The phylogeny of the living and fossil Sphenisciformes (penguins)
Abstract
We present the first phylogenetic analysis of the Sphenisciformes that extensively samples fossil taxa. Combined analysis of 181 morphological characters and sequence fragments from mitochondrial and nuclear genes (12S, 16S, COI, cytochrome b, RAG‐1) yields a largely resolved tree. Two species of the New Zealand Waimanu form a trichotomy with all other penguins in our result. The much discussed giant penguins Anthropornis and Pachydyptes are placed in two clades near the base of the tree. Stratigraphic and phylogenetic evidence suggest that some lineages of penguins attained very large body size rapidly and early in the clade's evolutionary history. The only fossil taxa that fall inside the crown clade Spheniscidae are fossil species assigned to the genus Spheniscus. Thus, extant penguin diversity is more accurately viewed as the product of a successful radiation of derived taxa than as an assemblage of survivors belonging to numerous lineages. The success of the Spheniscidae may be due to novel feeding adaptations and a more derived flipper apparatus. We offer a biogeographical scenario for penguins that incorporates fossil distributions and paleogeographic reconstructions of the Southern continent's positions. Our results do not support an expansion of the Spheniscidae from a cooling Continental Antarctica, but instead suggest those species that currently breed in that area are the descendants of colonizers from the Subantarctic. Many important divergence events in the clade Spheniscidae can instead be explained by dispersal along the paths of major ocean currents and the emergence of new islands due to tectonic events.
© The Willi Hennig Society 2006.
The Sphenisciformes (penguins) are one of the most highly derived clades of extant birds. Extensive morphological, physiological, and behavioral modifications from the typical avian condition have allowed penguins to achieve a largely aquatic lifestyle and invade some of the most inhospitable environments in the world. The fossil record of Sphenisciformes is extensive compared with most other avian groups. These fossils provide an excellent opportunity to explore the history of a unique clade.
Thomas Huxley (1859) provided the first description of a fossil penguin, founding Palaeeudyptes antarcticus on an incomplete tarsometatarsus from New Zealand. Subsequently, the discovery of “giant” penguin fossils in Antarctica by the Swedish South Polar Expedition of 1901–03 excited much interest. While some accounts exaggerated the size of these taxa (see Simpson, 1946), a few extinct species did reach sizes significantly greater than any extant species (Fig. 1). Since these early discoveries, New Zealand, Antarctica and South America have yielded a wealth of penguin material. Smaller collections of fossils have also been recovered from Australia and South Africa. Most recently, field work in South America (see Stucchi, 2002; Clarke et al., 2003; Stucchi et al., 2003; Acosta Hospitaleche et al., 2004; Acosta Hospitaleche and Canto, 2005), and Seymour Island, Antarctica (Myrcha et al., 2002; Tambussi et al., 2005), has uncovered new material and renewed interest in many aspects of penguin evolution.

Reconstruction of the giant Eocene penguin Anthropornis nordenskjoeldi from the La Meseta Formation of Seymour Island, Antarctica, with extant emperor penguin (Aptenodytes forsteri), chinstrap penguin (Pygoscelis antarctica) and little penguin (Eudyptula minor) for scale. Skeletal reconstruction incorporates the known elements of the appendicular girdle, forelimb and hindlimb of Anthropornis nordenskjoeldi with skull patterned from large penguin cranial elements from La Meseta (Myrcha et al., 1990) and Olson (1985). Missing elements are extrapolated from phylogeny. Standing height follows the estimates obtained by analyses of Jadwiszczak (2001). Artwork by Kristin Lamm.
The earliest penguin fossils reveal that the most recognizable hallmarks of the penguin skeleton, including pachyostotic bone structure and the conversion of the wing into a flipper, were in place by the Paleocene (Tambussi et al., 2005; Slack et al., 2006). Molecular evidence suggests the stem lineage leading to penguins had already branched off by the Late Cretaceous (Baker et al., 2006; Slack et al., 2006), though Cretaceous representatives of this lineage may well have lacked easily identifiable penguin apomorphies. Several authors have hypothesized that the disappearance of marine reptiles at the Cretaceous‐Tertiary mass extinction opened an ecological niche into which penguins quickly radiated.
The relationships of extant penguins have recently received much deserved attention and a solid morphological and molecular basis for their classification now exists. An early study by Zusi (1975) delineated several morphological characters useful for grouping penguins and Schreiwei (1982) provided a detailed comparative study of penguin myology. O'Hara (1989) was the first to conduct a cladistic analysis of penguins, arriving at several alternate arrangements depending on which of four outgroup taxa were used to root the tree. In the last few years, extensive morphological (Giannini and Bertelli, 2004; Bertelli and Giannini, 2005), molecular (Bertelli and Giannini, 2005; Baker et al., 2006) and combined (Bertelli and Giannini, 2005) analyses have provided robust hypotheses of extant penguin phylogeny.
A comprehensive phylogenetic analysis incorporating fossil penguins has not been previously attempted, despite the fact that the majority of known penguin species are extinct. Some authors (Fordyce, 1991) have cited the incomplete nature of most fossil penguin taxa as an obstacle to phylogenetic analysis. While the marine habitat and pachyostotic bone structure of penguins make them particularly amenable to fossilization, the vast majority of known specimens represent isolated elements scattered by the action of the sea. Hence, many forms are currently known from only a handful of postcranial elements.
Numerous theoretical and empirical studies have pointed out that a high percentage of missing data does not necessarily lead to lack of resolution in phylogenetic analysis (see Kearney and Clark, 2003; Wiens, 2003). It is the amount of informative data, rather than the proportion of missing data, which is crucial to resolution. Fortuitously, the two bones most often present in the hypodigms of fossil penguin taxa (humerus and tarsometatarsus) are also two of the most morphologically variable across taxa. As our results illustrate, these bones preserve enough phylogenetic signal to provide significant resolution of penguin phylogeny.
Taxonomic note
Clarke et al. (2003) suggested that the Linnean family name Spheniscidae be applied to the clade comprising all extant penguins. These authors also proposed that the name Pansphenisciformes be applied to the clade of all taxa more closely related to extant penguins than any other living avian taxon and that the name Sphenisciformes be applied to all members of that lineage that have lost the power of aerial flight. We use the name Spheniscidae in the sense proposed by Clarke et al. (2003) throughout this paper. Presently, no stem penguin lacking the apomorphic loss of aerial flight is known, and so the phylogenetic names Sphenisciformes and Pansphenisciformes denote the same set of taxa. We use the name Sphenisciformes for the clade including all fossil and living penguins throughout this paper.
Materials and methods
We examined fossils and casts from nine fossil penguin species, along with skeletal specimens, skins, and eggshells of all extant penguin species and the outgroup taxa listed in Appendix 1. For taxa we were unable to examine firsthand, we relied on descriptions and figures from the literature. For some fossils, we supplemented our firsthand observations with descriptions and figures from the literature. Sources utilized for codings include: Acosta Hospitaleche et al. (2004) (Eretiscus tonnii); Acosta Hospitaleche (2005) (Arthrodytes andrewsi); Ameghino (1905) (Arthrodytes andrewsi); Jadwiszczak (2006) (Anthropornis grandis, Palaeeudyptes gunnari, Palaeeudyptes klekowskii); Marples (1952) (Archaeospheniscus lowei); Marples (1960) (Marplesornis novaezealandiae); Myrcha et al. (1990) (Palaeeudyptes klekowskii); Myrcha et al. (2002) (Anthropornis grandis, Delphinornis larseni, Marambiornis exilis, Mesetaornis polaris, Palaeeudyptes klekowskii) Simpson (1981) (Eretiscus tonnii); Slack et al. (2006) (Waimanu manneringi, Waimanu tuatahi); Stucchi (2002) (Spheniscus megaramphus); and Stucchi et al. (2003) (Spheniscus urbinai).
Note on Palaeeudyptes
The taxonomic history of the genus Palaeeudyptes is complex. The Antarctic material assigned to this genus has been divided into two species, P. klekowskii and P. gunnari. We observe no problems with the diagnoses for these species provided by Myrcha et al. (2002) and utilize both species as terminals in our analysis. The New Zealand record of Palaeeudyptes is more complicated. The holotype specimen of Palaeeudyptes antarcticus is a tarsometatarsus missing much of metatarsal II. Marples (1952) assigned many specimens to this taxon in his review of New Zealand fossils. Brodkorb (1963) later erected the species P. marpelsi to accommodate large specimens originally assigned to P. antarcticus. Simpson (1971) called attention to the uncertain provenience of the holotype of P. antarcticus and regarded both the variation in size and the stratigraphic range of specimens assigned to P. antarcticus as unrealistically large. Given the problematic nature of the species level classification of Palaeeudyptes, Simpson referred all but the holotype specimens of P. antarcticus and P. marpelsi to Palaeeudyptes sp. In our analysis, we considered the New Zealand fossils listed in Appendix 1 as one terminal. All were referred to Palaeeudyptes antarcticus by Marples (1952) and can be identified to the genus level using the diagnosis provided by Simpson (1971).
Outgroup selection
Because the earliest known penguins are already highly specialized, reconstructing their relationships with other extant avian lineages is challenging. Simpson (1946) noted similarities shared by early penguins and Procellariiformes, and Marples (1962) discussed particular similarities with diving petrels. Subsequently, numerous authors have presented evidence for a close relationship between Sphenisciformes and Procellariformes (Ho et al., 1976; Saiff, 1976; Sibley and Ahlquist, 1990; Mey et al., 2002; Mayr and Clarke, 2003), though others have argued for a relationship with Gaviiformes (Cracraft, 1981; Olson, 1985; Groth and Barrowclough, 1999). Baker et al. (2006) used the gaviiform Gavia and the procellariform Diomedea as outgroups in a molecular analysis of extant penguins, and found Diomedea to be closer to penguins, though with low support.
Giannini and Bertelli (2004) and Bertelli and Giannini (2005) rooted their analyses with Gavia stellata, and used 11 additional outgroup terminals from the Procellariiformes. These included Diomedea melanophrys, Phoebetria palpebrata, Macronectes giganteus, Daption capense, Pterodroma incerta, Puffinus griseus, Procellaria aequinoctialis, Pachyptila desolata, Oceanites oceanicus, Oceanodroma leucorrhoa, and Pelecanoides urinatrix. We included the same outgroup terminals in our morphological analysis and added Gavia immer and Diomedea exulans to the combined analysis in order to improve sequence representation (see below).
Phylogenetic analyses
Two parsimony analyses were conducted. The first analysis utilized only the morphological data set (Appendix 2) and included all taxa (data matrix in Appendix 3). Two hundred replications (saving up to 10 trees per replication) were run, each followed by TBR branch swapping, with an extra TBR round on the optimal trees. As a measure of support, we calculated Bremer values (Bremer, 1994) as described in Giannini and Bertelli (2004). For searches and support estimation we used the program TNT (Goloboff et al., 2004).
Second, a combined analysis of non‐molecular + DNA data was performed using the method of Direct Optimization (DO; Wheeler, 1996). Under DO, unaligned sequences are optimized on a given topology, and the length implied by the topology is calculated on the basis of a predefined transformation costs. We followed Bertelli and Giannini (2005) in applying equal costs to all molecular transformations (i.e., indels and substitutions). Indels, if required by the topology under examination, are inserted in hypothetical ancestral sequences at internal nodes (the ancestral sequences are generated from the pair of descendant sequences in a downpass). A length is calculated for each tree examined, and lengths compared during a search; the tree or trees that minimize this total cost are chosen. Searches involve replicated tree building and TBR branch swapping, and were run as implemented in the program POY 3.11 (Wheeler et al., 2003). The morphological data are co‐optimized under equal costs such that a total DNA + morphological length is calculated for each tree examined. We followed Giannini and Simmons (2003) in using 100 replications (random addition sequence of taxa followed by TBR branch swapping) coupled with tree fusing (Goloboff, 1999). Bremer support values of the combined analysis were calculated from searches using the commands ‐bremer and ‐constrain as implemented in POY.
Sequences used include those analyzed by Bertelli and Giannini (2005) as well as new sequence data available from Baker et al. (2006) and Slack et al. (2006). Genes used include 12S (∼ 970 bp), 16S (∼ 1070 bp), COI (456 bp), cytochrome b (1143 bp), RAG‐1 (2862 bp). Coding genes (COI, cyt b, RAG‐1) were run as prealigned in POY. See Appendix 4 for accession numbers and authors.
Biogeographical analysis
Bertelli and Giannini (2005) mapped penguin breeding areas onto optimal combined topologies and interpreted the biogeographical patterns in terms of the dispersal–vicariance concept (Ronquist, 1997). That biogeographical reconstruction of the group was based on extant species and therefore limited in scope. With the phylogenetic analysis presented in this work, substantial extinct diversity is included, and a more comprehensive biogeographical hypothesis can be constructed using a similar approach. Briefly, each distinct breeding area of the Southern Hemisphere was assigned a character state in a single geographic character that was optimized in the consensus penguin subtree from the combined analysis. Nine breeding areas were recognized by Bertelli and Giannini (2005): the South American coasts and the islands of the continental shelf (hereafter “sam”); the Galapagos Islands (ga); the South African coast (saf); the isolated Atlantic Ocean Islands of Tristan da Cunha and Gough (tg); the isolated Bouvet Island (bu); the scattered Indian Ocean Islands including Marion, Prince Edwards, Kerguelen, and Heard Islands, most of which share a similar penguin fauna (io); the Australia–New Zealand region and nearby islands (az); the Antarctic Peninsula (ap); the islands of the Scotia arch (S. Orkney, S. Georgia, and S. Sandwich Islands; sa); and the shores of the Antarctic Continent (ac). The delimitation of the areas was based on known distribution of rookeries (penguin breeding grounds), faunistic composition, and degree of isolation/connectedness of the region (Bertelli and Giannini, 2005). In the current analysis, we used the same areas, assuming that the locality of a fossil terminal represented or at least was included in its breeding ground. For instance, Palaeeudyptes klekowskii is known from fossils from Seymour Island, so we scored this fossil “ap” for the geographic character. Minor modifications were introduced in the scorings of Bertelli and Giannini (2005) for extant species. While Bertelli and Giannini (2005) used simple mapping (i.e., full optimization), in this paper we use only the downpass to reconstruct the ancestral distributions. This is more consistent with Ronquist's (1997) dispersal–vicariance interpretation of historical biogeographical reconstruction.
Results
The analysis including only morphological characters resulted in eight most parsimonious trees of 449 steps (branches of length 0 collapsed). The strict consensus of these trees is presented in Fig. 2. Procellariformes and Sphenisciformes are reciprocally monophyletic. Throughout the clade Sphenisciformes, support values are low. This is not unexpected given the incomplete nature of many taxa, a problem exacerbated because some taxa are known from non‐overlapping sets of elements. Despite these issues, the branch leading to Sphenisciformes receives a Bremer support value of 4. This indicates robust support for penguin monophyly.

Strict consensus of four most parsimonious trees from the analysis using the morphological character set. Bremer support values are indicated above branches. See text for details of analysis.
The strict consensus tree is well resolved given the high percentage of missing data for many fossils. The two species of the recently described Paleocene penguin Waimanu fall outside of a clade of all other penguins. Three small Antarctic Eocene penguins are located near the base of the tree, with Delphinornis larseni most basal and Mesetaornis polaris and Marambiornis exilis forming a trichotomy with the remaining ingroup taxa. Towards the crown, an additional trichotomy is formed by two clades of large Eocene–Oligocene penguins and the remaining penguin taxa. The majority of the remaining fossils are placed in a largely pectinate arrangement leading to the crown clade Spheniscidae, with only Spheniscus urbinai and Spheniscus megaramphus nested within Spheniscidae. All extant genera are monophyletic, and Spheniscus + Eudyptula form the sister group to (Pygoscelis + Aptenodytes) + (Megadyptes + Eudyptes).
From the combined (DNA + morphology) analysis, two trees of 4308 steps were obtained. The topology of the fossil taxa (except for fossil species of Spheniscus) is identical to the topology recovered by the morphological analysis. The extant genera are all monophyletic in the strict consensus tree of the combined analysis (Fig. 3). Aptenodytes is the most basal extant taxon, and Pygoscelis is sister to (Spheniscus + Eudyptula) + (Eudyptes + Megadyptes). The fossil Spheniscus species form a clade with living members of the genus in this result. The relationships of the extant genera agree with those recovered by the combined analysis of Bertelli and Giannini (2005) and the molecular analysis of Baker et al. (2006). The species level relationships are more resolved but congruent with those recovered in the combined analysis of Bertelli and Giannini (2005) and differ from the results of Baker et al. (2006) in that the positions of Eudyptes sclateri and Eudyptes chrysocome and reversed. As discussed by Bertelli and Giannini (2005), the differences in the arrangement of the extant taxa are largely the effect of differences in the root preferred by the molecular versus the morphological data.

Strict consensus of four most parsimonious trees from the analysis using the combined molecular and morphological data set. Bremer support values are indicated above branches. See text for details of analysis.
Discussion
Implications for taxonomy
Simpson (1946) was the first to attempt a comprehensive revision of living and fossil penguin taxonomy. Simpson divided the fossil taxa into four subfamilies (Palaeospheniscinae, Paraptenodytinae, Palaeeudyptinae, Anthropornithinae) and placed all living taxa in the subfamily Spheniscinae. Accounting for the later synonymization of taxa listed by Simpson (1946), Palaeospheniscinae and Palaeeudyptinae comprised the genera Palaeospheniscus and Palaeeudyptes, respectively. Paraptenodytinae included Paraptenodytes and Arthrodytes andrewsi and Anthropornithinae included Anthropornis, Delphinornis and Pachydyptes. Marples (1952) subsumed Anthropornithinae into Palaeeudyptinae and assigned several newly described New Zealand genera (Platydyptes, Duntroonornis, Korora, Archaeospheniscus) to the Palaeeudyptinae. Subsequent authors (Brodkorb, 1963) generally followed Marple's system of classification, sometimes recognizing Anthropornithinae (Tambussi et al., 2005). Following decades of work on fossil penguins and the discovery of many new fossil taxa, Simpson (1971) abandoned his own subfamily level classification system pending discoveries of more complete materials. At that time, he noted that current knowledge would not provide the basis for a classification system that would be “of evolutionary significance or otherwise useful” (Simpson, 1971, p. 366).
In recent revisions, Acosta Hospitaleche et al. (2004) placed Palaeospheniscus and Eretiscus in the subfamily Palaeospheniscinae and Acosta Hospitaleche and Canto (2005) united Arthrodytes andrewsi and Paraptenodytes in the family Paraptenodytinae. However, neither study included a phylogenetic analysis. The study presented in this paper provides a test of whether the subfamilies of Simpson correspond to clades. Our result confirms the monophyly of Palaeospheniscinae. We included Arthrodytes andrewsi and Paraptenodytes antarcticus in our study, but were unable to completely resolve their positions relative to other taxa. In our result, these two taxa can form a clade or be arranged paraphyletically for the same cost. Testing the validity of Paraptenodytinae requires further study of these fossils and sampling of the other species included in the genus Paraptenodytes (P. robustus and P. brodkorbi).
Anthropornithinae as originally conceived by Simpson (1946) is non‐monophyletic in our result and needs be redefined or discarded. Palaeeudyptinae in the sense of Simpson (1946) is paraphyletic in our result as Pachydyptes ponderosus is nested within the genus Palaeeudyptes. Palaeeudyptinae in the expanded sense of Marples (1952) is certainly polyphyletic. As discussed above, we apply the phylogenetic name Spheniscidae to the least inclusive clade uniting all extant penguins. An important consensus has been reached from independent data regarding the phylogenetic relationships within this clade [compare topologies proposed by Bertelli and Giannini (2005) and Baker et al. (2006)]. However, we feel that applying names to other clades within the Sphenisciformes and within Spheniscidae is premature at this time, given that many important fossils are presently being studied. Over the next few years, we expect further progress will permit a stable phylogenetic taxonomy for the Sphenisciformes.
Body size and penguin evolution
Most living penguins are rather small, with members of the genus Aptenodytes (king and emperor penguins) far outstripping other species in size. The impressive size of some Eocene taxa has been central to many popular accounts of fossil penguins. Simpson (1946) pointed out that many estimates were wild exaggerations, but the fact remains that several fossil taxa reached heights and weights far exceeding all living taxa (Jadwiszczak, 2001; also see Figs 1, 9, and 12 of this paper for examples).

Humerus of Pygoscelis antarcticus (AMNH 26160) and cast of humerus of Pachydyptes ponderosus (BMNH A3262) in posterior view. Abbreviations: ld: scar for insertion of m. latissimus dorsi; sc, scar for insertion of m. supracoracoideus. Scale bar = 1 cm.

Cast of ulna of Palaeeudyptes sp. (USNM Acc 377506) and ulna of Eudyptes pachyrhynchus (AMNH 26509) in ventral view. Abbreviation: ctd, cotyla dorsalis. Scale bar = 1 cm.
Body size has figured prominently in discussions of penguin origins. Simpson (1946) put to rest two earlier hypotheses of penguin ancestry in his monographic description of Paraptenodytes and revision of other fossil taxa. The first hypothesis speculated that penguins evolved from a bipedal ancestor that had never possessed aerial flight capabilities. This bizarre scenario would require an origin for penguins separate from all other birds, and was discussed seriously only by Lowe (1933). The second hypothesis supposed that penguins evolved from a flightless ancestor that took to the sea, perhaps pressured by deteriorating Antarctic climate. Simpson argued against these scenarios and presented evidence for the evolution of penguins from a volant ancestor. He pointed out that the penguin flipper is most likely derived from a functional wing and noted evidence from other avian groups capable of underwater flight. The largest extant birds capable of both aerial and underwater flight are alcids close in size to the smallest extant penguin, Eudyptula minor. The large non‐volant alcid Pinguinus impennis (the Great Auk) was capable of underwater flight and is closely related to birds that retain aerial flight capabilities. The same appears true for the extinct Mancallinae. Based on these observations, Simpson predicted that the ancestor of penguins was most likely a small bird capable of both aerial and aquatic flight, and the shift to an entirely aquatic use of the forelimb freed penguins to attain sizes not feasible for an aerial flier.
If Simpson's hypothesis is correct, the earliest members of the Sphenisciformes were very small. The initial attainment of large body size may have occurred rapidly, as the early Paleocene Waimanu manneringi appears to have approached the extant emperor penguin in size (Slack et al., 2006) and the late Paleocene Crossvallia was larger than the emperor penguin (Tambussi et al., 2005). Our results reveal that large changes in body size occurred multiple times early in the clade's history. Three Eocene penguins (Delphinornis, Mesetaornis, Marambiornis) are placed near the base of the tree immediately above Waimanu. Estimates based on tarsometatarsus size suggest these penguins were only slightly larger than typical non‐Aptenodytes crown penguins (Jadwiszczak, 2001). Immediately above these small penguins in the tree, two clades of very large Eocene‐Oligocene penguins form part of trichotomy. The arrangement of Waimanu, small penguins, and very large penguins suggests high levels of body size plasticity in early penguins. The large size range spanned by Eocene penguins may have enabled numerous taxa to occupy a single area. Ten recognized species are known from the Late Eocene of Seymour Island and may have coexisted (Jadwiszczak, 2006). The size range of later penguins is more restricted, and no extremely large forms are known to survive into the Neogene.
Tambussi et al. (2005) asserted that Crossvallia provided evidence of penguins attaining large size independently in the Late Paleocene and Late Eocene, under different environmental conditions. Because of the poor preservation of the holotype, we were unable to reliably code Crossvallia for inclusion in our analysis. However, if Crossvallia is closely related to Anthropornis as Tambussi et al. (2005) tentatively suggested, it is more parsimonious to assume the clade including these taxa share large body size through common ancestry than to argue for two independent increases. In our result, the Anthropornis clade and the Palaeeudyptes–Pachydyptes clade are part of a polytomy, so it remains unclear whether these large penguins represent a single radiation of large forms or two separate lineages converging on large body size. A more rigorous exploration of body size shifts in penguin evolution using comparative methods is warranted, and would require hypotheses of relationships for additional fossils, especially the smaller taxa (i.e., Korora and Duntroonornis).
Feeding modifications
Penguins prey on a variety of organisms, including fish, cephalopods, and crustaceans. Zusi (1975) considered many features of penguin cranial osteology, myology and tongue structure in the context of diet. In that study, Zusi divided penguins into fish specialists, planktonic specialists, and generalists. The diet of individual penguin species was not well studied at the time, but subsequent research (see species accounts in Williams, 1995) broadly supports these generalizations. Zusi (1975) observed that penguins classified as fish specialists (Spheniscus) possessed powerful adductor musculature, stoutly constructed upper jaws, and smaller buccal papillae. Planktonic specialists (Eudyptes, Pygoscelis adeliae, Pygoscelis antarctica) were characterized by reduced adductor musculature, deep mandibles, large tongues and well‐developed papillae.
Few fossil penguin skulls are known and fewer still are relatively complete, making it difficult to compare most fossils to living species for beak shape, gape, mandibular depth, and other important features that may yield insight into their ecology. One relevant feature for which data are available is the morphology of the temporal fossae. All known fossil specimens possess deep, dorsally extended temporal fossae. Conversely, the Spheniscidae (with the exception of Spheniscus) show a marked reduction in the dorsal extent of the temporal fossae. The wide separation of the temporal fossae optimizes as an unambiguous synapomorphy of the Spheniscidae in the combined tree. The presence of dorsally extended fossae in Spheniscus is thus most parsimoniously interpreted as a reversal. The alternative topology of the morphology‐based tree places Spheniscus + Eudyptula as the sister group to all other extant taxa. Under this topology, the powerful adductor complex of Spheniscus may instead be interpreted as a retention of the primitive condition seen in fossil penguins.
The deep, extensive temporal fossae of the fossil penguins Waimanu, Paraptenodytes and Marplesornis[a similar condition is present in skulls assigned to the genus Palaeospheniscus (Acosta Hospitaleche and Canto, 2005) and an unnamed skull from the Eocene of Antarctica (Ksepka and Bertelli, in press)] indicate early penguins possessed powerful adductor musculature. Beak morphology is almost completely unknown for fossil penguins, though Waimanu tuatahi preserves a partial mandible and a few partial beaks from the Eocene of Antarctica have been reported (Olson, 1985; Myrcha et al., 1990). These beaks are nearly straight, strongly constructed, and lack notable deepening of the mandible. The available data suggest early penguins had a strong bite force and probably relied heavily on larger prey items such as fish. The exploitation of planktonic prey may well be a feeding strategy unique to the Spheniscidae. However, specimens preserving the palate, mandibles and upper jaws of fossil taxa are necessary to fully explore fossil penguin ecology.
Evolution of the penguin flipper
This study places trends in the evolution of the penguin flipper apparatus in a phylogenetic context. Modifications that occur in the transition from the flipper pectoral girdle and limb of primitive penguins to that of extant taxa include: narrowing of the sternal margin of the coracoid, a change from a strongly convex to a concave sternal margin of the coracoid, expansion of the tricipital fossa, division of the tricipital fossa into two chambers, reduction of the ulnar condyle, and flattening of the shaft of the humerus in the anteroposterior dimension. The flattening of the wing bones presumably serves to increase the aerodynamic efficiency of the wing for underwater swimming. Sacrifices in the resistance of the bone to shear forces incurred during the transition from a primitive subcircular cross‐section to a more elliptical cross‐section were likely unimportant given the pachyostotic structure of penguin bones.
Penguins generate thrust during both the downstroke and upstroke (Clark and Bemis, 1979). Raikow et al. (1988) measured joint mobility in the wings of wing‐propelled divers as well as non‐diving birds and found great reduction in the range of motion of the intrinsic wing joints of penguins compared with other groups, resulting in a degree of stiffening of the wing incompatible with aerial flight. The ulnar condyle in extant penguins is reduced to a nearly flat articular surface and the humerus is tightly bounded to the ulna and radius by a set of ligaments. The prominent, subhemispherical ulnar condyle of basal fossil taxa suggests a greater range of motion was possible at the joint between the humerus and forearm in primitive penguins, though the large shelf dorsal to the ulnar condyle developed in many fossil taxa may have served partially to limit movement at the humerus–forearm joint during the downstroke. Reducing the ulnar condyle may have been particularly useful in limiting movement at the humerus–forearm joint during the upstroke as well, and thus improving underwater flight efficiency. Exploring the functional significance of this and other modifications offers a promising avenue for future research.
Timing of the radiation of the Spheniscidae
The majority of fossil penguins fall outside the crown clade in both analyses.
Aside from the fossil species assigned to Spheniscus considered in our analysis, penguin fossils have been referred to at least three extant genera. Simpson (1975) considered some Pliocene fossils from New Zealand to be not clearly distinct from Eudyptula but did not describe this material in detail. Aptenodytes ridgeni is known from a partial postcranial skeleton from the Pliocene of New Zealand. Simpson (1972a) described the patellar groove as shallow in A. ridgeni. If this indicates a condition homologous to that in extant Aptenodytes, it would constitute one unambiguous synapomorphy supporting the allocation of A. ridgeni to the genus. Simpson (1972a) also described Pygoscelis tyreei from the Pliocene of New Zealand, noting overall similarities to extant Pygoscelis. However, none of the features discussed in the description allow confident assignment to that genus. One particular feature warns against including P. tyreei in Pygoscelis. A lamina of bone from the medial margin of the coracoid reaches the procoracoid process in the fossil. This feature is present in most penguins but absent in all living species of Pygoscelis. We did not observe variation in this feature in any of the dozens of Pygoscelis specimens examined in this study, and suspect intraspecific variation is extremely rare if it occurs at all. However, given the distribution of this character, it is possible the presence of the lamina in Pygoscelis tyreei is symplesiomorphic. The data available in Simpson's description allow us to place Pygoscelis tyreei within a clade including Paleospheniscinae, Marplesornis, and Spheniscidae, but do not permit further resolution of its position. We consider the affinities of P. tyreei uncertain given the available data.
Regardless of the phylogenetic position of P. tyreei, the three taxa discussed above only slightly extend the stratigraphic range of the Spheniscidae beyond the present. The taxa with the greatest implications for estimating the age of the crown radiation are Spheniscus megaramphus and Spheniscus urbinai. These Miocene taxa are nested within the Spheniscidae in the results of both the combined and morphological analyses and so predict the divergence of the Spheniscidae by the Miocene. Substantial ghost lineages are required to explain the late appearances of the other extant lineages under the combined topology. The late appearance of members of most crown lineages is surprising, given that numerous more basal penguin taxa are known from the interval during which members of the Aptenodytes, Pygoscelis and Eudyptes–Megadyptes lineages are predicted to have existed. It is worthwhile to note that the total length of these ghost lineages decreases under the morphology based tree, as the fossil Spheniscus species are part of the basal divergence in that tree. One possible explanation of the long ghost lineages is that they are an artifact of our poor understanding of some Neogene fossils. Inguza predemersus, Teringaornis moisleyi, Dege hendeyi, and Nucleaornis insolitus are all known from highly incomplete material. No apomorphies allowing confident assignment to a specific extant lineage are preserved in any of these fossils, as far as we can ascertain from the descriptions and figures presented in the literature. However, Inguza (Simpson, 1976) and Teringaornis (Scarlett, 1983) possess divided tricipital fossae (character 131: 1) and Nucleaornis (Simpson, 1979) has a deep intermetatarsal groove (character 159: 2) suggesting they belong close to or within Spheniscidae.
Biogeography
Our biogeographical reconstruction of the penguin clade is shown in Fig. 4. The oldest known penguins, Waimanu manneringi and W. tuatahi from New Zealand, appear at the base of our cladograms. As a consequence, the ancestral node Sphenisciformes is reconstructed as a New Zealand clade. Several Eocene taxa from Seymour Island (Delphinornis, Marambiornis, Mesetaornis) appear in the next three crownward backbone clades in the penguin subtree, which are unambiguously reconstructed as Antarctic Peninsula clades. In discussing penguin paleobiogeography it is critical to recognize that during the Paleocene and Eocene, New Zealand was closer to Antarctica than it is today, and that the Northern tip of the Antarctic Peninsula was displaced well north of its current position (Fig. 4A). Thus, the Antarctic Peninsula and Tierra del Fuego are most accurately considered as a single marine biogeographical area in the American quadrant of the Subantarctic region during the Early Tertiary.

(A) Map of the Southern Hemisphere in the Early Eocene with distributions of fossil penguins considered in this analysis represented by stars: SI, Seymour Island, Antarctica (A. nordenskjoeldi, A. grandis, P. klekowskii, P. gunnari, D. larseni, M. exilis, M. polaris); NZ, South Island, New Zealand (W. manneringi, W. tuatahi, A. lowei, P. novaezealandiae, Palaeeudyptes antarcticus, P. ponderosus, M. novaezealandiae); PB, Pisco Basin, Peru (off map) (S. megaramphus, S. urbinai); PT, Patagonia, Argentina (Paraptenodytes antarcticus, P. patagonicus, A. andrewsi, E. tonnii). (B) Map of the Southern Hemisphere showing the distribution of land masses in the present. Abbreviations: AC, shores of the Antarctic continent; AP, Antarctic Peninsula; AZ, Australia/New Zealand region; BU, Bouvet Island; IO, Southern Ocean Island; GA, Galapagos Islands; SA, islands of the Scotia Arch; SAF, South African coasts; SAM, South American coasts; TG, Islands of Tristan da Cunha and Cough. (C) Reconstructions of biogeographical events based on the downpass optimization of breeding territories on to the consensus tree from the combined analysis.
The fossil backbone clades show reconstructed areas that fluctuate between the Australian and American quadrant of the Subantarctic region. This reflects the placement of forms from those two Subantarctic regions, including the node of the common ancestor of Marplesornis and all extant genera, which is unambiguously reconstructed as an Australia/New Zealand clade. The distribution of the ancestor of all the extant genera is reconstructed as inhabiting the Antarctic Peninsula, the Scotia Arc, and New Zealand as it is the next backbone node that excludes Aptenodytes. The ancestors of the clades ((Eudyptula + Spheniscus) (Megadyptes + Eudyptes)) and (Megadyptes + Eudyptes) probably bred in New Zealand.
The distribution of the Aptenodytes ancestor is very wide including all regions within the Antarctic and Subantarctic shores. This is in contrast with the findings of Baker et al. (2006), who reconstructed the ancestor of Aptenodytes (and all modern genera) as exclusively Antarctic. The ancestor of Pygoscelis may have populated the Peninsula and islands of the Scotia Arch. The ancestor of Spheniscus + Eudyptula split into a New Zealand population that gave rise to Eudyptula, and a South American population that gave rise to Spheniscus. As discussed by Bertelli and Giannini (2005), this dispersal–vicariance event is consistent with the prevalent circumpolar current and the West Wind Drift. Further interpretation of this migration is now possible, thanks to the addition of data from fossil taxa. This dispersal event is at least Miocene in age, and the two fossil Spheniscus from Peru (the coast that likely received the vagrants from the East) are robust compared with modern Spheniscus. We speculate that large body size may have allowed the ancestor of Spheniscus to endure the journey across the immense expanse of the South‐west Pacific, which was (and is) devoid of islands. Our reconstruction also confirms another pair of dispersal–vicariance events within the Spheniscus clade proposed by Bertelli and Giannini (2005). A Pacific‐coast ancestral population of Spheniscus dispersed to the Galapagos and split forming the daughter species S. mendiculus (Atacama) and S. humboldti (Galapagos). This event is consistent with the north‐westward Humboldt current. Also confirmed is the dispersal–vicariance event that produced the daughter species S. demersus (South Africa) and S. magellanicus (South America). This event is consistent with the westward Brazil–Benguela current.
Other major dispersal–vicariance events seem to have occurred within established modern species. According to our reconstruction, the species Aptenodytes forsteri and A. patagonicus are the result of a major Antarctic–Subantarctic split. Two other events of Continental Antarctic coast invasion are reconstructed in the Pygoscelis clade; in this case, the ancestral breeders may have come from the American quadrant of the Antarctic (the Peninsula and the Scotia Arch). Other events include the expansion of some Eudyptes, which is reconstructed as an ancestrally Indian Ocean + Australia/New Zealand clade, toward the American quadrant. This indicates a trans‐Indian rather than a trans‐Pacific route. Although they must contend with the circumpolar current, vagrant penguins have established themselves in numerous islands of the South Indian and South Atlantic Oceans. Some of these volcanic islands became emergent during very recent tectonic activity (some 5 Ma). Therefore, it seems safe to conclude that tectonics of the Southern plates played a major part in the diversification of modern lineages of penguins by providing vagrants with grounds in which they could establish successful breeding populations. These islands are presently home to several distinct subspecies of penguins.
It is important to address the extent to which our analysis is biased by the fossil record. As with any empirical endeavor, our analysis depends on available facts, and likely reflects limitations of the data. However, several lines of evidence suggest that major aspects of our reconstruction are correct. First, in spite of some degree of ambiguity, breeding grounds on the shores of the Australian or American quadrant are reconstructed in many internal nodes of the backbone of the consensus tree. Part of that reconstruction depends on single‐locality fossil records, but another major part is based on accurately known breeding grounds of modern penguins. That points to a coherent biogeographical pattern in which the Australian and American quadrants of the Subantarctic were of capital importance for the evolution of penguins throughout the Tertiary. Second, reconstructions are consistent with major natural forces and geological features that dominate climate and life in the Southern Oceans. These include some ancient speciation patterns consistent with constant and strong marine currents (dispersal–vicariance events in Spheniscus), and other speciation patterns that are consistent with more recent availability of breeding grounds (tectonic activity that generated young volcanic islands in the Southern Indian and Atlantic Oceans). A further assurance of the reliability of our reconstructions stems from the fact that none of the volcanic islands are reconstructed as possible breeding grounds in deep internal nodes—they all appear in terminal branches and support recent speciation within well established modern species.
Our biogeographical reconstructions differ sharply with another recently published scenario. Baker et al. (2006) proposed that the cooling of the Southern Hemisphere forced penguin populations out of Antarctica, where they found new opportunities for speciation—a process that would account for modern penguin diversity. We find little support for this view as a general explanation. Instead, we emphasize the major importance of Subantarctic regions throughout penguin history, starting from the reconstruction of Australia/New Zealand (outside the Antarctic convergence) as the ancestral habitat of the penguin clade, and followed by South America and the Antarctic peninsula. Although in all likelihood the lack of fossil penguins from continental Antarctica is artifactual, none of the oldest fossils of our sample occupied regions of polar climate. Climate along the Antarctic Peninsula during the Paleocene and Eocene was likely similar to that of present‐day Patagonia. Cooling during mid‐ and late Tertiary may not have had as profound an impact on penguin evolution as marine currents and emergence of islands. Moreover, for the three modern species (Aptenodytes forsteri, Pygoscelis adeliae, P. antarctica) with strictly Continental Antarctic breeding populations, that geographic distribution is mapped only to their corresponding terminal branches in our trees—not to the base of the crown penguin clade. Therefore, we propose a scenario in which cooling provided speciation opportunities to colonize an extreme environment rather than forcing penguins to escape from it. This event probably occurred recently. In fact, the most extreme Antarctic breeder—the emperor penguin (Aptenodytes forsteri)—cannot be distinguished morphologically from its sister species, the king penguin (Aptenodytes patagonicus), except by a few integumentary characters and modest size and proportional differences. Furthermore, key features that enable the emperor to breed successfully on the ice during winter (e.g., absence of nest; presence of incubatory sac; incubation over the feet; laying single eggs; young forming large, compact creches; the prolonged, alternating male–female incubation cycle; etc.) are also present in the king—a typical Subantarctic breeder with just two of nine breeding localities marginally inside the Antarctic Convergence (Heard and South Georgia Islands). The presence of another Aptenodytes species in the Subantarctic—the Pliocene fossil A. ridgeni from New Zealand—further supports this view.
Conclusions
The inclusion of fossil taxa in phylogenetic analyses of the Sphenisciformes expands our understanding of penguin evolution and biogeography. Despite large amounts of missing data for many fossils, the strict consensus trees for both the morphology and combined analyses are largely resolved. The topology of these trees reveals multiple shifts in body size in early penguins, places transitions in the forelimb skeleton in a phylogenetic context, and reveals that a few cranial characters related to feeding (i.e., reduced adductor musculature) are restricted to the crown clade. Our results also reveal that the subfamily level taxonomy of penguins currently available does not reflect the group's phylogenetic relationships and should be revised. The inclusion of fossils provides new data for biogeographical reconstructions and supports a Subantarctic ancestral breeding range for the Spheniscidae.
Acknowledgments
Mark Florence, Storrs Olsen, and Chris Milensky (USNM), Carl Mehling, J. Cracraft, F. Vuilleumier, G. Barrowclough and Paul Sweet (AMNH), Pat Holroyd (UCMP), P. A. Tubaro (MACN); E. Alabarce (COL). Adams and R. Prys‐Jones (BMNH) D. Willard, J. Bates and S. Hackett (FMNH); and K. Garrett (LACM) provided access to specimens. Marcelo Stucchi kindly provided images and information on S. urbinai. Kristin Lamm created the artwork in Fig. 1. Two anonymous reviewers, Julia Clarke, and Kristin Lamm contributed helpful suggestions on this manuscript. D. Ksepka thanks the Doris and Samuel P. Welles fund for providing travel assistance to study the UCMP collections. S. Bertelli thanks the Chapman Memorial Fund at the AMNH and the Antorchas Foundation at the LACM for postdoctoral support.
Appendices
Appendix 1: List of specimens examined
Abbreviations of Institutions: AMNH, American Museum of Natural History, New York, USA; BMNH, Natural History Museum, Tring, UK; COL, Coleccion Ornitologica Lillo, Facultad de Ciencias Naturales e IML, Tucumán, Argentina; FMNH, Field Museum of Natural History, Chicago, USA; LACM, Natural History Museum of Los Angeles County, Los Angeles, USA; MACN, Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”, Buenos Aires, Argentina; MONZ, Museum of New Zealand, Wellington, New Zealand; OM, Otago Museum, Dunedin, New Zealand; RM, Riksmuseet, Stockholm, Sweden; UCMP, University of California Museum of Paleontology, Berkeley, CA, USA; USNM, National Museum of Natural History, Smithsonian Institution, Washington, DC, USA. Taxa are listed alphabetically.
(1) Fossil specimens
Anthropornis nordenskjoeldi BMNH A2024 (cast of RM A8), BMNH A2103 (cast of RM A43), USNM 402486; Delphinornis larseni USNM 404467; Palaeospheniscus patagonicus AMNH 3274, 3276, 3285, 3287, 3289, 3295, 3295, 3297, 3298, 3316, 3321, 3323, 3330, 3336, 3340, 3343, 3344, 3349, 3352, 3355, 3358; Paraptenodytpes antarcticus AMNH 3338; Pachydyptes ponderosus BMNH A3262 (cast of holotype); Platydyptes novaezealandiae USNM Acc 363732 [cast of NMNH (now MONZ) 1451 (holotype)]; Palaeeudyptes antarcticus BMNH A1048 (holotype), USNM Acc 377506 (cast of OM C.47.35), cast of OM C.50.28, cast of OM C.48.74, cast of OM C.47.17 and cast of OM C.50.25; Palaeeudyptes gunnari BNMH A2001 [cast of RM A7 (holotype)], BMNH A 3341; Palaeeudyptes klekowski UCMP 21486.
(2) Integumentary specimens
Aptenodytes forsteri AMNH 196281–2, 265389, 435602–7, 435609–11, 435613–14, 525862, 525867–71, 793545, LACM 69986; Aptenodytes patagonicus AMNH 132452–3, 132456, 132459, 132460–1, 269635, 428536, 435830, 435832, 443456, 442457, 525873–7, 525879–84, 525886, 525888–9, COL 10988; Daption capense AMNH 445493–4, COL 8089–93, 8095; Eudyptes chrysocome chrysocome AMNH 196414, 211923, 211991–2, 211994, 211996–7, 211999, 212002, 212009, 215732, 215759, 445213–20, 445222–6, 525731, 525733, 525736, 525738, 525740, 525743, 525745–52, 525754–8, 525760, 525762–3, 525765–73, COL 10986, 582–263; Eudyptes chrysolophus AMNH 196167–71, 349606, 445227–8, 5257926, BMNH 1905.12.30.164 [holotype], COL 10987; Eudyptes pachyrrhynchus AMNH 29898, 525703–4, 525707–8, 525713, 525726, BMNH 1845.1.13.29 [holotype]; Eudyptes schlegeli AMNH 211984, 525798–810, 525812; Eudyptes sclateri AMNH 196413, 212001, 212003–5, 212007, 212010–11, 525714, 525722, 525777, 525780–5, 525787, 525789, 525791, BMNH 1889.4.7.1 [holotype]; Eudyptes robustus AMNH 525720–1, 525723, 748427; Eudyptula minor albosignata AMNH 212016, 212018–19, 212021–5, 220909, 225701, 525702; Eudyptula minor minor AMNH 168804, 220910, 212013–15, 525528, 525599–603, 525605–21, 525623–4, 525627, 525629–30, 5256324, 525636–7, 525651, 525685, LACM 23909–11; Diomedea melanophrys AMNH 445399, 445410, 445417–18, 445420, COL 5936–9, 598–2729; Gavia stellata AMNH 525949, 525951, 525987, 703321, BMNH 1935.9.9.36, 1937.10.17.157, 1937.10.17.158, 1965.m.134; Macronectes giganteus AMNH 211703, 749313, COL 600, 5940–1; Megadyptes antipodes AMNH 461127–8, 525843–59, 525861–2, 5258646, LACM 101235; Oceanites oceanicus COL 8121, 8123, 8125–9; Oceanodroma leucorrhoa AMNH 70249, 70252, 79397, 349391, 528706; Pachyptila desolata AMNH 132519, 269679, 269682, 269285, 407676, COL 12831; Pelecanoides urinatrix AMNH 212088, 334637, 528738, 528760, 748433, COL 12445; Phoebetria palpebrata AMNH 26714, 132538, 196430, 211408, 527084, 749332, COL 10992; Procellaria aequinoctialis AMNH 211612, 211616, 527317, COL 10095–6; Pterodroma incerta AMNH 132491–2, 269656, 269659, 527987; Puffinus griseus AMNH 211986–7, 748419, 749268, 527606, COL 4219,5893–7; Pygoscelis adeliae AMNH196161–3, 325834, 325245, 325247–9, 435615–16, 442407–8, 442410–11, 525833, 525835–9, LACM 54430, 63897, COL 8138; Pygoscelis antarctica AMNH 196159, 132460, 168780,168806, 190158, 196156, 196160, 349607, 442419–21, 525840–2, 775711, COL 8136; Pygoscelis papua AMNH 132462–9, 196165–6, 196415, 269638–9, 435821–9, 442412–17, 442458–62, 445204–12, 525819–20, 525822, 525824–31, 775712, COL 8019, 8137, 14043, 15837–9, two specimens uncataloged; Spheniscus demersus AMNH 139880, 300427, 525902–3, 525905, 525908–11, 525913–16, BM 1911.12.18.13, 1939.5.11.1, 1939.5.11.2, 1987.2.4.191, 1987.24.211, 1987.24.212, LACM 20531; Spheniscus humboldti AMNH 147869, 196395, 305680, 305682–5, 308767, 357489, 44522931, 525937, 828554, BM 1913.10.11.112, LACM 18007, 18473, 18498, 25439–40, 50001–2, 84448, COL uncataloged; Spheniscus magellanicus AMNH 445240–1, 525890, LACM 25436–8, 69988, COL 4206, 7217, 7235–8, 12378, 16001, 581–2642, two specimens uncataloged; Spheniscus mendiculus AMNH 178170, 196305–6, 266348, 292163, 292233–4, 292636, 299219, 407655, 442424, 525917, 525919–34, 525936, 804313–15, LACM 18649, 18653, 20056, 30295–8, 35882, 84451.
(3) Osteological specimens
Aptenodytes forsteri AMNH 3745, 3725, 3727, 3767, 4856, 8110–12, 11634, 12002, FMNH 106829, 339515, LACM 99854; Aptenodytes patagonicus AMNH 83, 2611, 4382, 26471–2, FMNH 351211, 106475, 107781, LACM 86301, 111055; Daption capense AMNH 3126, 8868, 8915, LACM 102367, 102496, 102534, 102571, 102579, 102597; Diomedea melanophrys AMNH 3135, 23506, 23564–5, MACN 54409–10; Eudyptes chrysocome chrysocome AMNH 5398, 5972, 9173, 26477, LACM 99775; Eudyptes chrysocome moseleyi FMNH 291231–3, 312902, 314652, 345115–18; Eudyptes chrysolophus AMNH 5964, 26478, FMNH 431779, 339520, LACM 104403, 110995; Eudyptes pachyrhynchus AMNH 26509; Eudyptes robustus AMNH 27678, Eudyptes schlegeli AMNH 5399; Eudyptes sclateri BMNH 1952.1.38; Eudyptula minor AMNH 5314, 5620, 6957, 11561, FMNH 339521, 106434, 106492, 106505–6, 106997, LACM 90093, 102347–9, 102351, 102353–4, 103167–8, 103938; Gavia stellata AMNH 4974, 5971, 12905, 24132, LACM 86316, 90036, 99695, 100736, 101089–90, 112288, 112745, 112747–8, 112751; Macronectes giganteus AMNH 1634, 5400, LACM 102362–3, 103782, 104882, MACN 1655a, 4474a; Megadyptes antipodes AMNH 5613, 5615; Oceanites oceanicus AMNH 1357, 16599, LACM 102519, 102553, 102590, 103972, 104105, 104108–9, 104610; Oceanodroma leucorrhoa AMNH 22019, 22032, LACM 18127, 18129, 20081, 86402, 104314, 104316, 107468, 107969; Pachyptila desolata AMNH 3119, 4145, LACM 102501–2, 102510, 102522, 102532, 104299, 106728–9; Pelecanoides urinatrix AMNH 16043, FMNH 339530; Phoebetria palpebrata LACM 102361; Procellaria aequinoctialis AMNH 1638, 8929, LACM 101822, 102261, 102381–2, 103767, 104102, 110997; Pterodroma incerta AMNH 3120, 4693; Puffinus griseus AMNH 2902, 23554, LACM 86374, 103648, 103786–7, MACN 68380; Pygoscelis adeliae AMNH 26162–3, 26474–6, FMNH 96173, 104041, 339519, 378439, 104213–14, 106733, LACM 101721, 101722; Pygoscelis antarctica AMNH 26158–9; 26160–1, FMNH 104215–17, 390994; Pygoscelis papua AMNH 3191, 4361, 5766, 22679, 26164, FMNH 315111, 330150–4, 339516–18, 344864, 345119; Spheniscus demersus AMNH 1310, 1625, 3875, 4069, 10898, 10926, 12782, 22678, FMNH 289829, 291343, 290227, 291234, 375397, 378685, 378754, 104173, 104434, LACM 111061; Spheniscus humboldti AMNH 4920, 26165, FMNH 330157, 339522–5, 339527–9, LACM 18473, 18498, 88834; Spheniscus magellanicus AMNH 823, 8826, 22677, 26479, 26481, FMNH 379141, LACM 90632, 101723, MACN 54412, 54682–3, 54685; Spheniscus mendiculus AMNH 3648, 3772, FMNH 105189, 105390, 105405, 105586, LACM 16055, 18653, 86303–4, 89906.
(4) Oological specimens
Aptenodytes patagonicus AMNH 5843, 13389–91, 13547–8, 15536; Daption capense AMNH 6326, 13444–5; Eudyptes chrysocome AMNH 13392, 13412–15; Eudyptes chrysolophus AMNH 17297, 17301; Eudyptes robustus AMNH 17217; Eudyptula minor AMNH 5846; Gavia stellata AMNH 39, 42, 44, 46, 6288; Macronectes giganteus AMNH 6130, 6136, 13438, 13440, 13555; Oceanites oceanicus AMNH 13497 [type]; Oceanodroma leucorrhoa AMNH 6329, 15537; Pachyptila desolata AMNH 13464, 14572–3, 14576; Pelecanoides urinatrix AMNH 13513–14; Phoebetria palpebrata AMNH 6709–10, 13437; Procellaria aequinoctialis AMNH 13456, 13461–2; Puffinus griseus AMNH 3034, 5858, 6308, 13489–90; Pygoscelis antarctica AMNH 13411; Pygoscelis adeliae AMNH 5844, 13410; Pygoscelis papua AMNH 13393–6, 13400, 13549–51; Spheniscus demersus AMNH 17477; Spheniscus magellanicus AMNH 13416–19, 13604; Spheniscus humboldti AMNH 13603.
Appendix 2: Morphological character matrix
The matrix includes the osteological, integumentary, breeding, and myological characters compiled by Bertelli and Giannini (2005) for a recent analysis of extant penguin relationships. Coding modifications and one additional character from Bertelli et al., in press) are incorporated. We here add 26 novel osteological characters as well as several new character states for previously published characters to this foundation. Characters that are new to this analysis or have been modified are marked by an asterix and are illustrated. Full character description, discussion of sources, and illustrations of previously used characters are provided in Bertelli and Giannini (2005). Anatomical nomenclature and terms of orientation follow Baumel et al. (1993), and English equivalents of some terms are also provided. Many previous authors describing fossil penguins oriented the humerus as in a swimming animal. Therefore, the dorsal and ventral orientations in our descriptions correspond to the preaxial and postaxial orientations of earlier workers. Likewise, our anterior and posterior orientations correspond to the ventral and dorsal orientations, respectively, of earlier workers.
Bill
0. Tip of maxilla (rostrum maxillare): pointed (0); or hooked (1). We scored the fossils Spheniscus urbinai and Spheniscus megaramphus (1) because the tip of the premaxilla is strongly hooked in both species.
1. Tip of mandible (rostrum mandibulare): pointed (0); slightly truncated (1); strongly truncated, squared off (2); procellariform‐like (3). In Procellariiformes, the bill tip is squared but with a rounded margin in lateral view.
2. Longitudinal grooves on the base of the culmen: absent (0); present (1).
3. Longitudinal grooves on the base of latericorn and ramicorn: absent (0); present (1).
4. Feathering of maxilla (rostrum maxillare), extent: totally unfeathered (0); slightly feathered, less than half the length of maxilla (1); feathering that reaches half the length of maxilla (2); feathering surpassing half the length of maxilla (3). Ordered.
5. Ramicorn, inner groove at tip: absent (0); present and single (1); present and double (2). Ordered.
6. Orange or pink plate on ramicorn: absent (0); present (1).
7. Plates of ramphotheca, inflated aspect: absent (0); present (1).
8. Gape (rima oris), aspect: not fleshy (0); margin narrowly fleshy (1); margin markedly fleshy (2). Ordered.
9. Ramicorn color pattern: black (0); reddish (1); pink (2); yellowish (3); orange (4); green (5); blue (6).
10. Latericorn and ramicorn, light distal mark: absent (0); present (1).
11. Latericorn color: black (0); red (1); orange (2); yellow (3); green (4); blue (5).
12. Culminicorn color: black (0); red (1); orange (2).
13. Maxillary and mandibulary unguis, color: black (0); red (1); yellow (2); green (3); bluish gray (4).
14. Bill of downy chick, color: dark (0); reddish (1); pale, variably horn to yellowish (2); bluish (3).
15. Bill of immature, color: dark (0); bicolored reddish and black (1); red (2); yellowish (3); grayish (4).
16. Nostril tubes: absent in adult (0); present in adult (1).
*17. Nostril tubes: absent in hatchling (0); present in hatchling (1). Tube nostrils are present in hatchlings of Eudyptula, but not adults (Kinsky, 1960). In previous analyses the presence of nostril tubes in hatchlings was incorporated as a third state of character 16. We consider the adult and juvenile conditions separately in the present analysis.
18. External nares: present (0); absent (1).
Iris
19. Iris color: dark (0); reddish‐brown (1); claret red (2); yellow (3); white (4); silvery gray (5).
Feathers
20. Scale‐like feathers: absent (0); present (1).
21. Rhachis of contour feathers: cylindrical (0); flat and broad (1).
22. Rectrices: form a fan functional for steering (0); do not form a functional fan (1).
23. Remiges: differentiated from contour feathers and specialized for flight propulsion (0); indistinct from contour feathers (1).
24. Apteria: present (0); absent (1).
25. Molt of contour feathers: gradual (0); simultaneous (1).
Adult plumage
26. Yellow pigmentation in crown feathers (pileum): absent (0); present (1).
27. Head plumes (crista pennae): absent (0); present (1).
28. Head plumes (crista pennae), aspect: compact (0); sparse (1).
29. Head plumes (crista pennae), aspect: heading upward (0); heading backward, not drooping (1); heading backward, drooping (2).
30. Head plumes (crista pennae), position of origin: at base of bill close to gape (0); on the recess between latericorn and culminicorn (1); on forehead (2). Ordered.
31. Head plumes (crista pennae), color: yellowish (0); orange (1).
32. Nape (occiput), crest development: absent (0); slight (1); distinct (2). Ordered.
33. Periocular area (regio orbitalis), color: black (0); white (1); yellow (2); bluish gray (3).
34. Fleshy eyering (regio orbitalis): absent (0); present (1).
35. White eyering (regio orbitalis): absent (0); present (1).
36. White eyebrow (regio orbitalis, supercilium): absent (0); narrow, from postocular area (1); narrow, from preocular area (2); wide, from preocular area (3). Ordered.
37. Loreal area (lorum), aspect: feathered (0); with spot of bare skin in the recess between latericorn and culminicorn (1); with spot of bare skin contacting eye (2); bare skin extending to the base of bill (3). Ordered.
38. Auricular patch (regio auricularis): absent (0); present (1).
39. Throat pattern (jugulum): black (0); white (1); yellowish (2); irregularly streaked (3); with chinstrap (4).
40. Collar: absent (0); at most slight notch present (1); present, diffusely demarcated (2); black, strongly demarked (3). Ordered.
41. Breast (pectus), golden in color: absent (0); present (1).
42. Dorsum color: black (0); dark bluish gray (1); light bluish gray (2).
43. Black marginal edge of dorsum between lateral collar and axillary patch (axilla), contrasting with dorsum: absent (0); present (1).
44. Black dots irregularly distributed over white belly (venter): absent (0); present (1).
45. Flanks (ilia), dark lateral band reaching the breast (pectus): absent (0); present (1).
46. Distinct dark axillary patch of triangular shape (axila): absent (0); present (1).
47. Flanks (ilia), extent of dorsal dark cover into the leg: incomplete, not reaching tarsus (0); complete, reaching tarsus (1).
48. Rump (pyga): indistinct from dorsum (0); distinctly whitish (1).
49. Tail length (cauda): short, the quills barely emerge from the rump (0); the quills are distinctly developed (1).
50. Outer rectrices, color: same as inner rectrices (0); lighter than inner rectrices (1).
51. White line connecting leading edge of flipper with white belly (venter): absent (0); present (1).
52. Flipper (ala[membrum thoracicum]), upperside, light notch at base: absent (0); present (1).
53. Leading edge of flipper (ala[membrum thoracicum]), pattern of upperside: black (0); white (1).
54. Leading edge of flipper (ala[membrum thoracicum]), pattern of underside: white (0); incompletely dark (1); completely dark and wide (2). Ordered.
55. Flipper (ala[membrum thoracicum]), underside, dark elbow patch: absent (0); present (1).
56. Flipper (ala [membrum thoracicum]), underside, tip pattern: immaculate (0); patchy, in variable extent (1); small circular dot present (2).
Immature plumage
57. White eyebrow (supercilium): absent (0); or present (1).
58. Throat pattern (jugulum): black (0); or mottled (1); or white (2); or brown (3).
59. Flanks (ilia), dark lateral band: absent (0); or present (1).
Natal plumage
60. Chicks hatch almost naked: no (0); yes (1).
61. Dominant color pattern of first down: pale gray (0); distinctly brown (1); bicolored, dark above and whitish below (2); uniformly blackish gray (3).
62. Dominant color pattern of second down: pale gray (0); distinctly brown (1); bicolored, dark above and whitish below (2); uniformly blackish gray (3).
63. Chick, second down, collar: absent (0); present (1).
Feet
64. Feet (pedes), dorsal color: dark (0); pinkish (1); orange (2); white‐flesh (3); blue (4).
65. Feet (pedes), dark soles: absent (0); present (1).
66. Feet (pedes), unguis digiti: flat (0); compressed (1).
Breeding characters
67. Clutch size: two eggs (0); one egg (1).
68. Incubatory sac: absent (0); present (1).
69. Nest: no nest, incubation over the feet (0); nest placed underground, either burrowed on sand or inside natural hollow or crack (1); open nest, a shallow depression on bare ground or in midst of vegetation (2).
70. Size of first egg relative to the second egg: similar (0); dissimilar, first smaller (1); dissimilar, second smaller (2).
71. Crèche: absent (0); small, 3–6 birds (1); large, formed by dozens to hundreds of immatures (2). We added a coding for Spheniscus mendiculus based on data from Williams (1995).
72. Eggs, shape: oval (0); conical (1); spherical (2).
73. Ecstatic display: absence (0); presence (1).
Osteological characters
Cranium
74. Basioccipital, subcondylar fossa (os basioccipitale, fossa subcondylaris): absent or shallow (0); deep (1). Marples (1960) referred to this feature as the precondylar fossa and noted the morphology in Marplesornis is very similar to Spheniscus humboldti and unlike Paraptenodytes antarcticus.
75. Supraoccipital, paired grooves for the exit of versus occipitalis externae (os supraoccipitale, sulcus vena occipitalis externae): poorly developed (0); deeply excavated (1).
*76. Cerebellar prominence, posterior projection (Fig. 5): weak or moderate (0); strong (1). The cerebellar prominence is extremely well‐developed in extant penguins, but less so in outgroup taxa and some fossil penguins. Marples (1960) noted a weaker development in Marplesornis.

Dorsal view of the skulls of Paraptenodytes antarcticus (AMNH 3338) and Eudyptes pachyrhynchus (AMNH 22679). Abbreviations: fs, fossa temporalis; pc, prominentia cerebellaris. Scale bar = 1 cm.
77. Frontal, shelf of bone bounding salt‐gland fossa (os frontale, fossa glandulae nasalis) laterally: absent (0); present (1).
78. Squamosal, temporal fossa (os squamosum, fossa temporalis), size (Fig. 5): fossae separated by a wide surface (at least the width of the cerebellar prominence) (0); more extensive, fossae meeting or nearly meeting at midline of the skull (1).
79. Squamosal, temporal fossa (os squamosum, fossa temporalis), depth of posterior region: flat (0); shallow (1); greatly deepened (2). Ordered.
80. Squamosal (os squamosum), development of the opening/s that transmits a. ophthalmica externa in the caudoventral area of the fossa temporalis (near the crista nuchalis): small or vestigial (0); well‐developed (1).
81. Orbit (orbita), fonticuli orbitocraneales: small or vestigial (0); broad and conspicuous openings (1).
82. Ectethmoid (os ectethmoidale): absent (0); weakly developed, widely separate from the lacrimal (1); well developed, contacting or fused to the lacrimal (2).
83. Lacrimal (os lacrimale): unperforated (0); perforated (1).
*84. Lacrimal (os lacrimale): reduced, concealed in dorsal view (0); small portion exposed in dorsal view (1); well‐exposed in dorsal view (2). Previously this character was divided into two states. After further study, we opted to add an intermediate state to accommodate the condition in Pygoscelis. Ordered.
85. Lacrimal, dorsal border: closely applied to the frontal (0); separated by a wide split from the frontal (1).
86. Nasal cavity, external naris (cavum nasi, apertura nasi ossea), posterior margin: extended posterior to the anterior margin of the hiatus orbitonasalis (0); not extended posterior to the anterior margin of the hiatus orbitonasalis (1).
87. Nasal cavity (cavum nasi, pila supranasalis): slender, slightly constricted laterally (0); wide throughout its length (1).
88. “Basitemporal” plate (lamina parasphenoidalis), dorsoventral position with respect to the occipital condyle: ventral to the level of the condyle (0); at the level of the condyle (1); dorsal to the level of the condyle, surface depressed (2). Ordered.
89. Basipterygoid process: absent (0); vestigial or poorly developed (1); well developed (2). Ordered.
90. Eustachian tubes (tuba auditiva): open or very little bony covering near the posterior end of the tube (0); mostly enclosed by bone (1).
*91. Pterygoid (os pterygoideum), shape: elongated (0); slight lateral expansion of anterior end (1); anterior end broad, pterygoid subtriangular (2). We have added an additional state to this character to account for the moderate expansion in Marplesornis. Ordered.
92. Palatine (os palatinum), lamella choanalis: curved and smooth plate, slightly differentiated from main palatine blade (0); ridged, distinct from main blade by a low keel (1); extended vertically ventrally forming the crista ventralis (2). Ordered.
93. Vomer (os vomer): laterally compressed, vertical laminae and free from palatines (0); horizontally flattened laminae and ankylosed with palatines (1).
94. Facial foramen (ossa otica, fossa acustica interna, foramen n. facialis): absent (0); present (1).
95. Jugal arch (arcus jugalis), bar shape in lateral view: straight (0); slightly curved (1); ventrally bowed (2); strongly curved, sigmoid shape (3). Ordered.
96. Jugal arch (arcus jugalis), dorsal process: absent (0); present (1). This pointed process is located on the posterior end of the jugal, adjacent to the condyle for articulation with the quadrate.
97. Premaxilla, frontal process (os premaxillare, proc. frontalis), naso‐premaxillary suture: visible (0); obliterated (1).
98. Quadrate, otic process (os quadratum, proc. oticus), anterior border, process for attachment of the M. adductor mandibulae externus, pars profunda: absent (0); present, as a ridge (1); present, as a tubercle (2).
99. Tomial edge (crista tomialis): plane of tomial edge approximately at the level of the basitemporal plate (lamina parasphenoidalis) (0); dorsal to the level of the basitemporal plate (1).
100. Mandible, coronoid process (mandibula, processus coronoideus), position on the dorsal margin of the mandible with respect to the caudal fenestra (fenestra mandibulae caudalis): markedly anterior to fenestra (0); level with anterior end of the fenestra (1); posterior to fenestra (2). Ordered.
101. Mandible, rostral fenestra (mandibula, fenestra mandibulae rostralis): imperforate or small opening (0); large opening (1).
102. Mandible, caudal fenestra (mandibula, fenestra mandibulae caudalis): open, can be seen through from the medial or lateral aspects (0); nearly or completely concealed by the os spleniale medially, i.e., fenestra not visible in the medial aspect (1).
103. Mandible, mandibular ramus (mandibula, ramus mandibulae): depth subequal over entire ramus (0); pronounced deepening at midpoint (1).
104. Mandible, dentary, length of dorsal edge (ossa mandibulae, os dentale, margo dorsalis) relative to mandibular ramus length in lateral view: markedly more than half the length of ramus (0); approximately half the length of ramus (1).
105. Mandible, articular, medial process (os articulare, proc. medialis mandibulae) (Fig. 6): not hooked (0); hooked (1).

Ventral view of the articular end of the mandibles of Eudyptes pachyrhynchus (AMNH 26509), Pygoscelis papua (AMNH 22679) and Paraptenodytes antarcticus (AMNH 3338). Abbreviations: pm, proccesus medialis mandibulae; pr, proccesus retroarticularis. Scale bar = 1 cm.
106. Mandible, angular, retroarticular process (mandibula, os angulare, proc. retroarticularis), aspect in dorsal view in relation to the articular area with the quadrate (area between the lateral cotyle [cotyla lateralis] and medial cotyle [cotyla medialis]): broad, approximately equal to the articular area (0); moderately long, narrower than the articular area (1); very long, longer and narrower than the articular area (2). Ordered.
107. Mandible, angular (mandibula, os angulare), aspect in dorsal view: sharply truncated posteriorly (0); posteriorly projected, forming the retroarticular process (proc. retroarticularis) (1).
*108. Mandible, medial emargination between retroarticular process and medial process of angular (Fig. 6): absent (0); weak concavity (1); strong concavity (2). This character is easiest to evaluate in ventral view. This character is non‐comparable in taxa lacking a retroarticular process. Ordered.
109. Atlas (atlas), proc. ventralis: absent or slightly developed (0); well‐developed, high and prominent ridge on the dorsal surface of the arcus atlantis (1).
110. Transition to free cervicothoracic ribs (costae incompletae) arrives at: 13th cervical vertebrae (vertebrae cervicodorsales): (0); 14th cervical vertebrae (1); 15th cervical vertebrae (2). Ordered.
111. Cervical vertebrae (vertebrae cervicales), elongated dorsal process (processus spinosus) on the sixth cervical vertebra: absent (0); present (1).
112. Cervical vertebrae (vertebrae cervicales), transverse processes (processus transversus) in last five cervical vertebrae: not elongated laterally (0); greatly elongated laterally (1).
113. Cervical vertebrae (vertebrae cervicales), transverse processes (processus transversus) of vertebrae 12–13: laterally oriented (0); deflected dorsally (1).
*114. Posterior‐most thoracic vertebra: heterocoelous (0); weakly opisthocoelous; (1); strongly opisthocoelous (2). Ordered.
115. Caudal vertebrae (vertebrae caudales): seven (0); eight (1); nine (2). Ordered.
116. Ribs, uncinate processes (costae, proc. uncinatus): elongate, narrow (0); wide, spatulate (1); wide, bifurcated (2).
117. Sternum, external spine (sternum, spina externa rostri): absent (0); present (1).
118. Sternum, furcular facet (sternum, facies articularis furculae) projects as a distinctive process: absent (0); present (1).
119. Furcula, furcular process (furcula, apophysis furculae): absent or low blade‐like process (1); knob‐like process (2); long process (3).
120. Scapula, blade, caudal half (scapula, corpus scapulae, extremitas caudalis): blade‐like (0); expanded and paddle‐shaped (1).
121. Coracoid, medial margin (coracoideum, margo medialis) coracoidal fenestra (Fig. 7): complete (0); incomplete (1); absent (2). This character was described by O'Hara (1989; character 14) as the fenestrate lamella. Zusi (1975, p. 61) described this character as the coracoidal fenestra, and observed the same distribution of this feature within penguins as O'Hara (1989) and the present study. In some penguins (Aptenodytes, Pygoscelis, Eudyptula), the medial margin of the coracoid forms an incomplete ridge that does not reach the procoracoid process dorsally, leaving a wide notch between the medial margin ridge and the procoracoid process (state 1). In other penguins, the medial margin projects dorsally reaching the procoracoid process so that a large fenestra is enclosed in the substance of the medial ridge (state 0). This area is obscured in the holotype of Marplesornis.Marples (1960) considered the fenestra complete in Marplesornis, but Simpson (1972b) disagreed and observed that it was probably open laterally. Because of this discrepancy we have opted not to code Marplesornis for this character.

Coracoid of Procellaria aequinoctialis (AMNH 8929), cast of coracoid of Anthropornis nordenskjoeldi (BMNH A2024) and coracoid of Aptenodytes forsteri (AMNH 3767) in anterior view. Abbreviations: cf, coracoidal fenestra; fns, foramen nervi supracoracoidei; pp, processus procoracoideus. Scale bar = 1 cm.
122. Coracoid, foramen nervi supracoracoidei (Fig. 7): absent (0); present (1). This character is modified from Mayr (2004; character 29). Mayr (2004) cited ontogenetic evidence indicating this foramen is non‐homologous to the coracoidal fenestra of penguins (see character 121). We therefore treat the presence or absence of the foramen nervi supracoracoidei as a separate character.
123. Coracoid, sternal end (coracoideum, extremitas sternalis coracoidei) (Fig. 7): wide (0); narrow (1).
*124. Coracoid, sternal end (coracoideum, extremitas sternalis coracoidei) (Fig. 7): convex (0); concave (1), flat (2). This feature was discussed by Marples (1952), who considered it a striking difference between fossil and living penguins. Small isolated penguin coracoids from the Eocene of Seymour Island in the UCMP collection preserve a convex base, providing evidence this feature is not size related.
125. Forelimbs (ossa alae), flattened: absent (0); present (1).
126. Humerus, head (humerus, caput humeri): very developed and reniform, continuous with tubercucum dorsale: absent (0); present (1).
*127. Humerus, incisura captius (Fig. 10): deep and essentially confluent with sulcus transversus (0); shallow, clearly separated from sulcus transversus (1).

Humerus of Eudyptes chrysocome moseleyi (FMNH 291232) and cast of humerus of Anthropornis nordenskjoeldi (BMNH A2013) in anterior view. Abbreviations; ic: incisura capitus; sl: sulcus transversus; tb: tuberculum ventrale. Scale bar = 1 cm.
*128. Humerus, pit for ligament insertion on proximal surface adjacent to head (Fig. 8): absent or very shallow (0); deep (1).

Humeri of Palaeospheniscus patagonicus (AMNH 3340) and Spheniscus humboldti (AMNH 4921) in proximal view. Abbreviations: fp: proximal margin of fossa pneumotricipitalis; lp: pit for ligament insertion; st: sulcus transversus. Scale bar = 1 cm.
*129: Humerus, tuberculum ventrale (Fig. 10): powerfully developed (0); reduced, well exposed in posterior view (1); reduced, so as to be partially obscured in posterior view (2). Marples (1952) noted differences in the orientation of this feature and referred to it as the internal tuberosity.
130. Humerus, pneumatic fossa (humerus, fossa pneumotricipitalis ventralis), aspect: small with penetrating pneumatic foramina (0); moderate fossa without pneumatic foramen (1); deep fossa without pneumatic foramen (2).
131. Humerus, pneumatic fossa (humerus, fossa pneumotricipitalis ventralis), subdivided into cavities: single (0); bipartite (1). The strength of the crest dividing the chambers of the pneumatic fossa is variable in extant penguins, but all possess at least a weakly divided fossa.
132. Humerus, deltoid crest, impression for attachment of pectoral muscle (humerus, crista deltopectoralis, impressio M. pectoralis): superficial or shallow groove (0); deep oblong fossa (1).
*133. Humerus, insertion of m. supracoracoideus(Fig. 9): small, semicircular scar (0); greatly elongated with long axis parallel to main axis of humeral shaft (1), greatly elongated with long axis oblique to long axis of shaft (2). Marples (1952) discussed the orientation of this scar but referred to the muscle as m. pectoralis secundus.
*134. Humerus, attachment surface for m. supracoracoideus and m. latissimus dorsi (Fig. 9): separated by a wide gap (0); separated by small gap or confluent (1).
*135. Humerus, proximal margin of pneumotricipital fossa (fossa pneumotricipitalis) (Fig. 8): weak projection (0); projects so as to be well‐exposed in proximal view (1).
*136. Humerus, shaft dorso‐ventral width (Fig. 10): shaft thins or maintains width distally (0); shaft widens distally (1).
*137. Humerus, shaft, sigmoid curvature (Fig. 10): absent or weak (0); strong (1).
138. Humerus, development of dorsal supracondylar process (humerus, proc. supracondylar dorsalis): absent (0); compact tubercle (1); very long process (2).
139. Humerus, distal end, development of trochlear process for articulation with os sesamoideum m. scapulotricipitalis: absent (0); present (1). According to Slack et al. (2006), the sulcus scapulotricipitalis and sulcus humerotricipitalis of Waimanu tuatahi are deep, but the sulcus scapulotricipitalis does not form a trochlea.
140. Humerus, proximal‐most trochlear process: extends beyond the humeral shaft (0); does not extend beyond the humeral shaft (1). While this process clearly extends beyond the humeral shaft in some specimens of Palaeospheniscus, it is difficult to evaluate in many other fossil penguins because of weathering or damage.
*141. Humerus, angle between main axis of shaft and tangent of radial and ulnar condyles (Fig. 10): less than 45 degrees (0); more than 45 degrees (1); nearly 90 degrees (2). Several authors (Lowe, 1933; Simpson, 1946; Marples, 1953) discussed the angle between the humeral shaft and plane of the distal condyles in penguins. Both living and fossil penguins exhibit intra‐ and interspecific variation in this feature. We have chosen to divide this variation into two states that we feel discriminate between the highly acute angle in some fossils and the general condition in penguins. A third state represents the condition typical of the outgroups.
*142. Humerus, ulnar and radial condyles (condylus ventralis and condylus dorsalis) (Fig. 10): radial and ulnar condyles projected and rounded (0); radial condyle flattened, ulnar condyle anteriorly displaced from and projecting well distal to radial condyle (1); radial and ulnar condyles flattened, nearly level and confluent (2). Marples (1952) discussed this feature at length and compared the arrangement of the condyles in penguins with other avian groups. We code Marplesornis (1) based on Marples (1960) description of the ulnar condyle as flattened and indistinct in that taxon. Ordered.
*143. Humerus, shelf adjacent to ulnar condyle (condylus dorsalis) (Fig. 11): equal to or wider than condyle (0); narrower than condyle (1). We consider this character non‐comparable for taxa lacking this shelf.

Cast of humerus of Palaeeudyptes antarcticus (BMNH A 4079) and humeri of Paraptenodytes antarcticus (AMNH 3338) and Pygoscelis papua (AMNH 22679) in distal view. Abbreviations: cd, condylus dorsalis; cv, condylus ventralis; sh, shelf adjacent to condylus ventralis. Scale bar = 1 cm.
*144. Ulna, ventral border (Fig. 12): border forms smooth curve with apex located one‐fourth of way to distal end (0); well‐developed tab‐like projection arises proximally, very close to humeral articulation (1); acute projection, distally displaced from humeral facet (2); rounded (3). We formulated a state to accommodate the distinct shape of the ulnae of Archaeospheniscus lowei and Platydyptes described by Marples (1952; see pl. 1 and pl. 4, Fig. 4).
*145. Ulna, distinct process extends toward sulcus humerotricipitalis of humerus: absent (0); present (1). Marples (1952) noted the presence of this character in two specimens he assigned to Platydyptes novazealandiae. Simpson (1971) later designated Marple's specimen “A” (OM C.47.15) the holotype of Platydyptes marpelsi. The distribution of this feature remains unknown in some fossils with preserved ulnae because its delicate nature leaves it vulnerable to damage.
146. Phalanges of manus, free pollex (ossa digitorum manus, phalanx digiti alulae): absent (0); present (1).
147. Phalanges of manus, phalange digit III (ossa digitorum manus, phalanx digiti minoris), proximal process: absent (0); present (1).
148. Phalanges of manus (ossa digitorum manus, phalanxs): short (0); long (1)
*149. Fusion of ilium to synsacrum (Fig. 13): unfused (0); partially fused (1); well‐fused (2). Clarke et al. (2003) noted fusion of the ilium and synsacrum in individual specimens of Pygoscelis antarctica and Pygoscelis papua. We observed fusion in all adult specimens we examined. We also note that these elements are typically partially fused, but never completely fused, in specimens of Pygoscelis adeliae and so have included an intermediate state for that taxon. Ordered.

Pelvis of Pygoscelis papua (AMNH 22679) and Eudyptes pachyrhynchus (AMNH 26509) in dorsal and lateral views. Abbreviations: cn: concavity on lateral surface of pelvis; si, sutura iliosynsacralis; vt, vertebra thoracica. Scale bar = 1 cm.
*150. Pelvis, depression posterior to ilio‐ishiatic fenestra (foramen ilioischiadicum) (Fig. 12): absent (0); present (1). In penguins, a sheet of bone unites the ilium and ischium. In Eudyptes and Megadyptes, the area directly posterior of this fenestra is strongly concave.
151. Pelvis (pelvis et os coxae), size of ilio‐ischiatic fenestra (foramen ilioischiadicum) in relation to acetabulum (foramen acetabuli): smaller (0); similar or larger (1).
152. Pelvis, ischio‐pubic fenestra (pelvis et os coxae, fenestra ischiopubica): very wide and closed at its posterior end (0); slit‐like and open at its posterior end (1).
153. Ischium, most posterior extent in relation to postacetabular ilium (ala postacetabularis ilii): ischium shorter than postilium (0); ischium projects slightly beyond the postilium (1); ischium produced far posterior to postilium (2).
154. Femur, patella (femur, patella), sulcus m. ambiens: shallow groove (0); deep groove (1); perforated (2). A perforated patella probably belonging to Anthropornis was reported by Marples (1952), but it cannot be assigned to this species with complete certainty.
155. Tibiotarsus, patellar crest (tibiotarsus, crista patellaris): greatly enlarged (0); slightly developed (1).
156. Tarsometatarsus (tarsometatarsus): slender, proximodistal length much greater than mediolateral width (EI > 3) (0); shortened, 2.5 < EI < 3 (1); EI < 2.5 (2). Bertelli and Giannini (2005) used two states to describe the difference between the extremely shortened penguin tarsometatrsus and the elongate tarsometatarsus of outgroup taxa. We have added an additional state to encompass variation within Sphenisciformes. We obtained scorings from measurements of specimens as well as from Myrcha et al. (2002), who provided data for many fossil taxa. Ordered.
*157. Tarsometatarsus, margo medialis, pronounced convexity (Fig. 14): absent (0); present (1). Myrcha et al. (2002) listed this character as a diagnostic feature of the genus Anthropornis.

Left tarsometatarsus of Anthropornis nordenskjoeldi (USNM 402486) in dorsal view and right tarsometatarsus of Palaeeudyptes antarcticus (BMNH A 1048) in plantar view. Abbreviations: cih, crista intermediae hypotarsi; clh, crista lateralis hypotarsi, cmh: crista mediae hypotarsi; fpl, foramen vasculare proximale laterale; fpm, foramen vasculare proximale mediale; sdm: sulcus longitudinalis dorsalis medialis. Scale bar = 1 cm.
*158. Tarsometatarsus, intermediate hypotarsal crest (crista intermediae hypotarsi) (Figs 14 and 15): non‐distinct (0); distinct (1). In some fossil penguins, the intermediate crest(s) are clearly differentiated, while in extant penguins and other fossils only a lateral and medial crest are distinguishable.

Right tarsometatarsi of Delphinornis larseni (USNM 404467) and Spheniscus magellanicus (AMNH 26479) in dorsal view and right tarsometatarsus of Palaeospheniscus patagonicus (AMNH 3358) in plantar view. Abbreviations: crista lateralis hypotarsi, cmh: crista mediae hypotarsi, fpl, foramen vasculare proximale laterale; fpm, foramen vasculare proximale mediale; sdm, sulcus longitudinalis dorsalis medialis. Scale bar = 1 cm.
*159. Tarsometatarsus, sulcus longitudialis dorsalis medialis (Fig. 15): absent or barely perceptible (0); shallow groove (1); deep groove (2). Many early researchers considered the tighter fusion of the metatarsals in fossil penguins a primitive feature, and the weaker union in extant penguins a more derived condition. The lateral sulcus is well‐developed in all penguins. In this analysis we consider the development of the medial sulcus, and find considerable variation between the extremely weak sulcus of many fossils and the deep sulcus in extant penguins such as Spheniscus. Ordered.
*160. Tarsometatarsus, metatarsal four: distal end projects laterally (0); straight (1).
161. Tarsometatarsus, foramina vascularia proximalia medialis opening on fossa para hypotarsalis medialis: absent (0); present (1). This character is considered non‐comparable in taxa in which the foramina vascularia proximalia medialis is absent.
*162. Tarsometatarsus, proximal vascular foramina on plantar surface (Figs 14 and 15): medial foramina (foramen vasculare proximale mediale) present, lateral foramina (foramen vasculare proximale laterale) absent or vestigial (0); both foramina present (1); lateral foramina present, medial foramina absent or vestigial (2). The relative sizes of these foramina vary in penguins. In some taxa, one foramen is always well‐developed and the other is absent or highly reduced.
*163. Tarsometatarsus, foramen vasculare distale (Fig. 15): present, separated from incisura intertrochlearis lateralis by osseus bridge (0); present, open distally (1); absent (2). Ordered.
164. Tarsometatarsus, hypotarsus, tendinal canals (tarsometatarsus, hypotarsus, canales hypotarsi): present (0); absent (1).
Myological characters
165. M. latissimus dorsi, pars cranialis, accessory slip: absent (0); present (1).
166. M. latissimus dorsi, pars cranialis and pars caudalis: separated (0); fused (1).
167. M. latissimus dorsi, pars metapagialis, development: wide (0); intermediate (1); narrow (2). Ordered.
168. M. serratus profundus, cranial fascicle: absent (0); present (1).
169. M. deltoideus, pars propatagialis, subdivision in superficial and deep layers: undivided (0); divided (1).
170. M. deltoideus, pars major: triangular or fan‐shaped (0); strap‐shaped (1).
171. M. deltoideus, pars major, caput caudale: short (0); intermediate (1); long (2). Ordered.
172. M. deltoideus, pars minor, origin on the clavicular articulation of the coracoid: absent (0); present (1).
173. M. ulnometacarpalis ventralis: absent (0); present (1).
174. M. iliotrochantericus caudalis: narrow (0); wide (1).
175. M. iliofemoralis, origin: tendinous (0); mostly tendinous (1); mostly fleshy (2); totally fleshy (3). Ordered.
176. M. flexor perforatus digitis IV, rami II–III: free (0); fused (1).
177. M. flexor perforatus digitis IV, rami I–IV: free (0); fused (1).
178. M. flexor perforatus digitis IV, insertion of middle rami: on phalanx 3 (0); on phalanx 4 (1).
179. M. latissimus dorsi, pars caudalis, additional origin from dorsal process of vertebrae (spinous process of Schreiweis, 1982): absent (0); present (1).
Digestive tract
180. Mouth, oral mucosa (bucca, tunica mucosa oris), buccal papillae group (sensuWatson, 1883) on the medial surface of the lower jaw (ramus mandibularis) at the level of the rictus: small number of rudimentary papillae with no clear arrangement (0); two clear rows of short conical papillae (1); large, elongated papillae with no clear arrangement (2). In a previous analysis (Bertelli and Giannini, 2005) we considered only the presence or absence of the buccal papillae. We have here expanded this character to reflect variation within Sphenisciformes.
Appendix 3: Morphological data matrix
| Taxa | 0 0123456789 | 1 0123456789 | 2 0123456789 | 3 0123456789 | 4 0123456789 |
|---|---|---|---|---|---|
| Gavia stellata | 0000100000 | 0000C00002 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Diomedea melanophrys | 1300000004 | 0221001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Phoebetria palpebrata | 1300000000 | 0000001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Macronectes giganteus | 1300100003 | 03‐2231105 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Daption capense | 1300100000 | 0000001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Procellaria aequinoctialis | 1300000005 | 0403?01100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Puffinus griseus | 1300000000 | 0000001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Pterodroma incerta | 1300000000 | 0000001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Oceanodroma leucorrhoa | 1300000000 | 0000001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Oceanites oceanicus | 1300000000 | 0000001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Pachyptila desolata | 1300000006 | 0504341100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Pelecanoides urinatrix | 1300000000 | 00‐0001100 | 000000‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ | ‐‐‐‐‐‐‐‐‐‐ |
| Eudyptula minor | 1110120000 | 0000001005 | 11111100‐‐ | ‐‐03000001 | 0020000100 |
| Aptenodytes forsteri | 0000201000 | 0000000010 | 11111100‐‐ | ‐‐00000010 | 0111000101 |
| Aptenodytes patagonicus | 0000101000 | 0000000010 | 11111100‐‐ | ‐‐00000010 | 0111000101 |
| Pygoscelis antarctica | 1100210000 | 0000000012 | 11111100‐‐ | ‐‐01000004 | 0000001001 |
| Pygoscelis papua | 1100210001 | 0100110011 | 11111100‐‐ | ‐‐00010000 | 0000001011 |
| Pygoscelis adeliae | 1100310001 | 0100110014 | 11111100‐‐ | ‐‐10010000 | 0000001001 |
| Megadyptes antipodes | 1100120002 | 0111000013 | 11111110‐‐ | ‐‐02100002 | 0000000101 |
| Eudyptes c. moseleyi | 1100120101 | 0111020012 | 1111111112 | 1020000000 | 0000000101 |
| Eudyptes c. chrysocome | 1100120111 | 0111020012 | 1111111112 | 1020000000 | 0000000101 |
| Eudyptes chrysolophus | 1100120121 | 0111020012 | 1111111112 | 2100000000 | 0000000111 |
| Eudyptes schlegeli | 1100120121 | 0111020012 | 1111111112 | 2101000001 | 0000000111 |
| Eudyptes pachyrrhynchus | 1100120101 | 0111020016 | 1111111101 | 1000000000 | 0000000101 |
| Eudyptes robustus | 1100120121 | 0111020016 | 1111111101 | 1000000000 | 0000000101 |
| Eudyptes sclateri | 1100120121 | 0111020010 | 1111111100 | 0000000000 | 0000000111 |
| Spheniscus demersus | 1211010000 | 1000000000 | 11111100‐‐ | ‐‐00103200 | 1000110100 |
| Spheniscus humboldti | 1211010000 | 1000000001 | 11111100‐‐ | ‐‐00102303 | 0000110100 |
| Spheniscus magellanicus | 1211010000 | 1000000000 | 11111100‐‐ | ‐‐00102100 | 3000110100 |
| Spheniscus mendiculus | 1211010003 | 1000000000 | 11111100‐‐ | ‐‐00101303 | 2000110100 |
| Spheniscus urbinai | 1????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Spheniscus megarrhamphus | 1????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Waimanu manneringi | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Waimanu tuatahi | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Paraptenodytes antarcticus | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Paleospheniscus patagonicus | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Eretiscus tonnii | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Marplesornis novaezealandiae | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Anthropornis nordenskjoeldi | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Anthropornis grandis | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Pachydyptes ponderosus | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Platydyptes novaezealandiae | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Archaeospheniscus lowei | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| “Palaeeudyptes antarcticus” | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Palaeeudyptes gunnari | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Palaeeudyptes klekowskii | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Marambiornis exilis | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Mesetaornis polaris | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Delphinornis larseni | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Arthrodytes andrewsi | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Taxa | 5 0123456789 | 6 0123456789 | 7 0123456789 | 8 0123456789 | 9 0123456789 |
|---|---|---|---|---|---|
| Gavia stellata | 0000200010 | 0020100100 | 00000–001 | 1100000001 | ‐‐‐‐‐‐‐‐‐‐ |
| Diomedea melanophrys | 0021201002 | 0021100000 | 200000‐0‐0 | 0000010111 | ‐‐‐‐‐‐‐‐‐‐ |
| Phoebetria palpebrata | ?????0???2 | ?????????? | ?????????? | ?????????? | ‐‐‐‐‐‐‐‐‐‐ |
| Macronectes giganteus | 0121201002 | 0021100000 | 000000‐0‐0 | 0000?10110 | ‐‐‐‐‐‐‐‐‐‐ |
| Daption capense | 0121200002 | 0021100000 | 000000‐0‐0 | 0000010112 | ‐‐‐‐‐‐‐‐‐‐ |
| Procellaria aequinoctialis | 1121200002 | 0021100000 | 001000‐0‐0 | 0000010112 | ‐‐‐‐‐‐‐‐‐‐ |
| Puffinus griseus | 1121200002 | 0021100000 | 00A000‐0‐0 | 0100?10112 | ‐‐‐‐‐‐‐‐‐‐ |
| Pterodroma incerta | 0121200002 | 0021100000 | 001000‐0‐0 | ?????????? | ‐‐‐‐‐‐‐‐‐‐ |
| Oceanodroma leucorrhoa | 0121210011 | 0021100000 | ?01000‐0‐0 | 0000010112 | ‐‐‐‐‐‐‐‐‐‐ |
| Oceanites oceanicus | 0121210011 | 0021100000 | ?01000‐0‐0 | 0000?10112 | ‐‐‐‐‐‐‐‐‐‐ |
| Pachyptila desolata | 0121200012 | 0021100000 | 001000‐0‐0 | 0000?10112 | ‐‐‐‐‐‐‐‐‐‐ |
| Pelecanoides urinatrix | 0021200012 | 0021100000 | 101000‐0‐0 | ?????????? | ‐‐‐‐‐‐‐‐‐‐ |
| Eudyptula minor | 0011200020 | 1210010110 | 1000011120 | 0010221100 | 0101002020 |
| Aptenodytes forsteri | 1110200020 | 120012 A 011 | 1100010121 | 1111211000 | 0000000000 |
| Aptenodytes patagonicus | 1110200020 | 1200120011 | 1100010121 | 1111211000 | 0000111000 |
| Pygoscelis antarctica | 0010100020 | A200030011 | 2010110110 | 1011212100 | 1000111010 |
| Pygoscelis papua | 1011100020 | A20002A011 | 1000110110 | 1011212100 | 1001101000 |
| Pygoscelis adeliae | 1010100020 | 1200030011 | 2001110110 | 1011211100 | 0000101020 |
| Megadyptes antipodes | 0011200020 | 1200021010 | 2010111120 | 1110211100 | 0101010020 |
| Eudyptes c. moseleyi | 0011200020 | 1200031020 | 2001111120 | 1010211100 | 0000111010 |
| Eudyptes c. chrysocome | 0011200020 | 1200031020 | B001111120 | 1010211100 | 0000111010 |
| Eudyptes chrysolophus | 0011200020 | 1200031020 | 2011111120 | 0010211100 | 0000111010 |
| Eudyptes schlegeli | 0011200020 | 1200031020 | 2011111120 | 1010211100 | 0000111010 |
| Eudyptes pachyrrhynchus | 0011200020 | 1200031020 | 2001111120 | 2010211100 | 0000111020 |
| Eudyptes robustus | 0011200020 | 1210031020 | 2001111120 | 0010211100 | 0000211020 |
| Eudyptes sclateri | 0011200020 | 1200031020 | 2011111120 | ????211100 | 0000211020 |
| Spheniscus demersus | 0011001120 | 1210010120 | 0000112120 | 1110221100 | 0000111110 |
| Spheniscus humboldti | 0011001120 | 1210010120 | 0000112120 | 1110221100 | 0010111030 |
| Spheniscus magellanicus | 0011001120 | 1210010120 | 0000112120 | 1110221100 | 0010011111 |
| Spheniscus mendiculus | 0011001120 | 1210010120 | 0000112120 | 1110221100 | 0000111031 |
| Spheniscus urbinai | ??????11?? | ???????1?0 | ?0?0?????? | ?????????? | ?????????? |
| Spheniscus megarrhamphus | ??????112? | ?????????0 | 2000?1112? | ?????????? | ?????????? |
| Waimanu manneringi | ?????????? | ?????????? | ?????????? | ????1????? | ?????????? |
| Waimanu tuatahi | ????20???? | ?????A???? | ???0?????? | ????1????0 | ?????????? |
| Paraptenodytes antarcticus | ?0??????20 | 10??????2? | 2?0??00101 | ?????????? | ?????????? |
| Paleospheniscus patagonicus | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Eretiscus tonnii | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Marplesornis novaezealandiae | ????????20 | ?110??0??? | 2??????10? | ???????1?? | ?????????? |
| Anthropornis nordenskjoeld | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Anthropornis grandis | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Pachydyptes ponderosus | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Platydyptes novaezealandiae | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Archaeospheniscus lowei | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| “Palaeeudyptes antarcticus” | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Palaeeudyptes gunnari | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Palaeeudyptes klekowskii | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Marambiornis exilis | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Mesetaornis polaris | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Delphinornis larseni | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Arthrodytes andrewsi | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Taxa | 1 0 0123456789 | 1 1 0123456789 | 1 2 0123456789 | 1 3 0123456789 | 1 4 0123456789 |
|---|---|---|---|---|---|
| Gavia stellata | 0110000002 | 0000110111 | 0210100000 | 10000‐0010 | ‐20‐3?1002 |
| Diomedea melanophrys | 0000300102 | ‐000100000 | 0210200000 | 00000‐0020 | ‐20‐3?1012 |
| Phoebetria palpebrata | 0000300102 | ‐000?????? | ?????????? | 0????‐???? | ????3????? |
| Macronectes giganteus | 0330C00102 | ‐00011?011 | 0210?00000 | 110??‐?020 | ‐20‐3?1012 |
| Daption capense | 0000001102 | ‐000110011 | 0210200100 | 1000?‐0020 | ‐20‐3‐1010 |
| Procellaria aequinoctialis | 0110001101 | ‐000110011 | 0210200100 | 1000?‐0020 | ‐20‐3‐1010 |
| Puffinus griseus | 0000101101 | ‐00011?011 | 0210?00??? | 000??‐??20 | ??0?3?101? |
| Pterodroma incerta | 0??0101101 | ‐00001?011 | ?????????? | ?????‐???? | ????3????? |
| Oceanodroma leucorrhoa | 0000001101 | ‐000010000 | 0210100100 | 100??‐0020 | ‐20‐3?1010 |
| Oceanites oceanicus | 0110001101 | ‐000010000 | 0210?00100 | 100??‐0020 | ‐20‐3?1010 |
| Pachyptila desolata | 0A00401101 | ‐000110011 | 0210200100 | 1000?‐0020 | ‐20‐3?101? |
| Pelecanoides urinatrix | 0000401101 | ‐00001?011 | ?????????? | ???1?‐???? | ????3????? |
| Eudyptula minor | 0121301001 | 0101011002 | 1101111112 | 2112101001 | 1121200100 |
| Aptenodytes forsteri | 1000000110 | ‐211011001 | 1101111112 | 2112101001 | 0121200100 |
| Aptenodytes patagonicus | 1110000110 | ‐211011001 | 11011111A2 | 2112101001 | 0121200100 |
| Pygoscelis antarctica | 0000100002 | 0221011101 | 1101111102 | 2112101001 | 0121200102 |
| Pygoscelis papua | 0020200002 | 22210 A 1101 | 1101111112 | 2112101001 | 0121200102 |
| Pygoscelis adeliae | 0010110002 | 2221011101 | 1101111102 | 2112101001 | 0121200101 |
| Megadyptes antipodes | 0110101002 | 0001001102 | 1001111112 | 2112101001 | 0121200100 |
| Eudyptes c. moseleyi | 0220111002 | 1201001102 | 1001111112 | 2112101001 | 0121200100 |
| Eudyptes c. chrysocome | 0220111002 | 1201001102 | 1001111112 | 2112101001 | 0121200100 |
| Eudyptes chrysolophus | 0220111002 | 1201001102 | 1001111112 | 2112101001 | 0121200100 |
| Eudyptes schlegeli | 0220111002 | 1201001102 | 1001111112 | 2112101001 | 0121200100 |
| Eudyptes pachyrrhynchus | 0220111002 | 1201001A02 | 1A01111112 | 2112101001 | 0121200100 |
| Eudyptes robustus | 0220111002 | 1201001102 | 1001111112 | 2112101001 | 0121200100 |
| Eudyptes sclateri | 0220111002 | 1201001102 | 1001111112 | 2112101001 | 0121200100 |
| Spheniscus demersus | 0221001001 | 0101011012 | 1001111112 | 2112111001 | A 121200100 |
| Spheniscus humboldti | 0?21001001 | 0?01011012 | 1001111112 | 2112111001 | 1121200100 |
| Spheniscus magellanicus | 0?21001001 | 0201011012 | 1001111112 | 2112111001 | 1121200100 |
| Spheniscus mendiculus | 0?2?001001 | 0001011012 | 1001111112 | 2112111001 | 1121200100 |
| Spheniscus urbinai | ?????????? | ??????1012 | ?001111??2 | 2112101001 | 11???????0 |
| Spheniscus megarrhamphus | ?????????? | ??????012? | ?????????? | ?????????? | ?????????? |
| Waimanu manneringi | ?????????? | ?????????? | ?????????? | ?????????? | ?????????0 |
| Waimanu tuatahi | ?????????? | ????????12 | 0??011???1 | 2??1??01?0 | ??0?1????? |
| Paraptenodytes antarcticus | ?????????? | ????100012 | ??0??11001 | 2012001001 | 0111?????? |
| Palaeospheniscus patagonicus | ?????????? | ?????????? | ?????11102 | 2112?01001 | 011120???? |
| Eretiscus tonnii | ?????????? | ?????????? | ?????111?2 | 21121?1001 | ?1???????? |
| Marplesornis novaezealandiae | ?????????? | ????0?0?12 | 1??1111??? | 21????1001 | ?12??????? |
| Anthropornis nordenskjoeldi | ?????????? | ?????????? | ?000011001 | 20120?010? | ?01010???? |
| Anthropornis grandis | ?????????? | ?????????? | ?????11??1 | 2?1??????? | ?????????? |
| Pachydyptes ponderosus | ?????????? | ?????????? | ??00011001 | 20110?0001 | ?010?????? |
| Platydyptes novaezealandiae | ?????????? | ?????????? | ?????11102 | 2?1???1001 | ?11101???? |
| Archaeospheniscus lowei | ?????????? | ?????????? | ??00011001 | 20120?0101 | ?1?10????0 |
| “Palaeeudyptes antarcticus” | ?????????? | ?????????? | ???0011001 | 20110?0001 | ?0101????? |
| Palaeeudyptes gunnari | ?????????? | ?????????? | ?????110?1 | 2011??0101 | ?01??????? |
| Palaeeudyptes klekowskii | ?????????? | ?????????? | ?????11??? | 20110?0101 | ?0??1????? |
| Marambiornis exilis | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Mesetaornis polaris | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Delphinornis larseni | ?????????? | ?????????? | ?????????? | ?????????? | ?????????? |
| Arthrodytes andrewsi | ?????????? | ?????????? | ???00110?1 | 2012??1001 | ?1???????? |
| Taxa | 1 5 0123456789 | 1 6 0123456789 | 1 7 0123456789 | 1 8 0 |
|---|---|---|---|---|
| Gavia stellata | 0100‐00010 | 01100?021? | 1?01?????? | ? |
| Diomedea melanophrys | ?102‐00010 | 01100????? | ?????????? | ? |
| Phoebetria palpebrata | ?????????? | ?????????? | ?????????? | ? |
| Macronectes giganteus | 0102‐000?0 | 0???0????? | ?????????? | ? |
| Daption capense | 0102‐00010 | 01100????? | ?????????? | ? |
| Procellaria aequinoctialis | 0102‐00010 | 01100????? | ?????????? | ? |
| Puffinus griseus | ?102‐0???? | ????0????? | ?????????? | ? |
| Pterodroma incerta | ?????????? | ?????????? | ?????????? | ? |
| Oceanodroma leucorrhoa | 0102‐00010 | 01?00????? | ?????????? | ? |
| Oceanites oceanicus | 0102‐00010 | 01100????? | ?????????? | ? |
| Pachyptila desolata | 0102‐000?0 | 0??00????? | ?????????? | ? |
| Pelecanoides urinatrix | ?????????? | ?????????? | ?????????? | ? |
| Eudyptula minor | 0111112002 | 1112100210 | 0100100000 | 0 |
| Aptenodytes forsteri | 0011012002 | 1112111011 | 0110120011 | ? |
| Aptenodytes patagonicus | 0011012002 | 1112111011 | 0110120011 | 2 |
| Pygoscelis antarctica | 0011212001 | 1012111210 | 0211100111 | ? |
| Pygoscelis papua | 001111200B | 1012111211 | 0211100111 | 2 |
| Pygoscelis adeliae | 0011212001 | 1012111210 | 0211100111 | ? |
| Megadyptes antipodes | 1011112002 | 1112100101 | 0211001110 | ? |
| Eudyptes c. moseleyi | 1011112002 | 11121????? | ?????????? | 1 |
| Eudyptes c. chrysocome | 1011112002 | 1112100111 | 0000011100 | 1 |
| Eudyptes chrysolophus | 1011112002 | 1112100111 | 0000011100 | 1 |
| Eudyptes schlegeli | 1011112002 | 1112100111 | 0000011100 | ? |
| Eudyptes pachyrrhynchus | 1011112002 | 111210011A | 0000011100 | ? |
| Eudyptes robustus | 1011112002 | 11121????? | ?????????? | ? |
| Eudyptes sclateri | 101111200? | 11121????? | ?????????? | ? |
| Spheniscus demersus | 0111112002 | 1112100101 | 1011031100 | 0 |
| Spheniscus humboldti | 0111112002 | 1112100101 | 1011031100 | ? |
| Spheniscus magellanicus | 0111112002 | 1112100101 | 1011031100 | 0 |
| Spheniscus mendiculus | 0111112002 | 1112100101 | 1011031100 | 0 |
| Spheniscus urbinai | 01????2002 | 11121????? | ?????????? | ? |
| Spheniscus megarrhamphus | ?????????? | ?????????? | ?????????? | ? |
| Waimanu manneringi | ?1???110?0 | 0?101????? | ?????????? | ? |
| Waimanu tuatahi | ??????10?0 | 0?101????? | ?????????? | ? |
| Paraptenodytes antarcticus | ????2?20?0 | ?0121????? | ?????????? | ? |
| Palaeospheniscus patagonicus | ?????12001 | 1‐221????? | ?????????? | ? |
| Eretiscus tonnii | ??????1001 | 1‐2?1????? | ?????????? | ? |
| Marplesornis novaezealandiae | ?????????? | ?????????? | ?????????? | ? |
| Anthropornis nordenskjoeldi | ??????11?0 | 1002?????? | ?????????? | ? |
| Anthropornis grandis | ??????11?0 | 10021????? | ?????????? | ? |
| Pachydyptes ponderosus | ?????????? | ?????????? | ?????????? | ? |
| Platydyptes novaezealandiae | ?????????? | ?????????? | ?????????? | ? |
| Archaeospheniscus lowei | ?1??2????? | ?????????? | ?????????? | ? |
| “Palaeeudyptes antarcticus” | ??????2010 | 1?221????? | ?????????? | ? |
| Palaeeudyptes gunnari | ??????20?0 | 10221????? | ?????????? | ? |
| Palaeeudyptes klekowskii | ??????2010 | 10221????? | ?????????? | ? |
| Marambiornis exilis | ??????1010 | 10111????? | ?????????? | ? |
| Mesetaornis polaris | ??????1010 | 10111????? | ?????????? | ? |
| Delphinornis larseni | ??????1000 | 10101????? | ?????????? | ? |
| Arthrodytes andrewsi | ?????????? | ?????????? | ?????????? | ? |
Appendix 4: Accession numbers and authorship of DNA sequences
(1) GenBank accession numbers and authorship of sequences used
12S rDNA: Baker et al. (2006): DQ137187, DQ1371901, DQ1371934, DQ137196202, DQ137205; Cooper and Penny (1997): U88007, U88024; García‐Moreno et al. (unpublished): AY139621, AY139623, AY139630; Paterson et al. (1995): X8251718, X82522‐3, X82533; Slack et al. (2003): AY158677; NC_004538; van Tuinen et al. (2000): AF173573, AF1735778.
16S rDNA: Baker et al. (2006): DQ13714762, DQ13714765‐6; Slack et al. (2006): AY158677, AY293618.
cytochrome b: Baker et al. (2006): DQ137207‐10, DQ13721520, DQ137235; Nunn et al. (1996): U48943, U48955; Nunn and Stanley (1998): AF0760512, AF076046, AF076060, AF076062, AF076064, AF076068, AF076076, AF07608990, U74335, U74350, U74353; Slack et al. (2003): NC_004538; Stanley and Harrison (1999): AF158250.
COI: Baker et al. (2006): DQ13716772, DQ13717486; Hebert et al. (2004): AY666477, AY666284; Slack et al. (2006): NC_007172.
RAG‐1: Baker et al. (2006): DQ1372303, DQ13723547.
(2) GenBank accession numbers by species and locus
Taxa are listed alphabetically.
Aptenodytes forsteri: 12S: DQ137187; 16S: DQ137147; COI: DQ137185; cyt‐b: DQ137225; RAG‐1: DQ137246; Aptenodytes patagonicus: 12S: AY139221; 16S: DQ137148; COI: DQ137186; cyt‐b: AY139623; RAG‐1: DQ137247; Daption capense: 12S: X82517; cyt‐b: AF076046; Eudyptes chrysocome: 12S: AY139630; 16S: DQ137155; COI: DQ137172; cyt‐b: AF076051; RAG‐1: DQ137233; Eudyptes chrysolophus: 12S: DQ137197; 16S: DQ137157; COI: DQ137171; cyt‐b: AF076052; RAG‐1: DQ137232; Eudyptes pachyrrhynchus: 12S: U88007, X82522; 16S: DQ137152; COI: DQ137170; cyt‐b: DQ137210; RAG‐1: DQ137231; Eudyptes robustus: 12S: DQ137193; 16S: DQ137153; COI: DQ137176; cyt‐b: DQ137216; RAG‐1: DQ137237; Eudyptes schlegeli: 12S: DQ137196; 16S: DQ137156; COI: DQ137175; cyt‐b: DQ137215; RAG‐1: DQ137236; Eudyptes sclateri: 12S: DQ137194; 16S: DQ137154; COI: DQ137169; cyt‐b: DQ137209; RAG‐1: DQ137230; Eudyptula minor: 12S and cyt‐b: NC_004538; 16S: AF158677; COI: DQ137174; RAG‐1: DQ137235; Diomedea exulans: 12S: DQ137205; 16S: DQ137165; COI: DQ137168; cyt‐b: DQ137208; RAG‐1: DQ137229; Diomedea melanophrys: 12S: AY158677; 16S: AY158677; COI: NC_007172; cyt‐b: U48955; Gavia immer: 12S: AF173577; 16S: DQ137166; COI: DQ137167; cyt‐b: DQ137207; RAG‐1: DQ137228; Gavia stellata: 12S: AF173578; 16S: AY293618; COI: AY666477; cyt‐b: AF158250; Macronectes gigantea: 12S: X82523; cyt‐b: AF076060; Megadyptes antipodes: 12S: DQ137198; 16S: DQ137158; COI: DQ137184; cyt‐b: DQ137224; RAG‐1: DQ1372245; Oceanites oceanicus: cyt‐b: AF076062; Oceanodroma leucorrhoa: COI: AY666284; cyt‐b: AF076064; Pachyptila desolata: cyt‐b: AF076068; Pelecanoides urinatrix: 12S: X82518; cyt‐b: AF076076; Phoebetria palpebrata: cyt‐b: U48943; Procellaria aequinoctialis: cyt‐b: U74350; Pterodroma incerta: cyt‐b: U74335; Puffinus griseus: 12S: X82533, U88024; cyt‐b: U74353; Pygoscelis adeliae: 12S: AF173573; 16S DQ137149; COI: DQ137183; cyt‐b: DQ137223; RAG‐1: DQ137244; Pygoscelis antarctica: 12S: DQ137190; 16S: DQ137150; COI: DQ137181; cyt‐b: AF076089; RAG‐1: DQ137242; Pygoscelis papua: 12S: DQ137191; 16S: DQ137151; COI: DQ137182; cyt‐b: AF076090; RAG‐1: DQ137243; Spheniscus demersus: 12S: DQ137199; 16: DQ137159; COI: DQ137177; cyt‐b: DQ137217; RAG‐1: DQ137238; Spheniscus humboldti: 12S: DQ137201; 16S: DQ137161; COI: DQ137180; cyt‐b: DQ137220; RAG‐1: DQ137241; Spheniscus magellanicus: 12S: DQ137200; 16S: DQ137160; COI: DQ137178; cyt‐b: DQ137218; RAG‐1: DQ137239; Spheniscus mendiculus: 12S: DQ137202; 16S: DQ137162; COI: DQ137179; cyt‐b: DQ137219; RAG‐1: DQ137240.




