PAUCITUBERCULATA, AMEGHINO, 1894

Goin, Francisco J., Candela, Adriana M., Abello, M. Alejandra & Oliveira, Edison V., 2009, Earliest South American paucituberculatans and their significance in the understanding of ‘ pseudodiprotodont’ marsupial radiations, Zoological Journal of the Linnean Society 155 (4), pp. 867-884 : 879-880

AS PAUCITUBERCULATA

Among the overwhelming diversity of South American Palaeocene marsupials, known mainly from remains found at Tiupampa ( Bolivia; see Marshall & de Muizon, 1988; de Muizon, 1991) and Itaboraí ( Brazil; see de Paula Couto, 1952a, b, 1962, 1970; Marshall, 1987), most marsupial lineages that evolved during the Cenozoic in this continent were already present: Didelphimorphia , Microbiotheria , Sparassodonta , Polydolopimorphia , and Paucituberculata . The new taxa Riolestes capricornicus and Bardalestes hunco constitute the oldest evidence of undoubted Paucituberculata in South America. Note that the three synapomorphies (labially salient hypoconid, laterally compressed entoconid, and StB or StB and StC + StD larger than the paracone and metacone, respectively) that nested Bardalestes , Riolestes and the remaining Paucituberculata (i.e. Palaeothentoidea + Caenolestoidea ), are coincident with those features previously considered as distinctive of this group ( Abello, 2007). This agrees with the definition of the Paucituberculata as the clade stemming from the most recent common ancestor of Riolestes , Bardalestes , and the Palaeothentoidea + Caeonoestoidea clade. Within Paucituberculata , Bardalestes and Riolestes are the most basal and primitive members of the order, showing the plesiomorphic condition for each feature present in the remaining Paucituberculata . Riolestes and Bardalestes are informative on the early evolution of the molar pattern in the Paucituberculata . Among them, five features support the Caenolestoidea + Palaeothentoidea clade: paracristid of m1 lacking a notch (111); hypoconulid somewhat reduced and anteroposteriorly compressed (161); open centrocrista, with the postparacrista and premetacrista fused to the lingual slopes of the STB and STD, respectively (333); metaconule much larger than the paraconule; and labiolingually compressed (391) stylar cusps which are lingually expanded, but not reaching the lingual face of the protocone (372). Taking into account the plesiomorphic conditions of Riolestes and Bardalestes , the inferred ancestral molar pattern of the Paucituberculata is characterized by the presence of a paracristid notch in the m1, a well-developed hypoconulid, uncompressed stylar cusps, a closed and deeply ‘V’-shaped centrocrista, and a metaconule larger than the paraconule but not enlarged or ‘hypocone’-like. This latter feature gives to the upper molar a subtriangular occlusal design that differs from the typical quadrangular pattern of the remaining Paucituberculata .

Bardalestes and Riolestes not only shed light on the phylogenetic relationships of the remaining Paucituberculata (see below) but also give some hints on the very origin, and timing, of this marsupial clade. Meredith et al. (2008) have recently reviewed the relationships and timeline of all major marsupial clades. Their results, applying MULTIDIVTIME and BEAST analyses for the divergence estimates, conclude that the ‘All marsupials but Didelphimorphia’ clade (i.e. the Paucituberculata plus the Australidelphia of their phylogeny in Meredith et al. 2008, fig. 2) split between 71.3 and 85.6 Mya (see their table 5). That is, the late (but not latest) Cretaceous. In sharp contrast, their divergence estimates for the Paucituberculata are stated to be between 9.2 and 14.1 Mya. The empirical evidence here provided on the fossil record of the Paucituberculata supports none of these estimates. The oldest known Paucituberculata is the Late Palaeocene Riolestes capricornicus . Derorhynchids, which stand as a probable sister group of paucituberculatans, have their oldest record in the latest early Palaeocene (the latest Danian Banco Negro deposits; Bond et al., 1995). The much better known, earliest Palaeocene fauna of Tiupampa, lacks any record not only of paucituberculatans but also of derorhynchids.

CLADISTICS OF THE PAUCITUBERCULATA

The present analysis yielded several unexpected results regarding the traditional systematic concepts for Paucituberculata marsupials: (1) the exclusion of Pliolestes from the ‘Pichipilinae’ (sensu Marshall et al., 1990) and their grouping together with the Caenolestidae Stilotherium , Caenolestes , and Rhyncholestes ; and (2) the grouping of Pichipilus and Phonocdromus as sister taxa to Palaeothentidae + Abderitidae clade. These conclusions are discussed separately.

1. In his original description of Pliolestes tripotamicus, Reig (1955) pointed out the close affinities of this genus with Phonocdromus (see also Pascual & Herrera, 1973). Subsequently Marshall (1980), in his review of the South American Caenolestoidea , included Pliolestes in the new tribe Pichipilini , together with Phonocdromus and Pichipilus . Finally, Marshall et al. (1990) raised this taxon to a subfamiliar rank: Pichipilinae (Caenolestidae) .

The results of our analysis support the hypothesis that Pliolestes does not form a monophyletic group with Pichipilus and Phonocdromus , but instead with the ‘Caenolestini’ sensu Marshall (1980). Actually, the concept of Caenolestidae proposed in this paper is restrained to the following genera: Caenolestes , Pliolestes , Rhyncholestes , and Stilotherium , and is characterized by the following synapomorphies (see Fig. 7 ): high entocristid in m1–3 (171), curved entocristid in m1-3 (191), m4 single-rooted and greatly reduced in relation to m3 (251). Stilotherium is the sister taxon to ( Pliolestes ( Caenolestes + Rhyncholestes ). The latter group is supported by the synapomorphy: entoconid posteriorly located (201). Rhyncholestes plus Caenolestes share one synapomorphy: very shallow metacristid in m2–3 (291). Pliolestes shows one homoplasic character shared with the representatives of the Palaeothentoidea (see bellow): para- and metaconids of m3 very close or fused (231).

Note that neither the living caenolestid Lestoros nor the extinct Pseudhalmarhiphus were considered in this paper. The holotype (and only known remain) of the Deseadan species Pseudhalmarhiphus guaraniticus Ameghino (1899) (only species of the genus) has disappeared from Argentinean collections, so almost nothing can be said of its affinities. Marshall (1976, 1980), based on illustrations of the holotype published by Ameghino (1894), said that ‘... Pseudhalmarhiphus is structurally similar to and is probably ancestral to Stilotherium dissimile ’ ( Marshall, 1980: 35) . Finally the generic validity of Lestoros inca ( Thomas, 1917) – the only species of this genus geographically restrained to the Andean region south of Peru – has been discussed (e.g. Simpson, 1970). Marshall (1980) stated that Caenolestes and Lestoros can hardly be distinguished.

2. Another result of this analysis is the exclusion of Pichipilus and Phonocdromus from the Caenolestidae and their grouping as sister taxa of palaeothentids + abderitids. The features that, according to Marshall (1980), justify regarding the ‘Caenolestinae’ – including Stilotherium , Pseudhalmarhiphus , Caenolestes , Lestoros , Rhyncholestes , Pliolestes , Phonocdromus , and Pichipilus – as a monophyletic group, are the following: (1) doublerooted and functional p2; (2) p3 large, double-rooted and subequal or higher than the trigonid of m1; (3) M/m 1–4 tuberculo-sectorial, without lophs in unworn molars; (4) m1 with prominent metaconid; (5) nonmodified talonid and trigonid areas; (6) m2–3 with differentiated trigonids and talonids; (7) talonid much larger than trigonid in occlusal view; (8) M1–3 with metacone (‘intermediate conule’ sensu Marshall, 1980) labially orientated; (9) M1 quadritubercular; (10) neither P3 nor M1, p2 or m1 are modified as sectorial teeth; (11) presence of an antorbital vacuity between the nasal, maxillary, and frontal. However, none of these features are synapomorphies within the Paucituberculata , nor are present in all ‘Caenolestinae’ taxa: (1) a doublerooted and functional p2 is the generalized condition in all marsupials; (2) p3 is quite small compared with the development of this tooth in the Palaeothentidae and Abderitidae – and, in any case, its development is only moderate with respect to other marsupial groups; (3) the molars of the Palaeothentidae are also tuberculo-sectorial and lack lophs; (4) in the Palaeothentidae the metaconid is also prominent; (5) the statement that that the trigonid and talonid areas in the m1 of ‘Caenolestidae’ are ‘non-modified’ was not verified in our study: in Caenolestes , as well as in Rhyncholestes , Stilotherium , and Pliolestes , the para- and metaconids are separated and the trigonid lacks a basin; (6) trigonids and talonids are differentiated in all the Paucituberculata and, besides, both Pichipilus and Phonocdromus share the same differentiation between both regions as the Palaeothentidae ; (7) talonid wider than trigonid is not an exclusive feature of ‘Caenolestidae’ sensu Marshall (1980); rather, as it has been pointed out, it is already present in a generalized form of Paucituberculata : Riolestes capricornicus ; (8) the metacone, when present, is always labially orientated, adjacent to the stylar cusp D; (9) M1 is quadritubercular in all Paucituberculata – although with slight variants; see Character Analysis; (10) ‘P3, M1, p2, and m1 not differentiated in sectorial teeth’ is the generalized condition in all marsupials; (11) finally, the presence of the vacuities cannot be verified in Pichipilus or Phonocdromus (see Character Analysis). This latter character is important because, as Marshall (1980: 127) himself recognized, despite the other features listed in his diagnosis, this one would be actually the only synapomorphy in common with his ‘Caenolestinae’ (‘Caenolestini’ + ‘Pichipilini’). As mentioned above, the only specimen of ‘Pichipilini’ in which a skull has been preserved, with the rostrum area, is MLP 66-I-17-204, assignable to Pichipilus centinelus ; however, the presence or absence of antorbital vacuities cannot be determined because of the damage to the material (contra Marshall & Pascual, 1977). Thus, Pichipilus and Phonocdromus do not show derived features in common with the ‘Caenolestinae’ sensu Marshall et al. (1990), but with the Palaeothentidae and Abderitidae , with which they form a monophyletic group. Both genera, here included within the family Pichipilidae (new rank) do not show affinities with Pliolestes , a caenolestid related to the Rhyncholestes + Caenolestes clade. The Pichipilidae are the sister group of palaeothentids + abderitids. In turn, pichipilids, palaeothentids, and abderitids, here included within the Superfamily Palaeothentoidea nov., are the sister group of the Caenolestoidea . In our concept, this last taxon includes a single family: Caenolestidae , containing the following genera: Stilotherium , Pseudhalmarhiphus , Caenolestes , Lestoros , Rhyncholestes , and Pliolestes .

Figures

Figure 7. Strict consensus of five equally most parsimonious trees of ‘pseudodiprotodont’ marsupials. Numbers above each branch indicate character number, whereas those below are the character-states; absolute and relative Bremer support values are below each branch.