i7

A CRITIQUE OF CRACRAFT'S CLASSIFICATION OF BIRDS

STORRS L. OLSON National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560 USA

ABSTRACT.•A recently proposed "phylogenetic" classification of birds (Cracraft 1981b) is not constructed according to dadistic principles and contains little information to support most of the taxa proposed in it. That which is presented is frequently misleading or erro- neous. The nomenclature used is inconsistent and ungrammatical. In failing to provide synapomorphies to cluster taxa, in using data that are not presented in a primitive-derived sequence, in citing differences as evidence of nonrelationship, and using convergence to refute phylogenetic hypotheses, Cracraft commits the very methodological transgressions for which he has long criticized other systematists. Received 4 January 1982, accepted 24 April 1982.

IN the past decade or so, there has been great are postulated to be strictly monophyletic" controversy over methodology in systematics, (Cracraft 1981b: 682). occasioned by the rise of dadistics or "phylo- Division 1.•^This contains the orders Sphen- genetic systematics." Most of this debate has isciformes, Gaviiformes (including the Podi- taken place outside the field of ornithology, cipediformes and the Cretaceous toothed di- with the principal exception of a continuous vers of the order Hesperornithiformes), series of papers and reviev^^s by Cracraft (e.g. ProceUariiformes, and Pelecaniformes. Cracraft 1971, 1972, 1980, 1981a). Although these have (1981b: 686) states that "These four orders have primarily been discussions of methodology and often been placed near one another, although criticisms of the work of others, Cracraft (1981b) a strong argument for their interrelationships has now put forth a classification of the entire has not yet been made." Not a single synap- Class Aves that permits an examination of how onnorphy is advanced to justify "Division 1" as effectively he has applied his preferred meth- a monophyletic group. odology in actual practice. Division 2.•This contains the paleognathous ratites and tinamous. An increasing amount of CRACRAFT'S "DIVISIONS" evidence, including new data from paleontol- Because Cracraft (1981b; 681) considers that ogy (Houde and Olson 1981), suggests that the "most currently recognized orders and families few chiiracters that can be used to define this [of birds] are probably monophyletic," his group are primitive (plesiomorphous). Cracraft principal innovation would appear to be the (1980, 1981b) has specifically admitted this segregation of the Class Aves into nine major possibility but rationalizes the consequences categories, virhich he terms "Divisions." In- by saying that even if these characters "are deed, it is clear from his abstract that this is to primitive, that in itself does not constitute an be regarded as the most important aspect of his argument against monophyly but merely sig- classificahon. Yet these Divisions are not named nifies that the hypothesis of monophyly is not formally, "because in a number of cases their well corroborated" (Cracraft 1981b: 688). Not status as monophyletic groups is not yet well corroborated at all would be more accurate, be- documented" (Cracraft 1981b: 685). If no doc- cause, in a cladistic phylogeny, monophyletic umentation is to be provided for the focal point taxa are supposed to be defined by shared de- of his classification, then we might fairly ques- rived character states, without which there is tion from the outset why it was presented at no justification for Division 2. To cite neotenic all. Nevertheless, let us examine the basis for features of ratites as shared derived characters, Cracraft's "Divisions," bearing in mind that as Cracraft (1981b: 689) has done, is hardly likely the principal tenet of the cladist school is that to persuade nonomithologists of the mono- "Taxa are clustered on the basis of synapo- phyly of this group. By the same logic, all of morphies (Hennig 1966), and the taxa so formed the various neotenic species of salamanders 733 The Auk 99: 733-739. October 1982 734 STORRS L. OLSON [Auk, Vol. 99

(belonging to each of the eight famihes of Uro- . . . are placed together because a fossil is said dela) would form a monophyletic group. [emphasis added] to have a head only similar Division 3.•This category includes the Ci- to one taxon and a postcranial skeleton only coniiformes and the Falconiformes, the latter similar to the other taxon. The evidence of including the Strigiformes. One wiU search in comparative morphology and systematics sug- vain for a single synapomorphy that wiU define gests that mosaic evolution does not work as a group that contains both flamingos and owls absolutely as this" (Cracraft 1981b; 696). Thus, as monophyletic within the Class Aves. In- the concrete evidence offered by the fossils is stead, we are told that "The placement of such dismissed, because preconceptions about how disparate orders in the same Division may seem evolution works cannot accommodate it. Cra- unwarranted, and, admittedly, there is no clear craft offers no alternative explanation for the evidence for this .... Their placement in Di- relationships of Presbyornis and agrees with vision 3 is tentative and boldly hypothetical" Olson and Feduccia that a relationship be- (Cracraft 1981b: 690). No synapomorphies are tween Anseriformes and Galliformes "is not mentioned, and thus no evidence is presented well documented" (Cracraft 1981b: 695). that this weird taxon is monophyletic. It is de- Division 5.•In this category Cracraft in- cidedly misleading for Cracraft (1981b: 690) to cludes the Gruiformes, Charadriiformes, and cite Ligon (1967) as supporting the view that Columbiformes. He states that the interrela- the Ciconiiformes and Falconifonnes "may have tionships of these three orders "have not yet a relationship." Ligon argued for a close rela- been resolved satisfactorily" but that "there is tionship only between the Cathartidae and Ci- reason to maintain this association" because coniidae, which he placed in their own order "it stands as a working hypothesis" £md "be- (Ciconiiformes) apart from herons (Ardei- cause no alternative hypothesis of interordinal formes) and hawks (Accipitriformes). This is relationships seems better" (Cracraft 1981b: completely contrary to Cracraft, who regards 697). Once again, no synapomorphies are ad- the Ciconiiformes and Falconiformes each to vanced in support of monophyly of this group. be monophyletic. Division 6.•This includes only the order Division 4.•This contains the Galliformes and Psittaciformes, which most systematists con- Anseriformes. Arguments against a close re- sider to be composed of but a single family, lationship between these two orders have been Psittacidae. Never has it been doubted that the discussed elsewhere (Olson and Feduccia 1980a: parrots constitute a natural group, and it should 2). Practically the only morphological character have been possible for Cracraft to provide syn- that has been at all consistently cited as indi- apomorphies here to demonstrate monophyly, cating a possible relationship between Anser- but none is offered. Cracraft (1981b: 699) ig- iformes and Gcdliformes is the supposed sim- nores the evidence that suggests a possible close ilarity in the large, rounded "basipterygoid" relationship between the Psittaciformes and processes. Olson and Feduccia (1980a: 5) doc- Columbiformes (Sibley and Ahlquist 1972) and, umented that these structures are not really instead, says only that parrots "are so mor- similar between the two groups, that they have phologically distinct from other birds that their a different developmental history, and that they relationships to other groups have remained are not homologous. But Cracreift (1981b: 696) unresolved." Considering that he criticizes simply ignores this and continues to cite ba- other avian systematists for bestow^ing "high sipterygoid morphology as possibly indicating rank on taxa that are niorphologically divergent a relationship between Anseriformes and Gal- compared to their closest relatives" (p. 683), liformes. would it not have been preferable to reflect un- The fossil bird Presbyornis, which is known certainty about the relationships of parrots by from thousands of specimens, has, in essence, considering them incertae sedis, rather than the head of a duck and the body of a charad- raising the family Psittacidae to the same level riiform. It provides strong evidence for a der- as a taxon that contains penguins, loons, pel- ivation of the Anseriformes from the Charad- icans, grebes, and Hesperornis, among others? riiformes (Olson and Feduccia, 1980a). Cracraft, Division 7.•This division also consists of a however, seems to regard Presbyornis as a hy- single order, the Cuculiformes. Cracraft's pothetical creature and finds "it disquieting for (1981b: 699) justification for this taxon is that theoretical reasons , . . when two higher taxa "There is reasonably strong evidence for ac- October 1982] Critique of Phyhgenetic Classification 735 cepting a sister-group relationship between the Grues. This deficiency has still not been re- turacos (Musophagidae) and cuculids, but the medied, even though Cracraft (1981b) changed affinities of both families to other birds remain the composition of the taxon radically in the uncertain." What the "reasonably strong evi- meantime. Nor will one find any mention of dence" may be, or if it consists of synapomor- synapomoiphies that define the suborder Ralli, phies, is never stated. the infraorders Arami or Psophii [sic], or the Division 8.•The Caprimuigiformes and superfamilies Psophioidea or Rhynocheti [sic] Apodiformes are the constituents of this taxon. as monophyletic. Although Cracraft explains why he believes the Although at one point Cracraft (1981b: 691) Caprimuigiformes to be related to the Apodi- promises to "note some of the data that are dae and the Apodidae to the Trochilidae, he consistent with ciconiiform monophyly," no- does not present any synapornorphies that are where do I find a discussion of the supposed common to all members of this division and "characters, shared hierarchically, that are con- that would define it as monophyletic. sistent with ciconiiform monophyly" (p. 693). Division 9.•This is composed of the Pici- He states that flamingos, storks, and ibises formes, Coliiformes, Coraciiformes, and Pas- constitute a taxon that is the sister group of seriformes•a group of taxa for virhich "a pre- herons and Balaeniceps, but the only supposed cise hypothesis of . . . interrelationships has evidence for this is that "In terms of pure phe- not been supported" (Cracraft 1981b: 701). This netic resemblance they all have very similar is all that is said for the monophyly of Divi- humeri, pelves, and sterna. They all apparently sion 9. have the iliotrochantericus médius muscle poorly separated (or not at all) from the iliotro- chantericus anterior" (Cracraft 1981b: 692). But SUBORDINATE TAXA these are not presented as "an argument based If Cracraft's "Divisions" are without sub- on derived character sequences" such as Cra- stance, the same may be said of practically all craft (1972: 387) would have others do. One of his subordinate taxonomic categories. Par- need go no further than comparing the ster- ticularly egregious examples are the Grui- num of an ibis with that of Balaeniceps to dem- formes and Ciconiiformes. Apart from three onstrate the erroneousness of Cracraft's asser- families that are considered incertae seáis, Cra- tion that these and other bones of Ciconiiformes craft (1981b) divides the Gruiformes into two are "very similar." The iliotrochantericus méd- suborders: the Ralli, containing only the Ral- ius is one of a number of muscles that show lidae and Heliornithidae, and the Grues, con- no significant variation within Ciconiiformes taining the remaining families of the order. He (Vanden Berge 1970), where its condition has states (p. 697) that "The phylogenetic relation- not been shown to be different from that in ships of the suborder Grues have been docu- other birds (see also Olson and Feduccia 1980b: mented elsewhere (Cracraft 1973, MS)." The 17). contents of unpublished manuscripts [which Cracraft (1981b) repeatedly makes erroneous Cracraft (1981b) cites on 12 occasions] cannot or misleading statements and ignores pertinent be assessed, of course, but the 1973 reference data in discussing the supposed sister-group deals with a composition of the suborder Grues relationship between flamingos and storks, one entirely different from that now advocated by of few instances where he does provide some Cracraft (1981b). His Grues of 1981 contains, characters. He accuses Olson and Feduccia among others, the families Cariamidae, Rhyn- (1980b) of "overemphasizing differences" ochetidae, and Eurypygidae, which were not (Cracraft 1981: 692) between flamingos and Ci- included in his Grues of 1973, whereas the Ral- coniiformes, whereas in fact they consistently lidae, -which were included in his Grues of 1973, emphasized the many similarities between fla- are placed in a different suborder in the 1981 mingos and Charadriiformes. Cracraft's (1981b: classification. How can one concept of the sub- 692) statement that the hypotarsus of the fossil order Grues be justified by citing a paper con- flamingo Juncitarsus "is totally unlike all known taining a completely different concept of that phoenicopterid genera, both fossil and Recent" taxon? One of the criticisms made of Cracraft's is false, as it is actually very similar to the hy- 1973 paper (Olson 1974) was that characters potarsus in fossil flamingos of the genus Pa- were never provided to diagnose the suborder laelodus. This is an example of Cracraft himself 736 STOKRS L. OLSON [Auk, Vol. 99 citing differences between taxa as evidence of Cracraft's classification is not dichotomous nor nonrelationship, in contradiction to his own sequenced in any way in which a phylogenetic stated views (see below). tree could be consistently reconstructed from Olson and Feduccia (1980b: 42, fig. 18) dis- it. cussed and illustrated the fact that the super- The highest category in Cracraft's subclass ficial similarities between storks and flamingos Neornithes is the division; yet there are nine in the distal end of the tibiotarsus are not ho- of these, so it would be impossible to arrange mologous. Cracraft omits any reference to this, them dichotomously, even if categories be- however, and uses the deep intercondylar fos- tw^een the division and the subclass were pro- sa of the tibiotarsus as a character linking fla- vided, which they are not. There is only one mingos and storks. He fails to mention that the order in each of Divisions 2, 6, and 7, so in four-headed gastrocnemius, in addition to these cases an order is exactly equivalent to a being found in storks and flamingos, also oc- division and is therefore redundant. Division curs in the recurvirostrid Cladorhynchus, as 5 contains three orders, Vkrhich, in the absence clearly described and illustrated by Olson and of sequencing, obviously cannot be arranged Feduccia {1980b: 29, fig. 7a). Three more sup- dichotomously vihile being maintained at posed morphological similarities between equivalent rank. Division 9 contains four or- storks and flamingos that Cracraft cites are the ders but with no taxa between the division and lack of the peroneus brevis (= fibularis brevis), the order by which these four could be ar- the presence of a "short and stout" iliacus ranged in a dichotomous branching scheme. (= iliofemoralis internus), and the narrow (or Furthermore, it would seem to be impossible lack of) separation of the two heads of the pubo- to construct a dichotomous branching pattern ischio-femoralis. He does this in disregard of that contains only two taxa each at the level of the fact that Olson and Feduccia (1980b; 16, 24, "cohort," "infrasuborder," and "infrasuper- 29) show that each of these conditions is pres- family," but at the same time contains 26 sub- ent in Cladorhynchus. Thus, Cracraft has com- orders and 33 infraorders. Thus, although Cra- pletely misrepresented the evidence, and not craft (1981b: 682) tells us that phylogenetic one of the characters he cites in any w^ay refutes hypotheses are to be presented as cladograms, the charadriifonn relationships for flamingos it is not difficult to discern why cladograms proposed by Olson and Feduccia (1980b). were omitted from his own paper. For Cracraft to represent his as a cladistic, "phylogenetic" classification is dearly unjustified. THE INFORMATION CONTENT OF CRACRAFT'S CLASSIFICATION CRACRAFT'S NOMENCLATURE Succinctly stated, the factual basis for Cra- craft's (1981b) classification is nonexistent, as Even though Cracraft's classification is not virtually no data are provided to justify the dichotomous, it does contain the proliferation taxonomic categories used in it. Clearly, Cra- of higher taxa that such classifications neces- craft's discussions of the information content sitate. The names that are appended to these of classifications (Cracraft 1981b: 682) are of lit- taxa present many problems. TTie endings used tle relevance to his own classification, as it is for orders, families, subfamilies, and tribes, are obviously impossible to retrieve infonnation consistent, but among the superfamilies the from a system into which no information was name "Rhinocheti" should have been rendered put in the first place. as "Rhinochetoidea." For subordinal endings, One of the reasons that cladistic classifica- Wetmore (1960) used the plural of the generic tions should, in theory, lend themselves to in- stem, with which the ending is supposed to formation retrieval is that they are supposed to conform in gender. Cracraft has evidently at- be based optimally on strictly dichotomous tempted to form the endings not only for sub- branching sequences, with sister groups re- orders in this manner but for cohorts, infraor- ceiving the same taxonomic rank. Thus, even ders, infrasuborders, and infrasuperfamilies as though Cracraft does not present a cladogram well, with the result that there is no way to of the phylogeny of birds, we should be able distinguish one taxon from another by its end- to derive one from his classification, // it were ing. Furthermore, many of these endings do constructed according to cladistic principles. But not agree in gender with the generic stem and October 1982) Critique of Phyhgenetic Classification 737 in some cases the generic stem itself is incor- have not been organized into primitive-de- rect (e.g. "Scopiae," which would have to be rived sequences. If Cracraft does not take his derived from "Scopia" rather than Scopus). The own criticisms seriously, can others rightly be names "Ciconii," "Psophii," "Caprimulges," expected to do so? "Upupes," "Pitti," "Muscicapi," "Bombycil- Cracraft (e.g. 1972, 1981a,b) is reluctant to 11," "Sitti/' and "Fringilli" are among others attribute a significant role to convergent evo- that are ungrammatically constructed. In sev- lution in the history of birds. He has been crit- eral instances Cracraft has attempted to form ical of those who have invoked the argument two different names on the same stem in this of convergence in postulating relationships and manner, so at least one of the two names au- has repeatedly emphasized that differences be- tomatically has to be grammatically incorrect tw^een taxa do not constitute proof of nonrela- (e.g. Grues and Gruí, Arames and Arami, Al- tionship. Notwithstanding Cracraft's views, cedines and Alcedini). To make matters even convergent evolution is as prevalent among more confusing, he has interjected the ending birds as it is in other groups of organisms and "-morpha" [sic], apparently as the fancy struck his classification may encompass as many or him, because it occurs once as a subordinal more instances of convergence as do the alter- ending (Charadriomorpha) and twice among natives he seeks to supplant. Grebes would the infraorders (Tyrannomorpha, Passeromor- have to be convergent in their cervical myology pha). The carelessness of Cracraft's nomencla- with certain Gruiformes (Zusi and Storer 1969). ture alone is sufficient to make the adoption of Flamingos w^ould have to be convergent on the his classification inadvisable. Recurvirostridae in many aspects of their mor- phology and behavior (Olson and Feduccia 1980b). Ibises would have to be convergent with ADDITIONAL SHORTCOMINGS Charadriiformes and some Gruiformes in their Cracraft has been very outspoken in his crit- schizorhinal nostrils, occipital fontanelles, four- icism of the role of biochemical studies in elu- notched sternum, and other osteological char- cidating phylogenies because of their failure to acters (Olson 1979). Herons would have to be resolve data into primitive-derived sequences convergent with the Mesoenatidae in their very and because "phylogenetic affinity can be based distinctive pterylosis and larsal morphology solely on a recognition of shared character-states (Olson 1979). Parrots would be convergent with that were inherited (i.e. derived) from the pigeons in the morphology of their humérus common ancestor" (Cracraft 1971: 158). His (Wetmore 1926) and with cuckoos and Pici- criticism of Sibley's research on egg-white pro- formes in being zygodactyl. This list could be teins (Cracraft 1971) would seem to make his greatly extended. Cracraft has merely traded stand on this matter quite unequivocal. Al- one set of convergences for another. though Cracraft's objections to the egg-white Although Cracraft (1972, 1981a) criticizes protein data were based on arguable philo- others for citing differences as evidence of sophical grounds. Brush (1979) showed that nonrelationship and for invoking convergence problems with laboratory technique rendered to refute systematic hypotheses, he does the the egg-white electrophoretic data of dubious same himself. Take, for example, the flightless, systematic value, and Sibley himself now states two-toed running birds of the family Ergilor- that the egg-white analyses lead to what he nithidae that are knovirn from Tertiary fossils now regards as erroneous conclusions (Sibley from Asia. These have consistently been placed and Ahlquist 1980). Presumably, Cracraft has with the Gruiformes, yet they share clearly de- not forgotten his previous criticisms, and, be- rived characters of the tarsometarsus and toes cause he cites Brush's (1979) paper, we may with ostriches (Feduccia 1980). Cracraft (1973, assume that he was aware of its implications. 1981a), who also places the Ergilomithidae in Yet he uses the same studies of egg-w^hite pro- the Gruiformes, attributes the similarities be- teins (Sibley and Ahlquist 1972) to support his tween the Ergilomithidae and ostriches to con- phylogeny whenever it suits his purpose (Cra- vergence. As evidence of this, however, he craft 1981b: 699, 700). In addition, Cracraft merely lists 10 differences between the tarso- (1981b) cites as supporting his views a number metatarsi of ergilomithids and Struthio (Cra- of other papers that present biochemical, kary- craft 1973: 118), without discussing their phy- ological, or other nonmorphological data that logenetic significance. Is this not the very 738 STORRS L. OLSON [Auk, Vol. 99 approach that Cracraft (1972: 387) says "should has brought such controversy to systematics, be replaced by an argument based on derived has also created an environment in which un- character sequences"? substantiated speculation is not only condoned The similarities between the Shoebill (Ba- but encouraged. Yet before we will ever have laeniceps) and Pelecaniformes (Cottam 1957) are a phylogeny that reflects the probable evolu- attributed to convergence by Cracraft (1981b: tionary history of the Class Aves, someone will 687, 693) because of the facts that "characters have to do the work upon which such a phy- of the Shoebill do not fit into the hierarchical logeny must ultimately be based. pattern of derived characters shown by the pelecaniforms but that they do within the ACKNOWIEDGMENTS ciconiiforms" (p. 687), Yet Cracraft does not I am obliged to Alan Feduccia, Ernst Mayr, David present a "phylogenetic (primitive-derived W. Steadman, Richard P. Vari, and Richard L. Zusi sequence) analysis" (p. 687) for either the Pele- for criticisms of a substantially less concise draft of caniformes or the Ciconiiformes, and his clas- the manuscript. sification is therefore as deficient in this respect LITERATURE CTTED as the one by Cottam (1957) that he criticizes. In another instance, Cracraft (1981b; 691) even BRUSH, A. H. 1979. Comparison of egg-white pro- teins: effect of electrophoretic conditions. Bio- projects the argument of convergence into the chem. Syst. Evol. 7: 155-165. future: "I believe that the similarities between COTTAM, F. A. 1957. The petecanifornn characters flamingos and Cladorhynchus will eventually be of the skeleton of the Shoe-bill Stork, Balaeniceps interpreted as convergences." rex. Bull. British Mus. Nat. Hist. (Zool.) 5: 1-71. In the foregoing, w^e have seen that Cracraft CRACRAFT, J. 1971. Review of "A comparative study (1) has not clustered taxa on the basis of syn- of the egg-white proteins of passerine birds" by apomorphies, (2) has cited data that have not C. G. Sibley. Bird-Banding 42: 157-161. been resolved into primitive-derived se- . 1972. The relationships of the higher taxa quences, (3) has used differences between taxa of birds: problems in phylogenetic reasoning. as evidence of nonrelationship, and (4) has re- Condor 74: 379-392. . 1973. Systematics and evolution of the sorted to the argument of convergence to refute Gruiformes (Class Aves). 3. Fhylogeny of the phylogenetic hypotheses. In other words, he suborder Grues. Bull. Amer. Mus. Nat. Hist. 151: himself is guilty of exactly the things that he 1-128. has spent over a decade harshly criticizing oth- . 1980. Phylogenetic theory and methodol- er systematists for doing. ogy in avian paleontology: a critical appraisal. Contrib. Sei. Nat. Hist. Mus. Los Angeles Co. CONCLUSION 330; 9-16. . 1981a. Review of "The age of birds" by A. The one unequivocal message that has Feduccia. Syst. Zool. 30: 219-221. emerged from recent arguments is that not all 1981b. Toward a phylogenetic classification of the recent birds of the world (Class Aves). systematists are willing to accept one and the Auk 98; 681-714. same methodology. Mayr and others (see Mayr FEDUCCIA, A. 1980. The age of birds. Cambridge, 1981 and references cited therein) have pre- Massachusetts, Harvard Univ. Press. sented counterarguments and alternatives to HENNIG, W. 1966. Phylogenetic systematics. Ur- cladistics, for example. Dissent could continue bana, Illinois, Univ. Illinois Press. indefinitely without a consensus being reached. HouDE, p.. Se S. L. OLSON. 1981. Paleognathous Overlooked in the tumult is the basic fact that carínate birds from the early Tertiary of North advances in avian systematics are the result of America. Science 214; 1236-1237. hard, often tedious work, whether it entails LiGON, J. D. 1967. Relationships of the cathartid breaking up rocks and grinding out fossils, vultures. Occ. Pap. Mus. Zool. Univ. Michigan 651: 1-26. dissecting tiny muscles for hours under a mi- MAYR, E. 1981. Biological classification: toward a croscope, skinning birds and counting feathers synthesis of opposing methodologies. Science in pterylae, or other laborious procedures. Only 214: 510-516. through the accumulation of more and more OLSON, S. L. 1974. Review of "Systematics and data in this manner will our understanding of evolution of the Gruiformes (Class Aves). 3. avian evolution be promoted. Unfortunately, Phylogeny of the suborder Grues" by J. Cracraft. the era of Hennig, Popper, and Croizat, which Auk 91: 862-865. October 1982] Critique of Phylogenetic Classification 739

. 1979. Multiple origins of the Ciconiifor- as indicated by DNA x DNA hybridization. mes, Proc. Colonial Waterbird Group, 1978:165- Proc. 17th Intern. Omithol. Congr. : 1215-1220. 170. VANDEN BERGE, J. C. 1970. A comparative study , & A. FEDUCCIA. 1980a. Presbyornis and the of the appendicular musculature of the order Ci- origin of the Anseriformes (Aves: Charadriom- coniiformes. Amer. Midi. Natur. 84: 289-364. orphae). Smithsonian Contrib. Zool. 323: 1-24. WETMORE, A. 1926. Descriptions of additional fos- , & . 1980b. Relationships and evo- sil birds from the Miocene of Nebraska. Amer. lution of flamingos (Aves: Phoenicopteridae). Mus. Novitates 211: 1-5. Smithsonian Contrib. Zool. 316: 1-73. . 1960. A classification for the birds of the SiDLEY, C. G., & J. AHLQUIST. 1972. A comparative world. Smithsonian Misc. Coll. 139: 1-37. study of the egg-white proteins of non-passerine Zusi, R. L., & R. W. STOKER. 1969. Osteology and birds. Peabody Mus. Nat. Hist., Yale Univ., BuU. myology of the head and neck of the Pied-billed 39: 1-276. Grebes (Podilymbus). Misc. Pub!. Mus. Zool. , & . 1980. The relationships of the Univ. Michigan 139: 1^9. "primitive insect eaters" (Aves: Passeriformes)