A phylogenetic analysis was conducted to assess the phylogenetic relationships of specimen MLG 2630 with other pipoids. The present phylogeny is based on the extensive analysis published by Gómez (2016) with the addition of characters 166–176, based on personal observations and newly published contributions (S1). The coding of character 39 was modified and state 1 was split into two different states (S1). We added a state 2 to character 99 (S1). The coding of several character states of Saltenia ibanezi was modified from that of previous authors based on newly collected specimens (MAS-P/2; Scanferla et al., 2011) (see S2). Characters 18, 19, 32, 33, 36, 38, 39, 50, 59, 62, 74, 75, 82, 93, 94, 98, 101, 109, 115, 124, 125, 126, 135, 137, 158 and 168 were treated as ordered.

These modifications resulted in a data matrix composed of 176 characters scored for 38 taxa (S3).

The phylogenetic analysis was performed using TNT 1.1 (Goloboff et al., 2008). All characters were equally weighted and treated as unordered. Heuristic searches were performed after 1,000 pseudoreplicates of WAG + TBR search strategy, with 10 random addition sequences after each search, and 100 trees were saved at each replicate. Bremer Support and absolute bootstrap frequencies were calculated with the aim to test the robustness of tree branches (Fig. 4).

Anura Rafinesque, 1815

Pipoidea Fitzinger, 1843

Panpipidae nomen novum.

Diagnosis. The anuran clade is diagnosed based on the following synapomorphies: 1) anterior ramus of pterygoid dorsally located with respect to the maxilla (23-1); 2) straight distal margin of sacral diapophyses (99-1); 3) cleithrum covering posterior edge of suprascapular cartilage (114-1); 4) interiliac scar ample both ventrally and dorsally (149-2); and 5) presence of a poorly-developed lateral flange on pterygoid (167-1).

Definition. Stem-based clade consisting of Patagopipa and all species that share a more common ancestor with Shelania laurenti, Pipa, and Xenopus laevis than with Vulcanobatrachus mandelai, Avitabatrachus uliana, Cordicephalus gracilis, or Paleobatrachus grandipes.

Remarks. The clade Panpipidae is here coined with the aim to include the crown-group Pipidae, Shelaniinae, and another stem pipids. We follow previous authors (Ford and Cannatella, 1993 and Gómez, 2016) in the restriction of Pipidae to crown group taxa; this is Pipinae + Xenopodinae.

Shelaniinae nomen novum.

Diagnosis. This clade is diagnosed based on the following synapomorphies: 1) anterior ramus of pterygoid reaching the antorbital plane (24-1); 2) eight presacral vertebrae (82-1); 3) presacral vertebrae I–II not fused but imbricated medially (88-1); 4) marked forward orientation of the transverse process of presacra vertebrae IV (94-2); 5) cross-section of distal iliac shaft flattened, dorsoventrally compressed (137-2); 6) second pair of ribs anterolaterally oriented (174-1).

Definition. The stem-based clade consisting of Patagopipa and all species that share a more common ancestor with Shelania laurenti than with Pipa, Silurana, Xenopus, or Eoxenopoides.

Included taxa. Shelania pascuali Casamiquela, 1960; Saltenia ibanezi Reig, 1959; Kuruleufenia xenopoides Gómez, 2016; Patagopipa corsolinii nov.

Temporal distribution. From Late Cretaceous to Eocene (probably Pleistocene; Báez et al., 2007) times.

Patagopipa nov. gen.

Diagnosis. Small pipoid frog diagnosable based on the following combination of characters (autapomorphies marked by an asterisk): 1) elongate and transversely narrow pterygoids, with poorly expanded proximal flanges*; 2) frontoparietal dorsally flat with rounded anterior and posterior margins; 3) frontoparietal lacking interorbital constriction; and 4) notably robust and stout radioulna and humerus*.

Etymology. Patago, from Patagonia the region of Argentina in which the holotype specimen was found; Pipa, the type genus of the Pipidae family.

Type species. Patagopipa corsolinii nov. sp.

Patagopipa corsolinii nov. sp. (Fig. 2 and Fig. 3)

Diagnosis. The same as for genus by monotypy.

Holotype. MLG 2630, a single slab containing a nearly complete skeleton having partially disarticulated skull, incomplete vertebral column, and nearly complete right fore- and hindlimbs. Left forelimb is poorly preserved, and left hindlimb is only represented by the proximal end of the femur.

Etymology. The specific epithet honors Rodolfo Corsolini, director of the MLG Museum.

This article conforms to the requirements of the amended International Code of Zoological Nomenclature–except for the new names, which are defined by the phylogenetic nomenclature. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The LSID for this publication is: urn:lsid:zoobank.org:pub:2CF53EA3-50C6-48C0-9665-9A29D8F01931.

The Patagopipa corsolinii holotype shows a combination of characters that indicates that it was at a juvenile stage. It has a completely ossified humerus, radio-ulnae, femora, tibio-fibulae, ilia, and pedal phalanges. Trueb and Hanken (1992) indicate that in Xenopus all these bones ossify during stages 57–58 of Nieuwkoop and Faber (1956). Metacarpals, and coracoid epiphyses are not completely ossified, traits that in Xenopus (Trueb and Hanken, 1992) occur between the stages 58 and 60. Regarding the ribs, in Xenopus these ossify at stage 58 (Trueb and Hanken, 1992), but the transverse processes of all presacral vertebrae ossify at stage 64. Patagopipa shows well-developed ossified ribs, but these are not connected to vertebrae. In Patagopipa, the sacral diapophyses seem not completely ossified because they are smaller and poorly laterally expanded and do not form the straight (or nearly straight) lateral margin typical of pipids; interestingly, in Xenopus the diapophyses start to ossify at stage 63. Interestingly, in paleobatrachid tadpoles (Rócek, 2003), the sacral diapophyses in stage 63 are only weakly ossified, but in stage 66 the osseous part is much notorious, and even, shows a similar contour with those of P. corsolinii. The urostyle is formed by the fusion of the postsacral vertebrae with the hypocord (Trueb and Hanken, 1992), which occurs in Stage 66; this condition is already present in Patagopipa. Finally, P. corsilinii lacks distal tarsals, carpals and phalanges of the hand, which in Xenopus (Trueb and Hanken, 1992) ossify during stage 60 or after. For these reasons, the stage of the present specimen probably corresponds to stage 66.

The phylogenetic analysis recovered 14 most parsimonious trees (MPTs) that resulted in a strict consensus tree of 592 steps in length, having a consistency index of 0.40, and a retention index of 0.71. Most nodes have low Bremer Support values, in agreement with previous analyses made based on this data matrix (e.g., Gómez, 2016; Fig. 4).

The strict consensus tree differs from previous analyses in the position of some taxa (e.g., Báez and Púgener, 2003, Báez and Trueb, 1997, Gómez, 2016 and Trueb and Báez, 2006). As indicated above, probably the most important innovation is that a new clade (Shelaniinae) is coined with the aim to include South American aglossans previously regarded as basal pipids (Saltenia) or basal xenopodinomorphans (Kuruleufenia, Shelania). In this sense, after the phylogenetic analysis the interrelationships of these clades are modified, including the relationships between Pipinae and Xenopodinae. In this sense, Shelaniinae turn out to be the sister group of crown Pipidae (= Pipidae sensu stricto; Pipinae + Xenopodinae; Báez and Púgener, 2003, Báez and Trueb, 1997 and Gómez, 2016), and both clades constitute Panpipidae (see diagnoses above). As in previous works, Vulcanobatrachus turns out to be a basal pipoid (Gómez, 2016 and Trueb et al., 2005), being nested here as a stem-Panpipidae. Llankibatrachus, ‘Xenopus’ romeri and ‘Shelania’ laurenti were considered xenopodinomorphans by previous authors (Báez and Púgener, 2003 and Báez and Trueb, 1997). In this analysis, these taxa are more derived than Shelaniinae, being the successive sister groups of the clade Pipinae + Xenopodinae. Llankibatrachus resemble pipids, and differ from shelaniines, by the presence of three unambiguous synapomorphies, namely strongly expanded sacral diapophyses (98-2), clavicle fused to scapula (108-2), and anterior half of iliac shafts subparallel to each other (175-0). This combination of characters is absent in most shelaniines, including Patagopipa. Further, ‘Xenopus’ romeri and ‘Shelania’ laurenti share with crown pipids a rounded posterior margin of the frontoparietal bone (166-0), and posterior margin of otic capsules at the same level as or anterior to occipital condyles (168-0). Both features are different from the condition observed in Late Cretaceous and Paleogene pipoids from South America, including Llankibatrachus, Saltenia and Shelania, for example. Finally, the extinct genus Oumtkoutia from the early Late Cretaceous of Morocco (Rage and Dutheil, 2008) is excluded from its original position among Xenopodinae and is nested as stem-Pipinae. This result rests on weak evidence; Oumtkoutia shares with Pipinae a single unambiguous derived feature: poorly-developed basal processes on the otic capsules. Because of this, and because of the low statistical support of the phylogenetic position of Oumtkoutia, we regard this taxon as of uncertain phylogenetic position. Furthermore, because of the presence of several plesiomorphic traits of the braincase (see Rage and Dutheil, 2008) it is also possible that this genus may be better nested among basal pipoids. The phylogenetic position of this taxon should be resolved when more complete specimens become available. In sum, the new results here performed indicate that interrelationships among Pipimorpha are far from robust and well-established.

Patagopipa belongs to Panpipidae based on the presence of several derived features, namely the presence of a large azygous frontoparietal bone, large otic capsules, fused sacrum and urostyle, and strongly expanded sacral diapophyses (Báez, 1981, Báez and Trueb, 1997, Estes, 1975, Estes and Reig, 1973 and Trueb and Báez, 2006).

Patagopipa shares these characters with shelaniines: anterior ramus of pterygoid reaching the antorbital plane, marked forward orientation of the transverse process of presacral vertebrae IV, second pair of ribs anterolaterally oriented, and distal iliac shaft dorsoventrally flattened in cross section. In addition to these synapomorphies, Patagopipa shares with shelaniines such as Shelania and Saltenia common anatomical traits, including: short anterior nasal process, presence of a well-developed and acute antorbital process of the maxilla (39-2), and posteriorly expanded otic capsules that surpass the posterior level of occipital condyles (168-1) (see Báez, 1981, Báez and Púgener, 1998, Báez and Trueb, 1997, Casamiquela, 1960 and Casamiquela, 1961).

Among shelaniines, the present phylogenetic analysis indicates that Patagopipa constitutes the sister group of the genus Saltenia. Both genera are united by the derived presence of rounded anterior margin of frontoparietal bone (169-1), and frontoparietal lacking interorbital constriction (170-1). Furthermore, Saltenia and Patagopipa lack derived features present at the clade conformed by Kuruleufenia + Shelania, including neural arch lamina of sacrum nearly as long as wide (101-1), and vertebrae that are subrectangular in contour when viewed dorsally (172-1).

Llankibatrachus truebae was found in beds near those that yielded Patagopipa corsolinii (Báez and Púgener, 2003), and thus, detailed comparisons between both taxa are necessary. In the present phylogenetic analysis, Llankibatrachus nests outside Shelaniinae (lacking all the synapomorphies of the clade), as the sister group of the Pipidae clade. In this sense, Llankibatrachus lacks features observed in Patagopipa and other Shelaniinae, namely: the frontoparietal, which is three times longer than wide, lacking interorbital constrictions (170-1), and the transversely expanded presacral vertebral (172-1). Furthermore, Llankibatrachus shares derived features with crown pipids that are absent in Patagopipa (it is possible that some of these traits could be the result of the ontogenetic stage of the specimen), as for example, fused scapula and coracoid and notably expanded sacral diapophyses (see section 2). In addition, Llankibatrachus differs from Patagopipa by the presence of well-developed parasagittal crests on the frontoparietal bone, by the presence of a supraorbital flange, the frontoparietal with acute anterior margin, by posteriorly directed presacral ribs II-IV, and by the straight distal margin of sacral diapophyses, among other minor anatomical details.

From the Eocene–Oligocene of Brazil, Bedani and Haddad (2002) described the genus and species Paleopipa aiuruoca. In spite of this, the taxon was described only in a preliminary way, and these illustrations are not enough informative; the shape of the frontoparietal bone, the inflate and posteriorly extended otic capsules, the ilium with the distal end of shaft dorsoventrally flat, and second ribs strongly anterolaterally oriented indicate that this taxon may be included within Shelaniinae. Paleopipa differs from Patagopipa by having frontoparietals with mid-length transverse constriction, strongly expanded sacral diapophyses, and gracile and elongate humerus (Bedani and Haddad, 2002).

In sum, Patagopipa is a pipid frog well-nested among Shelaniinae, being the sister group of the Late Cretaceous Saltenia. In the phylogenetic analysis, both taxa, share: long metacarpals compared with radio-ulna length (135-1); otic capsules posteriorly expanded surpassing the posterior margin of the occipital condyles (168-1); frontoparietal rounded in shape (169-1); and vertebrae two or three times wider than long (rectangular-shaped when viewed dorsally) (172-1). Furthermore, Saltenia and Patagopipa share some features that deserve a more detailed study: absence of parasagittal crests and sacral diapophyses poorly developed and slightly more posteriorly than anteriorly expanded. Its unique combination of characters indicates that Patagopipa corsolini is a valid genus and species of pipimorph toad.

The present phylogeny agrees in some respects with previous phylogenetic hypotheses (Báez and Púgener, 1998, Báez and Trueb, 1997, Gómez, 2016, Trueb, 1999 and Trueb and Báez, 2006). All these analyses evidence a complex biogeographic scenario of pipimorph evolution. In fact, stem pipoids include both South American and African taxa. Panpipids include the South American Shelaniinae and Llankibatrachus, whereas crown Pipidae encompass both African and South American lineages. This complex pattern makes the biogeographic history of pipoids very difficult to interpret.

Most authors agreed in that South America and Africa were completely separated by the opening of the Atlantic Ocean by Turonian times (80 Mya). This resulted in the “splendid isolation” of South American floras and faunas, which remained isolated from other landmasses until Late Neogene times (see Pascual and Ortiz Jaureguizar, 2007). In spite of this, some authors proposed that faunistic interchanges between Africa and South America occurred through island chains or subaerial terrestrial bridges (see Ezcurra and Agnolín, 2012 and Oliveira et al., 2010).

Trueb et al. (2005) and Rage et al. (2013) proposed that pipoids were widespread along Western Gondwana by Cretaceous times, before the severing of both landmasses. Later, by Turonian times, both landmasses separate each other and pipoids evolved separately on both continents. The present analysis argues against such proposal. The interleaved occurrence of African and South American taxa may imply that, during Turonian times, a large number of pipoids (including living pipids) were present and diversified during Early Cretaceous times before the severing of both continents, which is unlikely and inconsistent with molecular and paleontological data. Thus, based on the present analysis and previous phylogenetic hypotheses, it is more likely that the occurrence of several pipoid clades on both sides of the Atlantic Ocean may be the result of a transoceanic dispersal. In this context, two main hypotheses have been proposed by previous authors.

Estes (1975) explored the hypothesis that panpipids may have crossed from Africa to South America in post-Aptian times by means of floating islands of vegetation, a hypothesis proposed by Hoffstetter (1974) for explaining monkey and caviomorph arrival to South America during Eocene times.

However, the present phylogeny contradicts Estes’ (1975) hypothesis. In fact, the presence of several interleaved South American and African panpipid clades contrasts with a single westward-directed floating island model. On the contrary, the present phylogenetic hypothesis sustains several bidirectional dispersals between both landmasses.

In this sense, the presence of several panpipid groups in both Africa and South America was interpreted by Cannatella (2015) as being the result of several eastward and westward dispersals of panpipids during the Paleogene, suggesting that vicariance does not seem a feasible explanation for panpipid distribution and phylogeny. In this way, Cannatella (2015) suggested that panpipids may have dispersed from South America to Africa and vice-versa during the Earliest Tertiary through an island chain connecting both continents.

This is in agreement with Bonaparte (1984), who proposed that important faunistic similarities between Tertiary faunas of South America and Europe might be explained by dispersion from Africa to South America and vice-versa using the Walvis Ridge–Rio Grande Rise terrestrial passage or island chain. This agrees with minimal divergence times between several South American and African extant plant and animal groups that are calibrated well after the purported latest separation between both landmasses, suggesting dispersal events between Africa and South America after their separation (see details in Ezcurra and Agnolín, 2012); see also Oliveira et al., 2010) built upon this evidence a new paleobiogeographical model to explain these unexpected patterns in several clades, mainly reptiles, indicating that Africa, South America, and Europe retained deep biogeographical ties up to Paleogene times. In this sense, recent authors sustained a late faunistic connection between Africa and South America based on herpetofaunal evidence (e.g., Agnolín, 2012, Hofmeyr et al., 2016, Pereira and Schrago, 2017 and Vidal et al., 2008). The multiple panpipid interchanges (at least three different bidirectional interchanges based on present phylogeny) between South America and African in post-Cretaceous times are in agreement with Ezcurra and Agnolín's (2012) model suggesting a late Africa–South America terrestrial biotic interchange up to the Earliest Tertiary times.