Pucadelphys andinus, Marshall & de Muizon, 1988

Beck, Robin M. D., Voss, Robert S. & Jansa, Sharon A., 2022, Craniodental Morphology And Phylogeny Of Marsupials, Bulletin of the American Museum of Natural History 2022 (457), pp. 1-353 : 309-312

† Pucadelphys

SPECIES SCORED: † Pucadelphys andinus (type and only described species).

GEOLOGICAL PROVENANCE OF SCORED SPECIMENS: Tiupampa, Santa Lucía Formation, Bolivia.

AGE OF SCORED SPECIMENS: The fossil-bearing beds at Tiupampa are within a reversed magnetostratigraphic polarity unit (Marshall et al., 1997; Sempere et al., 1997). Based on a range of geological evidence, Marshall et al. (1997) and Sempere et al. (1997) identified this reversal as corresponding to chron 26r, which spans from the late Danian (~62.15 Mya; Jehle et al., 2019) to the Selandian-Thanetian boundary (61.6 Mya; Coccioni et al., 2012: fig. 2; Cohen et al., 2013 [updated]). This interpretation was followed by Zimicz et al. (2020).

However, the study by Gelfo et al. (2009) and several subsequent studies (Woodburne et al., 2014a, 2014b; Muizon et al., 2015, 2018; Muizon and Ladevèze, 2020) have instead associated this reversal with the earlier chron 28r, which spans from 65.118 to 64.866 Mya (Sprain et al., 2015). This was driven in large part by the results of various cluster analyses presented by Gelfo et al. (2009), in which the Tiupampan mammal fauna grouped with North American early Paleocene (Puercan) faunas, to the exclusion of other South American Paleocene and Eocene mammal faunas. We critically review the evidence presented by Gelfo et al. (2009) in detail here (see also Zimicz et al., 2020).

The grouping of the Tiupampa and North American Puercan mammal faunas found by Gelfo et al. (2009) was largely due to the shared presence of the metatherian † Peradectes. However, this is problematic for a number of reasons. The Tiupampan † Peradectes is currently known from a single damaged upper molar (YPFB Pal 6132; Marshall and Muizon, 1988: fig. 12) and has not, to our knowledge, been included in a formal phylogenetic analysis to test whether it is indeed a member of this genus. The only other South American record of † Peradectes is † P. austrinum — which is the species to which Marshall and Muizon (1988) originally referred the Tiupampan specimen, although Muizon (1992) subsequently referred it to † P. cf. austrinum (see also Jaillard et al., 1993: 656)—from Laguna Umayo in Peru (Sigé, 1971; 1972; Crochet, 1980). Sigé et al. (2004) concluded that Laguna Umayo is either Chron 26r or Chron 24r (late Thanetian to early Ypresian; Coccioni et al., 2012), in which case † Peradectes survived in South America until at least the late Paleocene; thus, the presence of † Peradectes at Tiupampa does not clearly support an early Paleocene age. The Puercan (earliest Paleocene) faunas were the only North American mammal faunas considered by Gelfo et al. (2009), but the North American record of † Peradectes extends until the end of the Eocene (Korth, 2008; 2018), and so again the shared presence of † Peradectes at Tiupampa and in the Puercan of North America does not directly support an early Paleocene age for the former. Furthermore, the North American Puercan “† Peradectes ” does not unambiguously group with † Peradectes sensu stricto in the phylogenetic analyses of Williamson et al. (2012; 2014), and some of these Puercan specimens are now recognized as belonging to a different genus, † Thylacodon (Williamson et al., 2012; 2014); this casts further doubt on whether the Tiupampan and Puercan taxa are closely related, particularly in the absence of a formal phylogenetic analysis that includes the Tiupampan † Peradectes specimen.

The only other taxa shared by Tiumpampa and the Puercan faunas in Gelfo et al.’s (2009) dataset are “ Didelphimorphia ,” “ Didelphidae ,” and “Condylarthra”—all of which are arguably inappropriate to use in clustering analyses because they are clearly nonmonophyletic (at least as used by Gelfo et al., 2009)—and the “condylarth” family †Mioclaenidae (see Muizon and Cifelli, 2000). Zack et al. (2005; see also Zack, 2010) considered †Mioclaenidae to be equivalent to †Hyopsodontidae, and placed the Tiupampan and other South American members of this group within the hyopsodontid subfamily Kollpaninae (Zack et al., 2005: table 4 ). Regardless, mioclaenids/hyopsodontids survived in North America after the Puercan (Archibald, 1998; Williamson and Carr, 2007), and so their presence at Tiupampa is not clearly supportive of a Puercan age for the latter. Thus, evidence from these other taxa likewise does not clearly support an early Paleocene age for Tiupampa (Zimicz et al., 2020).

Muizon and Ladevèze (2020: 600–609) restated the evidence in favor of an early Paleocene (chron 28r) age for Tiupampa, and argued that the “primitiveness” of various Tiupampan mammals compared to relatives at other fossil sites in North and South America provides further support for this age interpretation. However, Muizon and Ladevèze (2020) did not provide any phylogenetic or other quantitative analyses in support of their arguments, and such “stage of evolution” evidence is not (in our opinion) compelling, given that the comparisons are made between fossil sites at very different paleolatitudes and on different continents (see also Zimicz et al., 2020).

In summary, while an early Paleocene age for Tiupampa is plausible, we do not find the evidence for correlating this fauna with chron 28r, rather than chron 26r, to be overwhelming (contra Gelfo et al., 2009; Muizon et al., 2015, 2018; Muizon and Ladevèze, 2020). Instead, we take a compromise position, and assign † Pucadelphys and the other Tiupampan terminals included here († Mayulestes and † Allqokirus ; see below) an age range corresponding to the maximum age of chron 28r (65.118 Mya; Sprain et al., 2015) and the minimum age of chron 26r, which corresponds to the Seldandian-Thanetian boundary (59.2 Mya; Coccioni et al., 2012; Cohen et al., 2013 [updated]).

ASSIGNED AGE RANGE: 65.118 –59.200 Mya.

REMARKS: † Pucadelphys andinus is represented by abundant, excellently preserved cranial and postcranial material from the Tiupampa Fauna in Bolivia (Marshall and Muizon, 1995; Ladevèze and Muizon, 2007; Ladevèze et al., 2011). As discussed above, the age of the Tiupampa Fauna, which is within the Santa Lucía Formation, remains somewhat controversial: originally suggested to be Late Cretaceous (Marshall et al., 1983; Muizon et al., 1983, 1984; Marshall and Muizon, 1988), it now appears to be either middle Paleocene (Marshall et al., 1997; Sempere et al., 1997; Zimicz et al., 2020) or early Paleocene (Gelfo et al., 2009; Woodburne et al., 2014a; Muizon et al., 2015, 2018; Muizon and Ladevèze, 2020). Twenty-two individuals of † P. andinus (12 adult females, six adult males, and four subadults) are known from mostly intact crania and mandibles, and numerous other individuals (including a single juvenile) are known from more fragmentary remains (Marshall and Muizon, 1995; Ladevèze and Muizon, 2007; Ladevèze et al., 2011). These specimens indicate the presence of marked sexual dimorphism and also a surprising degree of variation in molar morphology (Ladevèze et al., 2011).

Although † Pucadelphys was originally classified as a “didelphid” (Marshall and Muizon, 1988; 1995), it differs from crown-clade didelphimorphians ( Didelphidae sensu Voss and Jansa, 2009 ) in numerous morphological features, including absence of an ossified hypotympanic sinus floor, presence of a medial process of the squamosal, absence of a distinct rostral tympanic process of the petrosal, presence of posterior cingulids on the lower molars, absence of a proximal calcaneocuboid facet on the ventral surface of the calcaneus, and presence of a third trochanter of the femur, among others (Marshall and Muizon, 1988; 1995; Marshall and Sigogneau-Russell, 1995; Muizon et al., 1997; Argot, 2002; Muizon and Argot, 2003; Ladevèze and Muizon, 2007).

In fact, † Pucadelphys has consistently been recovered within Marsupialiformes sensu Beck (2017a; see also Vullo et al., 2009) but outside Marsupialia in published phylogenetic analyses based on morphology (Rougier et al., 1998, 2004, 2015; Luo et al., 2003, 2011; Asher et al., 2004; Ladevèze and Muizon, 2007; SánchezVillagra et al., 2007; Beck et al., 2008; Horovitz et al., 2008, 2009; Forasiepi, 2009; Ladevèze and Muizon, 2010; Williamson et al., 2012, 2014; Engelman and Croft, 2014; Forasiepi et al., 2014a; Suarez et al., 2015; Ni et al., 2016; Wilson et al., 2016; Beck, 2017b; Carneiro and Oliveira, 2017a, 2017b; Maga and Beck, 2017; Bi et al., 2018; Carneiro et al., 2018; Muizon et al., 2018; Carneiro, 2019; Rangel et al., 2019; Engelman et al., 2020; Ladevèze et al., 2020; Muizon and Ladevèze, 2020; Zimicz and Goin, 2020), total evidence (Asher et al., 2004; Beck et al., 2014; Beck, 2017b; Maga and Beck, 2017; Abello and Candela, 2019), and molecular-scaffold (Beck, 2012) datasets.

Two notable exceptions to this general consensus are the studies of Goin et al. (2006) and Velazco et al. (2022), both of which reported phylogenetic analyses that placed † Pucadelphys within Marsupialia . Goin et al.’s (2006) analysis is problematic in several respects, inter alia: (1) their study was based almost exclusively on dental characters; (2) plesiomorphic absence versus secondary loss of certain features was determined a priori (e.g., chars. 13, 16, 18); (3) fossil taxa were scored for unpreserved characters (e.g., sperm-pairing was scored as absent in † Kokopellia but present in † Itaboraidelphys), and (4) the monophyly of several clades was enforced a priori using topological constraints (Goin et al., 2006: supplementary data). Parsimony analysis of a corrected version of the Goin et al. (2006) resulted in a largely unresolved strict consensus with no support for † Pucadelphys within Marsupialia (R.M.D.B., personal obs.).

Velazco et al. (2022) used an expanded version of the “phenomic” morphological matrix of O’Leary et al. (2013), comprising 4541 osteological and soft tissue characters, which they analyzed in combination with 37 kb of DNA sequence data from 27 nuclear genes, using both maximum parsimony and maximum likelihood. These analyses placed † Pucadelphys sister to † Deltatheridium from the Late Cretaceous of Mongolia, with this clade sister to the Recent didelphid Didelphis , and the Recent microbiotheriid Dromiciops in turn sister to † Pucadelphys + † Deltatheridium + Didelphis . This topology implies that both † Pucadelphys and † Deltatheridium are marsupials, rather than (as was previously generally accepted) nonmarsupial metatherians. This surprising result warrants further scrutiny; in particular, Velazco et al. (2022) included only two Recent marsupials ( Didelphis virginiana and Dromiciops gliroides ) in their study, and it remains to be seen what impact the addition of more Recent marsupials would have. We also note that at least some of the synapomorphies identified by Velazco et al. (2022: 7) as supporting a † Pucadelphys + † Deltatheridium + Didelphis clade to the exclusion of Dromiciops appear to be questionable. The fossil microbiotheriid † Microbiotherium has a paracanine fossa (see char. 34), suggesting that its absence in Dromiciops may be secondary rather than plesiomorphic, and evidence from other fossil microbiotheriids (in particular, † Woodburnodon ; Goin et al., 2007c) indicates that the absence of well-developed stylar cusps in Dromiciops is likewise secondary. In addition, contact between the fibula and calcaneus is present in † Pucadelphys (Muizon, 1998; Argot, 2002; Szalay and Sargis, 2006) and † Deltatheridium (Horovitz, 2000: fig. 9), whereas it is absent in Dromiciops (Szalay, 1994; Szalay and Sargis, 2006) contra Velazco et al. (2022).

We consider that the majority of current evidence indicates that † Pucadelphys is a nonmarsupial metatherian contra Goin et al. (2006) and Velazco et al. (2022), and we follow the classification of Muizon et al. (2018), in which it is a member of the family †Pucadelphyidae, within the superorder †Pucadelphyda (which includes the order † Sparassodonta ; see † Mayulestes and † Allqokirus below). As such, we consider † Pucadelphys suitable for use as an outgroup taxon for our morphological and total-evidence analyses.