATURAL HISTORY MUSELrlVI