The Basal Penguin (Aves: Sphenisciformes) Perudyptes devriesi and a Phylogenetic Evaluation of the Penguin Fossil Record

2010·10.8 MB·English

by  Daniel T. Ksepka

THE BASAL PENGUIN (AVES: SPHENISCIFORMES)
PERUDYPTES DEVRIESI AND A PHYLOGENETIC
EVALUATION OF THE PENGUIN FOSSIL RECORD
DANIEL T. KSEPKA
Department of Marine, Earth, and Atmospheric Sciences
North Carolina State University, Raleigh, NC 27695
([email protected])
JULIA A. CLARKE
Division of Paleontology
American Museum of Natural History, and
Department of Geological Sciences
Jackson School of Geosciences
The University of Texas at Austin
1 University Station C1100 Austin, TX 78713
([email protected])
BULLETINOFTHEAMERICANMUSEUMOFNATURALHISTORY
Number337, 77pp., 30figures, 2tables
Issued June 3,2010
CopyrightEAmericanMuseumofNaturalHistory2010 ISSN0003-0090
CONTENTS
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Geological Context of the Perudyptes Type Locality . . . . . . . . . . . . . . . . . . . . . . . . . 5
Taxonomic Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Institutional Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Systematic Paleontology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Description and Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Skull . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Humerus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Carpometacarpus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Pelvis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Femur. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Tibiotarsus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Tarsometatarsus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Phylogenetic Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Primary Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Outgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Ingroup Taxonomic Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Treatment of Problematic Fossils. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Molecular Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Search Strategy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Additional Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Primary Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Identifying Extinct Parts of the Crown Penguin Radiation . . . . . . . . . . . . . . . . . . . . 26
Evaluating Potential ‘‘Direct Ancestors’’ of Spheniscidae . . . . . . . . . . . . . . . . . . . . . 30
A New Consideration of Fragmentary Fossils. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Taxonomic Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Recommended Calibration Points for Divergence Estimation . . . . . . . . . . . . . . . . . . 42
Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
ABSTRACT
We present the first detailed description of Perudyptes devriesi, a basal penguin from the
middleEocene(,42Ma)ParacasFormationofPeru,andanewanalysisofallpublishedextinct
penguin species as well as controversial fragmentary specimens. The Perudyptes devriesi
holotypeincludeskeyregionsoftheskullandsignificantpostcranialmaterial,thushelpingtofill
a major phylogenetic and stratigraphic (,20 million year) gap between the earliest fossil
penguins(WaimanumanneringiandWaimanutuatahi,,58–61.6Ma)andthenextoldestpartial
skeletons. Perudyptes devriesi is diagnosable by five autapomorphies: (1) an anteroventrally
directed postorbital process, (2) marked anterior expansion of the parasphenoid rostrum, (3)
posterior trochlear ridge of the humerus projecting distal to the middle trochlear ridge and
conformedasalarge,broadlycurvedsurface,(4)convexarticularsurfacefortheantitrochanter
ofthefemur,and(5)extremelyweakanteriorprojectionofthelateralcondyleofthetibiotarsus.
TheskullofPerudyptesischaracterizedbydeeptemporalfossaeandanelongate,narrowbeak
that differs from other reported stem penguins in its short mandibular symphysis. The wing
skeletonof Perudyptes preservesa combination of plesiomorphic featuresalso observed in the
basalpenguinWaimanuandderivedfeaturessharedwithmorecrownwardpenguins.Features
of the wing optimized as primitive for Sphenisciformes include retention of a discrete dorsal
supracondylar tubercle onthe humerus and presence of a modestly projected pisiformprocess
on the carpometacarpus. Derived features present in Perudyptes and all more crownward
penguins,butabsentinWaimanu,includeamoreflattenedhumerus,developmentofatrochlea
forthetendonofm.scapulotricepsatthedistalendofthehumerus,andbowingoftheanterior
face ofthe carpometacarpus.
A combined molecular and morphological dataset for Spheniciformes was expanded by
adding25osteologicalandsofttissuecharactersaswellas11taxa.Inagreementwithprevious
results, Perudyptes devriesi is identified as one of the most basal members of Sphenisciformes.
This analysis also confirms the placement of the middle/late Miocene (,11–13 Ma) fossil
SpheniscusmuizoniasamemberoftheSpheniscuscladeandplacesthelateMiocene(,10Ma)
Madrynornis mirandus as sister taxon to extant Eudyptes. These two species, known from
relatively complete partial skeletons, are the oldest crown clade penguin fossils and represent
well-corroborated temporal calibration points for the Spheniscus-Eudyptula divergence and
Megadyptes-Eudyptes divergence, respectively. Our results reaffirm that the Miocene penguin
taxon Palaeospheniscus, recently proposed to represent a member of the crown radiation,
belongsoutsideof thecrownclade Spheniscidae.
The phylogenetic positions of small Eocene Antarctic penguin taxa (Delphinornis,
Marambiornis, and Mesetaornis) recently proposed as possible direct ancestors to crown
Spheniscidaewerefurtherevaluatedusingalternatecodingstrategiesforincorporatingscorings
from isolated elements that preserve critical morphologies and are thought to represent these
taxa, although they cannot yet be reliably assigned to individual species. Under all scoring
regimes, Delphinornis, Marambiornis, and Mesetaornis were recovered as distantly related to
Spheniscidae.
Using synapomorphies identified in the primary analysis, we evaluated the phylogenetic
positionoffragmentaryspecimens,includingtheholotypesofvalidbutpoorlyknownspecies,
specimens currently unassignable to the species level, and morphologically distinct specimens
that have not yet been named. All pre-Miocene specimens can be excluded from Spheniscidae
based on presence of plesiomorphies lost in all crown penguins, consistent with a recent
radiationforthepenguincrownclade.Thisstudyprovidesadditionalsupportforascenarioof
penguinevolutioncharacterizedbyanoriginofflightlessnessneartheK-Tboundary,dispersal
throughouttheSouthernHemisphereduringtheearlyPaleogene,andalateCenozoicoriginfor
thecrowncladeSpheniscidae.Stratigraphicdistributionandphylogeneticrelationshipsoffossil
penguins are consistent with distinct radiations during the Eocene, Oligocene, and Miocene.
WhiletheEoceneandOligocenepenguinfaunasaresimilarinmanyrespects,theMiocenefauna
is characterized by smaller average size and novel cranial morphologies, suggesting that an
ecologicalshiftindietoccurred closeto the origin of crownSpheniscidae.
3
4 BULLETIN AMERICAN MUSEUM OFNATURALHISTORY NO.337
INTRODUCTION holotype of Perudyptes devriesi, by contrast,
includes a well-preserved skull associated
Sphenisciformes (penguins) are a charis- with significant postcranial material, provid-
matic group of flightless aquatic birds, ing new insight into this poorly understood
distributed broadly throughout the Southern interval of penguin evolution. In this paper,
Hemisphere. Fossil material is abundant for we present a detailed anatomical description
the clade, owing largely to the marine habits of this species and a comprehensive evalua-
and dense bone structure of these birds. tion of the penguin fossil record.
Penguins first appear in the fossil record in Seventy-fivefossilpenguinspecieshavebeen
theearlyPaleocene(,60.5–61.6Ma)ofNew namedovertime,butmanyofthesewerelater
Zealand (Slack et al., 2006)and expandtheir found to be nomina dubia or synonyms of
range to nearly the entirety of their present- previously named species (Simpson, 1946,
day geographical distribution by the late 1971a, 1972a; Jenkins, 1985; Myrcha et al.,
Eocene. Fossils indicate that stem penguins 2002; Acosta Hospitaleche, 2004; Jadwiszc-
reached Antarctica by the late Paleocene zak, 2006b; Ando, 2007). Fifty-two fossil
(Tambussi et al., 2005), South America by species are provisionally considered valid in
themiddleEocene(Clarkeetal.,2003,2007), thispaper,althoughsomeoftheseawaitmore
and Australia by the late Eocene (Simpson, detailed reevaluation (see discussion in
1957; Jenkins, 1974). Peruvian fossils reveal Ksepka, 2007). Additionally, two taxa have
that these Paleogene dispersals included been erected based on subfossil remains
multiple incursions into low latitude waters interpreted as belonging to recently extinct
by the late Eocene (Clarke et al., 2007). The penguin species. Megadyptes waitaha repre-
middleEocene(,42Ma)Perudyptesdevriesi, sents the extinct sister taxon of the extant
collected from rocks deposited at ,14uS Megadyptes antipodes (Yellow-eyedPenguin)
paleolatitude, is the oldest penguin known based on morphological features and ancient
from equatorial waters (Clarke et al., 2007). molecular data recovered from subfossil
Prior to the discovery of Perudyptes bones from New Zealand (Boessenkool et
devriesi, penguins were poorly represented al., 2009). Tasidyptes hunteri was also origi-
in the fossil record for the ,20-million-year nally identified as a recently extinct species
interval between the age of Waimanu tuatahi basedonremainsrecoveredfromamiddenin
from the Paleocene (Slack et al., 2006) and Tasmania (van Tets and O’Conner, 1983),
the age of the next oldest partial skeletons although as discussed below at least some of
from the late Eocene (Marples, 1952; For- thesebonesmayrepresentextantEudyptessp.
dyce and Jones, 1990). Penguin specimens The true total of extinct species is certainly
knownfromthisearlyPaleocene–lateEocene higherthanthoseformallyrecognizedhere,as
intervalwerelimitedtofragmentaryremains. several fossils representing unique species
SuchmaterialincludestheholotypeofCross- have been mentioned in the literature (For-
valiaunienwillia,comprisingthreeincomplete dyceandJones,1990)butarenotyetformally
bones from the late Paleocene of Seymour described and named (Ando, 2007). A con-
Island, Antarctica (Tambussi et al., 2005), a servative estimate places the total number of
singlepartialfemurfromthe?middleEocene pre-Quaternary fossil penguin specimens in
ofNewZealand(Marples,1952;seeSimpson scientific repositories at well over 5000
[1972] regarding stratigraphic uncertainty), a (Ksepka, personal obs.), ranging from artic-
partial hindlimb and pelvis from the middle ulated skeletons to isolated bones. Subfossil
Eocene of South America (Clarke et al., remains are even more abundant, with many
2003), and ,70 almost exclusively isolated thousandspecimenscollectedfromHolocene
postcranial elements and a single beak beachdeposits,abandonedbreedingcolonies,
fragment from the early-middle Eocene and middens throughout the world (e.g.,
deposits of Seymour Island, Antarctica (Jad- McEvey and Vestjens, 1973; van Tets and
wiszczak, 2006a, 2006b). The limited nature O’Conner, 1983; Worthy, 1997; Lambert et
of this material has obscured understanding al., 2002; Worthy and Grant-Mackie, 2003;
ofthemorphologicalchangesthattookplace EmslieandWoehler,2005;Emslieetal.,2007;
in the early evolution of penguins. The Emslie and Patterson, 2007). Although the
2010 KSEPKAANDCLARKE:FOSSIL RECORDOFPENGUINS 5
Fig. 1. Mapshowing the type locality of Perudyptes devriesi (star) and the extentof the PiscoBasin,
withsimplifiedstratigraphiccolumnreflectingourcurrentunderstandingofstratigraphywithinthePisco
Basin(after Devries etal., 2006).
fossilrecordofpenguinsisrich,manyspecies (1990) and DeVries (1998). Marine deposits
are known with certainty from only a single were laid down over the course of the
element. These incomplete specimens pose a Cenozoic during six major transgressions
challenge to phylogenetic analysis (discussed (DeVries, 1998) and preserve a wealth of
by Fordyce and Jones, 1990) and have fossil marinevertebrates (e.g., Muizon, 1981,
hindered the incorporation of biogeographi- 1984, 1988, 1993; Muizon et al., 2003, 2004;
cal, temporal, and morphological data from Stucchi, 2002, 2003, 2007; Stucchi and Ur-
such taxa into the emerging synthesis of bina,2004;StucchiandEmslie,2005;Stucchi
penguin evolution. Herein, we undertake an et al., 2003; Clarke et al., 2007; Esperante et
evaluation of the phylogenetic position of all al., 2008;Ksepkaetal.,2008;Lambertetal.,
fossil penguin taxa and multiple sets of 2008)andinvertebrates(e.g.,Rivera,1957;de
remains recognized as representing as yet Muizon and DeVries, 1985; DeVries, 2007).
unnamedtaxa,withthegoalofimprovingour Perudyptes devriesi is known from a single
understanding of the tempo of penguin specimen collected from the middle Eocene
evolution and contributing to the debate Paracas Formation (Choros Formation of
regardingthetimingoftheradiationofcrown Dunbaretal.,1990).TheParacasFormation
Spheniscidae (Bertelli and Giannini, 2005; directly overlies crystalline basement rocks
Baker et al., 2006; Ksepka et al., 2006; andisseparatedbyanunconformityfromthe
Jadwiszczak,2006b; Clarkeet al., 2007). overlying middle-upper Eocene Otuma For-
mation(DeVrieset.al.,2006;DeVries,2007).
GEOLOGICAL CONTEXTOFTHE PERUDYPTES The Paracas Formation comprises a basal
transgressive sandstone member representing
TYPE LOCALITY
acomparativelynearshorepaleoenvironment
Cenozoic marine sediments are exposed and grading upward into a fine-grained,
throughout the Pisco Basin of coastal south- tuffaceous,anddiatomaceoussiltysandstone
ernPeru(fig. 1).ThestratigraphyofthePisco member representing a more distal shelf
BasinwasdescribedindetailbyDunbaretal. paleoenvironment (DeVries, 2007). The Per-
6 BULLETIN AMERICAN MUSEUM OFNATURALHISTORY NO.337
udyptes devriesi holotype (MUSM 889) was collected(MUSM891,892,894),butbecause
collected from the southern wall of the Que- of incomplete preservation it remains uncer-
brada Perdida locality (14u349S, 75u529W) in tainwhetherallspecimensbelongtothesame
an orange, coarse-grained, thick-bedded, sili- taxon or multiple species are represented.
ciclastic sandstone bed that has been identi- However, these materials indicate that a
fied as part of the basal member of the diversepenguinfaunawaswellestablishedin
ParacasFormation(DeVries,1998).Thebeds Peru by the middle Eocene (Clarke et al.,
containing the Perudyptes devriesi holotype 2007). Only one extant species of penguin
are located a few tens of meters above the (Spheniscushumboldti,PeruvianBlack-footed
ruggedplatformofcrystallinebasementrock. Penguin) occurs regularly along the coast of
Based on the presence of late middle Eocene Peru today, and only this species plus
radiolarians (Cryptocarpium ornatum, Litho- Spheniscus mendiculus (Galapagos Penguin)
cycliaaristotelis,Lithocycliaocellus)higherin occur at latitudes below 20uS (Williams,
the section and the gastropod Turritella 1995). While time averaging must be taken
lagunillasensis in correlative sandstone beds, into account, the presence of at least three
the age of these beds is ,42 Ma (DeVries et distinct taxa in a relatively narrow strati-
al., 2006). graphic interval suggests greater sphenisci-
Littoral invertebrate remains support de- form diversity in Peru, and at low latitudes
position of the basal Paracas sandstone at in general, during the Eocene than in the
QuebradaPerdidaasclosetothepaleoshore- present.
line. During the Eocene, Quebrada Perdida
would have been located at a paleolatitude TAXONOMIC BACKGROUND
nearly equivalent to the 14u349S present-day
latitude, as there has been essentially no Clarke et al. (2003)proposed phylogenetic
latitudinal translation of the area since the definitionsforhighertaxawithinthepenguin
late Jurassic (Jesinkey et al., 1997; Somoza total group. Under these definitions, Pan-
and Tomlinson, 2002; Hartley et al., 2005). sphenisciformes is applied to the clade in-
Cold-waterupwellingalongthewesterncoast cluding all taxa more closely related to
of Peru appears to have been in place by the Spheniscidae than any other extant avian
Late Cretaceous or early Tertiary (Keller et lineage. Sphenisciformes is applied in a more
al., 1997), and this ‘‘proto-Humboldt’’ cur- exclusive sense to the clade including all
rent may have influenced low-latitude pen- Pansphenisciformes that share the apo-
guin diversity by cycling cold, nutrient-rich morphic loss of aerial flight. These taxa
waterintotheecosystem(Clarkeetal.,2007). currently include the same set of known
Sedimentary evidence indicates that land species. However, volant basal members of
adjacent to the coast was continuously arid the penguin lineage (not yet reported in the
or semiarid from the Jurassic to present day fossil record) would be placed within Pan-
(Hartley et al., 2005). sphenisciformes but excluded from Sphenis-
AdditionalfossilsfromtheParacasForma- ciformes.Spheniscidaeisappliedtothecrown
tion currently awaiting description suggest cladeofpenguins,comprisingthemostrecent
that a minimum of three penguin taxa commonancestorofalllivingpenguinspecies
occupying a range of body sizes occurred in and its descendants. We employ this recom-
themiddleEoceneofPeru.Basedonthesizeof mended taxonomy throughout this paper.
preservedlimbbones,theholotypeindividual
ofPerudyptesdevriesiwascomparableinsize INSTITUTIONAL ABBREVIATIONS
to the King Penguin (Aptenodytes patagoni-
cus),thesecondlargestextantpenguinspecies. AMNH, American Museum of Natural
A smaller penguin (,75% the size of Per- History; CADIC, Centro Austral de Investi-
udyptes devriesi) is represented by a partial gaciones Cient´ıficas, Tierra del Fuego, Ar-
tibiotarsus and a partial tarsometatarsus gentina; CM, Canterbury Museum, Christ-
(MUSM 890). Three large specimens signifi- church, New Zealand; IB/P/B, Prof. A.
cantly surpassing comparable elements of Myrcha University Museum of Nature, Uni-
Perudyptes devriesi in size have also been versityofBialystok,Bialystok,Poland;MLP,
2010 KSEPKAANDCLARKE:FOSSIL RECORDOFPENGUINS 7
Museo de La Plata, La Plata, Argentina; Asaltglandfossaembaysthesupraorbital
MNZS, Museum of New Zealand Te Papa margin of the frontals and extends ventrally
Tongarewa, Wellington, New Zealand; along the anterior surface of the postorbital
MUSM,MuseumofSanMarcosUniversity, process to near the tip of the process. In
Lima, Peru; NMV, National Museum of extant penguins and Paraptenodytes antarc-
Victoria, Melbourne, Australia; OM, Otago ticus this fossa varies in development on the
Museum,Dunedin,NewZealand;OU,Otago frontals, but it does not extend along the
University Geology Museum, Dunedin, New anterior face of the postorbital process. The
Zealand; SAM, South African Museum, surface occupied by the salt gland is unusu-
Cape Town, South Africa; SAMA, South ally smooth compared to extant penguins
Australian Museum, Adelaide, Australia; and other fossil taxa, where this surface is
UCMP, University of California Museum of roughanddeeplypitted.Althoughasmooth-
Paleontology,Berkeley,CA;USNM,Nation- er surface is characteristic of juvenile extant
al Museum of Natural History, Smithsonian penguins,MUSM889showsfullfusionofall
Institution,Washington, DC. preserved elements and does not exhibit
unfinished texturing of the limb bones,
suggesting that the holotype individual rep-
SYSTEMATIC PALEONTOLOGY
resents an adult. A lateral shelf of bone
AVESLINNAEUS,1758(SENSUGAUTHIER,1986) bounds the salt gland fossa in extant
Pygoscelis, Megadyptes, and most exemplars
NEOGNATHAEPYCRAFT, 1900
of Eudyptes as well as some outgroup taxa
SPHENISCIFORMESSHARPE, 1891 (SENSU
(Diomedeidae), but this shelf is absent in
CLARKEETAL., 2003)
Perudyptes devriesi and most other stem
PERUDYPTESDEVRIESI CLARKEETAL., 2007
Sphenisciformes (Ksepka and Bertelli [2006]
inferred the presence of a shelf in a single
DESCRIPTION AND COMPARISONS interorbitalfragmentfromthelateEoceneof
Antarctica). The postorbital process of Per-
SKULL: The cranium of the Perudyptes udyptes devriesi is anteriorly deflected, a
devriesi holotype (figs. 2–3) is nearly com- proposed apomorphy of the species. The
plete posterior to the nasal-lacrimal contact, relative width of the frontals between the
although it lacks palatal elements. The orbits is narrow, comparable to the width in
quality of preservation of the skull roof is Waimanu tuatahi (Slack et al., 2006: fig. 1).
high, but crushing has obscured much of the Deep temporal fossae incise the posterior
occipital region. A portion of the premaxilla portion of the skull roof. A clearly defined
was also recovered disarticulated from the sagittal crest is formed where the crests
remainder of the cranium. bounding these fossae meet at midline. The
The preserved portion of the premaxilla sagittal crest meets the nuchal crest at a
shows that the internarial bar is suboval in nearly 90u angle. The edges of the temporal
cross-section, as in Archaeospheniscus lop- fossae are particularly well demarcated ante-
delli. Most extant penguins also present a riorly. A small foramen for the external
similar premaxilla shape in cross-section. ophthalmic artery perforates the squamosal
However, the ventral surface of the inter- in the caudoventral area of the temporal
narialbarinSpheniscusismoreflattenedand fossa. The sharp, anteroposteriorly elongate
possesses slight ventral projections along the sagittal crest of Perudyptes devriesi most
lateral margins, giving the internarial bar an closely resembles that of Paraptenodytes
inverted U-shaped cross-section. The sutures antarcticus. In Waimanu tuatahi, skulls as-
between the left and right premaxillae, as signed to Palaeospheniscus, and extant and
well as those between the premaxillae and extinctspeciesofSpheniscus,thesagittalcrest
nasals, are completely obliterated on the tends to be wider, with greater separation of
dorsal surface in Perudyptes devriesi. Ven- the temporal fossae (Zusi, 1975; Acosta
trally,thepremaxillaeareconvexandsharea Hospitaleche and Canto, 2005; Slack et al.,
sutural contact demarcated by a shallow 2006). Intraspecific variation in the separa-
depression. tion of the temporal fossae occurs within
8 BULLETIN AMERICAN MUSEUM OFNATURALHISTORY NO.337
Fig.2. CraniumofMUSM889in(A)dorsalview,(B)leftlateralview,and(D)ventralview.Fragment
of upperbeak (C)inlateral view. Scale bar 5 1 cm.
2010 KSEPKAANDCLARKE:FOSSIL RECORDOFPENGUINS 9
Fig. 3. Line drawings of the cranium of MUSM 889 in same views as figure2. See appendix 5
for abbreviations.
10 BULLETIN AMERICAN MUSEUM OFNATURALHISTORY NO.337
Fig.4. MandibleofMUSM889:(A)leftmandibularramusinmedialview,(B)leftmandibularramus
inlateralview,(C)rightmandibularramusinmedialview,and(D)rightmandibularramusinlateralview.
Scalebar5 1cm. Seeappendix 5forabbreviations.
extant Spheniscus species (Ksepka and Ber- necessary to confirm this hypothesis. The
telli, 2006), but sample sizes in fossil taxa occipital condyle is wider than high. As in
remain too small to evaluate variation in other penguins, the supraoccipital projects
extinct taxa. caudally to form a hood over the occipital
Aportionofthemesethmoidisintact.This condyle. Dorsoventral compression at the
element attaches to the frontals along the posterior part of the cranium distorts the
midline, but there is no trace of the thin shape of the foramen magnum and obscures
lateral expansion for the olfactory chamber the morphology of surrounding elements.
present in extant penguins. The delicate Themandible(fig. 4)ispreservedfromthe
natureofthemesethmoidleavesitsusceptible symphysis tothecaudaltipofthedentary.It
todamage, butaswewere unableto identify is elongate and narrow, although it does not
a broken edge, the absence of the expansion exhibit the extreme spearlike morphology of
appears to be real. Icadyptes salasi (Clarke et al., 2007; Ksepka
The dorsal tympanic recess is located in a et al., 2008). The symphysis is short, con-
circularfossathatisposteriorlyboundedand trasting with the extensive symphysis in the
distinctfromthebasicranium-quadrateartic- stem sphenisciforms Waimanu tuatahi, Ica-
ulation. The parasphenoid rostrum flares dyptes salasi, and Archaeospheniscus lowei.
laterally, forming a well-developed shelflike The rostral tip is not entirely intact, but the
surface near its anterior end. Small ridges beak is tapering at the most distal preserved
that bound the ventral surface of the para- point, indicating that at most a small
sphenoidrostrumareespeciallyvisibleinthis fragment has been lost. The mandibular
slightly expanded region. We interpret this ramus is very straight with no noticeable
surface as a support for the palatines near deepening at the midpoint. A rostral man-
the palatine-pterygoid articulation, although dibular fenestra is absent. A conspicuous
a skull preserving the complete palate is elongate depression is present on the lingual