The Condor90:885-905 0 The CooperOrnithological Society 1988

PHYLOGENY OF THE PHALACROCORACIDAE’

DOUGLAS SIEGEL-CAUSEY Museum of Natural History and Department of Systematicsand Ecology, Universityof Kansas, Lawrence,KS 66045

Abstract. I undertooka phylogeneticanalysis of the Recenttaxa of Phalacrocoracidae usingqualitative osteological characters. The family comprisestwo subfamilies.The Phal- acrocoracinae(cormorants) comprise four generaof all-dark, littorine species:Microcarbo (microcormorants),Compsohalieus (marine cormorants), Hypoleucos (mesocormorants), and Phalacrocorax(macrocormorants). The Leucocarboninae(shags) comprise five generaof variablyplumaged, littorine, and pelagicspecies: Leucocarbo (guano shags), Notocarbo (blue- eyedshags), Nesocarbo (Campbell Island Shag), Euleucocarbo (New Zealandblue-eyed shags), and Stictocarbo(cliff shags).The relationshipof the anhingas(Anhingidae) to the Phalacro- coracidaeremains problematical and unresolved.Additional analyses using cranial or post- cranialcharacters produced comparable results, with the greatestdivergence obtained when only crania were compared.I discussthe nature of homoplasyin the family: cormorants are characterizedby convergences,shags by reversals.Plumage patterns have functional correlates,but phylogenetichistory may be the ultimate factor. Key words: Phalacrocoracidae;Anhingidae; Phalacrocoracinae;Leucocarboninae; cor- morants;phylogenetics; systematics; osteology.

INTRODUCTION ment (Sharpe 1899, Peters 193 1, von Boetticher Since the very first attempts to reconstruct the 1937, van Tets 1976, Dorst and Mougin 1979); phylogeny of the Class Aves, there has been little incomplete treatments dealing with regional fau- controversy over which taxa comprised the Phal- nas, selectedspecies groups, or collection hold- acrocoracidae. More pertinent at the time was ings were more common (Ogilvie-Grant 1898; determining the relationships of the family with- Hutton 1903; Hall 1920; Mathews and Iredale in Pelecaniformes, and that of the order to the 1921; Falla 1932, 1937; van Tets 1965). Most rest of birds. The most widely acceptedphylog- recently, Cracraft (1985) investigated the system- eny was by Wetmore (1934) who positioned the atic relationships among Pelecaniformes, and family between the Sulidae and Anhingidae at confirmed the monophyly of the family. the “primitive end’ of the avian phylogeny. Lit- The classification generally followed in the tle effort was expended by systematists before northern hemisphere is of a single genus Phal- or after this to determine relationships within acrocorux (cf. Dorst and Mougin 1979) some- the cormorants. times supplemented by the monotypic genus The Phalacrocoracidaehave been considered Nunnopterum for the Flightless Cormorant ofthe to comprise approximately 30 speciesin one to Galapagos(cf. Sharpe 1899, Witherby et al. 1940). five genera;the most recent treatment recognizes Other taxonomies often remove the microcor- 29 speciesin one genus(Dorst and Mougin 1979). morants (sensu van Tets 1976) from Phalucro- There has been little consensusamong authori- corax into the genus Huliaetor [sic] (cf. Peters ties on the status of certain speciesand higher- 193 1). In the southern hemisphere, Phalucro- order relationships within the family, possibly corax is partitioned further by the use of Leu- because of sporadic and incomplete systematic cocarbo and Stictocarbo for various shagsof the treatments and changing conceptsof speciesand southern oceans (cf. Falla 1932, 1937). genera. Although cormorants are found world- An ancillary (and essentially trivial) issue has wide, and are usually abundant members of the been whether the anhingas (darters) constitute a littorine avifauna, in the past century there have distinct family or are a subfamily of the Phala- been only a few attempts at a family-wide treat- crocoracidae.Peters (193 l), Olson (1985) Beck- er (1986) and others consideredtheir differences to be sufficient for family standing (i.e., Anhing- * Received23 March 1988.Final acceptance7 July idae), while Dorst and Mougin (1979) Cracraft 1988. (1985) and many others found their similarities

18851 886 DOUGLAS SIEGEL-CAUSEY with the Phalacrocoracidae to be great enough cent Phalacrocoracidaeand Anhingidae using 137 to restrict it to a subfamily (i.e., Anhinginae). osteological characters.I present a hypothetical The first worldwide treatment was by Sharpe evolutionary tree for the family, and discussits (1899) who, without indicating methodology or implications for classification, morphological charactersused, placed all forms from the early convergence, and plumage patterns. Miocene to the Recent into Phalacrocorax, re- This article is dedicatedto the memory ofRalph serving separategenera for an Eocene taxon Ac- W. Schreiber, friend and teacher, who encour- tiornis spp., the Galapagos Cormorant (Nannop- aged me early on to study Pelecaniformes, and terum harrisz),and the extinct Pallas’s Cormorant supported my continuing research on cormo- (Pallasicarboperspicillatus).Because he listed the rants with insight and humor. His sudden death latter speciesalso in Phalacrocorax, this gave it diminishes us all. the notable distinction of belonging to two genera simultaneously. METHODS Peters (193 1) followed a traditional linear ar- rangement of species placed into two genera TAXA AND SPECIMENS (Phalacrocorax, Halietor), but without justifi- I studied skeletons of all Recent taxa of Phala- cation or methodology. Von Boetticher (1936) crocoracidae, except for the Indian Cormorant, considered certain aspectsof external morphol- Hypoleucosjiiscicollis, and some of the island ogy (e.g., rectrix number, abdomen color, etc.) forms of Notocarbo atriceps found in Antarctic and biogeography,and clusteredall of the extant waters (e.g., N. a. melanogenis, N. a. purpuras- forms by general similarity into three genera and tens, etc.), of which no specimens are readily 10 subgenera. He later revised this taxonomy available (Wood and Schnell 1986). In taxa (von Boetticher 1937) and altered speciesallo- known to be particularly variable or encom- cations, generic and subgenericnames, and pro- passingmany subspecies,I studied as many forms poseda resolution of the blue-eyed shagcomplex as possible. In all, I examined 226 specimens of (Leucocarbos.1.) of the southern hemisphere. 36 putative species of cormorants, shags, and Systematists studying southern hemisphere anhingas (Table 1). Except for certain New Zea- Phalacrocoracidaehave been facedwith a greater land shags(Nesocarbo campbelli, Euleucocarbo diversity of speciesthan elsewherein the world, chalconotus,E. colensoi,E. onslowi,E. ranfurlyi: and perhaps as a direct result of this, have pro- one skeleton each), three microcormorants @Ii- posed a variety of taxonomies (cf. Mathews and crocarbocoronatus, M. niger, A4. pygmaeus:one Iredale 1913, Falla 1937, and others). Of these, skeleton each), and the extinct Spectacledor Pal- only van Tets ( 1976) consideredall extant species. las’s Cormorant (Compsohalieusperspicillatus: Using similarities in external morphology and unassociated skeletal elements), I examined at behavior, he apportioned all members of this least two specimens of each species. family into two genera (Phalacrocorax, Leuco- carbo) with three subgenera in the former, and ANALYSIS OF CHARACTERS two in the latter genus. I used 137 osteological characters for the phy- Most recently, Dorst and Mougin (1979) fol- logenetic analysis (Appendix 1); less than one- lowing Peters (193 l), lumped all extant species fourth of these have been described or defined into a single genus Phalacrocorax. Assessments previously, but most are illustrated without iden- of possible specific and superspecific affinities tification in referencesaccompanying the char- were given by footnote but without justification. acter descriptions. Where possible, anatomical Neither Dorst and Mougin nor Peters presented descriptions follow Howard (1929) and Owre an explicit phylogeny of the family, but listed (1967); in many cases,however, suitable names speciesin a linear arrangement, “in taxonomic for featurescould not be determined throughthese sequence” (Peters 1931:iv, Mayr and Cottrell sources, and characters were described instead 1979:vi). by appearance or location. In some taxa previ- To date, no attempt has been made to present ously considered to be of subspecificrank (e.g., anything other than a linear arrangement of E. chalconotus, N. bransjieldensis,Stictocarbo speciesand only van Tets (1976) gave distinct featherstoni), I found sufficient diagnostic char- characters to delineate higher-order relation- acters(autapomorphies) to discriminate them as ships. I undertook a phylogenetic analysis of Re- species (see McKitrick and Zink 1988). In the CORMORANT PHYLOGENY 887

FIGURE 1. General tree of the Recent speciesof Phalacrocoracidae(CI = 0.678, length = 227 steps).Character changesare given in the indicated figureswhich follow. Speciesnames in bracketsindicate presumptiveplacement where no specimenswere available for study. case of N. atricepsand albiventer, two taxa cur- these outgroups and the method of Maddison et rently considered specifically distinct, I was un- al. (1984) to construct a hypothetical ancestor to able to discover any diagnostic osteologicalchar- root the evolutionary tree. I found no differences acters. For this analysis, therefore, they were in tree topology of the Phalacrocoracidae using treated together under the senior synonym, atri- actual outgroup taxa or a hypothetical ancestor, ceps.Members of two genera of cormorants (Hy- although I obtained the most parsimonious so- poleucos, Phalacrocorax) showed substantial lution using the latter. Transformation serieswere geographic variation in osteology within cur- treated as linear, except in four caseswhere I was rently recognized species;in these cases,I used unable to determine linearity with confidence. specimens of the nominal taxon (e.g., H. oliva- These (8,60,62,97) I treated as unordered char- ceusolivaceus, P. carbo carbo, etc.). acters. To test the effect of these assumptions I used only qualitative osteologicalcharacters about the probable evolution of character states in this analysis becauselittle is known at present on tree stability, I performed subsequent anal- about behavior, life histories, and the intrataxon yses treating all charactersas unordered. variation in external morphology for approxi- The trees were derived using the PAUP pro- mately half of the extant species.In addition, it gram (Swofford 1984) and I used the method of is not possible at present to establish polarities searchdescribed by Livezey (1986). The search and states with many of these nonosteological for most parsimonious trees was accomplished characters. using the multiple parsimony (MULPARS) and Each of the characters I used was a discrete alternate swapping between global and local trait for which at least two discrete states could search (ALT) options. The accelerated trans- be defined. I made no attempt to develop an formation (ACCTRAN) optimization, which exhaustive list of diagnostic charactersfor each minimizes reversals within a tree, was used to species. position characters. Results of analysis using delayed transformation (DELTRAN), which minimizes parallel character stateswithin a tree, DERIVATION OF TREES did not alter topology; differencesinvolved only I determined polarities of each character by com- the placement of nine characters (see below in parison with selected taxa-Sula and Morus (Su- Results). lidae), Pelecanus(Pelecanidae), Fregata (Fregati- SeeAppendix 1 for character descriptions and dae), Diomedea (Diomedeidae), and Spheniscus polarities, and Appendix 2 for the data matrix (Spheniscidae)-each proposed as an outgroup of character state codings for the hypothetical to the Phalacrocoracidae(Cracraft 1985). I used ancestor,anhingas, shags,and cormorants. A list 888 DOUGLAS SIEGEL-CAUSEY

Leucocarboninae of species and specimens examined is available from the author.

RESULTS Phalacrocoracinae I found one tree using the charactersand polar- ities given in Appendix 1. The tree illustrated (Figs. 1-5) has a length of 227 steps,and a CI of 0.678. The normalized F-ratio was 0.266, with four out of 137 charactersunordered. Separate analyses using DELTRAN and ACCTRAN op- timizations (see Methods) differed only in the Anhingidae relative placement of nine characters(3, 27, 28, 83 la-b 1Aa-b 49, 50, 57, 110, 112, 135) in the basal portion ofthe tree; topologiesdid not differ between these runs. The tree illustrated here is the result of analyses using the ACCTRAN optimization.

ANHINGIDAE Ridgway, 1887 Hypothetical (anhingas,darters) Ancestor Two homoplasious synapomorphies support the FIGURE 2. Partial tree of the character changesat monophyly of the Anhingidae (Fig. 2) and are the subfamily level. Solid lines representderived char- convergent with features related to feeding and acter transformations; parallel lines, transformations diving found in the Phalacrocoracidae(see Stolpe convergent elsewhere in the tree. The number to the left of the characterstate symbol is the characternum- 1932, Owre 1967). Coding of eight characters ber describedin Appendix 1; the letter sequenceto the (27,28,49,50,79,80,97, 100) were problemati- right is the character state transformation. cal in this group, possibly because I examined only five specimens of two putative species(A.

I Phalacrocoracinae FIGURE 3. Partial tree of the character changesin the Phalacrocoracinae(cormorants). Solid lines represent unambiguouslyderived charactertransformations, parallel lines representcharacter transformations convergent to states found lower in the tree, crossesrepresent charactertransformations reversed to statesfound lower in the tree, and open circles represent complexly varying charactersshowing evidence of both convergenceand reversal within the family. Symbol legendsare described in Figure 2. CORMORANT PHYLOGENY 889

Fig

LeucocarboninaeI

FIGURE 4. Partial tree of the characterchanges in the Leucocarboninae(primitive shags).Symbols and legends are described in Figures 2 and 3. anhinga, A. novaehollandiae). In some speci- respectively, and are supported by various other mens, intermediate states for these characters charactersof plumage color, external morphol- were found; in other specimens,character states ogy, and behavior (van Tets 1976; Siegel-Causey, varied among supposed conspecifics. Because unpubl.). precise coding of these characters could not be determined with confidence, they were treated as polymorphic, i.e., both character statesfound gaimardi within the group. In subsequent analyses, these characterswere coded first as primitive and then as derived. In both cases,a single tree was found which differed only from that illustrated here in that the Anhingidae were placed between subfamilies of Phalacrocoracidae.In the former analysis, Anhingidae was the sister group to the shags;in the latter, to cormorants. Similar results were obtained using all characters unordered; otherwise, the trees were the same as in the pri- mary analysis. PHALACROCORACIDAE Bonaparte,1855 Three synapomorphies, one of which (137: ter- minal hook to the bill) is unambiguous, reaffirms the monophyly of the family (Fig. 2) implied earlier by Cracraft (1985). I determined that the

Phalacrocoracidaecomprise two primary clades, Stictocarbo which I treat here as subfamilies: the Phalacro- FIGURE 5. Partial tree of the character changesin coracinae and Leucocarboninae. These subfam- the continuation of the Leucocarboninae(cliff shags). ilies are referred to here as cormorants and shags, Symbols and legendsare describedin Figures 2 and 3. 890 DOUGLAS SIEGEL-CAUSEY

PHALACROCORACINAE Bonaparte,1855 1964) and behavior (van Tets 1965, 1976) lend (cormorants) additional validity for the clade. Ten character state changes,six of them unam- HypoleucosReichenbach, 1852 (mesocormo- biguoussynapomorphies(41,55,86,92,97, 106), rants). - Three synapomorphies, one of which support the monophyly of this subfamily (Fig. (5 1) is unambiguous and relates to adduction of 2). Of the remaining synapomorphies, three (4, the mandible (Owre 1967), establish the mono- 11, 96) are convergent and one is reversed (57) phyly of this genus of entirely dark, medium- farther up in the tree. The unambiguous synapo- sized cormorants (Fig. 3). My analysis indicates morphies relate to jaw musculature, distal wing that H. olivaceusis the earliest divergence within flexibility and strength,and hind limb action (see this genus and, together with H. auritus, are the Owre 1967), and correlate with an overall pref- only New World mesocormorants. As with P. erenceby cormorants for lesssustained flight and carbo, all of the speciesI examined in this genus feeding in deeper waters (Ainley 1984, Siegel- showeddistinct intrataxon variation in character Causey 1985). My analysis confirms van Tets’ states. The Indian Cormorant, H. fiscicollis, is (1976) implication that the cormorants consti- most likely a member of this genus;its plumage, tute a monophyletic group. behavior, and external morphology (Ogilvie- Microcarbo Bonaparte, 1855 (microcormo- Grant 1898; Ali and Ripley 1968; van Tets, pers. rants).-My analysis confirms the monophyly of comm.) indicate that its systematicposition may this derived genus by 16 character changes(Fig. lie between H. auritus and H. varius. Unfortu- 3), six of which are unambiguous (8, 46, 55, 58, nately, no skeletons exist of this species(Wood 103, 109). The unambiguously derived synapo- and Schnell 1986), so its phylogenetic position morphies representchanges in cranial shapeand remains undetermined. terminal enervation of the mandible. coracoidal Phalacrocorax Brisson, 1760 (macrocormo- musculature, possible increased furcular move- rants).-Two species,the Great Cormorant, P. ment, and hind limb musculature (see Owre carbo, and the JapaneseCormorant, P. capilla- 1967). tus,comprise this genusof large cormorants.Nine Large cormorants. -The remaining three gen- character changes, five of them unambiguous era of cormorants are united by six synapomor- synapomorphies(12,30, 32, 34,67) and relating phies, one of which (80) is unambiguous, and to palatal musculature, nasal glands, and flight relates to humeral rotation and more powerful (seeOwre 1967), establish the monophyly of this flight capabilities (see Owre 1967). Other behav- genus (Fig. 3). Most of the synapomorphies are ioral and morphological features(van Tets 1965, related to feeding and have been associatedpre- 1976) support this grouping. viously with a preference for shallower waters CompsohalieusRidgway, 1884 (marine cor- and a diet of upper-water fish (Kuroda 1922, morants).-Six synapomorphies, one of them Austin 1948, van Dobbin 1952, Ono 1980, H%r- unambiguous (86) and relating to distal wing k&en 1988). flexibility (Owre 1967), establishes the mono- phyly of this genus of heavy-bodied, widely dis- LEUCOCARBONINAE, new name (shags) tributed cormorants (Fig. 3). Members of this Seventeen character changes, six of which are genus are the deepestand strongestdivers in the unambiguous synapomorphies (50, 59, 63, 107, family (Ainley 1984, Siegel-Causey 1985); some 125, 129), establish the monophyly of this of the characters(6,68, 100) are convergent with subfamily (Figs. 4, 5). These derived characters those observed in shagsand may represent sim- relate to jaw musculature, and strengthenedhu- ilar adaptations to a pelagic habitat. moral and femoral musculature (seeOwre 1967). Continental cormorants. -This clade, consti- My analysis indicates the existence of five main tuting Phalacrocoraxand Hypoleucos,comprises clades, which I interpret as genera. All but one moderate to large cormorants breeding in the of thesegenera have been recognizedpreviously, marine and aquatic littoral. Two unambiguous but speciesallocation among them has differed synapomorphies(7,93) relating to bill shapeand widely among authors. For convenience, the first distal wing musculature (see Owre 1967) estab- four genera are referred to as “primitive shags.” lish the monophyly of this group (Fig. 3); simi- Leucocarbo Bonaparte, 1857 (guano or trek larities in plumage (Ogilvie-Grant 1898, Palmer shags).- My analysis establishesthe monophyly CORMORANT PHYLOGENY 891 of this basal clade with two synapomorphies,one the other shags of New Zealand and outlying of which (125) is unambiguous and related to islands, in particular, E. colensoiand E. ranfurlyi distal hind limb movement (Owre 1967) (Fig. 4). (Ogilvie-Grant 1898, 1905; Hutton 1903; Math- Some osteologicalcharacters (e.g., 6, 9 1) varied ews and Iredale 1921; Falla 1932, 1937). My among the three specimens of L. nigrogularis analysis showedN. campbelli to be osteologically that I examined, but most of the differenceswere distinctive (76) and a sister speciesto the New found in one of them (BMNH 1964.39.2), a Zealand blue-eyed shags(Fig. 4) consistent with probablejuvenile. The small number ofavailable Voisin’s (1973) designation of this form as a skeletons prevents knowing whether these dif- monotypic genus, Nesocarbo. ferenceswere ontogenetic or representednormal EuleucocarboVoisin, 1973 (New Zealand blue- variation within this form. No such variation eyed shags).- My analysis confirmed the mono- was seen in L. bougainvillii or L. capensis. typy of this genus with three character state The remaining genera in the Leucocarboninae changesrelated to flight (seeOwre 1967) (Fig. 4) are united by seven synapomorphies, four of all of which were homoplasious within shagsand which are unambiguous (43, 59, 77, 89) and re- convergent with cormorants, Microcarbo in par- lated to jaw movement and wing action (seeOwre ticular. The four taxa treated here are quite sim- 1967). The other three charactersvary complexly ilar looking, and, as in the other primitive shags, within the subfamily, showing repeatedreversals convergent for many external features. Skeletal and convergences. specimensare rare, and often misidentified (Sie- Notocarbo, n. gen. (blue-eyed shags).-This gel-Causey, unpubl. data). Because of this and clade of at least four speciesis supported by one the small number of reliable specimens, intra- unambiguous synapomorphy (19) found on the genericrelationships among this group should be mesethmoid, and one (79) homoplasious char- considered tentative until more detailed work is acter on the humerus that is convergent with possible. statesobserved in Compsohalieus(Fig. 4). Their StictocarboBonaparte, 1855 (cliff shags).-The possible functions are obscure. No pre-existing monophyly of this genus was confirmed by 13 genusname is available for this clade so I propose characterchanges, four of which (10,37,97, 108) Notocarbo(Noto- = G., “southern”; -carbo = L., are unambiguous (Fig. 5) and related to prey cap- coal, black: a generic root in the family) in light ture and femoral abduction (seeOwre 1967). The of their exclusively southern hemisphere distri- single most distinctive character in this clade is bution; I designateN. atricepsatriceps as the type the degree of dorsoventral compression of the speciesfor this genus. The diagnostic character cranium (10). This feature is most exaggerated for the genus is osteological:the prefrontal pro- in the North American taxa, S. pelagicusand S. cessof the mesethmoid is broadly produced into wile, where the dorsal surface of their crania is an anteriorally directed triangular projection. nearly planar. There are many convergenceswith External morphology of this group is similar to cormorants in characters related to swimming, other primitive shags,and discrimination by such especially to the marine cormorants, Compso- charactersat present is problematical. The genus halieus spp. Many of the synapomorphies ex- includes the South Georgia Shag(N. georgianus), pressedin this genusare associatedwith the hind the Antarctic Shag (N. bransfieldensis),the Im- limbs (e.g., 108, 116, 124) and related to their perial Shag(N. atriceps),and the Kerguelen Shag more upright resting posture (see Owre 1967). (N. verrucosus).Referred taxa are those currently The basal group of cliff shags(S. magellanicus, considered as subspeciesof N. atriceps or albi- S. pelagicus, S. wile, S. aristotelis) are charac- venter. terized by charactersof the cranium and tibiotar- The remaining shags are united by five syn- sus (22, 43, 133) that are likely associatedwith apomorphies. The only unambiguous synapo- foot flexion and feeding (see Owre 1967). The morphy is the attachment of M. iliotrochanter Red-leggedShag (S. gaimardz)of South America, on the femur (113) and relates to lessenedmo- and the Chatham Island and Spotted shags of bility of that element (Owre 1967). New Zealand (S. featherstoni and S. punctatus), Nesocarbo Voisin, 1973 (Campbell Island are the most derived Cliff Shags,and are distin- Shag).-The Campbell Island Shag (N. camp- guished from their congeners by 10 character belli), has long been considered closely related to changes, three of which (39, 94, 111) are un- 892 DOUGLAS SIEGEL-CAUSEY

TABLE 1. A Linnaean classification of the Recent ambiguous (seealso von Boetticher 1935). Func- species of Phalacrocoracidae.Genera and subfamily tions of these characters are obscure, but they limits were set using the conventions given in Wiley (198 1). Fossil taxa, and Recent taxa of Anhingidae, are may be associatedwith bill flexion, distal wing not included. * Speciesplacement referred without ex- movement, and stance (Owre 1967). One auta- amination; ** speciesextinct. pomorphy each establishesthe Chatham Island and Spotted shags as distinct species (cf. Mc- Family PhalacrocoracidaeBonaparte, 1855 Kitrick and Zink 1988) rather than subspecific Subfamily PhalacrocoracinaeBonaparte, 1855 forms of S. punctatus as was hypothesizedearlier Genus MicrocarboBonaparte, 1855 (Ogilvie-Grant 1905, Oliver 1930, Peters 193 1, M. africanus(Gmelin, 1789) Fleming 1939, van Tets 1976). M. coronatus(Wahlberg, 1855) M. pygmaeus(Pallas, 1773) [type by original designation] DISCUSSION M. niger (Vieillot, 18 17) M. melanoleucos(Vieillot, 1817) TAXONOMIC CLASSIFICATION Genus CompsohalieusRidgway, 1884 C. perspicillatus(Pallas, 181 l)** My findings indicate a greater diversity of form C. penicillatus(Brandt, 1837) [type by original in the family than previously recognized, and designation] prompt a revision of the classification of the C. harrisi (Rothschild, 1898) Phalacrocoracidae.Rather than submerge these C. neglectus(Wahlberg, 1855) differenceswith subgenericgroupings (cf. van Tets C. fuscescens(Vieillot, 1817) Genus HypoleucosReichenbach, 1852 1976), I have adopted Wiley’s (198 1) conven- H. olivaceus(Humboldt, 1805) [=brasilianus tions for a hierarchial classification. I recognize (Gmelin, 1789)] two subfamilies in the Phalacrocoracidae, nine H. auritus (Lesson, 1831) genera, and at least 35 speciesof cormorants and H. fuscicollis(Stephens, 1826)* H. varius(Gmelin, 1789) [type by original des- shags(Table 1). ignation] Systematic treatments of the past have placed H. sulcirostris(Brandt, 1837) the anhingas in various taxonomic categories, Genus PhalacrocoraxBrisson, ’ 1760 from a genus in the Phalacrocoracidae (Ripley P. carbo(Linnaeus, 1758) [type by tautonomy] 195 1, Ali and Ripley 1968) to a separatefamily P. caaillatus(Temminck and Schleael. 1850) Subfamily Leucocarboninae,new name- ’ (Becker 1986 and others).These effortshave dealt Genus LeucocarboBonaparte, 1857 primarily with the question of higher-order rank, L. nigrogularis(Ogilvie-Grant and Forbes, which ultimately is subjective. Yet undeter- 1899) mined are the phylogenetic relationships of the L. capensis(Sparrman, 1788) L. bougainvillii(Lesson, 1837) [type by subse- taxa comprising the Anhingidae, or their precise quent designation] relationship with the Phalacrocoracidae;a more GenusNotocarbo, n. gen. detailed survey of character and taxon variation N. verrucosus(Cabanis, 1875) in this group than was possible in this study is N. atricepsatriceps (Ring, 1828) [type by origi- needed to resolve the problematical characters nal designation] N. bransfieldensis(Murphy, 1936) discussedearlier. Until then, the most prudent N. georgianus(Lonnberg, 1906) course is to consider the Anhingidae of equal Genus NesocarboVoisin, 1973 rank to the Phalacrocoracidae. N. campbelli(Filhol, 1878) [type by monotypy] Cormorants (Phalacrocoracinae) can be gen- Genus EuleucocarboVoisin, 1973 E. carunculatus(Gmelin, 1789) [type by origi- eralized as heavy-bodied, deep-feeding, flat- and nal designation]* tree-nesting birds with indifferent or labored E. chalconotus(Gray, 1845) flight. The most distinct genus of cormorants, E. onslow(Forbes, 1893) perhaps of the entire family, are the microcor- E. colensoi(Buller, 1888) morants, Microcarbo. Unlike the rest of the fam- E. ranfurlyi (Ogilvie-Grant, 1901) Genus StictocarboBonaparte, 1855 ily, which are the size of a goose or duck, mi- S. magellanicus(Gmelin, 1789) crocormorants are nearer the size of a raven and 5’. pelagicus(Pallas, 1811) have quite distinct behaviors in courtship, feed- S. mile (Gmelin, 1789) ing, and reproduction (van Tets 1976). This group S. aristotelis(Linnaeus, 1761) S. gaimardi (Lesson and Garnot, 1828) is one of the few in the family previously rec- S. punctatus(Sparrman, 1786) [type by subse- ognized as a distinct genus (e.g., Peters 193 1); quent designation] however, Halietor is the junior synonym of the S. featherstoni(Buller, 1873) equally used Microcarbo (e.g., von Boetticher CORMORANT PHYLOGENY 893

1937, van Tets 1976). Therefore, I use Micro- the most studied members of the family (seevan carbo to designate this genus. Dobbin 1952, Palmer 1964, Siegel-Causey1985). Until recently (Crawford et al. 1982, Williams Recent studies indicate that geographic subdi- and Cooper 1983) M. coronatuswas considered vision within the New World species is much to be a subspecificform of M. africanus; in my greater than is currently recognized and that cer- results, unambiguous autapomorphies distin- tain populations may be reproductively isolated guishboth, and M. africanusis distinguishedfrom (Siegel-Causey, in press). Less is known about its sister speciesby two characters.Microcarbo the other members of this genus, especially the pygmaeusand M. niger have often been consid- Indian Cormorant, H. jiiscicollis, of which only ered conspecific (Ogilvie-Grant 1898, Deignan rudiments of distribution and behavior have been 1963) and in many accounts it is unclear which reported. speciesis being discussed.My analysis supports Macrocormorants, Phalacrocorax spp., are the specific standing of M. pygmaeus and M. among the largest extant speciesin the family. niger; less clear, however, is the relationship be- The Great Cormorant, P. carbo, is distributed tween M. melanoleucosand M. niger, but this over a vast area (eastern Canada to southern Af- may be a consequenceof the few skeletonsavail- rica, western Europe to China, and Australasia able. including some sub-Antarctic islands), and the Marine cormorants, Compsohalieusspp., are presentspecies definition includes forms that vary identified for the first time by this analysis; be- widely in certain character states(e.g., 82, 130). fore, members were grouped with other cormo- It is very likely that P. carbo, as currently rec- rants or shags,but rarely did any of the previous ognized, is a superspeciesdefined by superficial systematic treatments associateany of them to- similarities in plumage and external morphology gether. The constituent speciesare restricted to (cf. Witherby et al. 1940, Palmer 1964, Marion islands or coastlines adjacent to upwelling re- 1983). For this analysis, I used specimensof the gions of temperate and tropical oceans.Although nominal subspecies to represent P. carbo, al- species(C. penicillatus, C. fuscescens,C. neglec- though subsequent work indicated that certain tus) may become locally abundant, most have forms (e.g., P. c. lucidus,P. c. maroccanus,P. c. stronglycircumscribed distributions likely caused novaehollandiae) may be specifically distinct by factors related to breeding habitat, weak win- (Siegel-Causey, unpubl.). Little is known about ter dispersal,and diet (Rand 1960, Palmer 1964, P. capillatus, but its present distribution around Thomson and Morley 1966, Cooper 198 1, Ain- the Japan Sea, close resemblance to East Asian ley 1984, Siegel-Causey1985). One ofthe species populations of P. carbo, and behavior (Ogilvie- is flightless (C. harrisz). Another now extinct Grant 1898, Austin 1948, Yamamoto 1967, Ono species,C. perspicillatus,may have been flight- 1980) suggestthat it may be a population isolated less (Greenway 1958), but osteologicaland mor- during Pleistocene glaciations. phological evidence is equivocal (Shufeldt 19 15, Shags(Leucocarboninae) can be characterized Stegmann 1936). I did not include in this analysis as compact, pelagic, flat- and cliff-nesting birds the numerous autapomorphies associated with with fair flight abilities. The shagspossess many presumed or actual flightlessnesswhich are de- homoplasious characters related to diving and tailed elsewhere (Stejneger 1885, Stejneger and strong flying, and as a group, are the strongest Lucas 1889, Lucas 1895, Gadow 1902, Shufeldt fliers and most pelagic of the family. 1915, Ono 1980). Guano shags,Leucocarbo spp., are more cor- Continental cormorantsare entirely black, have morant-like in external morphology compared broad geographic distributions, and, including to the other members of this subfamily. The ex- certain microcormorants, are the only members ception is the Guanay, L. bougainvillii, whose of the family to inhabit the continental interiors. overall external appearance is quite similar to Mesocormorants, Hypoleucosspp., include two members of the other genera of shags.The So- of the most common cormorants of the New cotra Shag,L. nigrogularis,is the most primitive World: the Olivaceous and Double-crested cor- form in this subfamily, and externally resembles morants (H. olivaceus,H. auritus). These two the Cape Shag, L. capensis,in many features species have complementary distributions ex- (Archer 1937, Ripley and Bond 1966, Gallagher tending from Tierra de1 Fuego to Canada, from and Woodcock 1980). Little is known about its the Atlantic and Caribbean shoresto the Pacific, biology, but information on its behavior and and with Phalacrocorax carbo are undoubtedly nesting habits (Gallagher and Rogers 1978; Gal- 894 DOUGLAS SIEGEL-CAUSEY lagher et al. 1984; van Tets, pets. comm.) lend feed in nearshorewaters (Bernstein and Maxson support to its placement with the Leucocarbon- 1984, 1985). Subfossil evidence from Tierra de1 inae. The Cape Shag was regarded previously as Fuego supports a broader distribution, possibly having affinities with the Indian Cormorant (H. related to extensive postbreeding dispersal (Sie- fuscicollis) and Socotra Shag (Peters 1931) the gel-Causey, unpubl.). The South Georgia Shag Great Cormorant, P. carbo(von Boetticher 1937) (N. georgianus) is smaller and restricted to the or Bank and Brandt’s cormorants, C. neglectus Scotia Arc (Murphy 1936). The Kerguelen Shag and C. penicillatus (Dorst and Mougin 1979). (N. verrucosus)is smaller still, and appears to be Using overall similarity in courtship behavior, restricted to the Kerguelen Archipelago (Paulian van Tets (1976) was the first to place this species 1953, Voisin 1970). Previous records of this within the shags(Leucocarbo s.1.). The Guanay specieson the Crozet Islands were misidentifi- is the most derived member of this genus and cations (Voisin, pers. comm.). The interspecific possessesthe white abdomen and fleshy eye-ring relationships with Notocarbo are tentative, and of many of the other southern hemisphere shags. further study is proceeding to understand better The Socotra, Cape, and Guanay shagsare dis- their phylogeny and biogeography (Siegel-Cau- tinguished from all other shagsby their prefer- sey, unpubl.). ence to breed in very dense colonies, often in Blue-eyed shags of the New Zealand waters vast numbers on level ground, but other com- (Nesocarbocampbelli, Euleucocarbospp.) are ex- parative aspects of behavior have been little ternally quite similar (see Voisin 1973) but os- studied. teologically distinct from Notocarbo. All of these Blue-eyed shags (actually, the eye-ring is taxa have been regarded variously in the past as bluish- the irides are generally dark), Notocarbo subspecificforms of atriceps, albiventer, camp- spp., seem to be more pelagic than the other belli, or carunculatus(see Buller 1895, Falla 1937, generaof shags,and are found throughout coastal Falla and Stoke111945, Voisin 1973). I was able Fuego-Patagonia, Antarctica, and various sub- to discern autapomorphic characters for each Antarctic islands (Watson 1975, Siegel-Causey speciesexamined, but a more detailed study of 1986a). In addition to the specieslisted here, taxa all of the shagsof the Southern Ocean is essential currently regardedas subspeciesare referred into for the resolution of relationships in thesegroups. this genus (i.e., N. a. nivalis, N. a. melanogenis, All of these specieswere represented by only a N. a. purpurascens).Skeletons were not available single skeleton, so precise determination of re- for these forms, so precisedetermination of rank lationship among them is problematical for the and relationship is not yet possible. present. It has been unclear in the past whether the two The most derived genus of shagsis cliff shags, blue-eyed shagsof South America (atriceps, al- Stictocarbospp., comprising two groups of eco- biventer)were conspecificor distinct species(see logically differentiated birds. The basal group of Lataste 1893, Devillers and Terschuren 1978). species(S. magellanicus, S. pelagicus, S. urile, In a previous study on patterns of courtship be- S. aristotelis)are dark-colored, nest in a range of havior in various pairings between forms iden- habitats from gentle slopesof islands to perpen- tified in the field as atriceps or albiventer, I de- dicular cliff faces,and feed in the upper layers of tected no significant differences in type or marine coastalwaters (Witherby et al. 1940, Lack sequencebetween forms (Siegel-Causey 1986a). 1945, Snow 1960, Siegel-Causeyand Hunt 198 1, The extent of variation within morphological, Siegel-Causey 1986b). The most derived cliff osteological, and behavioral characters is such shags(S. gaimardi, S. punctatus, S. featherstom) that there are no diagnostic features to discrim- are the only members of the family with gray inate atricepsfrom albiventer. Therefore, I have plumage, and were recognized early on as being treated both forms under the senior synonym, closely related (Murphy 1936). These shagsare N. atriceps. able to stand nearly upright when at rest, and are The Antarctic Shag(N. bransfieldensis)is mas- able to breed on the steepestcliffs (Forbes 1893, sive (one of the largest shagsextant) and is dis- Koepckeand Koepcke 1953, Siegel-Causey1987). tinguished by at least four diagnostic characters Sibley and Ahlquist (unpubl.) have evaluated (Siegel-Causey, unpubl.). They are resident on the DNA-DNA hybridization of selectedtaxa of the Palmer Peninsula of Antarctica and islands the Class Aves (the “tapestry”) with the aim of along the Scotia Arc (Watson et al. 1971) and elucidating the phylogeny of birds. Their CORMORANT PHYLOGENY 895 coverage of the Pelecaniformes, and Phalacro- coracidae in particular, is quite sketchy, so com- parisonswith their resultsand the phylogeny pre- sented here is problematical. The Sibley and Ahlquist “tapestry” differs in that shagsare in- termixed with cormorants, but parallels my re- sults in distinguishing members of Microcarbo from those of Phalacrocorax and Hypoleucos. It is difficult at present, however, to interpret their a technique and results accurately (see Houde Hypothetical 1987) so meaningful comparisons are not yet Ancestor possible.

CRANIAL CHARACTERS Previous avian osteologistshave been concerned about problems inherent in using cranial char- acters(see Woolfenden 196 1) that may represent functional accommodations for food gathering, defense, etc., rather than particular evidence of Hypothetical phylogenetic relationship. To test for the possi- Ancestor bly confounding effectscaused by the integration FIGURE 6. Strictconsensus trees of segregated- - char- of cranial charactersin this analysis, I analyzed acter analysis.(a) Cranial charactersonly (n = 100, the character subsetsseparately (i.e., cranial vs. normalizedCF = 0.771. Rohlf’s CI = 0.860. length= postcranialcharacters). In general,the cranial and 86). Species relationships are polytomic, but generic assignmentsare unchangedfrom that shown in Figures postcranial subsetswere congruent with the en- 1-5, with the exception that C. perspicillatusand N. tire data set. With two exceptions, in both anal- verrucosusare unassigned.(b) Postcranial characters yses all specieswere grouped into genera iden- only (n = 100, normalized CF = 0.771, Rohlf’s CI = tically as found in the analysisusing all characters. 0.686, length = 132). Speciesrelationships are poly- Intrageneric relationships were not resolved tomic only in Notocarbo and Euleucocarbo,and all generic assignmentsare unchanged from that shown within most genera in these character subsets, in Figures l-5. however, and the number of equally parsimo- nious trees was correspondingly large. Interge- morants and shags.On the other hand, an anal- neric relationshipsin the postcranialanalysis (Fig. ysis employing only postcranial characters will 6b) were identical to those found in the analysis allow a fairly accurate generic assignment, and of all characters except that the Anhingidae, interspecific relationships will be resolved in se- Phalacrocoracinae, and Leucocarboninae were lected genera (e.g., Hypoleucos, Compsohalieus, resolved only to a polytomy. The analysis em- Mcrocarbo, Nesocarbo, Stictocarbo). Full reso- ploying exclusively cranial characters (Fig. 6a) lution will obtain only in using a full complement produced a much different topology than in the of characters. other analyses. Here, all except two speciesassignments (C. HOMOPLASY perspicillatus, N. verrucosus) were preserved,but Considerable convergence and reversal of char- genericrelationships differed somewhat from the acter statesis evident in the tree (Figs. 1-5) and analysis using all characters.Although the order by the consistencyofcharacter change(Appendix of genera is preserved within subfamilies, the 1). The extent to which some of these are related Phalacrocoracinae are paraphyletic using this to errors in assessinghomology is open to inter- subset,and the primitive shags(Leucocarbo spp.) pretation since anatomical studies in this family are polyphyletic. are rare and restricted to only a few taxa. The It appears that the head is prone to conver- majority of convergencesand reversals are as- gence as earlier workers surmised, and postcra- sociatedwith probable adaptations for flight and nial charactersare lessprone to convergence.An feeding (see Stolpe 1932, Owre 1967) and most analysis using cranial charactersalone will allow involve the wing elements (56, 64, 68, 79) ster- only generic assignment for most taxa of cor- num, quadrate, and mandible (6, 13, 21, 38,41, 896 DOUGLAS SIEGEL-CAUSEY

49). Some of these charactersco-occurred, pos- foraging efficiency. If the Phalacrocoracidaeuse sibly indicating functional adaptations to com- plumage as a means to reduce detection by prey, mon ecological problems. This effect is partic- then all-black speciesshould feed near the bot- ularly notable in comparisons between taxa, e.g., tom, white-bellied species near the top of the Anhinga and Euleucocarbo (83, loo), Phalacro- water column, and those with intermediate pat- corax and Microcarbo (14,45, 102, 116), and are terns (e.g., gray or light brown) highest of all. In supported in many casesby ecological associa- contrast to such functional explanations, it can tions (Fjeldsa 198 1, 1985). be useful to examine plumage patterns phylo- Reversals were just as numerous, and perhaps genetically. are more problematic. In some casesof reversal Cormorants (Phalacrocoracinae) are entirely (2, 13, 3 1, 40, 98, 124), the character change black and feed near the bottom, or near the 1% involved only a few taxa at a time, often as an incident light level (van Dobbin 1952, Ainley autapomorphic condition. Such charactersprob- 1984, Cairns 1986, Harkijnen 1988). The excep- ably reflect a simple, readily activated genetic tion, M. melanoleucos,is polymorphic with some mechanism, as may be the case for many of the populations having white abdomens, and is be- plumage charactersof the shags(cf. Baker 1973, lieved to feed on shoalingsprat in the upper water Jefferiesand Parslow 1976, Birkhead 1984). The layers (Falla and Stoke111945, Thomson and complex of repeated reversal of character states Morley 1966). The relation of plumage color to as evident within the Leucocarboninae (e.g., 6, feeding habitat, predicted by Simmons (1972), 14, 100, 102, 107, 110, 116, 130) are more per- is not supported in the Leucocarboninae. plexing, although some seem amenable to bio- Shagshave a greaterdiversity of plumage types geographic or ecological explanation (Siegel- than do cormorants. Two speciesof guano shags Causey, unpubl.). (L. nigrogularis, L. capensis)are entirely black, and feed in a similar manner as do cormorants PLUMAGE PATTERNS but higher in the water column (Gallagher and Plumage patterns in seabirdshave been hypoth- Woodcock 1980, Fumess and Cooper 1982). esized to be adaptive for thermoregulation Most other shags(L. bougainvillii,Notocarbo spp., (Hamilton and Heppner 1967, Heppner 1970) Nesocarbospp., Euleucocarbo spp.) are black with social communication (Armstrong 197 1, Ward white abdomens, and feed within the water col- and Zahavi 1973) reproductive isolating mech- umn from benthic to surface layers (Murphy anisms (Pierotti 1987) and foraging efficiency 1936, Kooyman et al. 197 1, Serventy et al. 197 1, (Simmons 1972, Siegfried et al. 1975). The most DufIy 1983). The cliff shags vary in plumage, recent scrutiny has been directed towards the from black with white abdomens (S. magellan- latter two areas. icus), to all-black (S. aristotelis, S. pelagicus, S. Pierroti (1987) examined the relationship be- urile), to light gray (S. gaimardi, S. punctatus, S. tween bill, face, and foot color among shagsand featherston& Less is known about their feeding cormorants grouped by areas of potential breed- habits, but they appear to have broad preferences ing sympatry to investigate whether external col- in feeding similar to other shags(Falla and Sto- oration served to reduce interspecific hybridiza- kell 1945, Lack 1945, Cawkell and Hamilton tion. If hybridization between species is more 196 1, Ainley et al. 198 1). In shags,plumage pat- likely than that between members of different terns track the phylogeny, with the most prim- genera, as his study assumed, then the hybrid- itive taxa being entirely black, the blue-eyed shags ization potential was overestimated becauseonly (s.1.)with white abdomens, and the most derived a single genus was used for the family Phalacro- cliff shagswith gray plumage. In the Leucocar- coracidae (e.g., his table 4, groups 2-S). Ques- boninae, functional adaptation such as camou- tions of rank aside, prereproductive isolating flage seemsless important in determining plum- mechanisms such as incompatible courtship be- age patterns than does phylogenetic history. havior, size, nesting and breeding preferences, etc., may have evolved much earlier before pres- ACKNOWLEDGMENTS ent speciesever bred sympatrically. I thank the following curatorsand museumsfor assis- Cairns (1986) surveyed the plumage color in tancein borrowing or examining specimens:A. Allison 61 speciesof pursuit-diving seabirds, and pos- (Bernice P. Bishop Museum), J. C. Barlow (Royal On- tulated adaptive significanceonly in referenceto tario Museum), G. F. Barrowclough(American Mu- CORMORANT PHYLOGENY 897 seum of Natural History), J. A. Bartle (National Mu- BULLER,W. L. 1888. A history of the birds of New seum of New Zealand), G. S. Cowles (British Museum Zealand. Wellington. 2nd ed. By author, London. of Natural History), J. Fjeldsa (Zoologisk Museum, BULLER.W. L. 1895. Notes on Phalacrocorux colen- Kobenhavn University), P. Haarhoff (South African soi,’ of the Auckland Islands, and P. onslowi, of the Museum), G. F. Mees (Rijksmuseumvan Natuurlijke), Chatham Islands. Trans. N.Z. Inst. 27:129-132. K. Ono (National Science Museum of Japan), R. B. CAIRNS,D. K. 1986. Plumagecolor in pursuit-diving Payne (Museum of Zoology, University of Michigan), seabirds:why do penguinswear tuxedos?Bird Be- J. V. Remsen (Museum of Zoology, Louisiana State hav. 6:58-65. Universitv). G. F. van Tets (Div. of Wildlife Research. CAWKELL,E. M., AND J. E. HAMILTON. 1961. The CSIRO), G. E. Woolfenden(University of South Flor: birds of the Falkland Islands. Ibis 103:1-27. ida), R. L. Zusi (U.S. National Museum of Natural COOPER,J. 1981. Biology of the Bank Cormorant, History). J. J. Becker, R. M. Chandler, J. Cracraft, J. Part 1: Distribution, population size, movements, Gauthier, W. Hoffman, P. S. Humphrey, B. C. Livezev, and conservation. Ostrich 52:208-215. M. C. McKitrick, J. L. Strauch, Jr., G. F. van Tets, COTTAM,P. A. 1957. The pelecaniform charactersof J.-F. Voisin. E. 0. Wilev. D. S. Wood. and R. M. Zink the Shoe-bill Stork Bulueniceps rex. Bull. Br. Mus. gave helpful comments’during initial drafts of this Nat. Hist. (Zool.) 5:5 l-71. manuscript.J. Neff, MD, provided me the opportunity CRACXAFT,J. 1985. Monophyly and phylogeneticre- to analyze the results.This researchwas supportedby lationships of the Pelecaniformes: a numerical the National ScienceFoundation grant BSR 84-07365 cladistical analysis.Auk 102:834-853. and by the Museum of Natural History, University of CRAWFORD,R.J.M., P. A. SHELTON,R. K. BROOKE, Kansas. AND J. COOPER. 1982. Taxonomy, distribution, population size, and conservationof the Crowned LITERATURE CITED Cormorant, Phalacrocorax coronutus. Gerfaut 72: 3-30. AINLEY,D. G. 1984. Cormorants, p. 92-101. In D. DEIGNAN,H. G. 1963. Checklist of the birds of Thai- Haley [ed.], Seabirdsof the Eastern North Pacific land. Bull. U.S. Natl. Mus. 226:1-263. and Arctic waters. Pacific Search Press, Seattle. DEVILLERS,P., AND J. A. TERSCHUREN.1978. Rela- AINLEY, D. G., D. W. ANDERSON,AND P. R. KELLY. tionships between the Blue-eyed Shags of South 1981. Feeding ecology of marine cormorants in America. Gerfaut 68:53-86. southwesternNorth America. Condor 83: 120-l 3 1. VANDOBBIN, W. H. 1952. The food ofthe cormorant ALI, A. S., AND S. D. RIPLEY. 1968. Handbook of the in the Netherlands. Ardea 40:1-63. birds of India and Pakistan. Vol. 1:35-43. DORST.J., AND J.-L. MOUGIN. 1979. Familv Phala- ARCHER,G. F. 1937. The birds of British Somaliland crocoracidae,p. 162-179. In E. Mayr and G. W. and the Gulf of Aden. Vol. 1. Gurney and Jackson, Cottrell [eds.],Checklist ofbirds ofthe world. Har- London. vard Univ. Press, Cambridge. ARMSTRONG,E. A. 1971. Social signallingand white DUFFY,D. C. 1983. The foragingecology of Peruvian plumage. Ibis 113:534. seabirds.Auk 100:800-8 10. AUSTIN,0. L. 1948. The birds of Korea. Bull. Mus. DULLEMEIJER,P. 195la. The correlation between Comp. Zool. lOl:l-300. muscle system and skull structurein Phulacroco- BAKER,A. J. 1973. Genetics of plumage variability rux turbo sinensis (Shaw and Nodder), I. K. Ned. in the Variable Oystercatcher (Huekzatopus uni- Akad. Wetensch. Ser. C 54:247-258. color). Notomis 20:330-345. DULLEMELIER,P. 1951b. The correlation between BAUMEL,J. J., A. S. KING, A. M. LUCAS,J. E. BREAZILE, muscle system and skull structurein Phalacroco- AND H. E. EVANS. 1979. Nomina Anatomica rux curb; sinensis (Shaw and Nodder), II. K. Ned. Avium. Academic Press, New York. Akad. Wetensch. Ser. C 54:4OwO4. BECKER,J. J. 1986. Reidentification of “Phalucro- cord subvoluns Brodkorb as the earliest record DULLEMEIJER,P. 195lc. The correlation between of Anhingidae. Auk 103:804-808. muscle system and skull structurein Phalucroco- BERNSTEIN,N. P., ANDS. J. MAXSON. 1984. Sexually rax curb0 sinensis (Shaw and Nodder), III. K. Ned. distinct daily activity patterns of Blue-eyed Shags Akad. Wetensch. Ser. C 54:533-536. in Antarctica. Condor 86: 15l-l 56. DULLEMEIJER,P. 1952. The correlationbetween mus- BERNSTEIN,N. P., AND S. J. MAXSON. 1985. Repro- cle system and skull structure in Phulacrocorax ductive energeticsof Blue-eyed Shagsin Antarc- turbo sinensis (Shaw and Nodder). IV. K. Ned. tica. Wilson Bull. 97:450-462. Akad. Wetensch. Ser. C 55:95-102: BIRKHEAD,T. R. 1984. Distribution of the bridled FALLA,R. A. 1932. New Zealand cormorants in the form of the Common Guillemot Uris ualge in the collection of the Auckland Museum, with notes North Atlantic. J. Zool. (Lond.) 202: 165-l 76. on field observations. Rec. Auckl. Inst. Mus. 1: VON BOETTICHER,H. 1935. Der Gaimardische Bunt- 139-154. kormoran. Vogel femer Lander 1935:81-83. FALLA,R. A. 1937. Birds. British, Australianand New VON BOETTICHER,H. 1936. Die geographischeVer- Zealand Antarctic Research Expedition 1929- breitung der Kormorane. Omit. Monatschr. 61: 1931. Reports (B)II: l-304. 101-l 15. FALLA,R. A., AND G. STOKELL. 1945. Investigation VONBOETTICHER, H. 1937. Zur Systematik der Kor- of the stomach contents of New Zealand fresh- morane. Festschr. f. Prof. Dr. Embrik Strand III: water shags.Trans. R. Sot. N.Z. 74:320-33 1. 586-594. FJELDSA,J. 1981. A comparisonof bird communities 898 DOUGLAS SIEGEL-CAUSEY

in temperate and subarcticwetlands in northern with special reference to cormorant (Phalucroco- Europeand the Andes. Proc. SecondNordic Congr. rux turbo) and shag(P. aristotelis). J. Anim. Ecol. Omithol. (1979):101-108. 14:12-16. FJELDSA,J. 1985. Classification of waterbird com- LAMBRECHT,K. 1933. Handbuch der Palaeomithol- munities in southeasternAustralia. Emu 85: 14 l- ogie, Pt. 1. Gebruden Bomtraeger Verslag, Berlin. 149. LATASTE,F. 1893. Liste d’oiseaux recuellis par M. FLEMING,C. A. 1939. Birds of the Chatham Islands. le Dr. Federico Delfin dans le dttroit de Magellan Emu 38:380413. et environs, et offerts par lui au Mu&e Zoologique FORBES,H. 0. 1893. On the birds inhabiting the de l’Ecole de Mtdecine de Santiago. Actes Sot. Chatham Islands. Ibis 1893:52l-54 1. Sci. Chili 3:121-123. FURNESS,R. W., AND J. COOPER. 1982. Interactions LIVEZEY,B. C. 1986. A phylogeneticanalysis of Re- between breeding seabird and pelagic fish popu- cent anseriform generausing morphological char- lations in the southernBenguela region. Mar. Ecol. acters. Auk 103:?‘37-754. - _ - Prog. Ser. 8:243-250. LOWE.P. R. 1926. More notes on the auadrate as a GADOW,H. 1902. The wingsand the skeletonofPhal- factor in avian classification.Ibis 19!26:152-l 88. acrocorax harrisi. Nov. Zool. 9: 169-l 76. LUCAS,F. A. 1895. The cranium of Pallas’s Cor- GALLAGHER,M. I)., AND T. D. ROGERS. 1978. On morant. Proc. U.S. Natl. Mus. 17:83-94. the breeding birds of Bahrain. Bonn. Zool. Beitr. MADDISON,W. P., M. J. DONOGHUE,AND D. R. MAD- 29:5-17. DISON. 1984. Outgroup analysisand parsimony. GALLAGHER,M. D., AND M. W. WOODCOCK.1980. Syst. Zool. 33:83-103. The birds of Oman. Quartet, London. MARION,L. 1983. Probltmes biogCographiques,tcol- GALLAGHER.M. D.. D. A. SCOTT.R.F.G. ORMOND. ogiques et taxonomique posCspar le grand cor- R. J. CANNER,AND M. C. JENNINGS.1984. The moran P. carbo. Rev. Ecol. (Terre et Vie) 38:65- distribution and conservationof seabirdsbreeding 99. on the coasts and islands of Iran and Arabia, { MATHEWS,G. M., AND T. IREDALE. 1913. A reference 421-456. In J. P. Croxall, P.G.H. Evans, and R. list of the birds of New Zealand, pt. II. Ibis 1913: W. Schreiberreds.], Status and conservationof the 402-435. world’s seabirds.int. Count. Bird Preserv. Tech. MATHEWS,G. M., AND T. IREDALE. 1921. The nature Publ. 2. ICBP. Cambridge. of the New Zealand avifauna. Emu 20:210-221. GREENWAY,J. C.,‘JR. 1958: Extinct and vanishing MAYR, E., ANDG. W. COTTRELL.1979. Check-list of birds of the world. Am. Comm. Int. Wild Life birds of the world. Museum of Comparative Zo- Prot. Spec. Publ. 13. New York. ology, Harvard Univ., Cambridge, MA. HALL, R. 1920. The Tasmanian and New Zealand MCKITRICK,M. C., AND R. M. ZINK. 1988. Species groups. Emu 19~275-287. conceptsin ornithology. Condor 90:1-14. HAMILTON,W. J., AND F. HEPPNER. 1967. Radiant MURPHY,R. C. 1936. Oceanic birds of South Amer- solar energy and the function of black homeo- ica, vol. 2. Am. Mus. Nat. Hist., New York. therm pigmentation: a hypothesis. Science 155: OGILVIE-GRANT,W. R. 1898. Order XVII. Steganop- 196-197. odes,p. 329-413. Catalogofbirds in the collection H~~RK~~NEN,T. J. 1988. Food-habit relationshiu of of the British Museum. Vol. 26. harbour sealsand black cormorants in Skage-&ak OGILVIE-GRANT,W. R. 1905. On the birds procured and Katteaat. J. Zool.. (Land.) 2 14:673-68 1. by the Earl of Ranfurly in New Zealand and the HEPPNER,F. 1970. The n%abolic significanceof dif- adjacent islands. Ibis 1905:543-575. ferential absorptionof radiant energyby black and OLIVER,W. R. B. 1930. The New Zealand double- white birds. Condor 72150-59. crested shags;with description of a new species. HOUDE, P. 1987. Critical evaluation of DNA hy- Trans. N.Z. Inst. 61:138-139. bridization studiesin avian systematics.Auk 104: 17-32. OLSON,S. L. 1985. The fossil record of birds. Avian Biol. 8:80-238. HOWARD,H. 1929. The avifauna of Emeryville shell- mound. Univ. Calif. Publ. Zool. 32:301-394. ONO,K. 1980. Comparativeosteology of three species HUTTON,F. W. 1903. The cormorants of New Zea- of Japanesecormorants of the genusPhalucroco- land: a study in variation. Emu 3: l-8. rux (Aves, Pelecaniformes). B&. Nat. Sci. Mus. Janan. Ser. C (Geol. and Paleont.) 6: 129-l 5 1. JEFFERIES,D. J., AND J.L.F. PARSLOW.1976. The ge- netics of bridling in guillemots from a study of OWRE,~0.T. 196;. Adaptations for locomotion and hand-reared birds. J. Zool. (Lond.) 179:41l-420. feeding in the Anhinga and the Double-crested KOEPCKE,H.-W., AND M. KOEPCKE.1953. Die war- Corm&ant. Ornithological Monograph No. 6. men Feuchtuftwiisten Perus (Eine Einteilung in American Omitholoeists’ Union. Washington.DC. Lebensstgtenunter besonderer Berucksichtinunn PALMER,R. S. 1964. Handbook ofNorth imekican der Viigel). Bonn. Zool. Bietr. 4:79-146. - - birds. Vol. 1. Loonsthrough Flamingos. Yale Univ. KOOYMAN,G. L., J. P. SCHROEDER,D. G. GREENE,AND Press,New Haven. V. A. SMITH. 1971. Effectsof deep dives on pen- PAUFAN, P. 1953. Pinnipedes, CCtacCs,Oiseaux des guins and Blue-eyed shags.U.S. Antarct. J. 6:95. Iles Kerguelen et Amsterdam. M&m. Inst. Sci. KURODA,N. 1922. Notes on the birds of Tsutshima Madagascar8: 11l-23 1. and iki Islands, Japan. Ibis 1922:75-105. PETERS,J.-L. 1931. Check-list of birds of the world. LACK, D. L. 1945. Ecology of closely related species Vol. 1. Harvard Univ. Press, Cambridge. CORMORANT PHYLOGENY 899

PIEROTTI,R. 1987. Isolating mechanismsin seabirds. gestorbenenScharbe Phalacrocorax perspicillatus Evolution 41559-570. Pall. Omithol. Monatsber. 44: 140-l 53. PYCRAFT,W. P. 1898. Contributions to the osteology STEJNEGER,L. 1885. Results of ornithological explo- of birds. Part I. Steganopodes.Ibis 1898:80-100. rations in the Commander Islands and in Kamt- RAND, R. W. 1960. The biology of guano-producing schatka.Bull. U.S. Natl. Mus. 9:1-382. sea-birds. 3. The distribution, abundance, and STEJNEGER,L., ANDF. A. LUCAS. 1889. Contribution feeding habits of the cormorants Phalacrocoraci- to the history ofPallas’ Cormorant. Bull. U.S. Natl. dae off the south-westerncoast of Cape Province. Mus. 12:83-94. S. Afr. Div. Fish. Inv. Rep. 42:1-32. STOLPE,M. 1932. Physiologisch-anatomischeUnter- RIPLEY,S. D. 1951. Birds collected and noted round suchungeniiber die hintere Extremitgt der Viigel. Dhahran, Saudi Arabia, and Bahrein Island. Pos- J. Omithol. 80: 163-247. tilla 9:1-11. SWOFFORD,D. L. 1984. PAUP: phylogeneticanalysis RIPLEY,S. D., AND G. M. BOND. 1966. The birds of using parsimony. Illinois Nat. Hist. Surv. (Mim- Socotra and Abd-el-Kuri. Smithson. Misc. Coll. eo). 151:1-37. TECHNAU,G. 1936. Die Nasendriiseder VGgel. Zug- SAIFF,E. I. 1978. The middle ear of the skull of birds: leich ein Beitragzur Morphologie der Nasenhiihle. the Pelecaniformes and Ciconiiformes. Zool. J. J. Omithol. 845 1l-6 17. Linn. Sot. 63:315-370. THOMSON,J. A., AND N. W. MORLEY. 1966. Phys- SERVENTY,D. L., V. N. SERVENTY,AND J. WARHAM. iological correlatesof habitat selectionin Austra- 1971. The handbook of Australian sea-birds. lian cormorants. Emu 66: 17-26. Reed, Sydney. VANTETS, G. F. 1965. A comparative study of some SHARPE,R. B. 1899. A hand-list of the genera and social communication patterns in the Pelecani- speciesof birds. Vol. 1. British Museum (Natural formes. Ornithological Monograph- _ No. 2. Amer- History), London. ican Ornithologists’ Union, Washington, DC. SHUFELDT,R. W. 1888. Observationsupon the Order VAN TETS,G. F. 1976. Australasia and the origin of Tubinaresand Steganopodes.Proc. U.S. Natl. Mus. shagsand cormorants, Phalacrocoracidae.Proc. 11:253-315. XVI Int. Omithol. Congr. (1974):121-124. SHUFELDT,R. W. 1902. The osteologyof the Stega- VOISIN.J.-F. 1970. On the soecific statusof the Ker- nopodes. Mem. Carnegie Mus. 1:109-223. g&en Shag and its affinities. Notomis 17:286- SHUFELDT,R. W. 1915. Comparative osteology of 290. Harris’s FlightlessCormorant (Nannopterum har- VOISIN, J.-F. 1973. Notes on the Blue-eyed Shags rise]. Emu 15:86-l 13. (genusLeucocarbo Bonaparte). Notomis 20:262- SIEGEL-CAUSEY,D. 1985. Cormorants, p. 60-62. In 271. C. M. Perrins and A.L.A. Middleton [eds.], The WARD, P., ANDA. ZAHAVI. 1973. The importance of encyclopediaof birds. Facts On File, New York. certain assemblagesof birds as “information- SIEGEL-CAUSEY,D. 1986a. The courtship behavior centres” for food finding. Ibis 115:517-534. and mixed-species pairing of King and Imperial WATSON,G. E. 1975. Birds of the Antarctic and Sub- shags(Phalacrocorax albiventer and P. atriceps). antarctic.American Geophysicists’ Union, Wash- Wilson Bull. 98:571-580. ington, DC. SIEGEL-CAUSEY,D. 1986b. The behaviour and phy- WATSON,G. E., J. P. ANGLE, P. C. HARPER,M. A. logenetic status of the Magellanic Cormorant BRIDGE,R. P. SCHLATTER,W.L.N. TICKELL,J. C. (Phalacrocorax magellanicus). Notomis 33:249- BOYD,AND M. M. BOYD. 1971. Birds of the Ant- arctic and Subantarctic.Antarctic Mao Folio Se- __751I. SIEGEL-CAUSEY,D. 1987. The behaviour of the Red- ries, Folio 14. American Geophysicists’ Union, leggedCormorant (Phalacrocoraxgaimard~]. Not- Washington, DC. omis 34: l-9. WETMORE,A. 1934. A systematic classificationfor birds of the world, revised and amended. Smith- SIEGEL-CALJSEY,D. In press. Cranial pneumatization son. Misc. Coll. 89:1-l 1. in the Phalacrocoracidae.Wilson Bull. WILEY, E. 0. 1981. Phylogenetics.The theory and SIEGEL-CAUSEY,D. In press. Extra rectrices in Oli- practice of phylogenetic systematics.John Wiley vaceousCormorants. Bull. Br. Omithol. Club. and Sons, New York. SIEGEL-CAUSEY,D., AND G. L. HUNT, JR. 198 1. Co- WILLIAMS,A. J., ANDJ. COOPER.1983. The Crowned lonial defensebehavior in Double-crestedand Pe- Cormorant. Breedingbiology, diet, and offspring- lagic cormorants. Auk 98:522-53 I. reduction strategy.Ostrich 54:2 13-2 19. SIEGFRIED,W. R., A. J. WILLIAMS,P.G.H. FROST,AND WITHERBY,H. F., F.C.R. JOURDAIN,N. F. TICEHURST, J. B. KINAHAN. 1975. Plumage and ecology of AND B. W. TUCKER. 1940. Handb. Br. Birds 4: cormorants. Zool. Afr. 10:183-l 92. l-1A SIMMONS, K.E.L. 1972. Some adaptive features of WOOD, D. S., AND G. D. SCHNELL. 1986. Revised seabirdplumage types. Br. Birds 65:465479,5 lC- worldinventory ofavian skeletalspecimens, 1984. 521. American Ornithologists’ Union and Oklahoma SNOW,B. K. 1960. The breeding biology of the shag Biological Survey, Norman, OK. Phalacrocorax aristotelis on the Island of Lundy, WOOLFENDEN,G. E. 1961. Postcranial osteology of Bristol Channel. Ibis 102:554-575. the waterfowl. Fl. State Mus. Bull. 6: l-l 29. STEGMANN,B. 1936. iiber das Flugvermijgender aus- YAMAMOTO, H. 1967. Phalacrocorax capillatus as a 900 DOUGLAS SIEGEL-CAUSEY

breedingbird on Iwate coast,Honshyu. Misc. Rep. midline; (b) extends laterally to edge of orbit. Yamashina Inst. Ornithol. 5:48-60. (Technau 1936: pl. IV, fig. 2) CI = 0.25. 15. Nasal gland depression:(a) posterior margin lin- APPENDIX 1 ear; (b) posterior margin distinctly serrate. (Sie- gel-Causey,in press) CI = 1.0. DESCRIPTION OF CHARACTERS 16. Mesethmoid: (a) unfenestrated;(b) fenestratedby The 137 charactersused in this analysisare numbered one or more foramina. (Shufeldt 1902: pl. IV) CI and groupedanatomically. Characterstates are lettered = 0.5. and correspondto the changesshown in Figures 2-4. 17. Prefrontal (lachrymal) process of mesethmoid: Plesiomorphic conditions are designated“a” and de- (a) with posterior and/or anterior accessory rived characterstates are ordered alphabetically.Char- flanges;(b) without accessoryflanges. (Shufeldt acter transformation serieswere assumedlinear; char- 1888: fig. 1) CI = 1.0. acters followed by “U” were analyzed as unordered. 18. Prefrontal (lachrymal) process of mesethmoid: Referencesindicate depictions of characters,but most (a) simple, unconnected; (b) superiorally con- are illustrated without identification. Taxa with prob- nectedto internal surfaceof orbit by stronglateral lematic state determinations are listed in parentheses flange. (Pycraft 1898: PI. 8, fig. 3) CI = 1.0. after the correspondingcharacter. Consistency indices 19. Prefrontal (lachrymal) process of mesethmoid: (CI) follow each character. Anatomical terminology (a) anterior surfacenormally produced,often into follows Howard (1929) and Owre (1967). a thin spine; (b) broadly produced into a trian- gular projection. CI = 1.0. 20. Palatine processof prefrontal (lachrymal): (a) an- SKULL teroventral surface produced, (b) surface exca- 1. Temporal crests:(a) separatedat midline; (b) meet vated into distinct cup; (c) ventral surfacedeeply sagittally. CI = 0.5. excavated. CI = 1.0. 2. Temporal and nuchalcrests: (a) separatedat mid- 21. Attachment of M. depressusmandibulus: (a) by line by sagittal crest; (b) meet sagittally, without distinct fossa:(b) laterallv nroduced.(Owre 1967: sagittal crest. (Ono 1980: fig. l-3a) CI = 0.05. fig. 50) CI = 5.5. . . . 3. Sagittalcrest: (a) producedat midline; (b) absent 22. Attachment of M. protractus pterygoideus: (a) or obsolete. CI = 0.5. fossa emarginate; (b) fossa with strongly pro- 4. Postorbital process:(a) simple and small; (b) ex- duced anterior ridge. CI = 1.O. cavated posterolaterally.(Shufeldt 1902: pl. VI, 23. Attachment of M_ rectus capitus: (a) posterior; fig. 25) CI = 0.33. (b) anterior to vaeusforamen. (Shufeldt 1902: ul. 5. Postorbital process:(a) produced into triangular Vi, fig. 29) CI =‘i .O (variable‘in atriceps). _ tuberosity; (b) connected by strong lateral shelf 24. Attachment of M. pterygoideusventralis fasci- to attachment of M. protractus pterygoideus. culus: (a) oriented in line with lateral edge of (Shufeldt 1902: pl. IV, fig. 14) CI = 0.5. basitemporal plate; (b) oriented sag&tallyto lat- 6. Secondpostorbital (temporal) process:(a) absent eral edge. (Owre 1967: fig. 52) CI = 1.0. or miniscule; (b) prominent, oriented on dorsov- 25. Eustachian canal: (a) lateralmost margin of an- entral surface;(c) prominent, oriented on later- terior edge closed or nearly so; (b) lateralmost almost surface.(Dullemeijer 195la: fig. 1b) CI = margin is broadly open. (Cracraft 1985: fig. 2) CI 0.4 (variable in nigrogularis). = 0.5. 7. Nasal prominence: (a) proximalmost width is 26. Basitemporal plate: (a) lateral edge between eus- greater-thanhalf the width of the ventral surface tachian canaland attachmentof M. rectuscapitus of maxilla; (b) is less than half the width. (Ono not prominent; (b) lateral edge is strongly pro- 1980: fig. l-2,-3) CI = 1.0. duced. (Dullemeijer 1951~: fig. 19) CI = 1.0. 8. Nasal shelf of premaxillary: (a) dorsally convex; 27. Foramen trigeminalis prooticus:(a) posterior; (b) (b) excavated: (c) ulanar. CI = 1.0. U anterior to upper tvmpanic recess. (Saiff 1978: 9. Maxillary: (a)‘nos&l groove superficial;(b) deep- fig. 2) CI = 0:5 (variable in hhingaj. ly excavated. (Pycraft 1898: pl. 8, fig. 3) CI = 28. Foramen trigeminalis prooticus: (a) smaller; (b) 0.33. largerthan upper tympanic recess.CI = 0.5 (vari- 10. Cranium: (a) approximately long as deep;(b) dor- able in Anhinga). soventrally compressed.(Ono 1980: fig. l-2,-3) 29. Upper tympanic recess:(a) open; (b) with acces- CI = 1.0. sory transversestrut. CI = 1.0. 11. Frontal: (a) preorbital lengthis subequalto width; 30. Upper tympanic recess: (a) extends between (b) length is much greater than sagittalwidth. CI quadrate articular surfaces;(b) restricted to pos- = 0.5. terior margin of articular surfaces. (Saiff 1978: 12. Nasal glanddepression on ventral surfaceof fron- fig. 2) CI = 1.0. tal: (a) small, barely extending into orbit; (b) 31. Palatine: (a) with well-defined lateral angles:(b) moderate, not longer than half the length of the are laterally rounded. (Cottam 1957: fig: lc) Ci frontal; (c) large, reaching posterior margin of = 0.5. orbit. (Technau 1936: figs. 14 and 23) CI = 0.5. 32. Palatine: (a) distinctly narrowed posteromedially 13. Nasal gland depression:(a) shallow; (b) strongly into abrupt neck at pterygoidal articulation; (b) excavated. CI = 0.5. pterygoidalarticulation is broad. (Shufeldt 1902: 14. Nasal gland depression:(a) medially restrictedto pl. VI, figs. 27 and 29) CI = 1.0. CORMORANT PHYLOGENY 901

33. Palatine: (a) articulatesimply with pterygoids;(b) pseudotemporalis:(a) very small; (b) subequalto with strong posterodorsalprocess. CI = 1.O. length of ventral external opening of fossaaditus; 34. Palatine: (a) ventral surface planar; (b) strongly (c) greaterin lengththan external openingof fossa concave. (Pvcraft 1898: ~1. 7. fia. 3) CI = 1.0. aditus. (Owre 1967: fig. 54b) CI = 0.67 (variable 35. Supraoccipitalcondyle (grticula~ionof occipital in Anhingu). style): (a) anterior; (b) posterior to occipital con- 51. Attachment of M. adductusmandibulae intemus dyle. CI = 0.5. pseudotemporalis: (a) distinct from external opening of fossa aditus; (b) connectsto external QUADRATE opening. (Dullemeijer 195la: fig. 9b) CI = 1.O. 36. Quadrate: (a) pneumatic; (b) apneumatic. CI = 52. Insertion of M. depressusmandibulus: (a) lateral 0.5. insertion slightly excavated, dorsal insertion in- 37. Orbital process: (a) separate from supraorbital distinct; (b) lateral insertion convex, dorsal in- process;(b) juncture to shaft is excavated me- sertion robust. CI = 0.5. dially into a deep fossabounded by a strongridge 53. Dorsal mandibular groove: (a) arisesposteriad to running superiorlyto supraorbitalprocess: (Dtk medial mandibular groove; (b) arisessubequally lemeijer 1951b: fig. 12) CI = 1.0. with medial mandibular groove. CI = 1.O. 38. Supraorbitalprocess: (a) reduced or obsolete;(b) 54. Symphysis:(a) accessorybone stronglyproduced strongly produced. (Lowe 1926: fig. 3) CI = 0.5. into commissure; (b) accessorybone does not 39. Supraorbitalprocess: (a) conical; (b) lateral shelf. reach into commissure. (Ono 1980: fig. 2) CI = CI = 1.0. 0.5.

MANDIBLE CORACOID 40. Caudal fossa: (a) entire; (b) bisected by distinct 55. Anterior intramuscularline: (a) changesdirection vertical ridge. (Dullemeijer 1951 b: fig. 16) CI = below level parallel with superior edge of ster- 0.5. nocoracoidalprocess; (b) changesdirection in line 41. Attachment of M. adductusmandibulae externus with the stemocoracoidalprocess; (c) changesdi- profundus: (a) absent, reduced, or indistinct; (b) rection above stemocoracoidal process. (Ono produced into robust tuberosity on dorsomedial 1980: fig. 4) CI = 1.0. surfaceof external articular process.(Owre 1967: 56. Anterior intramuscularline: (a) intersectssternal fig. 54d) CI = 1.0. facet nearest internal distal angle; (b) intersects 42. Attachment of M. adductusmandibulae extemus sternal facet nearest stemocoracoidal process. superficialis:(a) absent,reduced, or indistinct; (b) (Lambrecht 1933: fig. 104) CI = 0.5. expressedas strongtransverse ridge. (Dullemeijer 57 Subfurcularangle: (a) narrow, less than 90”; (b) 195la: fig. 3b) CI = 1.0. broad, greater than 90”. CI = 0.5. 43. Attachment of M. adductusmandibulae internus 58. Subfurcularfossa: (a) reducedor absent;(b) deep- pterygoideus:(a) reduced or present as indistinct ly excavated. CI = 1.0. impression on ventromedial surface of prearti- 59 Attachment of M. supracoracoideus:(a) planar; cular; (b) impression is deeply excavated; (c) (b) excavatedinto distinct fossa,immediately ad- impression deeply excavated and bilobed. (Dul- jacent to acrocoracoid surface; (c) excavation lemeijer 195la: fig. 9b) CI = 1.0. broad. (Owre 1967: fig. 12) CI = 1.O (variable in 44. Attachment of M. adductusmandibulae intemus Anhinga). pterygoideus:(a) posterodorsaledge with normal 60. Brachial tuberosity: (a) interior margin simple; margins; (b) posterodorsal edge produced into (b) interior margin produced into ridge superior strong ridge. CI = 1.O. to attachment of M. supracoracoideus;(c) inte- 45. Attachment of M. adductusmandibulae intemus rior margin rugose.CI = 1.O (variable in Anhin- pterygoideus:(a) anteriormostinsertion is planar; ga). u (b) anteriormost insertion is into distinct fossa, 61. Brachial tuberosity: (a) normally produced; (b) bounded by strong dorsal ridge. CI = 0.5. very deep scar between it and glenoid facet. CI 46. Fossa aditus: (a) greatly reduced in size, smaller = 1.0. than oval crista of coronoid process,restricted to 62. Accessoryfossa: (a) absent,reduced, or indistinct; upper third of prearticular; (b) larger than oval (b) deep, subcircularpit just superior to attach- crista, occursin lower half of prearticular. (Dul- ment ofM. supracoracoideus;(c)excavation ovo- lemeijer 195la: fig. 9a) CI = 1.0. idal and shallow. (Baumel et al. 1979: fig. 6b) CI 47. Internal attachment of M. adductusmandibulae = 1.0. u pseudotemporalisin fossa aditus: (a) indistinct 63. Procoracoid:(a) entire; (b) bisected by irregular or absent; (b) present as distinct line. CI = 1.0. canal;(c) cleaved into two separateprominences. 48. Attachment of M. adductusmandibulae intemus CI = 1.0. pseudotemporalis: (a) present as indistinct impression; (b) strongly produced. CI = 1.0. HUMERUS 49. Attachment of M. adductusmandibulae intemus 64. Ligamental furrow: (a) does not reach head; (b) pseudotemporalis:(a) rugose ridge; (b) robust distinctly notcheshead. CI = 0.5. ledge. (Dullemeijer 195la: fig. 9b) CI = 0.33 65. Ligamentalfurrow: (a) entire lengthofequal depth; (variable in Anhingu). (b)_ lateralmost^ part excavated into pit. CI = 1.O. 50. Attachment of M. adductusmandibulae internus 66. Ltgamentallurrow: (a) mesial marginssimple; (b) 902 DOUGLAS SIEGEL-CAUSEY

mesial marginmarked by stronglyproduced crest. distally into distinct triangular shape.(Baumel et (Ono 1980: fig. 7) CI = 1.0. al. 1979: fig. 9a) CI = 1.0. 67. Deltoid shaft: (a) medioproximal surface con- cave;(b) medioproximal surfacestrongly convex. CARPOMETACARPUS CI = 1.0. 89. Metacarpus II: (a) distal tuberosity absent, re- 68. Deltoid shaft:(a) lateroproximal surfaceconcave; duced, or obsolete; (b) tuberosity strongly pro- (b) lateroproximal surface more or less convex. duced. (Ono 1980: fig. 10) CI = 1.O (variable in CI = 0.33. Anhinga). 69. Deltoid shaft: (a) laterodistal surfaceconcave; (b) 90. Metacarpus III: (a) anterior carpal fossa absent, laterodistal surface strongly convex. CI = 1.O. reduced,or indistinct; (b) fossadeeply excavated. 70. Median crest: (a) linearly joins deltoid shaft; (b) CI = 0.5 (variable in Anhinga). strongly indented at junction with deltoid shaft. 91. Metacarpus III: (a) nroximal width lessthan half (Ono 1980: fig. 7) CI = 0.5. the width of metacarpusII; (b) proximal width 71. Bicipital crest:(a) distal aspectis smoothlycurved; greater than half the width of metacarpusII. CI (b) distal aspect sharply indented. CI = 1.0. = 1.O (variable in nigrogularis). 72. Bicipital furrow: (a) surfaceplanar; (b) with strong 92. Internal ligamental fossa: (a) nearly emarginate; lateral ridge. CI = 1.0. (b) proximalmost margins deeply excised. CI = 73. Bicipital furrow: (a) normally indented; (b) ex- 1.0. tremely excised. CI = 1.0. 93. Internal ligamental fossa: (a) surfacejust proxi- 74. Capital groove: (a) entire length of equal depth; mal is planar; (b) surfacejust proximal marked (b) transverseridge forms deep pit proximalmost by deep fossa. CI = 1.0. to head. (Ono 1980: fig. 7) CI = 1.O. 94. Pisiform process:(a) separate;(b) connected to 75. Capital groove: (a) excised, open at distal and pollical facet by strongproximal crest at midline. proximal ends;(b) excavated,closed by distal and (On0 1980: fig. 10) CI = 1.0. proximal ridges. CI = 1.O (variable in Anhingu). 76. Pneumaticfossa: (a) internal surfaceapneumatic; STERNUM (b) distinctly pneumatic. CI = 1.O. 95. Sternal plates: (a) angle with carina greater than 77. Pectoralattachment: (a) anglebetween it and del- 90”; (b) at most 90”. CI = 0.50. toid crest greater than 4.5”;(b) less than 45”. CI 96. Ventral manubrial spine: (a) small or absent; (b) = 1.O. (variable in atrice& strongly produced. CI = 0.33. 78. External tuberosity: (a) distal surface planar; (b) 97. Ventral manubrial spine: (a) internal surface distal surface deeply incised into groove. CI = planar; (b) concave; (c) convex. (Shufeldt 1902: 1.0. pl. VI, fig. 30) CI = 1.0 (variable in Anhingu). U 79. Ectepicondyle:(a) ligamental furrow very small; (b) thin and shallow; (c) wide and deep. CI = FURCULA’ 0.29. (variable in Anhingu) 98. Attachment of M. rhomboideus superficialis:(a) 80. Attachment of M. coracobrachialisposterior: (a) crest is absent, reduced, or indistinct; (b) crest is reduced or indistinct: (b) marked bv deep pit. strongly produced. (Owre 1967: fig. 1Id) CI = (Owre 1967: fig. 13) Ci I1 .O (variable in A-&n- 0.5. &W). 99. Attachment of M. rhomboideus superficialis:(a) 81. Attachment of M. proscapulohumerus:(a) re- crest extends from anterior surface to furcular duced or indistinct; (b) producedinto prominent angle; (b) crest does not reach furcular angle; (c) tuberositv. (Owre 1967: fia. 13) CI = 1.0. crest indistinct or obsolete.(Ono 1980: fig. 3) CI 82. Brachialis impression: (a)>educed or indistinct; = 0.67. (b) deeply excised.(Owre 1967: fig. 13) CI = 0.5. 100. Medial surfaceof furcular shaft proximal to cor- 83. Brachialisimpression: (a) strongventral ridgenear acoidal facet: (a) surfaceof shaft convex; (b) sur- impressionof supracoracoideus;(b) ventral ridge face concave or excavated. (Baumel et al. 1979: absent. (Owre 1967: fig. 14) CI = 0.5. fig. 6a) CI = 0.25 (variable in Anhinga). 101. Hypocleidus: (a) strong dorsomedial ridge; (b) ULNA ridge absent. CI = 0.5. 84. External condyle: (a) simple; (b) with strong 102. Symphysis: (a) internal aspect concave; (b) in- proximally producedridge. (Ono 1980: fig. 8) CI ternal aspectexpanded mesially into hollow cup. = 1.0. CI = 0.33. 85. Attachment of M. bicipitus: (a) separate;(b) con- 103. Furcular process:(a) normally produced; (b) ex- nectsbrachialis impression by strongridge. CI = panded mesially into shelf. CI = 1.O. 1.0. 86. Humero-ulnar depression:(a) small, shallow; (b) SCAPULA broad and deeply excised, (c) narrow and deeply 104. Acromion: (a) inferolateral surfaceplanar; (b) in- excised. (Ono 1980: fig. 8) CI = 1.0. ferolateral surface strongly produced into hook. 87. Attachment of M. scapulatricipitis:(a) shallow or CI = 1.0. indistinct; (b) deeply excavated.(Owre 1967: fig. 105. Attachment of M. tensor patagialis brevis: (a) 15b, “triceps, scapularhead”) CI = 1.O. reduced, absent, or indistinct; (b) strongly pro- 88. Attachment of anterior articular ligament: (a) duced into robust tuberosity. (Ono 1980: fig. 5) subquadrate;(b) foreshortened proximally and CI = 0.5. CORMORANT PHYLOGENY 903

PELVIS producedinto rugosecrest. (Owre 1967: fig. 40b 106. Postacetabularelements: (a) 8; (b) 9. (Shufeldt bottom) CI = 1.0. 1902: pl. V, fig. 22) CI = 1.0. 121. Internal condyle:(a) superiomedialfossa shallow, 107. Posterolateralimpression of M. extemus iliofib- emarginate;(b) fossadeeply excavated,bounded ularis: (a) lateral scar reduced, less than one in- by sharp crests.CI = 1.0. tervertebral foramen in length; (b) lateral scar 122. Flexor attachment on external condyle: (a) pro- robust, at least one intervertebral foramen in duced medially; (b) produced laterally. CI = 1.O. length; (c) lateral scar robust, at least one inter- 123. External condyle: (a) medial prominence broad vertebral foramen in length. (Steinegerand Lucas and rounded; (b) medial prominence produced 1889: pl. III, fig. 2) CI = 0.67. - - into thin blade. CI = 1.0. 108. Posterolateralimuression of M. extemus iliofib- 124. External condyle: (a) medial and lateral promi- ularis: (a) lateral scar separate; (b) lateral scar nences separate; (b) prominences connected by connectedto strongposterior scar.(Stejneger and transverseridge causinga deep pit. CI = 0.5. Lucas 1889: pl. III, fig. 2) CI = 1.0. 125. External condvle: (a) rotular groove shallow: (b) 109. Posterolateralimpression of M. extemus iliofib- rotular groove wide’and deep; (c) rotular groove ularis: (a) mesial scar absent or reduced, (b) me- narrow and deeo. (Ono 1980: fia. 12) CI = 1.0. sial scar connectsanterior scar. CI = 1.O. 126. External condyle: (a) internal &face broad and 110. Intramuscular line of M. iliacus preacetabulae: rounded;(b) internal surfacedistinctly narrowed. (a) ariseson posterior edgeof preacetabulum;(b) CI = 1.0. arises on lateral edge of preacetabulum.(Stejne- 127. External condyle: (a) superior surfaceshallow or ger and Lucas 1889: pl. III, fig. 2) CI = 0.25. planar; (b) superior surfaceexcavated into fossa. 111. Antitrochanteric ilium: (a) planar; (b) ventrally CI = 1.0. produced. CI = 1.0. 128. Trochanter: (a) anterior angle distinct; (b) ante- rior angle indistinct, curvilinear. (Ono 1980: fig. FEMUR 12) CI = 1.0. 112. Attachment of M. ischiofemoralis: (a) lateral TIBIOTARSUS impression is reduced, (b) impression ‘strongly producedgiving sharpangularity to posterolater- 129. Attachment of M. flexis cruris lateralis + me- al femoral surface.(Owre 1967: fig. 40a bottom) dialis: (a) mesial impression reduced or planar; CI = 0.33. (b) mesial impression strongly produced. (Owre 113. Attachment of M. iliotrochanter: (a) impression 1967: fig. 41) CI = 1.0. is reduced or indistinct; (b) impression deeply 130. Internal articular notch (incisura tibialis): (a) an- excavated. (Owre 1967: fig. 40a bottom) CI = gle is broad, greater than 90”; (b) angle is deep, 1.0. subequalto 90”. (Baumel et al. 1979: fig. 13b) CI 114. Attachment of M. iliotrochanter: (a) impression = 0.25. is broad and subcircular;(b) impression is very 131. Plantaris fossa: (a) shallow; (b) excavated. (Ono narrow. (Ono 1980: fig. 12) CI = 1.0. 1980: fig. 13) CI = 0.25. 115. Attachment of M. iliotrochanter: (a) impression 132. Supratendinal bridge: (a) inferior and superior is oriented transversely on femoral surface; (b) margins subparallel; (b) margins medially con- impression runs in superior-inferior orientation. stricted into hourglassshape. CI = 1.0. CI = 1.0. 133. Outer cnemial crest:(a) distolateralprocess weak- 116. Attachment of M. flexis perforatus digitalis III: ly produced into a ridge; (b) strongly produced (a) planar above internal condyle; (b) deeply ex- distally into robust hook. CI = 1.0. cavated above internal condyle. (Owre 1967: fig. TARSOMETATARSUS 40b bottom) CI = 0.33. 117. Attachment of M. flexis perforatus digitalis III: 134. Trochlea metatarsusII: (a) plantar curve simple; (a) superior impression is reduced or planar; (b) (b) strong medioplantar processextends distally. superior impression is deeply excavated along (Ono 1980: fig. 14-3b) CI = 1.0. shaft. CI = 1.0. 135. Trochlea metatarsusIII: (a) dorsal curve simple; 118. Attachment of M. femoritibialis intemus: (a) an- (b) strongly produced dorsally. (Ono 1980: fig. terior intramuscularimpression reduced or plan- 14-lc) CI = 1.0. ar; (b) intramuscularimpression is deeply exca- 136. Distal accessoryforamen: (a) with external open- vated. (Owre 1967: fig. 40d bottom) CI = 1.0. ing; (b) with internal openinginto distal foramen. 119. Attachment of M. iliacus iliotrochanteris medi- CI = 0.5. alis: (a) impression is reduced or planar; (b) RHAMPHOTHECA impression is produced into robust tuberosity. (Owre 1967: fig. 40a bottom) CI = 1.0. 137. Terminal end: (a) linear, pointed; (b) uncinate, 120. Intramuscular line between M. flexis perforatus sharply hooked. CI = 1.0. digitalis III and IV: (a) reduced or indistinct; (b)