ORIGINAL PAPER Persistence of a Mesozoic, non-therian mammalian lineage (Gondwanatheria) in the mid-Paleogene of Patagonia Francisco J. Goin & Marcelo F. Tejedor & Laura Chornogubsky & Guillermo M. López & Javier N. Gelfo & Mariano Bond & Michael O. Woodburne & Yamila Gurovich & Marcelo Reguero Received: 17 November 2011 /Revised: 23 April 2012 /Accepted: 25 April 2012 /Published online: 15 May 2012 # Springer-Verlag 2012 Abstract We describe two isolated molariforms recovered from early–middle Eocene (early Lutetian) levels of north- western Patagonia, Argentina. Comparisons with major lin- eages of therian and non-therian mammals lead us to refer them to a new genus and species of Gondwanatheria (Allo- theria). There is a single root supporting each tooth that is very short, wide, rounded, and covered by cementum; the steep sidewalls, lack of a neck between the crown and root, and the heavily worn stage in both molariforms suggest that they were of a protohypsodont type. Both teeth are strongly worn at their centers, all along their length, with the labial edge less worn than the lingual; they show strong transverse crests that alternate with lingual grooves. The protohypsodont aspect of the teeth, as well as the strong, transverse crests, are suggestive of sudamericid affinities; on the other hand, the thin enamel layer and the occlusal pattern formed by the crests and grooves shows more similarities to molariform teeth of the Ferugliotheriidae. The new taxon adds evidence regarding the (1) extensive radiation of the Gondwanatheria throughout the Southern Hemisphere, (2) persistence of several lineages well after the Cretaceous/ Paleogene boundary, and (3) early evolution of hypsodont types among South American herbivorous mammals. Keywords Gondwanatheria . South America . Paleogene . Eocene . Hypsodonty Communicated by: Robert Asher F. J. Goin (*) : J. N. Gelfo :M. Bond :M. Reguero CONICET and División Paleontología Vertebrados, Museo de La Plata, Paseo del Bosque s/n, B1900FWA La Plata, Argentina e-mail: fgoin@fcnym.unlp.edu.ar J. N. Gelfo e-mail: jgelfo@fcnym.unlp.edu.ar M. Bond e-mail: constantino1453@yahoo.com.ar M. Reguero e-mail: regui@fcnym.unlp.edu.ar M. F. Tejedor CONICET and Centro Nacional Patagónico, Bv. Almirante Brown 2915, 9120 Puerto Madryn, Chubut, Argentina e-mail: tejedor@cenpat.edu.ar L. Chornogubsky CONICET and Sección Paleontología Vertebrados, Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”, Av. Angel Gallardo 470, C1405DJR Buenos Aires, Argentina e-mail: lchorno@macn.gov.ar G. M. López División Paleontología Vertebrados, Museo de La Plata, Paseo del Bosque s/n, B1900FWA La Plata, Argentina e-mail: glopez@fcnym.unlp.edu.ar M. O. Woodburne Department of Geology, Museum of Northern Arizona, Flagstaff, AZ 86001, USA e-mail: mikew@npgcable.com Y. Gurovich School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia e-mail: yamilag@gmail.com Naturwissenschaften (2012) 99:449–463 DOI 10.1007/s00114-012-0919-z Abbreviations AH Andesitas Huancache Formation EDJ enamel-dentine junction IBC Ignimbrita Barda Colorada Formation IPM interprismatic matrix LIEB-PV Vertebrate Paleontology Collection, Laboratorio de Investigaciones en Evolución y Biodiversidad, Esquel m meters Ma Megannum MACN PV RN Sección Paleontología Vertebrados, Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”, Buenos Aires, Vertebrate Paleontology, Río Negro Collection MF upper molariform MLP División Paleontología Vertebrados, Facultad de Ciencias Naturales y Museo de La Plata (UNLP), La Plata mm millimeters OES outer enamel surface PLEX prismless outer enamel SALMA South American Land-Mammal Age SANU South American native ungulates TLH Tufolitas Laguna del Hunco Formation UNPSJB Universidad Nacional de la Patagonia “San Juan Bosco”, sede Esquel µm microns (micrometers) Introduction Gondwanatherians are a group of extinct, still poorly known, non-therian mammals endemic to the Southern Hemisphere. Their known Mesozoic record includes the Late Cretaceous of India, Madagascar, southern South America (Patagonia) and, possibly, the Cretaceous of Africa (Tanzania; Fig. 1). Their Cenozoic record includes the Pa- leocene of Patagonia and the Eocene of Antarctica, as well as still dubious records for the Paleogene of tropical South America (Fig. 1 and references cited therein). Their affini- ties are still a matter of debate (see below). Specific taxa within the order were originally regarded as basal xenar- thrans (as for Sudamerica ameghinoi, Sudamericidae; Scillato-Yané and Pascual 1985; Mones 1987; Bonaparte 1990), or as multituberculates (Ferugliotherium windhau- seni, Ferugliotheriidae; Bonaparte 1987). Later, Krause and Bonaparte (1990) recognized that both ferugliotheriids and sudamericids belong to the same group of non-therian mam- mals, and included them as a suborder (Gondwanatheria) of Multituberculata, and later as a superfamily (Gondwanather- ioidea) tentatively regarded as belonging to the plagiaulacoid clade of multituberculates (Krause and Bonaparte 1993). Additional remains of sudamericids were successively recognized from Cretaceous levels of other continents: Lavanify miolaka from Madagascar (Krause et al. 1997) and Bharattherium bonapartei from India (Krause et al. 1997; Prasad et al. 2007; Wilson et al. 2007). A second taxon of Ferugliotheriidae (Trapalcotherium matuastensis) was also reported from the Late Cretaceous of Patagonia (Rougier et al. 2009). Paleogene remains of gondwanatherians were also recovered from early–middle Eocene levels of the Antarctic Peninsula (Goin et al. 2006a), and from middle Eocene deposits of northern Perú (an undescribed “probable sudamericid gondwanathere”; Antoine et al. 2011, electronic supplementary material (ESM): 18). Finally an enigmatic tooth, possibly representing a ferugliotheriid, was reported from the ?late Eocene–?early Oligocene of Perú (Goin et al. 2004). Meanwhile, discussions regarding the affinities of the Gondwanatheria were still ongoing. Pascual et al. (1999; see also Kielan-Jaworowska et al. 2004) regarded the group as Mammalia incertae sedis, stating its possible allotherian (though not strictly multituberculatan) affinities. Pascual and Ortiz-Jaureguizar (2007) argued that gondwanatherians are allotherians derived from haramiyidans. Gurovich and Beck (2009; see also Gurovich 2006, 2008) concluded that ferugliotheriids and sudamericids constitute a natural group together with cimolodontan and “plagiaulacid” multitu- berculates. In the absence of more complete skeletal and dental remains of gondwanatherians, it is predictable that discussions will continue. Putting aside the possible affinities of gondwanatherians, a certain multitubercu- late described as a possible cimolodontan, Argentodites coloniensis (Kielan-Jaworowska et al. 2007; but see Rougier et al. 2009), is known from the Late Cretaceous of La Colonia Formation in Chubut Province, central Patagonia. In a recent contribution by Tejedor et al. (2009), two new mammalian associations of late early–early middle Eocene age (early Lutetian) from western Patagonia were described. They were recovered from two nearby fossil sites, the Laguna Fría and La Barda localities, near the town of Paso del Sapo, in northwestern Chubut Province, Argentina (Fig. 2). More than 50 species of mammals, including a variety of metatherians, xenarthrans and native ungulates, as well as South America’s oldest bats, have been recognized to date, many of them still under study. Based on the chronologic information provided by the mammals, as well as from isotopic dates of under- and overlying beds, Tejedor et al. (2009) estimated that the new sites and levels could possibly represent a new biochronolog- ical unit for the Paleogene of South America, occupying the temporal gap between the early Eocene Riochican SALMA and the middle Eocene, Vacan subage of the Casamayoran SALMA. 450 Naturwissenschaften (2012) 99:449–463 Here, we report the discovery of a new mammalian taxon recovered from one of these fossil localities, the La Barda site (Fig. 2c). The new mammal is represented by two upper molariform teeth, with an overall morphology that is so unique as to prevent us from allocating it to any known group of therian mammals. We conclude that the mammal represents a new genus and species of Gondwanatheria (Allotheria), whose affinities with either ferugliotheriids or sudamericids are still unclear. The description of the new taxon and a discussion of its significance are the main purposes of this work. Methods The anatomical terminology (“molariform”, “islets”, etc.) follows that of von Koenigswald et al. (1999), except for the term “groove,” which we apply in the same sense of “syncline” or “flexum/flexid” of these authors. By this, we acknowledge that the flexae in the new taxon are much more mesiodistally compressed than in other gondwanatherians. However, their meaning is the same: “…lateral infolds of the enamel band from the lingual or buccal (0labial) side, which may be filled with cementum” (von Koenigswald et al. 1999: 267). The use of the term “molariform” explicitly excludes any attempt of homologizing these teeth with those of the molar loci of other mammalian groups. Perikymata are growth lines on the enamel surface related to striae of Retzius (von Koenigswald and Sander 1997). As defined previously (von Koenigswald et al. 1999; see also von Koenigswald 2011) enamel islets of the first type are pits or fossae observable at the occlusal face of the teeth, which are not related to transverse flexae or grooves; they occur in the mesial or distal lobes of the teeth. In turn, enamel islets of the second type derive from transverse flexae. In order to observe the enamel microstructure of the molari- forms of Greniodon sylvaticum, we grounded and etched two enamel surfaces at the exposed mesial face of specimen LIEB- PV (Vertebrate Paleontology Collection, Laboratorio de Inves- tigaciones en Evolución y Biodiversidad, Esquel) 2000. Verti- cal and tangential views were obtained (areas 1 and 2, respectively, in Fig. 5a). The specimen was not imbedded in resin for this study. Etching was made by exposing the polished sections a few seconds with hydrocloric acid. Observations were made with the aid of a scanning electron microscope. The resulting micrographs are shown in Fig. 5. All dental measurements except enamel prism diameter are in millimeters and were taken with a digital caliper (Schwiz). Prism diameters (maximum width) are expressed in micrometers. Systematic paleontology Mammalia Linnaeus 1758 Allotheria Marsh 1880 Gondwanatheria Mones 1987 Family indet. Greniodon sylvaticus, gen et. sp. nov. (Figs. 3, 4, 5, and 6) Etymology Greniodon, after the late Juan Grenier, owner of the “26 de Mayo” farm, and his family, and –odon, from the Greek odontos (tooth). Gender is masculine. The specific name sylvaticus comes after the Latin sylva (or silva), forest, in reference to the inferred paleofloral environment in which the new mammal here described lived (see below). Fig. 1 Late Cretaceous reconstruction of the Southern Hemisphere (based on the ODSN Plate Tectonic Reconstruction Service, http:// www.odsn.de/odsn/services/paleomap/paleomap.html; 100 Ma) show- ing fossil localities that include gondwanatherians or presumed gond- wanatherians. References (from older to younger): 1 Cretaceous (s.l.) of Tanzania: unnamed ?Gondwanatheria (“Red Sandstone Group”; Krause et al. 2003); 2 Late Cretaceous (Maastrichtian) of Madagascar: Lavanify miolaka, Sudamericidae (Maevarano Formation; Krause et al. 1997); 3–6 Late Cretaceous (Maastrichtian) of India: Bharattherium bonapartei, Sudamericidae (Deccan intertrappean sediments; Prasad et al. 2007; Wilson et al. 2007) [3 Gokak, 4 Naskal, 5 Kisalpuri, and 6 Bacharam localities]; 7–9 Late Cretaceous (Campanian-Maastrichtian; Alamitian Age) of northern Patagonia, Argentina: Ferugliotherium wind- hauseni (Ferugliotheriidae), Trapalcotheriummatuastensis (Ferugliother- iidae), and Gondwanatherium patagonicum (Sudamericidae) [7 Los Alamitos (Los Alamitos Formation; Bonaparte 1986a), 8 Cerro Tortuga (Allen Formation; Rougier et al. 2009), 9 La Colonia (La Colonia Formation.; Pascual andOrtiz-Jaureguizar 2007: Figs. 11–12)]; 10medial Paleocene (early Selandian)of central Patagonia (Salamanca Formation, Punta Peligro locality; Scillato-Yané and Pascual 1985); 11 early–middle Eocene of northern Patagonia, Argentina: Greniodon sylvaticus gen. et sp. nov. (Family indet) (Volcanic–Pyroclastic complex of the Middle Chubut River, La Barda locality; Tejedor et al. 2009); 12 early–middle Eocene of the Antarctic Peninsula: Sudamericidae, gen. et sp. indet., “cf. Sudamerica ameghinoi” (La Meseta Formation, locality IAA 1/90; Goin et al. 2006a: 136); 13 middle–late Eocene of Perú: “Sudamericidae”, undescribed specimens and taxa (Contamana locality, Yahuarango Formation; Antoine et al. 2011 [Electronic supplementary material http://dx.doi.org/10.1098/rspb.2011.1732]); 14 ?late Eocene-early Oligo- cene of Perú: ?Gondwanatheria, gen. et sp. indet. (Santa Rosa locality, Yahuarango Formation; Campbell et al. 2004; Goin et al. 2004) Naturwissenschaften (2012) 99:449–463 451 http://www.odsn.de/odsn/services/paleomap/paleomap.html http://www.odsn.de/odsn/services/paleomap/paleomap.html http://dx.doi.org/10.1098/rspb.2011.1732 Holotype LIEB-PV 2001 (Figs. 3a–d and 6c), an upper left molariform (MF?3). Referred specimen LIEB-PV 2000 (Figs. 3e–i, 4, 5b), an upper right molariform (MF?2), lacking the mesialmost lobe. Collection data Specimens LIEB-PV 2000 and 2001 were discovered in 2010, not directly at the La Barda site but instead during “picking” efforts of sedimentary matrix con- centrates coming from this locality in 2005. Sorting was performed in the Laboratorio de Investigaciones en Evolución y Biodiversidad (LIEB, Esquel). Locality, horizon, and age La Barda fossil site, in “26 de Mayo” farm, 28 km southwest of Paso del Sapo Town, Chubut Province, Argentina, at 42° 46′ 48.5″S, 69° 51′ Fig. 2 Map of Chubut province (a, b in black) and of the area west of Paso del Sapo Town (c), including fossil localities mentioned in the text. La Barda is the fossil site from which the remains of Greniodon sylvaticus gen. et sp. nov. were recovered. Modified after Ruiz (2006). Other fossil localities mentioned in the text are indicated in b. LH Laguna del Hunco (early Eocene, plant macrofossils), RP Río Pichileufú (early–middle Eocene, plant macrofossils), PP Punta Peligro (medial Paleocene, Peligran SALMA) 452 Naturwissenschaften (2012) 99:449–463 43.3″W, which lies at 1,056 m above sea level (Figs. 1 and 2c). The La Barda deposits are part of the Middle Chubut River Volcanic–Pyroclastic Complex, reviewed by Aragón and Mazzoni (1997). This complex includes a great variety of volcanogenic bodies such as ignimbrites, domes, lava flows, necks, intrusives, tuffs, and volcaniclastic deposits (of predominantly lacustrine origin), all of them frequently interbedded. Further aspects of the regional geology, stratigraphy, geochemistry, petrogenesis, and paleobotany of this complex have been studied by Petersen (1946), Archangelsky (1974), Volkheimer and Lage (1981), Lage (1982), Aragón and Romero (1984), Rapela et al. (1984), Mazzoni and Aragón (1985), Aragón et al. (1987), and Mazzoni et al. (1989). Out of 12 formal stratigraphic units cropping out in the study area (Aragón and Mazzoni 1997), three of them, belonging to the Middle Chubut River Volca- nic–Pyroclastic Complex, are relevant for this study: Ignim- brita Barda Colorada Formation (IBC), Tufolitas Laguna del Hunco Formation (TLH), and Andesitas Huancache Forma- tion (AH). IBC overlies the Late Cretaceous/earliest Paleo- cene Lefipán Formation. It represents a major volcanic event (≥100 km3) that formed a large 80 m thick ignimbritic plateau related to the development of a 25-km wide caldera. TLH conformably overlies IBC and corresponds to the lacustrine deposits within the caldera. TLH bears an extraordinary fossil flora exposed at a few localities (Wilf et al. 2003). AH consists predominantly of lavas alternating with intercalated tuffs, volcanic agglomerates, and sandstones. Together with IBC, AH comprises the thickest deposits of the complex. The La Barda locality occurs within the AH interbedded tuffs. These tuffs occur above basalts dated by 40Ar/39Ar method at 47.89±1.21 Ma (Gosses et al. 2006). The AH to the west has yielded three 40K/40Ar dates near 43 Ma (Mazzoni et al. 1991). Judging from the isotopic dates of levels in the Laguna Fría and La Barda localities, and due to the similarities of both faunas, Tejedor et al. (2009) assumed a 45–47 Ma age (early Lutetian) for the deposition of the La Barda mammal-bearing tuffs. In the study area (Fig. 2c), several additional stratigraphic units can be distinguished, the oldest of which are the Late Cretaceous continental sedimentary deposits of the Paso del Sapo Formation. These deposits underlie the Late Creta- ceous–earliest Paleocene, marine to marginal deposits of the Lefipán Formation. A couple of kilometer east of the La Barda site, and within the uppermost levels of the Lefipán Formation, the earliest South American therian mammal known up to now was found: Cocatherium lefipanum (Marsupialia, Polydolopimorphia; Goin et al. 2006b; Fig. 2c). East and northeast of the La Barda locality crop out middle Miocene tuffs that have been referred to the Collón Curá Formation (Lage 1982; Ruiz 2006). Finally, near the Chubut River, there are several fluvial terraces, as well as Holocene– Recent deposits. Diagnosis Gondwanatheria differ from the Ferugliotherii- dae in its much larger size, protohypsodont molariforms that have its transverse lobes better defined by deep, narrow grooves, and by its wear pattern, which is quite marked at the central portion of the molariforms all along its lengths. Gondwanatheria differs from all known Sudamericidae in having a much thinner enamel layer in its molariforms. It differs from South American Sudamericidae in its larger size and specifically from S. ameghinoi in the marked asymme- try of its upper molariform grooves (exclusively lingual in Greniodon, both labial and lingual in Sudamerica). More- over, it differs from Madagascar and Indian Sudamericidae in its much larger size, thinner enamel layer, more complex occlusal design of its molariforms, absence of perikymata, and much deeper lingual grooves that divide the crown into several distinct lobes. Measurements LIEB-PV 2001: length, 7.80; maximum width, 4.16; lingual height, 2.66; labial height, 2.71. LIEB-PV 2000: length, 7.68; maximum width, 4.35; lingual height, 2.95; labial height, 3.81. Description Both molariform teeth share a similar morphology, though they seem to belong to different loci (even though fragmen- tary, LIEB-PV 2000 was larger judging from its maximum width). Because of their overall similarity with specimen MACN PV-RN 248 of F. windhauseni (Bonaparte 1986a; Fig. 6a), it is assumed that both LIEB-PV specimens more probably correspond to upper molariforms. The holotype of G. sylvaticus (LIEB-PV 2001) is more complete, though also more worn than the referred specimen LIEB-PV 2000, which lacks the mesial lobe. The occlusal surfaces of both teeth are noticeably worn at their central portions, all along the longitudinal axes; thus, the occlusal surfaces are not flat but labiolingually concave. The labial edges of both teeth are higher than the lingual edges. Their external sur- faces are nearly vertical; there is no indication of a neck between the crowns and the bases of the teeth. The bases, defined as the proximal portions of the molariforms above the closing of the grooves (von Koenigswald et al. 1999), include the roots, which are single, thick, short, rounded, and covered dorsally and laterally by a layer of cementum (Fig. 4). This feature can be observed on the lingual faces of both molariforms, where deep grooves invaginate the crown. These grooves do not reach the bases of the molari- forms, but instead they end at about one third of the pre- served crown height in lingual view (Figs. 3i and 4a). The labial and lingual faces (sidewalls) of the crown in each specimen are not smooth but somewhat irregular, or wrinkled, showing on some portions several vertical Naturwissenschaften (2012) 99:449–463 453 Fig. 3 Greniodon sylvaticus gen. et sp. nov. a–d, LIEB-PV 2001, holotype, an upper left molariform (MF?3) in occlusal (a), disto- occlusal (b, c), and labio-occlusal (d) views. b Detail of the mesial lobe is shown. e–i Referred specimen LIEB-PV 2000, a fragmentary right upper molariform (MF?2) in occlusal (e), mesio-occlusal (f), disto-occlusal (g), labio-occlusal (h), and linguo-occlusal (i) views. la labial, le leading edge, li lingual, m mesial, te trailing edge. Scale 1 mm 454 Naturwissenschaften (2012) 99:449–463 striations (see, e.g., the mesial half of LIEB-PV 2000 in Fig. 4b). They seem to correspond to areas where the outer enamel surface varies in thickness. In other gondwanather- ians (Gondwanatherium and Sudamerica), these irregular surfaces of the outer enamel surface represent adaptations for holding the cementum that surrounds the tooth (von Koe- nigswald et al. 1999: Fig. 10a). This suggests that the external faces of Greniodon’s molariforms were also at least partially covered by cementum (see below). In occlusal view, both LIEB-PV 2000 and LIEB-PV 2001 share the same basic morphology, characterized by the pres- ence of strong grooves and deep pits or fossae (named “islets” in von Koenigswald et al. 1999 nomenclature). LIEB-PV 2001 shows two major grooves: “A” and “D” (Fig. 6c), as well as islet “C”, representative of groove “C” in LIEB-PV 2000 (Fig. 6b). Also, an indentation of the enamel distal to groove “A” in the holotype seems to represent the (also slightly invaginated) groove “B” in LIEB-PV 2000 (“B” in Fig. 6b, c). Avery small islet (“3”) in the holotype seems to be homologous to a larger one, but similarly placed, in specimen LIEB-PV 2000. Additional, tiny islets mesial and distal to islet “2?” in the holotype (Fig. 6c) suggest that its unworn crown morphology may have been more complex than that of LIEB-PV 2000. Finally, islets numbered “1” and “4” in the holotype (Fig. 6c) are not evident in LIEB-PV 2000 (the former because of the lack of the mesial lobe, the latter because the distal lobe is broken; Fig. 3g). In summary, we interpret that at least five transverse lobes were present in unworn upper molariforms of G. sylvaticum, at least in those currently known. These lobes are defined by four major grooves, all of them present in LIEB-PV 2000 (though, due to the state of wear of the tooth, groove “B” is poorly distinguishable), while three of them can be seen in LIEB-PV 2001. Due to the different degrees of wear in each specimen, not all grooves are visible as such; this is the case for groove “C” in LIEB-PV 2000 (Fig. 6b), which we interpret as homologous to islet “C” in specimen LIEB-PV 2001 (Fig. 6c). With the probable exceptions of islets 1 and 4, which can be regarded as islets of the first type (sensu von Koenigswald et al. 1999: Fig. 4), islets 2 and 3 can be regarded as the second type, i.e., derived from the transverse grooves by wear. Root morphology, protohypsodonty, and wear pattern Figure 4 shows specimen LIEB-PV 2000 in (a) lingual, (b) labial, (c) mesial, and (d) dorsal views. This specimen is less worn than the holotype of G. sylvaticus and its basal portion is better preserved as well. It can be seen that there is no definite neck between the crown and the basal portion of the tooth, in such a way that the labial and lingual faces of the molariform are nearly vertical. There is a single root, which is wide and blunt, and covered by a thick layer of cementum (Fig. 4d). The labial face of the specimen is covered by cementum almost up to the occlusal plane. Apparently due to preservation, the cementum that cov- ers this side was irregularly eroded througout its extent, leaving a dorsal and distal cap with a thick layer of cementum (“ceT” in Fig. 4b), while the lower two thirds of the mesial half is thinner (“cet” in Fig. 4b). In the latter, the rugged pattern of the labial face can be appre- ciated, indicating that enamel wrinkles probably an- chored the layer of cementum. Dorsally, specimen LIEB-PV 2000 shows a thick layer of cementum (Fig. 4d). In lingual view (see also Fig. 3i), it can be seen that the lateral grooves that incise the molariform are closed at the middle portion of the molariform’s face. Above them, a slight thickening of the enamel could represent the dorsalmost portion of this tissue; however, the molariform is broken here. What seems evident is that no definite, discrete neck (as is present in brachyo- dont teeth) can be identified either in the labial or in the lingual sides of the tooth. Our interpretation is that both known specimens of G. sylvaticus represent old individuals that underwent substantial attrition on their crowns and had, at least in the adult stage, closed roots. Taking in account that the known specimens of G. syl- vaticus are longer than high, they cannot be regarded as hypsodont. (However, there is a gradient in increasing tooth growth and height that should be considered; see, e.g., MacPhee and Flemming 2003; von Koenigswald 2011). On the basis of: (1) the worn off condition of the molari- forms (suggesting a higher unworn crown); (2) their fairly steep sidewalls; and (3) the lack of a neck between the crown and the root, we consider that the molariforms of Greniodon may have been of a protohypsodont type. This type of growth is compatible with the development of roots (at least in adult stages; von Koenigswald et al. 1999; von Koenigswald 2011). The molariforms of Greniodon are characterized by the presence of grooves that define transverse lophs in relation to the dental axis. As cutting edges are most effective in mastication if they are oriented perpendicular to the power stroke, we infer a power stroke that was parallel to the molariform row (Rensberger 1973). A few additional infer- ences can also be made, assuming that the positional infor- mation provided by us proves to be correct (i.e., that both specimens are upper molariforms, and that the lobe distal to groove “A” is the mesialmost one). At least in the mesial lobe of the holotype of G. sylvaticum, it can be observed that the enamel and the dentine are eroded differentially (Fig. 3b). As in S. ameghinoi (see Krause et al 1992: Fig. 6c), the dentine at the mesial edge (leading edge) of the lobe is abraded more deeply than that of the distal edge (trailing edge; Fig. 3b). This indicates that the mandible was pulled Naturwissenschaften (2012) 99:449–463 455 backwards (palinally) during the power stroke of the masti- catory cycle. In turn, and different from other known suda- mericids, it can be observed in specimen LIEB-PV 2001 that the labial edge of the molariform is slightly higher than the lingual one (Fig. 3c). This implies that during mastication, the lower molariforms occluded not entirely face against face but instead slightly lingually in relation to the upper ones. Enamel microstructure The enamel of the molariforms of Greniodon is similar in thickness (0.105 mm) to that of Ferugliotherium (0.1 mm), thinner than that of Indian sudamericids (0.2 mm), and much thinner than that of Sudamerica and Gondwanatherium (0.2– 0.3 mm; see Krause et al. 1992; von Koenigswald et al. 1999). As in the molariforms of S. ameghinoi andGondwanatherium patagonicum, the entire enamel band inGreniodon is prismat- ic; i.e., it lacks a layer of prismless outer enamel (or PLEX). (Among gondwanatherians, PLEX is common only in the incisor enamel of Sudamerica and Gondwanatherium; von Koenigswald et al. 1999.) It is composed of a single layer of radial enamel. Prisms are separated by complete prism sheats at least in some of the prisms (Fig. 5b). In a few prisms, seams can be clearly observed; in S. ameghinoi prisms are open and arc-shaped. Within prisms, a few tubuli could be observed (Fig. 5b). We measured the maximum width of 20 prisms at the outer half of the enamel band. Their mean was of 5.57 μm, ranging from 4.29 to 7.14 μm. Their diameter is larger than that of the prisms measured near the outer enamel surface (OES) in S. ameghinoi (3.5–4.5 μm), and more closely resemble the measurements obtained near the OES in G. patagonicum (mean, 5.61 μm’ ranging from 4.06 to 7.07 μm; von Koenigswald et al. 1999). Prisms of F. wind- hauseni are smaller, having a diameter of 4.6 μm (Sigogneau- Russell et al. 1991). The interprismatic matrix (IPM) is not thick as in other gondwanatherians (e.g., Sudamerica; von Koenigswald et al. 1999), but rather thin. The IPM is not organized in compact interrow sheets as in Malagasy and Indian gond- wanatherians (von Koenigswald et al. 1999: Figs. 12–13). However, in low magnitude tangential view (Fig. 5c) it can be observed that prisms tend to show some alignment. In S. ameghinoi, this alignment can also be appreciated (von Koenigswald et al. 1999: Fig. 13b). In vertical section (Fig. 5e), it can be observed that the constant upward inclination of the prisms towards the OES is more marked than that of Sudamerica (cf. von Koenigs- wald et al. 1999: Fig. 9). Prisms arise radially at about 55° from the enamel–dentine junction and do not change direction near the OES. The IPM is neither parallel nor transversal to the orientation of the prisms but instead is set at a 65° angle to them. As mentioned, the OES lacks a PLEX. Fig. 4 Greniodon sylvaticus gen. et sp. nov. Referred specimen LIEB-PV 2001 (upper molariform, MF?3) in lingual (a), labial (b), mesial (c), and dorsal (d) views. ce Cementum, cet thinner layer of cementum (probably due to postmortem erosion of the specimen), ceT thicker layer of cementum (both areas are separated by the dotted line), de dentine, en enamel. Scale 1 mm 456 Naturwissenschaften (2012) 99:449–463 Possible homologies of the molariform loci and pattern of Greniodon Specimens LIEB-PV 2001 (holotype) and LIEB-PV 2000 of G. sylvaticus have several features in common with the best preserved upper molariform of the ferugliotheriid F. windhauseni, MACN PV RN248, referred to an “M1” by Krause et al. (1992: Fig. 2g–h; Fig. 6a of this work). Be- cause of this, we assume that both specimens of Greniodon may correspond to the upper molariform series, with one (LIEB-PV 2000) being from the right side, and the holotype from the left. Fig. 5 Greniodon sylvaticus gen. et sp. nov. Detail of the enamel microstructure of referred specimen LIEB-PV 2000 (upper molariform, MF?3; all figures are scanning electronic micrographs). a View of the mesial face of the molariform (which lacks the distal lobe), showing areas 1 and 2 (vertical and tangential views, respectively) polished for their study. b–d Tangential views of prisms at different magnifications, from polished section 2. b ×600, c ×1000, d ×2000. e View of polished sec- tion 1, vertical view, ×600. OES outer enamel surface, EDJ enamel–dentine joint Naturwissenschaften (2012) 99:449–463 457 The best-known molariform series of a gondwanatherian mammal is that of the sudamericid S. ameghinoi (see von Koenigswald et al. 1999: Fig. 5), in which the first upper molariform (MF1) has a mesial lobe that is much narrower transversely than seen in the distal teeth. By comparison, this suggests that none of the specimens referred to G. sylvaticus (and, we hypothesize, specimen MACN PV RN248 of F. windhauseni, as well) may not correspond to an MF1. In turn, the distalmost lobe in MF2 of S. ameghinoi is noticeably narrower than the medial lobes, which is the case in both LIEB-PV specimens. Also, the mesial molari- forms (MF1 and MF2) of Sudamerica have more lobes (four) than the distal ones (MF3 and MF4, with three lobes each). As Greniodon molariforms have a high number of lobes (five), it is reasonable to assume that they correspond to the mesial portion of the molariform series. If an MF1 locus can be discarded because of the relative size of the mesialmost lobe, we find that the probable locus for both known teeth of Greniodon may correspond to MF2 and MF3. It should be noted that measurements of LIEB-PV 2000 and 2001 are not the same (see above), with LIEB-PV 2000 being slightly larger and higher. This could be due either to the different stage of wear in each specimen, to individual variation, or to each belonging to a different molar locus. In that MF2 is slightly larger thanMF3 in S. ameghinoi, it could be the case that LIEB-PV 2000 is a MF2 while LIEB- PV 2001 is a MF3. Pending further discoveries of more complete specimens of Greniodon, we tentatively refer LIEB-PV 2000 to a MF?2 and LIEB-PV 2001 to a MF?3. In Fig. 6, we propose possible homologies for grooves and islets between the molariforms of F. windhauseni (Fig. 6a) and G. sylvaticus (Fig. 6b–c). Much of this inter- pretation rests on the assumption that features such as depressions or valleys (flexa) in the brachyodont molariform of Ferugliotherium can be homologized to more definite structures, such as islets and grooves, respectively, in the protohypsodont teeth of Greniodon. Consequently, the four major flexa present in the upper molariform MACN PV Fig. 6 Schematic drawings of upper molariforms of South American gondwanatherians. a Ferugliotherium windhauseni; specimen MACN PV RN248 in occlusal view. b–c Greniodon sylvaticus gen. et sp. nov. (b LIEB-PV 2000; c LIEB-PV 2001 [holotype, inverted], both in occlusal views); d–f Gondwanatherium patagonicum; specimen MACN PV RN1029 in labial (d) and occlusal (e) views; fMACN PV RN1025 in occlusal view (redrawn after Gurovich 2006); g–h Sudamerica ameghinoi; specimen MACN PV RN1983 in occlusal (g) and lingual (h) views (redrawn after Gurovich 2006). Figures are not to scale. An attempt to homologize mo- lar structures is made in a–c; capital letters major grooves separating molar lobes, numb- ers to islets. Grooves and islets are bordered by enamel and are surrounded by dentine. In all likelihood, the grooves as well as the sidewalls of the molariforms of Greniodon were filled with cementum, probably lost due to postmortem erosion of the specimen 458 Naturwissenschaften (2012) 99:449–463 RN248 of Ferugliotherium (“A”, “B”, “C”, and “D”; Fig. 6a) are also present, though developed as deep grooves in Greniodon (Fig. 6b, c). Because of wear, groove “C” is present as a long pit in LIEB-PV 2001. Islets of the first type (von Koenigswald et al. 1999; “1” and “4” in Fig. 6 of this work) are present in LIEB-PV 2001; due to its fragmentary nature, they were not preserved in LIEB-PV 2000. They would correspond to the closed depressions “1” and “4” present at the mesial and distal ends of the upper molariform of Ferugliotherium. In a protohypsodont molar, these depressions develop into fossae, surrounded by enamel and dentine. Islets of the second type (von Koenigswald et al. 1999) “2” and “3” are obvious in LIEB-PV 2000, while islet “2” is represented by several pits in LIEB-PV 2001. They would correspond to the labial ends of the flexa “B” and “C” (“2” and “3”, respectively; Fig. 6a) in Ferugliotherium. Overall, it is striking that, even though Greniodon shares more derived features with sudamericids, its occlusal molar pattern resembles more closely that of Feruglio- therium. This is so because in Ferugliotherium and Greniodon, flexa/grooves have a lingual origin and deeply carve the molari- form almost to its labial side, in contrast to those of S. ameghinoi (with labial and lingual flexa that are sub-equal in length). Due to the scarcity of upper molariforms of G. patagonicum, it is hard to define its precise occlusal morphology. Some specimens (e.g., MACN PV RN1029; Fig. 6d–e) have grooves that incise the occlusal face only partially. Others, likeMACNPVRN1025 (Fig. 6f, here regarded as a ?distalmost upper right molariform), have deep grooves much like those of Ferugliotherium or Greniodon, though fewer in number. Affinities and comparisons At first sight the absence, inGreniodon, of a tribosphenic pattern (see Davis 2011) leads to the exclusion of therians among its probable affinities. Nonetheless, several therian lineages have morphological types extremely derived from a typical tribos- phenic pattern. Among the Metatheria, polydolopid polydolopi- morphians show a series of cusps and valleys that somewhat resemble that of the Multituberculata, to which gondwanather- ians have been variously related (see Ameghino 1897). However, they lack any indication of grooves that are trans- versely oriented with respect to the dental axis. Polydolopids also lack any indication of hypsodonty. The only hypsodont types among polydolopimorphians occur within the Argyrola- goidea (Bonapartheriiformes), i.e., Argyrolagiidae and Patago- niidae. However, the latter (Patagoniidae, with simple, subcylindrical molars; Pascual and Carlini 1987) have a simpli- fied molar pattern; among the former (Argyrolagidae; Simpson 1970), inwhich ecto- and entoflexa are present, they are never as deep as in Greniodon. Another group that may merit compari- son with Greniodon in molar morphology is the derived pauci- tuberculatan family Abderitidae. However, the latter shows a quadrangular profile in its upper molars, the presence of defined cusps, and lack grooves or islets. No other metatherian lineage shows further similarities with Greniodon. Most xenarthran teeth (when present) are enamel-less and include homodont, simple-crowned molariforms (Thenius 1989), which are sub-oval in section and generally include a single, central crest (e.g., Dasypodidae). Megalonychids and bradypodids have subcylindrical teeth with extremely simplified crowns. Glyptodontids have more complicated, lobed crowns —actually, the first molariform specimens of S. ameghinoi were compared with those of glyptodonts (Scillato-Yané and Pascual 1985). However, Greniodon molariforms differ from those of glyptodonts in bearing strongly asymmetrical, transverse grooves, as well as distinct islets, some of them being remnants of grooves (in less worn teeth), and others more probably being pits or fossae. In occlusal view, the molariforms of G. sylvaticus differ from those of most basal South American native ungulates (SANU) in that the latter are bunodont and brachydont. The teeth of Kollpaniinae, Didolodontidae, and even bunodont litopterns such as Protolipternidae and Megadolodinae can be distinguished from those of Greniodon in their cusp arrangement and in their basic tribosphenic pattern. More advanced SANU, in which the cusps develop distinct lophs, and valleys turn into fossae, show convergent structures; however, none of the other Greniodon features are compa- rable to those of known SANU. Excluding the bilophodont pattern of Pyrotheria and Xenungulatha, which is quite different from the crown topography of Greniodon, neither Notoungulata, Litopterna, nor Astrapotheria show deep lin- gual grooves in their upper molars comparable to those identified in Greniodon. The absence of labial grooves in Greniodon prevents further comparisons with the ectoloph of any SANU. Finally, the Greniodonmolariform labial face is smooth and lacks the folds and grooves characteristic of Paleogene notoungulates, litopterns, and astrapotheres. Except for the presence of grooves and crests in their molars, there seem to be no further comparable structures or homologous features between the Greniodon molariform morphology and that of the upper (or lower) molars of rodents. Even taking into account early South American “caviomorph” rodents, such as Eoincamys, Eobranisamys, Eopicure (Frailey and Campbell 2004), Cachiyacuy, Can- aanimys (Antoine et al. 2011), or crown African hystricog- nath rodents such as Gaudeamus (Sallam et al. 2011), the Greniodon molariforms lack structures comparable to the obliquely oriented “metaloph”, or enamel connection between the transverse crests that occur inmost rodents. Also lacking is any indication of a “protoloph-posteroloph” connection. Greniodon also lacks the accessory cuspules usually present in the upper molars of these rodents (Sallam et al. 2011, Fig. 3). Among non-therian mammals, the lower and upper molar pattern of monotremes (i.e., those taxa in which hard teeth- Naturwissenschaften (2012) 99:449–463 459 bearing enamel and dentine still persist, such as Monotre- matum Pascual et al. 1992, Steropodon Archer et al. 1985, or Obdurodon Archer et al. 1992; see also Woodburne 2003), is quite distinct as compared to that of Greniodon. Monotreme molars have wide labial and lingual cinguli, and the crowns are divided into two separate lobes (Green 1937) by a deep valley that extends across their transverse axis from side to side (i.e., there are no grooves or flexa in a strict sense). Upper molars of monotremes have several roots, not one as in Greniodon. In the latter, the root is low, rounded, and lacks a neck at the base of the crown, in such a way that it is poorly distinct, as occurs in many protohypsodont mammals, including G. patagonicum (Bonaparte 1986b). Two major features suggest affinity between Greniodon and other members of the Gondwanatheria (see Kielan- Jaworowska et al. 2004): (1) the conspicuous presence of grooves, crests, and islets transversally oriented with respect to the long axis of its molariform teeth and (2) evidence of wear that suggests a palinal direction of mastication. This last feature, however, depends on the consistency of the molar position here inferred (i.e., that they are upper molari- forms, and that their inferred mesial lobe is in fact mesial). In the absence of more complete materials, the latter cannot be stated with certainty. Even though specimens LIEB-PV 2000 and 2001 do not exhibit unworn crown surfaces, all preserved grooves reflect a series of crests comprised of enamel, dentine and, possi- bly, cementum that filled the spaces within grooves, and all are oriented transversely. The inferred protohypsodonty of Greniodon resembles that of the sudamericid Gondwana- therium (see Gurovich 2008: text Fig. 5d), unlike the bra- chyodont ferugliotheriids. The rugged aspect of the outer enamel surface suggests a structure for holding an external layer of cementum that surrounds the molariforms, and, as mentioned, fills the grooves. This feature also characterizes the Sudamericidae (von Koenigswald et al. 1999) but not the Ferugliotheriidae. Finally, the number and orientation of grooves and islets are reminiscent of the upper molariforms of Ferugliotherium (see above and Fig. 6), while differing from those of the Sudamericidae—the latter being much more symmetrical in the distribution of flexa and islets. [However, as mentioned, Fig. 6f shows a specimen of G. patagonicum in occlusal view. If it effectively represents an upper molariform, as we suggest (see also Gurovich 2006), it can be appreciated that it also has very asymmetrical transverse grooves]. Considering the enamel microstructure, none of the ob- served features seem to prevent the allocation of G. sylvati- cum in the Gondwanatheria, while a few of them resemble similar features observed in sudamericids (which, however, are not exclusive of this group among mammals). For in- stance, the prism diameter in Greniodon resembles that of Gondwanatherium measured near the OES, which is neither “gigantoprismatic” nor “small prismatic” in size, as that characterizing many multituberculates (Carlson and Krause 1985; Krause and Carlson 1987). The incipient alignment of prisms resembles the pattern seen in the enamel of Suda- merica; a complete alignment of prisms, with distinct inter- row sheets formed by the IPM can be observed in L. mio- laka and B. bonapartei (Krause et al. 1997; von Koenigs- wald et al. 1999). The IPM is set at a large angle regarding the prisms (almost at a straight angle in Sudamerica). Dis- tinct from sudamericids, however, is the much thinner IPM in Greniodon. Also, the enamel thickness of Greniodon resembles that of Ferugliotherium rather than that of suda- mericids (much thicker). In summary, the aforementioned features, plus the ab- sence of features diagnostic of other mammalian groups (see above), suggest the attribution of Greniodon to the Gond- wanatheria. However, the mixture of features of both Suda- mericidae and Ferugliotheriidae prevents us from assigning Greniodon to either of the known families of this group of mammals. Discussion Significance of the new taxon G. sylvaticus gen. et sp. nov constitutes an outstanding addition to the Paleogene faunas of South America. It extends the record of gondwanatherian mammals in Patago- nia into the earliest middle Eocene (early Lutetian) and is the last known appearance of the Gondwanatheria in this region. No gondwanatherian taxa are recorded in immedi- ately older or younger ages/SALMAs (i.e., the early Eocene Riochican SALMA and the middle Eocene Vacan sub- SALMA), which have been relatively well-sampled in this region for several decades. Still, Greniodon would not be the only taxon exclusive to Paso del Sapo in the Patagonian Paleogene faunal scheme. Another rare taxon, Polydolops rothi, a derived species of Polydolopidae, is also present at the Laguna Fría and La Barda sites, as well as in Cerro Pan de Azúcar near Gaiman (Chubut), from levels that have been related in age to the “Sapoan” associations (Tejedor et al. 2009). This taxon is absent in other South American localities representative of older or younger chronological units. The presence of these rare taxa may be an additional argument for utilizing the “Sapoan” faunas to represent a new biochrono- logical unit, as suggested by Tejedor et al. (2009), intermedi- ate in time between the early Eocene Riochican SALMA and the middle Eocene Vacan subage of the Casamayoran SALMA. Another possibility is that the faunas from Paso del Sapo, located near the Southern Andean Range, represent a local variation compared to more eastern localities and levels. The Proto-Andean orogeny (and, thus, montane 460 Naturwissenschaften (2012) 99:449–463 biotopes and environments) was active in southern South America from the Late Cretaceous (see, e.g., Aragón et al. 2011), likely favoring the development of heterogeneous paleoenvironments that may have allowed the diversification of distinctive local faunas. A third alternative, probably related to the previous one, is that the populationsGreniodon were of very small size, and restricted to particular habitats; therefore, its discovery in Paleogene levels in other localities and levels would be more due to chance than being regularly recovered by prospecting efforts. The sample of all mammals recovered to date from the Paso del Sapo fossil localities (mainly by means of screen-washing efforts) consists of several hundred specimens. If this is representative of the combined population sizes of the more than 50 species currently recognized (Tejedor et al. 2009), it must be concluded that the medium- sizedG. sylvaticus, currently known from only two specimens, was not a dominant component of this mammalian association. The only older, Cenozoic record of gondwanatherian mammals in Patagonia is that of the sudamericid S. ameghi- noi, from the middle Paleocene (early Selandian; Gelfo et al. 2009) of Punta Peligro (Fig. 2b) in the San Jorge Gulf, Chubut Province (Scillato-Yané and Pascual 1985). In having a more advanced stage of hypsodonty, and a quite different molariform occlusal pattern, it can be inferred that Sudamer- ica does not represent an earlier stage in the evolution of Greniodon’s pattern. As argued above, it is the ferugliotheriid molariform pattern that may represent such an early stage, thus suggesting a closer phylogenetic affinity withGreniodon. If so, Greniodon represents a second, parallel evolutionary trend toward the acquisition of hypsodonty within Gondwa- natheria. Given that the Paso del Sapo faunas include exclu- sively brachydont native ungulates (Tejedor et al. 2009), these early “experiments” among Gondwanan non-therian mam- mals are particularly intriguing. Also, the ecological implica- tions of these patterns are still unclear. From the molariform morphology of Greniodon, probable herbivorous habits can be inferred (see von Koenigswald et al. 1999); however, a specific type of diet is impossible to infer at this stage. Discussing the latest Cretaceous-earliest Paleocene hia- tus, Pascual and Ortiz-Jaureguizar (2007) stated that a crit- ical turnover event occurred during this time gap among South American mammals. This event marked the end of the (mostly Mesozoic) Gondwanan Episode, and the beginning of the (Cenozoic) South American Episode, the two major phases considered by these authors in the evolution of South American mammals. Among other features characterizing this turnover, the extinction of most non-tribosphenic line- ages was regarded as prominent. Exceptions considered by Pascual and Ortiz-Jaureguizar (2007) include the survival of S. ameghinoi into the early Paleocene of Patagonia (see Scillato-Yané and Pascual 1985), and of another sudamer- icid into the Eocene La Meseta Formation in Antarctica (see Goin et al. 2006a). G. sylvaticus is one more case of persistence of a Gondwanan non-therian lineage well after the Cretaceous/Tertiary boundary in South America (more precisely, about 18 million years after the K/T boundary). If the claims of Goin et al. (2004) and Antoine et al. (2011) are considered, the gondwanatherian presence into the late Pa- leogene of the Peruvian Amazonia would add an even more remarkable, continuous persistence of Gondwanatheria for more than 25 Ma into the Early Cenozoic. Palaeogeographical and palaeoecological aspects Analyzing the possible links of the Paso del Sapo faunas, Tejedor et al. (2009) correlated them with those from the La Meseta Formation in the Antarctic Peninsula (Fig. 1). This correlation was based on several shared taxa, such as the metatherian Pauladelphys (Derorhynchidae) and a native ungulate taxon close to Notiolofos (Sparnotheriodontidae). Interestingly, the authors mentioned that the then absence of Gondwanatheria in the Paso del Sapo localities, and their presence in the Antarctic Peninsula, could have been due either to paleogeographic reasons or to “…a currently un- filled sampling bias in the Argentine Paso del Sapo sites” (Tejedor et al. 2009: 35). Now we know the latter scenario was the case. Nevertheless, it should be noted that the Antarctic gondwanatherian was regarded as a sudamericid, “…Genus and species indet., cf. Sudamerica ameghinoi” (Goin et al. 2006a: 136). As argued above, the gondwana- therian from La Barda, besides being much larger than the Antarctic specimen, seems to be more closely related to the Ferugliotheriidae, and thus shows a general, but not more specific, affinity with the La Meseta sample. The age of the La Barda mammalian association has been estimated as between 47 and 45 Ma (Tejedor et al. 2009, and literature cited therein). The closest (in time and space) floral association known to date is that of the nearby Río Pichileufú flora (a few kilometer east of the city of Bari- loche, in western Río Negro Province; Fig. 2b), whose age was estimated as 47 Ma (Wilf et al. 2005). With more than 130 formally identified entities, the Río Pichileufú flora is regarded as the most diverse Cenozoic plant association known for South America (Wilf et al. 2003, 2005). It shares numerous taxa with the older (52 Ma), also very rich, Laguna del Hunco flora, located some 25 km northwest of La Barda (Fig. 2b). Both plant associations have been esti- mated as sharing a similar paleoclimate and biome: subtrop- ical to tropical montane rainforest, with abundant moisture and a maritime climate that precluded frosts (Wilf et al. 2005). Though an early leaf physiognomic analysis led to the mean annual precipitation being assumed at about 1.5 m (Wilf et al. 2005), a more recent study of Papuacedrus (Cupressaceae) from Laguna del Hunco led Wilf et al. (2009) to argue in favor of a 2.5 m/year precipitation min- imum estimate for the Laguna del Hunco and Río Pichileufú Naturwissenschaften (2012) 99:449–463 461 floras. The authors point out that “…high precipitation and a lack of frost, combined with biotic interchange both with Antarctica (and beyond to Australasia) and the neotropics, emerge as the most logical explanations for the extraordi- nary diversity of plants, insects, and insect-feeding damage found at Laguna del Hunco and Río Pichileufú” (Wilf et al. 2009: 2044). Such were the environments that probably surrounded the evolution of G. sylvaticus and other mam- mals in western Patagonia by the early Lutetian. A series of stepwise global cooling events, beginning by the middle Eocene, may have played a role in the decline and subsequent extinction of several Mesozoic biotic relicts in South America. The successive opening of the Drake Passage by the middle and late Eocene (Scher and Martin 2006), an event of global consequences, may have been the trigger of such declines, particularly in nearby continental areas such as Patagonia. In short, Patagonian gondwanather- ians may have been one of the first casualties of the beginning of the end of the Greenhouse World. Acknowledgments We thank Gabriel M. Martin, Leonardo Avilla, Ariel Humai, Constanza Koefoed, María Luisa Pemberton, and Érika Abrantes for their assistance in the fieldwork at La Barda and in the laboratory; Marcela Tomeo for the figures that illustrate this work; Carolina Vieytes and Rafael Urréjola for the scanning electron micro- graphs; discussions with Carolina were also very helpful while con- sidering the enamel microstructure of Greniodon. To Alejandro Kramarz for facilitating the MACN (Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”, Buenos Aires) gondwanatherian collections for their study; Juan J. Molly for making casts of the new specimens. Editors of Naturwissenschaften, Dr. David Krause, and two anonymous reviewers provided numerous, useful comments to the original manuscript. FJG, MFT, and LC thank CONICET (Consejo Nacional de Investigaciones Científicas y Técnicas, Argentina, PIP 5621 and PIP 0361) for its financial support in the lab and in the field. FJG thanks the Alexander von Humboldt Foundation. The authors declare that they have no conflict of interest. References Ameghino F (1897) Mammifères crétacés de l´Argentine. (Deuxième contribution à la connaissance de la faune mammalogique des couches à Pyrotherium). Bol Inst Geogr Argent 18:406–521 Antoine P-O, Marivaux L, Croft DA, Billet G, Ganerød M, Jaramillo C, Martin T, Orliac MJ, Tejada J, Altamirano AJ, Duranthon F, Fanjat G, Rousse S, Salas Gismondi R (2011) Middle Eocene rodents from Peruvian Amazonia reveal the pattern and timing of caviomorph origins and biogeography, Proc. R. Soc. B, electronic suppl. material: http://dx.doi.org/10.1098/rspb.2011.1732. Accessed on 12 Oct 2011 Aragón E, Mazzoni MM (1997) Geología y estratigrafía del complejo volcánico piroclástico del río Chubut medio (Eoceno), Chubut, Argentina. Rev Asoc Geol Argent 52(3):243–256 Aragón E, Romero EJ (1984) Geología, paleoambientes y paleobotánica de yacimientos terciarios del occidente de Río Negro, Neuquén y Chubut. 9 Congr Geol Argent 4:475–507 Aragón E, Mazzoni MM, Merodio J (1987) Caracterización geoquí- mica de la Ignimbrita Barda Colorada en el Río Chubut medio. 10 Congr Geol Argent 4:171–173 Aragón E, Goin FJ, Aguilera YE, Woodburne MO, Carlini AA, Roggiero MF (2011) Palaeogeography and palaeoenvironments of northern Patagonia from the Late Cretaceous to the Miocene: the Palaeogene Andean gap and the rise of the North Patagonian High Plateau. Biol J Linn Soc 103:305–315 Archangelsky S (1974) Sobre una edad de la flora del Hunco, Provincia de Chubut. Ameghiniana 11:413–417 Archer M, Flannery TF, Ritchie A, Molnar RE (1985) First Mesozoic mammal from Australia—an early Cretaceous monotreme. Nature 318:363–366 Archer M, Jenkins FA Jr, Hand SJ, Murray P, Godthelp H (1992) Description of the skull and non-vestigial dentition of a Miocene platypus (Obdurodon dicksoni n. sp.) from Riversleigh, Australia, and the problem of monotreme origins. In: Augee ML (ed) Platypus and Echidnas. R Zool Soc NSW, pp 15–27 Bonaparte JF (1986a) Sobre Mesungulatum houssayi y nuevos mamí- feros cretácicos de Patagonia, Argentina. 4 Congr Argent Paleont Bioestrat 2:48–61 Bonaparte JF (1986b) A new and unusual Late Cretaceous mammal from Patagonia. J Vertebr Paleontol 6:264–270 Bonaparte JF (1987) The late Cretaceous Fauna of Los Alamitos, Patagonia, Argentina. In: Bonaparte JF (ed), Rev Mus Argent Cs Nat, Paleontol 3, Buenos Aires, pp 103–178 Bonaparte JF (1990) New Late Cretaceous mammals from the Los Alamitos Formation, northern Patagonia. Natl Geogr Res 6:63–93 Campbell KE Jr, Frailey CD, Romero-Pittman L (2004) The Paleogene Santa Rosa local fauna of Amazonian Perú: geographic and geologic setting. In: Campbell KE Jr (ed) The Paleogene Mam- malian Fauna of Santa Rosa, Amazonian Perú. Nat Hist Mus Los Angeles County, Sci Ser 40, pp. 3–14 Carlson SJ, Krause DW (1985) Enamel ultrastructure of multitubercu- late mammals: an investigation of variability. Contrib Mus Pale- ont Univ Mich 27:1–50 Davis DM (2011) Evolution of the tribosphenic molar pattern in early mammals, with comments on the “dual-origin” hypothesis. J Mamm Evol 18:227–244 Frailey CD, Campbell KE Jr (2004) Paleogene rodents from Amazonian Perú: The Santa Rosa local fauna. In: Campbell KE, Jr. (ed) The Paleogene Mammalian Fauna of Santa Rosa, Amazonian Perú. Nat Hist Mus Los Angeles County, Sci Ser 40, Los Angeles, pp. 71–130 Gelfo JN, Reguero MA, López GM, Carlini AA, Ciancio MR, Chornogubsky L, Bond M, Goin FJ, Tejedor MF (2009) Eocene mammals and continental strata from Patagonia and Antarctic Peninsula. In: Albright LB, III (ed.) Papers on Geology, Vertebrate Paleontology, and Biostratigraphy, in Honor of Michael O. Woodburne. Mus North Ariz Bul 64, Flagstaff, pp. 567– 592. Goin FJ, Vieytes EC,VucetichMG,Carlini AA, BondM (2004) Enigmatic mammal from the Paleogene of Perú. In: Campbell KE, Jr. (ed) The Paleogene Mammalian Fauna of Santa Rosa, Amazonian Perú. Nat Hist Mus Los Angel County Sci Ser 40, Los Angeles, pp. 145–153 Goin FJ, Reguero MA, Pascual R, Koenigswald W von, Woodburne MO, Case JA, Marenssi SA, Vieytes EC, Vizcaíno SF (2006a) First gondwanatherian mammal from Antarctica. In: Francis JE, Pirrie D, Crame JA (eds) Cretaceous-Tertiary High-Latitude Pale- oenvironments, James Ross Basin, Antarctica, Spec Publ Geol Soc London, pp. 145–161 Goin FJ, Pascual R, Tejedor MF, Gelfo JN, Woodburne MO, Case JA, Reguero MA, Bond M, López GM, Cione AL, Sauthier DU, Balarino L, Scasso RA, Medina FA, Ubaldón MC (2006b) The earliest Tertiary therian mammal from South America. J Vertebr Paleontol 26(2):505–510 Gosses J, Carroll A, Aragón E, Singer E (2006) The Laguna del Hunco Formation: Lacustrine and Sub-Aerial Caldera Fill Chubut Province, Argentina. Geol Soc Amer Ann Meet 38(7):502, Abstracts 462 Naturwissenschaften (2012) 99:449–463 http://dx.doi.org/10.1098/rspb.2011.1732 Green HLHH (1937) The development and morphology of the teeth of Ornithorhynchus. Phil Trans R Soc Lond B 228:367–420 Gurovich Y (2006) Bio-evolutionary aspects of Mesozoic mammals: description, phylogenetic relationships and evolution of the Gondwanatheria (Late Cretaceous and Paleocene of Gondwana). Dissertation, Universidad Nacional de Buenos Aires Gurovich Y (2008) Additional specimens of sudamericid (Gondwana- theria) mammals from the early Paleocene of Argentina. Palae- ontol 51:1069–1089 Gurovich Y, Beck R (2009) The phylogenetic affinities of the enigmatic mammalian clade Gondwanatheria. J Mammal Evol 16:25–49 Kielan-Jaworowska Z, Cifelli RL, Luo Z-X (2004) Mammals from the age of dinosaurs: origins, evolution, and structure. Columbia Univ Press, New York Kielan-Jaworowska Z, Ortiz-Jaureguizar E, Vieytes MC, Pascual R, Goin FJ (2007) First ?cimolodontan multituberculate mammal from South America. Acta Palaeontol Pol 52(2):257–262 Krause DW, Bonaparte JF (1990) The Gondwanatheria, a new suborder of Multitubercualta from South America. J Vertebr Paleontol 10:13A Krause DW, Bonaparte JF (1993) Superfamily Gondwanatherioidea: a previously unrecognized radiation of multituberculate mammals in South America. Proc Natl Acad Sci 90:9379–9383 Krause DW, Carlson SJ (1987) Prismatic enamel in multituberculate mammals: tests of homology and polarity. J Mammal 68:755–765 Krause DW, Kielan-Jaworowska Z, Bonaparte JF (1992) Ferugliothe- rium Bonaparte, the first known multituberculate from South America. J Vertebr Paleont 12:351–376 Krause DW, Prasad GVR, Koenigswald W, Sahni A, Grine FE (1997) Cosmopolitanism among Gondwanan Late Cretaceous mammals. Nature 390:504–507 Krause DW, Gottfried MD, O’Connor PM, Roberts EM (2003) A Cretaceousmammal from Tanzania. Acta Palaeontol Pol 48:321–330 Lage J (1982) Descripción geológica de la Hoja 43c, Gualjaina. Pro- vincia del Chubut. Serv Geol Nac Bol 189:1–72 Linnaeus C (1758). Systema naturae per regna tria naturae: secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis (10th ed.). Laurentius Salvius, Stockholm MacPhee RDE, Flemming C (2003) A possible Heptaxodontine and other caviidan rodents from the Quaternary of Jamaica. Amer Mus Novit 3422:1–42 Marsh OC (1880) Notice of new Jurassic mammals representing two new orders. Am J Sci Arts ser 3:235–239 Mazzoni MM, Aragón E (1985) El complejo volcánico piroclástico de la Formación Huitrera (Paleoceno-Eoceno), en el área del Río Chu- but medio, República Argentina. 4 Congr Geol Chil 3:275–300 Mazzoni MM, Aragón E, Merodio J (1989) La Ignimbrita Barda Colorada del Complejo Volcánico Piroclástico del Río Chubut medio. Rev Asoc Geol Argent 44:246–258 Mazzoni MM, Kawashita K, Harrison S, Aragón E (1991) Edades radimétricas eocenas en el borde occidental del Macizo Norpata- gónico. Rev Asoc Geol Argent 46:150–158 Mones A (1987) Gondwanatheria, un nuevo orden de mamíferos sudamericanos (Mammalia: Edentata: ?Xenarthra). Comun Pale- ontol Museo Hist Nat Montevideo 1:237–240 Pascual R, Carlini AA (1987) A new superfamily in the extensive radiation of South American Paleogene marsupials. Fieldiana Zool n ser 39:99–110 Pascual R, Ortiz-Jaureguizar E (2007) The Gondwanan and South Amer- ican episodes: twomajor and unrelated moments in the history of the South American mammals. J Mamm Evol 14:75–137 Pascual R, Archer M, Ortiz-Jaureguizar E, Prado JL, Godthelp H, Hand SH (1992) First discovery of monotremes in South America. Nature 356:704–705 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 19:375–384 Petersen CS (1946) Estudios geológicos en la región del río Chubut medio. Dir Gen Minas y Geol (Buenos Aires) 59:1–137 Prasad GVR, Verma O, Sahni A, Krause DW, Khosla A, Parmar V (2007) A new Late Cretaceous gondwanatherian mammal from central India. Proc Indian Natl Sci Acad 73:17–24 Rapela C, Spalletti L, Merodio J, Aragón E (1984) El vulcanismo Paleoceno-Eoceno de la provincia volcánica Andino-Patagónica. 9 Congr Geol Argent 1:189–213 Rensberger JM (1973) An occlusion model for mastication and dental wear in herbivorous mammals. J Paleontol 47:515–528 Rougier GW, Chornogubsky L, Casadío S, Paéz ArangoN,Giallombardo A (2009) Mammals from the Allen Formation, Late Cretaceous, Argentina. Cretaceous Res 30:223–238 Ruiz LE (2006) Estudio Sedimetológico y Estratigráfico de las Formaciones Paso del Sapo y Lefipán en el Valle Medio del Río Chubut. Dissertation, Universidad Nacional de Buenos Aires Sallam HM, Seiffert ER, Simons EL (2011) Craniodental Morphology and Systematics of a New Family of Hystricognathous Rodents (Gaudeamuridae) from the Late Eocene and Early Oligocene of Egypt. PLoS One 6. doi:10.1371/journal.pone.0016525 Scher HD, Martin EE (2006) Timing and climatic consequences of the opening of Drake Passage. Science 312:428–430 Scillato-Yané G, Pascual R (1985) Un peculiar Xenarthra del Paleo- ceno Medio de Patagonia (Argentina). Su importancia en la sis- temática de los Paratheria. Ameghiniana 21:173–176 Sigogneau-Russell D, Bonaparte JF, Robert RM, Escribano V (1991) Ultrastructure of dental tissues of Gondwanatherium and Suda- merica (Mammalia, Gondwanatheria). Lethaia 24:27–38 Simpson GG (1970) The Argyrolagidae, extinct South American Marsupials. Bull Mus Comp Zool 139:1–86 Tejedor MF, Goin FJ, Gelfo JN, López GM, 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 Pata- gonia, Argentina. Amer Mus Novit 3638:1–43 Thenius E (1989) Zähne und Gebiß der Säugetiere. Handbook of zoology, volume VIII, part 56. Walter de Gruyter, Berlin Volkheimer W, Lage L (1981) Descripción geológica de la Hoja 42 c, Cerro Mirador, provincia del Chubut. Bol Serv Geol Nac 181:1–71 von Koenigswald W (2011) Diversity of hypsodont teeth in mammalian dentitions—construction and classification. Palaeontographica Abt A294:63–94 von Koenigswald W, Sander PM (1997) Glossary of terms used for enamel microstructures. In: von Koenigswald W, Sander PM (eds) Tooth enamel microstructure. AA Balkema, Rotterdam, pp 267– 280 von Koenigswald W, Goin FJ, Pascual R (1999) Hypsodonty and enamel microstructure in the Paleocene gondwanatherian mammal Sudamerica ameghinoi. Acta Palaeontol Pol 44:263–300 Wilf P, Cúneo NR, Johnson KR, Hicks JF, Wing SL, Obradovich JD (2003) High plant diversity in Eocene South America: evidence from Patagonia. Science 300:122–125 Wilf P, Johnson KR, Cúneo NR, Smith ME, Singer BS, Gandolfo MA (2005) Eocene plant diversity at Laguna del Hunco and Río Pichileufú, Patagonia, Argentina. Amer Nat 165:634–650 Wilf P, Little SA, Iglesias A, Zamaloa MC, Gandolfo MA, Cúneo R, Johnson KR (2009) Papuacedrus (Cupressaceae) in Eocene Pata- gonia: a new fossil link to Australasian rainforests. Amer J Bot 96 (11):2031–2047 Wilson GP, Das Sarma DC, Anantharaman S (2007) Late Cretaceous sudamericid gondwanatherians from India with paleobiogeo- graphic considerations of Gondwanan mammals. J Vertebr Pale- ontol 27:521–531 Woodburne MO (2003) Monotremes as pretribosphenic mammals. J Mamm Evol 10:195–248 Naturwissenschaften (2012) 99:449–463 463 http://dx.doi.org/10.1371/journal.pone.0016525 Persistence of a Mesozoic, non-therian mammalian lineage (Gondwanatheria) in the mid-Paleogene of Patagonia Abstract Introduction Methods Systematic paleontology Description Root morphology, protohypsodonty, and wear pattern Enamel microstructure Possible homologies of the molariform loci and pattern of Greniodon Affinities and comparisons Discussion Significance of the new taxon Palaeogeographical and palaeoecological aspects References