The Skull of Epidolops ameghinoi from the Early Eocene Itaboraí Fauna, Southeastern Brazil, and the Affinities of the Extinct Marsupialiform Order Polydolopimorphia

Abstract

The skull of the polydolopimorphian marsupialiform Epidolops ameghinoi is described in detail for the first time, based on a single well-preserved cranium and associated left and right dentaries plus additional craniodental fragments, all from the early Eocene (53–50 million year old) Itaboraí fauna in southeastern Brazil. Notable craniodental features of E. ameghinoi include absence of a masseteric process, very small maxillopalatine fenestrae, a prominent pterygoid fossa enclosed laterally by a prominent ectopterygoid crest, an absent or tiny transverse canal foramen, a simple, planar glenoid fossa, and a postglenoid foramen that is immediately posterior to the postglenoid process. Most strikingly, the floor of the hypotympanic sinus was apparently unossified, a feature found in several stem marsupials but absent in all known crown marsupials. “Type II” marsupialiform petrosals previously described from Itaboraí plausibly belong to E. ameghinoi; in published phylogenetic analyses, these petrosals fell outside (crown-clade) Marsupialia. “IMG VII” tarsals previously referred to E. ameghinoi do not share obvious synapomorphies with any crown marsupial clade, nor do they resemble those of the only other putative polydolopimorphians represented by tarsal remains, namely the argyrolagids. Most studies have placed Polydolopimorphia within Marsupialia, related to either Paucituberculata, or to Microbiotheria and Diprotodontia. However, diprotodonty almost certainly evolved independently in polydolopimorphians, paucituberculatans and diprotodontians, and Epidolops does not share obvious synapomorphies with any marsupial order. Epidolops is dentally specialized, but several morphological features appear to be more plesiomorphic than any crown marsupial. It seems likely Epidolops that falls outside Marsupialia, as do morphologically similar forms such as Bonapartherium and polydolopids. Argyrolagids differ markedly in their known morphology from Epidolops but share some potential apomorphies with paucituberculatans. It is proposed that Polydolopimorphia as currently recognised is polyphyletic, and that argyrolagids (and possibly other taxa currently included in Argyrolagoidea, such as groeberiids and patagoniids) are members of Paucituberculata. This hypothesis is supported by Bayesian non-clock phylogenetic analyses of a total evidence matrix comprising DNA sequence data from five nuclear protein-coding genes, indels, retroposon insertions, and morphological characters: Epidolops falls outside Marsupialia, whereas argyrolagids form a clade with the paucituberculatans Caenolestes and Palaeothentes, regardless of whether the Type II petrosals and IMG VII tarsals are used to score characters for Epidolops or not. There is no clear evidence for the presence of crown marsupials at Itaboraí, and it is possible that the origin and early evolution of Marsupialia was restricted to the “Austral Kingdom” (southern South America, Antarctica, and Australia).

Electronic supplementary material
The online version of this article (doi:10.1007/s10914-016-9357-6) contains supplementary material, which is available to authorized users.

Full text

ORIGINAL PAPER
The Skull of Epidolops ameghinoi from the Early Eocene Itaboraí
Fauna, Southeastern Brazil, and the Affinities of the Extinct
Marsupialiform Order Polydolopimorphia
Robin M. D. Beck
1,2
Published online: 26 October 2016
#The Author(s) 2016. This article is published with open access at Springerlink.com
Abstract The skull of the polydolopimorphian marsupialiform
Epidolops ameghinoi is described in detail for the first time,
based on a single well-preserved cranium and associated left
and right dentaries plus additional craniodental fragments, all
from the early Eocene (53–50 million year old) Itaboraí fauna
in southeastern Brazil. Notable craniodental features of
E. ameghinoi include absence of a masseteric process, very small
maxillopalatine fenestrae, a prominent pterygoid fossa enclosed
laterally by a prominent ectopterygoid crest, an absent or tiny
transverse canal foramen, a simple, planar glenoid fossa, and a
postglenoid foramen that is immediately posterior to the
postglenoid process. Most strikingly, the floor of the
hypotympanic sinus was apparently unossified, a feature found
in several stem marsupials but absent in all known crown mar-
supials. BType II ^marsupialiform petrosals previously described
from Itaboraí plausibly belong to E. ameghinoi;inpublished
phylogenetic analyses, these petrosals fell outside (crown-clade)
Marsupialia. BIMG VII^tarsals previously referred to
E. ameghinoi do not share obvious synapomorphies with any
crown marsupial clade, nor do they resemble those of the only
other putative polydolopimorphians represented by tarsal re-
mains, namely the argyrolagids. Most studies have placed
Polydolopimorphia within Marsupialia, related to either
Paucituberculata, or to Microbiotheria and Diprotodontia.
However, diprotodonty almost certainly evolved independently
in polydolopimorphians, paucituberculatans and diprotodontians,
and Epidolops does not share obvious synapomorphies with any
marsupial order. Epidolops is dentally specialized, but several
morphological features appear to be more plesiomorphic than
any crown marsupial. It seems likely Epidolops that falls outside
Marsupialia, as do morphologically similar forms such as
Bonapartherium and polydolopids. Argyrolagids differ marked-
ly in their known morphology from Epidolops but share some
potential apomorphies with paucituberculatans. It is proposed
that Polydolopimorphia as currently recognised is polyphyletic,
and that argyrolagids (and possibly other taxa currently included
in Argyrolagoidea, such as groeberiids and patagoniids) are
members of Paucituberculata. This hypothesis is supported by
Bayesian non-clock phylogenetic analyses of a total evidence
matrix comprising DNA sequence data from five nuclear
protein-coding genes, indels, retroposon insertions, and morpho-
logical characters: Epidolops falls outside Marsupialia, whereas
argyrolagids form a clade with the paucituberculatans
Caenolestes and Palaeothentes, regardless of whether the Type
II petrosals and IMG VII tarsals are used to score characters for
Epidolops or not. There is no clear evidence for the presence of
crown marsupials at Itaboraí, and it is possible that the origin and
early evolution of Marsupialia was restricted to the BAustral
Kingdom^(southern South America, Antarctica, and Australia).
Keywords Epidolops .Polydolopimorphia .Marsupialia .
Marsupialiformes .Argyrolagidae .Itaboraí .Eocene
Introduction
Calcareous deposits in the Itaboraí Basin in Rio de Janeiro
State, southeastern Brazil, preserve one of the few diverse
Electronic supplementary material The online version of this article
(doi:10.1007/s10914-016-9357-6) contains supplementary material,
which is available to authorized users.
*Robin M. D. Beck
R.M.D.Beck@salford.ac.uk
1
School of Environment & Life Sciences, University of Salford, M5
4WT, Manchester, UK
2
School of Biological, Earth and Environmental Sciences, University
of New South Wales, Sydney, NSW 2052, Australia
J Mammal Evol (2017) 24:373–414
DOI 10.1007/s10914-016-9357-6
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

early Palaeogene vertebrate faunas known from South
America outside Patagonia (Bergqvist et al. 2008). The
Itaboraí fauna formed the basis for recognizing the
Itaboraian South American Land Mammal Age
(SALMA; Gelfo et al. 2009; Woodburne et al. 2014b). The
absolute age of the Itaboraian has been difficult to resolve:
older papers typically interpreted it as Paleocene (Marshall
1985; Medeiros and Bergqvist 1999), but more recent works
have proposed a younger age, namely Paleocene-Eocene or
early Eocene (Marshall et al. 1997; Gelfo et al. 2009;
Woodburne et al. 2014b;Goinetal.2016,in press). The most
recent published estimate for the absolute age of the Itaboraian
is 50–53 MYA (Woodburne et al. 2014b).
Among the vertebrate fossils collected from Itaboraí are
hundreds of mammal specimens. Most of these are isolated
teeth and jaw fragments (Paula Couto 1952a,b,c,d;Cifelli
1983b;Marshall1987; Oliveira and Goin 2006,2011;Goin
and Oliveira 2007;Goinetal.2009), but postcranial (Paula
Couto 1952a; Cifelli 1983a,b;Szalay1994; Szalay and Sargis
2001; Bergqvist et al. 2004; Bergqvist 2008; Oliveira et al.
2016) and cranial (Paula Couto 1952a,b,c,d; Ladevèze 2004,
2007; Ladevèze and Muizon 2010; Oliveira and Goin 2015)
remains are also present. The mammalian fauna comprises
both eutherian and marsupialiform species, and is highly di-
verse, with more than 25 genera currently recognized
(Oliveira and Goin 2006,2011; Bergqvist 2008; Woodburne
et al. 2014a). However, representatives of non-therian lineages
known to have survived into the Cenozoic in South America
(namely monotremes, meridiolestidans, and
gondwanatherians; Pascual et al. 1992,1999; Gelfo and
Pascual 2001;Goinetal.2012b; Rougier et al. 2012)have
not been described from Itaboraí.
SeveralItaboraítaxaarecandidates for being the oldest pu-
tative crown marsupials known from South America, notably
the apparent paucituberculatan Riolestes capricornensis (if this
is not based on a deciduous premolar of another taxon; see Goin
et al. 2009;Beckin press-b), plus isolated marsupialiform tar-
sals that Szalay (1994) referred to his BItaboraí Metatherian
Groups^(IMGs) V and XII and which have been identified as
possibly representing early didelphimorphians or another
crown marsupial lineage (Szalay 1994; Szalay and Sargis
2001;Beckin press-b).
One of the best preserved marsupialiform fossils from
Itaboraí is DGM 321-M, a crushed partial cranium and asso-
ciated left and right mandibles of Epidolops ameghinoi (Paula
Couto 1952c;Marshall1982a:figs.62–63; Bergqvist et al.
2008:fig.9B).Epidolops ameghinoi is also represented by
more than one hundred additional craniodental fragments
from Itaboraí (Marshall 1982a:74–82), making it by far the
most abundant marsupialiform in the fauna. In his original
description of Epidolops, Paula Couto (1952c) identified a
second species, E. gracilis, among the Itaboraí material.
However, Marshall (1982a) considered that all the specimens
could be referred to a single species, E. ameghinoi.
Subsequently, Szalay (1994) tentatively referred isolated tar-
sals comprising his IMG VII morphotype to E. ameghinoi.
Epidolops is a member of the extinct order
Polydolopimorphia (Case et al. 2005;Goinetal.2009,
2016,in press). Polydolopimorphians are usually described
as having a diprotodont lower dentition, sometimes referred
to as Bpseudodiprotodont^on the assumption that it is non-
homologous with that of diprotodontians (Ride 1962,1964;
Goin 2003;Goinetal.2009). Most polydolopimorphians ex-
hibit a relatively low-crowned, bunodont molar morphology
(Marshall 1982a;Goin2003; Chornogubsky 2010;Goinetal.
in press), but the Oligocene-Pliocene argyrolagoids include
forms with hypsodont and hypselodont molars (Simpson
1970b; Hoffstetter and Villarroel 1974; Pascual and Carlini
1987; Villarroel and Marshall 1988; Sánchez-Villagra and
Kay 1997; Flynn and Wyss 1999; Sánchez-Villagra et al.
2000;Goinetal.2010,in press; Zimicz 2011).
Given current definitions of the order, the South American
fossil record of Polydolopimorphia spans from the Paleocene
to the Pliocene (Goin et al. 2016,in press).
Polydolopimorphians are also known from the middle
Eocene La Meseta Fauna from Seymour Island, off the
Antarctic Peninsula (Woodburne and Zinsmeister 1982,
1984;Goinetal.1999;Chornogubskyetal.2009). Possible
polydolopimorphians have been described from the Late
Cretaceous of North America (Case et al. 2005)andthe
Cenozoic of Australia (Beck et al. 2008a;Sigéetal.2009),
but these more questionable records are based on very frag-
mentary dental evidence; their similarities may simply reflect
convergent evolution of a bunodont molar morphology (Beck
et al. 2008a).
Recent works (e.g., Case et al. 2005; Goin et al. 2010,
2016,in press; Oliveira and Goin 2011; Chornogubsky and
Goin 2015) have recognized three suborders within
Polydolopimorphia (see Table 1): Hatcheriformes (which con-
tains the dentally most plesiomorphic forms);
Polydolopiformes (which includes Roberthoffstetteria
nationalgeographica from the early or middle Paleocene
Tiupampa locality in Bolivia, Sillustania quechuense from
the late Paleocene-early Eocene Chulpas locality in Peru,
and the diverse polydolopids) and Bonapartheriiformes.
Within Bonapartheriiformes, two superfamilies are currently
recognized: Bonapartherioidea and Argyrolagoidea (the latter
including the dentally highly derived groeberiids, patagoniids,
and argyrolagids; Goin et al. 2010,2016,in press; Zimicz
2011). Goin et al. (2016: table 5.1) considered Epidolops to
be a member of Bonapatherioidea, within which they recog-
nized four families: Prepidolopidae, Bonapartheriidae,
Gashterniidae, and Rosendolopidae. Goin et al. (2016:
table 5.1) placed Epidolops in Bonapartheriidae, but in its
own subfamily, namely Epidolopinae (see also Goin et al.
2003a,in press; Goin and Candela 2004;Caseetal.2005).
374 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Tab l e 1 Selected metatherian classifications, modified from Goin et al. (2009:table2). As presented here, the classifications other than those of Simpson (1945) and Ride (1964) are restricted to
metatherians from North and South America only, and taxonomic levels below the level of family are not shown, except that the bonapatheriid subfamilies Epidolopinae (which includes Epidolops)and
Bonapartheriinae are indicated for the classifications of Case et al. (2005)andGoinetal.(2016)
Simpson (1945)Ride(1964) Aplin & Archer (1987) Marshall et al. (1990)Szalay(1994) Kirsch et al. (1997)Caseetal.(2005)Goinetal.(2016)
Order Marsupialia Infraclass Metatheria Supercohort Marsupialia Infraclass Metatheria Infraclass Metatheria Infraclass Marsupialia Supercohort Marsupialia Infraclass Metatheria
Superfamily
Didelphoidea
Superorder
Marsupialia
Cohort Ameridelphia Supercohort Marsupialia Cohort Ameridelphia Supercohort
Boreometatheria
Cohort Alphadelphia Cohort BAmeridelphia^
Family Didelphidae Order
Marsupicarnivora
Order Didelphimorphia Cohort Alphaldelphia Order Didelphida Order Peradectimorphia Order Peradectia Family Pediomyidae
Family
Caroloameghiniidae
Superfamily
Didelphoidea
Family Didelphidae Order Peradectia Suborder
Archimetatheria
Family Peradectidae Family Peradectidae Family Pucadelphyidae
Superfamily
Borhyaenoidea
Family Didelphidae Family Sparassocynidae Cohort Ameridelphia Family Stagodontidae Supercohort
Notometatheria
Family Pediomyidae Family
Jaskhadelphyidae
Family
Borhyaenidae
Superfamily
Borhyaenoidea
Order Paucituberculata Order Didelphimorphia Family Pediomyidae Cohort Didelphidia Family Stagodontidae Family Mayulestidae
Superfamily
Dasyuroidea
Family Borhyaenidae Superfamily
Caroloameghinioidea
Superfamily
Didelphoidea
Suborder
Sudameridelphia
Order Didelphimorphia Family
Caroloameghiniidae
Family Protodidelphidae
Family Dasyuridae Superfamily
Dasyuroidea
Family
Caroloameghiniidae
Family Didelphidae Infraorder
Itaboraiformes
Family Didelphidae Cohort Ameridelphia Family Derorhynchidae
Family Notoryctidae Family Dasyuridae Superfamily
Caenolestoidea
Family Sparassocynidae Family
Caroloameghiniidae
Cohort Pseudiprotodontia Order Didelphimorphia Family Sternbergiidae
Superfamily
Perameloidea
Family Thylacinidae Family Caenolestidae Order
Polydolopimorphia
Infraorder
Polydolopimorphia
Order Paucituberculata Superfamily
Didelphoidea
Family Herpetotheriidae
Family Peramelidae Order
Paucituberculata
Superfamily
Argyrolagoidea
Superfamily
Polydolopoidea
Family Prepidolopidae Superfamily
Caenolestoidea
Family Pucadelphidae Order Sparassodonta
Superfamily
Caenolestoidea
Family Caenolestidae Family Gashterniidae Family Protodidelphidae Family Polydolopidae Family Caenolestidae Family BDidelphidae^Family Hondadelphidae
Family
Caenolestidae
Family Polydolopidae Family Groeberiidae Family Prepidolopidae Family
Bonapartheriidae
Superfamily
Argyrolagoidea
Family Caluromyidae Family Hathliacynidae
Family
Polydolopidae
Order Peramelina Family Argyrolagidae Family Bonapartheriidae Infraorder
Sparassodonta
Family Argyrolagidae Family Herpetotheriidae Superfamily
Borhyaenoidea
Superfamily
Phalangeroidea
Family Peramelidae Superfamily
Polydolopoidea
Family Polydolopidae Family Borhyaenidae Family Patagoniidae Family Derorhynchidae Family Borhyaenidae
Family
Phalangeridae
Order Diprotodonta Family Prepidolopidae Order Sparassodonta Family
Thylacosmilidae
Superfamily
Groeberioidea
Family Protodidelphidae Family Proborhyaenidae
Family
Thylacoleonidae
Family Phalangeridae Family Bonapartheriidae Superfamily
Borhyaenoidea
Suborder
Glirimetatheria
Family Groeberiidae Order Sparassodonta Family Thylacosmilidae
Family
Phascolomidae
Family Wynyardidae Family Polydolopidae Family Stagodontidae Infraorder
Paucituberculata
Order Polydolopimorphia Superfamily
Borhyaenoidea
Supercohort Marsupialia
Family
Macropodidae
Family Vombatidae Order Sparassodonta Family Hondadelphidae Family Caenolestidae Superfamily
Polydolopoidea
Family Mayulestidae Order Didelphimorphia
Family
Diprotodontidae
Family
Diprotodontidae
Family Borhyaenidae Family Hathliacynidae Infraorder
Simpsonitheria
Family Polydolopidae Family Hondadelphidae Superfamily
Peradectoidea
Family Macropodidae Family Thylacosmilidae Family Borhyaenidae Family Gashterniidae Family Prepidolopidae Family Hathliacynidae Family Peradectidae
Marsupialia incertae
sedis
Cohort Australidelphia Family Proborhyaenidae Family Groeberiidae Family Bonapartheriidae Family Borhyaenidae Family
Caroloameghiniidae
Family Notoryctidae Order Microbiotheria Family Thylacosmilidae Family Argyrolagidae Family Protodidelphidae Family Proborhyaenidae Superfamily
Didelphoidea
Family Microbiotheriidae Order Paucituberculata Family Patagoniidae Family Thylacosmilidae Family Didelphidae
J Mammal Evol (2017) 24:373–414 375
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Tab l e 1 (continued)
Simpson (1945)Ride(1964) Aplin & Archer (1987) Marshall et al. (1990)Szalay(1994) Kirsch et al. (1997)Caseetal.(2005)Goinetal.(2016)
Superfamily
Caroloameghinoidea
Superfamily
Caenolestoidea
Suborder
Didelphimorphia
Family
Caroloameghinidae
Order Paucituberculata Family Sparassocynidae
Family Kollpaniidae Family Didelphidae Cohort Eometatheria Cohort Australidelphia Order Paucituberculata
Family Caenolestidae Family
Sparassocynidae
Order Microbiotheria Order Microbiotheria Superfamily
Caenolestoidea
Family Palaeothentidae Cohort Australidelphia Family Microbiotheriidae Family
Microbiotheriidae
Family Caenolestidae
Family Abderitidae Order
Gondwanadelphia
Marsupialia incertae
sedis
Superfamily
Palaeothentoidea
Superfamily
Argyrolagoidea
Suborder
Microbiotheria
Order
Polydolopimorphia
Family Pichipilidae
Family Argyrolagidae Family
Microbiotheriidae
Suborder
Hatcheriformes
Family Palaeothentidae
Family Groeberiidae Suborder
Bonapartheriiformes
Family Abderitidae
Cohort Australidelphia Superfamily
Bonapartherioidea
Cohort Australidelphia
Order Microbiotheria Family Prepidolopidae Order Microbiotheria
Superfamily
Microbiotheroidea
Family Bonapartheriidae Family
Woodburnodontidae
Family Pediomyidae Subfamily
Bonapartheriinae
Family
Microbiotheriidae
Family
Microbiotheriidae
Subfamily Epidolopinae Order
Polydolopimorphia
Family Gashterniidae Family Glasbiidae
Superfamily
Argyrolagoidea
Suborder
Bonapartheriiformes
Family Groeberiidae Superfamily
Bonapartherioidea
Family Patagoniidae Family Prepidolopidae
Family Bonapartheriidae
Subfamily
Bonapartheriinae
Subfamily Epidolopinae
Family Gashterniidae
Family Rosendolopidae
Superfamily
Argyrolagoidea
Family Groeberiidae
Family Patagoniidae
Family Argyrolagidae
Suborder
Polydolopiformes
Family Polydolopidae
376 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Goin et al.‘s(2016) classification of Polydolopimorphia
received partial support from the phylogenetic analyses of
Goin et al. (2009) and Chornogubsky and Goin (2015): in
both analyses, clades equivalent to Polydolopiformes,
Bonapartheriiformes, and Argyrolagoidea were recovered.
The unpublished phylogenetic analyses of Chornogubsky
(2010), meanwhile, recovered clades equivalent to
Polydolopiformes and Bonapartheriiformes, but these analy-
ses were focused on relationships within Polydolopidae and
included only four non-polydolopid polydolopimorphians
(Epidolops,Bonapartherium,Prepidolops,and
Roberthoffstetteria). Ultimately, taxon sampling in these and
other analyses (e.g., Goin et al. 2006;OliveiraandGoin2011;
Forasiepi et al. 2013) is too limited to adequately test relation-
ships within Polydolopimorphia.
The relationship of polydolopimorphians to other
marsupialiforms has proved difficult to resolve (summarized
in Table 1). Most early studies argued for a close relationship
between Polydolopimorphia and the South American order
Paucituberculata (which includes the living caenolestid
Bshrew opossums^), largely based on the shared presence of
diprotodonty (Gregory 1910;Simpson1928,1945,1948;
Paula Couto 1952c). In a major review of BPolydolopidae^
(= polydolopids and Epidolops), Marshall (1982a) concluded
that the enlarged anterior Bgliriform^tooth of the lower jaw of
polydolopids is probably the canine. If so, diprotodonty
must have arisen independently in polydolopids and
paucituberculatans, because the paucituberculatan gliriform
tooth is unequivocally an incisor (Ride 1962; Abello 2013).
Some authors that accepted Marshall’s(1982a) conclusion
that the polydolopid gliriform tooth is the lower canine nev-
ertheless continued to link polydolopimorphians with
paucituberculatans (e.g., Aplin and Archer 1987; Marshall
1987; Kirsch et al. 1997). Kirsch et al. (1997) named the
grouping of Polydolopimorphia and Paucituberculata as the
cohort Pseudiprotodontia. The classifications of Aplin and
Archer (1987), Marshall (1987), and Kirsch et al. (1997)did
not group argyrolagoids or gashterniids with other
polydolopimorphians, with Kirsch et al. (1997) instead plac-
ing them within Paucituberculata.
The classifications of Marshall et al. (1990), Szalay
(1994), and Case et al. (2005), by contrast, did not endorse
a specific relationship between Polydolopimorphia and
Paucituberculata. In Marshall et al.‘s(1990:fig.2)phylog-
eny, Polydolopimorphia is sister to Didelphimorphia
(which includes living didelphid opossums), whilst
Paucituberculata is sister to Sparassodonta (an extinct order
of South American carnivorous marsupialiforms), with
these four orders collectively forming a clade. Marshall
et al. (1990) referred to this clade as Ameridelphia, which
is a name originally proposed by Szalay (1982) to refer to
non-australidelphian marsupialiforms. Marshall et al.
(1990) placed the argyrolagoid families Groeberiidae and
Argyrolagidae within Paucituberculata, rather than
Polydolopimorphia.
Szalay (1994) classified Polydolopimorphia as an infraor-
der in his suborder Sudameridelphia, and recognized
Paucituberculata as an infraorder within a different suborder,
Glirimetatheria. Szalay (1994) also erected the infraorder
Simpsonitheria for the argyolagoid families Groeberiidae,
Argyrolagidae, and Patagoniidae, plus Gashterniidae. Szalay
(1994) placed Simpsonitheria together with Paucituberculata,
in Glirimetatheria. Finally, Case et al. (2005) classified
Polydolopimorphia (including argyrolagoids and
gashterniids) as BMarsupialia^(= Marsupialiformes here)
incertae sedis, but placed Paucituberculata together with
Didelphimorphia and Sparassodonta in Ameridelphia.
In several papers (Goin et al. 1998b,2009,2016,in press;
Goin 2003; Goin and Candela 2004; Oliveira and Goin, 2006,
2011; Chornogubsky and Goin 2015), Goin and co-
authors have proposed a very different hypothesis of
polydolopimorphian relationships. Specifically, they have ar-
gued that Polydolopimorphia is closely related to the order
Microbiotheria, which is known from South America (includ-
ing the extant Dromiciops gliroides) and the middle Eocene of
Seymour Island off the Antarctic Peninsula, and the
Australian order Diprotodontia, which includes the koala,
wombats, Bpossums,^kangaroos, and a range of extinct
forms. Recently, isolated tarsals from the early-middle
Eocene (Lutetian) La Barda locality in Patagonia have also
been identified as representing a probable diprotodontian
(Lorente et al. 2016); if so, this is the first South American
record of Diprotodontia. If polydolopimorphians are close rel-
atives of microbiotherians and diprotodontians, it would mean
that they are also members of the trans-Gondwanan marsupial
superorder Australidelphia (Szalay 1982,1994; Beck et al.
2008b; Nilsson et al. 2010;Beck2012,in press-b), which in
turn would have significant implications for our understand-
ing of marsupialiform biogeography.
To date, hypotheses regarding polydolopimorphian affini-
ties have relied almost exclusively on dental features (Goin
2003;Goinetal.2006,2009; Oliveira and Goin 2011).
Cranial anatomy is obviously a key source of phylogenetic
(as well as functional) data within mammals, but few crania
of polydolopimorphians are known. Several relatively com-
plete crania of argyrolagids have been described (Simpson
1970b; Sánchez-Villagra and Kay 1997; Sánchez-Villagra
et al. 2000), but these represent relatively late (Oligocene or
younger), craniodentally apomorphic taxa. The groeberiid
Groeberia is also known from partial crania, but these are less
well preserved, and the known craniodental morphology of
this taxon is also highly apomorphic (Patterson 1952;
Simpson 1970a; Pascual et al. 1994). Among older
polydolopimorphians, the skull of Epidolops ameghinoi from
Itaboraí, DGM-321-M, is one of the best preserved, and is
therefore a critically important specimen. However, despite
J Mammal Evol (2017) 24:373–414 377
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

having been illustrated in several published works (Paula
Couto 1952c;Marshall1982a:figs.62–63; Bergqvist et al.
2008: fig. 9B), DGM 321-M has never been described in
detail.
In this paper, I provide the first detailed description of
the cranial morphology of Epidolops ameghinoi, based
largely on DGM 321-M but supplemented by information
provided by the additional specimens from Itaboraí. I do not
present a detailed description of the dentition, because this
has been well covered in previous publications (Marshall
1982a; Goin and Candela 1996; Zimicz 2014), but I
present a novel interpretation for the dental formula of
E. ameghinoi. I argue that the marsupialiform BType II^
petrosals described by Ladevèze (2004) plausibly belong
to E. ameghinoi, and I accept that the IMG VII tarsals re-
ferred to this taxon by Szalay (1994) are correctly attribut-
ed. I compare the morphology of Epidolops with that of
other taxa currently included in Polydolopimorphia. I qual-
itatively assess the available morphological evidence re-
garding the relationship of Polydolopimorphia to other
marsupialiforms. I discuss the implications of the new in-
formation presented here for our understanding of the af-
finities of argyrolagids and other argyrolagoids. As a
quantitative test of the position of polydolopimorphians
within Metatheria, I add Epidolops and the argyrolagids
Argyrolagus and Proargyrolagus to modified versions of
the total evidence matrix of Beck et al. (2014)andanalyze
them using a Bayesian undated approach. I conclude with a
discussion of the implications of this study for our under-
standing of marsupialiform biogeography, specifically re-
garding the origin and early evolution of Marsupialia.
Materials and Methods
Specimens
All specimens of Epidolops ameghinoi that I examined are
currently housed at the Museu de Ciências da Terra (prefix
DGM) in the Departamento Nacional de Produção Mineral,
and at the Museu Nacional do Rio de Janeiro (prefix MNRJ),
both in Rio de Janeiro. Comparative specimens of other taxa
examined in the course of this study are from the Department
of Mammalogy at the American Museum of Natural History
(prefix AMNH M-), the University of New South Wales (pre-
fix UNSW), the Museo Municipal de Ciencias Naturales
BLorenzo Scaglia,^Mar del Plata (prefix MMP), and the
Natural History Museum, London (prefix BMNH).
Anatomical Terminology and Abbreviations
Terminology for cranial anatomy follows Beck et al. (2014;
see also Wible 2003; Voss and Jansa 2009). Terminology and
abbreviations for the dental formula follow Voss and Jansa
(2009: table 7), in which the maximum metatherian dental
formula is assumed to be I1–5C1P1–3M1–4 in the upper
dentition and i1–4c1p1–3m1–4inthelowerdentition.
Recent papers by Goin and co-authors (e.g., Oliveira and
Goin 2011) have instead followed Hershkovitz (1982,1995)
in assuming that metatherians have lost the anteriormost lower
incisor, and so have referred to the lower incisors as i2–5(see
also Voss and Jansa 2009: table 7).
Assumed Classification
I tentatively follow Goin et al.’s(2016) classification of
Epidolops within Polydolopimorphia (see also Goin and
Candela 2004;Caseetal.2005;Goinetal.2010,in press).
However, the results of the current study cast doubt on wheth-
er argyrolagids (and possibly other argyrolagoids) are
polydolopimorphians; I believe it more likely that
argyrolagids are in fact members of Paucituberculata (see be-
low). I follow Sereno’s(2006: table 10.1) stem-based phylo-
genetic definition for Metatheria, namely the most inclusive
clade containing Didelphis marsupialis but not Mus musculus.
I restrict the name Marsupialia to the crown-clade only (see
Rougier et al. 1998; Flynn and Wyss 1999), and I use the
phylogenetic definition of Beck et al. (2014: 131), namely
the least inclusive clade containing Didelphis marsupialis,
Caenolestes fuliginosus,andPhalanger orientalis. Vullo
et al. (2009) proposed the name Marsupialiformes for the
clade corresponding to Btraditional,^more inclusive defini-
tions of Marsupialia (e.g., Kielan-Jaworowska et al. 2004); I
follow Beck’s(in press-a: Table 1) definition of
Marsupialiformes here, namely the most inclusive clade con-
taining Didelphis marsupialis but not Deltatheridium
pretrituberculare.
Regression Analysis of Petrosal Size
Ladevèze and Muizon (2010) ruled out referral any of the
eight marsupialiform petrosal morphotypes (Types I-VIII) de-
scribed from Itaboraí to E. ameghinoi based on incompatibilty
in relative size. However, Ladevèze and Muizon (2010)based
this conclusion on regressions of molar area (for M2, M3, m2,
and m3) against promontorium area, and Szalay (1994:
Tab le 6.3) r emark ed that E. ameghinoi has Brelatively small
molars [that] are unlikely to reflect body size accurately.^As
an alternative approach, I regressed promontorium area
against total cranial length for the set of 12 extant and fossil
marsupialiform taxa used by Ladevèze and Muizon
(2010: table 2). I then plotted estimated skull length for
E. ameghinoi (55 mm –see below) and promontorium area
for the eight Itaboraí petrosal morphotypes to see if any of the
eight morphotypes is an appropriate size for referral to
E. ameghinoi. Following Beck (2012), all measurements were
378 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

log
10
-transformed prior to analysis, and reduced major axis
regression was used (as implemented by the R package
smatr; Warton et al. 2012; R Development Core Team
2016). Measurements and sources for these are given in the
Electronic Supplementary Material.
Phylogenetic Analysis
As a test of the evolutionary relationships of Epidolops and
argyrolagids, I carried out a phylogenetic analysis using mod-
ified versions of the total evidence matrix of Beck et al.
(2014). This matrix comprises DNA sequence data from five
nuclear protein-coding genes (APOB,BRCA1,IRBP,RAG1,
and VWF), plus indels in the sequence data, retroposon inser-
tions, and morphological characters (see Beck et al. 2014 for
full details). This dataset was enlarged by adding 20
retroposon insertion characters taken from Gallus et al.
(2015), and 15 novel morphological characters. Epidolops
and the argyrolagids Proargyrolagus and Argyrolagus were
then added to this expanded matrix.
I produced two versions of the matrix: in the first (BMatrix
A^), I scored Epidolops based solely on DGM 321-M and
other isolated craniodental specimens from Itaboraí that could
be unequivocally identified as belonging to this taxon based
on dental morphology; in the second (BMatrix B″), I assumed
that the Type II petrosal morphotype described by Ladevèze
(2004) and the IMG VII tarsal morphotype described by
Szalay (1994) also represent Epidolops, and used these addi-
tional specimens for scoring purposes (see below). Scores for
Epidolops were based on firsthand observation of craniodental
specimens in the DGM and MNRJ collections, plus the
descriptions of the Type II petrosals by Ladevèze (2004)and
the IMG VII tarsals by Szalay (1994). Scores for Argyrolagus
were taken from Simpson (1970b), whilst those for
Proargyrolagus were taken from Sánchez-Villagra and Kay
(1997), Sánchez-Villagra et al. (2000), and Sánchez-Villagra
(2001). A full list of the morphological characters and scorings
for Epidolops,Argyrolagus,andProargyrolagus is given in
Electronic Supplementary Material. The full morphological and
total evidence matrices can be downloaded from Morphobank
(http://www.morphobank.org, Project 2436).
The complete total evidence matrix was analyzed using a
Bayesian non-clock approach in MrBayes 3.2.6, following
Beck et al. (2014). As in Beck et al. (2014), an eight partition
scheme was used for the DNA sequence data, and the nuclear
indel and retroposon insertion partitions were assigned sepa-
rate restriction site (binary) models, with the assumption
that only variable characters were coded. For the morpholog-
ical partition, an Mk model was specified; because
autapomorphies were present, I specified that variable charac-
ters were scored (Bcoding = var^). As in Beck et al. (2014), a
gamma distribution with four rate categories was used to to
model rate heterogeneity between morphological characters.
The MrBayes 3.2.6 analysis comprised two independent
runs of four chains (three Bheated,^one Bcold^), running for
50 × 10
6
generations and sampling trees every 2000 genera-
tions. The temperature of the heated chains was decreased
from 0.2 to 0.1. An average standard deviation of split fre-
quencies of 0.01–0.02 indicated that the chains had con-
verged. The first 25 % were discarded as burn-in. A minimum
ESS of >500 and PSRF of 1.00 for all parameters confirmed
that stationarity was reached among the post-burn-in trees, as
also indicated by plots of log likelihood against generation
number. 50 % majority rule consensus was used to summarize
the post-burn-in trees, with Bayesian posterior probabilities
(BPPs) calculated as support values.
SYSTEMATIC PALEONTOLOGY
METATHERIA HUXLEY, 1880 (SENSU SERENO, 2006)
MARSUPIALIFORMES VULLO ET AL., 2009 (SENSU
BECK, IN PRESS-A)
POLYDOLOPIMORPHIA ARCHER, 1984
BONAPARTHERIIFORMES PASCUAL, 1980
BONAPARTHERIOIDEA PASCUAL, 1980
BONAPARTHERIIDAE PASCUAL, 1980
EPIDOLOPINAE PASCUAL AND BOND, 1981
EPIDOLOPS PAULA COUTO, 1952
EPIDOLOPS AMEGHINOI PAULA COUTO, 1952
Diagnosis
Marshall (1982: 73–74) presented a detailed but non-
differential diagnosis for the subfamily Epidolopinae, which
was based on E. ameghinoi only, but a differential diagnosis is
presented here.
Medium-sizedmarsupialiform(estimatedbodymass~400g;
Zimicz 2014) with probable dental formula I1–3/i1–3C1/c1
P1–3/p1–3M1–4/m1–4. Differs from most marsupialiforms in
the combined presence of diprotodonty, enormous and
plagiaulacoid P3 and p3, and bunodont molars. Differs from
paucituberculatans in that its i2–3 and c1 are well developed
and procumbent (rather than reduced and single-rooted or ab-
sent), a large diastema is present behind c1, P3 and p3 are
enormous and plagiaulacoid, maxillopalatine fenestrae are
very small, and an alisphenoid tympanic process is absent.
Differs from diprotodontians in that its i2–3 and c1 are well
developed and procumbent (rather than reduced and single-
rooted or absent), the floor of its hypotympanic sinus is
unossified, its glenoid fossa is simple and planar (rather than
complex, with a separate articular eminence and mandibular
fossa), and its postglenoid foramen is posterior to the
postglenoid process (rather than shifted medially).
Among taxa currently included by Goin et al. (2016)in
Polydolopimorphia, E. ameghinoi: differs from hatcheriforms
J Mammal Evol (2017) 24:373–414 379
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

in that stylar cusp C is absent, stylar cusp B and stylar cusp D
are positioned relatively closer to the paracone and metacone,
respectively, and the paracone and metacone are connected by
weak lophs to the protocone and metaconular hypocone, respec-
tively; differs from polydolopiforms in having better developed
molar crests and in lacking a well-developed paraconule and
well-developed supernumerary cusps; differs from rosendolopids
in that stylar cusp B and stylar cusp D are positioned relatively
closer to the paracone and metacone, respectively, and the
paracone and metacone are connected by weak lophs to the
protocone and metaconular hypocone, respectively; differs from
prepidolopids in that P3 and p3 are plagiaulacoid (with a distinct
serrated edge), in having a more procumbent anterior dentition,
in having stylar cusp B and stylar cusp D positioned relatively
closer to the paracone and metacone, respectively, and in having
paracone and metacone connected by weak lophs to the
protocone and metaconular hypocone, respectively; differs from
Gashternia in that its P3 has many more cuspules forming a
serrated edge and lacks a lingual shelf; differs from
Bonapartherium in that P2 is much smaller and single-rooted,
P3 is plagiaulacoid and lacks a lingual platform, stylar cusp B
and stylar cusp D are positioned relatively further from the
paracone and metacone, respectively, and the lower incisors are
enlarged and procumbent; differs from Patagonia in that its i1 is
more procumbent, its i2–3 and c1 are well developed and pro-
cumbent (rather than reduced and single-rooted or absent), its P3
and p3 are enormous and plagiaulacoid (rather than absent), its
molars are bunodont (rather than hypsodont), and a total of four
(rather than three) molars are present; differs from Groeberia in
that its i1 is more procumbent, its i2–3 and c1 are well developed
and procumbent (rather than reduced and single-rooted or ab-
sent), its P3 and p3 are enormous and plagiaulacoid (rather than
very reduced or absent), its molars are bunodont (rather than
hypsodont), its rostrum is relatively longer, it lacks a distinct
masseteric process, its maxillopalatine fenestrae are much small-
er, and its dentary is relatively longer and shallower and lacks a
medial platform; differs from Klohnia in that its i1 is more pro-
cumbent, its i2–3 and c1 are well developed and procumbent
(rather than reduced and single-rooted or absent), its molars are
bunodont (not hypsodont), and a total of four (rather than three)
molars are present; differs from argyrolagids in that its i1 is more
procumbent, its i2–3 and c1 are well developed and procumbent
(rather than reduced and single-rooted or absent), its molars are
bunodont (not hypsodont or hypselodont), a large diastema is
present behind c1, P3 and p3 are enormous and plagiaulacoid
(rather than small and hypsodont or hypselodont, or entirely
absent), maxillopalatine fenestrae are present but very small,
and an alisphenoid tympanic process is absent.
Distribution and Temporal Range
All known specimens of E. ameghinoi are from the Itaboraí
Basin, Rio de Janeiro State, southeastern Brazil (Paula Couto
1952c;Marshall1982a). Three distinct depositional phases,
S1–3, have been recognized at Itaboraí (Medeiros and
Bergqvist 1999; Bergqvist et al. 2008); all E. ameghinoi spec-
imens are reported as being from fissure fills (Paula Couto
1952c; Marshall 1982a), which formed during phase S2.
The absolute age of the Itaboraí fauna (Itaboraian SALMA)
remains somewhat uncertain and it seems likely that the fossil-
bearing deposits span a considerable age range (Gayet et al.
1991; Marshall et al. 1997;Rage1998;Pinheiroetal.2012).
An ankaramite flow at the northern border of the Itaboraí
Basin has a K/Ar date of 52.6 +/−2.4Ma(Riccominiand
Rodrigues-Francisco 1992) and may postdate phase S2
(Bergqvist et al. 2008). Another Itaboraian fauna is known
from the Las Flores Formation in central Patagonia (Goin
et al. 1997; Woodburne et al. 2014b); Ar/Ar dating of an
overlying tuff suggests a minimum age of 49.5 Ma for this
fauna (Woodburne et al. 2014b). Based largely on these two
radiometric dates, Woodburne et al. (2014b) suggested that the
Itaboraian spans ~53–50 Ma, and I tentatively follow this
here.
Notes
A smaller species, E. redondoi (estimated body mass 127 g;
Zimicz 2014: tabla 4), has been described from the Cerro
Redondo locality in Patagonia (Goin and Candela 1995).
Simpson (1935) identified 14 stratigraphic levels (a-n) at
Cerro Redondo and reported that mammals were found in
levels h and m. Simpson (1935) concluded that level h was
slightly older than the Carodnia Zone of the Peñas
Coloradas Formation, which is currently interpreted as
62 Ma old (Clyde et al. 2014; Woodburne et al. 2014b).
Woodburne et al. (2014a: fig. 6) proposed that the
stratigraphically higher level m at Cerro Redondo falls
within the Ernestokokenia Faunal Zone, which represents
the Riochican SALMA, currently estimated at ~49 Ma old
(Woodburne et al. 2014b). It is uncertain from which level
at Cerro Redondo the only known specimen of E. redondoi
was collected (Goin and Candela 1995); however, assum-
ing that is from one of the two levels known to bear mam-
mals, and that an age of 50–53 Ma for the Itaboraían is
accurate, it seems more likely that it is from level m and
henceis49Maold.
Gayetetal.(1991) mentioned the presence of Epidolops
sp. at Estancia Blanco Rancho, Santa Lucia Formation,
Bolivia. If Estancia Blanco Rancho is similar in age to a
much better known Santa Lucia Formation fauna,
Tiupampa, it is probably early or middle Paleocene in age
(Marshall et al. 1997; Woodburne et al. 2014b). However,
this would mean the Epidolops material from Estancia
Blanco Rancho is at least 9 Myr older than that from
Itaboraí; as such, this record should be treated with caution
pending a description of the relevant material.
380 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Two as-yet undescribed species of Epidolops listed by
Zimicz (2014: 108) are from the Itaboraian-aged Las
Flores Formation of Patagonia (F.J. Goin, pers. comm.).
Description
Overall Morphology of DGM-321 M
DGM-321 M comprises a cranium plus associated left and
right mandibles (Figs. 1-3,5,and6). A large fragment of what
may be fossilized bone is present in the same box; it is labelled
as 321-M, but it differs in color and texture from the other
material, and its shape does not correspond to any missing
region of the cranium of E. ameghinoi (it appears to be too
large and too thick to be part of the posterior cranial roof). I am
confident that this fragment does not pertain to E. ameghinoi -
indeed, it is not unambiguously mammalian - and I do not
discuss it further here.
The total preserved length of the cranium is approximately
51 mm, whilst maximum width across the zygomatic arches is
approximately 37 mm. Total intact length of the cranium was
probably approximately 55 mm. The cranium is crushed dorso-
ventrally, with the degree of crushing much greater posterior to
the rostrum (Fig. 2). The dorsal surface of the rostrum is largely
missing, with only fragments of the nasals remaining. The fron-
tals are largely intact, but the parietals are missing.
The posterior part of the cranium (basioccipital,
supraoccipital, paired exoccipitals, and paired parietals, plus the
interparietal, if the last bone was present) is not preserved. The
ventral floor of the anterior part of the braincase is visible in
dorsal view (Fig. 1): the foramen rotundum (the exit of the max-
illary branch of the trigeminal nerve; Fig. 1: fro) and the internal
opening of the carotid foramen (Fig. 1: cf) are visible on both the
left and right sides. Flynn and Wyss (2004) noted that loss of the
posterior braincase (as also seen in cranial specimens of
the polydolopimorphians Kramadolops mckennai and
Bonapartherium hinakusijum; Pascual 1981;FlynnandWyss
2004) resembles the damage produced by modern predatory
birds (such as owls) when feeding on mammals.
The auditory region of DGM-321 M is slightly more com-
plete on the left side (Figs. 3and 5), but neither petrosal is
preserved. The squamosal contribution to the sidewall of the
braincase is largely missing. The zygomatic arches are largely
complete but damaged, and the glenoid fossa and postglenoid
process are complete on both sides. Foramina for the postglenoid
venous system are preserved. The left and right mandibles are
largely complete, but the left and right incisor arrays, right c1,
and left and right m4 are missing or broken (Fig. 6).
Nasal
Both nasals are badly damaged and fragmentary, with the left
slightly more complete (Fig. 1: na). Anteriorly, fragments of
Fig. 1 Cranium of Epidolops ameghinoi (DGM 321-M - holotype) in
dorsal view. aphotograph; binterpretative drawing. Abbreviations:
cf = carotid foramen; end = endocranial cavity; fr = frontal; fr-
pa = reconstructed path of frontal-parietal suture; fro = foramen
rotundum; gpa = glenoid process of the alisphenoid; ju = jugal;
?lac = ?lacrimal; mx = maxilla; na = nasal; pdp = posterodosal process
of the premaxilla; pmx = premaxilla = pop = postorbital process;
sq = squamosal
J Mammal Evol (2017) 24:373–414 381
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

left and right nasals are preserved in contact with the facial
processes of the maxillae, where they form the roof of the
nasal cavity. More posteriorly, crushing of the skull means
that the nasals are poorly preserved, but the posterior contact
Fig. 2 Cranium of Epidolops ameghinoi (DGM 321-M - holotype) in
lateral view. aphotograph of left lateral view; binterpretative drawing of
left lateral view; cphotograph of right lateral view; dinterpretative
drawing of right lateral view. Abbreviations: C1a = upper canine
alveolus; fr = frontal; gf = glenoid fossa; I1a = first upper incisor
alveolus; I2a = second upper incisor alveolus; I3a = third upper incisor
alveolus; iof = position of infraorbital foramen; ju = jugal;
?lac = ?lacrimal; M1 = first upper molar; M2 = second upper molar;
M3a = third upper molar alveoli; M4a = fourth upper molar alveolus;
mx = maxilla; na = nasal; pdp = posterodosal process of the premaxilla;
P2 = second upper premolar; P3 = third upper premolar;
pgp = postglenoid process; pmx = premaxilla; sgf = supraglenoid
foramina; sma = sulcus for masseter muscle; sq = squamosal
Fig. 3 Cranium of Epidolops ameghinoi (DGM 321-M - holotype) in
ventral view. aphotograph; binterpretative drawing. Abbreviations:
?apf = ?accessory palatal foramen; appf = accessory posterolateral
palatal foramen; C1a = upper canine alveolus; cf = carotid foramen;
ecpc = ectopterygoid crest; enpc = entopterygoid crest; gf = glenoid
fossa; gpa = glenoid process of the alisphenoid; hs = hypotympanic
sinus; I1a = first upper incisor alveolus; I2a = second upper incisor
alveolus; I3a = third upper incisor alveolus; if = incisive foramen;
ju = jugal; M3a = third upper molar alveoli; M4a = fourth upper molar
alveolus; ?mpf = ?major palatine foramen; mls = midline suture;
mx = maxilla; pal = palatine; ?pc = ?pterygoid canal; pf = pterygoid
fossa; pgf = postglenoid foramen; pgp = postglenoid process;
plpf = posterolateral palatal foramen; pmx = premaxilla =
ppt = postpalatine torus; prgp = preglenoid process; sph = sphenoid
complex; sq = squamosal
382 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

with the frontals is intact. The posterior margins of the nasals
form a gentle convex curve. The nasals terminate well poste-
rior to the anterior margin of the orbit. It is unclear exactly
how far the nasals extend laterally, but, based on the right side
of the cranium, it seems likely that the maxilla and frontal
(rather than the nasal and lacrimal) were in contact.
Antorbital vacuities (a highly distinctive feature of living
caenolestids; Osgood 1921,1924; Patterson and Gallardo
1987; Voss and Jansa 2009: 29; Ojala-Barbour et al. 2013)
appear to be absent.
Premaxilla
The anterior and dorsal parts of the premaxillae are dam-
aged, particularly on the right side (Figs. 1-3: pmx). In
lateral view (Fig. 2), the suture with the maxilla can be
identified, extending posterodorsally from level with the
anterior margin of the canine alveolus. A distinct
posterodorsal process (sensu Wible 2003) extends poste-
riorly to a point approximately level with the middle of
the diastema between C1 and P2 (Figs. 1and 2:pdp).The
incisive foramina (Fig. 3: if) are crushed, obscuring their
exact morphology, but they are short, with their posterior
margins approximately level with the posterior half of the
C1 alveolus. The posterior borders of the incisive foram-
ina are formed by the maxillae.
The left and right premaxillae both preserve evidence of
four closely-packed alveoli (Figs. 2and 3). Posteriorly, the
premaxilla forms the anterior margin of a large alveolus for
a large, single-rooted C1, with the remainder formed by the
maxilla (Figs. 2and 3: C1a). Immediately anterior to this,
three alveoli are present, which presumably housed three
single-rooted incisors. The posteriormost alveolus (Fig. 3:
I3a) is roughly circular, whereas the middle alveolus
(Fig. 3: I2a) is somewhat rectangular, being slightly longer
mesiodistally than labiolingually. Only the posterior part of
the anteriormost alveolus (Fig. 3:I1a)ispreserved,butthe
incisor it housed was probably the largest of the three. The
anterior end of the premaxilla is not preserved, and hence
the presence of one or two additional anterior incisors (as-
suming a maximum incisor count of five) cannot be entire-
ly ruled out. However, in metatherians with four or five
upper incisors, I2–4 are usually similar in size (pers.
obs.), and thus the large size of the anteriormost alveolus
suggests that no more than three incisors were present. On
the (admittedly questionable) assumption that teeth are lost
from the posterior end of the incisor array (Ziegler 1971), I
tentatively identify them as I1–3. If only three incisors are
present, then I1 must have been set back posteriorly some-
what from the anterior end of the premaxilla. The I1 alve-
olus appears relatively shallow, and hence this tooth is
unlikely to have been open-rooted.
Maxilla
The exact position of the infraborbital foramen cannot be
determined in DGM 321-M, but its rough location can be
inferred on the right side (Fig. 2c and d: iof), directly above
P2 and well anterior of the suture with the jugal. This
interpretation is confirmed by some isolated maxillary
fragments that preserve the infraorbital foramen in this
position (e.g., DGM 198-M, 204-M, 913-M). In other
maxillary specimens, the infraorbital foramen is level with
the anterior margin of P3 (e.g., DGM 201-M, 205-M). The
suture with the jugal is essentially straight, but a slight saw-
edge is visible on the right side of DGM 321-M (Fig. 2cand
d). There is no antorbital fossa. The maxilla does not form a
distinct masseteric process at the base of the zygomatic arch in
DGM 321-M (Figs. 2and 3). Other specimens of
E. ameghinoi preserve a very weakly raised area on the
maxilla corresponding to the likely area of origin of the
superficial masseter (e.g., DGM 898-M; Fig. 4:osm),butin
none of these can this structure be reasonably described as
forming a distinct masseteric process.
In dorsal view (Fig. 1), the anterior root of the zygomatic
arch is anteroposteriorly elongate (as noted by Flynn and
Wyss 2004: 88); this is somewhat exaggerated in DGM 321-
M due to dorsoventral crushing, but isolated maxillary frag-
ments (e.g., DGM 898-M; Fig. 4)showthatthismorphology
is not entirely artefactual. Crushing means that the exact
Fig. 4 Isolated partial right maxilla of Epidolops ameghinoi (DGM 898-
M) in ventral view, with inferred extent of maxillopalatine fenestra
indicated. Abbreviations: C1a = upper canine alveolus; if = incisive
foramen; M4a = upper fourth molar alveolus; mpf = maxillopalatine
fenestra; osm = area of origin of superficial masseter; P1a = upper first
premolar alveolus; P3a = upper third premolar alveoli
J Mammal Evol (2017) 24:373–414 383
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

contribution of the maxilla to the orbital fossa, and its
relationships to the other bones in this region, is unclear in
DGM 321-M. However, a robust zygomatic process of the
maxilla extends posterolaterally on the inside of the
zygomatic arch of the jugal (Fig. 1: zpm). Isolated maxillae
(DGM 205-M and 898-M) show that the maxillary foramen
was completed dorsally by the lacrimal, rather than being
entirely enclosed by the maxilla. DGM 898-M also indicates
that the exposure of the maxilla within the orbital fossa was
relatively small, and hence the palatine and lacrimal were
probably in contact.
In ventral view (Fig. 3), the maxilla forms the majority of
the palate, from its anterior contact with the premaxilla (where
it forms the posterior borders of the incisive foramina, level
with C1) posteriorly, with the palatine forming the
posteromedial section. The palate between P3-M4 is damaged
in its midline in DGM 321-M, with the right maxilla broken at
the labial roots of M1–2 and its palatal process displaced
dorsomedially (contra Paula Couto 1952c:fig.2).Asaresult,
it is difficult to determine whether palatal vacuities are present
or absent in DGM 321-M. However, DGM 898-M (Fig. 4)
and MNRJ 2879-V (both isolated maxillae) preserve the an-
terior margin of a small palatal vacuity, which extends anteri-
orly to approximately level with the bony septum between the
P3 and M1 alveoli. Further posteriorly, the path of the
maxillopalatine suture can be traced in DGM 898-M
(Fig. 4), suggesting that the vacuity was enclosed posteriorly
by the palatine and hence that it is a maxillopalatine fenestra
sensu Voss and Jansa (2009).
The exact size of the intact maxillopalatine fenestrae in
E. ameghinoi cannot be determined, but they appear to have
been very short anteroposteriorly, probably only extending
posteriorly as far as M2 (Fig. 4: mpf). The palatal suture be-
tween the maxilla and the palatine is somewhat difficult to
identify in DGM 321-M due to the presence of obscuring
adhesive, but it appears to be complex and interdigitating me-
dial to M2–3(Fig.3; see also Paula Couto 1952c:fig.2:left
[anatomical right] side). The maxilla forms the lateral border
of the posterolateral palatal foramen sensu Voss and Jansa
(2009 = minor palatine foramen sensu Wible 2003), with the
maxillopalatine suture passing through this foramen (Fig. 3:
plpf).
The maxilla preserves alveoli for at least seven teeth
(Figs. 2and 3). At its anterior end, it clearly formed the ma-
jority of the alveolus for the large, single-rooted C1 (Figs. 2
and 3: C1a); the premaxilla seems to have formed the anterior
margin of the C1 alveolus in DGM 321-M (Figs. 2and 3:C1a)
and also in DGM 898-M (Fig. 4: C1a) and 917-M, but this
alveolus is entirely within the maxilla in MNRJ 2879-V (poly-
morphism in this feature occurs in a few living marsupials,
namely the caenolestid Lestoros inca and several
peramelemorphians; pers. obs.). The C1 is unknown in
E. ameghinoi. However, based on the size and position of its
alveolus, it was probably similar in morphology to the C1 of
Bonapartherium hinakusijum (see Pascual 1981: figs. 1-3),
which is large and possibly also somewhat procumbent (if this
is not an artefact of the dorsoventral crushing of the best pre-
served B. hinakusijum cranium, MMP 1408).
There is a large diastema separating C1 from P2; this region
is damaged on both left and right sides of DGM 321-M
(Figs. 2and 3), and so the presence of P1 cannot be ruled
out. In fact, DGM 898-M (Fig. 4: P1a), 917-M, and MNRJ
2879-Vall indicate the presence of a very small, single-rooted
P1 ~ 1 mm behind the posterior margin of the C1 alveolus.
DGM 917-M preserves the root and base of the crown of P1,
demonstrating that this tooth was slightly procumbent.
Marshall (1982a) stated that P1 was sometimes absent in
E. ameghinoi, but all three specimens in which the region of
the maxilla immediately posterior to C1 is well preserved
(DGM 898-M, 917-M, and MNRJ 2879-V) have a P1.
Marshall (1982a) also reported that P1 is double-rooted in
MNRJ 2879-V, but I interpret the posterior Balveolus^in this
specimen as an artefact due to damage.
P2 is a very small, button-like tooth located at the base of
the enormous, plagiaulacoid P3. It is single-rooted in DGM
321-M (Fig. 3: P2) and in several other specimens (DGM
912-M, 918-M), but double-rooted (MNRJ 2879-V; DGM
898-M –Fig. 4: P2) or incipiently double-rooted (DGM
917-M) in others. The relative sizes and arrangement of P2
andP3inE. ameghinoi are strongly reminiscent of the con-
dition seen in the living Australian diprotodontian Burramys
parvus (the mountain pygmy possum; see Ride 1956). DGM
898-M reveals the root morphology of P3 (Fig. 4: P3a): the
anterior alveolus is single, whereas the posterior alveolus is
incipiently divided and so has a mediolaterally-oriented
figure-of-8 shape in dorsal view; however, the roots within
the posterior alveolus are fully divided in DGM 898-M, i.e.,
there are two posterior roots, one posterolabial and one
posterolingual. The maxilla flares distinctly laterally where
it houses the P3, and hence the skull broadens markedly at
this point (Figs. 1,3and 4); this is in contrast to the rostrum,
which is relatively constant in width (Figs. 1and 3).
Posteromedial to P3, M1–3 are each housed in three alveoli:
two small roots on the labial side, and a single, broader root
on the lingual half (Figs. 3and 4). An alveolus for a small,
single-rooted M4 is present posterior to the lingual root of M3
(Figs 3and 4: M4a). The molar row is oriented roughly
anteroposteriorly, but is positioned distinctly medial to P3:
the labial roots of M1 are posterior to lingual root of P3
(Figs. 3and 4).
In dorsal view, the exact relationship between the maxilla
and the other bones forming the roof of the anterior region of
the cranium is unclear due to crushing and displacement
(Fig. 1); however, based on the right side, it seems likely that
the maxilla and frontal were in contact (as in most marsupials),
rather than the nasal and lacrimal.
384 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Lacrimal
A few fragments of lacrimal may be preserved on both sides of
DGM 321-M (Figs. 1and 2: ?lac). These possible remnants do
not give any indication of (for example) the presence or ab-
sence of a distinct orbital crest, or the number and arrangement
of the lacrimal foramina. However, the facial exposure of the
lacrimal appears to have been relatively small, and the right
side of DGM 321-M suggests that the maxilla and frontal were
in contact, rather than the nasal and lacrimal (Fig. 1).
The arrangement of bones in the orbital mosaic is also
unclear in DGM 321-M. However, isolated maxillae (DGM
205-M, 898-M) indicate that the maxillary foramen was com-
pleted dorsally by the lacrimal, and suggest that the lacrimal
andpalatinewereprobablyincontact.
Palatine
In ventral view (Fig. 3), the palatine forms the posteromedial
section of the hard palate, contacting the maxilla along a com-
plex, interdigitating suture (see also Paula Couto 1952c:
fig. 2). Large palatine fenestrae sensu Voss and Jansa (2009)
are absent, but a distinct foramen is visible within the palatal
process of the palatine (most obviously on the left side), me-
dial to M3 (Fig. 3: ?mpf). A much smaller foramen appears to
be present lateral to this, close to or within the suture with the
maxilla Fig. 3: ?apf). The larger foramen is plausibly the major
palatine foramen for the major palatine artery, vein, and nerve
(if these did not pass through the maxillopalatine fenestrae,
which are located further anteriorly). The smaller foramen
may be an accessory palatine foramen, which transmits
branches of the accessory palatine nerve and artery (Wible
and Rougier 2000; Wible 2003).Thepalatineformsadistinct,
raised postpalatal torus (Fig. 3: ppt), and also forms the pos-
terior and medial borders of the posterolateral palatal foramen
(which is completed anteriorly and laterally by the maxilla;
Fig. 3: plpf). An accessory posterolateral palatal foramen is
also identifiable (Fig. 3: appf), extending anteroposteriorly
through the postpalatine torus, posteromedial to the postero-
lateral palatal foramen. The palate lateral to the posterolateral
palatal foramen does not form distinct Bcorners,^unlike the
condition in most didelphids (Voss and Jansa 2009,2003).
Posterior to the postpalatine torus, the palatines contribute
to the lateral walls of the nasopharyngeal region. The anterior
part of this region is badly damaged, but there is a faint mid-
line suture more posteriorly (Figs. 3and 5: mls); this probably
represents midline contact between either the palatines or the
pterygoids, but damage and obscuring glue mean that these
alternatives cannot be distinguished. Ride (1956) proposed
that midline contact between the palatines seen in
diprotodontians with very large plagiaulacoid P3s is an adap-
tation to strengthen the palatal region of the cranium; if so, this
may explain its possible presence in E. ameghinoi, which has
a similar P3 morphology.
In lateral view, crushing and general damage mean that the
exact contribution of the palatine to the orbital mosaic is un-
clear, as is the location and morphology of the sphenopalatine
foramen. Based on the left side, it seems likely that the pala-
tine prevented contact between the maxilla and alisphenoid.
Jugal
Isolated maxillary specimens demonstrate that the jugal did
not contribute to the slightly raised area for origin of the su-
perficial masseter (Figs. 2,3and 4: osm; see above). The jugal
is deep dorsoventrally, and together with the squamosal forms
a robust zygomatic arch (Figs. 1-3).
The concave dorsal margin of the jugal suggests that the
orbit was relatively large (Fig. 2). The anterior part of the
jugal is buttressed medially by a prominent zygomatic pro-
cess of the maxilla (Fig. 1: zpm). More posteriorly, there
does not appear to be a distinct frontal process marking the
attachment of the postorbital ligament on the dorsal margin
of the jugal (Fig. 2). The jugal extends under the squamosal
as far as the glenoid fossa, terminating in a ventrally deep
but mediolaterally narrow preglenoid process (Fig. 3:
prgp); this process is better preserved on the left side of
DGM 321-M than on the right. The posterior end of the
preglenoid process terminates in a distinct facet that is ori-
ented posterolateral to anteromedial. The lateral face of the
zygomatic process of the jugal is marked by a prominent
ventral sulcus for the masseter muscles (Fig. 2:sma),while
the medial face is strongly concave.
Frontal
In dorsal view, the postorbital process forms a gently-rounded
lateral protuberance (Fig. 1: pop). A relatively sharp postor-
bital constriction is present immediately posterior to the post-
orbital process. A faint temporal line can be traced
posteromedially back from the postorbital process, reaching
the midline ~3.5 mm anterior from the posterior edge of the
frontals. The median frontal suture is unfused. Either side of
the posterior end of the median suture, areas of the frontal that
were overlapped by the parietal when the skull was intact are
identifiable: the suture between the paired frontals and parie-
tals was evidently W-shaped in dorsal view, with the base of
the W oriented anteriorly (Fig. 1: fr-pa). The posterior end of
the median frontal suture is slightly raised, suggesting that a
sagittal crest may have been present on the parietals.
Squamosal
Parts of both the left and right squamosal are preserved in
DGM 321-M (Figs. 1-3and 5: sq), with the glenoid region
J Mammal Evol (2017) 24:373–414 385
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

largely intact on both sides (Figs. 3and 5). However, only part
of the squamosal contribution to the lateral braincase is pre-
served, and the region posterior to the postglenoid process
(including the part surrounding the external auditory meatus)
is missing. Inventral view, the glenoid fossa (Figs. 3and 5:gf)
is mediolaterally broad and gently concave, forming a
smoothly curved surface that extends posteroventrally onto
the anterior face of the postglenoid process (Figs. 2,3and 5:
pgp). There is no raised articular eminence anteriorly, whereas
this structure is found in most diprotodontians (Aplin 1987,
1990). The postglenoid process is broad mediolaterally and
low, with its anterior face slightly concave and its posterior
face slightly convex. The posteromedial edge of the
postglenoid process is grooved for the passage of the
postglenoid vein. The postglenoid foramen itself (Figs. 3
and 5: pgf) is located slightly more dorsal, namely medial
and slightly posterior to the postglenoid process. Although
the region is damaged on both sides of DGM 321-M, the
slightly better preserved left side suggests that the postglenoid
foramen was probably fully enclosed by squamosal. Two
supraglenoid foramina, visible on the left side of DGM 321-
M dorsal and slightly posterior to the postglenoid process in
lateral view (Fig. 2a and b: sgf), appear to be continuous with
the postglenoid foramen (confirmed by breakage on the right
side).
The contact between the squamosal and alisphenoid is not
obvious. On the right side, the part of the alisphenoid that
contacted the squamosal appears to have flaked away, but
the suture can still be traced, coursing posteromedially from
the anteromedial corner of the glenoid fossa. This morphology
is confirmed on the left side, in which the alisphenoid is more
intact, and which indicates that a distinct glenoid process of
the alisphenoid (= the entoglenoid process of the alisphenoid
sensu Muizon 1998,1999;Figs1,3and 5: gpa) was present,
extending along the anterior margin of the medial part of the
glenoid fossa. Based on the left side of DGM 321-M, it is
unlikely that the squamosal contributed to the roof of the
hypotympanic sinus.
In lateral view, the zygomatic process of the squamosal is
deep, and together with the underlapping jugal, forms a robust
zygomatic arch (Fig. 2).
In dorsal view (Fig. 1), the zygomatic process of the squa-
mosal forms a prominent ridge on its dorsal margin; posteri-
orly, where this ridge merges with the squamosal contribution
to the braincase, it forms the posterior and lateral wall of a
distinct, roughly triangular depression. This depression pro-
vides attachment for the temporalis, and a number of small
foramina are visible within it.
Pterygoid
Grooves that presumably housed the pterygoid are visible in
the sphenoid complex on the right side of DGM 321-M, but
the right pterygoid itself appears to be largely or entirely ab-
sent; based on the disposition of these grooves, it is unlikely
that the pterygoid extended posteriorly as far as the external
opening of the carotid foramen (Figs. 3and 5: cf). The ptery-
goid appears to be at least partially preserved on the left side,
where it contributes to the entopterygoid crest (Figs. 3and 5:
enpc). The pterygoid is damaged, but the preserved part does
not extend posteriorly as far as the carotid foramen. The pre-
cise extent of the pterygoids when intact cannot be unambig-
uously inferred in DGM 321-M, and so it is uncertain whether
the midline suture visible in the nasopharyngeal region
(Figs. 3and 5: mls) represents midline contact by the palatines
or by the pterygoids (see above).
Sphenoid Complex
The sphenoid complex (Figs. 3and 5:sph) comprises the
presphenoid, basisphenoid, and paired orbitosphenoids and
alisphenoids (Wible 2003). Distinct sutures between these
Fig. 5 Left basicranial region of Epidolops ameghinoi (DGM 321-M -
holotype) in ventral view. aphotograph; binterpretative drawing.
cf = carotid foramen; ecpc = ectopterygoid crest; enpc = entopterygoid
crest; gf = glenoid fossa; gpa = glenoid process of the alisphenoid;
hs = hypotympanic sinus; ju = jugal; mls = midline suture;
pf = pterygoid fossa; pgf = postglenoid foramen; pgp = postglenoid
process; prgp = preglenoid process; sph = sphenoid complex;
sq = squamosal
386 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

bones are not identifiable in DGM 321-M, and so this region
will be described as a whole. In lateral view, neither the
sphenorbital fissure nor the foramen rotundum (both of which
open within the sphenoid complex) are identifiable with any
certainty. On the left side, there is a piece of bone that is in
contact with the palatine anteriorly and the frontal dorsally;
this is presumably part of the alisphenoid. However, damage
to the region posterior to this means that the full extent of the
alisphenoid (and whether it was in contact with the parietal, or
whether instead the frontal and squamosal were in contact) is
unclear, as it is on the right side.
In ventral view, the sphenoid complex preserves some par-
ticularly significant features (Figs. 3and 5). Posteriorly, the
ventral part of the back of the cranium appears to have broken
away along the basisphenoid-basioccipital suture. Prominent
entopterygoid crests extend posteriorly (Figs. 3and 5:enpc);
on the left side, the pterygoid also contributes to this crest, but
this bone is not preserved on the right. Lateral to these crests
are well-excavated pterygoid fossae (Figs. 3and 5:pf),andon
the left side there is also evidence for an ectopterygoid crest
(Figs. 3and 5: ecpc) that encloses at least the anterior half of
the pterygoid fossa laterally, indicating that the pterygoid mus-
culature of E. ameghinoi was well developed. A tiny foramen
appears to be present within this fossa on both sides of DGM
321-M (Figs. 3and 5: ?tcf); it may be the transverse canal
foramen, but CT data will be required to confirm this. The
external opening of the carotid foramen is visible at the pos-
terior margin of the sphenoid complex on both sides of DGM
321-M (Figs. 3and 5: cf). It is located slightly further anterior
on the right side compared to the left, but slight bilateral asym-
metry in the position of this foramen is not uncommon among
marsupials (pers. obv.); alternatively, this might be a tapho-
nomic artefact.
A glenoid process of the alisphenoid (Figs. 3and 5:gpa)
extends along the anterior margin of the medial part of the
glenoid fossa, and more posteriorly the suture between the
alisphenoid and squamosal extends in a posteromedial direc-
tion. Medial to the posterior part of this suture, the part of the
alisphenoid forming the hypotympanic sinus appears to be
preserved on the left side of DGM 321-M, sloping dorsally
where it starts to form the anterolateral part of the roof of this
sinus (Figs. 3and 5: hs). This preserved part of the
hypotympanic sinus roof is not strongly excavated.
Apart from a slight rise medially, there is no evidence of an
alisphenoid tympanic process along the anterior border of the
putative hypotympanic sinus (Figs. 3and 5: hs). This apparent
absence may be an artefact due to the obscuring adhesive;
however, there is no sign of the broken base of an alisphenoid
tympanic process as is clearly visible in, for example, fossil
crania of the Australian marsupialiform Yalkaparidon coheni
(see Beck et al. 2014: figs. 2 and 8) or the peramelemorphian
Yarala burchfieldi (see Muirhead 2000: figs. 1 and 3). There
is also no tympanic process of the squamosal.
It is possible that another bone formed an ossified floor for
the anterior part of the hypotympanic sinus, but that this bone is
not preserved in DGM 321-M. For example, it is possible that
the petrosal enclosed the hypotympanic sinus in E. ameghinoi,
as it does in acrobatid diprotodontians (Aplin 1987,1990).
However, multiple isolated marsupialiform petrosals are known
from Itaboraí (including the Type II petrosals of Ladevèze
2004, which are plausibly referable to E. ameghinoi –see be-
low), and none preserve evidence of extensive tympanic pro-
cesses that could enclose the hypotympanic sinus (Ladevèze
2004,2007; Ladevèze and Muizon 2010). Furthermore, the
entire auditory region of acrobatids is highly autapomorphic
(Aplin 1987,1990), whereas the preserved morphology of this
region in DGM 321-M appears relatively plesiomorphic within
Marsupialiformes (see below). Alternatively, the hypotympanic
sinus could have been floored by one or more entotympanics;
however, entotympanics do not occur in any known
metatherian (with the possible exceptions of acrobatids and
some specimens of Phalanger orientalis; Maier 1989;Aplin
1990;Norris1993; Sánchez-Villagra 1998), and it seems un-
likely that E. ameghinoi was an exception to this general rule.
Instead, based on available evidence, I conclude that
E. ameghinoi probably lacked an ossified hypotympanic sinus
floor. A notch medial to the slightly raised area at the
anteromedial corner of this region may represent the anterior
margin of the foramen ovale.
In dorsal view, the sphenoid contribution to the endocranium
is visible; the internal openings of the carotid foramina (Figs. 3
and 5: cf) can be seen either side of the hypophyseal fossa, and
the left foramen rotundum (Figs. 3and 5: fro) is also identifi-
able. Flynn and Wyss (2004: 89) observed that the braincase of
E. ameghinoi appears to be proportionately much smaller than
that of Kramadolops mckennai, but it seems similar to
Bonapartherium hinakusijum in this regard (see Pascual
1981: figs. 1-3; Goin et al. 2016:fig.5.10).
Mandible
The left and right mandibles of DGM 321-M are preserved
largely intact (Fig. 6). They are joined by adhesive at the
symphysis (Fig. 6c: mas), but the symphysis is nevertheless
clearly unfused. Multiple additional dentaries of E. ameghinoi
are present in the DGM and MNRJ collections, and they have
been used to supplement the description here.
The mandibular ramus appears short and robust, and its
lateral wall bulges out below p3, due to the enlarged roots of
this tooth. Anterior to this bulge, the dorsal and ventral mar-
gins of the ramus slope distinctly dorsally. A large mental
foramen (Fig. 6a and b: mf) is present anteroventral to the
small p2, level with the exposed roots of p3. A distinct sulcus
on the dorsal surface of the mandibular ramus extends anteri-
orly from medial to the anterior root of p3. The symphysis
(Fig. 6c: mas) appears to have extended posteriorly as far as
J Mammal Evol (2017) 24:373–414 387
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

the vestigial p2; in some other specimens the symphysis is
posteriorly more extensive, reaching as far back as p3 in
DGM 903-M and 904-M. Posterior to the bulge formed by
p3, the lateral face of the mandibular ramus is slightly con-
cave. A second, smaller mental foramen (Fig. 6b: ?mf) ap-
pears to be present below the midpoint of m1 on the right
mandible, but this region is damaged on the left. Two mental
foramina are present in several other specimens, e.g., DGM
903-M and DGM 904-N; the position of these foramina varies
slightly between specimens, but the larger anterior foramen is
typically ventral or anteroventral to p2, and the smaller poste-
rior foramen is typically ventral or anteroventral to m1.
The coronoid process (Fig. 6a and b: cor) is tall, its anterior
margin rising at an angle of ~55° relative to the horizontal, lateral
to the m4. There is no large foramen (the Bretromolar canal^
sensu Hoffstetter and Villarroel 1974) within the retromolar
space behind m4, whereas such a canal is present in argyrolagids
(Simpson 1970b; Hoffstetter and Villarroel 1974; Sánchez-
Villagra et al. 2000; Voss and Jansa 2009: 46; Babot and
García-López 2016). A distinct foramen is also present in the
retromolar space of at least some caenolestids (Simpson 1970b;
Voss and Jansa 2009: 46), but Babot and García-López (2016)
argued that this structure is not homologous with the retromolar
canal of argyrolagids (see BAffinities of Argyrolagoids^below).
On the lateral face of the anterior margin of the coronoid
process, a thick coronoid crest (Fig. 6a and b: coc) is present,
continuing ventrally into the body of the mandibular ramus and
defining the anterior limit of the well-excavated masseteric fossa
(Fig. 6a and b: maf). In the midsection of the coronoid process, a
shallow sulcus is visible on the anterior face of the coronoid
crest. The posterior margin of the coronoid process is gently
concave and slightly more vertical than its anterior margin.
The coronoid process forms a blunt hook at its posterodorsal
extremity. The condylar process (Fig. 6: con) is mediolaterally
broad and roughly cylindrical. Posteroventrally, the angular pro-
cess (Fig. 6: ang) is strongly medially inflected, forming a prom-
inent medial platform, the posteromedial edge of which forms a
posterodorsally-oriented hook. The mandibular foramen is eas-
ily identifiable on the medial face of the right mandible of DGM
321-M, but this region is damaged on the left.
Laterally, multiple foramina (Fig. 6a: ?mafo) are present
within the masseteric fossa of at least some specimens (see
Abbie 1939). In DGM 903-M, the largest of these foramina is
on the anterior wall of masseteric fossa, and is concealed by
the coronoid crest in lateral view.
The ventral margin of the masseteric fossa is formed by a
prominent masseteric line; as this line extends posteriorly, it
becomes crestlike and particularly extensive laterally, forming
a distinct posterior shelf of the masseteric fossa (Fig. 6:psmf;
Marshall and Muizon 1995; Wible 2003).
Lower Dental Formula
The lower dentition of E. ameghinoi will be discussed starting
with the molars and then moving anteriorly, because it is the
antemolar formula that warrants the most detailed discussion.
DGM 321-M and other mandibular specimens preserve evi-
dence of four molars (Fig. 6c), with m1–3 doubled-rooted and
m4 single-rooted (Fig. 6c: m4a). The molars show a clear
decreasing gradient in size moving from m2 to 4, whereas
m1 and m2 are similar in size. The double-rooted p3 is by
far the largest tooth (Fig. 6: p3), and is preceded by a vestigial,
single-rooted p2 (Fig. 6: p2). Anterior to this is a prominent
diastema, ~4.5 mm long, between the p2 and the first procum-
bent tooth, which I interpret here as c1 (Fig. 6a and c: c1; see
below), as did Marshall (1982a). On the right side of DGM
321-M, there appears to be a very small alveolus (Fig. 6band
c: p1a), even smaller than that of p2, ~3 mm anterior to p2 and
Fig. 6 Left and right mandibles of Epidolops ameghinoi (DGM 321-M -
holotype). aleft mandible in lateral view; bright mandible in lateral view;
ca left and right mandibles in dorsal view. Abbreviations: ang = angular
process; c1 = lower canine; coc = coronoid crest; con = mandibular
condyle; cor = coronoid process; m4a = fourth lower molar alveolus;
maf = massteric foramen; mas = mandibular symphysis; mf = mental
foramen; ?mafo = ?massteric foramen; ?mf = ?mental foramen;
p1a = first lower premolar alveolus; p2 = second lower premolar;
p3 = third lower premolar; psmf = posterior shelf of the massteric fossa
388 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

~1.5 mm posterior to c1; this alveolus seems to be absent from
the left side of DGM 321-M. A groove extends anteriorly
from this alveolus, suggesting that, if it did house a tooth, then
it must have been small, single-rooted, and distinctly procum-
bent. Other mandibular specimens of E. ameghinoi preserve
this alveolus (e.g., DGM 899-M 901-M, 903-M, 908-M), sug-
gesting that it was normally present. A root is preserved within
the alveolus in DGM 901-M, confirming that a procumbent
tooth was indeed present. Although the crown of this tooth is
not preserved in any E. ameghinoi specimen, it was probably
very similar in morphology to the similarly-positioned tooth in
the polydolopid Kramadolops abanicoi that Flynn and Wyss
(1999: fig. 1) identified as p1, but which Goin et al. (2010:86)
referred to as c1 and Chornogubsky (2010) identified as ?c1. I
concur with Flynn and Wyss (1999) that this tooth is p1 in
K. abanicoi and also E. ameghinoi (see BComparisons with
Other Taxa Currently Included in Polydolopimorphia^below).
More anteriorly, there is a large, anteriorly-facing alveolus for
an elongate, procumbent tooth. This tooth is preserved on the left
side of DGM 321-M (Fig. 6a and c: c1); it is characterized by an
elongate dentine root, and enamel is restricted to the distinctly-
hooked tip. The enamel extends slightly further down the root
labially than lingually, but it still extends only approximately
3 mm on the labial side and approximately 2.7 mm on the lingual
side,comparedwithatotaltoothlengthofapproximately7mm.
The elongate, dentine root and small enamelled tip of this tooth is
strongly reminiscent of the canine morphology seen in older
individuals of many marsupial taxa (e.g., didelphids,
peramelemorphians, and dasyuromorphians), in which the ca-
nine root is extruded continuously throughout life and, as a result,
the enamel becomes increasingly restricted to the tip of the tooth
(Jones 1997: 2572; Jones and Stoddart 1998: 240; Voss and
Jansa 2009: 48; Aplin et al. 2010: 15). By contrast, in marsupials
that have an enlarged, procumbent anterior tooth in their lower
dentition that can be unambiguously identified as an incisor
(diprotodontians, paucituberculatans), the enamel extends far
down the root. When the dentaries of DGM 321-M are placed
in approximate articulation with the cranium, the procumbent
lower tooth of the left dentary could plausibly occlude with C1
(which is missing), but its tip is distinctly posterior to the upper
incisor alveoli. This evidence, together with the morphology of
the more anterior teeth (discussed below), strongly suggests that
the large procumbent lower tooth is c1 in E. ameghinoi,asalso
concluded by Paula Couto (1952c) and Marshall (1982a).
Paula Couto (1952c: figs. 3, 5A, 6A, 7A) illustrated the
presence of two procumbent incisors anterior to the alveolus
for c1 on the right side of DGM 321-M. These teeth appear to
have broken off some time in the following 30 years, because
they appear to be absent in Fig. 63a–bofMarshall(1982a);
they are currently not in the box containing DGM 321-M and
are presumably now lost. Paula Couto (1952c: figs. 3, 5A,
6A, 7A) indicated that these two incisors were arranged
mediolaterally, with the more medial tooth slightly longer.
Examining DGM 321-M today, the roots of both of these
incisors can be identified, with more of the root of the lateral
tooth preserved; intriguingly, however, there appears to be
evidence of at least one additional alveolus, dorsal to the roots
of the two procumbent teeth. This raises the possibility that
E. ameghinoi has three lower incisors, an interpretation that
receives further support from examination of other isolated
E. ameghinoi dentaries.
Particularly informative are DGM 171-M, a partial left
dentary, and MNRJ 2880-V, a partial right dentary (Fig. 7).
There is a single large ventral alveolus (Fig. 7: i1a) at the
anterior end of both specimens, which presumably housed
i1. Dorsolateral to this, is an elongate opening that is oriented
dorsomedial to ventrolateral. This opening appears to bifur-
cate into two alveoli deep within the substance of the dentary
(Fig. 7: i2a and i3a). Based on this, I propose that this opening
housed two teeth: ventrolaterally, a procumbent tooth that cor-
responds to the lateral incisor that was originally present in the
right dentary of DGM 321-M and was illustrated by Paula
Couto (1952c. figs. 3, 5A, 6A, 7A); dorsomedially, a procum-
bent tooth the dorsal edge of which would have been slightly
higher than the dorsal edge of i1, and the which would have
been directed slightly more dorsally than either of the other
two incisors.
The dorsomedial position of this latter tooth relative to the
other two incisors is strongly reminiscent of the Bstaggered^i2
seen in most polyprotodont metatherians (Hershkovitz 1982,
1995), but with a far greater degree of Bstaggering.^Based on
this arrangement, I therefore identify this tooth as i2, and con-
clude that E. ameghinoi retains i1–3. Under this interpretation,
the two incisors illustrated by Paula Couto (1952c)arei1andi3,
Fig. 7 Partial right mandible of Epidolops ameghinoi (MNRJ 2880-V) in
anterior view. Abbreviations: c1a = lower canine alveolus; i1a = first
lower incisor alveolus; i2a = second lower incisor alveolus; i3a = third
lower incisor alveolus; mf = mental foramen; p1a = first lower premolar
alveolus
J Mammal Evol (2017) 24:373–414 389
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

rather than i1 and i2 as he suggested. Thus, I conclude that the
complete lower dental formula of Epidolops ameghinoi is i1–3
c1 p1–3m1–4. Based on the depth of their alveoli, the roots of
the procumbent teeth (i1–3 c1) of E. ameghinoi are not particu-
larly extensive within the mandibular ramus; their posterior ex-
tent within the mandible is limited by the enormous roots of p3.
Probable Additional Material of E. ameghinoi
Based on craniodental specimens, E. ameghinoi is by far the
most common named marsupialiform species known from
Itaboraí (contra Ladevèze and Muizon 2010: 749), with 115
craniodental specimens known and a Minimum Number of
Individuals (MNI) of 43 (Table 2); this represents ~38 % of the
total number of marsipialiform craniodental specimens and
~33 % of the total marsupialiform MNI from the fauna. The next
most abundant single species, the much smaller Marmosopsis
juradoi, is less than half as common, with an MNI of 19
(Table 2). All other named marsupialiform species from
Itaboraí have an MNI of 8 or less (Table 2). It therefore seems
likely that E. ameghinoi is represented among the non-dental
marsupialiform specimens from Itaboraí.
Two major types of non-dental marsupialiform material have
been described from the site: petrosals (Ladevèze 2004,2007;
Ladevèze and Muizon 2010) and postcranial elements (Szalay,
1994; Szalay and Sargis, 2001). Of these, only Szalay’s(1994)
BItaboraí Metatherian Group^(IMG) tarsal morphotype VII has
been referred to E. ameghinoi. However, available evidence sug-
gests that a petrosal morphotype can also be tentatively referred
to E. ameghinoi.
Petrosals
Ladevèze (2004,2007) and Ladevèze and Muizon (2010)
identified eight marsupialiform petrosal morphotypes (Types
I-VIII) from Itaboraí that can be distinguished based on both
relative size and morphology. Ladevèze (2004,2007)and
Ladevèze and Muizon (2010) used phylogenetic and morpho-
metric approaches to try to associate these petrosal
morphotypes with marsupialiform taxa from Itaboraí that have
been named based on dental specimens. The morphometric
approach involved plotting the areas of M2, M3, m2, and
m3 against the area of the petrosal promontorium for a range
of modern and fossil metatherians, and then calculating pre-
dictive regression equations from these data that could then be
applied to the named Itaboraí marsupialiforms (Ladevèze and
Muizon 2010:fig.5,table3;Ladevèze2007:fig.1,
table 2). Based on these regression equations, Ladevèze
(2007) and Ladevèze and Muizon (2010) concluded that E.
ameghinoi could not be associated with any of the eight pe-
trosal morphotypes: its molars appeared to be too big for as-
sociation with Types I and III-VIII, but too small for
association with Type II (Ladevèze and Muizon 2010:
table 3; Ladevèze 2007: table 2). I reach a different conclu-
sion, based on two lines of evidence.
Firstly, the molars of E. ameghinoi appear smaller relative
to the overall size of its skull than those of most other
metatherians (see Szalay 1994: table 6.3). Thus,
E. ameghinoi should be expected to have petrosals with a
larger promontorial area than predicted by the regression
equations of Ladevèze (2007) and Ladevèze and Muizon
(2010). As an alternative approach, I carried out reduced ma-
jor axis regression of promontorium area of the 12
marsupialiform taxa used by Ladevèze and Muizon (2010)
Tab l e 2 Relative abundance of named marsupialiforms from the
Itaboraí Fauna. Taxonomy, identity, and number of specimens follow
Marshall (1978, 1981, 1982a, 1987), Goin and Oliveira (2007), Goin
et al. (2009), Oliveira and Goin (2011,2015), and Oliveira et al. (2016)
Species total number
of craniodental
specimens
minimum
number
of individuals
Epidolops ameghinoi 115 43
Marmosopsis juradoi 40 19
Protodidelphis
mastodontoides
a
16 4
Monodelphopsis travassoi 15 7
Protodidelphis vanzolinii 15 6
Patene simpsoni 14 5
Gaylordia mater 13 4
Gaylordia macrocynodonta
b
12 8
Didelphopsis cabrerai 10 3
Mirandatherium alipioi 10 7
Itaboraidelphys camposi 74
Guggenheimia crocheti 53
Minisculodelphis modicum 53
Carolopaulacoutoia
itaboraiensis
42
Derorhynchus singularis 42
Minisculodelphis minimus 32
Eobrasilia coutoi 21
Bobbschaefferia fluminensis 11
aff. Bobbschaefferia sp. 2 1
Carolocoutoia ferigoloi 11
Gashternia carioca 11
Guggenheimia brasiliensis 11
cf. Nemolestes sp. 1 1
Periprotodidelphis
bergqvistae
11
Procaroloameghinia pricei 11
Riolestes capricornicus 11
Zeusdelphys complicatus 11
a
=BRobertbutleria mastodontoidea^
b
includes Gaylordia Bdoelloi^
390 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

in their regression analyses, using cranial length as the predic-
tor variable (see Fig. 8), rather than molar size. This analysis
found a stronger correlation between promontorium area and
total skull length (R
2
=0.944;p=1.38×10
−7
) than the cor-
relations between promontorium area and molar size found by
Ladevèze and Muizon (2010: fig. 5A-D), and gives the fol-
lowing regression equation: log
10
(promontorium ar-
ea) = 1.258609*log
10
(cranial length) - 0.8755767. Assuming
a total skull length of 55 mm based on DGM 321-M (see
above), this gives an expected promontorium area of
20.6 mm
2
for Epidolops ameghinoi; this value is almost iden-
tical to that of the Type II petrosals, namely 20.72–21.91 mm
2
(Ladevèze and Muizon 2010: table 3; see Fig. 8).
Secondly, Ladevèze and Muizon (2010: 749) explicitly
rejected relative abundance as a criterion for associating the
petrosal morphotypes with dental taxa. However, I argue that
relative abundance should be taken into account, particularly
given that E. ameghinoi is by far the most common
marsupialiform at Itaboraí (Table 2). Ladevèze and Muizon
(2010) suggested that Type II belongs to either
Bobbschaefferia fluminensis or Procaroloameghinia pricei;
however, these two taxa are much rarer than E. ameghinoi,
with both having an MNI of 1 (Table 2). Based on the com-
bined evidence of relative size and relative abundance, I
conclude that the Type II petrosals most likely belong to
E. ameghinoi.
Ladevèze (2004) gave a detailed description of the Type II
petrosals. They exhibit a number of features that are rare or
absent in relative to crown marsupials, but which are found in
the Tiupampan stem marsupials Pucadelphys,Andinodelphys,
and Mayulestes. These include a tiny rostral tympanic process,
a deep groove for the internal carotid artery at the anterior pole
of the promontorium, and the posterior part of the
hypotympanic sinus excavated in the petrosal lateral to the
promontorium. Ladevèze and Muizon (2007:characters
162–165; 2010: characters 15 and 19) implied that this sinus
is not homologous with the hypotympanic sinus; however, I
interpret them as homologous based on their position and
structural relations (see e.g., Muizon et al. 1997:fig.2;
Muizon 1999: fig. 4; Ladevèze and Muizon 2007:text-
fig. 3). The similarity in petrosal structure between the Type
II morphotype and Pucadelphys,Andinodelphys,and
Mayulestes is particularly interesting given that all three
Tiupampantaxalackanossifiedhypotympanicsinusfloor,
as also seems to be the case for Epidolops (see above); if, as
I believe, the Type II petrosals belong toE. ameghinoi,thenall
four taxa appear to have a similar morphology of the auditory
region that differs from all known crown marsupials.
Tars a ls
Szalay (1994) identified 12 distinct morphotypes among isolat-
ed marsupialiform tarsals from Itaboraí, which he referred to as
BItaboraí Metatherian Groups^(IMGs) I-XII. Of these, (Szalay
1994:174–177) tentatively referred IMG VII to E. ameghinoi,
because it is the second most abundant IMG (comprising 14
calcanea and two astragali) and it is compatible in size with the
craniodental remains (the more common IMG II is far too small
to be plausibly referred to E. ameghinoi). The IMG VII
calcanea (Fig. 9a) are distinctly apomorphic: the peroneal pro-
cess (Fig. 9a: pp) is very reduced, with the groove for the
tendon of the peroneus longus muscle (Fig. 9a: gtpl) present
on the ventral (rather than dorsal) side of this process, a prom-
inent calcaneofibular facet (Fig. 9a: CaFi) is present lateral to
the ectal facet (Fig. 9a: Ec), and the calcaneal tuber is elongate
(Fig. 9a: ct).
On the other hand, the calcaneocuboid facet (Fig. 9a:
CaCu) of IMG VII is a single surface, unlike the apomorphic
bipartite calcaneocuboid facet characteristic of didelphids
(Szalay 1982,1994). The australidelphian apomorphies of a
tripartite calcaneocuboid facet and merged ectal and
sustentacular facets (= the continuous lower ankle joint pattern
[CLAJP]) are also absent (Fig. 9e; Szalay 1982,1994). The
overall morphology of IMG VII is well-adapted for extensive
flexion-extension, with little capacity for inversion or eversion
of the foot, suggesting a terrestrial (perhaps cursorial) locomo-
tor mode (Szalay 1994:174–177).
Fig. 8 Reduced (‘standardized’) major axis regression of log
10
-
transformed measurements of promontorium area against cranial length
for 12 extant and fossil marsupialiforms (see Ladevèze and Muizon 2010:
table 2and electronic supplementary material). The solid line represents
the line of best fit. The dotted horizontal lines represent the
log
10
(promontorium area) of the eight petrosal morphotypes (Types I-
VIII) from Itaboraí described by Ladevèze (2004,2007) and Ladevèze
and Muizon (2010). The dotted vertical line represents log
10
(estimated
cranial length) of Epidolops ameghino i, based on DGM 321-M. The Type
II petrosal morphotype is suitably-sized for referral to Epidolops
ameghinoi, whereas the other morphotypes are too small
J Mammal Evol (2017) 24:373–414 391
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Comparisons with Other Taxa Currently Included
in Polydolopimorphia
Dentition
The occlusal morphology and cusp homologies of the molar
dentition of E. ameghinoi and other taxa currently included in
Polydolopimorphia (see Goin and Candela 2004;Caseetal.
2005;Goinetal.2010,2016,in press) have been discussed at
length in previous papers (Pascual and Bond 1981;Marshall
1982a; Goin and Candela 1995,1996,2004; Goin 2003;Goin
et al. 2003a,2010;Chornogubsky2010;Zimicz2011,2014),
and will not be repeated here. Instead, in comparing
E. ameghinoi with other polydolopimorphians, I focus on oth-
er aspects of the dentition that show potentially phylogeneti-
cally informative variation.
Upper Anterior Dentition
Assuming that incisors are lost from the posterior end of the
incisor series (Ziegler 1971), the upper dental formula of
E. ameghinoi was probably I1–3C1P1–3M1–4, although
the presence of one or two additional incisors (I4 and I5)
cannot be completely ruled out (see above). There is no
diastema between I3 and C1, and (based on its alveolus) C1
was clearly a large tooth.
It is unclear exactly how many upper incisors are present
in the bonapartheriid Bonapartherium hinakusijum,but
Pascual (1981) concluded that there were probably five.
Bonapartherium hinakusijum also has a very large C1 that,
based on the morphology in MMP 148, appears semi-
procumbent (Pascual 1981: lamina I); however, this semi-
procumbency may be an artefact, because this specimen is
distinctly crushed dorsoventrally. The rosendolopid
bonapartherioid Hondonadia feruglioi preserves at least
four upper incisors (of which the posteriormost is the
smallest), but the presence of five teeth cannot be ruled
out (Goin and Candela 1998). Hondonadia feruglioi also
preserves a very large, subvertical C1 (Goin and Candela
1998). Both B. hinakusijum and H. feruglioi exhibit a prom-
inent diastema between the incisors and C1, within which
there is a distinct paracanine fossa (Pascual 1981;Goinand
Candela 1998). The lack of a diastema between the incisors
and C1 in E. ameghinoi may be connected with the
presence of a more procumbent c1 that presumably no
longer fits between the incisors and C1 during occlusion.
Flynn and Wyss (2004) interpreted the upper dental formu-
la of the polydolopid Kramadolops mckennai as I?1–2C1
P1–3M1–3, and argued that the three alveoli preserved at
Fig. 9 Isolated calcanea of a range of marsupialiforms in dorsal (flexad)
and anterior (distal) views. aBItaboraian Metatherian Group^(IMG) VII
tarsal morphotype from Itaborai, which Szalay (1994) tentatively referred
to Epidolops ameghinoi (redrawn from Szalay 1994: fig. 6.25); b
Unnamed argyrolagid from the Colhuehuapian (early Miocene) Gaiman
locality (redrawn from Szalay 1994: fig. 7.28); cArgyrolagus scagliai
(redrawn from Szalay 1994: fig. 7.28); dCaenolestes fuliginosus
(AMNH M-62915 –see Beck 2012); eDromiciops gliroides
(unregistered UNSW Palaeontology Laboratory specimen–see Beck
2012). Abbreviations: CaCu = calcaneocuboid facet; CaCua = auxiliary
calcaneocuboid facet; CaCul = lateral calcaneocuboid facet;
CaCum = medial calcaneocuboid facet; CaFi = calcaneofibular facet;
CLAJP = continuous lower ankle joint pattern; ct = calcaneal tuber;
Ec = ectal facet; gtpl = groove for the tendon of the peroneus longus
muscle; pp. = peroneal process; Su = sustentacular facet. Red represents
the ectal (Ec) and sustentacular (Su) facets or continuous lower ankle joint
pattern (CLAJP –formed by fusion of the ectal and sustentacular facets);
green represents the calcaneocuboid (CaCu) facet or auxiliary
calcaneocuboid (CaCua) facet (which are probably homologous –see
Szalay 1994;Beck,2012); yellow represents the lateral calcaneocuboid
(CaCul) facet; blue represents the medial calcaneocuboid (CaCum) facet.
Specimens are not drawn to scale. Note that the lateral calcaneocuboid
(CaCul) facet is perpendicular to the page and so is not visible in
Argyrolagus scagliai (c; compare with b,dand e)
392 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

the anterior end of the only known cranium of K. mckennai,
SGOPV 3476 (see Flynn and Wyss 2004: fig. 6.1), housed an
incisor, C1, and P1. However, such an anterior position for P1
would be highly unusual among metatherians, and I think it
more likely that the posteriormost alveolus of the three more
likely housed an incisor or C1 (see also Chornogubsky 2010).
Discovery of additional polydolopid specimens will be re-
quired to confidently infer their anterior dental formula.
Among argyrolagids, Proargyrolagus bolivianus exhibits four
upper incisors (of which I1 appears by far the longest
mesiodistally, followed by I2) and a very small canine, with no
distinct diastema separating these teeth (Sánchez-Villagra and
Kay 1997). Argyrolagus scagliai has only two teeth anterior to
P3, namely two similarly-sized, apparently open-rooted incisors
(presumably I1–2), followed by a very long diastema that seems
to lack a paracanine fossa (Simpson 1970b). Species of
Groeberia have two open-rooted upper incisors, immediately
followed by a small C1 that Pascual et al. (1994) described as
premolariform; no distinct diastema is present (Pascual et al.
1994: fig. 1D; Chimento et al. 2014:fig.2A).Klohnia charrieri
also has at least two (possibly open-rooted) upper incisors, of
which I1 is the larger, but it is uncertain whether or not a canine
or a diastema was present (Flynn and Wyss 1999).
Lower Anterior Dental Formula
I interpret the lower anterior dental formula of E. ameghinoi as
i1–3 c1 (see above); this is in contrast to previous authors,
whoinferredaformulaofi1–2 c1 (Paula Couto 1952c;
Marshall 1982a). All four of these anterior teeth appear to be
procumbent and enlarged, but i1 and c1 are markedly larger
than i2–3(Figs.7and 10a). Based on the arrangement of the
alveoli, i2 appears to be staggered, as in deltatheroidans and
most marsupialiforms (Hershkovitz 1982,1995; Cifelli and
Muizon 1997; Rougier et al. 1998; Kielan-Jaworowska et al.
2004), but with a much greater degree of staggering (Fig. 7).
Pascual (1981: 513) reported that there are four lower incisors
in Bonapartherium hinakusijum, and that i2 appears to be stag-
gered. Based on alveolar evidence, none of the lower incisors
were enlarged or procumbent in this taxon (Pascual 1981). The
c1 of B. hinakusijum is also well developed; Pascual (1981)
concluded that this tooth was probably similar in size and orien-
tation to the c1 of didelphids, implying that it is also non-
procumbent in B. hinakusijum.Thus,B. hinakusijum cannot be
described as diprotodont. Pascual’s(1980b) description of
Prepidolops didelphoides indicates that at least three incisors
were present in this taxon, that i2 was staggered, and that c1 is
relatively well developed. Intriguingly, the anterior dentition of
P. didelphoides is not procumbent in young individuals (e.g.,
MLP 78-V-6-1), but becomes increasingly procumbent with
age (Pascual 1980b); that is to say, the degree of diprotodonty
increases over the course of ontogeny. Assuming that
Bonapartherium and Prepidolops are indeed
polydolopimorphians, these taxa suggest that diprotodonty is
not a synapomorphy of the order as a whole (unless
Bonapartherium and Prepidolops have secondarily lost
diprotodonty, which seems unlikely). The evidence from
Prepidolops, with diprotodonty developing over ontogeny, is
particularly intriguing, as this may represent an intermediate mor-
phology between non-diprotodont and diprotodont dentitions; if
so, it may give general insight into how diprotodonty arises in
mammals.
Polydolopids have only a single enlarged and procumbent
(Bgliriform^) anterior tooth in the lower jaw (Fig. 10b); a
Fig. 10 Dentaries of Epidolops ameghinoi and putative relatives. a
Epidolops ameghinoi (DGM 321-M; modified from Paula Couto 1952:
fig. 6A); bthe polydolopid Kramadolops abanicoi (SGOPY 2941
[reversed]; modified from Flynn and Wyss 1999:fig.1);cthe
argyrolagid Anargyrolagus primus (MPEF-PV 5299 [reversed];
modified from Goin and Abello 2013: fig. 4.18); dthe caenolestid
Lestoros inca (USNM 194383; modified from Martin 2013: fig. 4A).
Abbreviations: c1 = lower canine; i1 = first lower incisor; i2a = second
lower incisor alveolus; i3 = third lower incisor; p1 = first lower premolar;
p1a = first lower premolar alveolus; p2 = second lower premolar;
p3 = third lower premolar
J Mammal Evol (2017) 24:373–414 393
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

critical, still-debated question is what locus this tooth repre-
sents (Marshall 1982a; Flynn and Wyss 1999,2004;
Chornogubsky 2010;Goinetal.2010). The two most likely
possibilities are i1 (Goin et al. 2010) and c1 (Marshall 1982a;
Flynn and Wyss 1999,2004); this is based on the general
assumption that incisors are lost from the posterior end of
the dental series, and on comparison with E. ameghinoi,in
which both i1 and c1 are enlarged and procumbent (see
above). The c1 of E. ameghinoi and the gliriform tooth of
polydolopids are somewhat different in morphology (Figs. 6,
10a-b): the former is distinctly caniniform, with enamel re-
stricted to the curved tip of the tooth, and it is less procumbent
than i1 (Figs. 6,10a; Paula Couto 1952c). By contrast the
gliriform tooth of polydolopids such as Kramadolops
abanicoi is not caniniform but is instead laterally compressed,
and it also appears proportionately much larger than the c1 of
E. ameghinoi (Fig. 10b; Flynn and Wyss 1999); however, it
does not seem impossible that this morphology evolved from
a more caniniform precursor.
A potentially key specimen for determining the homol-
ogy of the polydolopid gliriform tooth is MACN 10340a,
an edentulous partial right dentary of a polydolopid. This
specimen was originally described as the holotype of
BPromysops acuminatus^by Ameghino (1902), but
Marshall (1982a) referred it to Eudolops tetragonus,and
more recently Chornogubsky (2010) referred it to
Eudolops caroliameghinoi. Marshall (1982a) reported that
MACN 10340a preserves a large alveolus at its anterior
end, with two much smaller alveoli present mesial to this.
Ameghino (1903: Figs. 3 and 8) illustrated this specimen,
but indicated the presence of only two alveoli: a larger al-
veolus laterally and a smaller alveolus medially. Assuming
the more recent description of Marshall (1982a) is accurate,
it implies that the gliriform tooth of MACN 10340a (and
hence presumably of other polydolopids) is c1, with the
two more mesial alveoli housing reduced incisors.
Alternatively, one of these two alveoli might have housed
two incisors (i2 and i3), as appears to be the case in
E. ameghinoi (Fig. 7; see above). Ultimately, however, the
complete absence of teeth in MACN 10340a means that its
referral to Eudolops tetragonus is not assured, and the iden-
tity of the gliriform lower tooth of polydolopids remains
contentious.
Based on tooth counts in Proargyrolagus bolivianus (see
Sánchez-Villagra and Kay 1997) and Anagyrolagus primus
(see Goin and Abello 2013), the gliriform lower tooth of
argyrolagids is unequivocally an incisor, identified here as i1
(Fig. 10c). There are four small, single-rooted teeth
between the gliriform incisor and p3 in the lower jaw in
Proargyrolagus bolivianus (see Sánchez-Villagra and Kay
1997)andAnagyrolagus primus (Fig. 10c; Goin and Abello
2013), and hence there must be at least one additional incisor
present in both of these taxa; assuming that incisors are lost
from posterior to anterior, the tooth immediately posterior to
i1 can therefore be identified as i2. Sánchez-Villagra and Kay
(1997: 720) reported that, based on its alveolus, the i2 of
Proargyrolagus bolivianus was probably procumbent and that
it was Bpressed along the side^of i1, suggesting that it may
have been staggered, but this requires confirmation. The i2 of
Anagyrolagus primus does not appear obviously staggered
(Goin and Abello 2013:fig.4.13–18). I consider the homolo-
gies of the remaining three teeth between i2 and p3 in
Proargyrolagus bolivianus and Anagyrolagus primus to be
unclear, but Goin and Abello (2013) interpreted them as c1
p1–2. In Argyrolagus scagliai, there is one tooth between the
gliriform i1 and p3, which is small, single-rooted, procumbent
and immediately behind i1 (Simpson 1970b); based on com-
parison with Proargyrolagus bolivianus, it seems likely that
this tooth is i2, as was concluded by Simpson (1970b).
In the lower dentition of Groeberia minoprioi, a large, rel-
atively vertically-oriented (rather than procumbent) tooth is
present anteriorly, followed by a diastema, after which there
arefourmolars(Pascualetal.1994); the molars can be clearly
identified as such because they retain clear evidence of a
tribosphenic cusp pattern (Patterson 1952;Simpson1970a).
In the holotype of Groeberia minoprioi, MMP 738, there is no
evidence of teeth within the lower diastema (Patterson 1952;
Chimento et al. 2014). However, Pascual et al. (1994)con-
cluded that there are two unicuspid teeth between the enlarged
anterior tooth and m1 in another G. minoprioi specimen, MLP
85-IX-24-1. Subsequently, Chimento et al. (2014)arguedthat
only a single unicuspid is present in this region in this speci-
men. By itself, this information is insufficient to identify the
identity of the enlarged anterior lower tooth in Groeberia
minoprioi; however, given that its two occlusal counterparts
in the upper jaw are unequivocally incisors (I1–2; Simpson
1970a; Pascual et al. 1994;Chimentoetal.2014), it seems
reasonable to conclude that it is also an incisor, i1. The identity
of the unicuspid (where present) within the lower diastema
(i.e., whether an incisor, canine, or premolar) is uncertain.
Pascualetal.(1994) suggested that it might be c1; however,
given that it is closely appressed to i1, it is tempting to identify
it as i2, as assumed by Chimento et al. (2014). If it is i2,
published studies do not clearly indicate whether or not it is
staggered.
In the lower jaw of Klohnia charrieri, there is a single
gliriform tooth followed by a large diastema and then p3; the
diastema is reportedly entirely edentulous (Flynn and Wyss
1999). Two large, open-rooted incisors are present in the up-
per jaw of Klohnia charrieri (see Flynn and Wyss 1999),
making it likely that the lower gliriform tooth is also an inci-
sor, i.e., i1. The lower anterior dentition of the enigmatic
Patagonia peregrina comprises a single lower gliriform tooth
of uncertain homology, but which has been interpreted as
incisor (Pascual and Carlini 1987; Goin and Abello 2013),
immediately followed by a very small, single-rooted,
394 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

somewhat procumbent tooth that was identified as a probable
c1 by Pascual and Carlini (1987, see also Goin and Abello
2013) but whose homology is likewise uncertain; it may in
fact be i2.
Premolar Number and Morphology
In E. ameghinoi, P1 and p1 were clearly small, single-rooted
teeth (although neither tooth is preserved in any known spec-
imen), P2 and p2 are tiny and buttonlike, while P3 and p3 are
enormously hypertrophied and plagiaulacoid (Figs. 2-4,6and
11a). Other polydolopimorphians show a diversity of premo-
lar numbers and morphologies.
In Bonapartherium hinakusijum, P1 is small but double-
rooted, whilst P2 and P3 are both large, three-rooted, and
bladelike, with a distinct lingual platform or talon not present
in E. ameghinoi; P3 is considerably wider and taller than P2
(Fig. 11b; Pascual 1980a,1981). In the lower dentition of
B. hinakusijum, p1 and p2 are much smaller and lower than
the large, bladelike p3 (Pascual 1980a,1981). Unlike
E. ameghinoi, neither the P3 nor p3 of B. hinakusijum is truly
plagiaulacoid sensu Simpson (1933), i.e., they lack a serrated
edge (Fig. 11b). In Hononadia feruglioi, the morphology of
P3 is unknown, but both P1 and P2 are small and
premolariform, with P2 slightly larger (Goin and Candela
1998). Hondonadia (= BPascualdelphys^)fierroensis (see
Goin et al. 2010) is known from a single lower dentary, the
premolars of which appear more plesiomorphic than those of
other taxa currently placed in the order Polydolopimorphia: all
three are double-rooted and they increase in size slightly from
anterior to posterior (Flynn and Wyss 1999).
In the prepidolopids Prepidolops didelphoides and
Punadolops alonsoi, P3 is the only upper premolar currently
known; it is very large and bladelike but not plagiaulacoid,
and it lacks a lingual platform or talon (Pascual 1980a,b;Goin
et al. 1998a). In the lower dentition of prepidolopids, meanwhile,
p1 is very small (based on Prepidolops didelphoides) and possi-
bly single-rooted (based on Prepidolops molinai), p2 is some-
what larger and double-rooted, with the roots and crown oriented
strongly obliquely relative to the major dental axis, and p3 is
enormous, double-rooted, and similar in crown morphology to
P3 (Pascual 1980b;Goinetal.1998a;). P3 is also the only upper
premolar known for Gashternia carioca:it is bladelike, and
somewhat plagiaulacoid, with five cusps aligned along its labial
margin, and a distinct shelf supporting a single blunt cusp is also
present lingually (Fig. 11c; Goin and Oliveira 2007). Lower
premolars of Gashternia calehor are not preserved in the only
known specimen (AMNH 28533), but, based on their alveoli, p1
was small and single-rooted, p2 was larger, double-rooted, and
implanted obliquely, and p3 was larger still and also double-
rooted (Simpson 1948). In Wamradolops tsullodon,P3islarge
and bladelike but apparently not plagiaulacoid (Goin and
Fig. 11 Upper premolars and
first upper molar of Epidolops
ameghinoi and putative relatives.
aEpidolops ameghinoi (DGM
800-M; modified from Marshall
1982: fig. 67b); bthe
bonapartheriid Bonapartherium
hinakusijum (MMP 1416); cthe
gashterniid Gashternia carioca
(MCN-PV 1801 [reversed];
modified from Goin and Oliveira
2007: fig. 1); dthe polydolopid
Polydolops thomasi (MACN
10338; modified from Marshall
1982: fig. 32);. ethe argyrolagid
Anargyrolagus primus (MACN-
ch-1305; modified from Carlini
et al. 2007: fig. 2A).
Abbreviations: M1 = first upper
molar; P2 = second upper
premolar; P3 = third upper
premolar
J Mammal Evol (2017) 24:373–414 395
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Candela 2004); a small posterolingual platform appears to be
present, but it is far less developed than in either Gashternia
carioca or Bonapartherium hinakusijum.
Based on SGOPV 3476, P1 must have been very small or
entirely absent in the polydolopid Kramadolops mckennai
(the alveolus that Flynn and Wyss 2004 argued as having
housed P1 is more likely for an incisor or C1 –see above),
and it has not been reported in other members of the family.
Both P2 and P3 are well developed, double-rooted and
plagiaulacoid in K. mckennai, with P2 the larger of the two
(Flynn and Wyss 2004). Among other polydolopids, Marshall
(1982a) reported that P2 is apparently absent but P3 is present
and double-rooted in Amphidolops serrula (see also Simpson
1948: plate 6.5), whereas in Eudolops tetragonus P2 and P3
are similarly-sized, relatively small, double-rooted, and some-
what bladelike without being plagiaulacoid (see Marshall
1982a: fig. 58). In most polydolopids (e.g., Polydolops
thomasi; Fig. 11d), however, P2 is considerably larger than
P3, and both teeth are distinctly plagiaulacoid (Marshall
1982a; Chornogubsky et al. 2009).
In the lower dentition, comparison with E. ameghinoi per-
suades me that the lower postcanine formula proposed for the
polydolopid Kramadolops abanicoi by Flynn and Wyss
(1999)–namely p1–3m1–3–is correct. If so, the p1 of
K. abanicoi is very similar in morphology to that of
E. ameghinoi, being very small, single-rooted, and procum-
bent (Fig. 10b; Flynn and Wyss 1999). The p2 is present but
very small and either double- or single-rooted in many
polydolopids (e.g., Kramadolops abanicoi -Fig.10b;
Polydolops spp.; Marshall 1982a) but entirely absent in others
(e.g., Antarctolops spp.; Chornogubsky et al. 2009).
Roberthoffstetteria nationalgeographica (grouped with
polydolopids in Polydolopiformes by Goin et al. 2016)has
three double-rooted, premolariform premolars, of which P3
was probably slightly larger than P2 (Marshall et al. 1983;
Muizon et al. 1984;G
oinetal.2003a). The lower premolars
of R. nationalgeographica are less well known, but p3 appears
to be double-rooted and premolariform (Muizon et al. 1984).
Among argyrolagids (e.g., Anargyrolagus primus –
Fig. 11e), P3 and p3 are consistently present, small, and ap-
pear to be hypsodont or hypselodont, similar to the molars;
with wear, they form a continuous dental series with the mo-
lars (Simpson 1970b; Sánchez-Villagra and Kay 1997;Goin
and Abello 2013). Both Anagyrolagus primus (Fig. 11e) and
Proargyrolagus bolivianus have a small, single-rooted P1 and
P2 in the upper jaw, whilst in the lower jaw there are four
small, single-rooted teeth between the gliriform incisor and
p3, and hence there could be a total of one, two, or three lower
premolars (Sánchez-Villagra and Kay 1997;Carlinietal.
2007; Goin and Abello 2013). In Argyolagus scagliai,there
are no teeth between I2 and P3 in the upper jaw, whilst in the
lower jaw there is one tooth between the gliriform incisor and
p3, which is probably i2 (Simpson 1970b; see above).
Groeberia minioproi has three small, single-rooted upper
premolars (i.e., P1–3) that differ little in size in the upper jaw
(Pascual et al. 1994). As discussed, the holotype of
G. minioproi lacks any teeth in the diastema between i1 and
m1, and hence lower premolars are absent in this specimen,
but a single-rooted unicuspid of uncertain homologyis present
in the diastema in MLP 85-IX-24-1 (Pascual et al. 1994;
Chimento et al. 2014). Thus, at most G. minioproi sometimes
retained a single lower premolar. However, Pascual et al.
(1994) suggested that the unicuspid in MLP 85-IX-24-1 is a
canine, and Chimento et al. (2014)arguedthatitismorelikely
an incisor; in either case, this would mean that lower premo-
lars are consistently absent in G. minioprioi. Upper premolar
number in Klohnia charrieri is uncertain based on available
specimens, but P3 is clearly present; its occlusal morphology
is unknown, but it clearly was not hypertrophied (Flynn and
Wyss 1999). In the lower jaw, p3 is apparently the only lower
premolar present: it is small, peg-like, and probably single-
rooted, but it appears somewhat hypsodont and with wear its
morphology resembles that of the p3 of argyrolagids (compare
Flynn and Wyss 1999:fig.2EwithSimpson,1970b: fig. 1A-
C). Assuming that the three posteriorly-located, quadrilateral,
hypselodont teeth in the upper and lower jaw of Patagonia
peregrina are molars (i.e., M1–3andm1–3), this taxon entire-
ly lacks premolars (Pascual and Carlini 1987; Goin and
Abello 2013).
Molar Number
Epidolops ameghinoi retains four molars, but M4 and m4 are
very small and single-rooted (Figs. 3,4and 6). Four molars
are present in Bonapartherium hinakusijum,Prepidolops
didelphoides,andP. molinai, but M4 and m4 are consider-
ably smaller than M3 and m3 (Pascual 1980a,b,1981). All
known polydolopids lack the fourth molar (Marshall 1982a),
as does the prepidolopid Punadolops alonsoi (see Goin et al.
1998a). Hondonadia (= BPascualdelphys^)fierroensis re-
tains four lower molars, with m4 double-rooted and only
slightly smaller than m3 (Flynn and Wyss 1999).
Argyrolagids retain four molars, with M4 and m4 usually
markedly smaller than M1–3andm1–3(Fig.10c; Simpson
1970b; Sánchez-Villagra and Kay 1997; Carlini et al. 2007;
Goin and Abello 2013), and a similar morphology is seen in
Groeberia (see Pascual et al. 1994). Klohnia charrieri has
only three molars (Flynn and Wyss 1999), as does
Patagonia peregrina (see Pascual and Carlini 1987;Goin
and Abello 2013).
Rostrum
Epidolops ameghinoi lacks a distinct masseteric process (for
attachment of the superficial masseter) at the anterior end of
the zygomatic arch (Figs. 2-4). A masseteric process also
396 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

appears to be absent in Bonapartherium hinakusijum,butdor-
soventral crushing of the best preserved cranial specimen,
MMP 1408, means that this is uncertain (Pascual 1981:lamina
I). The only known cranium of Kramadolops mckennai is too
badly crushed to determine whether or not a masseteric pro-
cess is present (Flynn and Wyss 2004: figs. 6.1–6.2). A dis-
tinct, raised masseteric process is, however, clearly present in
argyrolagids (Simpson 1970b: fig. 4A-B) and Groeberia (see
Pascualetal.1994).
Palate
Epidolops ameghinoi is characterized by a relatively imperfo-
rate palate, with a single pair of very small maxillopalatine
fenestrae extending from level with the anterior margin of
M1 to level with M2 (Fig. 4). Most definitive crown marsu-
pials have comparatively much larger palatal vacuities, al-
though their number, exact size, and sutural relations vary
(Archer 1984a,b; Voss and Jansa 2009).Afew,however,have
a largely imperforate palate, among them the didelphid
Caluromys, which is the modern marsupial that most closely
resembles E. ameghinoi in terms of the probable size of its
maxillopalatine fenestrae (see e.g., Bucher and Hoffmann
1980:fig.2; Cáceres and Carmignotto 2006:fig.2;Vossand
Jansa 2009:fig.38).
Among other polydolopimorphians, the palate of
Bonapartherium hinakusijum appears very similar to that of
E. ameghinoi, with a single, small, relatively centrally-placed
pair of maxillopalatine fenestrae (Pascual 1981). Pascual
(1981) stated that the fenestrae extend from P3 to M2 in
Bonapartherium hinakusijum, but Fig. 4.1 of Zimicz (2014)
and Fig. 5.10b of Goin et al. (2016) suggest that the fenestrae
are damaged in the best preserved cranium of this specimen,
MMP 1408, and that they may have only extended the length
of M1–2 when intact, as in E. ameghinoi. Flynn and Wyss
(2004: 85) stated that Ba distinct foramen or vacuity occurs
near the posterobuccal corner of the palate^in the polydolopid
Kramadolops mckennai. In fact, this opening (clearly identi-
fiable on the right [= anatomical left] side of SGOPV 3476 in
Flynn and Wyss, 2004:fig.6.1) appears to be the posterolat-
eral palatal foramen (see Voss and Jansa 2009:fig.14;Wible
2003:figs.4,5B). As illustrated (Flynn and Wyss 2004:
fig. 6.1), the palate of SGOPV 3476 seems too badly damaged
to determine whether or not there are true fenestrae; if
present, however, such fenestrae must have been small.
Other polydolopid specimens are insufficiently well preserved
to determine whether or not palatal fenestrae are present.
Pascual et al. (1994) reported the presence of an elongate
pair of palatal fenestrae in Groeberia spp., but they did not
identify the bones enclosing them; however, they appear to be
maxillopalatine fenestrae (pers. obs.). Elongate palatal vacui-
ties are present in all known argyrolagids (Simpson 1970b;
Sánchez-Villagra and Kay 1997; Sánchez-Villagra et al. 2000;
Carlini et al. 2007; García-López and Babot 2015); in
Hondalagus altiplanensis at least, they are between the max-
illa and palatine (pers. obs.). The precise extent of these fenes-
trae is unclear in Hondalagus altiplanensis, but in
Proargyrolagus bolivianus,Anagyrolagus primus,and
Argyrolagus scagliai they appear to extend from P3 to M4
(Simpson 1970b; Carlini et al. 2007). As noted by Simpson
(1970b: 22), a narrow median septum may have originally
divided left and right fenestrae in Argyrolagus scagliai,but,
if so, has broken away in known specimens; alternatively, a
septum might have been absent, with the fenestrae forming a
single, very large opening (this region is less well preserved in
Hondalagus altiplanensis,Proargyrolagus bolivianus and
Anagyrolagus primus). Finally, Goin and Abello (2013)re-
ported that palatal fenestrae are absent in Patagonia
peregrina, but more complete material is probably required
to confirm this.
Postpalatal Region
A notable apomorphy of the postpalatal region of E. ameghinoi
is the presence of a very well-defined pterygoid fossa, with a
distinct ectopterygoid crest laterally enclosing at least the an-
terior half of the fossa (Figs. 3and 5); in most other
metatherians, this fossa is shallow or indistinct, and lacks an
obvious ectopterygoid crest. Known crania of most other
polydolopimorphians are too poorly preserved to determine
the morphology of the pterygoid fossa (Simpson 1970a;
Pascual 1981; Pascual et al. 1994; Sánchez-Villagra and Kay
1997; Sánchez-Villagra et al. 2000;FlynnandWyss2004);
however, this fossa does not appear to be well developed in
the argyrolagid Argyrolagus scagliai (Simpson 1970b:
fig. 3B).
The transverse canal foramen is either absent or (based on a
tiny foramen present bilaterally within the pterygoid fossa)
very small in E. ameghinoi (Figs. 3and 5). Most other
polydolopimorphians are insufficiently well-preserved to de-
termine whether or not the transversecanal foramen is present,
but this foramen is present and large in the argyrolagids
Hondalagus altiplanensis (there are three foramina on the
right side of MNHN-Pal-BoIV-006,330, but only one on the
left side; Sánchez-Villagra et al. 2000)andArgyrolagus
scagliai (Simpson 1970b).
Auditory Region
The most striking aspect of the auditory region of E. ameghinoi
is the apparent lack of an ossified floor to the hypotympanic
sinus (see above; Fig. 12a). The morphology of the auditory
region is unknown for most other polydolopimorphians.
However, all known argyrolagids differ markedly from
E. ameghinoi in possessing a large alisphenoid tympanic
process flooring the hypotympanic sinus (Fig. 12d; Simpson
J Mammal Evol (2017) 24:373–414 397
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1970b; Sánchez-Villagra and Kay 1997; Sánchez-Villagra et al.
2000).
Glenoid Region
The glenoid region of E. ameghinoi appears unspecialized
(Figs. 3,5and 12a). The glenoid fossa forms a smooth curve,
and the postglenoid process is well developed but
unpneumatized. Complete loss of the ectotympanic in DGM
321-M suggests that this bone was neither fused nor tightly
sutured to the postglenoid process (fusion of the ectotympanic
with adjacent bones is observed in many diprotodontians;
Aplin 1987,1990; Springer and Woodburne 1989). The
postglenoid foramen is immediately posterior to the
postglenoid process, and was probably entirely enclosed by
the squamosal in the intact skull. The only other group cur-
rently included in Polydolopimorphia for which the glenoid
region is known is Argyrolagidae. Simpson (1970b:24–25)
reported that the glenoid fossa of Argyrolagus scagliai is
Balmost perfectly flat,^and that the ecotympanic resembles
that of Caenolestes, namely forming an incomplete ring that is
unfused to the adjacent bones. Simpson (1970b)didnotde-
scribe the morphology of the postglenoid process, but he iden-
tified a foramen immediately dorsal to the external auditory
meatus as a possible homologue of the postglenoid foramen
(Simpson 1970b: 25); however, it seems more likely that this
is in fact the subsquamosal foramen (= suprameatal foramen
sensu Wible 2003), in which case the postglenoid foramen of
argyrolagids has yet to be identified.
Mandible
The overall morphology of the mandible ofE. ameghinoi does
not appear particularly derived, except for the presence of
multiple small foramina within the masseteric fossa
(Figs. 6and 10a). Pascual (1980b,1981) did not discuss the
presence of masseteric foramina in Prepidolops spp. or
Bonapartherium hinakusijum, nor are they mentioned in pub-
lished descriptions of polydolopids, Klohnia charrieri,or
Patagonia peregrina (Marshall 1982a; Pascual and Carlini
Fig. 12 Basicranial region of Epidolops ameghinoi and other
marsupialiforms. aEpidolops ameghinoi (DGM 321-M); bthe
mayulestid Mayulestes ferox (MHNC 1249; modified from Muizon
1998:fig.8A); cthe caenolestid Caenolestes convelatus (modified
from Animal Diversity Web); dthe argyrolagid Argyrolagus scagliai
(MMP 5538 –specimen is damaged and is missing the occipital
region). Abbreviations: atp = alisphenoid tympanic process;
cf = carotid foramen; ect = ectotympanic; fo = foramen ovale;
gf = glenoid fossa; hs = hypotympanic sinus; pgf = postglenoid
foramen; pgp = postglenoid process; pr = promontorium of the
petrosal; rtpp = rostral tympanic process of the petrosal;
tcf = transverse canal foramen. Note that a-c are in ventral view,
whereas dis in ventromedial view
398 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1987; Flynn and Wyss 1999;Goinetal.2010;Goinand
Abello 2013). Rusconi (1933,1936) and Simpson (1970b:
20) both stated that a masseteric foramen is present in
Argyrolagus spp., but it is not mentioned in descriptions
of other argyrolagids in which this region is preserved
(Sánchez-Villagra and Kay 1997;GoinandAbello2013).
Epidolops ameghinoi lacks a retrodental canal sensu
Hoffstetter and Villarroel (1974, = maxillary canal sensu
Babot and García-López 2016), and this structure
has not been reported in Bonapartherium hinakusijum,
Prepidolops spp. or polydolopids (Pascual 1980b,1981;
Marshall 1982a). It is also clearly absent in Groeberia
(pers. obs.). A very large retrodental canal is, however,
consistently present in argyrolagids (Fig. 13a; Rusconi
1933; Simpson 1970b; Hoffstetter and Villarroel 1974;
Sánchez-Villagra et al. 2000; Babot and García-López
2016) - Goin and Abello (2013) reported that a retrodental
canal is absent in Proargyrolagus, but Sánchez-Villagra
et al. (2000: 292) identified the canal in Proargyrolagus
bolivianus after additional preparation of already-
described material (Sánchez-Villagra and Kay 1997).
Petrosal
As discussed above, Ladevèze’s(2004) Type II marsupialiform
petrosals from Itaboraí plausibly represent E. ameghinoi.Notable
features of this morphotype include the presence of a deep
groove for the internal carotid artery at the anterior pole of the
promontorium, presence of a very small rostral tympanic process,
and an incomplete dorsal roof of the geniculate ganglion of the
facial nerve (Ladevèze 2004).
Most other polydolopimorphians are not known from pe-
trosal specimens, but the ventral surface of the petrosal has
been described for the argyrolagids Proargyrolagus
bolivianus and Hondalagus altiplanensis: in both taxa, the
petrosal differs markedly from the Itaboraí Type II petrosals
in having a prominent, anteroposteriorly elongate rostral tym-
panic process (Sánchez-Villagra and Kay 1997; Sánchez-
Villagra et al. 2000). A similarly prominent rostral tympanic
process is present in Argyrolagus scagliai (Fig. 12d).
Hondalagus altiplanensis lacks a deep promontorial groove
for the internal carotid artery (Sánchez-Villagra et al. 2000), as
does Aryrgolagus scagliai (pers. obs.), but it is unclear wheth-
er or not this structure is present in Proargyrolagus bolivianus
(see Sánchez-Villagra and Kay 1997).
Tars a ls
Notable features of the IMG VII tarsal morphotype that Szalay
(1994)referredtoE. ameghinoi include: a very small peroneal
process that is positioned at the distal terminus of the calcane-
us, and in which the groove for the tendon of the peroneus
longus muscles is on the ventral (rather than dorsal) surface;
an unspecialized calcaneocuboid facet; an elongate tuber; sep-
arate ectal and sustenacular facets with a distinct sulcus
calcanei between them (= the separate lower ankle joint pat-
tern [SLAJP]); and a calcaneofibular facet that is lateral to
(and continuous with) the ectal facet (Fig. 9a).
Tarsal remains have also been described for argyrolagids
(Simpson 1970b; Szalay 1994; Babot and García-López 2016;
see Fig. 9b-c). These specimens show several derived similar-
ities that are markedly different from IMG VII: the
calcaneofibular facet is extremely broad, and in Argyrolagus
scagliai and Microtragulus bolivianus it is largely isolated
from (and much larger than) the ectal facet, whilst the
sustentacular facet is small and faces almost directly medially
(Fig. 9b and c). Although still not particularly well developed,
the peroneal process of argyrolagids is larger than that of IMG
VII and is set back from the distal end of the calcaneus, and the
groove for the tendon of the peroneal longus is on the dorsal
(not ventral) surface. Perhaps the most striking difference be-
tween the argyrolagid tarsals and IMG VII is the morphology
of the calcaneocuboid facet: it is a single facet in IMG VII
(Fig. 9a), whereas in argyrolagids it is tripartite and distinctly
Bstepped^(Fig. 9b and c), with a distally-facing proximal
Fig. 13 Comparison of the morphology of the retromolar space of the
argyrolagid Argyrolagus and the caenolestid Caenolestes.aPartial left
mandible of Argyrolagus sp. (MACN 17590) in medial view, with arrows
indicating the path of the retrodental canal (= maxillary canal sensu Babot
and García-López 2016; modified from Babot and García-López 2016:
fig. 9.1); bLeft mandible of Caenolestes caniventer (BMNH 1954.302)
in medial view, with vertical red line corresponding to the plane of the
coronal section shown in c;ccoronal section of left mandible of
Caenolestes fuliginosus (KU 124015) in anterior view, based on CT
scan data (modified from Digimorph). Abbreviations: can = canal;
cor = coronoid process; m4 = fourth lower molar; manc = mandibular
canal; manf = mandibular foramen; p3 = third lower premolar
J Mammal Evol (2017) 24:373–414 399
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

facet, a more distal facet that faces medially, and a distalmost
face that faces distally (see also Simpson 1970b:30–31;
Szalay 1994).
Affinities of Epidolops and Other
Polydolopimorphians
There are three main hypotheses for the affinities of Epidolops
and the other taxa currently included in Polydopimorphia,
namely that they are closely related to paucituberculatans
(Gregory 1910; Simpson 1928,1948; Paula Couto 1952c;
Simpson 1945; Aplin and Archer 1987;Marshall1987), that
they are closely related to microbiotherians and
diprotodontians in the marsupial superorder Australidelphia
(Goin et al. 1998b,2009,2016,in press;Goin2003;Goin
and Candela 2004; Oliveira and Goin 2006,2011;
Chornogubsky and Goin 2015), or that they are not closely
related to any other marsupialiform order (Szalay 1994;Case
et al. 2005). Here, I review these alternatives based on the
evidence provided by the E. ameghinoi specimens from
Itaboraí.
Evidence for Paucituberculatan Affinities
A close relationship between polydolopimorphians and
paucituberculatans was originally proposed based on the pres-
ence in both groups of an enlarged and procumbent
(Bgliriform^) anterior tooth in the lower jaw (Gregory 1910;
Simpson 1928,1948; Paula Couto 1952c; Simpson 1945).
However, the antemolar dentition of E. ameghinoi and
paucituberculatans differ markedly: specifically, E. ameghinoi
has four large, procumbent teeth (i1–3andc1,withi1andc1
particularly large; Figs. 6,7,and10a) whereas
paucituberculatans have only a single gliriform tooth followed
by a series of very small unicuspids (Fig. 10d; Osgood 1921;
Marshall 1980;Martin2007; Voss and Jansa 2009;Abello
2013;M
artin2013). Among other polydolopimorphians,
Bonapartherium hinakuaijum apparently lacks a true gliriform
lower tooth (Pascual 1981), whilst in Prepidolops didelphoides
procumbency of the anterior lower dentition develops over the
course of ontogeny (Pascual 1980b). Assuming that
Bonapartherium and Prepidolops form a clade with Epidolops
and other taxa currently included in Polydolopimorphia (Goin
et al. 2009; Chornogubsky and Goin 2015), and that the absence
of true diprotodonty in Bonapartherium and Prepidolops is
plesiomorphic rather than secondary, then this represents com-
pelling evidence that diprotodonty evolved independently in
Polydolopimorphia and Paucituberculata.
Epidolops ameghinoi has several foramina within the mas-
seteric fossa, and a masseteric foramen is also present in some
paucituberculatans (e.g., Caenolestes spp., Lestoros inca,and
Stilotherium dissimile;Osgood1921; Simpson 1970b; Voss
and Jansa 2009). However, the presence of this foramen has
not been reported in any other putative polydolopimorphian
besides the argyrolagid Argyrolagus spp. (Simpson 1970b).
Furthermore, a masseteric foramen is only variably present
in some other paucituberculatans (e.g., Palaeothentes spp.,
Rhyncholestes raphanurus; Forasiepi et al. 2014), whereas it
is consistently present in many diprotodontians, the
dasyuromorphian Myrmecobius fasciatus,Notoryctes spp.,
Yalkaparidon coheni, and the microbiotherian
Microbiotherium gallegosense (pers. obs.; Abbie 1939;
Marshall 1982b;Becketal.2014). Thus, the presence of this
foramen in E. ameghinoi does not constitute strong evidence
for a close relationship with paucituberculatans.
The remainder of the cranium of E. ameghinoi appears
more plesiomorphic than paucituberculatans and other crown
marsupials in lacking an ossified hypotympanic sinus floor
(Figs. 3,5,and12a). All known paucituberculatans have the
hypotympanic sinus enclosed anteriorly and ventrally by an
alisphenoid tympanic process (Fig. 12c; Osgood 1921,1924;
Patterson and Gallardo 1987;Goinetal.2003b;Ojala-
Barbour et al. 2013; Forasiepi et al. 2014), a morphology that
is likely plesiomorphic for Marsupialia as a whole (Horovitz
and Sánchez-Villagra 2003).
The very small palatal vacuities and absent or tiny
tranverse canal foramen of E. ameghinoi are also unlike
the morphology seen in paucituberculatans and most other
marsupials, in which palatal vacuities are normally well
developed and the transverse canal foramen is usually
prominent (Fig. 12c; Osgood 1921,1924; Patterson and
Gallardo 1987; Sánchez-Villagra and Wible 2002;Goin
et al. 2003b,2007a;Martin2013; Ojala-Barbour et al.
2013;Rincónetal.2015); these features of E. ameghinoi
may be plesiomorphies, although it should be noted there is
considerable homoplasy in both features within Metatheria
(see below).
If the Type II petrosals do indeed belong to E. ameghinoi,
they share a few features with some paucituberculatans: like
some specimens of Caenolestes spp. the dorsal roof for the
geniculate ganglion is incomplete, and they share with speci-
mens of Lestoros inca the presence of an anteroventral groove
on the promontorium for the internal carotid artery (pers. obs.;
Wible 1990;Ladevèze2004). However, they differ in their
tiny rostral tympanic process, which is a probable
plesiomorphic feature; all known paucituberculatan petrosals
exhibit a prominent rostral tympanic process (Fig. 12c;
Sánchez-Villagra and Wible 2002;Goinetal.2003b;
Forasiepi et al. 2014).
If the IMG VII tarsals represent E. ameghinoi, then they
also differ markedly from those of paucituberculatans
(compare Fig. 9a and d). IMG VII is apomorphic in that the
peroneal process is very small, with the groove for the tendon
of the peroneus longus muscle on the ventral (rather than
dorsal) surface, and the tuber is relatively elongate, but
400 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

plesiomorphic in that the calcaneocuboid facet is a single facet
(Fig. 9a; Szalay 1994); the very distal position of the peroneal
process in also distinctive, but is of uncertain polarity. By
contrast, calcanea of paucituberculatans (Fig. 9c) retain a
well-developed peroneal process with a dorsal groove for
the peroneus longus and a tuber that is not obviously elongate,
whereas the calcaneocuboid facet appears distinctly tripartite,
particularly in Palaeothentes minutus (Szalay 1982,1994;
Abello and Candela 2010).
Goin and co-workers (see e.g., Goin 2003;Goinetal.
2009) have shown that molar morphology differs marked-
ly between polydolopimorphians and paucituberculatans,
and they have concluded that their derived similarities,
such as the presence of enlarged stylar cusps B and D
and a metaconular hypocone sensu Beck et al. (2008a),
evolved independently. This conclusion has been support-
ed by published phylogenetic analyses (Goin et al. 2009;
Forasiepi et al. 2013; Chornogubsky and Goin 2015), al-
though these have focused almost exclusively on dental
characters and have not incorporated the cranial or post-
cranial evidence discussed here.
Evidence for Microbiotherian Affinities
Support for a close relationship between Polydolopimorphia
and the extant South American australidelphian order
Microbiotheria is based largely on the proposal that the
polydolopimorphian molar pattern is derivable from a some-
what Bmicrobiotherian-like^ancestor (Goin et al. 1998b,
2007b,2016;Goin2003; Goin and Candela 2004).
However, molars of definitive microbiotherians and
polydolopimorphians differ markedly in that the stylar cusps
are reduced and labiolingually compressed (forming a
crestlike structure along the labial margin of the tooth) in
microbiotherians (Marshall 1982b;Goinetal.2007b,2016;
Goin and Abello 2013), whereas stylar cusps B and D are
distinctly enlarged in polydolopimorphians (Fig. 11a-d; Goin
and Candela 1996;Goin2003;Goinetal.2016). Even the
most plesiomorphic described microbiotherian, the middle
Eocene Woodburnodon casei, appears far too derived in terms
of its stylar shelf morphology to represent a plausible struc-
tural ancestor for polydolopimorphians (Goin et al. 2007b).
Centrocrista morphology also differs between the two groups:
the centrocrista is straight in all known microbiotherians, whereas
the centrocrista of polydolopimorphians (where identifiable) is
open, with the postparacrista terminating at stylar cusp B and the
premetacrista terminating at stylar cusp D (Fig. 11a-d;;Goin
2003;C
aseetal.2005;Goinetal.2009: character 33). Finally,
all known polydolopimorphians have an enlarged metaconule
that is usually posterolingually displaced to form a metaconular
hypocone sensu Beck et al. (2008a;seeFig.11a-d), whereas all
known microbiotherians have very reduced conules (Marshall
1982b;Goinetal.2007b,2016; Goin and Abello 2013). Thus,
there are major differences in molar morphology between
microbiotherians and polydolopimorphians, and on available ev-
idence it seems more plausible to me that polydolopimorphians
evolved from an ancestor with a more generalized
marsupialiform molar morphology, namely in which the stylar
cusps and conules were well developed, and the centrocrista was
v-shaped.
Other aspects of the dentition and cranium of microbiotherians
also differ markedly from those of Epidolops and other
plesiomorphic polydolopimorphians. Microbiotherians are un-
usual in that i2 is not staggered (Hershkovitz 1982,1995,
1999), whereas i2 appears to be staggered in several
polydolopimorphians that preserve the anterior end of the man-
dible, including E. ameghinoi (Fig. 7), Bonapartherium
hinakusijum,andPrepidolops spp. (see Pascual, 1980b,1981).
In the cranium, microbiotherians have large palatal vacuities and
a complete auditory bulla that encloses the hypotympanic sinus,
formed by an alisphenoid tympanic sinus and fused rostral and
caudal tympanic processes of the petrosal (Hershkovitz 1999;
Sánchez-Villagra and Wible 2002; Giannini et al. 2004), but they
lack a groove for the internal carotid artery on the promontorium
(Sánchez-Villagra and Wible 2002). As already noted, palatal
vacuities are tiny in both E. ameghinoi and B. hinakusijum,whilst
E. ameghinoi lacks an ossified floor to the hypotympanic sinus
(Figs. 3,5,and12a), and, if the Type II petrosals belong to this
taxon, also has a tiny rostral tympanic process of the petrosal and
a distinct groove for the internal carotid artery (Ladevèze 2004).
Finally, in the tarsus, the microbiotherian Dromiciops
gliroides exhibits the combination of a tripartite calcaneocuboid
facet (also present in paucituberculatans –see above) and
CLAJP (Fig. 9e; Szalay 1982,1994) characteristic of
australidelphians (Szalay 1982,1994;Becketal.2008b;Beck
2012). The IMG VII tarsals, which are probably referable to
Epidolops, lack both of these apomorphies (Fig. 9a; Szalay
1994).
Evidence for Diprotodontian Aff inities
A close relationship between Polydolopimorphia and the ex-
tant Australian order Diprotodontia has been proposed by
Goin and co-workers (Goin 2003; Goin and Candela 2004).
Recent phylogenies consistently place Diprotodontia and
Microbiotheria in the clade Australidelphia, and some have
supported a sister-taxon relationship between the two orders
(see Beck in press-b for a review). The hypothesis that
Polydolopimorphia and Microbiotheria are closely related
(discussed above) implies that Polydolopimorphia must also
be closely related to Diprotodontia. The shared presence of
diprotodonty in polydolopimorphians and diprotodontians
represents an obvious putative apomorphy supporting this re-
lationship. I have already discussed why I believe a close
relationship between Polydolopimorphia and Microbiotheria
J Mammal Evol (2017) 24:373–414 401
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

is unlikely (see above), which weakens support for
Polydolopimorphia-Diprotodontia link.
A consideration of the craniodental and tarsal morphology
of diprotodontians and polydolopimorphians also does not
support this proposed relationship. Diprotodontians and most
polydolopimorphians are diprotodont, but (as discussed
above) the absence of diprotodonty in Bonapartherium
hinakusijum and the ontogeny-related diprotodonty of
Prepidolops didelphoides indicate that this derived feature
must have originated independently in polydolopimorphians
and diprotodontians –unless, that is, Bonapartherium and
Prepidolops are early diverging members of a combined
Polydolopimorphia + Diprotodontia clade. In addition, the
craniodental anatomy of E. ameghinoi is very different and
far more plesiomorphic than any diprotodontian. As already
discussed, E. ameghinoi lacks an ossified hypotympanic sinus
floor and the Type II petrosals that probably belong to this
taxon lack a well-developed rostral tympanic process, whereas
all diprotodontians have an at least partially ossified bulla (as
do all crown marsupials –see below) and usually also a prom-
inent rostral tympanic process of the petrosal (Archer 1984a;
Aplin 1987,1990; Springer and Woodburne 1989). In contrast
to the Type II petrosals, a promontorial groove for the internal
carotid artery is usually absent, although it is observed in a few
taxa, e.g., some specimens of vombatids Vombatus ursinus
and Lasiorhinus latifrons (Sánchez-Villagra and Wible 2002;
Aplin 1987,1990). Again unlike the Type II petrosals, the
geniculate ganglion is roofed dorsally in all diprotodontians
that I have examined with the exceptions of the macropodoid
Notamacropus agilis and thylacoleonid Thylacoleo carnifex
(pers. obs.).
In the glenoid region, most diprotodontians exhibit a
distinct Bcomplex^morphology, with a raised articular em-
inence anteriorly and grooved mandibular fossa, and the
postglenoid foramen is usually in a medial position,
often in the posteromedial corner of the glenoid fossa
(Aplin 1987,1990; Springer and Woodburne 1989). In
Epidolops, the glenoid region is much more plesiomorphic:
the glenoid fossa forms a continuous plane, and the
postglenoid foramen opens posterior to the postglenoid
process (Figs. 5and 12a).
Tarsal morphology within Diprotodontia is highly vari-
able, reflecting the variety of locomotor modes observed
within the order. However, the tarsals of small-bodied
Bpossums^such as burramyids, acrobatids, pseudocheirids
and petaurids, closely resemble those of the
microbiotherian Dromiciops gliroides (Fig. 9e) and likely
approach the ancestral diprotodontian morphotype (Szalay
1982,1994). The calcanea of these Bpossums^share with
Dromiciops the apomorphies of a tripartite calcaneocuboid
facet and CLAJP (Szalay 1982,1994), neither of which are
present in the IMG VII calcanea referred by Szalay (1994)
to Epidolops (Fig. 9a).
Evidence for a Position Outside Marsupialia
A third hypothesis for the affinities of Polydolopimorphia is
that the order is not closely related to any other marsupial
order. Indeed, available evidence suggests to me that
Epidolops, and other taxa that are probably closely related
such as Bonapartherium,Prepidolops, and polydolopids,
most likely fall outside Marsupialia.
Most striking is the apparent absence of an ossified floor to
the hypotympanic sinus in E. ameghinoi.Presenceofan
alisphenoid tympanic process flooring at least the anterior part
of the hypotympanic sinus is probably plesiomorphic for
Marsupialia (Horovitz and Sánchez-Villagra 2003). An ossi-
fied hypotympanic sinus floor formed by the alisphenoid is
also present in some fossil metatherians that fall outside
Marsupialia in recent published phylogenetic analyses; these
include Asiatherium and the as-yet-named BGurlin Tsav
skull^from the Late Cretaceous of Mongolia (Szalay and
Trofimov 1996), and Herpetotherium fugax from the early
Oligocene of North America (Gabbert 1998; Sánchez-
Villagra et al. 2007;Horovitzetal.2008).
Several other non-marsupial metatherians, however, lack
an ossified hypotympanic sinus floor, namely the Cretaceous
Asian deltatheroidans (G.W. Rougier, pers. comm. in
Forasiepi 2009;Bietal.2015:fig.2E),Pucadelphys,
Andinodelphys,andMayulestes (Fig. 12b) from the early or
middle Paleocene Tiupampa Fauna of Bolivia (Muizon 1994,
1998; Marshall and Muizon 1995;Muizonetal.1997), the
early Eocene North American peradectid Mimoperadectes
(contra Horovitz et al. 2009 - see Horovitz et al. 2009:fig.
S3 and comments by Beck 2012: electronic supplementary
material and Jansa et al. 2014: supporting information) and
many South American sparassodonts (Muizon 1999;
Forasiepi 2009). Thus, the distribution of this feature within
Metatheria is complex and shows some homoplasy.
Nevertheless, the apparent absence of an ossified
hypotympanic sinus floor in E. ameghinoi is a striking feature
not seen in any crown marsupial.
Another potentially plesiomorphic feature of E. ameghinoi
is its very small palatal fenestrae. Palatal morphology is vari-
able among metatherians that have been found to lie
outside Marsupialia in recent phylogenetic analysis: palatal
fenestrae are absent in deltatheroidans (the sister-taxon of
Marsupialiformes; Rougier et al. 1998;Bietal.2015),
Pucadelphys,Mayulestes, and sparassodonts (Muizon 1994,
1998; Marshall and Muizon 1995;Forasiepi2009;Engelman
and Croft 2014; Forasiepi et al. 2015), but present in the
BGurlin Tsav skull,^various Late Cretaceous marsupialiforms
(e.g., stagodontids; Fox and Naylor 1995,2006),
herpetotheriids, peradectids, and Andinodelphys (Muizon
et al. 1997; Fox and Naylor 2006). This variability makes it
difficult to determine the polarity of this feature. However,
given their broad distribution among crown marsupials, it seems
402 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

likely that the presence of well-developed maxillopalatine fenes-
trae is plesiomorphic for Marsupialia. The very small size of
these fenestrae in E. ameghinoi (and in Bonapartherium
hinakuaijum and probably also in Kramadolops mckennai;
Pascual 1981; Flynn and Wyss 2004;Goinetal.2016:
fig. 5.10b) may be indicative of a position outside Marsupialia.
However, it should be noted that many crown marsupials have
secondarily lost or greatly reduced maxillopalatine fenestrae,
such as caluromyine didelphids (Voss and Jansa 2009)and
diprotodontoid diprotodontians (Archer 1984a), and reduction
in the size of these fenestrae may also have occurred in
polydolopimorphians.
A prominent transverse canal foramen is present in most
marsupials and is likely plesiomorphic for Marsupialia, al-
though the precise morphology and position of this foramen
varies among marsupials (Sánchez-Villagra 1998; Sánchez-
Villagra and Wible 2002; Horovitz and Sánchez-Villagra
2003). Among non-marsupial metatherians, a prominent
transverse canal foramen is present in the Herpetotherium
fugax,Mimoperadectes houdei,Andinodelphys
cochabambensis, and probably also Asiatherium reshetovi,
but it is absent in Mayulestes ferox (Fig. 12b) and probably
also the BGurlin Tsav skull^(Szalay and Trofimov 1996;
Muizon et al. 1997; Muizon 1998; Sánchez-Villagra et al.
2007; Horovitz et al. 2008,2009). Among sparassodonts,
which also fall outside Marsupialia (Rougier et al. 1998,
2004; Forasiepi 2009; Engelman and Croft 2014; Forasiepi
et al. 2015; Beck in press-b;), an obvious transverse canal
foramen is absent in some taxa (e.g., Arctodictis) but present
in others (e.g.,Prothylacynus;Forasiepi2009). The transverse
canal foramen has been reported as absent in Pucadelphys
andinus (see Marshall and Muizon 1995), but based on my
own examination of a large collection of crania (see Ladevèze
et al. 2011), a transverse canal foramen appears to be present
in a few specimens. The condition in the sister-taxon to
Marsupialiformes, Deltatheroida, is currently unknown. The
transverse canal foramen is either absent or tiny in
E. ameghinoi, in contrast to the prominent foramen observed
in most marsupials; this may be an indication that it lies out-
side Marsupialia, but (as for palatal vacuities) the distribution
of this feature shows considerable homoplasy.
The Type II petrosals that I argue probably belong to
E. ameghinoi also appear more plesiomorphic than those of
crown marsupials, most obviously in the very small rostral
tympanic process and groove for the internal carotid artery
(Ladevèze 2004); the published phylogenies that have includ-
ed the Type II petrosals place them outside Marsupialia,
either in a clade with the Tiupampan Pucadelphys and
Andinodelphys (Ladevèze 2004,2007; Ladevèze and
Muizon 2010), or in an even more basal position within
Marsupialiformes (Ladevèze and Muizon 2007).
Szalay (1994: 326) considered that the IMG VII tarsals
from Itaboraí that he referred to E. ameghinoi to be derivable
from a Bprimitive itaboraiform^ancestral morphology.
Szalay’s(1994) concept of BItaboraiformes^is explicitly
paraphyletic, representing a grade from which Marsupialia
presumably originated; an Bitaboraiform^ancestry for
Epidolops is therefore compatible with a position close to,
but outside, Marsupialia.
One striking probable tarsal apomorphy of Marsupialia is
complete superposition of the astragalus on the calcaneus
(Szalay 1984,1993; Horovitz 2000); in non-marsupial
metatherians such as the deltatheroidan Deltatheridium
and the marsupialiforms Pucadelphys,Andinodelphys,
Mayulestes,andHerpetotherium, the astragalus is positioned
more medially relative to the calcaneus, as indicated by the
orientation of the ectal and sustentacular facets, and is in great-
er contact with the substrate (Horovitz 2000;Szalayand
Sargis 2001,2006; Horovitz et al. 2008). It is therefore note-
worthy that the sustentacular facet of the IMG VII calcanea
faces distinctly medially, particularly at its anterior end
(Fig. 9a), implying incomplete superposition by the astraga-
lus. The angle formed by the ectal and sustentacular facets
also suggests incomplete superposition (Fig. 9a).
Presence of a very large astragalar medial plantar tubercle
is another indicator that the astragalus is not completely su-
perposed on the calcaneus (Szalay 1993,1994; Szalay and
Sargis 2001); Szalay (1994:176–177) identified two astragali
(DGM 1.148-N and 1.151-M) as belonging to IMG VII, but
did not illustrate or describe them in detail, and hence the size
of the astragalar medial plantar tubercle is unclear. However,
Szalay (1994:176–177) reported that these two astragali are
very similar in morphology to most of the other Itaboraian
marsupialiform astragali, which suggests that the astragalar
medial plantar tubercle in these specimens was large (Szalay
1994; Szalay and Sargis 2001). If, as this evidence suggests,
Epidolops lacked complete astragalar superposition, it would
provide further support for a position outside Marsupialia.
Australidelphian Tarsals from the Early-Middle
Eocene La Barda Locality
Lorente et al. (2016) described isolated australidelphian-type
tarsals from the early-middle Eocene (Lutetian) La Barda lo-
cality in Patagonia, and argued that they belong to one of four
taxa currently included in Polydolopimorphia, namely either
Gashternia (Gashterniidae), Polydolops (Polydolopidae),
Amphidolops (Polydolopidae), or Palangania (family incertae
sedis). The La Barda tarsals differ markedly from the IMG VII
that Szalay (1994)referredtoE. ameghinoi, and they were
placed within Diprotodontia in Lorente et al.‘s(2016)phylo-
genetic analysis. This suggests that either IMG VII or the La
Barda tarsals (or possibly both) do not belong to the dental
taxa to which they have been tentatively referred, or that
Polydolopimorphia as currently recognized is polyphyletic.
J Mammal Evol (2017) 24:373–414 403
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Gashternia and the polydolopids Polydolops and
Amphidolops share with Epidolops the apomorphic presence
of an enlarged and bladelike P3 (Figs. 11a, c-d), and I have
already discussed additional craniodental similarities between
Epidolops and Kramadolops, which is currently the best
known polydolopid genus (see BComparisons with Other
Taxa Currently Included in Polydolopimorphia^above).
Premolar morphology is unknown for Palangania, and sever-
al dental similarities between this taxon and microbiotherians
have been noted (Goin et al. 1998b;Goin2003). If IMG VII
represents E. ameghinoi, then I consider Palangania to be the
most plausible candidate for referral of La Barda tarsals, out of
the four taxa suggested by Lorente et al. (2016). However, this
remains speculative in the absence of associated postcranial
remains of the taxa under consideration.
Affinities of Argyrolagoids
The superfamily Argyrolagoidea includes some of the dentally
most derived taxa currently placed within Polydolopimorphia,
namely the families Argyrolagidae, Groeberiidae, Patagoniidae,
together with Pradens,Klohnia,andEpiklohnia, which are cur-
rently classified as Argyrolagoidea incertae sedis
(Goin et al. 2010,2016;Zimicz2011). Recently, Chimento
et al. (2014) argued that Groeberia and Patagonia are not
polydolopimorphians or even therian mammals, but are in fact
members on the non-therian order Gondwanatheria. A full reas-
sessment of their work is beyond the scope of the current paper;
however, I do not accept their conclusions, as briefly sum-
marized here.
Perhaps most importantly, Groeberia retains obvious traces
of a tribosphenic molar pattern (Patterson 1952;Simpson
1970a), whereas gondwanatherians are non-tribosphenic
(Gurovich 2006;Krause2014). In addition, the phylogenetic
analysis of Chimento et al. (2014) is flawed because it does
not include any other polydolopimorphians besides Groeberia
and Patagonia. Particularly problematic is the absence of
argyrolagids, which share with Groeberia and Patagonia a
hypsodont or hypselodont molar dentition and a gliriform
lower incisor, but which are nevertheless unequivocally
marsupialiform based on their cranial morphology (Simpson
1970b; Sánchez-Villagra and Kay 1997; Sánchez-Villagra
et al. 2000). With argyrolagids (and other key groups with
superficially similar dentitions, such as rodents) absent from
Chimento et al.‘s(2014) matrix, it is unsurprising that
Groeberia and Patagonia ended up grouping with the only
hypsodont/hypselodont taxa with a procumbent lower incisor
present, namely gondwanatherians. I prefer the interpretation
of most recent authors (e.g., Szalay 1994;Kirschetal.1997;
Goin et al. 2016,in press), namely that Groeberia and
Patagonia are marsupialiforms, although their precise
affinities will remain unclear without the discovery of more
complete material.
Argyrolagids are by far the best known argyrolagoids, rep-
resented by multiple crania and also postcranial remains, in
addition to plentiful dental material (Rusconi 1933; Simpson
1970b; Hoffstetter and Villarroel 1974; Villarroel and
Marshall 1988; Szalay 1994;Sánchez-Villagra and Kay
1997; Sánchez-Villagra et al. 2000; Sánchez-Villagra 2001;
Carlinietal.2007; Garcia-Lopez and Babot 2015;Babot
and García-López 2016). Comparison of these specimens with
known material of E. ameghinoi reveals numerous major
differences.
As discussed above, the cranium of E. ameghinoi exhibits a
number of strikingly plesiomorphic features, including
very small maxillopalatine fenestrae (as also seen in
Bonapartherium hinakuaijum and probably also the
polydolopid Kramadolops mckennai; Pascual 1981; Flynn
and Wyss 2004;Goinetal.2016:fig.5.10b), the
hypotympanic sinus appears to lack an ossified floor, and
the petrosal(if the Type II petrosals represent Epidolops)lacks
a rostral tympanic process but has a deep groove for the inter-
nal carotid artery. By contrast, argyrolagids are far more de-
rived: the maxillopalatine fenestrae are enormous (possibly
forming a single, confluent opening in the palate), the
hypotympanic sinus is enclosed ventrally by a very large
alisphenoid tympanic process, and the petrosal has a promi-
nent, elongate rostral tympanic process but lacks a deep
groove for the internal carotid artery (Fig. 12d; Simpson
1970b; Sánchez-Villagra and Kay 1997; Sánchez-Villagra
et al. 2000; Babot and García-López 2016).
In the mandible, E. ameghinoi has four large, procumbent
teeth anteriorly (i1–3 and c1) and lacks a retrodental foramen
(Figs. 6,7,and10a), whereas in argyrolagids the only pro-
cumbent tooth is a gliriform incisor, followed by one or more
very small unicuspids, and a very large retrodental foramen is
present (Figs. 10cand13a; Rusconi 1933;Simpson1970b;
Hoffstetter and Villarroel 1974; Sánchez-Villagra and Kay
1997; Sánchez-Villagra et al. 2000; Sánchez-Villagra 2001;
Carlini et al. 2007;GoinandAbello2013; Babot and García-
López 2016). Based on its alveolus, the C1 of E ameghinoi was
clearly a very large tooth (as in Bonapartherium hinakuaijum
and Hondonadia feruglioi; Pascual 1981; Goin and Candela
1998), whereas C1 of argyrolagids is either very small or entirely
absent (Simpson 1970b; Sánchez-Villagra and Kay 1997;
Sánchez-Villagra et al. 2000; Carlini et al. 2007). The P3 and
p3 of E. ameghinoi are enormous and bladelike (Figs. 2,3,6,
10a, and 11a), as they are in several other polydolopimorphians,
including Bonapartherium hinakuaijum (Fig. 11b), Prepidolops
spp., Gashternia carioca (Fig. 11c), and polydolopids (Fig. 11d),
although the precise morphology differs somewhat between taxa
(Pascual 1980a,b,1981; Marshall 1982a; Goin and Oliveira
2007). By contrast, P3 and p3 of argyrolagids are small,
hypsodont or hypselodont teeth (Figs. 10c, 11e, and 13a;
404 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Rusconi 1933;Simpson1970b; Hoffstetter and Villarroel 1974;
Sánchez-Villagra and Kay 1997; Sánchez-Villagra et al. 2000;
Sánchez-Villagra 2001;Carlinietal.2007; Goin and Abello
2013) that would appear more easily derived from a generalised
premolariform morphology than from enlarged bladelike precur-
sors. Finally, the IMG VII tarsal morphotype referred
E. ameghinoi by Szalay (1994) differs markedly from known
argyrolagid tarsals, most notably in that the calcaneocuboid facet
of Epidolops is a single facet whereas that of argyrolagids is more
derived in being tripartite and distinctly stepped (Fig. 9a-c).
A prominent foramen is present in the retromolar space of
most paucituberculatans that I have examined, namely
Caenolestes,Lestoros,Palaeothentes,Stilotherium, and some
but not all specimens of Rhyncholestes (see also Simpson
1970b; Voss and Jansa 2009; Ojala-Barbour et al. 2013), and
it also seems to present in Abderites (Abello and Rubilar-
Rogers 2012: fig. 6.4). CT scans demonstrate that this fora-
men leads into an elongate canal that extends ventrally and
connects to the mandibular canal, within the substance of the
dentary (Fig. 13c). Sánchez-Villagra et al. (2000: character 13)
scored this foramen and canal in Lestoros as homologous with
the retrodental canal of argyrolagids (Fig. 13a).However,the
retromolar canal of argyrolagids is very short and does not
connect with the mandibular canal, but instead opens on the
medial surface of the dentary, dorsal to the mandibular fora-
men (Fig. 13a). In addition to the retrodental canal, Babot and
García-López (2016) reported the presence of small foramina
in the retromolar space (immediately posterior to m4) in
argyrolagids; they argued that these small foramina, and not
the retrodental canal, are homologous with the foramen in the
retromolar space observed in paucituberculatans.
Scattered tiny foramina are present within the retromolar
fossa of several other marsupials (pers. obs.), but these ap-
pear to be nutrient foramina that are variable in number and
position, and which do not lead into a distinct canal, unlike
the single, relatively large foramen observed in
paucituberculatans. It is unclear whether the small
retromolar foramina in argyrolagids lead into a distinct
canal; if they do not, I consider it more likely that the dis-
tinct retromolar foramen and canal found in most
paucituberculatans is homologous with the retrodental ca-
nal of argyrolagids (contra Babot and García-López 2016),
despite the differences in position and morphology (see
Brocklehurst et al. 2016 for a discussion of the homology
of a morphologically similar canal, which they refer to as
the Bcoronoid canal,^in afrotherian placentals). If these
structures are indeed homologous, then they would repre-
sent a striking potential synapomorphy uniting
paucituberculatans and argyrolagids. However, for the phy-
logenetic analysis presented here, I have elected to score the
argyrolagids Argyrolagus and Proargyrolagus as having a
retrodental canal, and the paucituberculatans Caenolestes
and Palaeothentes as unknown for this character.
A tripartite calcaneocuboid facet is also present in
paucituberculatans and australidelphians (Fig. 9d-e; Szalay
1982,1994; Horovitz and Sánchez-Villagra 2003;Beck
2012). This feature has been identified as an australidelphian
apomorphy (Szalay 1982,1994; Szalay and Sargis 2006;
Beck et al. 2008b; Beck 2012), but Horovitz and Sánchez-
Villagra (2003: 185, fig. 3) noted that a tripartite
calcaneocuboid facet is shared by paucituberculatans and
australidelphians. Szalay and Sargis (2006: 205) downplayed
this apparent resemblance, writing that Bthe alleged special
similarity of the cuboid proximal surface of Caenolestes and
australidelphians is so completely out of context of the well-
understood dynamics of the entire tarsal character complex of
these taxa that it is difficult to comment on.^Szalay and Sargis
(2006) did not provide any specific evidence in support of this
conclusion, however, and comparison of the calcaneocuboid
facet of the calcanea of Caenolestes with a range of
australidelphians reveals obvious similarities in morphology
(see Fig. 9d-e). The fossil paucitiberculatan Palaeothentes al-
so has a tripartite calcaneocuboid facet (Abello and Candela
2010: 1523, fig. 8D).
The position of the root within Marsupialia has yet to be
confidently resolved. However, the retroposon analysis of
Gallus et al. (2015) found statistically significant support for
Didelphimorphia to be the first living order to diverge, leaving
Paucituberculata and Australidelphia as sister-taxa. A tripar-
tite calcaneocuboid facet is a therefore a potential morpholog-
ical synapomorphy for Paucituberculata + Australidelphia,
with the fossil argyrolagids included within this clade. The
tripartite calcaneocuboid facet morphology appears to have
evolved from a didelphimorphian-like bipartite precursor
(Szalay 1994; Beck 2012), supporting the hypothesis that
Didelphimorphia was the first living order to diverge within
Marsupialia (Gallus et al. 2015). Thus, the tarsal morphology
of argyrolagids is congruent with a close relationship to
paucituberculatans.
In summary, there is no compelling morphological evi-
dence that argyolagids are closely related to Epidolops and
craniodentally similar forms such as Bonapartherium and
polydolopids, which (as discussed above) may fall outside
Marsupialia. Instead, the known anatomy of argyrolagids sug-
gests that they are crown marsupials, and closely related to, or
within, Paucituberculata. If so, Polydolopimorphia sensu Goin
et al. (2016) is polyphyletic. This conclusion is somewhat sim-
ilar to that of Szalay (1994)and Kirsch et al. (1997)whoplaced
Argyrolagidae and also Gashterniidae, Groeberiidae, and
Patagoniidae closer to undoubted paucituberculatans than to
polydolopimorphians such as Epidolops and polydolopids
(see Table 1). However, I consider that Gashterniidae is prob-
ably more closely related to Epidolops (and dentally similar
forms such as Bonapartherium and polydolopids) than to
argyrolagids, based on the shared presence in Epidolops and
Gashternia of a large, bladelike P3. Szalay (1994:335)also
J Mammal Evol (2017) 24:373–414 405
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

argued that argyrolagids may have originated from Bpichipilin
caenolestines,^but I do not think there is sufficient evidence at
present to link argyrolagids with a specific paucituberculatan
clade.
Sánchez-Villagra (2001) presented the first formal phylo-
genetic analysis of argyrolagids, using dental, cranial, and
postcranial characters, and found that they formed the sister-
taxon of caenolestids (other paucituberculatan families were
not included), with strong support; the results of this analysis
are congruent with my own conclusions. The phylogenetic
analyses of Goin et al. (2009) and Chornogubsky and
Goin (2015), by contrast, supported monophyly of
Polydolopimorphia sensu Goin et al. (2016), i.e., including
argyrolagids.
However, Goin et al. (2009) and Chornogubsky and Goin
(2015) relied almost exclusively on dental characters, and
interpreting the molar morphology of argyrolagids is difficult:
their molars of are highly derived and, because they are
hypsodont or hypselodont, their occlusal morphology is soon
lost through wear (Simpson 1970b; Sánchez-Villagra et al.
2000;Zimicz2011;GoinandAbello2013). To my knowl-
edge, the only argyrolagid molar to be described to date that is
sufficiently unworn to preserve distinct cusps is an m1 of
Proargyrolagus bolivianus (see Goin and Abello 2013:
figs. 1.14 and 4.1); all other specimens have lost their cusps
through wear, and hence uncertainties remain regarding cusp
homologies (Sánchez-Villagra and Kay 1997; Zimicz 2011;
Goin and Abello 2013). A priori interpretations regarding
cusp homologies that lack clear evidential support may bias
phylogenetic analyses towards particular topologies (O’Meara
and Thompson 2014). Assumptions that, for example, the
labialmost cuspid of the trigonid in argyrolagids is a
neomorphic Bectostylid^(as in Zimicz 2011; Goin and
Abello 2013) rather than the protoconid (as in Sánchez-
Villagra and Kay 1997), remain questionable in the absence
of unworn argyrolagid dentitions that would clarify the topo-
logical and occlusal relations between cusps.
Ultimately, the hypothesis of polydolopimorphian
polyphyly will require further testing. CT data from
argyrolagid crania are likely to prove particularly useful, as
will the discovery of more complete remains (ideally including
associated poscranial material) of other polydolopimorphians.
The affinities of argyrolagids could also potentially be tested
by molecular data: the youngest known argyolagids are from
the late Pliocene (Marplatan South American Land Mammal
Age; Goin et al. 2016), ~3.3–2.0 MYA, which is considerably
older than the oldest successfully sequenced DNA (~430 kya;
Dabney et al. 2013; Meyer et al. 2014,2016) but younger than
the oldest collagen peptide sequences (Rybczynski et al. 2013)
obtained from fossils to date. Thus it may be possible to obtain
collagen peptide sequences from late Pliocene argyolagids. If
my hypothesis is correct, such sequences should form a clade
with those from the living paucituberculatan caenolestids.
Alternatively, if Goin et al. (2016) are correct, they should be
more closely related to those of the living microbiotherian
Dromiciops and diprotodontians, within the clade
Australidelphia.
Results of Phylogenetic Analyses
The results of the Bayesian non-clock total evidence analyses
are shown in Fig. 14. Although taxon sampling is limited, the
results for both versions of the total evidence matrix (i.e.,
either using the Type II petrosals and IMG VII tarsals to score
characters for Epidolops or not) are congruent with the qual-
itative comparisons presented above. Most significantly,
Polydolopimorphia sensu Goin et al. (2016) is diphyletic in
both analyses, and Epidolops does not form a clade with either
paucituberculatans (represented here by the extant caenolestid
Caenolestes and the fossil palaeothentid Palaeothentes),
microbiotherians (represented here by the extant
microbiotheriid Dromiciops), or diprotodontians.
When the Type II petrosals and IMG VII tarsals are not
used for scoring Epidolops, the resultant phylogeny is highly
unresolved (Fig. 14a) with Epidolops part of a large polytomy
that includes crown marsupials and several probable stem
marsupials (Asiatherium,Herpetotherium, Peradectidae).
Argyrolagidae is monophyletic (BPP = 1.00) and forms a
relatively strongly-supported clade (BPP = 0.89) with the
paucituberculatans Caenolestes and Palaeothentes,congruent
with the earlier analysis of Sánchez-Villagra (2001). The
Australian Yalkaparidon is sister to this clade, albeit with
weak support (BPP = 0.54).
When the Type II petrosals and IMG VII tarsals are used to
score Epidolops, the resultant phylogeny is somewhat more
resolved (Fig. 14b). Epidolops is in a clade with Mayulestes,
Andinodelphys,andPucadelphys (BPP = 0.89), all of which
are from the early or middle Paleocene Tiupampa fauna of
Bolivia. As in the other analysis (Fig. 14a). Argyrolagidae is
monophyletic (BPP = 1.00) and forms a clade with the
paucituberculatans Caenolestes and Palaeothentes
(BPP = 0.90), with Yalkaparidon sister to this clade
(BPP = 0.88). A full list of the synapomorphies supporting
the clades present in both phylogenies (under both accelerated
and delayed transformation) is given in the Electronic
Supplementary Material.
Implications for the Biogeographical Origin
and Early Evolution of Marsupialia
Recent molecular divergence dates (e.g., Beck 2008; Meredith
et al. 2009,2011; Mitchell et al. 2014) suggest that
Marsupialia had originated and the modern orders had di-
verged from each other prior to the current estimate for the
406 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

age of the Itaboraian, namely 50–53 Ma; as such, crown mar-
supials could, in principle, be present at Itaboraí. As already
discussed, I conclude that Epidolops is not a crown marsupial.
But what about the other marsupialiform taxa? Voss and Jansa
(2009) observed that loss of the postcingulid from the lower
molars optimizes as a synapomorphy of Marsupialia, although
a postcingulid is present in dasyurids, thylacinids, and the
stem australidelphian Djarthia murgonensis, indicating a de-
gree of homoplasy in this feature (Beck in press-b).
Nevertheless, the presence of a postcingulid in several of
the Itaboraí marsupialiform taxa (e.g., Bobbschaefferia
fluminensis, Gaylordia mater, Minusculodelphis modicum,
Protodidelphis mastodontoides; Oliveira and Goin 2011,
2015;Oliveiraetal.2016) suggests that they may fall outside
Marsupialia. This conclusion receives support from recent
studies indicating that Gaylordia is a stem marsupial
(Oliveira and Goin 2015) and that Minisculodelphis cannot
be confidently placed within Marsupialia (Oliveira et al.
2016).
Several other Itaboraí marsupialiforms lack a postcingulid
(e.g., Guggenheimia crocheti,Procaroloameghinia pricei,
Protodidelphis vanzolinii; Oliveira and Goin 2011), and so
are better candidates for being members of Marsupialia.
However, loss of the third trochanter of the femur character-
izes crown marsupials except paucituberculatans (given my
proposal that argyrolagids are probably members of
Paucituberculata, it is interesting to note that the femur of
the argyrolagid Argyrolagus scagliai has a third trochanter;
Simpson 1970b) and a few Australian taxa that are secondarily
specialized for fossoriality (Szalay and Sargis 2001;Horovitz
Fig. 14 Phylogenetic relationships of Epidolops, argyrolagids, and other
metatherians, based on Bayesian undated analyses of a total evidence
matrix modified from Beck et al. (2014); both analyses comprised four
independent runs of 50 × 10
6
generations each, sampling trees every 2000
generations, and discarding the first 25 % (i.e., 12.5 × 10
6
generations) as
burn-in. a50 % majority rule consensus of post-burn-in trees from
analysis of matrix in which the Type II petrosals and IMG VII tarsals
from Itaboraí were not used to score characters for Epidolops (BMatrix
A^); the harmonic mean of lnL across all four runs was −62,311.01; b
50 % majority rule consensus of post-burn-in trees from analysis of
matrix in which the Type II petrosals and IMG VII tarsals from Itaboraí
were used to score characters for Epidolops (BMatrix B″); the harmonic
mean of lnL across all four runs was −62,336.95. Values at nodes are
Bayesian posterior probabilities. Extinct taxa are identified by daggers,
and Epidolops and the argyrolagids are highlighted in bold
J Mammal Evol (2017) 24:373–414 407
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

et al. 2008; Abello and Candela 2010;Becketal.2016).
Szalay and Sargis (2001) described ten marsupialiform femur
morphotypes, all of which exhibit a distinct third trochanter,
suggesting that they represent stem marsupials or possibly
paucituberculatans.
Riolestes capricornis was described as a paucituberculatan
based on a single lower molar from Itaboraí (Goin et al. 2009),
but this specimen may in fact be a dp3 of another, already
named marsupialiform taxon (indeed, this possibility was
considered by Goin et al. 2009). Szalay (1994)proposedthat
Carolopaulacoutoia (=BSternbergia^)itaboraiensis may be a
paucituberculatan, but subsequent studies have disagreed with
this conclusion (Goin 2003; Oliveira and Goin 2011). The
peculiar Deroryhnchus singularis has large and procumbent
anterior incisors (Paula Couto 1952b; Marshall 1987), and
Goin et al. (2009:872–873) noted that derorhynchid molar
morphology Banticipates^that seen in paucituberculatans.
Congruent with this, the phylogenetic analysis of Goin et al.
(2009), which was based on craniodental characters, placed
this taxon as sister to Paucituberculata. Szalay (1994:333) also
observed that the IMG 1 tarsal morphotype, which is of an
appropriate size for referral to D. singularis,Bshares some of
the primitive [tarsal] pattern with caenolestids.^However,
D. singularis also has an enlarged, procumbent canine
(Paula Couto 1952b; Marshall 1987), whereas this tooth is
greatly reduced or absent in definitive paucituberculatans
(Marshall 1980;Martin2007,2013;Abello2013). In addi-
tion, Forasiepi and Rougier (2009) tentatively referred an iso-
lated metatherian petrosal (MPEF-PV 2235) from the early
Paleocene (Clyde et al. 2014) Punta Peligro locality in south-
ern Argentina to Derorhynchus aff. D. minutus. The petrosal
preserves apparently plesiomorphic features not seen in crown
marsupials. If this specimen does represent Derorhynchus aff.
D. minutus, it suggests that derorhynchids probably lie outside
Marsupialia. Other published phylogenetic analyses that have
included Derorhynchus have not supported a close relation-
ship with Paucituberculata (Goin et al. 2006; Ladevèze and
Muizon 2010; Forasiepi et al. 2013). No other putative
paucituberculatans have been identified at Itaboraí based on
dental remains; it therefore seems likely that the femoral re-
mains (all of which retain a third trochanter) represent stem
marsupials.
Szalay (1994) identified two marsupialiform tarsal
morphotypes from Itaboraí, IMGs V and XII, as representing
didelphimorphians based on the presence of a bipartite
calcaneocuboid joint morphology similar to that of living
didelphids. The didelphid-like bipartite morphology is
plausibly ancestral to the tripartite morphology seen in
paucituberculatans (including argyrolagids; see above) and
australidelphians (Szalay 1994; Beck 2012), in which case
its reported presence in IMG V and XII does not necessarily
imply that the taxa represented by these tarsal remains are
members of Didelphimorphia. In any case, Szalay and
Sargis (2001: 257) wrote that they now doubted that IMG V
and XII represent didelphimorphians, although they did not
give any details regarding this change of opinion, nor did they
propose alternative affinities for these specimens. A reapprais-
al of these potentially highly significant specimens is desper-
ately needed. Regardless, IMG V and XII cannot be regarded
as representing unambiguous crown marsupials.
The phylogenetic analyses of Ladevèze and Muizon
(2010), Oliveira and Goin (2011), and Oliveira et al. (2016)
suggest that some of the Itaboraí marsupialiforms are crown
marsupials, but these analyses show major conflicts both with
each other and with more comprehensive metatherian
phylogenies. For example, Ladevèze and Muizon (2010)
found Gaylordia to be a crown marsupial, whereas Oliveira
and Goin (2015) found it to fall outside Marsupialia, whilst the
Paucituberculata + Peramelemorphia clade recovered by
Ladevèze and Muizon (2010) has not been found in other
recent analyses (see Beck in press-b). In summary, then, two
of the best-preserved Itaboraí marsupialiforms, namely
Epidolops and Gaylordia, appear to fall outside Marsupialia,
and none of the remaining taxa can be confidently identified
as crown marsupials.
The oldest known definitive crown marsupials are the stem
australidelphian Djarthia murgonensis and an isolated
Bameridelphian^marsupial calcaneus, both from the
54.6 Ma old Tingamarra fauna in northeasten Australia
(Beck et al. 2008b; Beck 2012). The oldest unequivocal
crown marsupials from South America, meanwhile are from
the BSapoan^(= 47–49 Ma old) Laguna Fría and La Barda
localities in Chubut Province, southern Argentina, namely the
paucitutuberculatan Bardalestes, microbiotherians, and isolat-
ed tarsals of an australidelphian (Goin et al. 2009; Tejedor
et al. 2009; Lorente et al. 2016).
Influenced by the work of Morrone (2002,2004,2006),
Goin and co-workers have emphasized in recent publications
(Goin et al. 2007b, 2012a, 2016, in press; Lorente et al. 2016)
that South America should be viewed as comprising two dis-
tinct biogeographical Bkingdoms^(Goin et al. 2012a:fig.3.1;
Goin et al. 2016: fig. 4.2): northern South America is part of
the Holotropical Kingdom, whilst southern South America is
part of the Austral Kingdom, which also includes Antarctica
and Australia. Goin et al. (2007b,2016,in press) and Lorente
et al. (2016) proposed that the origin and early evolution of
Australidelphia occurred in the Austral Kingdom.
As reviewed above, unequivocal crown marsupials have
not been identified at Itaboraí, which falls within the
Holotropical Kingdom, whereas they are known from
similarly-aged sites in the Austral Kingdom (Tingamarra,
Laguna Fría, and La Barda). Definitive marsupials
(paucituberculatans and microbiotherians) are known from
Santa Rosa, in the Amazon Basin of eastern Peru (Goin and
Candela 2004), which is part of the Holotropical Kingdom,
but this site is considerably younger than Itaboraí, namely
408 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

middle or late Eocene or early Oligocene. Collectively, this
raises the possibility that the origin and early evolution of
Marsupialia as a whole, rather than just Australidelphia, was
restricted to the Austral Kingdom. This hypothesis needs to be
tested by the discovery and full description of Late Cretaceous
and early Paleogene mammal faunas from areas that lie within
the Holotropical Kingdom (for example in the Bogotá
Formation of Colombia; Bloch et al. 2012) and from the
Austral Kingdom (e.g., the Las Flores and Punta Peligro
faunas of southern Argentina; Goin et al. 2002; Forasiepi
and Rougier 2009;Goinetal.2016). However, it may be that
much of the early evolution of marsupials occurred in regions
for which the Late Cretaceous-early Paleogene fossil record of
mammals is poor (Australia) or as yet non-existent (mainland
Antarctica).
Acknowledgments My thanks to the staff at the Museu de Ciências da
Terra (Departamento Nacional de Produção Mineral), Museu Nacional do
Rio de Janeiro, and Museo Municipal de Ciencias Naturales BLorenzo
Scaglia,^Mar del Plata, for facilitating my visits to their collections and
for their warm welcome. My particular thanks to Rodrigo Machado
(DNPM) for giving me permission to describe DGM 321-M, and also
for going far beyond the call of duty bylending me a pair of trousers when
I was denied entry to DNPM for transgressing Brazilian governmental
regulations by wearing shorts –let this be a warning for future visitors to
DNPM! I thank Marcelo Weksler, Eugenia Zandonà, and Andre Pinheiro
for hospitality during my visits to Rio de Janeiro. Sandrine Ladevèze
generously supplied high quality images of the Type II petrosals from
Itaboraí. This study has benefitted from discussions with Laura
Chornogubsky, Pancho Goin, Lílian Bergqvist, John Flynn, and
Rebecca Pian, although they may find much to disagree with in the
published paper. I am also extremely grateful to two anonymous re-
viewers, and the editor, John Wible, for their meticulous and constructive
reviews, which greatly improved both the structure and the content of the
final paper. Funding for this research has been provided by the Australian
Research Council (via Discovery Early Career Researcher Award
DE120100957) and the University of Salford.
Open Access This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a link
to the Creative Commons license, and indicate if changes were made.
References
Abbie AA (1939) A masticatory adaptation peculiar to some diprotodont
marsupials. Proc Zool Soc Lond B109(2):261–279
Abello MA (2013) Analysis of dental homologies and phylogeny of
Paucituberculata (Mammalia: Marsupialia). Biol J Linn Soc
109(2):441–465
Abello MA, Candela AM (2010) Postcranial skeleton of the Miocene
marsupial Palaeothentes (Paucituberculata, Palaeothentidae): paleo-
biology and phylogeny. J Vertebr Paleontol 30(5):1515–1527
Abello MA, Rubilar-Rogers D (2012) Revisión del género Abderites
Ameghino, 1887 (Marsupialia, Paucituberculata). Ameghiniana
48(3):605–620
Ameghino F (1902) Notices préliminaires sur les mammifères nouveaux
des terrains crétacés de Patagonie. Boletín de la Academia Nacional
de Ciencias de Córdoba 17:5–70
Ameghino F (1903) Los diprotodontes del orden de los plagiaulacoideos
y el origen de los roedores y de los polimastodontes. An Soc Cient
Argent 56:193–208
Aplin KP (1987) Basicranial anatomy of the early Miocene
diprotodontian Wynyardia bassiana (Marsupialia: Wynyardiidae)
and its implications for wynyardiid phylogeny and classification.
In: Archer M (ed) Possums and Opossums: Studies in Evolution.
Surrey Beatty & Sons, Sydney, pp 369–391
Aplin KP (1990) Basicranial regions of diprotodontian marsupials: anat-
omy, ontogeny and phylogeny. Unpublished Ph.D. Dissertation,
University of New South Wales, School of Biological Sciences,
Sydney
Aplin K, Archer M (1987) Recent advances in marsupial systematics with
a new syncretic classification. In: Archer M (ed) Possums and
Opossums: Studies in Evolution. Surrey Beatty & Sons, Sydney,
pp xv–lxxii
Aplin KP, Helgen KM, Lunde DP (2010) A review of Peroryctes
broadbenti, the giant bandicoot of Papua New Guinea. Am Mus
Novitates 3696:1–41
Archer M (1984a) The Australian marsupial radiation. In: Archer M,
Clayton G (eds) Vertebrate Zoogeography and Evolution in
Australasia. Hesperian Press, Perth, pp 633–808
Archer M (1984b) Origins and early radiations of marsupials. In: Archer
M, Clayton G (eds) Vertebrate Zoogeography and Evolution in
Australasia. Hesperian Press, Perth, pp 585–631
Babot MJ, García-López DA (2016) Redescription of the argyrolagid
Microtragulus bolivianus (Metatheria, Polydolopimorphia,
Bonapartheriiformes) based on new remains from northwestern
Argentina. Palaeontol Electron 19.2.23A:1–23
Beck RMD (2008) A dated phylogeny of marsupials using a molecular
supermatrix and multiple fossil constraints. J Mammal 89(1):175–189
Beck RMD (2012) An ‘ameridelphian’marsupial from the early Eocene
of Australia supports a complex model of Southern Hemisphere
marsupial biogeography. Naturwissenschaften 99(9):715–729
Beck RMD (in press-a) The biogeographical history of non-marine
mammaliaforms in the Sahul region. In: Ebach MC (ed)
Handbook of Australasian Biogeography. CRC Press, Boca Raton
Beck RMD (in press-b) Current understanding of the phylogeny of
Metatheria: a review. In: Goin FJ, Forasiepi AM (eds) New World
Marsupials and Their Extinct Relatives: 100 Million Years of
Evolution. Springer, Berlin
Beck RMD, Archer M, Godthelp H, Mackness BS, Hand SJ, Muirhead J
(2008a) A bizarre new family of Marsupialia (incertae sedis) from
the early Pliocene of northeastern Australia: implications for the
phylogeny of bunodont marsupials. J Paleontol 82:749–762
Beck RMD, Godthelp H, Weisbecker V, Archer M, Hand SJ (2008b)
Australia’s oldest marsupial fossils and their biogeographical impli-
cations. PLoS ONE 3(3):e1858
Beck RMD, Travouillon KJ, Aplin KP, Godthelp H, Archer M (2014)
The osteology and systematics of the enigmatic Australian Oligo-
Miocene metatherian Yalkaparidon (Yalkaparidontidae;
Yalkaparidontia; ?Australidelphia; Marsupialia). J Mammal Evol
21(2):127–172
Beck RMD, Warburton NM, Archer M, Hand SJ, Aplin KP (2016) Going
underground: postcranial morphology of the early Miocene marsu-
pial mole Naraboryctes philcreaseri and the evolution of fossoriality
in notoryctemorphians. Mem Mus Vic 74: 151–171
Bergqvist LP (2008) Postcranial skeleton of the upper Paleocene
(Itaboraian) BCondylarthra^(Mammalia) of Itaboraí Basin, Brazil.
In: Sargis EJ, Dagosto M (eds) Mammalian Evolutionary
Morphology: a Tribute to Frederick S. Szalay. Springer,
Dordrecht, pp 107–133
J Mammal Evol (2017) 24:373–414 409
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Bergqvist LP, Abrantes EAL, Avilla L (2004) The Xenarthra (Mammalia)
of São José de Itaboraí Basin (upper Paleocene, Itaboraian), Rio de
Janeiro, Brazil. Geodiversitas 26:323–337
Bergqvist LP, Mansur K, Rodrigues MA, Rodrigues-Francisco BH, Perez
RAR, Beltrão MC (2008) Itaboraí Basin, State of Rio de Janeiro -
The cradle of mammals in Brazil. In: Winge M, Schobbenhaus C,
Souza CRG, Fernandes ACS, Berbert-Born M, Queiroz ET (eds)
Sítios Geológicos e Paleontológicos do Brasil SIGEP 123. CPRM,
Brasılia, pp 413–432
Bi S, Jin X, Li S, Du T (2015) A new Cretaceous metatherian mammal
from Henan, China. PeerJ 3:e896
Bloch JI, Rincon AF., Head JJ, Herrera F, Jaramillo CR (2012) Early
Eocene mammals from the hot tropics of northern South America.
In: 72nd Annual Meeting Society of Vertebrate Paleontology,
Raleigh, North Carolina, 2012, p 64
Brocklehurst RJ, Crumpton N, Button E, Asher RJ (2016) Jaw anatomy
of Potamogale velox (Tenrecidae, Afrotheria) with a focus on cranial
arteries and the coronoid canal in mammals. PeerJ 4:e1906
Bucher JE, Hoffmann RS (1980) Caluromys derbianus. Mammal Species
140:1–4
Cáceres NC, Carmignotto AP (2006) Caluromys lanatus.Mammal
Species 803:1–6
Carlini AA, Pascual R, Goin FJ (2007) A new argyrolagid marsupial from
the early Miocene of Patagonia (Argentina). Neues Jahrbuch für
Geologie und Paläontologie - Abhandlungen 245(3):323–330
Case JA, Goin FJ, Woodburne MO (2005) BSouth American^marsupials
from the Late Cretaceous of North America and the origin of mar-
supial cohorts. J Mammal Evol 12(3–4):461–494
Chimento NR, Agnolin FL, Novas FE (2014) The bizarre ‘metatherians’
Groeberia and Patagonia, late surviving members of
gondwanatherian mammals. Hist Biol 27(5):603–623
Chornogubsky L (2010) Sistemática de la familia Polydolopidae
(Mammalia, Marsupialia, Polydolopimorphia) de América del Sur
y la Antártida. Unpublished Ph.D. thesis, Universidad de Buenos
Aires
Chornogubsky L, Goin FJ (2015) A review of the molar morphology and
phylogenetic affinities of Sillustania quechuense (Metatheria,
Polydolopimorphia, Sillustaniidae), from the early Paleogene of
Laguna Umayo, southeastern Peru. J Vertebr Paleontol 35(6):
e983238
Chornogubsky L, Goin FJ, Reguero M (2009) A reassessment of
Antarctic polydolopid marsupials (middle Eocene, La Meseta
Formation). Antarct Sci 21(3):285
Cifelli RL (1983a) Eutherian tarsals from the late Paleocene of Brazil. Am
Mus Novitates 2761:1–31
Cifelli RL (1983b) The origin and affinities of the South American
Condylarthra and early Tertiary Litopterna (Mammalia). Am Mus
Novitates 2772:1–49
Cifelli RL, Muizon C de (1997) Dentition and jaw of Kokopellia juddi,a
primitive marsupial or near-marsupial from the Medial Cretaceous
of Utah. J Mammal Evol 4(4):241–258
Clyde WC, Wilf P, Iglesias A, Slingerland RL, Barnum T, Bijl PK,
Bralower TJ, Brinkhuis H, Comer EE, Huber BT, Ibañez-Mejia M,
Jicha BR, Krause JM, Schueth JD, Singer BS, Raigemborn MS,
Schmitz MD, Sluijs A, Zamaloa MC (2014) New age constraints
for the Salamanca Formation and lower Río Chico Group in the
western San Jorge Basin, Patagonia, Argentina: implications for
Cretaceous-Paleogene extinction recovery and land mammal age
correlations. Geol Soc Am Bull 126(3–4):289–306
Dabney J, Knapp M, Glocke I, Gansauge MT, Weihmann A, Nickel B,
Valdiosera C, Garcia N, Paabo S, Arsuaga JL, Meyer M (2013)
Complete mitochondrial genome sequence of a middle Pleistocene
cave bear reconstructed from ultrashort DNA fragments. Proc Natl
Acad Sci USA 110 (39):15758–15763
Engelman RK, Croft DA (2014) A new species of small-bodied sparassodont
(Mammalia, Metatheria) from the middle Miocene locality of Quebrada
Honda, Bolivia. J Vertebr Paleontol 34(3):672–688
Flynn JJ, Wyss AR (1999) New marsupials from the Eocene-Oligocene
transition of the Andean Main Range, Chile. J Vertebr Paleontol
19(3):533–549
Flynn JJ, Wyss AR (2004) A polydolopine marsupial skull from the
Cachapoal Valley, Andean Main Range, Chile. Bull Am Mus Nat
Hist 285:80–92
Forasiepi AM (2009) Osteology of Arctodictis sinclairi (Mammalia,
Metatheria, Sparassodonta) and phylogeny of Cenozoic metatherian
carnivores from South America. Monogr Mus Argent Cienc Nat 6:
1–174
Forasiepi AM, Babot MJ, Zimicz N (2015) Australohyaena antiqua
(Mammalia, Metatheria, Sparassodonta), a large predator from the
late Oligocene of Patagonia. J Syst Palaeontol 13(6): 502–525
Forasiepi AM, Goin FJ, Abello MA, Cerdeño E (2013) A unique, late
Oligocene shrew-like marsupial from western Argentina and the
evolution of dental morphology. J Syst Palaeontol 12(5): 549–564
Forasiepi AM, Rougier GW (2009) Additional data on early Paleocene
metatherians (Mammalia) from Punta Peligro (Salamanca
Formation, Argentina): comments based on petrosal morphology. J
Zool Syst Evol Res 47(4):391–398
Forasiepi AM, Sánchez-Villagra MR, Schmelzle T, Ladevèze S, Kay RF
(2014) An exceptionally well-preserved skeleton of Palaeothentes
from the early Miocene of Patagonia, Argentina: new insights into
the anatomy of extinct paucituberculatan marsupials. Swiss J
Palaeontol 133(1):1–21
Fox RC, Naylor BG (1995) The relationships of the Stagodontidae, prim-
itive North American Late Cretaceous mammals. In: Sun A, Wang
Y (eds) Sixth Symposium on Mesozoic Terrestrial Ecosystems and
Biota. China Ocean Press, Beijing, pp 247–250
Fox RC, Naylor BG (2006) Stagodontid marsupials from the Late
Cretaceous of Canada and their systematic and functional implica-
tions. Acta Palaeontol Pol 51(1):13–36
Gabbert S (1998) Basicranial anatomy of Herpetotherium (Marsupialia:
Didelphimorphia) from the Eocene of Wyoming. Am Mus Novitates
3235:1–13
Gallus S, Janke A, Kumar V, Nilsson MA (2015) Disentangling the rela-
tionship of the Australian marsupial orders using retrotransposon and
evolutionary network analyses. Genome Biol Evol 7(4):985–992
Garcia-Lopez DA, Babot MJ (2015) A late Miocene Argyrolagidae
(Mammalia, Metatheria, Bonapartheriiformes) from northwestern
Argentina. Ameghiniana 52(3):314–323
Gayet M, Marshall LG, Sempere T (1991) The Mesozoic and Paleocene
vertebrates of Bolivia and their stratigraphic context: a review. In:
Suarez-Soruco R (ed) Fosiles y facies de Bolivia - Vol. I Vertebratos,
vol 12. Revista Technica de Yacimientos Petroliferos Fiscales
Bolivianos, Santa Cruz, pp 393–433
Gelfo JN, Goin FJ, Woodburne MO, Muizon C de (2009)
Biochronological relationships of the earliest South American
Paleogene mammalian faunas. Palaeontology 52(1):251–269
Gelfo JN, Pascual R (2001) Peligrotherium tropicalis (Mammalia,
Dryolestida) from the early Paleocene of Patagonia, a survival from
a Mesozoic Gondwanan radiation. Geodiversitas 23(3):369–379
Giannini NP, Abdala F, Flores DA (2004) Comparative postnatal ontog-
eny of the skull in Dromiciops gliroides (Marsupialia:
Microbiotheriidae). Am Mus Novitates 3460:1–17
Goin FJ (2003) Early marsupial radiations in South America. In: Jones M,
Dickman C, Archer M (eds) Predators with Pouches: the Biology of
Carnivorous Marsupials. CSIRO (Commonwealth Scientific and
Industrial Research Organization), Collingwood, pp 30–42
Goin FJ, Abello MA (2013) Los Metatheria sudamericanos de comienzos
del Neógeno (Mioceno temprano, edad mamífero Colhuehuapense).
Parte 2: Microbiotheria y Polydolopimorphia. Ameghiniana 50(1):
51–78
410 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Goin FJ, Abello MA, Chornogubsky L (2010) Middle Tertiary marsu-
pials from central Patagonia (early Oligocene of Gran Barranca):
understanding South America’sGrande Coupure. In: Madden RH,
Carlini AA, Vucetich MG, Kay RF (eds) The Paleontology of Gran
Barranca: Evolution and Environmental Change through the Middle
Cenozoic ofPatagonia. Cambridge University Press, Cambridge, pp
71–107
Goin FJ, Candela AM (1995) Una nueva especie de Epidolops Paula
Couto, 1952 (Marsupialia, Polydolopimorphia, Polydolopidae).
Consideraciones sobre el patrón molar inferior de los
Epidolopidae. VI Congreso Argentino de Paleontología y
Bioestratigrafía, Trelew, Actas:143–148
Goin FJ, Candela AM (1996) A new early Eocene polydolopimorphian
(Mammalia, Marsupialia) from Patagonia. J Vertebr Paleontol 16(2):
292–296
Goin FJ, Candela AM (1998) Dos nuevos marsupiales
Bpseudodiprotodontes^del Eoceno de Patagonia, Argentina.
Publicación Especial de la Asociación Paleontológica Argentina 5:
79–84
Goin FJ, Candela AM, Abello MA, Oliveira EV (2009) Earliest South
American paucituberculatans and their significance in the under-
standing of ‘pseudodiprotodont’marsupial radiations. Zool J Linn
Soc 155:867–884
Goin FJ, Candela AM (2004) New Paleogene marsupials from the
Amazon Basin of Eastern Peru. In: Campbell Jr KE (ed) The
Paleogene Mammalian Fauna of Santa Rosa, Amazonian Peru.
Natural History Museum of Los Angeles County, Science Series
40, Los Angeles, pp 15–60
Goin FJ, Candela AM, Bond M, Pascual R, Escribano V (1998b) Una
nueva Bcomadreja^(Mammalia, Marsupialia, ?Polydolopimorphia)
del Paleoceno de Patagonia, Argentina. Publicación Especial de la
Asociación Paleontológica Argentina 5:71–78
Goin FJ, Candela AM, Forasiepi A (1997) New, middle Paleocene mar-
supials from central Patagonia. J Vertebr Paleontol 17(3):49A
Goin FJ, Candela A, López G (1998a) Middle Eocene marsupials from
Antofagasta de la Sierra, northwestern Argentina. Geobios 31(1):
75–85
Goin FJ, Candela AM, Muizon C de (2003a) The affinities of
Roberthoffstetteria nationalgeographica (Marsupialia) and the ori-
gin of the polydolopine molar pattern. J Vertebr Paleontol 23(4):
869–976
Goin FJ, Case JA, Woodburne MO, Vizcaino SF, Reguero MA (1999)
New discoveries of Bopposum-like^marsupials from Antarctica
(Seymour Island, Medial Eocene). J Mammal Evol 6(4):335–365
Goin FJ, Forasiepi AM, Candela AM, Ortiz Jaureguizar E, Pascual R,
Archer M, Godthelp H, Muirhead J, Augee M, Hand S, Wroe S
(2002) Earliest Paleocene marsupials from Patagonia. In: I
International Palaeontological Congress, Sydney, Sydney, 2002, p 68
Goin FJ, Gelfo JN, Chornogubsky L, Woodburne MO, Martin T (2012a)
Origins, radiations, and distribution of South American mammals:
from greenhouse to icehouse worlds. In: Patterson BD, Costa LP
(eds) Bones, Clones, and Biomes: an 80 Million Year History of
Modern Neotropical Mammals. University of Chicago Press,
Chicago pp 20–50
Goin FJ, Pascual R, Tejedor MF, Gelfo JN, Woodburne MO, Case JA,
Reguero MA, Bond M, Lopez GM, Cione AL, Udrizar Sauthier D,
Balarino L, Scasso RA, Medina FA, Ubaldon MC (2006) The ear-
liest Tertiary therian mammal from South America. J Vertebr
Paleontol 26(2):505–510
Goin FJ, Sánchez-Villagra MR, Abello A, Kay RF (2007a) A new gen-
eralized paucituberculatan marsupial from the Oligocene of Bolivia
and the origin of ‘shrew-like’opossums. Palaeontology 50:1267–
1276
Goin FJ, Sánchez-Villagra MR, Kay RF, Anaya-Daza F, Takai M (2003b)
New palaeothentid marsupial from the middle Miocene of Bolivia.
Palaeontology 46(2):307–315
Goin FJ, Tejedor MF, Chornogubsky L, Lopez GM, Gelfo JN, Bond M,
Woodburne MO, Gurovich Y, Reguero M (2012b) Persistence of a
Mesozoic, non-therian mammalian lineage (Gondwanatheria) in the
mid-Paleogene of Patagonia. Naturwissenschaften 99(6):449–463
Goin FJ, Woodburne MO, Zimicz AN, Martin GM, Chornogubsky L
(2016) A Brief History of South American Metatherians:
Evolutionary Contexts and Intercontinental Dispersals. Springer,
Dordrecht
Goin FJ, Zimicz AN, Forasiepi AM, Chornogubsky LC, Abello MA (in
press) The rise and fall of South American metatherians: contexts,
adaptations, radiations, and extinctions. In: Rosenberger AL,
Tejedor MF (eds) Origins and Evolution of Cenozoic South
American Mammals. Springer, New York
Goin FJ, Zimicz N, Reguero MA, Santillana SN, Marenssi SA, Moly JJ
(2007b) New marsupial (Mammalia) from the Eocene of Antarctica,
and the origins and affinities of the Microbiotheria. Rev Asoc Geol
Argent 62(4):597–603
Gregory WK (1910) The orders of mammals. Bull Am Mus Nat Hist 27:
1–524
Gurovich Y (2006) Bio-evolutionary aspects of Mesozoic mammals: de-
scription, phylogenetic relationships and evolution of the
Gondwanatheria, (Late Cretaceous and Paleocene of Gondwana).
Unpublished doctoral thesis, Universidad de Buenos Aires,
Facultad de Ciencias Exactas y Naturales, Buenos Aires
Hershkovitz P (1982) The staggered marsupial lower third incisor (I
3
).
Geobios Mem Spec 6:191–200
Hershkovitz P (1995) The staggered marsupial third lower incisor:
hallmark of cohort Didelphimorphia, and description of a
new genus and species with staggered i3 from the Albian
(Lower Cretaceous) of Texas. Bonner Zool Beitr 45:153–169
Hershkovitz P (1999) Dromiciops gliroides Thomas, 1894, last of the
Microbiotheria (Marsupialia), with a review of the family
Microbiotheriidae. Fieldiana Zool 93:1–60
Hoffstetter R, Villarroel C (1974) Découverte d’un marsupial
microtragulidé (= argyrolagidé) dans le Pliocène de l’Altiplano
bolivien. Comptes Rendus de l’Académie des Sciences, Paris 278:
1947–1950
Horovitz I (2000) The tarsus of Ukhaatherium nessovi (Eutheria,
Mammalia) from the Late Cretaceous of Mongolia: an appraisal of
the evolution of the ankle in basal therians. J Vertebr Paleontol
20(3):547–560
Horovitz I, Ladevèze S, Argot C, Macrini TE, Martin T, Hooker JJ, Kurz
C, Muizon Cd, Sánchez-Villagra MR (2008) The anatomy of
Herpetotherium cf. fugax Cope, 1873, a metatherian from the
Oligocene of North America. Palaeontographica Abteilung A
284(4–6):109–141
Horovitz I, Martin T, Bloch J, Ladevèze S, Kurz C, Sánchez-Villagra MR
(2009) Cranial anatomy of the earliest marsupials and the origin of
opossums. PLoS ONE 4(12):e8278
Horovitz I, Sánchez-Villagra MR (2003) A morphological analysis of
marsupial mammal higher-level phylogenetic relationships.
Cladistics 19:181–212
Jansa SA, Barker FK, Voss RS (2014) The early diversification history of
didelphid marsupials: a window into South America’sBsplendid
isolation.^Evolution 68(3):684–695
Jones M (1997)Character displacement in australian dasyurid carnivores:
size relationships and prey size patterns. Ecology 78(8):2569–2587
Jones ME, Stoddart DM (1998) Reconstruction of the predatory behav-
iour of the extinct marsupial thylacine (Thylacinus cynocephalus). J
Zool Lond 246:239–246
Kielan-Jaworowska Z, Cifelli RL, Luo Z-X (2004) Mammals from the
Age of Dinosaurs: Origins, Evolution, and Structure. Columbia
University Press, New York
Kirsch JAW, Lapointe FJ, Springer MS (1997) DNA-hybridization stud-
ies of marsupials and their implications for metatherian classifica-
tion. Aust J Zool 45:211–280
J Mammal Evol (2017) 24:373–414 411
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Krause DW (2014) Dental morphology of Vintana sertichi (Mammalia,
Gondwanatheria) from the Late Cretaceous of Madagascar. J Vertebr
Paleontol 34 (suppl to no. 6):137–165
Ladevèze S (2004) Metatherian petrosals from the late Paleocene of
Itaboraí (Brazil), and their phylogenetic implications. J Vertebr
Paleontol 24:202–213
Ladevèze S (2007) Petrosal bones of metatherian mammals from the late
Palaeocene of Itaboraí (Brazil), and a cladistic analysis of petrosal
features in metatherians. Zool J Linn Soc 150:85–115
Ladevèze S, Muizon C de (2007) The auditory region of early Paleocene
Pucadelphydae (Mammalia, Metatheria) from Tiupampa, Bolivia,
with phylogenetic implications. Palaeontology 50(5):1123–1154
Ladevèze S, Muizon C de (2010) Evidence of early evolution of
Australidelphia (Metatheria, Mammalia) in South America: phylo-
genetic relationships of the metatherians from the late Palaeocene of
Itaboraí (Brazil) based on teeth and petrosal bones. Zool J Linn Soc
159(3):746–784
Ladevèze S, Muizon C de, Beck RMD, Germain D, Céspedes-Paz R
(2011)Earliest evidence ofmammalian social behaviour in the basal
Tertiary of Bolivia. Nature 474(7349):83–86
Lorente M, Chornogubsky L, Goin FJ (2016) On the existence of non-
microbiotherian australidelphian marsupials (Diprotodontia) in the
Eocene of Patagonia. Palaeontology 59(4):533–547
Maier W (1989) Morphologische Untersuchungen am Mittelohr der
Marsupialia. Z Zool Syst Evolutionsforsch 27:149–168
Marshall LG (1980) Systematics of the South American marsupial family
Caenolestidae. Fieldiana Geol (New Series) 5:i–viii, 1–145
Marshall LG (1982a) Systematics of the extinct South American marsu-
pial family Polydolopidae. Fieldiana Geol 12:1–109
Marshall LG (1982b) Systematics of the South American marsupial fam-
ily Microbiotheriidae. Fieldiana Geol 10:1–75
Marshall LG (1985) Geochronology and land-mammal biochronology of the
transamerican faunal interchange. In: Stehli FG, Webb SD (eds) The
Great American Biotic Interchange. Plenum, New York, pp 49–85
Marshall LG (1987) Systematics of Itaboraian (middle Paleocene) age
Bopossum-like^marsupials from the limestone Quarry at Sao Jose
de Itaborai, Brazil. In: Archer M (ed) Possums and Opossums:
Studies in Evolution. Surrey Beatty & Sons, Sydney, pp 91–160
Marshall LG, Case JA, Woodburne MO (1990) Phylogenetic relation-
ships of the families of marsupials. Curr Mammal 2:433–505
Marshall LG, Muizon C de (1995) Pucadelphys andinus (Marsupialia,
Mammalia) from the early Paleocene of Bolivia. Mem Mus Natl
Hist Nat 165:1–164
Marshall LG, Muizon C de, Sigé B (1983) Late Cretaceous mammals
(Marsupialia) from Bolivia. Geobios 16:739–745
Marshall LG, Sempere T, Butler RF (1997) Chronostratigraphy of the
mammal-bearing Paleocene of South America. J South Am Earth
Sci 10(1):49–70
Martin GM (2007) Dental anomalies in Dromiciops gliroides
(Microbiotheria, Microbiotheriidae), Caenolestes fuliginosus and
Rhyncholestes raphanurus (Paucituberculata, Caenolestidae). Rev
Chil Hist Nat 80(4):393–406
Martin GM (2013) Intraspecific variability in Lestoros inca
(Paucituberculata, Caenolestidae), with reports on dental anomalies
and eruption pattern. J Mammal 94(3):601–617
Medeiros RA, Bergqvist LP (1999) Paleocene of the São José de Itaboraí
Basin, Río de Janeiro, Brazil: lithostratigraphy and biostratigraphy.
Acta Geol Leopold XXII(48):3–22
Meredith RW, Janecka JE, Gatesy J, Ryder OA, Fisher CA, Teeling EC,
Goodbla A, Eizirik E, Simao TL, Stadler T, Rabosky DL, Honeycutt
RL, Flynn JJ, Ingram CM, Steiner C, Williams TL, Robinson TJ,
Burk-Herrick A, Westerman M, Ayoub NA, Springer MS, Murphy
WJ (2011) Impacts of the Cretaceous Terrestrial Revolution and
KPg extinction on mammal diversification. Science 334 (6055):
521–524
Meredith RW, Westerman M, Springer MS (2009) A phylogeny of
Diprotodontia (Marsupialia) based on sequences for five nuclear
genes. Mol Phylogenet Evol 51(3):554–571
Meyer M, Arsuaga JL, de Filippo C, Nagel S, Aximu-Petri A, Nickel B,
Martinez I, Gracia A, Bermudez de Castro JM, Carbonell E, Viola
B, Kelso J, Prufer K, Paabo S (2016) Nuclear DNA sequences from
the middle Pleistocene Sima de los Huesos hominins. Nature
531(7595):504–507
Meyer M, Fu QM, Aximu-Petri A, Glocke I, Nickel B, Arsuaga JL,
Martinez I, Gracia A, de Castro JMB, Carbonell E, Paabo S
(2014) A mitochondrial genome sequence of a hominin from Sima
de los Huesos. Nature 505(7483):403–406
Mitchell KJ, Pratt RC, Watson LN, Gibb GC, Llamas B, Kasper M,
Edson J, Hopwood B, Male D, Armstrong KN, Meyer M,
Hofreiter M, Austin J, Donnellan SC, Lee MS, Phillips MJ,
Cooper A (2014) Molecular phylogeny, biogeography, and habitat
preference evolution of marsupials. Mol Biol Evol 31(9):2322–2330
Morrone JJ (2002) Biogeographical regions under track and cladistic
scrutiny. J Biogeogr 29:149–152
Morrone JJ (2004) Panbiogeografía, componentes bióticos y zonas de
transición. Rev Bras Entomol 48(2):149–162
Morrone JJ (2006) Biogeographic areas and transition zones of Latin
America and the Caribbean islands based on panbiogeographic
and cladistic analyses of the entomofauna. Annu Rev Entomol 51:
467–494
Muirhead J (2000) Yaraloidea (Marsupialia, Peramelemorphia), a new su-
perfamily of marsupial and a description and analysis of the cranium
of the Miocene of Yarala burchfieldi. J Paleontol 74(3):512–523
Muizon C de (1994) A new carnivorous marsupial from the Palaeocene of
Bolivia and the problem of marsupial monophyly. Nature 370:208–
211
Muizon C de (1998) Mayulestes ferox, a borhyaenoid (Metatheria,
Mammalia) from the early Palaeocene of Bolivia: phylogenetic
and palaeobiologic implications. Geodiversitas 20(1):19–142
Muizon C de (1999) Marsupial skulls from the Deseadan (late Oligocene)
of Bolivia and phylogenetic analysis of the Borhyaenoidea
(Marsupialia, Mammalia). Geobios 32(3):483–509
Muizon C de, Cifelli RL, Céspedes Paz R (1997) The origin of the dog-
like borhyaenoid marsupials of South America. Nature 389:486–
489
Muizon C de, Marshall LG, Sigé B (1984) The mammal fauna from
the El Molino Formation (Late Cretaceous-Maastrichtian) at
Tiupampa, southcentral Bolivia. Bull Mus natl Hist nat Paris 4e
série, section C 6(4):315–327
Nilsson MA, Churakov G, Sommer M, Tran NV, Zemann A, Brosius J,
Schmitz J (2010) Tracking marsupial evolution using archaic geno-
mic retroposon insertions. PLoS Biol 8(7):e1000436
Norris CA (1993) Changes in the composition of the auditory bulla in
southern Solomon Islands populations of the grey cuscus,
Phalanger orientalis breviceps (Marsupialia, Phalangeridae). Zool
J Linn Soc 107:93–106
O’Meara RN, Thompson RS (2014) Were there Miocene
meridiolestidans? Assessing the phylogenetic placement of
Necrolestes patagonensis and the presence of a 40 million year
meridiolestidan ghost lineage. J Mammal Evol 21(3):271–284
Ojala-Barbour R, Pinto CM, Brito MJ, Albuja VL, Lee TE Jr, Patterson
BD (2013) A new species of shrew-opossum (Paucituberculata:
Caenolestidae) with a phylogeny of extant caenolestids. J Mammal
94(5):967–982
Oliveira EV, Goin FJ (2006) Marsupiais do início do Terciário do Brasil:
origem, irradiação e história biogeográfica. In: Cáceres NC,
Monteiro Filho ELA (eds) Os Marsupiais do Brasil: Biologia,
Ecologia e Evolução. UFMS, Campo Grande, pp 299–320
Oliveira EV, Goin FJ (2011) A reassessment of bunodont metatherians
from the Paleogene of Itaboraí (Brazil): systematics and age of the
Itaboraian SALMA. Rev Brasil Paleontol 14(2):105–136
412 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Oliveira EV, Goin FJ (2015) A new species of Gaylordia Paula Couto
(Mammalia, Metatheria) from Itaboraí, Brazil. Rev Bras Paleontol
18(1):97–108
Oliveira EV, Zimicz N, Goin FJ (2016) Taxonomy, affinities, and paleo-
biology of the tiny metatherian mammal Minusculodelphis,fromthe
early Eocene of South America. Naturwissenschaften 103 (1–2):6
Osgood WH (1921) A monographic study of the American marsupial
Caenolestes. Field Mus Nat Hist Zool Ser 14:1–162
Osgood WH (1924) Review of living caenolestids with description of a
new genus from Chile. Field Mus Nat Hist Publ 222 Zool Ser 14(2):
163–172, pl. 123
Pascual R (1980a) Nuevos y singulares tipos ecológicos de marsupiales
extinguidos de América del Sur (Paleoceno Tardío o Eoceno
Temprano) del Noroeste argentino. II Congresso Latinoamericano
de Paleontología y Bioestratigrafía, Buenos Aires, Actas 2:151–173
Pascual R (1980b) Prepidolopidae, nueva familia de Marsupialia
Didelphoidea del Eoceno sudamericano. Ameghiniana 17:216–242
Pascual R (1981) Adiciones al conocimiento de Bonapartherium
hinakusijum (Marsupialia, Bonapartheriidae) del Eoceno temprano
del Noroeste argentino. Anais II Congreso Latino-Americano
Paleontologico (Porto Alegre, Brasil) 2:507–520
Pascual R, Archer M, Ortiz-Jaureguizar E, Prado JL, Godthelp H, Hand
SJ (1992) First discovery of monotremes in South America. Nature
356:704–705
Pascual R, Bond M (1981) Epidolopinae subfam. nov. de los
Polydolopidae (Marsupialia, Polydolopoidea). Anais do II
Congresso Latino-americano de Paleontologia 2:479–488
Pascual R, Carlini AA (1987) A new superfamily in the extensive radia-
tion of South American Paleogene marsupials. Fieldiana Zool 39:
99–110
Pascual R, Goin FJ, Carlini AA (1994) New data on the Groeberiidae:
unique late Eocene-early Oligocene South American marsupials. J
Vertebr Paleontol 14(2):247–259
Pascual R, Goin FJ, Krause DW, Ortiz-Jaureguizar E, Carlini AA (1999)
The first gnathic remains of Sudamerica: Implications for
gondwanathere relationships. J Vertebr Paleontol 19(2):373–382
Patterson B (1952) Un nuevo y extraordinario marsupial deseadiano.
Revista del Museo Municipal de Ciencias Naturales Mar del Plata
1:39–44
Patterson BD, Gallardo MH (1987) Rhyncholestes raphanurus.Mammal
Species 286:1–5
Paula Couto C de (1952a) Fossil mammals from the beginning of the
Cenozoic in Brazil. Condylarthra, Litopterna, Xenungulata, and
Astrapotheria. Bull Am Mus Nat Hist 99:355–394
Paula Couto C de (1952b) Fossil mammals from the beginning of the
Cenozoic in Brazil. Marsupialia: Didelphidae. Am Mus Novitates
1567:1–26
Paula Couto C de (1952c) Fossil mammals from the beginning of the
Cenozoic in Brazil. Marsupialia: Polydolopidae and Borhyaenidae.
Am Mus Novitates 1559:1–27
Paula Couto C de (1952d) Fossil mammals from the beginning of the
Cenozoic in Brazil. Notoungulata. Am Mus Novitates 1568:1–16
Pinheiro AEP, Fortier DC, Pol D, Campos DA, Bergqvist LP (2012) A
new Eocaiman (Alligatoridae, Crocodylia) from the Itaboraí Basin,
Paleogene of Rio de Janeiro, Brazil. Hist Biol 25(3):327–337
R Development Core Team (2016) R: a language and environment for
statistical computing. R Foundation for Statistical Computing,
Vienna
Rage JC (1998) Fossil snakes from the Palaeocene of Sao Jose de
Itaborai, Brazil. Part I. Madtsoiidae, Aniliidae. Palaeovertebrata
27(3–4):109–144
Riccomini C, Rodrigues-Francisco BH (1992) Idade potássio-argônio do
derrame de ankaramito da Bacia de Itaboraí, Rio de Janeiro, Brasil:
Implicações tectônicas. Boletim de Resumos Do 37° Congresso
Brasileiro de Geologia:469–470
Ride WDL (1956) The affinities of Burramys parvus Broom a fossil
phalangeroid marsupial. Proc Zool Soc Lond 127(3):413–429
Ride WDL (1962) On the evolution of Australian marsupials. In: Leeper
GW (ed) The Evolution of Living Organisms. Melbourne
University Press, Melbourne pp 281–306
Ride WDL (1964) A review of Australian fossil marsupials. J Proc Roy
Soc West Australia 47:97–131
Rincón AD, Shockey BJ, Anaya F, Solórzano A (2015) Palaeothentid
marsupials of the Salla Beds of Bolivia (late Oligocene): two new
species and insights into the post-Eocene radiation of
palaeothentoids. J Mammal Evol 22(4): 455–471
Rougier GW, Wible JR, Beck RMD, Apesteguia S (2012) The Miocene
mammal Necrolestes demonstrates the survival of a Mesozoic
nontherian lineage into the late Cenozoic of South America. Proc
Natl Acad Sci USA 109 (49):20053–20058
Rougier GW, Wible JR, Novacek MJ (1998) Implications of
Deltatheridium specimens for early marsupial history. Nature 396:
459–463
Rougier GW, Wible JR, Novacek MJ (2004) New specimens of
Deltatheroides cretacicus (Metatheria, Deltatheroida) from the
Late Cretaceous of Mongolia. Bull Carnegie Mus Nat Hist 36 (1):
245–266
Rusconi C (1933) New Pliocene remains of diprotodont marsupials from
Argentina. J Mammal 14 (3):244–250
Rusconi C (1936) La supuesta afinidad de Argyrolagus con los
Typotheria. Boletin de la Academia Nacional de Ciencias en
Córdoba 33:173–182
Rybczynski N, Gosse JC, Harington CR, Wogelius RA, Hidy AJ,
Buckley M (2013) Mid-Pliocene warm-period deposits in the High
Arctic yield insight into camel evolution. Nature Comm 4:1550
Sánchez-Villagra MR (1998) Patterns of morphological change in the
ontogeny and phylogeny of the marsupial skull. Unpublished
Ph.D. Dissertation, Duke University, Durham
Sánchez-Villagra MR (2001) The phylogenetic relationships of
argyrolagid marsupials. Zool J Linn Soc 131:481–496
Sánchez-Villagra MR, Kay RF (1997) A skull of Proargyrolagus,the
oldest argyrolagid (Late Oligocene Salla Beds, Bolivia), with brief
comments concerning its paleobiology. J Vertebr Paleontol 17(4):
717–724
Sánchez-Villagra MR, Kay RF, Anaya-Daza F (2000) Cranial anatomy
and palaeobiology of the Miocene marsupial Hondalagus
altiplanensis and a phylogeny of argyrolagids. Palaeontology
43(2): 287–301
Sánchez-Villagra MR, Ladevèze S, Horovitz I, Argot C, Hooker JJ,
Macrini TE, Martin T, Moore-Fay S, Muizon C de, Schmelzle T,
Asher RJ (2007) Exceptionally preserved North American
Paleogene metatherians: adaptations and discovery of a major gap
in the opossum fossil record. Biol Lett 3(3):318–322
Sánchez-Villagra MR, Wible JR (2002) Patterns of evolutionary transfor-
mation in the petrosal bone and some basicranial features in marsu-
pial mammals, with special reference to didelphids. J Zool Syst Evol
Res 40:26–45
Sereno PC (2006) Shoulder girdle and forelimb in multituberculates:
evolution of parasagittal forelimb posture in mammals. In: Carrano
MT, Gaudin TJ, Blob RW, Wible JR (eds) Amniote Paleobiology:
Perspectives on the Evolution of Mammals, Birds, and Reptiles.
University of Chicago Press, Chicago, pp 315–366
Sigé B, Archer M, Crochet J-Y, Godthelp H, Hand S, Beck RMD (2009)
Chulpasia and Thylacotinga, late Paleocene-earliest Eocene trans-
Antarctic Gondwanan bunodont marsupials: new data from
Australia. Geobios 42(6):813–823
Simpson GG (1928) Affinities of the Polydolopidae. Am Mus Novitates
323:1–13
Simpson GG (1933) The Bplagiaulacoid^type of mammalian dentition a
study of convergence. J Mammal 14(2):97–107
J Mammal Evol (2017) 24:373–414 413
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Simpson GG (1935) Occurrence and relationships of the Río Chico fauna
of Patagonia. Am Mus Novitates 818:1–21
Simpson GG (1945) The principles of classification and a classification of
mammals. Bull Am Mus Nat Hist 85:1–350
Simpson GG (1948) The beginning of the age of mammals in South
America. Bull Am Mus Nat Hist 91(1):1–232
Simpson GG (1970a) Addition to the knowledge of Groeberia
(Mammalia, Marsupialia) from the mid-Cenozoic of Argentina.
Breviora 362:1–17
Simpson GG (1970b) The Argyrolagidae, extinct South American mar-
supials. Bull Mus Comp Zool 139:1–86
Springer MS, Woodburne MO (1989) The distribution of some
basicranial characters within the Marsupialia and a phylogeny of
the Phalangeriformes. J Vertebr Paleontol 9(2):210–221
Szalay FS (1982) A new appraisal of marsupial phylogeny and classifi-
cation. In:Archer M(ed) Carnivorous Marsupials. Royal Zoological
Society of New South Wales, Mosman, pp 621–640
Szalay FS (1984) Arboreality: Is it homologous in metatherian and eu-
therian mammals? In: Hecht M, Wallace B, Prance G (eds)
Evolutionary Biology, vol 18. Plenum Publishing Corporation,
New York, pp 215–258
Szalay FS (1993) Pedal evolution of mammals in the Mesozoic: tests for
taxic relationships. In: Szalay FS, Novacek MJ, McKenna MC (eds)
Mammal Phylogeny, Vol. 1: Mesozoic Differentiation,
Multituberculates, Early Therians, and Marsupials. Springer-
Verlag,NewYork,pp108–128
Szalay FS (1994) Evolutionary History of the Marsupials and an Analysis
of Osteological Characters. Cambridge University Press, Cambridge
Szalay FS, Sargis EJ (2001) Model-based analysis of postcranial osteol-
ogy of marsupials from the Palaeocene of Itaboraí (Brazil) and the
phylogenetics and biogeography of Metatheria. Geodiversitas 23(2):
139–302
Szalay FS, Sargis EJ (2006) Cretaceous therian tarsals and the
metatherian-eutherian dichotomy. J Mammal Evol 13:171–210
Szalay FS, Trofimov B (1996) The Mongolian Late Cretaceous
Asiatherium, and the early phylogeny and paleobiogeography of
Metatheria. J Vertebr Paleontol 16:474–509
Tejedor MF, Goin FJ, Gelfo JN, López G, Bond M, Carlini AA, Scillato-
Yané GJ, Woodburne MO, Chornogubsky L, Aragón E, Reguero
MA, Czaplewski NJ, Vincon S, Martin GM, Ciancio MR (2009)
New early Eocene mammalian fauna from western Patagonia,
Argentina. Am Mus Novitates 3638:1–43
Villarroel A. C, Marshall LG (1988) A new argyrolagoid (Mammalia:
Marsupialia) from the middle Miocene of Bolivia. J Paleontol 62(3):
463–467
Voss RS, Jansa SA (2003) Phylogenetic studies on didelphid marsupials
II. Nonmolecular data and new IRBP sequences: separate and
combined analyses of didelphine relationships with denser taxon
sampling. Bull Am Mus Nat Hist 276:1–82
Voss RS, Jansa SA (2009) Phylogenetic relationships and classifica-
tion of didelphid marsupials, an extant radiation of New World
metatherian mammals. Bull Am Mus Nat Hist 322:1–177
Vullo R, Gheerbrant E, Muizon C de, Neraudeau D (2009) The oldest
modern therian mammal from Europe and its bearing on stem mar-
supial paleobiogeography. Proc Natl Acad Sci USA 106 (47):
19910–19915
Warton DI, Duursma RA, Falster DS, Taskinen S (2012) SMATR 3 - an R
package for estimation and inference about allometric lines. Meth
Ecol Evol 3:257–259
Wible JR (1990) Late Cretaceous marsupial petrosal bones from North
America and a cladistic analysis of the petrosal in therian mammals.
J Vertebr Paleontol 10:183–205
Wible JR (2003) On the cranial osteology of the short-tailed opossum
Monodelphis brevicaudata (Marsupialia, Didelphidae). Ann
Carnegie Mus 72(3):137–202
Wible JR, Rougier GW (2000) Cranial anatomy of Kryptobataar
dashzevegi (Mammalia, Multituberculata) and its bearing on
the evolution of mammalian characters. Bull Am Mus Nat Hist
247:1–124
Woodburne MO, Goin FJ, Bond M, Carlini AA, Gelfo JN, López
GM, Iglesias A, Zimicz AN (2014a) Paleogene land mammal
faunas of South America; a response to global climatic
changes and indigenous floral diversity. J Mammal Evol
21(1):1–73
Woodburne MO, Goin FJ, Raigemborn MS, Heizler M, Gelfo JN,
Oliveira EV (2014b) Revised timing of the South American
early Paleogene land mammal ages. J South Am Earth Sci
54:109–119
Woodburne MO, Zinsmeister WJ (1982) Fossil land mammal from
Antarctica. Science 218:284–286
Woodburne MO, Zinsmeister WJ (1984) The first land mammal from
Antarctica and its biogeographic implications. J Paleontol 58:913–
948
Ziegler AC (1971) A theory of the evolution of therian dental formulas
and replacement patterns. Quart Rev Biol 46 (3):226–249
Zimicz N (2011) Patrones de desgaste y oclusión en el sistema
masticatorio de los extintos Argyrolagoidea (Marsupialia,
Polydolopimorphia, Bonapartheriiformes). Ameghiniana 48 (3):
358–379
Zimicz N (2014) Paleoecología de los Bonapartherioidea (Marsupialia,
Polydolopimorphia, Bonapartheriiformes). Ameghiniana 51(2):
106–128
414 J Mammal Evol (2017) 24:373–414
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com