Descriptions and phylogenetic relationships of a new genus and two new species of Oligo‐Miocene cormorants (Aves: Phalacrocoracidae) from Australia

Abstract

Tertiary cormorant fossils (Aves: Phalacrocoracidae) from Late Oligocene deposits in Australia are described. They derive from the Late Oligocene – Early Miocene (26–24 Mya) Etadunna and Namba Formations in the Lake Eyre and Lake Frome Basins, South Australia, respectively. A new genus, Nambashag gen. nov., with two new species (Nambashag billerooensis sp. nov., 30 specimens; Nambashag microglaucus sp. nov., 14 specimens), has been established. Phylogenetic analyses based on 113 morphological and two integumentary characters indicated that Nambashag is the sister taxon to the Early Miocene Nectornis miocaenus of Europe and all extant phalacrocoracids. As Nambashag, Nectornis, and extant phalacrocoracids constitute a strongly supported clade sister to Anhinga species, the fossil taxa have been referred to Phalacrocoracidae. Sulids and Fregata were successive sister taxa to the Phalacrocoracoidea, i.e. phalacrocoracids + Anhinga. As phalacrocoracids lived in both Europe and Australia during the Late Oligocene and no older phalacrocoracid taxa are known, the biogeographical origin of cormorants remains unanswered. The phylogenetic relationships of extant taxa were not wholly resolved, but contrary to previous morphological analyses, considerable concordance was found with relationships recovered by recent molecular analyses. Microcarbo is sister to all other extant phalacrocoracids, and all Leucocarbo species form a well-supported clade.© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314.

Full text

Page 1

Descriptions and phylogenetic relationships of a new
genus and two new species of Oligo-Miocene
cormorants (Aves: Phalacrocoracidae) from Australia
TREVOR H. WORTHY*
School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney,
NSW 2052, Australia
Received 3 June 2010; revised 10 August 2010; accepted 10 August 2010
Tertiary cormorant fossils (Aves: Phalacrocoracidae) from Late Oligocene deposits in Australia are described. They
derive from the Late Oligocene – Early Miocene (26–24 Mya) Etadunna and Namba Formations in the Lake Eyre
and Lake Frome Basins, South Australia, respectively. A new genus, Nambashag gen. nov., with two new species
(Nambashag billerooensis sp. nov., 30 specimens; Nambashag microglaucus sp. nov., 14 specimens), has
been established. Phylogenetic analyses based on 113 morphological and two integumentary characters indicated
that Nambashag is the sister taxon to the Early Miocene Nectornis miocaenus of Europe and all extant
phalacrocoracids. As Nambashag, Nectornis, and extant phalacrocoracids constitute a strongly supported clade
sister to Anhinga species, the fossil taxa have been referred to Phalacrocoracidae. Sulids and Fregata were
successive sister taxa to the Phalacrocoracoidea, i.e. phalacrocoracids + Anhinga. As phalacrocoracids lived in both
Europe and Australia during the Late Oligocene and no older phalacrocoracid taxa are known, the biogeographical
origin of cormorants remains unanswered. The phylogenetic relationships of extant taxa were not wholly resolved,
but contrary to previous morphological analyses, considerable concordance was found with relationships recovered
by recent molecular analyses. Microcarbo is sister to all other extant phalacrocoracids, and all Leucocarbo species
form a well-supported clade.zoj_693 277..314
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314.
doi: 10.1111/j.1096-3642.2011.00693.x
ADDITIONAL KEYWORDS: Nambashag – Oligocene – Pelecaniformes – phylogenetic analyses – shags –
taxonomy.
INTRODUCTION
Amongst extant taxa, the Pelecaniformes is tradition-
ally considered to include tropic birds (Phaethon),
frigate birds (Fregata), pelicans (Pelecanus), gannets
and boobies (Morus, Sula, and Papasula), darters
(Anhinga), and the cormorants or shags (Phalacroco-
rax) (e.g. Peters, 1931; Lanham, 1947; Mayr & Cot-
trell, 1979; Cracraft, 1985). The composition of the
order has recently been the subject of much debate;
see for example, the discussion in Kennedy, Gray &
Spencer (2000) and Livezey & Zusi (2007). Genetic
evidence supports Phaethon being separated from
other pelecaniforms, and Pelecanidae is also rela-
tively distant to the remaining pelecaniforms and
mayhave affinity with
Kennedy & Spencer, 2004; Ericson et al., 2006; Brown
et al., 2008; Hackett et al., 2008). This emerging
understanding of relationships led Christidis & Boles
(2008) to transfer Pelecanus to Ciconiiformes and
use Phalacrocoraciformes for Fregatidae, Sulidae,
Anhingidae, and Phalacrocoracidae.
Within traditional Pelecaniformes, using osteo-
logical comparisons, anhingids have long been con-
sidered the sister to phalacrocoracids (e.g. Pycraft,
1898; Lanham, 1947; Owre, 1967; Cracraft, 1985;
some ciconiiforms (e.g.
*E-mail: t.worthy@unsw.edu.au
Zoological Journal of the Linnean Society, 2011, 163, 277–314. With 5 figures
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314
277

Page 2

Siegel-Causey, 1988). Although this relationship is
supported bysome recent
Hackett et al., 2008), some molecular analyses recov-
ered phalacrocoracids as sister to sulids + anhingids
(Fain & Houde, 2004; Kennedy & Spencer, 2004;
Kennedy et al., 2005), so the issue remains unre-
solved. In contradiction to these recent molecular
analyses, a recent morphological analysis of birds
examining their higher order phylogeny recovered
traditional Pelecaniformes, with Phaethon basal and
Anhinga sister to Phalacrocorax (Livezey & Zusi,
2007).Thesevaried results
thesistertaxon ofPhalacrocoracidae,
nearestoutgroupshave
confidence.
Relationships within Phalacrocoracidae are still
being determined. Siegel-Causey (1988) used cladistic
methodology in a morphology-based phylogenetic
analysis of most cormorant species and derived well-
supported clades on the basis of which he advocated
that Phalacrocorax be subdivided into several genera
within a fundamental dichotomy. Siegel-Causey’s
hypothesis was fundamentally different from conven-
tional wisdom, however, as for example, Leucocarbo
as defined by Voisin (1973) appeared markedly para-
phyletic, and overall groupings bore little relationship
to the behavioural and morphologically based group-
ings of van Tets (1976). His apparently robust analy-
sis came under scrutiny when Kennedy et al. (2000)
employed molecular data (mitochondrial DNA) to
examine Siegel-Causey’s hypothesis and found little
support for it. In their most recent contribution,
Kennedy, Valle & Spencer (2009) extended this analy-
sis to include 24 phalacrocoracid species, although the
data were limited to three mitochondrial genes. Of
the species included in their analysis, Microcarbo
melanoleucos was sister to all other phalacrocoracids
and, moving crownward, Phalacrocorax gaimardi was
the next branch. Remaining taxa fell into two well-
supported clades: (1) Phalacrocorax auritus, Phalac-
rocorax brasilianus, and Phalacrocorax harrisi were
sister to all Leucocarbo species as herein defined; and
(2), a relatively poorly resolved group of other taxa,
wherein some subclades had significant support as
follows: Phalacrocorax penicillatus + Phalacrocorax
urile + Phalacrocoraxpelagicus,
aristotelis,Phalacrocorax
capillatus + Phalacrocorax
nigrogularis,Phalacrocorax
corax featherstoni, and Phalacrocorax varius + Phala-
crocorax sulcirostris. However, in the absence of more
comprehensive taxon sampling and corroborating
data from other sources, e.g. nuclear genes, these
authors refrained from advocating any taxonomic rec-
ommendations. Further, these analyses have so far
lacked the necessary mix of anhingids, sulids, and
moleculardata(e.g.
show thatneither
nor
with
its
beenestablished
Phalacrocorax
capensis + Phalacrocorax
carbo,Phalacrocorax
punctatus + Phalacro-
more distant outgroups in their analyses (Kennedy
et al., 2005).
FOSSIL HISTORY OF CORMORANTS
Göhlich & Mourer-Chauviré (2010) summarized the
fossil record of Oligocene and Miocene phalacrocorac-
ids from Europe. With the addition of ?Borvocarbo
tardatus of these authors, 13 species are now recog-
nized from this period in Europe and two from North
America (Table 1, and references therein). Most fossil
cormorants were initially described in Phalacrocorax,
but following Lambrecht (1933), the older species
(Oligocene and Early Miocene) have been accepted as
properly belonging in separate genera (Paicheler
et al., 1978; Cheneval, 1984; Mayr, 2001, 2007, 2009;
Mourer-Chauviré, Berthet & Hugueney, 2004). Most
recently, and influenced by Livezey & Zusi (2007),
Mayr (2007, 2009) accepted the superfamily Phalac-
rocoracoideatoinclude
Anhingidae. He included the Oligocene and Early
Miocene taxa Borvocarbo, ?Borvocarbo stoeffelensis,
Oligocorax, Nectornis, and ?Limicorallus carbunculus
in Phalacrocoracoidea, but outside of crown group
Phalacrocoracidae (i.e. the clade comprising all extant
phalacrocoracids and their most recent common
ancestor), he considered that their assignment to the
stem lineage of Phalacrocoracidae was not supported
by present data. Göhlich & Mourer-Chauviré (2010)
accepted Oligocorax, Nectornis, and Limicorallus as
members of Phalacrocoracidae, but left Borvocarbo in
Phalacrocoracoidea, thereby leaving open the possi-
bility that this fossil taxon might be ancestral to both
anhingas and cormorants.
The relationships of any of these fossils with crown
group anhingids and phalacrocoracids have yet to be
the subject of a phylogenetic analysis. If the European
taxa are cormorants, then they are the earliest fossils
of the group and show that phalacrocoracids arose no
later than the Late Oligocene in Europe. Middle to
Late Miocene taxa are poorly known, but by the Late
Miocene, taxa more derived than the Late Oligocene
– Early Miocene taxa have morphologies that do not
preclude them from being within Phalacrocorax. A
partial premaxilla was referred to Phalacrocoracidae
from the Jebel Qatrani Formation of Early Oligocene
age, Egypt (Rasmussen, Olson & Simons, 1987), but
its fragmentary nature makes this referral tenuous.
Outside of Europe, the group has a Late Miocene
appearance in North America (Table 1). From Austra-
lia, undescribed cormorant fossils were reported from
Miocene lacustrine sediments of South Australia
(Vickers-Rich, 1991) and ‘Phalacrocoracidae’ from the
Pinpa Fauna of the Namba Formation, South Austra-
lia, Late Oligocene – Middle Miocene by Rich et al.
(1991). In addition, unspecified fragmentary fossils
Phalacrocoracidaeand
278
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 3

Table 1. Fossil species of Oligocene and Miocene age attributed to Phalacrocoracidae or Phalacrocoracoidea, sensu
Livezey & Zusi (2007), but which are not Anhingidae (Brodkorb, 1963; Cheneval, 1984; Mlíkovský, 2002; Göhlich &
Mourer-Chauviré, 2010). Limicorallus saiensis Kurochkin, 1968, originally described as a rallid was assigned to
Phalacrocorax by Mlíkovský & Švec (1986), although this was restricted to the Phalacrocoracoidea by Mayr (2009).
Phalacrocorax subvolans Brodkorb, 1956 was assigned to Anhingidae by Becker (1986) so is not listed here
TaxonAge Location Specimen/s
Borvocarbo guilloti
Mourer-Chauviré, Berthet
& Hugueney (2004)
?Borvocarbo stoeffelensis
Mayr (2007)
?Borvocarbo tardatus Göhlich
& Mourer-Chauviré (2010)
Oligocorax littoralis
(Milne-Edwards, 1863)
Late OligoceneFrancecoracoid
Late Oligocene (MP 30)Germany skeleton on a slab
Early Miocene (MN 4b) GermanyL tibiotarsus, pL ulna, pR
radius
Two humeri, pR ulna, L
coracoid, R femur,
part 2 tibiotarsi, pR
tarsometatarsus (Cheneval,
1984)
Distal L humerus
Early Miocene (MN 1–3) France, Germany
& Czechia
Limicorallus saiensis
Kurochkin (1968)
?Limicorallus carbunculus
Mayr (2009)
Nectornis miocaenus
(Milne-Edwards, 1867), in
Milne-Edwards
(1867–1871a)
Nectornis anatolicus
(Mourer-Chauviré, 1978) in
Paicheler et al. (1978)
Late OligoceneKazakhstan
Early Miocene (MN 1)GermanyR tarsometatarsus
Early Miocene (MN 2–4)France, Germany,
and Czechia
Many bones (Cheneval, 1984)
Middle Miocene (MN 6–8)Turkey L coracoid, pL humerus, dL
ulna, dL radius, part &
counterpart of part R wing,
dL humerus, pL radius, pL
ulna on slab
L coracoid, pL humerus, L
carpometacarpus, a
vertebra
Phalacrocorax intermedius
(Milne-Edwards, 1867), in
Milne-Edwards
(1867–1871a)
Phalacrocorax ibericus
Villalta (1963)
Phalacrocorax lautus
Kurochkin & Ganea (1972)
Phalacrocorax serdicensis
Burchak-Abramovic &
Nikolov (1984)
Phalacrocorax longipes
(Tugarinov, 1940)
Early & Middle Miocene (MN
3–5)
France, Germany,
and Czechia
Late Miocene (MN 9)SpaindR humerus
Late Miocene (MN ?9)MoldaviapR femur
Late Miocene (MN 11–13) BulgariapR humerus
Late Miocene (MN 11–13)
and Early Pliocene (MN
15)
Late Miocene
Ukraine R tarsometatarsus, R femur
Phalacrocorax femoralis
Miller (1929)
Phalacrocorax marinavis
Shufeldt (1915)
CaliforniaPt skeleton
Late Miocene OregonPt 2 humeri, pR ulna, dL
tarsometatarsus
L, left, R right; p or d preceding L or R means proximal or distal part of that element.
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
279
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 4

were referred to Phalacrocoracidae from two sites in
Riversleigh Oligocene – Miocene deposits by Boles
(1997, 2006), but these are now identified as anser-
anatid (pers. observ.).
In 2006, the author partook in an expedition to
collect fossils in the Namba Formation in South Aus-
tralia, where a number of cormorant fossils was
found. Given that the Southern Hemisphere has most
extant cormorants and Australasia is their centre of
diversity (van Tets, 1976; Marchant & Higgins, 1990),
and that the Australian fossils are equally as old as
any cormorant fossils in the world, this early record
in Australia is significant for understanding the evo-
lution of the group. In addition, the taxa are repre-
sented by most major skeletal elements, potentially
allowing a phylogenetic assessment of their relation-
ships. Therefore, the aims of the present contribution
are to describe these and other cormorant fossils from
the Oligocene – Early Miocene lacustrine sediments
of South Australia subsequently identified in collec-
tions and to assess the phylogenetic relationships of
the included taxa.
METHODS
ABBREVIATIONS
Institutions
AM, Australian Museum, Sydney, Australia; AMNH,
Fossil Amphibian, Reptile, and Bird Collections, Divi-
sion of Paleontology, American Museum of Natural
History, New York, USA; ANWC, Australian National
Wildlife Collection,CSIRO,
BMNH, The Natural History Museum, Tring, UK;
NMNZ, Museum of New Zealand Te Papa Tongarewa,
Wellington(formerly National
Zealand, Dominion Museum, and Colonial Museum),
NewZealand; KUNHM,
Natural History Museum & Biodiversity Research
Centre, Lawrence, Kansas, USA; MV, Museum Victo-
ria, Melbourne, Victoria, Australia; QM, Queensland
Museum, Brisbane, Queensland, Australia; SAM,
South Australian Museum, Adelaide, South Australia,
Australia; UCR, Department of Earth Sciences, Uni-
versity of California, Riverside, California, USA;
USNM,DivisionofBirds,
History, Smithsonian Institution, Washington D.C.,
USA.
Canberra, Australia;
Museum ofNew
UniversityofKansas
MuseumofNatural
Common abbreviations
Some common terms are abbreviated as follows: L is
left and R is right element; p or d preceding L or R
means proximal or distal part of that element; artic.,
articularis; indet., indeterminate; LF, local fauna; lig.,
ligamentum; M., musculus; Mya, million years ago;
proc., processus; tuber., tuberculum.
Measurements were performed with Tesa dial cal-
lipers and rounded to 0.1 mm: TL, total length; PW,
proximal width; SW, shaft width; DW, distal width.
NOMENCLATURE
Names for specific bone landmarks follow Baumel &
Witmer (1993). Taxonomic nomenclature of modern
species follows Christidis & Boles (2008), Checklist
Committee (O.S.N.Z.), 2010, and Dickinson (2003).
Therefore I use Microcarbo for the small taxa Micro-
carbo pygmaeus, Microcarbo melanoleucos, Micro-
carbo africanus, and their relatives, which are now
demonstrated to be the sister group of other extant
cormorants (e.g. Kennedy & Spencer, 2004; Kennedy
et al., 2009). I also follow the recent checklists (e.g.
Christidis & Boles, 2008) in using Leucocarbo for the
morphologicallydistinctive
species, which are a genetically and morphologically
distinct group, but whose relationships to other cor-
morants are unresolved (Voisin, 1973; Kennedy et al.,
2009). Within this group, traditionally known as the
blue-eyed shags, there are well-defined species groups
(Voisin, 1973) within which several taxa were for-
merly considered subspecies (e.g. Mayr & Cottrell,
1979) as follows: (1), Leucocarbo atriceps assemblage
(includes L. atriceps, Leucocarbo bransfieldensis, Leu-
cocarbo georgianus, Leucocarbo nivalis, and Leuco-
carbo purpurescens, plus Leucocarbo albiventer and
Leucocarbo melanogenis not studied herein); (2), the
Leucocarbo carunculatus group (L. carunculatus, Leu-
cocarbo chalconotus, and Leucocarbo onslowi); (3), the
Leucocarbo campbelli group, although this is often
included with the L. carunculatus group (Leucocarbo
campbelli, Leucocarbo colensoi, and Leucocarbo ran-
furlyi); (4), Leucocarbo bougainvillii; (5), Leucocarbo
magellanicus; and (6), Leucocarbo verrucosus. All
other cormorants are listed under Phalacrocorax
herein.
southernpink-footed
COMPARATIVE MATERIAL
The following modern specimens were examined:
Anhinga
O.65077; Anhinga anhinga, KUNHM 37543, 30244.
Microcarbomelanoleucos,
O.60218, O.60449, O.60450, O.60486, O.64725; Micro-
carbo africanus, ANWC 21997; Microcarbo niger,
ANWC 21999;
Microcarbo
Museum, University of Copenhagen, unregistered
specimen from‘Siam’.
KUNHM 78251, 78253; P. auritus, ANWC 21826,
21828;
Phalacrocoraxbrasilianus
ANWC 21833, 21834; Phalacrocorax nigrogularis,
BMNH S/1973.7.8; Phalacrocorax pelagicus, ANWC
21949; Phalacrocorax urile, ANWC 21953; Phalacro-
novaehollandiae,AM S.291,S.1258,
AMS.374, O.4704,
pygmaeus,Zoological
Phalacrocorax gaimardi,
(=
olivaceus),
280
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 5

corax penicillatus, ANWC 21869; Phalacrocorax aris-
totelis, ANWC 21947; Phalacrocorax neglectus, ANWC
21893; Phalacrocorax featherstoni, ANWC 21945;
Phalacrocorax punctatus, AM O.70873; Phalacrocorax
varius, AM O.56973, O.65570; Phalacrocorax fusce-
scens, AM O.65567, O.65568; Phalacrocorax sulciros-
tris, AM O.60418, O.64602, NMNZ 21115, 21116,
21117, 26914; Phalacrocorax capensis, ANWC 21890;
Phalacrocorax carbo, AM O.65437, O.65569. Leuco-
carbo bougainvillii, USNM 500600; Leucocarbo atri-
ceps atriceps, KUNHM 79848, 81205; Leucocarbo
purpurescens, AM S.1370, ANWC 21983, NMNZ 1524;
Leucocarbo nivalis, ANWC 22006; Leucocarbo verru-
cosus, ANWC 21971; Leucocarbo campbelli, ANWC
21966; Leucocarbo chalconotus, ANWC 22003; Leuco-
carbo onslowi, ANWC 22001, NMNZ 15154, 15496,
15498, 21496, 23767, 24430, 25637, 26358, 26435,
26436, 27191. Phaethon rubricauda, AM O. 70575,
O.71399;
Pelecanusconspicillatus,
71383; Fregata minor, MV B.24450, QM O.27761;
Morus serrator, AM O.65332, O.56723; Sula leuco-
gaster, AM O.57509, QM 28039, QM 29534.
AMO.57540,
Data for Nectornis miocaenus were taken from descrip-
tions and images in Milne-Edwards (1867–71a,b),
Cheneval (1984), Mourer-Chauviré et al. (2004) and
images provided by Dr Didier Berthet (Muséum de
Lyon) and Dr Abel Prieur (Université Claude Bernard
Lyon 1), Lyon, France. Images of the following speci-
mens of N. miocaenus were examined: collections of
Geology, University of Lyon 1, Lyon, France: R ulna,
proximal part, FSL 443.584; R ulna, distal part, FSL
443.588; R carpometacarpus FSL 443.582; distal L
tibiotarsus FSL 443.561; L tarsometatarsus, proximal
part, FSL443.598; R tarsometatarsus, distal part, FSL
443.560. Muséum de Lyon, Lyon, France: StG317 R
femur; StG1463 L femur; StG311 R humerus; StG313
L humerus; StG315 L coracoid.
IDENTIFICATION OF FOSSIL MATERIAL
Fossil collections in the following institutions were
searched for pelecaniforms of Oligocene and Miocene
age from Australia: AM, MV, and SAM. A large col-
lection of fossil avian material housed at AM, includ-
ing materialfrom the
Museum of Paleontology, AMNH, and UCR, which
had been assembled for other purposes, was also
examined. All specimens at these institutions provi-
sionally identified as pelecaniforms were borrowed for
this study.
Universityof California
PHYLOGENETIC ANALYSES
A set of morphological characters was defined and
scored for all taxa in the study. I chose not to use
the morphological character set and database of
Siegel-Causey (1988) because the relationships these
retrieved have very little concordance with those
derived from molecular data (Kennedy et al., 2000,
2009). Examination of the Siegel-Causey (1988)
character set during a detailed comparison with
molecular data (Holland et al., 2010) raised many
problems, including issues relating to extensive non-
independence of characters, e.g. the character pairs or
groups 2–3, 4–5, 12–15, 27–28, 38–39, 41–43, 55–56,
67–68, 74–75, 82–85, 92–93, 98–99, 113–115 are con-
sidered overlapping or part of the same functional
complex. Moreover, many characters had states that
reflected relative degrees of expression of a feature,
e.g. 12, 13, 28, 41–43, 45, 46, 48, 59, 62, 107, 116–121,
127, and 129, that were likely to be unduly affected by
intraspecific variability and homoplasy. Essentially,
as Siegel-Causey’s (1988) data set does not retrieve
the probable real phylogenetic relationships, I deter-
mined to start again and constructed an entirely new,
or independent, character set derived from features
that distinguished taxa, from direct observation of
skeletons based on my experience of the usefulness of
characters in separating problematic avian taxa
(Worthy & Lee, 2008). An initial character set was
constructed using characters that differentiated at
least one species from another amongst the analysed
taxa. I examined pertinent literature, such as Owre
(1967) and Mayr (2001, 2007, 2009), for additional
characters. Care was taken to ensure that for all
characters, the designated states encompassed the
analysed taxa and that the feature was homologous
across the ingroup of cormorants and the outgroup of
various other pelecaniforms. The enlarged set of taxa
studied herein, i.e. representatives of all pelecani-
forms and the fossil phalacrocoracids, required differ-
ent charactersandor
character states used here, compared to Siegel-
Causey (1988), to capture variation. Additionally, as
the primary aim of this work was to assess the
relationships of the fossils, characters were mostly
selected from skeletal elements represented amongst
the fossils. For this reason and as no fossil cranium
was known, only a few cranial characters were
employed, but those that were, had significance in
resolving the relationships of the families considered
and/or of the extant phalacrocoracids studied. In
total, a set of 113 osteological characters was defined,
to which were added two integumentary characters
with potential to help resolve the relationships of
extant taxa (Appendix 1). Of these, 21 were either the
same as or in part those used by Siegel-Causey
(1988). This data set differed from that of Siegel-
Causey (1988) by having far fewer cranial characters
(15 vs. 35) and fewer mandible characters (eight vs.
15), but many more of the tibiotarsus (11 vs. five) and
tarsometatarsus (17 vs. three). One point of difference
differentdefinitionsfor
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
281
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 6

with Siegel-Causey’s (1988) choice of characters is
that I avoided features to do with muscle impressions,
which I consider are prone to intraspecific variation.
Further, I avoided dense sampling of the entire skull,
especially the mandible, because it is known to be a
source ofmuchhomoplasy
Sheldon, 1998). The phylogenetic analyses had the
principal aim of determining the relationships of the
Tertiary Australian taxa. As Oligocene taxa have
already been referred to Phalacrocoracoidea, e.g.
Mayr (2007, 2009), the choice of outgroup and ingroup
taxa was critical to assess the scenarios that this
hypothesis raises. For example, Anhinga cannot be
chosen as the outgroup, as to do so would preclude the
possibility of the fossil having a sister group relation-
ship with the anhingid-phalacrocoracid lineage. For
this reason, I assessed the use of Phaethon, Peleca-
nus, Fregata, and sulids alone or in combination to
determine the best outgroup.
Exploratory analyses to determine the most appro-
priate outgroup gave conflicting results. With all char-
acters unordered and no weighting used, and Phaethon
alone as the sole outgroup, the tree could not be rooted
such that the specified ingroup was monophyletic:
Phaethon, Fregata, Pelecanus, anhingids, sulids, and
phalacrocoracids formed an unresolved polytomy. With
both Phaethon and Pelecanus constituting the out-
group, the same topology was obtained. As much
genetic data suggest that Phaethon is only remotely
related to the remaining pelecaniforms, it was
excluded from further analyses to remove the possibil-
ity that homoplasy was affecting character optimiza-
tion. If Pelecanus alone was used as the outgroup, the
tree again could not be rooted such that the specified
ingroup was monophyletic: Pelecanus, the sulids, and a
clade of all remaining taxa formed a trichotomy. In the
remaining taxa, anhingids were sister to phalacroco-
racids, and Fregata sister to both. When Pelecanus and
Fregata were designated as the outgroup, the same
topology was obtained; however, when the sulids were
added to the outgroup, the ingroup was found to be
monophyletic with anhingids sister to the phalacroco-
racids.As in all analyses, anhingids were always found
to be sister to phalacrocoracids, the outgroup was
hereafter defined as the sulids (Sula leucogaster,
Morus serrator), Pelecanus, and Fregata. This arrange-
ment allowed the testing of the hypothesis that the
fossils were stem taxa of Phalacrocoracidae and
Anhingidae, that is, the superfamily Phalacrocora-
coidea, as was suggested for fossil European taxa
(Mayr, 2007, 2009).
As thereare significant
A. anhinga and A. novaehollandiae, as shown by both
skeletal and genetic data (Harrison, 1978; Kennedy
et al., 2005), both species were sampled to avoid inter-
specific bias affecting placement of the genus.
(e.g.McCracken&
differencesbetween
To construct a robust framework of crown group
taxa with which to compare the fossils, I sampled 27
representatives of 39 generally recognized extant and
one recently extinct phalacrocoracids (Nelson, 2005)
that reflected all major clades in genetic analyses
(Kennedy et al., 2000, 2009). Three of the five Micro-
carbo species (M. pygmaeus, M. africanus, and M.
melanoleucos) were used in the analyses. Examina-
tion of a partial skeleton of M. niger revealed that it
was consistent with other Microcarbo taxa, but was
omitted because of missing data. The Leucocarbo
species were represented by eight species (L. bouga-
invillii, L. atriceps, L. purpurescens, L. nivalis,
L. verrucosus, L. campbelli, L. chalconotus, and
L. onslowi) with the unsampled taxa (L. magellani-
cus, L. bransfieldensis, L. georgianus, L. melanogenis,
L. carunculatus, L. colensoi, L. ranfurlyi) considered
closely related to those that were sampled (Voisin,
1973; Nelson, 2005). Of the remaining cormorants,
the Galapagos cormorant P. harrisi was not sampled,
but this is a flightless insular derivative of the
P. auritus–P. brasilianus lineage (Kennedy et al.,
2009) and is unlikely to be related to the Australian
fossils. The great cormorants were represented by P.
carbo and P. capensis, with the closely related P.
capillatus of Japan unavailable. The only other taxa
not sampled were P. fuscicollis, the Indian cormorant
of tropical Asia, and P. perspicillatus, Pallas’s cormo-
rant, which is extinct and formerly lived in the Bering
Sea. Neither taxon is likely to be related to the fossil
Australian taxa on biogeographical grounds.
In addition to the extant taxa and the Australian
fossil taxa, I added data from N. miocaenus from the
Early Miocene of Europe. Other fossil taxa of compa-
rable age are poorly represented (Table 1), so much
missing data made it impractical to score them in
these analyses.
In all, a total of 34 terminal taxa was used in the
analysis of the fossils. These were scored for 111
osteological characters by examination of specimens
and for two integumental characters from Nelson
(2005). As with most morphological matrices, there
are missing data. In a few instances within the out-
group taxa, some characters could not be objectively
scoredowing toextensive
homology, and so were coded as ‘-’. Missing data,
where the state for characters was not preserved,
were coded as ‘?’. Although these are distinguished in
my matrix (Appendix 2), I treated them both as
missing in my PAUP analyses. A total of 19 multistate
characters (* in Appendix 1) was considered to vary
as morphoclines. Analyses were performed with these
multistate characters treated either as unordered
or ordered. This morphological data set may have
problems with homoplasy caused by convergent mor-
phologies from diving habits (e.g. McCracken et al.,
divergenceobscuring
282
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 7

1999; Callaghan & Harshman, 2005; Worthy & Lee,
2008). Many of the characters found exclusively in
diving anatid taxa (Worthy & Lee, 2008) have corre-
lates in the cormorants, such as lateral expansion of
the fibular condyle on the femur, height of the
cnemial crest on the tibiotarsus, flattening of the
anterior facies of the tibiotarsus, and pelvic charac-
ters (e.g. post-acetabular length and relative width),
that are perhaps most strongly related to function
(Raikow, 1970). Although all species of cormorants are
foot-propelled divers (Townsend, 1909), some taxa are
shallow divers and others are primarily deep divers.
Moreover, some typically roost or nest in trees, some
on cliffs and others on flatter substrates. Both these
behaviour categories are likely to be reflected differ-
entially as adaptive variation in the pelvic girdle
elements. For this reason, all 48 pelvic girdle charac-
terswere identifiedas
homoplasy and were treated in some analyses with
differential weighting schemes. Analyses with the
pelvic characters variously weighted from 0.9 to 0.1
varied little when less than 0.6, so I conducted final
analyses with weights of 0.5.
The phylogenetic analyses used PAUP* 4.0b10
(Swofford, 2001). Parsimony analyses used heuristic
searcheswith tree-bisection-reconnection
swapping, and 1000 random addition replicates per
search. Strict consensus trees of all most parsimonious
trees were created and rooted with outgroups forming
a polytomy at the base of the tree. When calculating
tree lengths, multistate taxa were treated as polymor-
phisms rather than ambiguity. Bootstrapping used
1000 heuristic searches and the same options. Trees
were manipulated in MrEnt (Zuccon & Zuccon, 2010).
Initially, analyses were performed with all charac-
ters unordered and equally weighted with no con-
straints enforced. Then analyses with 19 characters
ordered (as above), with and without weighting of the
pelvic girdle elements were carried out. Next, the
effect of weighting the pelvic girdle characters was
examined, with and without character ordering.
Lastly, because genetic data suggest that P. gaimardi
is the sister to cormorants other than Microcarbo
species (Kennedy et al., 2009), the effect of enforcing
this relationship was examined. I made the simplest
possible constraint to reflect this relationship as
follows: (A. anhinga (M. melanoleucos (P. gaimardi (P.
auritus P. carbo P. pelagicus P. fuscescens P. aristotelis
P.punctatusL.chalconotus)))),
Microcarbo taxa representing all clades found by
Kennedy et al. (2009) were constrained to group
inside A. gaimardi. Thus a majority of species (25,
including all the fossils and 16 phalacrocoracids) were
free to vary as their phylogenetic signal dictated, and
for those seven taxa constrained as sister to P. gaima-
rdi, internal relationships were not dictated.
potentially affected by
branch
i.e. seven non-
FOSSIL SITES
The specimens described below derive from localities
in two depositional basins of the Lake Eyre Basin: the
western Tirari Sub-basin (formerly Lake Eyre Sub-
Basin) and eastern Callabonna Sub-basin (formerly
Tarkarooloo Sub-Basin) (Tedford, Williams & Wells,
1986; Vickers-Rich, 1991; Woodburne et al., 1994;
Callen, Alley & Greenwood, 1995; Alley, 1998).
Pelecaniform fossils have been recorded from Lake
Palankarinna 28°46-47′S, 138°24′E in the Tirari Sub-
basin and Lake Yanda, 31°0.05′S, 140°18.5′E, Lake
Pinpa (= Pine Lake) 31°8′S, 140°13′E, and Billeroo
Creek, 31°6′S, 140°14′E in the Callabonna Sub-basin
(Rich et al., 1991; Vickers-Rich, 1991). The fossils
derive from the Etadunna Formation in the Tirari
Sub-basin (Woodburne et al., 1994) and the Namba
Formation in the Callabonna Sub-basin (Callen &
Tedford, 1976; Tedford et al., 1977). I follow Woodburne
et al. (1994) for the nomenclature of local faunas and
fossil mammal zones in accepting a Late Oligocene
26–24 Mya age for the Etadunna and Namba Forma-
tions. Woodburne et al. (1994) correlated Zone A, the
oldest mammal zone in the Etadunna Formation,
which has produced the Minkina LF, with the Pinpa
LF of the Namba Formation, and correlated the super-
jacent Zone B, containing the Ditjimanka LF at Lake
Palankarinna, with the Ericmas Fauna in the upper
part of the Namba Formation.
RESULTS
SYSTEMATIC PALAEONTOLOGY
ORDER PELECANIFORMES SHARPE, 1891
FAMILY PHALACROCORACIDAE REICHENBACH, 1849:
SHAGS AND CORMORANTS
The following fossil taxa are referred to Phalacro-
coracidae rather than other pelecaniform families
(Pelecanidae, Fregatidae, Sulidae, Anhingidae) by the
following uniquecombination
humerus with a deep, rather than shallow, fossa
pneumotricipitalis ventralis; (2) humeral proc. flexo-
rius with equal or greater prominence than the epi-
condylusventralis,not
supracondylaris dorsalis located distal to the proxi-
mal margin of the condylus dorsalis (exception: Nam-
bashag microglaucus gen. et sp. nov. is primitive for
this character); (4) ulna with the tuber. collateralis
ventralis elongate and distinctly separated from the
cotylar margin; (5) carpometacarpus with the ventral
rim of the trochlea carpalis ventralis joining the shaft
markedly distal to the proc. pisiformis; (6) coracoid
with the protuberant boss on the cranial margin of
the facies artic. sternalis dorsalis occupying no more
than half the total width of the articular facet; and (7)
femur with the tuber. gastrocnemialis lateralis elon-
of characters:(1)
less; (3)humeralproc.
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
283
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 8

gate and distinctly separated from the trochlea fibu-
laris. Phalacrocoracids are also diagnosed by the
following additional characters that were not assess-
able in the fossil material at hand: premaxillae with
rostral grooves extending to near the tip, os lacri-
males fused to the os frontales, large fonticuli cranio-
orbitales in the caudal walls of the orbits, and os
palatines only partly or not fused medially.
NAMBASHAG GEN. NOV.
Type species: Nambashag billerooensis sp. nov.
Diagnosis: A phalacrocoracid diagnosed by the follow-
ing unique combination of characters: ulna with proc.
cotyla dorsalis not distinctly hooked distally; coracoid
with cotyla scapularis having a shallow sulcus; femur
with insertion for M. caudofemoralis a short ovate
crista immediately proximal to the nutrient foramen
well separated from the crista marking the insertion
of M. flexor ischiofemoralis more proximally; tibiotar-
sus with the thickened tip of the crista cnemialis
cranialis in line with rest of the crista, rather than
offset anteriorly, sulcus extensorius laterally placed,
and pons supratendineus proximodistally short; tar-
sometatarsus with tendinal canal for M. flexor halli-
cus longus open plantarly, the tendinal canal for M.
flexor digitorum longus nearly enclosed, the tendinal
canal for M. flexor perforatus digiti 2 broadly open
laterally, groove for M. extensor hallicus longus where
it passes over the dorsomedial shaft margin signifi-
cantly wider than shaft width, lateral margin slightly
concave in anterior view, sulcus for M. abductor digiti
IV deep over whole length of shaft so crista plantaris
dorsalis is parallel and close to the dorsal facies over
whole shaft length, and foramen vasculare distale
relatively small and well separated from incisura
intertrochlearis lateralis.
Etymology: Noun in the nominative singular, masculine;
from Namba, from which formation most fossils herein
derive, and shaahg, the European name for Phalacroco-
rax aristotelis, from Old Norse skegg (the beard), so
called on account of the recurved crest of feathers that its
headisadornedwithinspringplumage.Thename‘shag’
first appeared in Merrett (1667).
NAMBASHAG BILLEROOENSIS SP. NOV. (FIGS 1, 2)
Holotype: SAM P.29079, right tarsometatarsus.
Diagnosis:
size of P. carbo with the sulcus for the origin of M. flexor
hallicus longus on the femur proximal to the trochlea
fibularis shallow with little craniocaudal depth.
A speciesof
Nambashag
about the
Figure 1. Holotype tarsometatarsus of Nambashag billerooensis sp. nov. SAM P.29079, in dorsal (A), plantar (B),
medial (C), lateral (D), and proximal (E) views. Plantar wall of fhl broken in this specimen. Scale bar = 10 mm.
Abbreviations: cl, cotyla lateralis; cm, cotyla medialis; ehl, groove for the M. extensor hallicus longus; ei, eminentia
intercotylaris; fdl, tendinal canal for M. flexor digitorum longus; fhl, tendinal canal for M. flexor hallicus longus; fmI, fossa
metatarsi I; fp2, tendinal canal for M. flexor perforatis digiti 2; fpm, fossa parahypotarsalis medialis; fvd, foramen
vasculare distale; fvp, foramen vascularis proximalis lateralis; ire, impressiones retinaculi extensorii; saIV, sulcus for M.
abductor digiti IV; tmt, tuberositas M. tibialis cranialis; II, trochlea metatarsi II; III, trochlea metatarsi III; IV, trochlea
metatarsi IV.
284
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 9

Figure 2. Nambashag billerooensis sp. nov.: left ramus of mandible (SAM P.32573) in dorsal (A) and medial (B) views;
right quadrate (SAM P.32580) in cranial (C) and caudal (D) views; humeri, left (SAM P.32581) in cranial view (E),
proximal left (SAM P.32569) in cranial (F) and caudal (G) views; ulnae, proximal right (SAM P.41264) in ventral aspect
(H), distal right (UCR16097) in dorsal aspect (I); left carpometacarpus (SAM P.32574) in dorsal aspect (J); left tibiotarsus
(SAM P.32565) in cranial (K) and proximal (L) aspects; left coracoid (SAM P.41289) in dorsal (M) and ventral (N) aspects;
left femur (SAM P.32567) in cranial (O), caudal (P), and lateral (Q) aspects. Scale bars = 10 mm. Abbreviations: ccl, crista
cnemialis lateralis; cd, crista deltopectoralis; ci, crista intercotylaris; cl, cotyla lateralis; co, capitulum oticum; coc,
condylus caudalis; col, condylus lateralis; com, condylus medialis; cs, capitulum squamosum; ct, crista transversa fossae;
fac, facies artic. clavicularis; fb, fossa M. brachialis; fcc, fovea carpalis caudalis; fp, fossa pneumotricipitalis; ic, impressio
coracohumeralis; Mc, insertion for M. caudofemoralis; mc, margo caudalis; Mfhl, origin for M. flexor hallicus longus; Mfi,
insertion for M. flexor ischiofemoralis; Mp, insertion for M. psoas; omm, os metacarpale minus; pa, proc. acrocoracoideus;
pcd, proc. cotyla dorsalis; po, proc. orbitalis; pp, proc. procoracoideus; ps, pons supratendineus; sc, cotyla scapularis; tc,
tuber. carpale; tcv, tuber. lig. collateralis ventralis; tsv, tuber. supracondylare ventrale; tub, tuberculum.
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
285
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 10

Etymology: After Billeroo Creek in South Australia,
where many of the fossils described here were
collected.
Type locality: White Sands Basin, Lake Palankarinna,
South Australia, Australia. Collected by M. Fuller on
9.vii.1987.
Horizon: Stratigraphy/age/fauna: Zone B, Etadunna
Formation, Late Oligocene 24–26 Mya, Ditjimanka LF.
Distribution: Late Oligocene (24–26 Mya), South Aus-
tralia, Australia: Lake Palankarinna, Etadunna For-
mation, Minkana LF, Zone A, and Ditjimanka LF,
Zone B; Lake Pinpa, Namba Formation, Pinpa LF;
Billeroo Creek, Namba Formation; Lake Yanda,
Namba Formation, Yanda LF.
Measurements of holotype: Length 48.7 mm, proximal
width 11.1 mm, proximal depth 15.4 mm, mid shaft
width 6.1 mm, mid shaft depth 4.7 mm, maximum
distal width 13.6 mm, width trochlea III 5.2 mm, depth
trochlea III 6.6 mm, length of distal foramen 2.0 mm,
distance foramen to intertrochear notch 4.1 mm.
Paratypes: Site 2 Billeroo Creek, Frome Downs
Station, 31°06.205′S, 140°13.912′E, Namba Forma-
tion, Late Oligocene, all collected THW & A. Camens
5–6.vi.2007 – SAM P.32566, shaft R tarsometatarsus;
SAM P32565, L tibiotarsus; SAM P.32567, L femur;
SAM P.32568, R femur; SAM P.32569, proximal L
humerus; SAM P.32571, distal R ulna; SAM P. 32574,
L carpometacarpus; SAM P.32580, R quadrate; SAM
P.32572,posteriorpart
P.32573, posterior part L side mandible; Site 7,
31°07.837′S, 140°12.636′E, Lake Pinpa, South Austra-
lia, Namba Formation, Late Oligocene, Pinpa LF –
SAM P.43135, proximal L tarsometatarsus, collected
THW et al. 29.v.2007.
Rsidemandible; SAM
Referred material: Site 2, 31°06.205′S, 140°13.912′E,
Billeroo Creek, Frome Downs Station, South Austra-
lia, Namba Formation, Late Oligocene, all collected
THW & A. Camens 5–6.vi.2007 – SAM P.32570, frag-
ment distal L humerus; SAM P. 32575–577, three
cervical vertebrae; SAM P. 32579, R fibula; SAM P.
32578, distal R radius. Billeroo Creek, Frome Downs
Station, collection data THR 1977–229, grid 319151
CURNAMONA sheet (? = Site 2 Worthy & Camens) –
MV P.222423, proximal R humerus; MV P.222422,
proximal R ulna.
Lake Pinpa, Frome Downs Station, SA, Namba
formation, Pinpa LF, Late Oligocene – SAM P.32581,
L humerus in unjoined proximal and distal parts,
Site 8, 31°08.167′S, 140°12.636′E, collected THW &
Camens vi.2007; SAM P.43022, a very fragmented R
humerus, Site 10, 31°08.088′S; 140°12.622′E, col-
lected THW & Camens vi.2007; MV P.230791, distal R
tibiotarsus, Lake Pinpa, immediately south of the
east-west track across the floor of the lake, resting on
the Namba Fm, collected Tom Rich et al. 4.vii.1980,
THR 1980–16 (= PVR 80–1).
Lake Palankarinna, SA, Etadunna Formation, Late
Oligocene – SAM P.41275, distal L tarsometatarsus,
Steve’s Site, south of Tedford Locality (RV-8447),
Ditjimanka LF, Zone B, collected 6.vii.1987; SAM
P.42008, very worn distal R humerus, White Sands
Basin, Ditjimanka LF, Zone B, collected N.S. Pledge
11.viii.1995; SAM P.41264, proximal R ulna, White
Sand Locality, Ditjimanka LF, Zone B; SAM P41289,
complete L coracoid assembled from a cranial end
collected from Mother Lode (RV-8504) in ix.1992 and
a sternal part collected at Neville’s Nirvana on
4.viii.1999 (these sites are 20 m apart, pers. comm., J.
McNamara, South Australia Museum, 19.viii.2009),
Minkana LF, Zone A; UCR 16097, distal R ulna,
Turtle South (RV-7252), Minkana LF, Zone A.
Lake Yanda, Yanda LF, Namba Formation, Upper
Oligocene – MV P.222430, anterior sternum, MV
P.222420, L quadrate.
Measurements: See Table 2.
Description: Numbers in parentheses refer to charac-
ters in Appendix 1.
Mandible (SAM P. 32572, SAM P.32573; Fig. 2A, B).
The mandible, indeed the whole bill, is represented
only by these two caudal fragments. The right frag-
ment (P.32572) is more complete anteriorly with the
os surangular preserved to and including the angulus
mandibulae and the os prearticulare preserved to a
few millimetres anterior of this point, but the caudal
part of the cotyla lateralis, the crista transversa
fossae,andtheproc.mandibulae
missing. The left fragment (P.32573) is less complete
anteriorly, but the cotyla lateralis and crista trans-
versa fossae are complete and it lacks the proc.
medialis (Fig. 2A, B). Both specimens show that the
crista intercotylaris is a prominent caudomedially
directed boss, separated from the cotyla lateralis (23)
typical of phalacrocoracids. A large tubercle is present
dorsally at the anterior end of the cotyla lateralis (24):
it is present in all phalacrocoracids, but is reduced
and low in some taxa such as those included in
Leucocarbo. Breakage obscures the form of the fossa
caudalis, but the medial facies below the anterior end
of the cotyla medialis is deeply concave (26). The fossa
aditus canalis mandibulae was an extensive fossa
extending close to the ventral margin of the mandible,
which lacked a caudal pocket (27). The proc. coronoi-
deus has a well-developed tuberosity dorsal to the
medialis are
286
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 11

anterior end of the fossa aditus canalis mandibulae,
about 10 mm anterior of the tubercle adjacent to the
cotyla lateralis (28).
Quadrate (SAM P.32580, MV P.222420; Fig. 2C, D).
SAM P. 32580 has a maximum height of 16.2 mm and
width of 11.1 mm and MV P.222420, a height of
13.8 mm. These agree with the general form of crown
group phalacrocoracids with an obsolete proc. orbit-
alis, widely separated condyli medialis et lateralis
withadeepnotchbetween,
elevated c. 5 mm relative to the other condyles, the
capitulum squamosum about twice as large as the
capitulum oticum, from which it is separated by a
shallow notch. The tubercle, laterally, below the
capitulum squamosum is broken but its remaining
extent indicates it would have been anteriorly promi-
nent (21). Medially, a small foramen pneumaticum is
present at mid height (22).
Humerus
(MVP.222423,
P.32569, SAM P.32581, SAM P.32570, SAM P.43022;
Fig. 2E–G). The available humeral fragments are
similar to those of phalacrocoracids in their propor-
tions and reveal the following features. The tuber.
dorsale is discrete and elevated from the surrounding
facies; the crista deltopectoralis is concave dorsally;
there is a prominent margo caudalis directed towards
the centre of the caput humeri and linked to the crus
dorsale fossae. The incisura capitis is broad with a
round ligamental attachment for the M. subscapu-
condyluscaudalis
SAMP.42008, SAM
laris (Owre, 1967: 24) at its dorsal end abutting the
end of the margo caudalis, resulting in the dorsal end
of the incisura being elevated to enclose a sulcus in
the middle of the incisura, which varies in depth
intraspecifically. The insertion of the coracobrachialis
posterior is a shallow round sulcus on the dorsal side
of the tuber. ventrale. The fossa pneumotricipitalis
extends deeply under the tuber. ventrale (31) and is
floored by an osseous lamina that is penetrated by
several small pneumatic foramina. The incisura
capitis interrupts the proximal profile as a shallow
notch (32). The sulcus lig. transversus extends ven-
trally from adjacent to the caput humeri as a shallow
groove that in ventral view only makes a shallow
notch less than half the depth of the crista bicipitalis,
and whose ventral end is partly blocked by a thin
osseous wall (33). The crista deltopectoralis in SAM
P.32569 is 34.5 mm long and extends only 12 mm
distal of the crista bicipitalis (34). The width of the
ventral fossa pneumotricipitalis is greater than the
width dorsally of the crus dorsale fossa to the base of
the tuber. dorsale (35). The ventral facies immedi-
ately distal of the sulcus ligamentum transversus is
convex, not with a distinct groove as seen in some
phalacrocoracids (36). The impressio coracobrachialis
is a broad and deep sulcus smoothly rising to the
intumescentia humeralis ventrally, i.e. not emargin-
ated by a distinct crest (37) that narrows distally
(38). The tuber. supracondylare ventrale is directed
Table 2. Measurements (mm) of bones of Nambashag billerooensis sp. nov.
LengthProximal width Proximal depthShaft width Shaft depthDistal width
Tarsometatarsi
SAM P.29079
SAM P.43135
SAM P.41275
SAM P.32566
Tibiotarsi
SAM P.32565
MV P.230791
Femora
SAM P.32567
SAM P.32568
Humeri
MV P.222423
SAM P.32569
SAM P.32581
Ulnae
MV P.222422
SAM P.32571
SAM P.41264
Carpometacarpus
SAM P.32574
48.7
–
–
–
11.1
11.8
–
–
15.4
–
–
–
6.1
–
7.1
6.8
4.7
–
4.1
4.9
13.6
–
–
–
89.6
–
–
–
–
–
7.0
6.6
5.0
–
10.9
11.2
56.0–
c. 14.1
9.2
–
6.1
6.3
6.6
6.5
–
–
–
–
c.120
20.7
20.8
21.7
–
6.4
–
–
–
7.4
–
–
–
–
–
14.1
–
–
–
10.3
–
10.1
–
–
–
–
–
5.3
–
–
–
–
10.2
–
64.0–––––
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
287
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 12

cranially (39), rather than being rotated ventrally as
in most phalacrocoracids. Although in the best pre-
served distal humerus (SAM P.32581), the cranio-
proximal sides of both condyles are broken, the
broken end of the proc. flexorius is equally ventrally
prominent with the epicondylus ventralis indicating
that when complete it would have been more promi-
nent (40) as in all phalacrocoracids, but unlike
Anhinga. The proc. supracondylaris dorsalis is a
rounded prominence (preserved in SAM P.32570)
located distal to the proximal margin of the condylus
dorsalis (41), again as in phalacrocoracids and unlike
Anhinga. The fossa M. brachialis extends diagonally
from the dorsal margin of the shaft as a parallel sided
impression distally to abut the tuber. supracondylare
ventrale, where it is relatively deeper than proximal
parts of the fossa (42). The proc. flexorius and condyli
ventralis et dorsalis have a similar distal projection,
and the fossa olecrani is broad and deep.
Ulna (MV P.222422, SAM P.32571, SAM P.41264,
UCR 16097; Fig. 2H, I). The proximal fragments MV
P.222422 and SAM P.41264 reveal that the proc. cotyla
dorsalis is prominent cranially and, although about
4.0 mm long proximodistally, does not form a marked
distally directed hook (43); the tuber. lig. collateralis
ventralis is separated from the cotyla by a shallow
groove, is not ventrally prominent of the cotyla (44), is
elongate and separated from the cotylar margin (45),
the attachment of the M. bicipitis below the cranial lip
of the cotyla ventralis is elongate; the impressio bra-
chialis is 26 mm long and undercuts the ventral facies
in its proximal half; the incisura radialis is shallow,
bound by a single elevated ligamental attachment
aligned at about 45 degrees to the shaft axis and
extending proximoventrally from the base of the cotyla
dorsalis; the cranial facies is flat and roughly at right
angles to the adjacent impressio brachialis; and the
cranial intermuscular line is not prominent. The ole-
cranon is broken, so its extent proximal to the cotyla
dorsalis cannot be ascertained. The distal fragments
SAM P.32571 and UCR 16097 reveal that the tuber.
carpale is robust and shorter than crown phalacroco-
racids, for example P. carbo, and is more like Anhinga;
the incisura tuber. carpale forms a distinct notch that
is proximally offset from the sulcus intercondylaris in
ventral view, more like Anhinga than Phalacrocorax;
the condylus ventralis ulnaris has a ventrally promi-
nent tubercle at the distal margin of the depressio
radialis whose cranial face is visible in ventral view, as
in P. carbo; the condylus ventralis ulnaris projects only
slightly cranially of this tuberculum, rather than pro-
jecting cranially a distance equal to tuberculum width
as in crown phalacrocoracids, e.g. P. carbo; caudally,
the condylus dorsalis has an oval sulcus in its cranial
half from which the incisura tendinosa emanates,
which is bound cranially by a more robust ridge
compared with P. carbo (more like Anhinga), but
unlike both Anhinga and P. carbo, the most proximal
part of this sulcus is terminated by a triangular
prominence related to greater angularity of the shaft
in Nambashag.
Carpometacarpus (SAM P.32574; Fig. 2J). This
specimen preserves the complete length, but is
missing the proc. extensorius, part of the ventral rim
and cranial end of the trochlea carpalis, and the os
metacarpale minus. The length of the proximal synos-
tosis, between the proc. alularis and the spatium
intermetacarpale,is longer
(4.5 mm) (47). The fovea carpalis caudalis is deep,
short, and wide (width approximates length) (49). The
width of the os metacarpale minus at the proximal
synostosis is much less than that of the os metacar-
pale majus (50), but it is broken at the synostosis
precluding the determination of whether it was
grooved immediately distal to that point. The groove
extending proximally from the spatium intermetacar-
pale is short and does not link to the sulcus cranial to
the proc. pisiformis (52). The ventral rim of the tro-
chlea carpalis joins to the shaft well distal to the proc.
pisiformis (53) not adjacent to as in Anhinga.
Coracoid (SAM P.41289; Fig. 2M, N). Medial length
59.8 mm, maximumdorsoventral
facies artic. humeralis 9.7 mm, length from cotyla
scapularis to tip of the proc. acrocoracoideus 20.1 mm.
This generally well-preserved specimen, reassembled
from two halves collected about seven years apart,
lacks the entire proc. lateralis and the tip of the proc.
medialis. The proc. acrocoracoideus only slightly over-
hangs the ventrolateral facies (55), not markedly so
as in Anhinga. It has the pelecaniform apomorphy of
the facies artic. clavicularis being ovate and on the
ventral facies of the proc. acrocoracoideus (56), but
this facies is similar to those of phalacrocoracids in
being craniosternally elongate (8.1 by 6.6 mm), rather
than lateromedially elongate as in Anhinga. The
sulcus M. supracoracoidei lacks pneumatic foramina
and does not undercut the ridge linking the facies
artic. humeralis and the proc. acrocoracoideus (57), as
it does in crown group phalacrocoracids. The area
between the facies artic. humeralis and the tip of the
proc. acrocoracoideus is excavated in a shallow
parallel-sided groove (58), rather than a deep groove
narrowing ventrally because of a ridge running from
the tip, as in crown group phalacrocoracids. The proc.
acrocoracoideus does not extend medially over the
sulcus M. supracoracoidei, is deeper than wide, and
the facies artic. humeralis is elongate (11.3 ¥ 5.3 mm)
and projects laterally. The cotyla scapularis has a
shallow but distinct sulcus, rather than being flat or
convex as in crown group phalacrocoracids (59). This
depression is bound by a row of pneumatic foramina
separating the cotyla from the sulcus M. supracora-
(6.5 mm)thanwide
depth through
288
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 13

coidei. The articular facet is continuous with the
faciesartic.humeralis
extends onto the proc. procoracoideus. A distinct, cra-
nially directed prominence defines the mediodorsal
margin of the cotyla, which itself is directed crani-
olaterally. This contrasts with crown group phalacro-
coracids, where the cotyla is more elongate, directed
more laterally, and somewhat sloped sternally. The
proc. procoracoideus is stout, short, not extending
above the cotyla, and lacks a foramen nervi supraco-
racoidei. The shaft is narrow and elongate as typical
of phalacrocoracids. Ventrally, the linea muscularis
ventralis is prominent, laterally enclosing a shallow
sulcus, and straight, not recurved medially near the
sternal end as in P. carbo, and joins the facies artic.
sternalis no less than 13 mm from the angulus media-
lis. As it is most improbable that this articular facet
was 26 mm wide, it is inferred that the linea joins the
facet in its lateral half (60). The facies artic. sternalis
dorsalis has a prominent boss 9.0 mm long and, maxi-
mally, c. 3 mm wide, which is estimated to occupy less
than half the width of the sternal articulation (61).
The facies artic. sternalis ventralis is worn but lacked
a prominent lip such as seen in P. carbo, and thus is
more like M. melanoleucos in this feature.
Sternum (MV P.222430). This tentatively referred
anterior fragment preserves part of the coracoidal sulci
and the base of the carina, revealing a wide notch
between the labri interni and so the absence of a spina
interna rostri, and a wide flattened spine externa.
Femora (SAM P. 32567, P. 32568; Fig. 2O–Q). There
is damage to the femoral ball and the condylus media-
lis in P.32567, and P.32568 has the distal parts of both
condyles lost, but the following features are determin-
able. The dorsal pretrochanteric surface is excavated
in a shallow sulcus below the facies artic. antitrochan-
terica (69). Distally of this sulcus, the insertion for the
M. psoas forms an obvious oval area centred on the
pretrochanteric facies, rather than being located close
to the medial side (70). The insertion for the M.
femorotibialis internus on the medial facies is ovate
and not raised into a prominent tubercle (71). The
crista trochanteris is slightly shorter than proximal
width and is low. The femoral ball does not extend
proximal to the crista trochanteris. The linea inter-
muscularis cranialis connects to the medial side of the
crista trochanteris smoothly, without a distinct rugos-
ity as in most phalacrocoracids (72). The lateral facies
between the impressions for the M. obturator internus
and the M. ilio-trochantericus medialis (Owre, 1967) is
smooth with no deep sulcus (73), such as seen in
Leucocarbo chalconotus. Caudally, there is a broad
shallow fossa adjacent to the facies artic. antitrochan-
terica (74). There is a short ovate rugosity slightly
proximal to the nutrient foramen for the insertion of
M. caudofemoralis, which is separated by c. 5.0 mm in
mediallyand, laterally,
both
ischiofemoralis more proximally (75). The latter inser-
tion extends around onto the lateral facies. This
arrangement is seen in Anhinga and other pelecani-
forms, but not in any phalacrocoracids. In Microcarbo
taxa, the insertions for M. caudofemoralis and M.
flexor ischiofemoralis are separated by a short gap and
in remaining crown group taxa the insertions form a
continuous crista, so the wide separation of the inser-
tions in Nambashag is plesiomorphic. The tuber. M.
gastrocnemialis lateralis is located distinctly proximal
to the condylus fibularis and is elongate at c. 7.0 mm
long (76). The origin of M. flexor hallicus longus (Owre,
1967) forms a small shallow sulcus just proximal to the
condylus fibularis, which does not extend as far as the
tuber. M. gastrocnemialis lateralis and is restricted to
the caudal half of shaft depth, but which is emargin-
ated cranially by a low crest (77). The fossa poplitea is
broad and shallow. The caudal parts of both the
condylus lateralis and the condylus medialis are not
preserved. Although the shaft is caudally deflected
slightly in its distal third of length, as it is to varying
degrees in all phalacrocoracids, it was definitely not
markedly caudally rotated as in some modern phala-
crocoracids (80). The tuber. M. gastrocnemialis media-
lis is triangular, abuts the medial margin of the shaft,
is proximally separated from the condylus medialis
(82), and is integrated as a single tuberosity with the
crista supracondylaris medialis (81), rather than the
latter passing mesad of it. Despite the distal parts of
the condylus medialis not being preserved, the remain-
ing parts suggest that the condyle did not extend
mesial to the shaft (83). Overall, the femora are
relatively elongate compared to phalacrocoracids and
their length exceeds the length of associated tar-
sometatarsi (84), unlike the situation in most phalac-
rocoracids where femur length is equalled or exceeded
by tarsometatarsi length.
Tibiotarsus (SAM P.32565, MV P.230791; Fig. 2K,
L). SAM P.32565: Total length 89.6 mm, length from
proximal articular surface 84.6 mm, mid shaft width
6.96 mm, mid shaft depth 5.0 mm, distal width
10.9 mm, length fibular crest 23 mm, maximum width
of shaft through fibular crest 9.5 mm. This bone is
relatively robust. The crista cnemialis lateralis has its
tip thickened and formed on a plane with the rest of
crest (85), as in Anhinga, not displaced anteriorly of
the crista as in all phalacrocoracids; slightly so in
Microcarbo and P. neglectus, but markedly so in all
other Phalacrocorax species. The crista cnemialis lat-
eralis is not greatly elongated proximally, so that the
tip proximodistally overlaps the area interarticularis
(86). The crista cnemialis cranialis is prominent ante-
riorly and extends down the medial margin of the
shaft overlapping the fibular crest. The fossa flexoria
has a distinctly broad lateral part with the medial
specimens fromthecrista for M.flexor
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
289
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 14

part obsolete (87). The impressio lig. collateralis
medialis is not medially prominent and comprises two
distinct impressions (88), one for the insertion of M.
flexor cruris lateralis towards the caudal side of the
medial facies and, well separated from the latter, a
more elongate one for M. flexor cruris medialis on the
anterior margin (Owre, 1967: 80–81). This is similar
to the state in Microcarbo and P. gaimardi, but differs
from all other phalacrocoracids where they are both
in the anterior half of shaft depth and close together.
Some breakage and loss of the facies artic. lateralis
obscure the structure, but the length from its distal
side to the crista fibularis is about 10.5 mm, much
greater than shaft width at this point (7.0 mm) (89).
The epicondylus medialis forms a prominent tuberos-
ity on the medial facies (90). The tuber. retinaculi M.
fibularis extends about 5.5 mm proximal of the pons
supratendineus and is prominent laterally with no
anterior elevation (contrasts with Microcarbo species
that are prominent anteriorly) so that the lateral
margin of the sulcus extensorius is separated from
the lateral margin of the tuberculum by a rounded
ridge (91). The condylus medialis has slightly greater
anterior projection than the condylus lateralis, but its
projection anterior to the incisura intercondylaris is
much less than the depth posterior of that point (92).
The distal extent of the condyli medialis et lateralis
are equal (93), unlike all crown group phalacrocorac-
ids where the condylus medialis has markedly more
distal extent. The sulcus extensorius is positioned
laterally of centre on the shaft (94), and its width at
right angles to the axis on the proximal side of the
pons supratendineus is 3.8 mm, much wider than the
proximodistal 2.1 mm length of the pons supratend-
ineus (95). The shape of MV P.230791 conforms to the
above in all aspects.
Tarsometatarsus (SAM P.29079, P.43135, P.41275,
P.32566; Fig. 1). The plantar depth of the proximal
end is significantly greater than proximal width (99)
as in all members of Phalacrocoracidae, but contrast-
ing with Anhinga, Fregata, Sula, and Morus. Nam-
bashag billerooensis has an arrangement of tendinal
canals similar to that seen in all phalacrocoracids
and which differs from anhingids. The tendinal canal
for M. flexor hallicus longus (fhl) is the most laterally
and dorsally placed of the three main canals and is
closed plantarly (97) by a osseous bridge that bears a
shallow sulcus for more plantar tendons; seen in
SAM P.43135, broken in SAM P.29079. The canal for
M. flexor digitorum longus (fdl) lies medially to and
slightly plantar of fhl adjacent to the large robust
crista medialis plantaris (as in all phalacrocoracids)
and is not quite closed in its lateroplantar zone (96).
The tendinal canal for M. flexor perforatis digiti 2
(fp2) is a broadly open laterally directed groove
located plantar of and slightly laterad of fdl (98)
unlike in Anhinga where it is enclosed. The plantar
facies of the hypotarsus is a flattened oval plate
forming a distally directed hook in lateral view (102).
The eminentia intercotylaris has a similar proximal
extent as the plantar cotylar margins (100), rather
than exceeding them as in Anhinga. The groove
marking the track of the M. extensor hallicus longus
over the dorsal medial margin of the shaft is 7.6 mm
wide (measured as length down shaft), which is sig-
nificantly greater than shaft width (101), unlike all
crown phalacrocoracids where the trace width is
equal to or less than shaft width. The fossa parahy-
potarsalis medialis is obsolete (103), but a small crest
lies on the medial facies extending about 7 mm proxi-
mally from the groove for M. extensor hallicus
longus. The impressiones retinaculi extensorii are
low crests, a medially placed one medial to the
foramen vascularis proximalis medialis and the other
on the dorsomedial margin of the sulcus extensorius
with the intervening surface flat and sloping and
much elevated above the base of the sulcus extenso-
rius (104). The tuberositas M. tibialis cranialis
comprises two discrete tuberosities that are latero-
medially adjacent to each other and separated by a
shallow groove, with both lying immediately distal to
the foramina vascularia
extensorius extends distal to mid length but is sepa-
rated from the foramen vasculare distale by c. 7 mm.
In anterior view, the lateral margin is slightly
concave distal to the foramina vascularia proximalia
(105), not planar as in Anhinga and Microcarbo taxa.
In lateral view, the sulcus for M. abductor digiti IV is
broad and bound dorsally by the crista plantar lat-
eralis that is parallel and close to the dorsal facies
over the whole shaft length (106), as in Anhinga and
Microcarbo species, but unlike all other phalacroco-
racids where the crista plantar lateralis is more
widely separated from the dorsal surface at mid
length than it is either more proximally or more
distally. The shaft is wider than it is dorsoplantarly
deep (107). Distally, trochlea metatarsi II has equal
distal extent as trochlea metatarsi IV, with both
ending slightly proximal to trochlea metatarsi III.
Trochlea metatarsi II, however, ends only slightly
proximal to trochlea metatarsi III (108), rather than
distinctly so as in most phalacrocoracids other than
Microcarbo. Similarly, trochlea metatarsi IV ends
only slightly proximal to trochlea metatarsi III (109),
not markedly so as in Anhinga and Microcarbo. Tro-
chlea metatarsi IV is longer than it is wide at its
base (110). The foramen vasculare distale is rela-
tively small (2.04 mm long) and well separated from
the incisura intertrochlearis lateralis by 4.12 mm
(111), more like the small foramen seen in Leuco-
carbo taxa than in other phalacrocoracids. The fossa
metatarsi I is obvious in plantar aspect and c. 9 mm
proximalia. Thesulcus
290
T. H. WORTHY
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314

Page 15

long, but it barely impinges on the medial profile of
the shaft in anterior view.
Comparisons with fossil taxa
Borvocarbo guilloti is much smaller than N. bille-
rooensis and distinguished by a circular and deeply
concave cotyla scapularis and a convex, not flattened,
facies articularis clavicularis (Mourer-Chauviré et al.,
2004). ?Borvocarbo stoeffelensis is also small, and
shares several plesiomorphic features with Nam-
bashag, e.g. distal end of the proc. cotylaris dorsalis
on the ulna not forming a hook, the omal portion of
the facies artic. humeralis directed essentially later-
ally rather than dorsolaterally with its ventral
margin projecting laterally as in crown group phala-
crocoracids, the thickened tip of the crista cnemialis
lateralis on the tibiotarsus forming a plane with the
rest of the crest (character 9 in Mayr, 2007) and, on
the tibiotarsus, the condylus medialis protruding only
slightly distal to the condylus lateralis. It is distin-
guished from Nambashag by the shorter crista bicipi-
talis, which meets the humeral shaft at a steeper
angle (as in Oligocorax and Nectornis), a deeper cup
to the cotyla scapularis on the coracoid (Mayr, 2007:
fig. 5A), and the sulcus extensorius being located cen-
trally on the tibiotarsus (Mayr, 2007).
?Borvocarbo tardatus is of similar size to N. bille-
rooensis but is distinguished by a less hooked proc.
cotyla dorsalis on the ulna, the tip of the crista
cnemialis lateralis being displaced cranially of the
crista, a centrally located sulcus extensorius, a more
proximodistallyelongate
aligned pons supratendineus,
medialis projecting more distally on the tibiotarsus
(Göhlich & Mourer-Chauviré, 2010: figs 4, 5).
Oligocorax littoralis, the only species in the genus,
is of similar size to N. billerooensis. The coracoid has
the scapular articular surface forming a mediodistally
directed point on the proximal side of the proc. pro-
coracoideus and the proc. procoracoideus is robust
and globular (fide Mourer-Chauviré et al., 2004),
which are features shared with N. billerooensis, but it
differs by having the cotyla scapularis more deeply
concave, the facies artic. clavicularis much more cran-
iosternally elongate, and the linea intermuscularis
ventralis located more laterally (Milne-Edwards,
1867–71b: pl. 43, figs 5–7). The humerus differs from
that of Na. billerooensis by its relatively narrower
impressio coracobrachialis, which is more sharply
defined on its ventral margin and a relatively shorter
crista bicipitalis (Milne-Edwards, 1867–71b: pl. 44,
figs 1, 2). The ulna of O. littoralis shares with Na.
billerooensis the absence of a distal hook to the cotyla
dorsalis (Milne-Edwards, 1867–71b: pl. 44, figs 6, 7),
distinguishing both from crown group phalacrocorac-
ids. The tibiotarsus of Oligocorax has the proximodis-
and more
and
horizontally
the condylus
tal width of the condylus medialis narrower than in
extant Phalacrocorax species (see Milne-Edwards,
1867–71b: pl 42, fig. 7), and the crista medialis hypo-
tarsi on the tarsometatarsus protrudes less plantarly,
and the distal margin of the crista medialis hypotarsi
is near vertical rather than sloping distally as in
Phalacrocorax (Milne-Edwards, 1867–71b: pl. 42,
fig. 7; Mayr, 2001). The tarsometatarsus of O. littora-
lis is further distinguished from that of Na. bille-
rooensis by the tendinal canal for M. flexor hallicus
longus (fhl) being plantarly open (Milne-Edwards,
1867–71b: pl. 44, fig. 8).
Both Nectornis species are much smaller than Na.
billerooensis. Both are similar in having a slightly
concave cotyla scapularis, but differ from Na. bille-
rooensis and all crown group phalacrocoracids by: the
humerus having a short crista bicipitalis with a very
convex medial profile (caudal view) and that abruptly,
not gradually, joins to the shaft; the scapular articu-
lar surface of the coracoid extends over the proximal
surface of the proc. procoracoideus and forms a medio-
distally directed point; and the proc. procoracoideus is
more pointed (Mourer-Chauviré, in Paicheler et al.,
1978; Cheneval, 1984). In Nectornis, the trochlea
metatarsi II extends as far distally as trochlea meta-
tarsi III, as in Na. billerooensis and Microcarbo
species, but unlike Phalacrocorax species, in which it
is shorter (Mayr, 2001; data herein). Nectornis ana-
tolicus differs from Na. billerooensis by having a
foramen nervi supracoracoidei, a hollow in the sulcus
M. supracoracoidei that undercuts the ridge leading
from the facies artic. humeralis to the acrocoracoid tip
(Mourer-Chauviré, in Paicheler et al., 1978).
The Phalacrocorax intermedius humerus appears to
have a more projecting tuber. dorsale than Na. bille-
rooensis, which may relate to a more cranially
directed crista deltopectoralis (Milne-Edwards, 1867–
71b: pl. 43, figs 8, 9).
Limicorallus saiensis is a very small species based
only on a distal humerus of distal width 8.6 mm
(Kurochkin, 1968), which apart from size, differs from
Na. billerooensis by a ventrally inclined tuber. supra-
condylare ventrale (Mlíkovský & Švec, 1986) and also
by the fossa m. brachialis not extending alongside the
tuber. supracondylare ventrale, as judged from the
depiction in Kurochkin (1968).
?Limicorallus carbunculus is the smallest fossil cor-
morant taxon, with a tarsometatarsus two-thirds the
length of that for M. pygmaeus, and was referred to cf.
Phalacrocoracidae (Mayr, 2009). It differs from Na.
billerooensis and all crown group phalacrocoracids by:
(1) having the sulci for the tendons of M. flexor
perforatus digiti II and M. flexor hallicus longus on
the same dorsoplantar plane and separated by a low
ridge, rather than the former being displaced plan-
tarly and separated from the latter by a more pro-
AUSTRALIAN OLIGO-MIOCENE CORMORANTS
291
© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, 277–314