PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3566, 54 pp., 10 figures, 3 tables May 16, 2007 New Jurassic Mammals from Patagonia, Argentina: A Reappraisal of Australosphenidan Morphology and Interrelationships GUILLERMO W. ROUGIER,1,2 AGUSTI´N G. MARTINELLI,3 ANALI´A M. FORASIEPI,2 AND MICHAEL J. NOVACEK4 ABSTRACT A new mammal, Henosferus molus, n.gen. and n.sp., from the Callovian–Oxfordian (latest Middle to earliest Late Jurassic) Can˜ado´n Asfalto Formation from Chubut Province (Argentina) is described. This taxon corresponds to a new species clearly different from Asfaltomylos patagonicus from the same locality and stratigraphic level. This new species is based on three lower jaws with relatively well- preserved dentition. The lower jaw shows a primitive morphology having a Meckelian groove, a prominent medial flange associated with a lateral ridge of the dentary, and a deep dentary trough, which possibly indicates the presence, even though reduced, of postdentary bones still attached to the dentary. The lower dental formula is i4, c1, p5, m3. The premolars are simple, bearing a main cusp, while the molars appear to be tribosphenic, with an obtuse to right-angled trigonid and a basined talonid with three cusps. This association of plesiomorphic features in the jaw and derived features in the molars is documented in several taxa of the recently proposed Australosphenida. A phylogenetic analysis of mammaliaforms nests the new species with Asfaltomylos from the same locality and stratigraphic level; Henosferidae, new family, is recognized for Asfaltomylos and Henosferus, representing the basal radiation of Australosphenida. Henosferidae is the sister group to Ambondro from the Middle Jurassic of Madagascar, which, in agreement with previous phylogenies, is the sister taxon to the remaining australosphenidans. Additionally, our phylogenetic analysis does not support 1 Department of Anatomical Sciences and Neurobiology, University of Louisville, Louisville, KY 40292 (grougier@ louisville.edu). 2 Department of Anatomical Sciences and Neurobiology, University of Louisville, Louisville, KY 40292 (amfora01@ gwise.louisville.edu). 3 Museo Argentino de Ciencias Naturales ‘‘Bernardino Rivadavia’’, Av. Angel Gallardo 470, C1405DJR, Buenos Aires, Argentina ([email protected]). 4 Division of Paleontology, American Museum of Natural History ([email protected]). Copyright E American Museum of Natural History 2007 ISSN 0003-0082 2 AMERICAN MUSEUM NOVITATES NO. 3566 the inclusion of australosphenidans within eutherians. Henosferids likely retained some connection of the postdentary elements with the dentary; therefore, if the inclusion of Monotremata within Australosphenida is confirmed, final freeing of the postdentary elements and development of a tri- ossicular middle ear would be convergent events in Monotremata and Theria.Finally,the distinctiveness of the yet sparse South American record of Jurassic mammals when compared with the slightly better documented Cretaceous data is emphasized. The clear faunistic break between the Middle Jurassic and Early/Late Cretaceous underlies our rudimentary understanding of the evolution of Mesozoic mammals in Gondwana. INTRODUCTION tina (Rauhut et al., 2002; Martin and Rauhut, 2005). The record of Jurassic mammaliaforms from Asfaltomylos patagonicus represents the Gondwana is restricted to a few localities from only bona fide mammaliaform hitherto known Africa, Madagascar, and South America, where from the South American Jurassic (Rauhut their known diversity and abundance are, as et al., 2002; Martin and Rauhut, 2005). yet, much lower than those of boreal land- The ichnospecies Ameghinichnus pataganicus masses (Bonaparte, 1986a, 1990, 1995; from the Middle Jurassic La Matilde Bonaparte and Kielan-Jaworowska, 1987; Luo Formation (Chubut, Argentina) has tradition- et al., 2002; Kielan-Jaworowska et al., 2004). ally (Casamiquela, 1961, 1964; Bonaparte, Megazostrodon rudnerae and Erythrotherium 1978; Kielan-Jaworowska and Gambaryan, parringtoni (Crompton and Jenkins, 1968, 1994) been interpreted as representing a 1979; Crompton, 1974; Jenkins and Par- mammal (or mammaliaform). In addition, rington, 1976; Gow, 1986) are the only relative- Brasilichnium elusivum and an unnamed icho- ly complete Gondwanan mammaliaforms from taxon from the lower Jurassic Botucatu the Early Jurassic (Stormberg Group of Formation (Sa˜o Paulo, Brazil; Leonardi, Lesotho and South Africa). Isolated teeth from 1994; Rainforth and Lockley, 1996) have also the likely Early Jurassic of India are suggestive been regarded as mammals, but the systematic of a large diversity of ‘‘triconodontid’’ and status of these species is uncentain. ‘‘symmetrodont’’ animals, but their incomplete The Cretaceous record of mammaliaforms nature makes them difficult to integrate into in Gondwana is incomplete, but relatively more general interpretations of the early di- more diverse and abundant than that from versification of the mammalian lineage (Datta older rocks. Mammals or close relatives have et al., 1978; Datta, 1981; Yadagiri, 1984, 1985; been described from the Early Cretaceous of Datta and Das, 1996; Prasad and Manhas, Australia (Archer et al., 1985; Flannery et al., 1997, 2002). 1995; Rich et al., 1997, 1999a, 2001a,b, On the other hand, the Gondwanan latest 2002; Rich and Vickers-Rich, 2004), Moro- Middle through Late Jurassic shows a relatively cco (Sigogneau-Russell, 1991a,b, 1995, 2003; greater diversity of mammaliaforms, including Sigogneau-Russell and Ensom, 1998), Cam- the archaic ‘‘triconodont’’ Tendagurodon ja- eroon (Brunet et al., 1988, 1990; Jacobs et al., nenschi (Heinrich, 1998), the first occurrence of 1988), and probably Tanzania (Krause et al., cladotherians Brancatherulum tendagurense and 2003), and from the Late Cretaceous of Tendagurutherium dietrichi (Branca, 1916; Madagascar (Krause et al., 1994; Krause and Dietrich, 1927; Simpson, 1928a; Prothero, Grine, 1996; Krause, 2001) and India (Prasad 1981; Heinrich, 1998), and the probable har- and Sahni, 1988; Prasad et al., 1994; Prasad amiyid Staffia aenigmatica (Heinrich, 1999, and Godinot, 1994; Anantharaman and Das 2001) from the Late Jurassic Tendaguru beds Sarma, 1997; Krause et al., 1997). In South (Tanzania). Tribosphenic-like forms are repre- America, the diversity of mammaliaforms is sented by the Middle Jurassic Ambondro relatively better known than in the rest of mahabo (Flynn et al., 1999) from the Isalo III Gondwana; they have been discovered in the levels of Mahajanga Basin, Madagascar, Early and Late Cretaceous of Argentina and Asfaltomylos patagonicus from the latest (Bonaparte, 1986a, 1990, 1994, 1995, 2002; Middle to earliest Late Jurassic Can˜ado´n Bonaparte and Rougier, 1987; Pascual et al., Asfalto Formation, Chubut Province, Argen- 2000) and in the Late Cretaceous of Brazil 2007 ROUGIER ET AL.: SOUTH AMERICAN JURASSIC MAMMAL 3 (Bertini et al., 1993) and Bolivia (Gayet et al., Ausktribosphenos, and Bishops but also 2001). Among these finds, several taxa are Monotremata as well (Luo et al., 2001a, based on fragmentary and isolated elements 2002; Rauhut et al., 2002; Martin and that result in a uncertain taxonomic position Rauhut, 2005). This hypothesis of inclusion for many of them. Moreover, some of these of monotremes within Australosphenida was remains show a peculiar combination of critized by some (e.g., Rich et al., 2002; characters that has opened new questions Woodburne, 2003; Woodburne et al., 2003) about the evolutionary history of mammalian who alternatively suggested that australosphe- features. A wealth of new materials has nidans (excluding Monotremata) are a mono- recently been reported on briefly from a variety phyletic group, although these authors of Cretaceous formations from Patagonia, placed it close to, or inside, Placentalia. Argentina (e.g., Rougier et al., 2003a) that Under this hypothesis of a restricted, mostly span a wide geographical, temporal, and Mesozoic Australosphenida (i.e., excluding systematic range. These new specimens will Monotremata), monotremes would be basal enrich the material basis for discussion of the to Multituberculata plus Zatheria (Wood- faunal changes in the late Mesozoic of South burne et al., 2003). In short, under both America and of Gondwana in general. competing hypotheses there is a clade of The discovery of taxa interpreted as having Mesozoic gondwanan mammals that can be a tribosphenic molar pattern, in some cases dubbed Australosphenida; its various phylo- associated with primitive mandibular features genetic positions (nested inside Eutheria, in the Middle Jurassic of Madagascar (Flynn forming a monophyletic group together with et al., 1999) and Argentina (Rauhut et al., Monotremata, or anywhere in between) are 2002; Martin and Rauhut, 2005) and the Early still controversial. Cretaceous of Australia (Archer et al., 1985; In this contribution, we report new Jurassic Rich et al., 1997, 1999a, 2001a, b), has led to mammalian specimens from South America the postulation of a diphyletic acquisition of discovered during several field seasons carried the tribosphenic molar pattern (Luo et al., out jointly by staff of the University of 2001a, 2002). The core of this new interpreta- Louisville, American Museum of Natural His- tion is that the tribosphenic molar pattern tory, and Museo Paleontolo´gico ‘‘Egidio Feru- evolved in two distinctive lineages, named glio’’ during 2002–2004 at the Queso Rallado Australosphenida