Didolodontidae (Figs 8B–G, 9B–E)

Didolodontidae currently includes 10 genera [or 12 if we include two unnamed genera (Tejedor et al. 2009)] with an exclusive Palaeogene distribution (Fig. 2B; Supporting information, Table S1). There has been recent major detailed revision of this family (Gelfo 2006, 2010), so to avoid unnecessary repetition, only the most relevant taxonomic aspects are discussed here, along with the advances since these revisions. The family Didolodontidae was proposed by Scott (1913) to classify a group of bunodont ‘condylarths’ (see ‘Condylarthra’ section in Supporting information, File S1) endemic to South America with a Palaeogene fossil record consisting mostly of isolated teeth. Initially, Ameghino (1897) described the first known Eocene didolodontid species Didolodus multicuspis as a phenacodontid. Osborn (1910) identified additional South American taxa as ‘Condylarthra’ incertae sedis such as Notoprotogonia (Ameghino 1904b [later synonymized with Ernestokokenia Ameghino, 1901 by Simpson (1948)], Lambdaconus [later considered an early proterotheriid (Soria 2001)], and Proectocion [later considered an early adianthid (Cifelli 1983a)]. Scott (1913) later grouped these four taxa as part of the family Didolodidae within the order Litopterna (Table 1). Simpson (1934) subsequently corrected it to Didolodontidae, and grouped them again within

‘Condylarthra’. After adding two new species from the Rio Chico Formation [latest Danian–Late Lutetian in age (Krause et al. 2017)] of the genus Ernestokokenia (Ernestokokenia yirunhor Simpson, 1935 and Ernestokokenia chaishoer Simpson 1935), Simpson (1948) revised the whole group, giving a new definition for Didolodontidae and emending many of Ameghino’s (1906) taxonomic classifications of the Patagonian taxa, including the following genera in the family: Didolodus, Ernestokokenia, Argyrolambda Ameghino 1904b, Paulogervaisia Ameghino, 1901, Proectocion, Enneoconus Ameghino, 1901, Asmithwoodwarsia Ameghino, 1901, and Archaeohyracotherium Ameghino, 1906. Simpson soon recognized the close similarities between this group and North American ‘condylarths’, including species of Phenacodontidae and Mioclaenidae, stating that they were impossible to separate on a purely morphological basis (Simpson 1948, 1980).

Paula Couto (1952) added two species of Ernestokokenia (Ernestokokenia protocenica Paula Couto, 1952 and Ernestokokenia parayiruhnor Paula Couto, 1952) and the new genus Lamegoia from fissure fills in the Itaboraí Formation, Brazil, to the family Didolodontidae. McKenna (1956) added Megadolodus molariformis from La Venta, Colombia, which Cifelli and Villarroel (1997) later considered to be a bunodont proterotheriid, a hypothesis that recently has found some support in phylogenetic analyses (Carrillo et al. 2023). Paula Couto (1978) added Asmithwoodwardia scotti Paula Couto, 1952 from Itaboraí to the family, a taxon that he previously classified within the North American ‘Condylarthra’ family Hyopsodontidae based on dental similarities (Paula Couto 1952). In a revision of the family Didolodontidae, Cifelli (1983a), based on an indirect tarsal association, removed Ernestokokenia parayirunhor (renamed as Miguelsoria parayirunhor Cifelli 1983a) and Asmithwoodwardia scotti from Didolodontidae and placed them in Litopterna (see Protolipternidae section for more information). This tarsal association was based on the fact that the relative size and abundance of these tarsals in the Itaboraí sites was what you would expect for the size and abundance of the dentition of Miguelsoria parayirunhor (Cifelli 1983b). Cifelli (1983a) reclassified Ernestokokenia protocenica as Paulacoutoia protocenica [now known as Ricardocifellia protocenica (Mones, 2015) because of preoccupation], because he considered it sufficiently morphologically distinct from the genus Ernestokokenia. According to Cifelli (1983a), didolodontids are characterized by having a primitive dentition very similar to mioclaenids, but possessing a hypocone on the M3 (absent in mioclaenids) and a primitive ankle structure similar to many South and North American ‘condylarths’ (i.e., the astragalar facet for the medial malleolus is enlarged anteriorly and the calcaneal cuboid facet is dorsally expanded). Soria (2001), based on dental features, removed Xesmodon Berg, 1899 from Anisolambdidae and placed it within Didolodontidae.

In a detailed revision of Didolodontidae, Gelfo (2006) synonymized Argyrolambda with Didolodus, Archaeohyracotherium with Asmithwoodwardia, and Enneoconus with Ernestokokenia, thus reducing the total number of valid didolodontid genera. Later, Gelfo (2007), based on dental similarities and a phylogenetic analysis, added Escribania chubutensis Bonaparte et al., 1993 and Raulvaccia peligrensis Bonaparte et al. 1993 from Punta Peligro, Argentina, to Didolodontidae. Escribania chubutensis and Raulvaccia peligrensis were previously classified within the ‘condylarth’ North American family Mioclaenidae (Bonaparte et al. 1993). The most recent generic additions to Didolodontidae have been Umayodus Gelfo & Sigé, 2011 from the Lower Muñani Formation, Peru, and Saltaodus Gelfo et al., 2020a from the Lumbrera Formation, both of uncertain age (Gelfo and Sigé 2011, Zimicz et al. 2023).

In terms of the phylogenetic affinities of Didolodontidae, early phylogenetic analyses found a close relationship between didolodontids and sparnotheriodontids (Cifelli 1993, Bergqvist 1996; Fig. 1C), but these were based on tarsal associations that have been questioned by later authors (e.g.,Soria 2001, Gelfo and Sigé 2011; see also Sparnotheriodontidae and Protolipternidae sections). Muizon and Cifelli (2000), in a phylogenetic analysis that included North American mioclaenids, South American kollpaniines, didolodontids, and protolipternids, found that protolipternids were nested within Didolodontidae and that both groups together formed the sister group of kollpanines, a result that was also found in a later study (Gelfo 2006). This means that Didolodontidae excluding Protolipternidae is a paraphyletic group. Other phylogenetic studies that have included didolodontids in very small matrices have focused on their internal relationships, assuming their monophyly (e.g., Gelfo 2010; Table 2).

When didolodontids have been included with a limited taxon sample (i.e., usually Didolodus) in broader analyses that include other SANUs, they have been found in a branch at the stem of some North American ‘condylarths’ and other SANUs (e.g., Billet et al. 2015), or in a basal position in Laurasiatheria as the sister group of Protolipternidae (O’Leary et al. 2013). When a larger sample of didolodontids has been included, they have been found at the stem of litopterns but not forming a monophyletic Didolodontidae (Kramarz et al. 2021), or as a paraphyletic group among SANUs and North American ‘condylarths’ (Zimicz et al. 2023). In sum, there is still a considerable uncertainty in the phylogenetic position of didolodontids and their relationship among SANUs, and whether they form a monophyletic group. In fact, there has not been any phylogenetic analysis that includes didolodontids alongside representatives of all litoptern families, protolipternids, and the different SANUs’

MATERIALS AND METHODS