A phylogenetic analysis of all Recent genera of the Anseriformes using 120 morphological characters supports much of the current consensus regarding intraordinal relationships. I found that (1) Anseranas should be placed in a monotypic family; (2) Dendrocygna, Thalassornis, geese and swans, and Stictonetta are paraphyletic to the rest of the Anatidae; (3) Cereopsis is the sister group to Anser and Branta, and Coscoroba is the sister group to Cygnus and Olor; (4) Plectropterus is the sister group to the Tadorninae (shelducks) and the Anatinae (typical ducks); (5) the shelducks are monophyletic and include Sarkidiornis (provisionally), Malacorhynchus, Hymenolaimus, Merganetta, and Tachyeres; (6) the tribe "Cairinini" ("perching ducks") is an unnatural, polyphyletic assemblage and is rejected; (7) the dabbling ducks (including the smaller "perching ducks") comprise an unresolved, probably paraphyletic group; (8) tribal monophyly of the pochards (including Marmaronetta and Rhodonessa), sea ducks (including the eiders), and stiff-tailed ducks (including Heteronetta) is confirmed; and (9) the retention of Mergellus and resurrection of Nomonyx are recommended based on clarifications of intratribal relationships. Problematic groups, effects of homoplasy, phenetic comparisons, life-history correlates, biogeographic patterns, and fossil species are discussed, and a phylogenetic classification of Recent genera is proposed. Received 18 November 1985, accepted 2 April 1986.

Museum of Natural History and Department of Systematics and Ecology, University of Kansas, Lawrence, Kansas 66045 USA THE order Anseriformes is considered to comprise the families Anhimidae (2 genera, 3 species) and Anatidae (approximately 43 gen- era and 150 species). The family Anatidae is undoubtedly one of the best-studied groups of birds, owing largely to the historical impor- tance of waterfowl for hunting (Weller 1964a), domestication (Delacour 1964a), and aviculture (Delacour 1964b). The classification of the Anatidae proposed by Delacour and Mayr (1945) has been fol- lowed, with only minor revisions, in recent de- cades (e.g. Delacour 1954, 1956, 1959, 1964c; Johnsgard 1961a, 1962, 1965a, 1978, 1979; Woolfenden 1961; Frith 1967; Bellrose 1976; Palmer 1976; A.O.U. 1983; Bottjer 1983; Scott 1985). Perhaps the most innovative aspect of this system (inspired by the works of Salvadori 1895; Phillips 1922, 1923, 1925; and Peters 1931) was the erection of "tribes," groups of genera that were considered to be closely related with- in the subfamilies of the Anatidae. These tribes became the primary focus of subsequent works on anatid classification, many of which ad- dressed the tribal assignments of problematic genera (e.g. Humphrey and Butsch 1958; Johnsgard 1960a, 1961b; Humphrey and Ripley 1962; Davies and Frith 1964; Raikow 1971; Kear and Murton 1973). Most authors assumed the validity of the tribes and used them as working units in phylogenetic analyses of the family (e.g. Johnsgard 1961a, Bottjer 1983). A few workers named additional tribes (Moynihan 1958, Delacour 1959, Woolfenden 1961, Weller 1968b) or attempted to test the naturalness of those originally proposed (Cotter 1957, Wool- fenden 1961, Brush 1976). Behavioral characters have been accorded considerable weight in classifications of water- fowl. Delacour and Mayr (1945) based their revision on characters they considered to be "non-adaptive," including behavioral displays, nesting and feeding habits, and selected mor- phological characters (e.g. posture, body pro- portions, head shape, syringeal bulla). Reliance on comparative ethology in anatid systematics was furthered by the studies of Lorenz (1951- 1953), McKinney (1953), and Myres (1959) and was increased significantly by Johnsgard (1960a-c, 1961a-d, 1962, 1964, 1965a, b, 1966a, b, 1967, 1978), whose work was largely etho- logical and influenced profoundly by that of Delacour (1954, 1956, 1959, 1964c). This em- phasis, work on interspecific hybridization (Sibley 1957; Gray 1958; Johnsgard 1960d, 1963), and study of plumage patterns of downy young (Delacour 1954, 1956, 1959; Frith 1955, 1964b; Kear 1967) were prompted in part by the op- portunity to observe waterfowl in avicultural collections. Other data used in the classification of wa- terfowl include syringeal anatomy (Humphrey 1955, 1958; Johnsgard 1961e), cytogenetics (Ya- mashina 1952), serology (Cotter 1957, Bottjer 1983), osteology (DeMay 1940, Verheyen 1955, Humphrey and Butsch 1958, Woolfenden 1961, Humphrey and Ripley 1962, Raikow 1971), feather lice (Timmermann 1963), eggshell structure (Tyler 1964), egg-white proteins (Sib- ley 1960, Sibley and Ahlquist 1972), feather proteins (Brush 1976), royology (Zusi and Bentz 1978), lipids from the uropygial gland (Jacob and Glaser 1975, Jacob 1982), and mitochon- drial DNA (Kessler and Avise 1984). These studies, with the possible exceptions of those by Lorenz (1953) and Kessler and Av- ise (1984), estimated the evolutionary relation- ships of groups by assessments of overall sim- ilarities; no attempts were made to determine primitive conditions or to distinguish shared primitive characters from shared derived char- acters ("special" similarity). Moreover, the "evolutionary trees" presented in most of these works lack references to the specific characters used to support the branching patterns (e.g. Delacour and Mayr 1945; Johnsgard 1961a, 1978; Woolfenden 1961). I performed a phylogenetic (cladistic) anal- ysis of Recent genera of Anseriformes using 120 morphological characters. I present a hy- pothetical evolutionary tree for the order, con- sider the taxonomic implications, and discuss selected life-history and biogeographic corre- lates and the classification of selected fossil species. Many of the characters were described first in the pioneering work of Woolfenden (1961), to whom I dedicate this paper. METHODS Taxa and specimens.--Both genera of Anhimidae and all Recent genera of Anatidae were studied. I ana- lyzed separately several subgenera (sometimes con- sidered genera), including Olor, Lophonetta, Pteronet- ta, Amazonetta, Callonetta, Mergellus, Lophodytes, and Nomonyx. Several other "subgenera" were found to be identical to the taxa with which they generally are merged and are not labeled separately in the trees: Casarca (= Tadorna), Metopiana (= Netta), Oidemia (= Melanitta), and Charitonetta (= Bucephala). For Anas, species from several subgenera were examined: Mal- lard (Anas platyrhynchos), Northern Pintail (A. acuta), American Wigeon (A. americana), Green-winged Teal (A. crecca), and Northern Shoveler (A. clypeata). Oth- er species of Anas were studied for certain characters. Salvadori's Duck [Anas (Salvadorina) waigiuensis], pro- visionally assigned to Anas but considered problem- atic by some (Mayr 1931, Kear 1975), was not includ- ed because no skeletal specimens were available (Wood et al. 1982). Except for Rhodonessa (monotypic, probably extinct; one complete skeleton) and Camp- torhynchus (monotypic, extinct; casts of two partial skeletons), all genera analyzed were represented by at least two complete skeletons. For all polytypic genera at least two species were studied, and a num- ber of the common, diverse, or problematic genera were represented by large series. For Camptorhynchus, character states for unavail- able elements either were assumed provisionally (for characters invariant within the anatines) or coded as "missing." Assumption of anatine characters for Camptorhynchus is conservative (cf. Humphrey and Butsch 1958, Zusi and Bentz 1978) and did not alter its position in the resultant tree (compared with anal- yses without this assumption), but permitted more efficient computation of trees and a shorter final so- lution. Analysis of characters.--For the phylogenetic anal- ysis presented, 120 characters were used (Appendix 1); a majority of the osteological characters were de- scribed in Woolfenden (1961) and illustrated in Howard (1929). Some characters were rejected be- cause variation prevented even modal state assign- ments for some genera or because discrete states could not be distinguished. Sources for data on the postcranial skeleton were Wetmore (1951), Rand (1954), Verheyen (1955), Woolfenden (1961), Humphrey and Clark (1964), and Raikow (1971). Additional sources were (by anatom- ical region): integument and molt (DeMay 1940, Siegfried 1970, Palmer 1976), trachea and syrinx (Wetmore 1926; Niethammer 1952; Wolff and Wolff 1952; Humphrey 1955, 1958, unpubl. data; Hum- phrey and Butsch 1958; Johnsgard 1961b, e; Hum- phrey and Ripley 1962; Humphrey and Clark 1964; Weller 1968b), and skull (Abbott 1938, Harrison 1958, Raikow 1970a, Olson and Feduccia 1980a). I included only qualitative characters because the polarities and states of mensural characters are especially difficult to determine. Each character is an anatomical trait for which two or more discrete character states were de- fined. Derivation of trees.--Polarities of each character (primitive states) were determined by comparison with outgroups--Burhinus and Larus (Charadri- iformes), Ortalis and Meleagris (Galliformes), Ciconia (Ciconiiformes), and Phoenicopterus (Ciconiiformes or Charadriiformes)--each of which has been proposed as closely related to the Anseriformes (Delacour and Mayr 1945; Delacour 1954; Mainardi 1962; Simonetta 1963; Sibley et al. 1969; Sibley and Ahlquist 1972; Brush 1976; Feduccia 1977, 1978; Olson and Feduccia 1980a, b; Olson 1985). Generally, the galliforms were most useful for establishing polarities. Outgroups were used to construct a hypothetical ancestor (a vec- tor of primitive character states) for the Anseri- formes, which was used to root the evolutionary tree; the primitive condition of nine characters could not be determined and were coded as missing (Appendix 1). Transformation series were treated as linear un- less they appeared to be nonlinear or problematic; the latter were treated as unordered (Appendix 1). The syringeal bulla was given a weight of 2 because it is a locomotion-independent character complex in- volving enlargement, symmetry, and fenestration; all other characters were assigned unit weight. The logic and terminology of phylogenetic analy- sis are discussed in Wiley (1981). The tree was de- rived using the PAUP program (Swofford 1984), a program that seeks trees of maximum parsimony (i.e. requires the least number of character-state changes; see Kluge 1984) and that permits the examination of series of "equally short" trees. The large size of the data set prohibited an exhaustive search guaranteed to find the shortest tree(s), but two thorough meth- ods-alternate and global branch swapping--were employed and produced identical topologies. A data matrix for all outgroups and the Anseri- formes and a list of specimens examined are available from the author on request. RESULTS General findings.--Of 50 equally short trees examined that resulted from minor changes in character distributions, only three distinct to- pologies were found. The tree illustrated (Fig. 1) has the topology of 46 of these trees (consis- tency index = 0.59). Two trees reversed the or- der of branching of Thalassornis with the geese and swans, and two others altered relation- ships in the goldeneye-merganser clade (Bu- cephala, Mergellus, Lophodytes, Mergus). Anhimids and Anseranas.--Monophyly of the Anhimidae and the waterfowl and the early branching of Anseranas were confirmed (Fig. 2). The "primitive" status of Anseranas has been recognized widely (Miller 1919; Boetticher 1943; Delacour 1954; Johnsgard 1961c, e, 1962, 1978, 1979; Woolfenden 1961; Olson and Feduccia 1980a). Notable exceptions were the placement of Anseranas with the superficially similar Plec- tropterus (Peters 1931, Delacour and Mayr 1945) and the proposition that Anseranas is an aber- rant "true goose" (Davies and Frith 1964, Frith 1967). Both the anhimids and Anseranas have undergone substantial autapomorphic change since divergence, much of which is unique in the order. Geese, swans, and proto-ducks.--! found that the "Anserinae," as currently defined (e.g. Johnsgard 1978), is paraphyletic to the rest of the family (Fig. 2), in contrast to the monophy- ly depicted by Delacour and Mayr (1945), Boet- ticher (1952), and Woolfenden (1961). Johns- gard (1961a, e) depicted the group as paraphyletic to the "Anatinae," but later (1978) diagrammed it as monophyletic, as did Bellrose (1976). These variations, however, may reflect different approaches to tree construction as much as changing perceptions of relationships. The branching sequence (Fig. 2) differs from conventional schemes (Johnsgard 1967, 1978; Kear 1967; Raikow 1971; Brush 1976) in that Dendrocygna and Thalassornis are not sister gen- era but instead comprise a grade. !n an equally parsimonious topology, Thalassornis diverged immediately after the goose-swan branch. Most of the 13 autapomorphies in Thalassornis rep- resent adaptations for diving and include sev- eral convergences with diving ducks in other clades. Until the works of Johnsgard (1967) and Raikow (1971), Thalassornis was considered to be an aberrant stiff-tailed duck and allied with Oxyura (e.g. Peters 1931; Delacour and Mayr 1945; Delacour 1959, 1964c), a treatment re- peated recently (Howard and Moore 1984, Scott 1985). Ten character changes confirmed the mono- phyly of the geese and swans (Fig. 2), as hy- pothesized by most authorities in recent de- cades (Delacour and Mayr 1945; Delacour 1954; Johnsgard 1961a, e, 1965a, 1978; Woolfenden 1961; Bottjer 1983). Also, there is a sister-group relationship between Coscoroba and the "typi- cal" swans (Cygnus, Olor; 6 characters) and be- tween Cereopsis and the "typical" geese (Anser, Branta; 2 characters). My analysis demonstrated monophyly of Olor, but no apomorphies distin- guished Cygnus from the common ancestor of Cygnus and Olor (Fig. 2); hence, the topology of Cygnus remains unresolved. The generic monophyly of Anser and Branta also was not established. Traditionally, Coscoroba and Cer- eopsis have been considered to be either "links" ANCESTOR Chauna Anhima PART 1 Anseranas Dendrccygna Thalassornis Cygnus Coscoroba Anser Branta Cereopsis plectropterus Sarkidiornis pART 2 Tadorna Malacorhynchus Necchen Chloephaga Alopochen Cyanochen Tachyeres Merganetta pleC%aimus Pteronetta Lophonetta Nettapus Ana$ - Camionefta - Chenonetta - Amazonetta Marmaronetta Rhodonessa -- Netta PART Aythya Somateria Camptorhynchus -Melanitta Clangula Mergellus Bucephala -Lophodytes PART 4 Nomonyx Biziura Fig. 1. Phylogenetic tree of Recent anseriform genera and selected subgenera based on 120 morphological characters listed in Appendix 1. Lengths of horizontal lines correspond to the number of character changes (apomorphies) in the lineages. Sections of the tree are detailed in Figs. 2-5. between tribes, early branches from the com- mon ancestor to the "true" geese and swans, or, for Cereopsis, an aberrant shelduck or the sole member of a separate tribe (Peters 1931; Delacour and Mayr 1945; Delacour 1954, 1964c; Johnsgard 1961a, e, 1978; Woolfenden 1961; Frith 1967; Kear and Murton 1973; Bottjer 1983). My analysis showed Stictonetta to be the last branch in the grade of waterfowl with reticu- late tarsi (Figs. I and 2). Stictonetta long was believed to be an aberrant member of the shelducks (near Tadorna; Peters 1931, Boettich- er 1952) or the dabbling ducks (e.g. Anas; De- lacour and Mayr 1945; Delacour 1956, 1964c). Based on anatomical comparisons, however, other workers suggested that the genus was de- rived from an earlier "anserine" branch of the Anatidae (Verheyen 1955; Johnsgard 1960c, 1961a, b, 1962, 1965a, b, 1978; Woolfenden 1961; Frith 1964a, b, 1967; Brush 1976; Olson and Feduccia 1980a). Plectropterus and the shelducks.--I found that Plectropterus is not related closely to the "perching ducks" (e.g. Sarkidiornis, Cairina, and Nettapus) as generally recognized since Dela- cour and Mayr (1945). Instead, Plectropterus is the earliest branch of the waterfowl with scu- tellate tarsi, and lacks five synapomorphies uniting more-derived members of the family (Fig. 3). Woolfenden (1961) concluded that Plectropterus was most similar osteologically to the shelducks, and Tyler (I 964) found that egg- 52- 55- 64- 65 78- 53 a-b 58 a -b m J= 70 a -b 86 I a-b 105 a-c 17 a-b 94 a-b 5 a-b I 0 a -b 4 o -b 31 a -b 9 a -b 40 a-b 12 a-b 42 a-b 18 a-b 50 a -b 88 a-b 91 a -b ] O0 a -b ]16 a-b 106 a-b I05 b-c o 85 a-b 78 79 89 92 93 96 99 I09 120 74 a-b 4 b-c 45 - a-b 63 - a-b 68 - a-b 72 - a-b 77 - a-b b-a 38 + a-b 33 a-b 79 c -d 23 a-b 43 a-b Fig. 2. Detailed diagram of Part 1 of the phylogenetic tree of the Anseriformes shown in Fig. 1. Characters are listed in Appendix I. shells of Plectropterus were intermediate be- tween those of anserines and shelducks in structure. Bottjer (1983) found that Plectropterus differed greatly from other "perching ducks" serologically but attributed the result to exper- imental error. My analysis supports the monophyly of the shelducks, although only by a single, possibly convergent character; further study may show the shelducks to be a grade of relatively prim- itive ducks. The clade includes a polytomy in- volving Tadorna, Malacorhynchus, and two clades containing three genera each (Fig. 3). I found Malacorhynchus to be a highly derived shel- duck; this genus has been assigned most fre- quently to the dabbling ducks (e.g. Delacour 1956, Woolfenden 1961, Frith 1967, Johnsgard 1978), although Frith (1955) noted that the pat- tern of its downy young differed greatly from those of Anas and Brush (1976) found that Ma- lacorhynchus was distinctly different from Anas in its feather proteins. A single synapomorphy Fig. 3. Detailed diagram of Part 2 of the phylogenetic tree of the Anseriformes shown in Fig. 1. Characters are listed in Appendix 1. Graphical proximity of branches within the polytomy in the shelducks does not reflect relatedness. of the skull supports the monophyly of the three genera of "sheldgeese"--Alopochen, Chloephaga, and Neochen (Fig. 3). The Comb Duck (Sarkidiornis melanotos), tra- ditionally placed in the "Cairinini" (Delacour and Mayr 1945, Johnsgard 1978), appears to be an early branch of the shelducks. However, three humeral characters that were important in distinguishing the "anserines" and shel- ducks from the "anatines" [capital shaft ridge (character 22), deltoid crest (25), and external tuberosity (32)] were of equivocal or "inter- mediate" condition in this species. In addition, the enlarged, uniquely distally directed meta- carpal I of Sarkidiornis, a character complex herein considered to incorporate two aspects [orientation (41) and length (42)], proved dif- ficult to characterize. Perhaps Sarkidiornis branched immediately before Cairina or, alter- natively, after Plectropterus but before the di- vergence of the shelducks from other "ana- tines." Three problematic genera--Hymenolaimus, Merganetta, and Tachyeres--comprise a highly derived clade of shelducks (Figs. 1 and 3). Sev- eral of the characters uniting these genera are evidently related to diving and are shared by diving ducks in other clades (see Discussion). Hymenolaimus and Merganetta have been treat- ed as allied either with the shelducks, "perch- ing ducks," or as exceptional, possibly primi- tive dabbling ducks (Delacour and Mayr 1945, 1946; Delacour 1956; Ripley 1957; Johnsgard 1965a, 1966a). Some workers placed Merganetta in its own tribe, Merganettini (Woolfenden 1961; Kear and Steel 1971; Kear 1972, 1975; Brush 1976; Johnsgard 1978). My analysis does not support the suggestion (Olson and Feduc- cia 1980a: 22) that "... the typical members of the 'subfamily' Anserinae and the typical mem- bers of the 'subfamily' Anatinae are more closely related to one another than to Stictonet- ta, Malacorhynchus, or Merganetta." The third and most derived member of this clade is Tachyeres (Fig. 3), a neotropical genus generally placed in the shelducks or in a sep- arate tribe allied with the shelducks (Delacour 1954; Moynihan 1958; Johnsgard 1965a, 1978; Weller 1976; but see Ripley 1957, Woolfenden 1961). Like all shelducks, Merganetta and Tachy- eres show an enlargement of metacarpal I (adorned with keratin spurs in Merganetta) that is sexually dimorphic and age related (Weller 1968a, Livezey unpubl. data). Cyanochen is hy- pothesized to be the sister genus to the Hymen- olaimus-Merganetta-Tachyeres clade, although this relationship is supported by only two syn- apomorphies. Bottjer (1983) suggested that Cyanochen may have branched before the other shelducks. "Perching" ducks and "dabbling" ducks.--The sister group to the shelducks comprises four groups (Fig. 1): a poorly resolved grade of "perching" and "dabbling" genera (Pteronetta, Cairina, Aix, Lophonetta, Nettapus, Anas, Callonet- ta, Chenonetta, and Amazonetta), which in turn gave rise to the pochards and independently to the sea ducks and stiff-tailed ducks. The first group (Fig. 4), henceforth termed "dabbling ducks," is a paraphyletic group of genera pre- viously allocated to either the "Anatini" or the "Cairinini" (Delacour and Mayr 1945; Delacour 1956; Johnsgard 1960c, d, 1961a, e, 1962, 1965a, 1978). The polyphyletic character of the "Cairinini" was inferred by Woolfenden (1961). The tribe has been recognized by subsequent workers in spite of the equivocal allocation of several gen- era (e.g. Callonetta and Amazonetta; Johnsgard 1960a, 1965a, 1978), the widely recognized het- erogeneity of its members in behavior, mor- phology, and biochemistry (Johnsgard 1960c, 1961a, 1962, 1965a, 1978; Woolfenden 1961; Ty- ler 1964; Brush 1976; Bottler 1983), the lower incidence of interspecific hybridization within the tribe than between its members and those of other tribes (Johnsgard 1960d), and the con- spicuous lack of a single character (or combi- nation of characters) that uniquely distinguish- es its members from other anatines. Johnsgard (1965a, 1978) admitted that retention of the tribe was partly a taxonomic convenience to avoid creation of "a comparatively large tribe" (1978: xxi) and omitted it as a suprageneric taxon in his latest list (Johnsgard 1979). The genera of dabbling ducks (on the basis of three variable multistate characters) form a grade from relatively primitive (e.g. Cairina, Lo- phonetta) to more-derived forms (e.g. Anas, Cal- lonetta). A single osteological synapomorphy supports a close relationship between Cairina and Aix (Fig. 4), a relationship suggested pre- viously by karyotypic and serological compar- isons (Yamashina 1952, Cotter 1957, Bottjer 1983). Pochards.--The pochards are a monophyletic group in an unresolved polytomy that involves 28 a -b 55 56  33 b-c  97 o-b  Fig. 4. Detailed diagram of Part 3 of the phylo- genetic tree of the Anseriformes shown in Fig. 1. Characters are listed in Appendix 1. Graphical prox- imity of branches within the polytomy in the dab- bling ducks does not reflect relatedness. Anas, Callonetta, Chenonetta, Amazonetta, and the sea ducks and stiff-tailed ducks (Fig. 4), sug- gesting that the pochards arose independently of other diving ducks. I found that Rhodonessa is the sister group to Netta and Aythya, which agrees with most studies (Verheyen 1955; Johnsgard 1961a, e, 1962, 1978, 1979; Woolfen- den 1961; Humphrey and Ripley 1962; Brush 1976) since Delacour and Mayr (1945, 1946) and Delacour (1956) provisionally placed Rhodones- sa in the dabbling ducks. Marmaronetta, a genus believed to "link" the Anatini with the po- chards but retained within the Anatini (Johns- gard 1961a, b, e, 1978; Delacour 1964c; Brush 1976), is supported in my study as the sister genus to the pochards by two osteological syn- apomorphies (Fig. 4). This relationship is cor- roborated by the secondary loss of metallic col- oration in the speculum (Delacour and Mayr 1946). Sea ducks.--The sea ducks comprise a mono- phyletic group related to the stiff-tailed ducks, although this relationship is supported by only a few, possibly convergent characters related to diving (Fig. 5). My result contradicts the pre- viously proposed close relationship between sea ducks and "perching ducks" (Delacour and Mayr 1945, Delacour 1959, Bottler 1983). With the possible exception of a few workers who advocated (largely on the retention of primi- tive Anas-like characters) the tribal separation of the eiders (Somateria and Polysticta) from the other sea ducks (Humphrey 1955, 1958; Dela- cour 1959; Brush 1976; Todd 1979), the mono- phyly of the sea ducks has not been questioned recently (e.g. Johnsgard 1960b, 1961a, e, 1964, 1978; Woolfenden 1961; Bottjer 1983). The proposed sequences of genera within the group has varied (e.g. Delacour and Mayr 1945; Delacour 1959; Johnsgard 1960b, 1961a, 1965a, 1978, 1979). My analysis (Fig. 5) indicates that Polysticta, Somateria, Histrionicus, and Campto- rhynchus comprise a less specialized, basal grade of genera retaining primitive, unfenestrated syringeal bullae. This series of genera gives rise to a well-supported clade of, in order of in- creasing relatedness, Melanitta, Clangula, and the goldeneye-merganser clade. The eiders (Poly- sticta, Somateria) appear to be paraphyletic to the other sea ducks; this paraphyly is support- ed only weakly, and downy patterns suggest that the eiders may be monophyletic (see Dis- cussion). Placement of Camptorhynchus must re- main tentative because of the limited material available. Humphrey and Butsch (1958) placed Camptorhynchus after Melanitta but before Clan- gula, and Zusi and Bentz (1978) allied the genus with eiders, evidently on the basis of shared primitive characters. The very close relation- ship of goldeneyes and mergansers has had unanimous support in recent decades (e.g. De- lacour and Mayr 1945; Boetticher 1952; Hum- phrey 1955; Delacour 1959; Myres 1959; Johns- gard 1960b, d, 1961a, 1978; Brush 1976). This analysis shows the Sinew (Mergellus al- bellus) to be either the sister genus to Bucephala (Fig. 5) or the sister group to the Lophodytes- Mergus clade. The former topology is corrobo- rated by the relatively high frequency of Mer- gellus x Bucephala hybrids in the wild (Phillips 1925, Ball 1934, Gray 1958, Nilsson 1974, Johns- gard 1978). Previous workers either listed the Sinew between the goldeneyes (Bucephala) and the mergansers (Mergus, Lophodytes) as a mono- typic genus (Peters 1931, Woolfenden 1961, A.O.U. 1983), or merged it (with Lophodytes) into Mergus (Delacour and Mayr 1945; Boettich- er 1952; Humphrey 1955; Delacour 1959, 1964c; Johnsgard 1960c, 1961a, d, 1965a, 1978, 1979). Stiff-tailed ducks.--My study supports the monophyly of the stiff-tailed ducks, wherein Heteronetta is the sister genus to the more typ- ical members (Fig. 5). The position of the clade as closely related to the sea ducks and highly derived (especially Biziura) agrees with recent orderings of genera by taxonomists (e.g. Johns- gard 1979) but disagrees with suggestions of a pre-dabbling duck (Raikow 1970b, Johnsgard 1978) or pre-shelduck (Johnsgard 1965b, Bottjer 1983) origin for the group. A few workers have expressed doubts about the relationships of Heteronetta (Johnsgard 1960c, Brush 1976), and others accepted the relationship between Het- eronetta and other stiff-tailed ducks but sug- gested that the genus be accorded tribal rank (Weller 1967, 1968b; Rees and Hillgarth 1984). My analysis shows that the Masked Duck [Nomonyx (Oxyura) dominica] is the sister group to the highly derived Oxyura-Biziura clade (Fig. 5), i.e. Oxyura is related more closely to Biziura than to the very similar Nomonyx. This topol- ogy is supported as well by the derived loss of a speculum in Oxyura and Biziura, which is re- tained in Nomonyx (Delacour 1959). This find- ing supports the resurrection of Nomonyx as advocated by Woolfenden (1961), a recom- mendation rejected by Delacour (1964c), Johns- gard (1967), and most subsequent workers. I did not examine skeletons of all species of Oxy- ura, so monophyly of the genus was not es- tablished with certainty. DISCUSSION Diving habit and homoplasy.--Considerable homoplasy (convergence) of characters is shown in the tree (Figs. 1-5) and by the consistencies of characters (Appendix 1). The majority of the convergences are associated with adaptations for diving, and most involve the leg elements (characters 52, 55, 56, 64, 65, 69, 75), pelvis (119), and skeletal pneumaticity (28, 78). These fea- tures tend to co-occur, especially within ele- ments. Convergence between Thalassornis and the stiff-tailed ducks is particularly pervasive (Figs. 2 and 5). It appears, however, that the moderately large number of characters includ- ed in this analysis reduced the impact of such homoplasy on the resultant tree, although deletions or heavy weighting of selected char- acters can produce topological changes. For ex- ample, because of a number of diving-related homoplasies, postulation of the Hymenolaimus- Merganetta-Tachyeres clade as the sister group to the sea ducks and stiff-tailed ducks is only slightly less parsimonious than the topology presented (Fig. 1). Similarly, heavier weighting of appendicular characters places Heteronetta as the sister group to both the sea ducks and other stiff-tailed ducks. Patterns of downy young.--Although an anal- ysis of downy patterns for the entire order is not possible at present, a cladistic reevaluation of the downy young illustrated in Delacour (1954, 1956, 1959) permits an independent test of two parts of my phylogenetic hypothesis (using Anas as the outgroup). Patterns in downy stiff-tailed ducks agree well, wherein (1) Het- eronetta retains virtually all dabbling-duck characters; (2) Nomonyx, Oxyura, and Biziura share a synapomorphic, dark cheek stripe; (3) Oxyura and Biziura are united by the derived loss of the pale supraorbital stripe; and (4) Bi- ziura shares a loss of dorsal and wing spotting with the Peruvian Ruddy Duck (O. ferruginea) and dark cheeks with the Australian Blue-billed Duck (O. australis). Patterns of downy Mergini also are informative: (1) eiders retain the su- praorbital stripe of dabblers but, in contrast to osteological evidence, appear monophyletic in their dusky undersides, obsolete dorsal spots, and dark cheeks; (2) other Mergini lack the su- praorbital stripe and dorsal spotting of the dab- bling ducks (loral spot and vestigial back spots retained in Histrionicus); (3) Melanitta and its sister genera are synapomorphic in their dark breast bands, a character secondarily lost in Mergus; (4) the Black Scoter (Melanitta nigra) and Surf Scoter (M. perspicillata) are united by the derived darkening of the lower breast and bel- ly; (5) Bucephala, Mergeflus, Lophodytes, and Mer- gus share a reversal in (presence of) dorsal spot- ting; (6) obscured (Lophodytes) to dark (Mergus) cheeks unite the mergansers; and (7) Mergus is derived further in the anteriorly incomplete breast band and pale suborbital stripe. Similarity vs. relatedness.--Recognition of the different types of character change is important in light of the conspicuously unequal rates of morphological evolution in different lineages, e.g. autapomorphies of Branta vs. Cereopsis (Fig. 2) and Oxyura vs. Biziura (Fig. 5). The inade- quacy of simple distance techniques was dem- onstrated using these data through a compari- son of phylogenetic relationships with "path lengths" or patristic distances. Selected results were: (1) Anseranas is roughly equidistant from the anhimids and other anatids [corroborated immunologically by Bottjer (1983)], but is the sister group to the latter (Fig. 2); (2) Thalassornis is most similar to Dendrocygna but is more closely related to other anatids, excluding An- seranas (Fig. 2); (3) Heteronetta appears "nearer" to Anas than to Oxyura, a member of the sister group of Heteronetta (Figs. 4 and 5); and (4) be- cause of autapomorphies in Oxyura, Biziura is phenetically "closer" to Nomonyx than it is to its sister genus Oxyura (Fig. 5). Life-history correlates.--Diving, at least as an escape behavior, occurs throughout the order except in the anhimids and possibly Anseranas (Johnsgard 1962, Todd 1979). Groups that rou- tinely dive for food are fewer, but occur in six lineages throughout the family (Weller 1964b): Dendrocygna, Thalassornis, Hymenolaimus-Mer- ganetta-Tachyeres, pochards, sea ducks, and stiff- tailed ducks. Perching habit, the probably primitive char- acter used traditionally to define the polyphy- letic "perching ducks," occurs in many genera, including Anseranas, Dendrocygna, Plectropterus, Sarkidiornis, Tadorna, Malacorhynchus, Cairina, Aix, Chenonetta, Rhodonessa, and Amazonetta (All 1960, Johnsgard 1978, Todd 1979). A related trait (also used to justify the "Cairinini"), nesting in tree cavities, occurs in Dendrocygna, some shelducks (Sarkidiornis, Neochen, Alopochen, Ta- dorna, Malacorhynchus), a number of dabbling ducks (e.g. Aix, some Anas), and some sea ducks (Histrionicus, Bucephala, Mergellus, Lophodytes, Mergus). Use of terrestrial cavities for nesting also oc- curs in some shelducks (Tadorna, Hymenolaimus, Merganetta, Tachyeres), in some Anas, and in the sea ducks cited above (Hobbs 1957, Warham 1959, Johnsgard 1962, Johnson 1963, Weller 1964c, Kear 1970, Moffett 1970, Humphrey and Livezey 1985). Other species nest on the ground or over water (Weller 1964c, Kear 1970). Only the Black-headed Duck (Heteronetta atricapilla) is an obligate nest parasite, although infre- quent nest parasitism occurs in a number of other genera including Dendrocygna, Branta, Anas, Aythya, and Mergus (Weller 1959, 1968b). Although ground nesting appears to be prim- itive for the order (Johnsgard 1965a, Kear 1970), nesting habit is probably unreliable for intraor- dinal phylogenetic inferences. Clutch size, proportion of yolk in eggs, incubation period, parental carrying of young, and sexual dimor- phism also appear to be quite plastic (Johns- gard 1961f, 1966b; Lack 1967, 1968, 1974; Johns- gard and Kear 1968; Kear 1970; Livezey and Humphrey 1984). Selected reproductive characteristics, how- ever, show distinct primitive-to-derived se- quences (Kear 1970). Most change near the di- vergence of the goose-swan clade but may be confounded by an evolutionary trend toward reduced body size: (1) nest bowl unlined vs. lined with down (secondarily lost in stiff-tailed ducks); (2) biparental nest construction, incu- bation, and attendance of young vs. female alone responsible; and (3) brooding period and pair bond long ("anserines," roughly 6 months) vs. moderately long (shelducks, roughly 4 months) vs. comparatively short (dabblers and divers, less than 2 months). Participation of males in brood rearing is variable within Anas, however, wherein several neotropical species are characterized by protracted, perhaps per- manent pair bonds (Johnsgard 1978). Biogeographic patterns.--Despite the early ac- knowledgment of the diversity of "aberrant and primitive" genera in Australia (Delacour and Mayr 1945: 51), most previous biogeographers contended that the Anseriformes originated in the Northern Hemisphere, probably the Pale- arctic (Howard 1950, Weller 1964d). Although the fossil record of waterfowl is more complete for the Northern Hemisphere (Howard 1964), early forms are known from both hemispheres, and the apparent disparity in representation probably reflects intensity of paleontological research. The only essentially northern groups are Olor, Anser-Branta, Cyanochen, Rhodonessa, and the sea ducks; genera that have roughly equal distributions in both hemispheres are Cygnus, Tadorna, Alopochen, Anas, Netta, Aythya, and Oxyura. The remaining 20 genera are lim- ited to or most speciose in the Southern Hemi- sphere, and, with the anhimids, include most of the early branches in the order. Further- more, the earliest branches (Fig. 2) in the swan and goose clades are genera limited to the Southern Hemisphere. The numerous holarctic species of Anas, Aythya, and the sea ducks (Weller 1964d) may represent radiations has- tened by widespread glaciations (cf. Ploeger 1968). Consequently, I agree with Cracraft (1980) that the Anseriformes probably origi- nated in the Southern Hemisphere. Taxonomic implications.--The tribes of Anati- dae originally proposed by Delacour and Mayr (1945) were defined primarily in terms of gen- era of the Northern Hemisphere. Although most tribes were assigned members from both hemispheres, only the monotypic "Merganet- tini" was limited to the Southern Hemisphere. A number of "aberrant" southern genera were sorted tentatively among these tribes: Anser- anas and Plectropterus to the Cairinini; Cereopsis, Tachyeres, and Lophonetta to the Tadornini; Stic- tonetta, Malacorhynchus, and Hymenolaimus to the Anatini; and Thalassornis to the Oxyurini. Since then four of these genera have been placed in their own subfamilies or tribes, one has been moved to another subfamily, and the others have remained problematic (Wolfenden 1961, Johnsgard 1978). Several findings in my study involve these genera, and suggest a revision of the classification of waterfowl (Appendix 2). The dabbling ducks are paraphyletic and should be considered a phylogenetically un- resolved group. I therefore place these genera in a provisional taxon, the "Anatini" (Appen- dix 2). If paraphyly of this group is corrobo- rated by further work, the erection of addition- al tribes corresponding to the branches in the grade would be warranted. Classification of fossil groups.--Several early fossils can be classified tentatively on the basis of published descriptions (Howard 1964). Ro- mainvillia (upper Eocene or lower Oligocene), Cygnopterus (upper Oligocene), and Paranyroca (lower Miocene) possess the primitive procor- acoidal foramen (character 92) retained among Recent genera only by anhimids and Anseranas. Conformation of the tarsometatarsal trochlea (68) indicates that at least Romainvillia and Par- anyroca are derived with respect to Anseranas, and the hypotarsus (72) of Paranyroca shows it to be more primitive than Dendrocygna. Ac- cordingly, these fossils should be listed after Anseranas and before Dendrocygna and se- quenced (provisionally by epoch of occur- rence) as Romainvillia, Cygnopterus, and Parany- roca; the fossils may be given familial names or be designated "piesions" (Wiley 1981) at fa- milial rank. Phylogenetic reappraisals of sev- eral other fossil Anseriformes [e.g. Anas(?) blan- chardi, the tadornines Anabernicula and Brantadorna, and Chendytes; Howard 1964], and the probable anseriform Presbyornis (Olson and Feduccia 1980a), should provide minimum ages of branch points in the phylogeny. ACKNOWLEDGMENTS This work was supported in part by National Sci- ence Foundation grant BSR-83-19900 and the Uni- versity of Kansas Museum of Natural History. I am most grateful for the enthusiasm and insights of P. S. Humphrey; his comments on the manuscript, and those of G. E. Woolfenden, D. Siegel-Causey, R. M. Chandler, and W. Hoffman are greatly appreciated. I also benefited from discussions with E. O. Wiley, D. L. Swofford, L. M. Witruer, D. Frost, R. L. Mayden, L. D. Martinß D. E. Seibel, and D. M. Hillis. R. Mengel and M. Jenkinson permitted access to specimens and work space, and loans of specimens were arranged by S. L. Olson (National Museum of Natural Histo- ry), R. W. Storer (Museum of Zoology, University of Michigan), G. E. Woolfenden (Department of Biolo- gy, University of South Florida)ß and P. Devillers (In- stitut Royal des Sciences Naturelies de Belgique). J. Cracraft and A. H. Bledsoe provided useful criticisms of the manuscript. I thank M. A. Masten for graphical help and K. McManness for typing the manuscript. LITERATURE CITED AOTT, C.E. 1938. The skulls of ducks. Wilson Bull. 50: 249-253. ALI, $. 1960. The Pink-headed Duck Rhodonessa car- yophylacea (Latham). Wildfowl 11: 55-60. AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list of North America birds, 6th ed. Washington, D.C., Amer. Ornithol. Union. BALLß $. C. 1934. Hybrid ducks, including descrip- tions of two crosses of Bucephala and Lophodytes. Peabody Mus. Nat. Hist. Bull. 3: 1-26. B^UIEI,, J. J. (Ed.). 1979. Nomina anatomica avium. New York, Academic Press. BELLROSE, F. C. 1976. Ducksß geese and swans of North America. Harrisburg, Pennsylvaniaß Stackpole Books. ]OETTICHER, H. VON. 1943. Die phylogenetisch-sys- tematische stellung yon Anseranas. Zool. Anz. 142: 55-58. --. 1952. Ganse- und Entenv6gel aus aller Welt. Leipzig, Germany, Geese &Portig K.-G. BOTTJER, P.D. 1983. Systematic relationships among the Anatidae: an immunological study: with a history of anatid classificationß and a system of classification. Unpublished Ph.D. dissertation, New Havenß Connecticut, Yale Univ. BRVSH, A. H. 1976. Waterfowl feather proteins: analysis of use in taxonomic studies. J. Zool. London 179: 467-498. COTTERß W. ]., JR. 1957. A serological analysis of some anatid classifications. Wilson Bull. 69: 291- 300. CRACRAFT, J. 1980. Avian phylogeny and intercon- tinental biogeographic patterns. Pp. 1302-1308 in Acta 17th Congr. Intern. Ornithol., vol. II (R. Nohring, Ed.). Berlinß Deutschen Ornithologen- Gesellschaft. DAVIES, $. J. J. F., & H. J. FRITH. 1964. Some com- ments on the taxonomic position of the Magpie Goose Anseranas semipalmata (Latham). Emu 63: 265-272. DELACOUR, J. 1954. The waterfowl of the world, vol. 1. Londonß Country Life. 1956. The waterfowl of the world, vol. 2. London, Country Life. 1959. The waterfowl of the world, vol. 3. London, Country Life. 1964a. Domestic waterfowl. Pp. 154-166 in The waterfowl of the world, vol. 4 (J. Delacour, Ed.). London, Country Life. 1964b. Aviculture. Pp. 145-153 in The wa- terfowl of the world, vol. 4 (J. Delacour, Ed.). Londonß Country Life. 1964c. Corrections and additions. Pp. 327- 354 in The waterfowl of the world, vol. 4 (J. De- lacour, Ed.). Londonß Country Life. ß & E. MAYR. 1945. The family Anatidae. Wil- son Bull. 57: 3-55. -, & ---. 1946. Supplementary notes on the family Anatidae. Wilson Bull. 58: 104-110. DEMAY, I. $. 1940. A study of the pterylosis and pneumaticity of the screamer. Condor 42: 112- 118ß FEDUCCIA, A. 1977. Hypothetical stages in the evo- lution of modern ducks and flamingos. J. Theor. Biol. 67: 715-721. --. 1978. Presbyornis and the evolution of ducks and flamingos. Amer. Sci. 66: 298-304. FRITH, H.J. 1955. The downy ducklings of the Pink- eared and White-eyed ducks. Emu 55: 310-312. ß 1964a. Taxonomic relationship of Stictonetta naevosa (Gould). Nature 202: 1352-1353. 1964b. The downy young of the Freckled Duck Stictonetta naevosa. Emu 64: 42-47. 1967. Waterfowl in Australia. Honolulu, Hawaiiß East-West Center Press. GRAY, A.P. 1958. Bird hybrids: a checklist with bib- liography. Bucks, England, Commonwealth Agr. Bur. HARRISONß J. G. 1958. Skull pneumaticity in wild- fowl in relation to their mode of life. Wildfowl 9: 193-196. HOBBS, J. N. 1957. Notes on the Pink-eared Duck. Emu 57: 263-268. HOWARD, H. 1929. The avifauna of Emeryville shellmound. Univ. California Publ. Zool. 32: 301- 394. 1950. Fossil evidence of avian evolution. Ibis 92: 1-21. 1964. Fossil Anseriformes. Pp. 233-326 in The waterfowl of the world, vol. 4 (J. Delacour, Ed.). London, Country Life. HOWARDß R., & A. MOOREß 1984. A complete check- list of the birds of the world. Londonß Macmil- lan. HUMIHRE3/4, P.S. 1955. The relationships of the seaducks (Tribe Mergini). Unpublished Ph.D. dissertationß Ann Arborß Univ. Michigan. ß 1958. Classification and systematic position of the eiders. Condor 60: 129-135. ß & R. S. BUTSCH. 1958. The anatomy of the Labrador Duckß Camptorhynchus labradorius (Gmelin). Smithsonian Misc. Coil. 135: 1-23. , & G. A. CLARK, JR. 1964. The anatomy of waterfowl. Pp. 167-232 in The waterfowl of the worldß vol. 4 (J. Delacour, Ed.). Londonß Country Life. -, & B.C. LIVEZE3/4. 1985. Nest, eggs, and downy young of the White-headed Flightless Steamer- Duck. Pp. 949-954 in Neotropical ornithology (P. A. Buckley, M. S. Foster, E. S. Mortonß R. S. Ridgely, and F. G. Buckley, Eds.). Ornithol. Monogr. No. 36. ß & S. D. RIPLEY. 1962. The affinities of the Pink-headed Duck (Rhodonessa caryophyllacea). Postilia 61: 1-21. JACOB, J. 1982. Integumentlipide--ihre chemische Struktur und ihre Bedeutung als systematisches Merkmal in der Zoologie. Funkt. Biol. Med. 1: 83-90. -, & A. GLASER. 1975. Chemotaxonomy of An- seriformes. Blochem. Syst. Ecol. 2: 215-220. JOHNSGARD, P.A. 1960a. The systematic position of the Ringed Teal. Bull. Brit. Ornithol. Club 80: 165-167. 1960b. Classification and evolutionary re- lationships of the sea ducks. Condor 62: 426-433. ß 1960c. Comparative behaviour of the Anati- dae and its evolutionary implications. Wildfowl 11: 31-45. 1960d. Hybridization in the Anatidae and its taxonomic implications. Condor 62: 25-33. 1961a. The taxonomy of the Anatidae--a behavioural analysis. Ibis 103a: 71-85. ß 1961b. The systematic position of the Mar- bled Teal. Bull. Brit. Ornithol. Club 81: 37-43. 1961c. Breeding biology of the Magpie Goose. Wildfowl 12: 92-103. 1961d. The behavior and systematic posi- tion of the Hooded Merganser. Wilson Bull. 73: 227-236. 1961e. Tracheal anatomy of the Anatidae and its taxonomic significance. Wildfowl 12: 58- 69. ß 1961f. Evolutionary relationships among the North American Mallards. Auk 78: 3-43. 1962. Evolutionary trends in the behaviour and morphology of the Anatidae. Wildfowl 13: 130-148. 1963. Behavioral isolating mechanisms in the family Anatidae. Pp. 531-543 in Proc. 13th Intern. Ornithol. Congr., vol. I (C. G. Sibley, Ed.). Amer. Ornithol. Union. 1964. Comparative behavior and relation- ships of the eiders. Condor 66: 113-129. 1965a. Handbook of waterfowl behavior. London, Constable. 1965b. Observations on some aberrant Aus- tralian Anatidae. Wildfowl 16: 73-83. 1966a. The biology and relationships of the Torrent Duck. Wildfowl 17: 66-74. 1966b. Behavior of Australian Musk Duck and Blue-billed Duck. Auk 83: 98-110. 1967. Observations on the behaviour and relationships of the White-backed Duck and the stiff-tailed ducks. Wildfowl 18: 98-107. 1978. Ducks, geeseß and swans of the world. Lincoln, Univ. Nebraska Press. 1979. Order Anseriformes. Pp. 425-506 in Checklist of birds of the world, 2nd ed., vol. 1 (E. Mayr and C. W. Cottrell, Eds.). Cambridge, Massachusettsß Mus. Comp. Zool. --, & J. KEAR. 1968ß A review of parental car- rying of young by waterfowl. Living Bird 7: 80- 102. JOHNSON, A. W. 1963. Notes on the distribution, reproduction and display of the Andean Torrent Duck Merganetta armata. Ibis 105: 114-116. KEAR, J. 1967. Notes of the eggs and downy young of Thalassornis leuconotus. Ostrich 38: 227-229. 1970. The adaptive radiation of parental care in waterfowl. Pp. 357-392 in Social behaviour in birds and mammals (J. H. Crook, Ed.). New York, Academic Press. 1972. The Blue Duck of New Zealand. Liv- ing Bird I1: 175-192. 1975. Salvadori's Duck of New Guinea. Wildfowl 26: 104-111. --, & R. K. MURTON. 1973. The systematic status of the Cape Barren Goose as judged by its photo- responses. Wildfowl 24: 141-143. --,  T. H. STEEL. 1971. Aspects of social be- haviour in the Blue Duck. Notornis 18: 187-198. KESSLER, L. G., & J. C. AVISE. 1984. Systematic rela- tionships among waterfowl (Anatidae) inferred from restriction endonuclease analysis of mito- chondrial DNA. Syst. Zool. 33: 370-380. KLVGE, A. G. 1984. The relevance of parsimony to phylogenetic inference. Pp. 24-38 in Cladistics: perspectives on the reconstruction of evolution- ary history (T. Duncan and T. F. Stuessy, Eds.). New York, Columbia Univ. Press. LACK, D. 1967. The significance of clutch-size in waterfowl. Wildfowl 18: 125-128. 1968. The proportion of yolk in the eggs of waterfowl. Wildfowl 19: 67-69. 1974. Evolution illustrated by waterfowl. Oxford, England, Blackwell Sci. Publ. LIVEZEY, B.C., & P.S. HUMPHREY. 1984. Sexual di- morphism in continental steamer-ducks. Condor 86: 368-377. LORENZ, K. 1951-1953. Comparative studies on the behaviour of Anatinae. Avicult. Mag. 57: 157- 182; 58: 8-17, 61-72, 86-94, 172-184; 59: 24-34, 80-91. MAINARDI, D. 1962. Immunological data on the phylogenetic relationships and taxonomic posi- tion of flamingoes (Phoenicopteridae). Ibis 104: 426-428. MAYR, E. 1931. Anatomie und systematischen Stel- lung der Salvadori-Ente (Salvadorina waigiuensis Rothsch. & Hartert). Ornithol. Monatsberichte 39: 69-70. MCKINNEY, F. 1953. Studies on the behaviour of the Anatidae. Unpublished Ph.D. dissertation, Bris- tol, England, Univ. Bristol. MILLER, W. DEW. 1919. Notes on the structure of Anseranas semipalmata. Auk 36: 562-564. MOFFETT, G. M., JR. 1970. A study of nesting Torrent Ducks in the Andes. Living Bird 9: 5-28. MOYNIHAN, M. 1958. Notes on the behavior of the Flying Steamer Duck. Auk 75: 183-202. MYRE$, M.T. 1959. Display behavior of Buffiehead, scoters, and goldeneyes at copulation. Wilson Bull. 71: 159-168. NIETHAMMER, G. 1952. Zur Anatomie und sys- tematischen Stellung der Sturzback-Ente. J. Or- nithol. 93: 357-360. NILSSON, L. 1974. The behaviour of wintering Smew in southern Sweden. Wildfowl 25: 84-88. OLSON, S. L. 1985. The fossil record of birds. Pp. 79-238 in Avian biology, vol. 8 (D. S. Farner, J. R. King, and K. C. Parkes, Eds.). New York, Ac- ademic Press. --, & A. FEDUCCIA. 1980a. Presbyornis and the origin of the Anseriformes (Aves: Charadrio- morphae). Smithsonian Contrib. Zool. 323: 1-24. --, & --. 1980b. Relationships and evolu- tion of flamingos (Aves: Phoenicopteridae). Smithsonian Contrib. Zool. 316: 1-73. PALMER, R. S. (Ed.). 1976. Handbook of North American birds, vols. 2 and 3. New Haven, Con- necticut, Yale Univ. Press. PETERS, J. L. 1931. Check-list of birds of the world, vol. 1. Cambridge, Massachusetts, Harvard Univ. Press. PHILLIPS, J. C. 1922. A natural history of the ducks, vol. I. Boston, Houghton Mifflin. ß 1923. A natural history of the ducks, vol. 2. Boston, Houghton Mifflin. 1925. A natural history of the ducks, vol. 3. Boston, Houghton Mifflin. PLOEGER, P.L. 1968. Geographical differentiation in arctic Anatidae as a result of isolation during the Last Glacial. Ardea 56: 1-159. RAIKOW, R.J. 1970a. The function and evolution of the supraorbital process in ducks. Auk 87: 568- 572. 1970b. Evolution of diving adaptations in the stifftail ducks. Univ. California Publ. Zool. 94: 1-52. 1971. The osteology and taxonomic position of the White-backed Duck, Thalassornis leucono- tus. Wilson Bull. 83: 270-277. RAND, A. L. 1954. On the spurs on birds' wings. Wilson Bull. 66: 127-134. REMS, E. C., & N. HILLGARTH. 1984. The breeding biology of captive Black-headed Ducks and the behavior of their young. Condor 86: 242-250. RIPLEY, S.D. 1957. A paddling of ducks. New York, Harcourt, Brace, and Co. SALVADORI, T. 1895. Catalogue of the Chenomor- phae (Palamedeae, Phoenicopteri, Anseres), Crypturi and Ratitae in the collection of the Brit- ish Museum. London, Longmans. SCOTT, P. 1985. Duck. Pp. 157-161 in A dictionary of birds (B. Campbell and E. Lack, Eds.). Vermil- lionß South Dakota, Buteo Books. SIBLEY, C.G. 1957. The evolutionary and taxonomic significance of sexual dimorphism and hybrid- ization in birds. Condor 59: 166-191. 1960. The electrophoretic patterns of avian egg-white proteins as taxonomic characters. Ibis 102: 215-284. ß & J.E. AHLQUIST. 1972. A comparative study of the egg white proteins of non-passerine birds. Peabody Mus. Nat. Histß Bullß 39: 1-276. ß K. W. CORBIN, & J. H. HAAVIE. 1969. The relationships of the flamingos as indicated by the egg-white proteins and hemoglobinsß Con- dor 71: 155-179. SIEGFRIED, W. R. 1970. Double wing-moult in the Maccoa Duckß Wildfowl 21: 122. SILEO, L., C. H. TUITE, & J. B. D. HOPCRAFT. 1977ß Plumages and primary moult in Lesser Flamin- gosß Wildfowl 28: 139-142. SIMONETTA, A.M. 1963. Cinesi e morfologia del cranio negli uccelli non passeriformi. Studio su varie tenenze evolutive. Parte 1. Archß Zool. Ita- lia (Torino) 48: 53-135. SWOFFORD, D.L. 1984. PAUP: phylogenetic analysis using parsimony. Illinois Nat. Histß Surv. (Mim- eo0 TIMMERMANN, G. 1963. Fragen der Anatidensyste- matik in parasitologischer Sichr. Pp. 189-197 in Proc. 13th Intern. Ornithol. Congr., vol. I (C. G. Sibley, Ed.). Amer. Ornithol. Unionß TODD, F.S. 1979. Waterfowl: ducks, geese and swans of the worldß New York, Harcourt Brace Jova- novichß TYLER, C. 1964. A study of egg shells of the Anati- dae. Proc. Zool. Soc. London 142: 547-583ß VERHEYEN, R. 1955. La systmatique des Ansri- formes base sur l'ostologie compare. Bullß Instß Royal Soc. Natur. Belgique 31(35): 1-18, (36): 1- 16, (37): 1-22, (38): 1-16. WARHAM, J. 1959. The nesting of the Pink-eared Duckß Wildfowl 9: 118-127. WELLER, M. W. 1959. Parasitic egg laying in the Redhead (Aythya americana) and other North American Anatidae. Ecol. Monogr. 29: 333-365. 1964a. Fowling. Pp. 121-127 in The water- fowl of the world, vol. 4 (J. Delacour, Ed.). Lon- don, Country Lifeß 1964b. General habits. Pp. 15-34 in The wa- terfowl of the world, vol. 4 (J. Delacour, Ed.). London, Country Lifeß ß 1964c. The reproductive cycle. Pp. 35-79 in The waterfowl of the worldß vol. 4 (J. Delacour, Ed.). Londonß Country Life. 1964d. Distribution and species relation- shipsß Pp. 108-120 in The waterfowl of the world, vol. 4 (J. Delacour, Ed.). London, Country Lifeß 1967. Notes on plumages and weights of the Black-headed Duck, Heteronetta atricapilla. Condor 69: 133-145. ß 1968a. Plumages and wing spurs of Torrent Ducks Merganetta armata. Wildfowl 19: 33-40ß 1968b. The breeding biology of the parasitic Black-headed Duckß Living Bird 7: 169-207. 1976. Ecology and behaviour of steamer ducksß Wildfowl 27: 45-53ß WETMORE, A. 1926. Observations on the birds of Ar- gentina, Paraguay, Uruguay and Chile. U.S. Natl. Mus. Bull. 133. ß 1951. Observations on the genera of the swans. J. Washington Acad. Sci. 41: 338-340. WILEY, E. O. 1981. Phylogenetics: the theory and practice of phylogenetic systematics. New York, J. Wiley and Sons. WOLFF, E., & E. WOLFFß 1952. La dterminisme de la diffrenciation sexuelle de la syrinx du canard cultive in vitro. Bull. Biol. 86: 325-349. WOOD, D. S., R. L. ZusI, & M. A. JENKINSON. 1982. World inventory of avian skeletal specimens, 1982. Norman, Oklahoma, Amer. Ornithol. Union and Oklahoma Biol. Surv. WOOLFENDEN, G. E. 1961. Postcranial osteology of the waterfowl. Florida State Mus. Bull. 6: 1-129. YAMASHINA, Y. 1952. Classification of the Anatidae based on cytogenetics. Pap. Coord. Comm. Res. Gen. 3: 1-24. ZusI, R. L., & G. D. BENTZ. 1978. The appendicular royology of the Labrador Duck (Camptorhynchus labradorius). Condor 80: 407-418. APPENDIX 1 The 120 characters used in this analysis are listed below and are numbered and grouped anatomically. Character states are lettered and correspond to the character changes in Figs. 2-5. Plesiomorphic (prim- itive) conditions generally are designated "a" and derived character states are ordered alphabetically thereafter (implying a linear transformation series); characters followed by a "U" were analyzed as unor- dered. Characters for which the primitive state was not determined are marked with "U*." Characters judged to be unusually variable, generally necessi- tating determinations of modal conditions, are indi- cated with a "V." Taxa with problematic state deter- minations are listed in parentheses after the corresponding character. Consistency indices (CI) follow each character. Anatomical terminology fol- lows Howard (1929) and Woolfenden (1961) unless annotated otherwise. Integument 1. Molt of remiges: (a) sequential; (b) synchronous, once annually; (c) synchronous, twice annually. (Variable in Phoenicopteridae; Sileo et al. 1977.) CI = 1.0. 2. Tarsal sheath: (a) scutellate anteriorly and posteriorly; (b) reticu- late anteriorly and posteriorly; (c) scutellate anteriorly (at least distally) and reticulate posteriorly. CI = 1.0. 3. Spongy subcutaneous layer: (a) absent; (b) present. CI = 1.0. 4. Interdigital webbing of feet (excluding hallux): (a) lacking (slight webbing in anhimids); (b) incomplete (semipalmate); (c) complete (incised in Cereopsis and Branta sandwicensis). CI = 1.0. 5. Apteria: (a) present; (b) obsolete. CI = 1.0. Trachea 6. Bulla ossea of males (U, weight = 2): (a) not enlarged; (b) sym- metrically enlarged; (c) asymmetrically enlarged, unfenestrated; (d) asymmetrically enlarged, fenestrated; (e) enlargement re- duced and symmetrical, or obsolete. (Thalassornis, Malacorhynchus, Nettapus.) CI = 0.67. 7. Extrasternal, subdermal 1ooping of trachea in males (U*): (a) pres- ent; (b) absent. CI = 0.33. 8. Inflatable tracheal air sacs: (a) absent; (b) present. CI = 1.0 Skull 9. Occipital fontanelles: (a) absent; (b) present. CI = 1.0. 10. Lacrymals (U): (a) not fused to skull; (b) fused to skull dorsally, small, nonpneumatic; (c) fused dorsally, moderately thick, long, ventrally directed (lacking flange), nonpneumatic; (d) fused to skull dorsally, and (typically) also fused to postorbital process, slightly pneumatic; (e) fused dorsally, of variable shape and pneumaticity, posterioventrally directed, often with flanged ven- tral terminus. CI = 1.0. 11. Supraorbital process (U; best developed in adult males): (a) absent or small, straight, essentially coplanar with dorsal surface of skull; (b) large, flat, medially appressed to dorsal margin of orbit; (c) large, thick, rugose, dorsolaterally directed; (d) long, slender, dor- sally directed, often curved. (Aythya, Bucephala albeola.) CI = 0.38. 12. Anterior terminus of premaxillae: (a) strongly ventrally hooked, typically pointed; (b) strongly ventrally hooked, moderately rounded; (c) not ventrally hooked (directed anteriorly), rounded, spatulate. C1 = 1.0. 13. Bill lamellae: (a) absent; (b) present. (Nonhomologous lamellae in flamingos, vestigial in anhimids; Olson and Feduccia 1980a.) CI = 1.0. 14. Retroarticular processes of mandible: (a) lacking, small, or re- curved and rounded; (b) recurved, pointed, and bladelike. CI = 1.0. 15. Quadrate, lateral view: (a) not squarish, with variably deeply curved dorsal margin between orbital and otic processes; (b) squarish, with dorsal margin straight. CI = 0.33. 16. Frontonasal region of skull (U): (a) essentially continuous with profile defined by premaxillae and frontals; (b) enlarged into con- spicuous, laterally compressed, dorsal prominence (larger in males); (c) with rounded, pneumatic swelling (especially in adult males). CI = 0.50. 17. Frontals: (a) without dorsally directed hornlike prominence; (b) with small, ossified "horn" on midline. CI = 1.0. 18. Pterygoid-palatine articulation: (a) a simple abutment; (b) a ball- and-socket arrangement involving two extensions of the ptery- gold. CI = 1.0. 19. Dorsum of upper bill in region of external nares: (a) essentially continuous with curvature of skull to somewhat convex; (b) sub- stantially dorsally bowed. CI = 1.0. 20. Basipterygoid processes: (a) lacking or (in Galliformes) present but without basal supports; (b) present, lipped, almost pedicellate. CI = 1.0. Vertebrae 21. Number of cervical vertebrae (U, V): (a) 17; (b) 18-20; (c) 21; (d) 22-25; (e) 16. (Some Tadorninae.) CI = 0.67. Humerus 22. Capital shaft ridge: (a) prominent and directed toward head; (b) prominent and directed toward external tuberosity; (c) obsolete proximally or absent completely. (Sarkidiornis, Cyanochen, Hymeno- laimus, Chenonetta.) CI = 0.50. 23. Capital groove: (a) short, essentially directed distally; (b) extend- ing laterally toward external tuberosity, undercutting head. (Thal assornis.) CI  0.50. 24. Proximo-anconal region: (a) variably rounded by shaft; (b) tra- versed by a deep, uninterrupted trenchlike depression from un- der head to internal edge immediately distal to bicipital crest. CI = 0.33. 25. Deltoid crest (V): (a) margin rounded, laterally flaring, concave anconally; (b) margin angular or squared, depressed around shaft toward palmar side, convex anconally. (Sarkidiornis, Hymenolai- mus.) CI = 0.50. 26. Surface of attachment for anterior articular ligament: (a) not ele- vated, essentially parallels shaft; (b) elevated, angled distally; (c) elevated, angled medially. CI = 0.40. 27. Internal tuberosity: (a) proximally rotated, exposing completely the pneumatic foramen in anconal view, lacking a distinct, dis- tally directed prominence; (b) produced distally so as to largely or completely obscure pneumatic foramen in anconal view, typ- ically with prominent, distally directed point. CI = 0.50. 28. Pneumatic fossa (U): (a) open, usually containing numerous bony struts; (b) closed by bony shell except for a small central opening; (c) completely closed; (d) closed but perforated by numerous small holes. (Lophodytes.) C1 = 0.43. 29. Attachment site of M. latissimus dorsi posterioris: (a) well medial to external edge of pectoral attachment; (b) in line with outer edge of pectoral attachment, on anconal surface of shaft; (c) in line with outer edge of pectoral attachment, on raised ridge. CI = 1.0. 30. Distal portion of anconal surface of bicipital crest: (a) poorly de- veloped or shelflike; (b) produced medially with distinct proximal cuplike depression, visible as translucent window in palmar view. CI = 0.50. 31. Distal terminus of deltoid crest: (a) essentially continuous with lateral edge of shaft; (b) produced into prominent tuberosity on palmar surface of shaft. CI = 1.0. 32. External tuberosity: (a) prominent, buttressed, typically with at- tachment site elevated, parallel to shaft, not sloping away with ancohal surface of deltoid crest; (b) reduced, lacking buttress, with attachment site sloping and essentially flush with ancohal surface of deltoid crest. (Stictonetta, Sarkidiornis, Cyanochen, Merganetta, Oxyura.) CI = 0.33. 33. Relative anconal heights of ectepicondyle and entepicondyle: (a) ectepicondyle distinctly higher than entepicondyle; (b) condyles essentially equally high; (c) ectepicondyle lower than entepicon- dyle. (Hymenolaimus.) CI = 0.40. 34. Pit for attachment of M. flexor carpi ulnaris: (a) prominent; (b) reduced to obsolete. CI = 1.0. 35. External condyle and brachial depression (palmar side, distal end): (a) separated by smooth strip of bone; (b) connected by rounded ridge. CI = 1.0. 36. Attachment site for external head of triceps: (a) immediately distal to head, typically in excavation under head; (b) displaced distally on lobe of bone and obscures external terminus of capital groove. CI = 1.0. Carpometacarpus 37. Distal end of internal rim of carpal trochlea (external view): (a) with prominent swelling; (b) without prominent swelling; (c) deeply excavated. (Marmaronetta, Hymenolaimus, Clangula, Mergel- lus, Heteronetta.) CI = 0.33. 38. External rim of carpal trochlea: (a) essentially continuous, un- notched; (b) with prominent notch distally. CI = 1.0. 39. Dorsal surface of metacarpal II: (a) flattened proximally (can ap- pear angular); (b) rounded proximally. CI = 1.0. 40. Tuberosity of metacarpal II: (a) small; (b) prominent, spurred. CI = 41. Angle of process of metacarpal I: (a) perpendicular to or proxi- mally directed relative to shaft; (b) angled distally. (Plectropterus.) CI = 1.0. 42. Process of metacarpal 1 (U): (a) not enlarged, length less than width of trochlea; (b) an enlarged, pointed spur, longer than width of trochlea; (c) enlarged, blunt, typically with rugose-capped spur, longer than width of trochlea. (Plectropterus, Hymenolaimus, Lopho- netta.) CI = 0.50. 43. Attachment site of M. extensor metacarpi ulnaris (Zusi and Bentz 1978; "flexor" of Woolfenden 1961) (U, V): (a) completely proxi- mal to proximal fornix of metacarpal II and III: (b) opposite, at least partly, fornix; (c) completely distal to fornix. (Tachyeres.) CI = 0.29. 44. Lower proximal surface of metacarpal III: (a) ungrooved, round- ed; (b) distinctly grooved. CI = 0.50. 45. Facets for digits II and III (U*): (a) facet for digit III extending farther distally than facet for digit II; (b) facets essentially equal in distal extent. CI = 0.33. 46. Cuneiform fossa: (a) shallow to moderately deep; (b) deep, round- ed, ovate, with distinct rim. CI = 1.0. APPENDIX 1. Continued. 47. Distal portion of internal rim of carpal trochlea: (a) of uniform thickness with proximal portion; (b) distinctly thickened. CI - 0.50. 48. Internal rim of carpal trochlea (posterior view): (a) in line with internal margin of shaft; (b) sharply deflected laterally. (Callonet- ta.) CI 1.0. Radial carpal 49. Size and shape: (a) small, short, and blunt; (b) elongated into large pointed spur. CI - 1.0. Appendicular pneumaticity 50. Distal alar and pelvic elements: (a) essentially nonpneumatic; (b) pneumatic with one or more large foramina. CI - 1.0. Femur 51. Head, relative to plane of external surface of shaft: (a) oriented posteriorly; (b) perpendicular. CI - !.0. 52. Anterior extent of trochanter: (a) relatively great, such that an- terior-posterior depth of trochanter substantially exceeds depth of head; (b) reduced, such that depth of trochanter only equals that of head. CI = 0.33. 53. Distal extent of internal condyle: (a) distinctly less than that of external condyle; (b) equal to that of external condyle. CI 1.0. 54. Rotular depression: (a) shallow to moderately deep, margin rel- atively indistinct; (b) deep, distinctly bordered proximally. CI = 1.0. 55. Curvature of shaft, lateral view (U): (a) straight to slight; (b) mod- erate; (c) strong, subangular. CI - 0.22. 56. Popliteal fossa: (a) shallow; (b) deep, typically pitted. CI -- 0.25. 57. Lobe at midpoint of posterior surface of shaft: (a) not prominent; (b) prominent. CI - 1.0. 58. Posterior intermuscular line (U): (a) relatively distinct, following internal edge of shaft; (b) relatively distinct, swings laterally to- ward trochanter; (c) indistinct. CI = !.0. 59. Internal edge of distal end of shaft: (a) smoothly curving and continuous with proximal portion; (b) leveled by raised ridge to internal condyle. CI = 1.0. 60. Posterior intermuscular line: (a) distinguishable only as fine etch- ing; (b) forming overhanging ridge proximally. CI = 1.0. Tibiotarsus 61. Proximal articulating surface: (a) in line with shaft, squares with distal condyles; (b) strongly rotated counterclockwise about shaft (proximal view). CI = 0.50. 62. Rim of internal condyle: (a) distinctly notched; (b) lacking notch. CI = 1.0. 63. Inner cnemial crest: (a) not deflected laterally; (b) laterally de- flected. CI = 1.0. 64. Anterior extent of condyles: (a) internal distinctly greater than external; (b) approximately equal. (Aythya.) CI = 0.25. 65. Inner cnemial crest: (a) lacking distinct ridge extending distally along anterior surface of shaft; (b) continued by distinct ridge distally along anterior surface of shaft to point well beyond prox- imal terminus of fibular crest. CI = 0.25. 66. Internal condyle, posterior view: (a) with relatively rounded in- ternal edge; (b) flared, with squared medial edge. CI - !.0. 67. External ligamental prominence: (a) essentially continuous with curvature of shaft; (b) produced laterally, ridgelike. CI = !.0. Tarsometatarsus 68. Trochlea for digit 11: (a) approximately equal to trochlea for digit IV in distal extent; (b) proximal to trochlea for digit IV. CI - 1.0. 69. Anterior (of two) ligamental passages between trochlea for digits III and IV (in distal wall of distal foramen) (V): (a) obscured from view anteriorly by bone; (b) largely or completely exposed ante- riorly because of reduction of bony covering. CI - 0.25. 70. Internal calcaneal ridge of hypotarsus: (a) slightly to moderately exceeds other calcaneal ridges in posterior extent; (b) greatly ex- ceeds other (more external) calcaneal ridges. CI - 0.50. 71. Facet for metatarsal I: (a) deep; (b) obsolete. CI - 1.0. 72. Calcaneal ridges of hypotarsus: (a) 2, lateral to midline of shaft, bordered medially by depression (deep in Anseranas); (b) 3 or 4, situated on midline of shaft, without depression on internal mar- gin. CI = 1.0. 73. Wing on trochlea for digit lI: (a) not prominent medially; (b) medially prominent, thickened. CI = !.0. 74. Groove in trochlea for digit II: (a) absent; (b) present, but posterior terminus of groove variable in extent. CI - 1.0. 75. Anterior extent of internal and external ridges of shaft: (a) essen- tially equal, no twisting of shaft about its long axis; (b) internal ridge less prominent anteriorly than external, becoming flush with shaft immediately distal to proximal foramen, associated with moderate twisting of shaft; (c) internal edge of shaft depressed below level of shaft anteriorly, associated with strong twisting of shaft. (Hymenolaimus, Merganetta, Amazonetta.) CI = 0.33. 76. External margin of shaft: (a) concave (in anterior profile), curving smoothly to external surface of trochlea for digit IV; (b) essen- tially straight, trochlea for digit IV internally deflected. CI = 1.0. 77. Posterior opening of distal foramen: (a) directed posteriorly, flush with surface of shaft; (b) directed distoposterinrly, recessed in depression immediately proximal to symphysis of trochlea for digits III and IV. (Anseranas.) CI = 1.0. Sternum 78. Pneumatic foramen (U*): (a) open, ovoid; (b) pitted, largely oc- cluded by medial bar; (c) closed (sometimes marked by small depression). CI = 0.25. 79. Ventral manubrial region (U*, V): (a) keel-like, laterally com- pressed medial flange; (b) thick medial wedge; (c) lacking median protuberance(s); (d) long, peglike spine; (e) small lump; (f) a pair of small pointed prominences separated at midline by a deep excavation, typically with an ovoid pit at base; (g) a wide, mod- erately long, dorsoventrally compressed flange; (h) a pair of points partially separated by a shallow midline excavation; (i) a small, unforked, dorsoventrally compressed flange. (Cyanochen, Hymen olaimus, Cairina, Lophonetta, Chenonetta, Polysticta.) CI = 0.50. 80. Carina (keel) shape, lateral profile: (a) well developed, ventral margin curved throughout length; (b) reduced, ventral margin essentially straight for posterior half. CI - 0.33. 81. Posterior-lateral processes (U*): (a) extend well posterior to post- pectoral line of sternal plate; (b) approximately equal to post- pectoral line in posterior extent. CI = 0.33. 82. Dorsal manubrial region (U, V): (a) rounded notch; (b) rounded notch with small point on midline; (c) rounded notch with mod- erately large point on midline; (d) even shelf. (Malacorhynchus, Hymenolaimus, Marmaronetta, Callonetta.) CI = 0.43. 83. Abdominal plate (dense, symmetrical extension of sternal plate posterior to both post-pectoral line and posterior-lateral process- es): (a) absent; (b) present. CI !.0. 84. Sternal notches (posterior margin of plate medial to posterior- lateral processes): (a) typically open posteriorly; (b) typically closed posteriorly, forming fenestrae. CI 1.0. 85. Xiphial area: (a) posterior margin approximately straight or con- cave; (b) with medial, irregularly shaped, roughly circular exten- sion of thin bone (anterior to posterior-lateral processes). CI = 1.0. 86. Costal margin: (a) comprises less than half of basin length; (b) comprises more than half of basin length. CI - 1.0. 87. Carina: (a) lacking pneumatic foramen in anterior margin, appar- ently solid; (b) with small pneumatic foramen in anterior edge, but carina uninflated; (c) hollow, containing loop of trachea, with large pneumatic foramen in anterior edge. CI 1.0. 88. Intermuscular line: (a) angles medially to carlhal base well ante- rior to posterior edge of plate; (b) extends posteriorly to posterior margin of plate. CI 0.33. 89. Foramina of basin (U*): (a) limited to midline and anterior mar- gin; (b) essentially absent; (c) present on anterior margin, mid- line, and scattered across plate (often among transverse bony striatinns). CI = 0.50. 90. Midpoint of coracoidal sulcus: (a) solid; (b) having oval pneu- matic foramen. CI = !.0. Costae 91. Uncinate processes: (a) present; (b) absent. CI - 1.0. Coracoid 92. Procoracoidal foramen: (a) present (variable in Chauna); (b) ab- sent. CI = 1.0. 93. Pneumatic foramen on dorsal surface anterior to sternal facet: (a) present; (b) absent. CI = 1.O. 94. Dorsal sternal facet: (a) with anterior border essentially smoothly curving; (b) with prominent circular internal lip. CI 1.0. 95. Brachial tuberosity (U): (a) essentially without foramina under posterior edge; (b) with small foramina under posterior edge; (c) with small foramina, typically contained within larger foramina, under posterior edge. CI = 0.50. 96. Depression on ventral surface anterior to sternal facet (U*): (a) present, typically deep; (b) absent. (Anseranas.) CI - 0.25. 97. Furcular facet: (a) with posterior margin complete or slightly re- duced; (b) posterior margin deeply notched. CI - 1.0. 98. Angle of head: (a) coplanar to slightly ventral to plane of blade; (b) distinctly ventral to plane of blade. CI = 1.0. 99. Sternocoracoidal process: (a) wide, long, and rounded flange, ex- tending farther laterally than sternal facet; (b) variably shaped, rounded or angular process, approximately equal to sternal facet in lateral extent; (c) long pointed process, extending farther lat- erally than sternal facet. CI = 0.67. 100. Ventral (external) sternal facet (V): (a) anterior margin moderately raised or continuous with blade; (b) anterior margin with distinct buttress. CI - 0.33. Furculum 101. Coracoidal tuberosities (U*): (a) present; (b) obsolete. CI = 0.25. 102. Furcular process: (a) a flattened point; (b) variable, but reduced, essentially continuous with curvature of clavicles; (c) swollen truncate lobe. CI = 1.0. 103. Clavicular symphysis: (a) without foramina; (b) with medial fo- ramina. CI - 1.0. 104. Clavicles: (a) roughly circular in cross-section; (b) distinctly flat- tened antero-posteriorly. CI = 1.0. 105. Lateral surfaces of clavicles: (a) smooth, unperforated; (b) with depressions containing several small foramina; (c) with depres- sion containing large pneumatic foramen. CI 0.50. 106. Region of clavicular symphysis: (a) a continuous smooth curve; (b) markedly extended posterodorsally, forming a U-shaped ac- commodation for tracheal loop and associated modification of ca- rina. CI 1.0. Scapula 107. Coracoidal articulation: (a) flush with blade; (b) base protruding ventrally as rounded hump. CI = 1.0. 108. Taper (profile) of blade (U): (a) of uniform width or tapering continuously throughout length; (b) width maximal at midpoint; (c) width maximal at terminus. (Cygnus, Olor.) CI - 1.0. 109. Coracoidal articulation: (a) equal to acromion in proximal extent; (b) distinctly distal to acromion. CI = 1.0. 110. Internal surface, immediately posterior to glenoid facet: (a) essen- tially smooth; (b) having deep depression. CI -- 1.0. 111. Anterior edge (U*): (a) containing pneumatic fossa; (b) without pneumatic fossa. (Cairina.) CI = 0.25. 112. Dorsal surface of neck: (a) marked by single distinct raised at- tachment scar; (b) marked by two prominent raised attachment scars. CI 1.0. Pelvis 113. Preacetabular iliac fossa: (a) smoothly curved surface; (b) contain- ing a deep, irregularly shaped depression. CI = 1.0. 114. Caudal margin: (a) ischium extending well caudad to ilium; (b) variable, but ischium and ilium roughly equal in caudal extent, forming an obliquely sloping margin, with elements typically separated posteriorly by a distinct notch. CI 1.0. 115. Body of pubis (V): (a) concave dorsally (rarely almost straight); (b) convex dorsally. CI - 0.50. 116. Orientation of postischiac pubis: (a) directed posteriorly; (b) di- rected ventrally. CI = 1.0. 117. Shape of postischiac pubis: (a) of uniform width or evenly wid- ening caudally; (b) widened into roughly circular flange, espe- cially extensive anteroventrally. CI = 1.0. 118. Dorsolateral crests: (a) distinct to caudal margin of pelvis; (b) be- comes obsolete cranial to caudal margin. CI = 1.0. 119. Anterior terminus of shield (posterior terminus of fusion of me- dian dorsal ridge): (a) cranial to acetabula; (b) essentially coinci- dent with acetabula; (c) well caudad to acetabula. CI = 0.67. 120. Recessus iliacus (Baumel 1979; a pneumatic pocket at caudal ter- minus of renal depression): (a) present; (b) absent. CI 1.0. APPENDIX 2. A Linnean classification of the Recent genera of Anseriformes. I follow the conventions of Wiley (1981), with the exception of the provi- sional recognition of the paraphyletic "Anatini" (annotated incertae sedis). I have retained, where possible, the names and taxonomic ranks of pre- vious classifications. Names of subtribes are given endings of -eae after the names for the "sections" of Boetticher (1952) and are derived from the old- est included genus. Sedis mutabilis follows taxa in which the order of included groups is unresolved. * = two subfamilies may be in reverse order; ** = subtribes possibly are sister groups; *** = probable sister groups. Order Anseriformes Suborder Anhimae Family Anhimidae Genus Anhima Genus Chauna Suborder Anseres Family Anseranatidae Genus Anseranas Family Anatidae Subfamily Dendrocygninae Genus Dendrocygna Subfamily Thalassorninae* Genus Thalassornis Subfamily Anserinae* Tribe Anserini Genus Cereopsis Genus Anser Genus Branta Tribe Cygnini Genus Coscoroba Genus Cygnus Genus Olor Sub family Stictonettinae Genus Stictonetta Sub family Plectropterinae Genus Plectropterus Sub family Tadorninae Tribe Sarkidiornini Genus Sarkidiornis Tribe Tadornini sedis mutabilis Subtribe Tadorneae Genus Tadorna Subtribe Malacorhyncheae Genus Malacorhynchus Subtribe Chloephageae sedis mutabilis Genus Alopochen Genus Neochen Genus Chloephaga Subtribe Cyanocheneae** Genus Cyanochen Subtribe Merganetteae** Genus Hymenolaimus APPENDIX 2. Continued Genus Merganetta Genus Tachyeres Subfamily Anatinae [Tribe] "Anatini" incertae sedis Genus Pteronetta Genus Cairina Genus A/x Genus Lophonetta Genus Nettapus Genus Anas Genus Callonetta Genus Chenonetta Genus Amazonetta Tribe Aythyini Genus Marmaronetta Genus Rhodonessa Genus Netta Genus Aythya APPENDIX 2. Continued Tribe Mergini Genus Polysticta Genus Somateria Genus Histrionicus Genus Camptorhynchus Genus Melanitta Genus Clangula Genus Bucephala* * * Genus Mergellus*** Genus Lophodytes Genus Mergus Tribe Oxyurini Genus Heteronetta Genus Nomonyx Genus Oxyura Genus Biziura