The African cynodont Aleodon (Cynodontia, Probainognathia) in the Triassic of southern Brazil and its biostratigraphic significance

In this contribution we report the first occurrence of the enigmatic African probainognathian genus Aleodon in the Middle-early Late Triassic of several localities from the state of Rio Grande do Sul in southern Brazil. Aleodon is unusual among early probainognathians in having transversely-expanded postcanine teeth, similar to those of gomphodont cynognathians. This genus was previously known from the Manda Beds of Tanzania and the upper Omingonde Formation of Namibia. The Brazilian record of this genus is based upon multiple specimens representing different ontogenetic stages, including three that were previously referred to the sectorial-toothed probainognathian Chiniquodon theotonicus. We propose a new species of Aleodon (A. cromptoni sp. nov.) based on the specimens from Brazil. Additionally, we tentatively refer one specimen from the upper Omingonde Formation of Namibia to this new taxon, strengthening biostratigraphic correlations between these strata. Inclusion of A. cromptoni in a phylogenetic analysis of eucynodonts recovers it as the sister-taxon of A. brachyrhamphus within the family Chiniquodontidae. The discovery of numerous specimens of Aleodon among the supposedly monospecific Chiniquodon samples of Brazil raises concerns about chiniquodontid alpha taxonomy, particularly given the extremely broad geographic distribution of Chiniquodon. The discovery of Brazilian Aleodon and new records of the traversodontid Luangwa supports the hypothesis that at least two subzones can be recognized in the Dinodontosaurus Assemblage Zone.


Introduction
The South American fossil record of non-mammalian eucynodonts is one of the most diverse known worldwide [1][2][3] and includes representatives of most of the major Triassic subclades (e.g., cynognathids, gomphognathids, traversodontids, chiniquodontids, ecteniniids, based on direct examination of specimens housed at the UMZC (Cambridge) and NHMUK (London). Measurements of the Aleodon specimens from Brazil are detailed in Table 2.
The database of localities and tetrapod species from the Dinodontosaurus AZ (Fig 2) was based on the literature (see S1 File) and personal observation of various specimens (other than probainognathians), particularly those deposited in institutions in Rio Grande do Sul (e.g., MCN-PV, MMACR-PV-T, MPDC, UFRGS-PV-T, UFSM; see Institutional Abbreviations). For the study of similarities among localities we used the Jaccard Similarity Coefficient to construct the dendrograms, using the software Past v3. 13 [37]. The GPS coordinates of the mentioned Dinodontosaurus AZ localities are available by contacting the corresponding author.

Phylogenetics
The new species named herein (Aleodon cromptoni) was included in the phylogenetic data matrix of Ruta et al. [33], which includes 54 cynodont operational taxonomic units (including A. cromptoni) and 150 discrete cranial, dental, and postcranial characters. Modifications of character-states of some taxa (i.e., Chiniquodon, Aleodon brachyrhamphus, Probainognathus) and the complete data matrix are available in the S1 and S2 Files.
The modified data matrix (S2 File) was analyzed under equally-weighted parsimony using TNT 1.5 [38]. A heuristic search of 500 replicates of Wagner trees, followed by TBR branchswapping algorithm (holding 10 trees per replication), was performed. All characters were treated as non-additive. Bremer support [39] and bootstrap resampling analysis [40] were conducted.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: urn:lsid:zoobank.org:pub:BFF11AF1-742F-4446-9871-3908F70F2E93. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central and LOCKSS.  Table 2. Measurements (in mm) of Aleodon cromptoni sp. nov. specimens from the Middle-early Late Triassic of southern Brazil. Measurement parameters were taken from Abdala and Giannini [14]. *Estimated measure. Canine labio-lingual length (right) 6.1

Systematic Paleontology
Therapsida Broom, 1905 Cynodontia Owen, 1861 Eucynodontia Kemp, 1982 Probainognathia Hopson, 1990 Aleodon Crompton, 1955 Type species. Aleodon brachyrhamphus Crompton, 1955 Diagnosis. Probainognathian cynodonts characterized by the following combination of features (autapomorphies denoted by an asterisk): dental formula of I4/i3, C1/c1, PC9-11/ pc10-12, long osseous secondary palate extending to the posterior end of the tooth row; dorsoventrally deep zygomatic arch formed by jugal and squamosal; strongly labially concave upper postcanine tooth row; upper and lower postcanines ovoid to ellipsoid in outline in occlusal view, bearing a sectorial labial margin with two to three cusps in line and a well-developed lingual platform (more expansive than the lingual cingulum of other basal probainognathians) without a transverse crest Ã ; cusps of upper and lower postcanines less recurved posteriorly and apicobasally shorter than in Chiniquodon species.
Aleodon brachyrhamphus Crompton, 1955 Holotype. UMZC T906, partial skull and lower jaws from the Ruhuhu Valley, Tanzania [20]. Lifua Member, Manda Beds, Ruhuhu Basin [21]. Temporal region length 147.8 Emended diagnosis. Distinguished (autapomorphies denoted by an asterisk) from its congener A. cromptoni sp. nov. by a gradual increase in upper postcanine size between PC1 and PC4; mesiodistally narrower lingual platform of the upper and lower postcanines Ã ; relatively labiolingually elongate upper and lower postcanine crowns, with the occlusal surface of the uppers notably constricted between the labial cusp and the lingual platform Ã ; slightly concave mesial edge and slightly convex distal edge of middle and posterior lower postcanines; enlargement of canine-postcanine diastema during growth and, consequently, fewer upper and lower postcanines at adult size.
Diagnosis. Species of Aleodon distinguished (autapomorphies denoted by an asterisk) from A. brachyrhamphus by a distinct size increase between PC2 and PC3 Ã ; mesiodistally wider lingual platform of upper and lower postcanines Ã ; less labiolingually elongate upper and lower postcanine crowns, without distinct constriction between labial cusp and lingual platform in the uppers; reduced transverse contact between lower postcanine crowns along tooth row Ã ; and very reduced or absent canine-postcanine diastema in all ontogenetic states, resulting in an elevated number of postcanines compared with A. brachyrhamphus (up to 12).
Etymology. Named in honor of Alfred W. "Fuzz" Crompton, original describer of the genus Aleodon, for his many contributions to cynodont research.
Locality and horizon. The holotype MPDC-501-117 was found in a railroad cut that crosses the Cria Farm ("Fazenda Cria"), about 300 meters north from the main house, in the municipality of Vale Verde (Fig 1). UFRGS-PV-0071-T and UFRGS-PV-0122-T were found in Sanga Pinheiro (also written as Sanga Pinheiros), UFRGS-PV-0125-T and MMACR-PV-018-T in Sanga Nicanor (also known as Sanga do Zé), and UFRGS-PV-0146-T in Sanga Pascual. Sanga Pinheiro, Sanga Nicanor, and Sanga Pascual outcrops are located about 10 to 20 kilometers south of the city of Candelária; the first two are in the Pinheiro region and the latter in the Bom Retiro region. MCN-PV 10338 was found in the Cortado site (also called Rincão da Porta), next to the highway RST-287 [41], municipality of Novo Cabrais (Fig 1). UFRGS-PV-0274-T lacks precise locality data. Based on its style of preservation it could possibly come from the classical Pinheiro region. The tentatively-referred specimen MCP-PV-1695T comes from the Porto Mariante 2 locality, municipality of Bom Retiro do Sul. All specimens come from the state of Rio Grande do Sul, Brazil.
All Brazilian specimens of Aleodon cromptoni are from the Dinodontosaurus AZ, Pinheiros-Chiniquá Sequence, Santa Maria Supersequence [35, 42] (Fig 2). Based on radiometric dating of the overlying unit (Santa Cruz Sequence; [43]) and tetrapod correlation with the Chañares Formation of Argentina (Ischigualasto-Unión Basin; [44-45]), the Dinodontosaurus AZ is considered to be Ladinian-early Carnian in age. The only potential non-Brazilian specimen of A. cromptoni is the isolated skull GSN EN-3 from the upper Omingonde Formation of Namibia [13]. The upper Omingonde Formation was previously correlated with the Dinodontosaurus AZ based on the shared presence of the dicynodont Stahleckeria potens [46].
Remarks on specimens. The holotype MPDC-501-117 was found in 1988 by the twin brothers Daniel (1930Daniel ( -2002 and Abraão (1930Abraão ( -2006 Cargnin, the renowned fossil-hunter priests of Rio Grande do Sul. For many years, MPDC-501-117 was listed as an indeterminate traversodontid cynodont in the exhibition of the Museu Padre Daniel Cargnin, located in the   Barberena) at this locality were previously catalogued as coming from the municipality of General Câmara. As was noted by Raugust [47], this area belonged to the 5˚District of the municipality of General Câmara at the time, but in 1995 it became an independent district and was renamed municipality of Vale Verde (Fig 1). Drawings in situ of some specimens from this site (e.g., the rauisuchian UFRGS-PV-0152-T) are available in the field notes of Valdor Costa, deposited at the UFRGS. UFRGS-PV-0122-T and UFRGS-PV-0274-T were previously listed as large specimens of Chiniquodon theotonicus by Abdala and Giannini [14], and UFRGS-PV-0146-T was described by Oliveira et al. [19] as Chiniquodon cf. C. theotonicus. The African cynodont Aleodon in the Triassic of South America UFRGS-PV-0125-T is catalogued as having been found in Sanga Nicanor, Candelária, by Padre Daniel Cargnin in 1969. Historical data at the MMACR indicates that Sanga Nicanor refers to the old name for Sanga do Zé (located in the Pinheiro region), where specimen MMACR-PV-018-T was also found.
The Cortado site (also called Rincão da Porta), which yielded MCN-PV 10338, has also yielded specimens of Candelaria barbouri, dicynodonts, and fragmentary cynodont material [41,71]. Based on firsthand examination of this material by the lead author, we identify the synapsid specimens from this site as Dinodontosaurus sp. and Massetognathus sp. One specimen (UFSM 11079) from this locality, published as Cynodontia indet. [41], may belong to a new specimen of the probainognathian Candelariodon barberenai. The holotype of the archosauriform Barberenasuchus brasiliensis is also considered to be from this locality [52,72]. The African cynodont Aleodon in the Triassic of South America

Description and comparisons
Descriptions of the known specimens of Aleodon cromptoni are presented below. The postcranial elements of specimen UFRGS-PV-0146-T are not described here, as they have been studied in detail by Oliveira et al. [19].
MPDC-501-117 (holotype). This specimen consists of a partial right maxilla (Fig 3) bearing the base of the canine crown and the subsequent seven postcanines. The maxillary facial process is partially preserved, and is tall and dorsoventrally convex. There is also strong surficial convexity around the bulging canine root, followed by a conspicuous constriction delimiting a shallow concave surface above the postcanine row. The dorsal tips of the canine root and last three postcanine roots are exposed due to abrasion of the dorsal surface of the maxilla. A The African cynodont Aleodon in the Triassic of South America number of small vascular foramina are distributed across the lateral face of the maxilla. The majority is associated with the canine root, but a few larger foramina are also present above the postcanine tooth row. The alveolar edge is straight in lateral view, at the level of the canine; posteriorly, interdental processes are present as sharp projections that define successive ventrally-concave alveolar outlines for each postcanine. There is no maxillary platform, as in most non-gomphodont cynodonts (e.g., Procynosuchus, Thrinaxodon, Lumkuia, Chiniquodon, Probainognathus, Bonacynodon, Irajatherium; [3,32,[73][74][75][76][77][78][79][80]). In ventral view, the palatal process of the maxilla is partially preserved. It bears two large foramina, the larger one positioned medial to the level of PC1-PC2 and the smaller one medial to the level of PC3-PC4. Just medial to the canine root, the maxillary wall is broken, but there is clearly a space (the paracanine fossa) for housing the lower canine. This fossa is deep and is positioned anteromedial to the upper canine. The lateral suture for the palatal process of the palatine starts at the level of the PC4-PC5 contact and is strongly interdigitated. Medial and parallel to the postcanine tooth row, there is no longitudinal groove on the ventral surface of the maxilla as in A. brachyrhamphus and in contrast to the condition in most probainognathians (e.g., Chiniquodon, Prozostrodon; [32,78,[81][82]) where this groove houses the tips of sectorial lower postcanines. In medial view, the maxillary surface is strongly concave, delimiting the wall of the nasal cavity.
The canine of MPDC-501-117 lacks most of its crown, but based on the cross-section of the preserved root it was a relatively large tooth, similar to other specimens of A. cromptoni. Its cross-section at base is ovoid, with the main axis slightly rotated in an anteromedial-to-posterolateral direction. The alveolus of the canine has a rounded pit at its posterior edge, which could represent a replacement pit. However, an erupting canine is not evident. The medial edge of the canine is offset lateral to the beginning of the postcanine tooth row. The African cynodont Aleodon in the Triassic of South America The postcanine tooth row of MPDC-501-117 starts just posterior to the canine alveolus, without a diastema or evidence of resorbed teeth (Fig 3). There are seven postcanines, but this count is unlikely to reflect the original number of upper postcanines, based on breakage after PC7 and the condition seen in other specimens of Aleodon (see below). The postcanine crowns are in contact with each other and they diverge posterolaterally, forming a laterally concave tooth row. The general crown morphology consists of a bulbous main labial cusp that forms a sectorial crest and a lingual occlusal platform. They increase in absolute size posteriorly. However, the most evident variation in the teeth is that posteriorly, they exhibit greater expansion in transverse width than mesiodistal length. Consequently, in occlusal view, the postcanines go from being roughly circular anteriorly to transversely ovoid posteriorly. There is a kind of mesio-distal rudimentary interlocking among the postcanine teeth, where the convex distal edge of a preceding tooth contacts the lingually-offset convex mesial edge of the subsequent one (for a well-developed example of this, see the contact between PC5 and PC6 in Fig 3B). The tooth row shows no clear evidence of tooth replacement; most of the crowns are at the same height (with the exception of PC7), with a comparable degree of wear. There is a clear distinction between crown and root-a layer of enamel is visible labially and lingually on the former and it is also distinctly broader than the root. PC1 is the smallest of the series. It has a bulbous main cusp on its mesiolabial corner that bears a distal crest. The labial and mesial edges of the main cusp form a strong convex surface, whereas the distolingual edge is flat-toconcave and slopes gradually towards the lingual platform. The presence of a distal accessory cusp on the distal crest of the main cusp is not evident. A weak crest arises at the mesiolingual corner of the main cusp and descends to border the lingual platform. It bounds an oval concave surface restricted to the distolingual portion of the tooth. PC2 is similar in morphology to PC1 but the main cusp has a more labial position, with a convex labial edge and a slightly convex lingual edge that forms an abrupt angle with the lingual platform. The main cusp is mesiodistally asymmetrical with the distal crest more elevated than the mesial one, possibly indicating the presence of a worn-off accessory distal cusp. In this postcanine, the lingual platform extends across the mesiodistal length of the crown. It is slightly transversely broader in its distal portion. The lingual platform is bounded by a rim of enamel and the occlusal surface exposes the underlying dentine. Between PC3 and PC2 there is a considerable increase in size, whereas in the rest of the row this change is more gradual. PC3 has a main cusp with mesial and distal crests that in occlusal view form an almost 90˚angle. The lingual platform is circular in outine in occlusal view, with a shallow fossa on its distolabial side. The PC4 and PC5 are similar in morphology to PC3 but the distolabial fossa on the lingual platform is deeper. PC5 seems to be slightly smaller than PC4, but this is an artifact, because most of the lingual enamel layer of PC5 is broken. PC6 is not well preserved. The crown has a more ovoid shape than the preceding teeth. The main cusp is bulbous and situated on the mesiolabial side where the lingual platform develops from the mesial to the distal positions of the crown. The last postcanine, PC7, is also badly preserved. The crown seems to be slightly nearer to the alveolar level, with no exposure of the root in labial view. Consequently, it could represent a more recently erupted postcanine, at the back of the series. In PC7, the lingual platform is more developed in comparison to the other postcanines.
In general, the postcanine morphotype of MPDC-501-117 is most similar to that of Aleodon brachyrhamphus among cynodonts ( [20]; NHMUK PV R9390, UMZC T906) (Fig 4), differing considerably from that of other known taxa [3,14,32]. Although gomphodont cynodonts also possess transversely wide postcanines, the crown morphology in these taxa is fundamentally different: the upper postcanines have a conspicuous transverse crest, often with three discrete cusps, that divides or bounds the occlusal basin and opposes a transverse crest with two main cusps generally present on the lower postcanines (e.g., [80]).

MCN-PV 10338.
This specimen consists of a partial mandible with the best preserved lower dentition for A. cromptoni (Fig 5). Both mandibular rami are broken, the left at the base of the coronoid process and the right somewhat anterior to that, at the level of pc9. The dentaries are fused at the symphysis. The symphysis is long and extends posteriorly to the fourth postcanine. It has a flat anterior surface that projects anteroventrally. The anteroventral edge of the symphysis and the ventral edge of the dentary form a conspicuous mentum. The horizontal ramus of the dentary is anteroposteriorly short and dorsoventrally tall, with a robust aspect. The alveolar border is almost straight whereas the ventral edge is gently concave. The base of the coronoid process is preserved on the left side. It is stout and its preserved portion has an angle of~140˚relative to the alveolar edge. Based on the preserved portion of left dentary, the masseteric fossa is relatively shallow and extends anteriorly to the level of the antepenultimate tooth. In medial view, the splenial is a flat and tall bone covering the Meckelian groove ( Fig 5B). The groove is positioned below the mid-height of the dentary and runs parallel to the ventral edge of the dentary.
There are five partial incisors (left i1-i2 and right i1-i3) and the alveolus for an extra one (left i3) preserved in the lower jaw. The whole crown is not preserved for any of these incisors. The incisors are large with a mesiodistally flattened cross-section. The i1 and i2 are similar in size and slightly larger than i3, based on the preserved cross-sections. The incisors are tightly packed, and lack a diastema with the canine. As in the previous specimen, the canine is large, with an oval cross-section. The canine tip is acute and prominently recurved. The left and right canines are close to one another across the midline (separated by only 9mm), although this may be due in part to postmortem deformation. There is not a clear diastema with the postcanine tooth row, although this portion of the jaw is poorly preserved. The left mandibular ramus preserves 12 postcanines whereas the right ramus partially preserves nine teeth. The postcanines increase in size posteriorly and go from being circular (in the anterior half) to oval (in the posterior half) in occlusal view, with a short labial sectorial crest and a lingual platform with basin. Due to wear or breakage, a cuspidate sectorial labial edge is not observed in pc1 to pc9.
The left pc1 to pc3 only preserve the root. The left pc1 has no crown and the small portion of root seems to have been in the process of resorption, which also seems to be the case for the right pc1 and pc2. The left pc4 is circular in cross-section and exhibits an unusually tall root, which probably slipped out of the alveolus post-mortem. The shape of the left pc5 is distorted by a large, transverse crack running across it. The left pc6 and pc7 are the last circular-shaped teeth of the row. In these teeth, the crown is low, with elevated labial and lingual edges bordering a concave occlusal basin. This basin is not covered by enamel, although a thick layer of enamel covers the crown externally. The pc8 and pc9 are ovoid, with completely worn-off crowns. The enamel layer that covers the crown on the labial surface is very thin. The pc10 to pc12 are the best preserved elements, with nearly unworn crowns. They are oval in cross-section, with a low crown. The crown is conspicuously wider than the root, which is circular in cross-section. The labial crest is low with a main cusp a that is curved posteriorly and slightly laterally. There are three accessory cusps: a tiny mesial cusp b and two distal cusps, c and d (Fig 5D and 5E). The main cusp a is separated from cusp c by a deep notch. The lingual surface of the sectorial crest slopes towards the lingual platform gradually, without a distinctive step. This lingual surface and the occlusal platform have a thin layer of enamel, whereas the enamel that covers the crown externally is relatively thick. Accessory cusps on the lingual edge of the lingual platform are not evident. In lateral view, the last three postcanines have a similar cusp pattern to that of Chiniquodon (see Discussion), although the crown is lower and the cusps are more discrete. The right postcanines do not provide further details and their crowns are largely worn off.

UFRGS-PV-0071-T.
This specimen consists of the anterior half of a skull without lower jaws (Fig 6). This specimen is poorly-preserved overall, having suffered from the calcitic replacement and secondary expansion typical of Santa Maria fossils. It is referred to A. cromptoni because of the presence of transversely-expanded postcanines, with a labial cusp and broad lingual platform ( Fig 6B). The snout as preserved is similar to that of other Aleodon specimens. UFRGS-PV-0071-T is much smaller than the holotype and represents one of the smaller known specimens of A. cromptoni (58.6mm palatal length and 56.8mm muzzle length). Four upper incisors are preserved on each side of the snout. They are sub-conical, evenly spaced, posteroventrally angled, and decrease in size posteriorly. The right I2 and left I1 and I3 were in process of eruption at the time of the animal's death, with I2 and I3 less erupted than I1. The canine is large and ovoid in cross section. There is a deep paracanine fossa positioned anteromedial to the canine. All of the postcanine crowns in this specimen are damaged, with numerous cracks across their surface, but the postcanine tooth row is somewhat better preserved on the right side of the skull. The right tooth row consists of seven postcanines plus an alveolus at the rear with an erupting tooth. The postcanines in this specimen are circular to oval in cross section, and their overall morphology indicates the presence of a lingual platform, which is clearly evident in the right PC5-6. The right PC5 exhibits an elevated labial crest and a very small accessory cuspule on the distolingual corner of the lingual platform. This cuspule is worn off apically, but can be distinguished from the rest of the crown by the presence of a short vertical notch. Cuspules of this kind would be quickly eroded during mastication and, therefore, are worn out in functional teeth. The postcanine tooth row is concave laterally and projects lateral to the subtemporal fenestra. The last postcanine is positioned posterior to the level of the anterior orbital edge and just posterior to the anterior edge of the subtemporal fenestra. The secondary palate is long, reaching the level of the last postcanines ( Fig 6A). It is made up almost equally by the maxilla and palatine. The anterior half of the secondary palate is almost flat, outside of the conspicuous paracanine fossae, whereas the posterior half is transversely wider and slightly convex. Just medial to the last postcanine teeth there is a shallow concavity on the palate. The medial process of the palatine, which contacts the primary palate, projects posteromedially onto the pterygoid body by means of a triangular process. The pterygoid wings are wide and, based on the preserved portion, are shorter and less sharply pointed at tip than in Chiniquodon [14]. They are posteriorly and slightly ventrally projected. Along the midline, the pterygoids unite to form a sharp medial crest. The dorsal and lateral aspects of the skull are in need of futher preparation.
MMACR-PV-018-T. This specimen consists of a left maxilla with the root of the canine and nine postcanine alveoli (Fig 7). The size of this specimen is similar to UFRGS-PV-0071-T; together these represent the smallest known individuals of A. cromptoni. The maxilla is tall with a convex surface on the bulging canine root and a concave constriction posteriorly. There is a large maxillary foramen, positioned above the level of the alveolus for PC2. The canine root is oval in cross section and, as in other specimens of A. cromptoni, the postcanine row diverges posterolaterally. The first postcanine alveolus is oval. Posteriorly, the alveoli are transversely elongated until the alveolus of PC6, which is the transversely widest of the row. After this point, the PC8 and PC9 alveoli become transversely narrow and oval, with the main axis mesiodistally oriented (Fig 7). The presence of transversely narrow alveoli (or teeth) at the rear of the upper tooth row is also observed in specimens UFRGS-PV-0071-T (see above) and UFRGS-PV-0125-T (see below). This indicates the presence of a few sectorial postcanines in the last position of the tooth row, another feature convergent on the condition in gomphodonts like Diademodon. The alveolus of PC10 preserves only its mesial edge. It apparently housed a sectorial tooth, as did two preceding alveoli.

UFRGS-PV-0122-T.
Although very poorly preserved, this specimen is important in representing the largest known individual of Aleodon (310mm basal skull length). It consists of the skull (Fig 8) and a complete right and partial left lower jaw. The preorbital region is shorter than the postorbital, and this does not appear to be due to deformation. The temporal region is wide, and bounded by a deep, prominently arched zygoma (Fig 8). The zygomatic arch is similar to that of UFRGS-PV-146T. Its ventral edge is strongly concave ventrally and at its anteroventral corner there is a marked angulation, as in Chiniquodon [14]. The parietal crest is tall and the skull roof tapers anteriorly. The secondary palate is also well-developed posteriorly, differing from gomphodont cynodonts. The precise number of incisors and postcanines is unknown. The upper canines are extremely large, with the preserved portion of the right crown being 46 mm long. The shape of most postcanines is indiscernible but a few ones from the right side clearly exhibit a circular cross-section in occlusal view, indicating a lingual platform. The first postcanine is positioned medially to the level of the canine, similar to other specimens (small and large) of Aleodon, and differing from Chiniquodon, in which the canine and first postcanines are aligned [73,[81][82]. The jaw is robust with an elongated symphysis, high horizontal ramus, and tall coronoid process (Fig 8). The angular process of the dentary is posteriorly broken and part of the post-dentary complex is preserved. The lower postcanines are not available.
UFRGS-PV-0125-T. This specimen consists of a laterally compressed, poorly preserved cranium (Fig 9) suffering from calcitic expansion. The majority of the dentition has fallen out, with the exception of the first three left postcanines. There are four alveoli for incisors, which decrease slightly in size posteriorly. The canine alveoli are large, as in other specimens of A. cromptoni. The left postcanine tooth row preserves the PC1 to PC3 plus five alveoli, and the right side preserves at least nine alveoli. The postcanine tooth row is laterally concave, with its posterior end angled posterolaterally even though the skull is compressed (unlike the condition described for UFRGS-PV-0146-T below). The PC1 and PC2 are smaller than PC3, peglike in shape and without a diastema with the canine. Their crowns are poorly preserved. The PC3 is an intact but heavily worn crown. PC3 bears a low labial cusp and lingual platform and is transversely expanded and ovoid, similar to that of other Aleodon specimens. The last two alveoli in the right tooth row are markedly narrower in their transverse dimensions than the preceding ones, suggesting that these teeth may have been sectorial (Fig 9). In addition to postcanine morphology, identification of this specimen as Aleodon rather than as a gomphodont is supported by the great length of the secondary palate (terminating posterior to the postcanine tooth row), lack of maxillary platform, and the proportionally small size of the incisor alveoli. The zygomatic arch of this specimen (best preserved on the right side) and is extremely tall and robust, as in other large specimens of Aleodon. The dorsal profile of the skull is also similar to other Aleodon specimens, with a tall parietal crest at its posterodorsal edge.
UFRGS-PV-0146-T. Oliveira et al. [19] previously described this specimen in detail, albeit focusing on the postcranial elements. It was referred to Chiniquodon because of the presence of a relatively short snout, very deep zygomatic arch (Fig 10), angulation between the posterolateral portion of the maxilla and anteroventral margin of the zygomatic arch (Fig 10), and long secondary palate [19], following the taxonomic proposal of Abdala and Giannini [14]. With regard to the postcanines, they mentioned that because the postcanine crowns are broken, their presumable sectorial morphology could not be observed. The few differences observed between the postcranium of this specimen and that previously known for Chiniquodon [83] were considered to be attributable to intraspecific variation [19].
Although the postcanines of UFRGS-PV-0146-T are poorly preserved, they are not all broken (contra [19]) (see Figs 12 and 14). Several of the crowns are intact, albeit heavily worn, and exhibit the characteristic postcanine morphology of Aleodon (especially the left PC5; see Fig 14 and description below). The morphology of the dentition and its biostratigraphic (i.e., Dinodontosaurus AZ) and geographic provenance allow us to refer UFRGS-PV-0146-T to A. cromptoni. This specimen is larger than the holotype, but smaller than UFRGS-PV-0122-T (see S1 File).
The skull is slightly laterally compressed and bears several cracks that obscure the recognition of cranial sutures. The snout is shorter than the temporal region and the orbits face anterolaterally. Although deformed, the temporal fossae are clearly sub-rectangular, widening posteriorly. The zygomatic arch, postorbital bar, parietal and lambdoidal crests are robust. In lateral view, the profile of the skull is roughly triangular, with an elevated parietal crest that gradually slopes downwards anteriorly. The maxillary facial process is tall, with a prominent convexity produced by the canine root. The premaxilla is small and has a conspicuous ascending internarial process. The septomaxilla has a large lateral process extending posterodorsally between the nasal and maxilla. The zygomatic arch is composed of the squamosal and jugal. The jugal projects far posteriorly to reach the ventrolateralmost portion of the squamosal, longer than in most Chiniquodon specimens [14]. The squamosal projects anteriorly over the jugal, terminating below the posterior portion of the postorbital bar. The ventral edge of the zygomatic arch is highly concave, producing a prominent angulation at its anterior edge (below the postorbital bar) as in Chiniquodon. The deepest part of the zygomatic arch is in its posterior portion, which is as tall as the snout height at the middle of the tooth row. In ventral view, the secondary palate is long, extending to the level of the last postcanine. Based on a partial suture preserved on the left side, the lengths of the maxilla and palatine in the secondary palate are similar, as in other specimens of A. cromptoni (see above). The pterygoids are not well preserved and the primary palate and basicranial region are badly distorted. In posterior view, the occipital plate is triangular, with a deep, v-shaped notch separating the lambdoidal crest from the zygomatic arch.
The dental formula is 4I/3i, 1C/1c, 8PC/9pc. The upper incisors are subconical and slightly posteroventrally projected. The incisors are closely spaced. They increase in size posteriorly and there is a short diastema with the canine. The canine is a very large tooth (see S1 File) that is oval in cross section at the base. The upper and lower canines are recurved, although this is less evident in the somewhat-distorted left upper canine (Fig 10). There is no evidence of canine serrations. The postcanine tooth rows are not well preserved and slightly deformed, but it is evident that there is no diastema with the canine. The left tooth row is laterally concave and projects lateral to the subtemporal fossa, as in the holotype of A. cromptoni. On the other hand, the right tooth row seems to be straighter due to deformation of the maxillary body and zygomatic arch. The left tooth row preserves evidence of eight teeth, of which only four retain remnants of crowns. One in particular, the PC5, is largely intact and exhibits the diagnostic morphology of Aleodon, with a single labial cusp and large, transversely expanded lingual platform (Fig 14). The circular to ovoid cross-section of the crown, due to the development of the lingual platform, is also evident in PC3 to PC7 on the left side and the middle postcanines of the right side. On the right side, the precise number of teeth is uncertain due to poor preservation; parts of at least five crowns are preserved.
UFRGS-PV-0274-T. This specimen is a relatively large skull, with a total length of 277mm. The skull is dorsoventrally compressed and there are several cracks on its surface ( Fig  15). Part of the zygomatic arch, all the incisors, and both canines are reconstructed with plaster and resin. It has the same features observed in the aforementioned specimens, including a short and wide snout, with a conspicuous bulging at the canine root. The secondary palate is long and slightly convex at its posterior portion. The postcanine tooth rows only preserve portions of the right PC2 and PC7 and left PC6. The right postcanine row includes a very mesiodistally short alveolus for PC1. The PC2 crown is broken with a circular cross section. The alveoli for PC3 to PC6 are oval with their main axis, transversely oriented. The PC7 preserves part of the crown. It is oval shaped, in occlusal view, with a lingual platform. It seems to be an erupting tooth but due to the poorly preserved portion of the maxilla this is uncertain. At this point, the maxilla is obliquely crushed and the posterior portion slightly shifted laterally on the anterior portior, so the exact orientation of these teeth is difficult to discern. For example, the alveolus for PC8 seems to be almost at the same tranverse level as the alveolus of PC7. The alveoli for PC7 and PC8 are transversely narrow and decrease in size posteriorly, indicating the presence of sectorial teeth at the back of the tooth row. On the left side, the postcanine row preserves two circular alveoli for PC1-PC2. The alveolus for PC3 is considerably larger transversely, indicating a conspicuous change in tooth size at this point. The following alveoli are also transversely wide until reaching the alveolus for PC7. The last alveoli are partially preserved and the total number in this side is unknown. This row preserves part of the crown of PC6. It is somewhat oval in cross-section and bears the lingual platform characteristic of Aleodon. As in the previously-described specimens and differing from large specimens of A. brachyramphus, there is no diastema between the canine and postcanine teeth.

Assignment of the Brazilian specimens to Aleodon
MPDC-501-117 is referred to the genus Aleodon based on the presence of labiolinguallyexpanded postcanines with a tall main labial crest, dominated by a main cusp and a well-developed lingual platform in the preserved upper postcanine teeth. The lingual platform is not developed to the same degree in any other Triassic probainognathian, including chiniquodontids [14,15,[78][79]84], ecteniniids [85][86][87], probainognathids [3,74], and prozostrodontians [3,34,78,79,[88][89][90][91]. A posterior lower postcanine of Candelariodon barberenai, from the Dinodontosaurus AZ, was originally thought to be similar to Aleodon [92]; however, it has since been demonstrated that its dentition more closely resembles that of some prozostrodontians (e.g., Prozostrodon) and that the tooth in question is actually broken [93]. Another interesting taxon in this regard is the Middle Triassic Cromptodon mamiferoides (holotype PVL 3858) from Mendoza Province, Argentina [94]. It also has postcanines with a wide lingual cingulum, but unfortunately only the lower jaws are known. Nonetheless, the lower dentition of Cromptodon is clearly distinct from that of both Aleodon species. In Cromptodon, the lower postcanines have a sectorial labial crest with a conspicuous main cusp a, mesial cusp b and distal cusps c and d (cusp height = a>b>c>d), whereas in Aleodon the sectorial labial crest bears a conspicuous and bulbous main cusp, one very tiny mesial cusp, and two small distal cusps. The sectorial labial border of the lower postcanines of Aleodon is more similar to the morphology of Chiniquodon and ecteniniids, with the main cusp strongly posteriorly angled, than that of Cromptodon, in which the main cusps are more vertically positioned. In Cromptodon, the lingual cingulum is less developed, bears discrete cusps, and reaches its maximum width in the middle of the tooth row (pc3-pc4), then decreases in width in the posterior teeth. In contrast, in the well-preserved jaw MCN-PV 10338 of Aleodon cromptoni the lingual cingulum is the most conspicuous structure in all the postcanine teeth, being greatly expanded in both labiolingual and mesiodistal dimensions relative to the labial portion of the crown. The only known specimen of Cromptodon is extremely small (~27mm dentary length), contrasting considerably with the size sample for Aleodon cromptoni. Nonetheless, the presence of a canine-postcanine diastema in the former taxon would indicate a sub-adult or adult individual; therefore, the size difference between both taxa is notable and probably reflective of real taxonomic distinction.
The postcanine dentition of Aleodon is distinctive even when compared with gomphodont cynodonts. In the latter group, the upper postcanines bear a transverse crest, with a different position in each family, which is connected to the labial sectorial crest [80,[95][96]. The lower postcanines are usually subrectangular with a transverse crest, usually positioned on the mesial edge in traversodontids. Both upper and lower postcanine teeth develop conspicuous basins that, in conjunction with the crests, produce a mechanism for grinding [80,96]. In Aleodon, the upper and lower postcanines have a sectorial labial crest and a lingual platform, capable of rudimentary crown-to-crown occlusion, without a transverse crest. Also, the upper and lower postcanine teeth of Aleodon exhibit a similar pattern of cusp and basin, differing from traversodontids in which the morphology of the upper postcanine teeth is clearly differentiated from that of the lowers.
The additional Brazilian specimens described above can also be referred to Aleodon based on the presence of postcanine teeth with a large lingual platform, and excluded from gomphodonts due to some combination of lengthy secondary palate, lack of tranverse crest on upper/ lower postcanine teeth, and lack of maxillary platform. Although some of these specimens are poorly preserved, taken together they provide a nearly-complete view of various aspects of the osteology of A. cromptoni, such as dentition (

Aleodon brachyrhamphus and A. cromptoni
Aleodon cromptoni and A. brachyrhamphus are clearly closely-related taxa, and at present can only be distinguished based on features of the dentition. The publication of better preserved cranial specimens of A. brachyrhamphus (hopefully forthcoming) will be required to establish additional autapomorphies of these species. Here we discuss the key points of comparison currently available between Aleodon cromptoni and A. brachyrhamphus: 1. Although both species have a well-developed lingual platform on the upper and lower postcanine teeth, in A. cromptoni it is mesiodistally wider and labiolingually narrower than in A. brachyrhamphus. Consequently, in the latter taxon, the upper postcanine crowns are more transversely elongated.
2. Within the analyzed sample, the upper postcanine teeth of A. cromptoni have a distinct size increase between PC2 and PC3, as clearly shown in the holotype (Fig 4). By contrast, A. brachyrhamphus has a gradual increase in upper postcanine size between PC1 and PC4. Similar-sized specimens of both species (Fig 4) show different proportions in the absolute size of the postcanines. A. cromptoni has larger postcanine teeth in relation to A. brachyrhamphus, with the lingual platform two to three times wider than the sectorial labial crest in the middle and posterior postcanines.  (Fig 4). In fact, in none of the specimens referred to A. cromptoni is there a conspicuous diastema. Therefore, it is possible that there are some differences in tooth replacement between these species, with limited or absent alveolar resorption in the anterior postcanine row of A. cromptoni.
4. The lower postcanines of the two Aleodon species are different in crown morphology. In A. brachyrhamphus (UMZC T906), the first three teeth are circular in occlusal view and they increase in width abruptly, with the fourth tooth considerably wider than the previous ones. From pc4 to pc7, they widen slightly and maintain almost constant width. They are more transversely elongated than in A. cromptoni, being about two times broader than long. However, in A. cromptoni the teeth remain circular in occlusal outline until pc5, and then have a less-developed lingual platform than in A. brachyrhamphus. The labial edge of the crown bears a main cusp in A. brachyrhamphus, whereas in A. cromptoni there is a sectorial crest with three discrete cusps descending in size posteriorly plus a tiny mesial one. The contact area between teeth of the same row is larger in A. brachyrhamphus than in A. cromptoni, due to the more elongated lingual platform in the former. As was mentioned in point (c), the smaller number of postcanines in A. brachyrhamphus would be related to the loss of anterior teeth, as clearly observed in UMZC T906, which has a large diastema and fewer postcanines than MCN-PV 10338.
5. In the holotype of A. brachyrhamphus (UMZC T906), only two lower incisors were described in each jaw [20], differing a priori from the condition of A. cromptoni, which has three incisors, a number common in probainognathians [32]. However, based on personal observations of UMZC T906 (by AGM), it seems to have three incisors in the left side and two in the right, with the left mediolabial border of the dentary broken off. Therefore, we assume that A. brachyrhamphus had three lower incisors, like A. cromptoni. Upper incisors are unknown in the holotype of A. brachyrhamphus. A referable specimen, NHMUK PV R9390, has four upper incisors, also as in A. cromptoni.
6. The upper postcanines of the A. brachyrhamphus specimen NHMUK PV R9390 have a longitudinal depression on the lingual face of the crown, such that there is a distinct 'notch' on the lingual platform ( Fig 4A). This feature is more evident in postcanines in the mid-length of the row. The consistent presence of this character-state in A. brachyrhamphus specimens is uncertain, in large part because of poor preservation of the postcanine crowns in other specimens of this species. However, this feature is clearly not present in any specimen of A. cromptoni. 8. Based on the size range present in the sample of A. brachyrhamphus from the Manda Beds (for those specimens deposited in the UMZC and NHMUK collections), it seems that the Brazilian A. cromptoni reached considerably larger skull size than its congener, as illustrated by specimens UFRGS-PV-0122-T, UFRGS-PV-0146-T, and MCN-PV 10338, and also by the tentatively-referred specimen GSN EN-3 from Namibia. In the sample of A. brachyrhamphus complete skulls are unavailable and, consequently, its total skull length is unknown. The holotype UMZC T906 represents one of the largest A. brachyrhamphus specimens; its jaw is smaller than the A. cromptoni jaw MCN-PV 10338 (see Fig 5), here considered a mid-sized specimen.

Identification of the Namibian specimen GSN EN-3 as A. cromptoni
Abdala and Smith (2009) described an isolated skull with worn dentition (GSN EN-3) from the upper Omingonde Formation of Namibia as a specimen of Aleodon (Fig 16). This specimen was referred to Aleodon on the basis of its snout proportions, secondary palate length, and tooth morphology (specifically the labiolingually-expanded, ovoid-ellipsoid postcanine crowns). Abdala and Smith [13] did not identify this specimen to species (leaving it as Aleodon sp.), presumably because of its great geographic separation from the Manda Beds of Tanzania. The only possible difference they noted between this specimen and the Tanzanian material was the presence of a pineal foramen in UMZC T906 (holotype of A. brachyrhamphus), which was mentioned as present by Crompton [20]. However, they also noted that they could find no evidence of a pineal foramen when re-examining UMZC T906, and subsequent examination of that specimen by the authors of this paper (AGM and CFK) corroborates its absence. The cranial morphology of GSN EN-3 is typical for Aleodon, with a tall, well-developed parietal crest and deep, robust zygomatic arches (Fig 16A and 16B), but the cranium alone does not exhibit any characters that permit identification as either A. brachyrhamphus or A. cromptoni. The postcanine tooth rows in this specimen are externally concave and point posterolaterally, terminating lateral to the subtemporal fossa in ventral view. The right postcanine tooth row preserves only one tooth (PC8), but also 11 alveoli. The left postcanine tooth row preserves 11 intact but badly worn crowns. The high degree of wear on these teeth makes species referral based on their morphology somewhat suspect, which is why we consider the following identification to only be tentative. With this noted, two features of the postcanine tooth row of GSN EN-3 are notable. One is the marked increase in tooth size between PC2 and PC3 (seen via alveolar dimensions on the right side) (Fig 16D and 16E). The other is the very broad, rounded lingual platform present in the best-preserved postcanine crowns (particularly the left PC6, 7, and 11). These are both characteristics that distinguish A. cromptoni from A. brachyrhamphus. Although it is possible that the Namibian material represents a third, endemic species of Aleodon, based on the above characters, the close geographic proximity of Rio Grande do Sul and Namibia during the Triassic, and the recent discovery of other characteristic elements of the Dinodontosaurus AZ in the upper Omingonde Formation (i.e., Stahleckeria potens; [46]), we consider it more likely that this specimen represents A. cromptoni.

Aleodon versus Chiniquodon
Abdala and Giannini [14] revised the taxonomic composition of Chiniquodontidae, which at the time was restricted to the Middle-Late Triassic of Argentina and Brazil. They synonymized Probelesodon [73,75,82,84] and the large-bodied taxon Belesodon [97] with the genus Chiniquodon [97], a taxon initially described from the Santa Maria Formation of southern Brazil. Previously stated differences between these taxa were explained as ontogenetic factors. Subsequently, material referred to Chiniquodon sp. was recognized in the Santacruzodon AZ of southern Brazil [98] and the upper Omingonde Formation of Namibia [13], and a new species, C. kalanoro, was described from the basal "Isalo II" beds of Madagascar [15]. As a result, the Chiniquodontidae currently includes: Chiniquodon theotonicus from the Chañares Formation of Argentina and Dinodontosaurus AZ of Brazil, C. sanjuanensis from the Ischigualasto Formation of Argentina, and C. kalanoro from basal "Isalo II" beds of Madagascar. Consequently, this genus spans a remarkably broad geographic and temporal range, extending across much of Gondwana and minimally from the Carnian to the Norian. Aleodon was excluded from the Chiniquodontidae by Abdala and Giannini [14], in contrast to the previous proposals of Hopson [26] and Hopson and Kitching [32].
The features diagnosing Chiniquodontidae include (see [14]): (a) robust zygomatic arch, flared laterally; (b) angulation between the ventral edge of the maxillary zygomatic process and the anteroventral margin of the jugal; (c) pterygoid flanges greatly elongated, ending in a thin projection; (d) long secondary osseous palate; and (e) postcanines with backwardly recurved main cusps, lacking cingula or with tiny lingual cingular cusps. The new described specimens of Aleodon clearly demonstrate that both Chiniquodon and Aleodon possess the same general cranial morphology, because they share the aforementioned traits (a-d), differing only in the dental feature (e). Aleodon is characterized by having non-sectorial postcanine teeth bearing a conspicuous, basined lingual platform, not sharply-recurved sectorial teeth throughout the postcanine row. The posteriormost upper postcanines of Aleodon are more sectorial in shape and bear recurved main cusps, however. The morphology of these teeth adds additional support for a close relationship between these taxa, as previously proposed by Hopson and Kitching [32] and Ruta et al. [33]. It is noteworthy how congruent the skull and jaw morphologies are between Chiniquodon and Aleodon despite their disparate dental forms.
Skull morphology is fairly consistent among the specimens referred to A. cromptoni, in which the skull length ranges from~13cm (estimated from UFRGS-PV-0071-T) to 31cm (UFRGS-PV-0122-T). They have a short, wide snout, large zygomatic arches, robust jaws, and large canines (even in the smallest specimen). By contrast, the sample of specimens referred to Chiniquodon theotonicus is noticeably more heterogeneous. Specimens from the Chañares Formation from Argentina are usually smaller in size and more slenderly built, with long, narrow snouts. This morphotype is also observed in at least one specimen that comes from the Dinodontosaurus AZ, the holotype of Probelesodon kitchingi (MCP-PV-1600-T; [81,82]) (Fig 17). Most other specimens of Chiniquodon from Brazil are poorly preserved [14,97] and several referred specimens are still unprepared and in concretions. Independent of size, this morphotype bears a short snout and robust zygomatic arches (UFRGS-PV-0066-T, UFRGS-PV-1331-T (ex UFRGS-PV-0066-Tg by Abdala and Giannini [14]), with a marked concave outline between the parietal crest and the roof of the snout in lateral view (see dotted line in Fig 17). In contrast, the sample from the Chañares Formation and the specimen MCP-PV-1600-T from the Dinodontosaurus AZ have a more straight dorsal outline in lateral view, slightly tapering anteriorly (Fig 17). This outline in the Chiniquodon sample from Brazil is not considered a product of postmortem processes because it appears in several specimens, with different sizes and states of preservations. In both cases, robust or gracile forms, all the postcanine teeth are sectorial (Fig 17), distinguishing Chiniquodon from Aleodon. Further preparation of the extensive Brazilian sample of "Chiniquodon" would be greatly beneficial towards resolving the Chiniquodon-Belesodon-Probelesodon complex, and permit testing whether the aforementioned heterogeneity is attributable to taxonomy, ontogeny, and/or sexual dimorphism.
The referred sample for Aleodon cromptoni includes specimens reaching skull lengths greater than any known for Chiniquodon. The specimens UFRGS-PV-0274-T (27.7cm) and UFRGS-PV-122-T (31cm), both cited as the largest Chiniquodon individuals by Abdala and Giannini [14], are here considered to belong to A. cromptoni. Therefore, the largest specimen currently retained in Chiniquodon is the holotype of Belesodon magnificus (GPIT/RE/7112) at 26cm. The remaining specimens from Brazil and Argentina range from~5 to~20cm, with the most common size range between 10 and 15cm (see [14]).

Phylogenetic placement of Aleodon cromptoni
The cladistic analysis resulted in 75 most parsimonious trees of 529 steps (Ci = 0.35, Ri = 0.72). A selected part of the strict consensus tree is shown in Fig 18 (see S1 File for the complete tree), depicting Aleodon cromptoni as sister-taxon of A. brachyrhamphus, within Chiniquodontidae. Both species are supported by three ambiguous sinapomorphies: tooth-to-tooth contact because of widened postcanines (ch. 108 (3)), bucco-lingually expanded upper postcanines (ch. 110(3)), and well developed lingual cingulum in lower postcanines (ch. 124 (2)). Chiniquodontidae, including Chiniquodon plus Aleodon species, is supported by four ambiguous synapomorphies: large extranasal process of the premaxilla but not contacting nasal (ch. 2(1)), suborbital angulation between maxilla and jugal (ch. 26(1)), posterolateral end of maxilla forms right angle ventral to jugal contact (ch. 49(1)), and length of the secondary palate longer than the anterior border of orbit (ch. 52 (2)). Although the inter-relationships of probainognathians are not the scope of the present contribution, the clade Chiniquodontidae and the placement of the new species (A. cromptoni) are well-supported within Probainognathia, with high Bremer support values (see Fig 18). The African cynodont Aleodon in the Triassic of South America

Biostratigraphic implications
In recent years, radioisotopic dating from various continental formations has substantially altered our understanding of Triassic chronology and the relative ages of the major tetrapod clades. In particular, these dates have cast doubts on the reliability of vertebrate biostratigraphy for establishing the age of Triassic faunal assemblages. However, interbasinal correlations were historically made at very rough levels, often establishing equivalency based on "family"-level taxa. More recently, new fieldwork and re-evaluation of historic specimens has revealed significant lower-level taxonomic overlap between cynodont faunas in South America and southern Africa (Fig 19). We discuss these biostratigraphic issues below, establishing context for the importance of Aleodon in Brazil.
Although the new dates from Chañares and the Santacruzodon AZ shift the age of these faunas from Middle (Ladinian) to Late (Carnian) Triassic, the effects of this change on tetrapod correlation are relatively minor. The taxa such as traversodontid cynodonts, rhynchosaurs, and rauisuchians that dominate these assemblages were previously known from Carnian deposits, so a younger age for these faunas is not particularly surprising; the major novelties are the implications for the diversification of dinosauromorphs (e.g., [45]). By contrast, dates produced by Ottone et al. [106] for the Quebrada de los Fósiles Formation (San Rafael Group, Argentina) require a substantial overhaul of our understanding of Triassic biochronology if accurate. Ottone et al. [106] provided a SHRIMP 238 U/ 206 Pb age of 235.8±2.0Ma from an ignimbrite blow at the top of the Quebrada de los Fósiles Formation, placing the tetrapod association of the younger Río Seco de la Quebrada Formation in the beginning of the Carnian. A similar age had been obtained by Valencio et al. [107] based on a 40 K/ 40 Ar dating of 232±10Ma from an ignimbrite at the middle section of the Río Seco de la Quebrada Formation. This proposal is extremely problematic in terms of biochronology, because the Río Seco de la Quebrada Formation is the only South American unit bearing cynodonts (Cynognathus crateronotus and Diademodon tetragonus; [6,9,10]) typical of the Cynognathus AZ of the Karoo Basin in South Africa, an assemblage traditionally considered Early to Middle Triassic (Olenekian-Anisian) [7,[108][109][110][111]. Cynognathus crateronotus and Diademodon tetragonus have been widely-utilized in biostratigraphic correlations, since they achieved a broad Gondwanan distribution (Argentina, South Africa, Namibia, Zambia, Antarctica) during the Triassic and are relatively abundant in the fossil record [8,10,109,110,112]. Based upon the new radiometric dating from San Rafael Group and the traditionally older inferred age of the Cynognathus AZ of the Karoo (although see [113]), Ottone et al. [106] suggested two possible scenarios: (a) that the classical Cynognathus AZ of South Africa is much younger than previously thought or (b) that the biochron of Cynognathus and Diademodon is longer than previously thought, and thus their records in Argentina reflect the persistence of this fauna for nearly 20 million years. Both of these scenarios are highly problematic for Triassic faunal correlation (Fig 19). Other studies dating the top of the lower section of the Quebrada de los Fósiles Formation produced an age of 245Ma [114,115]. Based on these dates, the upper section of the Quebrada de los Fósiles Formation would have a span of about 9Ma (i.e., almost all of the Ladinian) [115].
The other tetrapod assemblage from the Triassic of Argentina is the traditional "Río Mendoza local fauna" of the Cerro Bayo of Potrerillos (Mendoza Province) [94,101] referred to the Cerro de Las Cabras Formation (Uspallata Group, Cuyo Basin [102]; it was originally assigned to the Rio Mendoza Formation, see also [116]). The Cerro de Las Cabras Formation includes the dicynodont Vinceria andina, the traversodontid cynodont Andescynodon mendozensis, and the probainognathian cynodont Cromptodon mamiferoides [94,101]. Due to the shared presence of Vinceria, this association was correlated with the Río Seco de la Quebrada Formation of the Puesto Viejo Group (e.g., [117][118]), with both considered as Anisian in age. Martinelli et al. [10] suggested that the tetrapod association of the Cerro de las Cabras Formation may prove to be somewhat younger than the Puesto Viejo association (as also suggested by [119]) based on the phylogenetic relationships of the traversodontids from Cerro de las Cabras and Puesto Viejo [96,120] in combination with the presence of Cynognathus and Diademodon in the latter unit. Nevertheless, based on dating from the Cuyana Basin [121][122], the tetrapod association from the Cerro de Las Cabras Formation does appear to be Middle Triassic, possibly Anisian, in age (Fig 19).
Based upon the dates from the Puesto Viejo Group [106] and those from the Cerro de Las Cabras Formation (Cuyana Basin; [121][122]), biostratigraphic correlations between both units are quite problematic and do not fit with previous faunistic interpretations (e.g., [10,117,123]), especially because of the occurrence of Cynognathus and Diademodon in the supposed early Carnian Río Seco de la Quebrada Formation. As was briefly commented on by Sues [124] and Martinelli and Soares [4], the tetrapod composition of the Río Seco de la Quebrada Formation is highly bizarre when compared to other basins from South America (i.e., Ischigualasto-Villa Unión Basin in Argentina and Paraná Basin in Brazil). Even if the hypothesis of Ottone et al. [106] is accepted, in which the biochron of Cynognathus and Diademodon is longer than the African one, the absence of both genera in other well-sampled basins of Argentina and Brazil of early Carnian age would be a remarkably aberrant distribution (Fig 19). Also, the "evolutionary grade" of the entire tetrapod fauna, not only the cynodonts, is much more derived in the Carnian-Norian formations of the Ischigualasto-Villa Unión and Paraná basins. It is possible that the San Rafael Group may be younger than the typical Cynognathus AZ of the Karoo, but its proposed position as coeval to the Chañares Formation in Argentina and the Dinodontosaurus AZ in Brazil makes little sense in the evolutionary scenario of Triassic tetrapod communities. Determination of the exact stratigraphic provenance of Cynognathus and Diademodon within the Puesto Viejo Group, their stratigraphic correlation with the traversodontid Pascualgnathus, and new radiometric data are urgently needed to elucidate these biostratigraphic issues.
Brazilian Aleodon and African biostratigraphy. With regard to the continental African Triassic assemblages, several biostratigraphic problems have been also highlighted (e.g., [13]). The Karoo Supergroup has traditionally been used as the standard for correlating Permo-Triassic units worldwide, due to its vast temporal span, the abundance of collected fossils, and their taxonomic diversity (e.g., [111]). In particular, the Lystrosaurus and Cynognathus AZs of the Karoo have been fundamental to our understanding of the evolution of Early and Middle Triassic communities after the end-Permian mass extinction. More recently, studies on the Cynognathus AZ have permitted the recognition of three subzones based initially on temnospondyl amphibian taxa [110,125] and subsequently reinforced by other faunal elements [126]. For cynodonts, the oldest Subzone A includes Cynognathus, Langbergia, and indeterminate cynodonts with allotherian-like teeth [127] and has traditionally been considered upper Olenekian. Subzone B corresponds to the 'classical' Cynognathus AZ and besides Cynognathus and Diademodon also includes Trirachodon, Bolotridon and Lumkuia [32,109] and has traditionally been considered lower Anisian. Finally, in the youngest Subzone C [126,[128][129][130] the dominant cynodont is Cricodon [126], with rarer records of Cynognathus and Diademodon. This subzone has traditionally been considered upper Anisian. Thus, the Cynognathus AZ as a whole has been thought to span from the upper Olenekian (late Early Triassic in age) to the upper Anisian (mid Middle Triassic in age).
The discovery of the genus Aleodon in the Dinodontosaurus AZ of Brazil bears directly on these biostratigraphic questions. Aleodon brachyrhamphus comes from the Lifua Member of the Manda Beds in Tanzania [20][21] where it is found in association with gomphodont cynodonts: the trirachodontid Cricodon metabolus [20] and the traversodontids Scalenodon angustifrons [20], Mandagomphodon hirschsoni, and M. attridgei [120,136] (Fig 19). Recently, new stratigraphically controlled collections have subdivided the Lifua Member into lower and midupper sections, but these faunal associations hold: the higher section includes Aleodon, trirachodontids and traversodontids, whereas the lower includes Cynognathus (e.g., [112,137]).
In Namibia, in the upper Omingonde Formation, GSN EN-3 (the specimen here tentatively referred to Aleodon cromptoni; Fig 16) was reported exclusively associated with traversodontids (Luangwa sp. and an indeterminate form) and Chiniquodon sp. in the upper levels of the unit. Significantly, Aleodon in Namibia is not found in association with Cynognathus and Diademodon, because these taxa occur only in the middle and lower portions of the upper Omingonde Formation [13]. Abdala and Smith [13] suggested the distribution of these taxa indicates that the middle and lower levels can be correlated with the Cynognathus AZ of the Karoo Basin and that the upper level can be correlated with the upper portion of the Ntawere Formation in Zambia (based on the shared presence of the traversodontid Luangwa; [11,135]) and part of the Santa Maria Formation in Brazil (based on Luangwa sudamericana and Chiniquodon theotonicus; [12]). In addition, the presence of Stahleckeria potens [13] and a large rauisuchian [138] in the upper Omingonde Formation strengthens its relationships with the Dinodontosaurus AZ of southern Brazil.
The presence of Aleodon in Brazil permits a biostratigraphic correlation with the upper levels of the Manda Beds in Tanzania [112,137] and the upper Omingonde Formation in Namibia, and its association with the basal traversodontid Luangwa sudamericana extends this correlation to the upper horizons of the Ntawere Formation in Zambia (Fig 19), as was previously argued by Abdala and Sá-Teixeira [12].
The recognition of Luangwa sudamericana in the Triassic of Brazil, closely related to Luangwa drysdalli from Zambia, opened new queries about the temporal significance of this taxon [12] and its age and possible inclusion into the Dinodontosaurus AZ. This taxon suggested the occurrence of a hidden faunal association, older than the known typical fauna of the Dinodontosaurus AZ. Problematically, the precise location of the holotype specimen of L. sudamericana is unknown, but comes from an area between Vera Cruz and Candelária cities [12]. After that contribution, there were brief reports of new Luangwa material coming from the Dona Francisca region (Antonini Bortolin site; [156][157]), but these specimens remain unpublished. In addition, the specimen UFRGS-PV-0265-T (Fig 21; see description in S1 File) can be referred to Luangwa and it comes from the same outcrop as the holotype of Aleodon cromptoni. Another possible specimen deposited in the UFRGS-PV collection that can be referred to Luangwa is UFRGS-PV-0140-T, found in the Sanga Hintz (Bom Retiro region, Candelária), but this specimen needs further preparation to be certain of this referral. As previously mentioned, at least one specimen of A. cromptoni (UFRGS-PV-0146-T) comes from Sanga Pascual, located nearby Sanga Hintz.
Taking into consideration the new Brazilian Aleodon and the already known Luangwa sudamericana and Brasinorhynchus mariantensis, there appears to be evidence for an older faunal assemblage in the Pinheiros-Chiniquá Sequence, mixed within the traditional Dinodontosaurus AZ. However, there are several problems with this idea: (1) several records from traditional outcrops lack stratigraphic control; therefore, the stratigraphic and taxonomic characterizations of two possible subzones or zones become impossible. For example, the lack of this information precludes the recognition of two subzones or fossiliferous levels in the same outcrop.  The African cynodont Aleodon in the Triassic of South America prepared, hampering the recognition of diagnostic features. Traditionally, the presence of any Dinodontosaurus-like kannemeyeriiform remains was enough to place strata in the Dinodontosaurus AZ, but these specimens may not be robustly referable to Dinodontosaurus on a strict apomorphy basis. (5) The taxonomic referral of numerous Dinodontosaurus AZ specimens remains controversial, as should be evident from the Aleodon case detailed here.
The association of Aleodon and Luangwa, their relationship with other Gondwanan faunas (see Fig 19), and the stenaulorhynchine affinities of Brasinorhynchus are likely indicative of the occurrence of an older subzone within the Dinodontosaurus AZ, although this is admittedly based on a 'stage of evolution' argument. This older subzone can be characterized by the presence of Aleodon-Luangwa-Brasino rhynchus. However, with available stratigraphic data we cannot determine their precise temporal span. Langer et al. (Fig 2 in [52]) suggested that the Sanga Baum (Chiniquá region) and the Sanga Pascual and Bom Retiro area (see S1 File for references of these localities) could have a younger faunal association than the remaining Dinodontosaurus AZ localities (e.g., Cynodontier Sanga, Weg Sanga, Pinheiro region, Dona Francisca, Cortado sites) due to the presence, in the former, of the toothless dicynodont Stahleckeria potens. However, due to the aforementioned problems this hypothesis is difficult to test.
For example, in the Pinheiro region, without precise biostratigraphic control, Aleodon is associated with Dinodontosaurus, Massetognathus, Chiniquodon, and Prestosuchus (see S1 File). The African cynodont Aleodon in the Triassic of South America Therefore, for this locality the possibilities are: (1) both subzones are represented and the fossil sample is mixed; (2) the biochron of the taxa is not strictly bounded and the Dinodontosaurus AZ spans from the Ladinian to the early Carnian with a succession of forms that do not necessarily define two clearly distinctive subzones.
In Chiniquá region, there is no record of Aleodon, Luangwa, or Brasinorhynchus at the moment. Of note, in that region was found the doswelliid Archeopelta arborensis, which is phylogenetically closer to Tarjadia ruthae [161] from the lower member of the Chañares Formation (e.g., [44,162]).
In the area of Dona Francisca (Figs 1 and 20) there is a predominance of Dinodontosaurus, Massetognathus, and the rauisuchians Decuriasuchus and Prestosuchus [154,[163][164]. At present there is no record of Aleodon and there are mentions of possibly Chiniquodon and Luangwa based only on abstract meetings [156][157] and have not been confirmed by thorough description.
Outwardly the genus Dinodontosaurus seems to be present in most localities (e.g., [41,54,152]), although several specimens merit further taxonomic analysis. With the taxonomic base at hand, this suggests a lengthy temporal span of the genus. In fact, a critical aspect we noted is that during fieldwork prospections the presence of dicynodont elements (skulls and/or postcranial elements) is ambiguously thought as indicative of the Dinodontosaurus AZ.
The distribution of taxa in Table 3 visually suggests that the taxa of the Dinodontosaurus AZ are randomly distributed and that there is a predominance of taxa indicative of a presumably older age (i.e., Aleodon, Luangwa, and Brasinorhynchus) in the east portion of the Triassic outcrops. Although not conclusive, it would suggest that at these areas the two subzones appear in the same outcrops or region, and the lack of stratigraphic control of collected fossils biases the recognition of both fossil assemblages. The obtained dendrogram, based on the analysis of 11 localities versus 19 taxa (S1 File), shows two main clusters, with two main sub-clusters in each one (Fig 22). The outcrops of the Chiniquá region are clustered together with Rinção do Pinhal and Linha Várzea sites, by the absence of Aleodon and Luangwa. The position of Porto Mariante 1 on the dendrogram is due to its incomplete fossil record. Cortado site and Pinheiro region are clustered together, being more similar than Dona Francisca, Bom Retiro region, and Vale Verde, which are clustered together. Particularly, Cortado site and Pinheiro region have the higher number of taxa from the center and east regions. As aforementioned, the dissimilarity of the west and east regions is conspicuous, and localities from the central region seems to represent "transitional" faunas. This dissimilarity would be an artifact result of selective effort for paleontological collections and thus a bias in this analysis. However, if the Dinodontosaurus AZ includes in fact mixed faunal elements of two biozones (or subzones) the present analysis would be indicating that these are more likely to be represented in the central and east outcrops, due to the shared presence of Aleodon and Luangwa, two cynodonts that are supposed to indicate an older age. Moreover, the amount of unpublished, unprepared specimens, and the lack of accurate geographical and stratigraphical data are also issues that hamper a better resolution of the similarities among Triassic outcrops in Rio Grande do Sul. In addition, the Triassic rocks in the center of the state of Rio Grande do Sul have different thickness along their east-west distribution. The younger portions of the Triassic package and its greater thickness are located in the central portion of the state [165]. To the east and west, the Triassic package thins out and the basal most portions (i.e., oldest rocks) of the Santa Maria Supersequence are exposed. Consequently, there are several outcrops of the Dinodontosaurus AZ on the east and west sides (see Figs 1 and 22 and Table 3); however, in the west side, the basal most portions are partially covered by the Guará Formation (e.g., [42,165]).
Dinodontosaurus and Santacruzodon AZs. The Santacruzodon AZ was recently included within the Santa Cruz Sequence (see historical and geological review in [35]), being considered "intermediate" between the Dinodontosaurus (older) and the Hyperodapedon (younger) AZs. Representative outcrops of this AZ are geographically confined to the east portion of the Triassic rocks, in the municipalities of Venâncio Aires, Santa Cruz do Sul, and Vera Cruz [35,98]. The fossil content of the Santacruzodon AZ is limited when compared to other assemblages from the Santa Maria Supersequence. This is represented by the traversodontids Santacruzodon hopsoni [93], the most common form, Menadon besairiei [18], and Massetognathus sp. [98], the probainognathians Chiniquodon sp. [98] and Santacruzgnathus abdalai [3], the archosauriform Chanaresuchus bonapartei [166], the rauisuchian Dagasuchus santacruzensis [167], and few unpublished fragmentary cranial remains of tusked dicynodonts, compared in morphology to Dinodontosaurus (Pers. Obs.).
The shared presence of Massetognathus sp., Chiniquodon sp., and Chanaresuchus; the close phylogenetic relationship of Santacruzodon with Massetognathus [120,168] and of Dagasuchus with Prestosuchus [167]; and the presence of Dinodontosaurus-like specimens are significant points that do not clearly demonstrate the distinctiveness of the Santacruzodon AZ compared to the Dinodontosaurus AZ (Fig 23). In addition, the traversodontid Menadon, apparently restricted to the Santacruzodon AZ in Brazil (and also recorded in Madagascar), also has a close phylogenetic relationship with Protuberum, from the Dinodontosaurus AZs, being positioned basal to the former taxon in some phylogenetic hypotheses (e.g., [120,168]). To this scheme, the age obtained for the Santacruzodon AZ as early Carnian (236±1.5 Ma; [43]) is the same that the typical faunal association of the lower section of the Chañares Formation [44,45]. However, traditional correlations have been done directly between the Dinodontosaurus AZ with the Chañares Formation (e.g., [51,143]). The taxa recovered in both Brazilian AZs and the Chañares Formation have an interdigited pattern and only further detailed taxonomic studies and stratigraphically controlled specimens, mainly from Brazil, will elucidate these issues summarized in Fig 23. As such, the standard idea that the Santacruzodon AZ (and the Santa Cruz Sequence as a whole) is a clearly distinctive tetrapod assemblage from the Dinodontosaurus AZ requires reevaluation with new paleontological data. The African cynodont Aleodon in the Triassic of South America

Conclusions
We describe a new species of Aleodon, A. cromptoni, based on several specimens from the Dinodontosaurus AZ of the Pinheiros-Chiniquá Sequence, Santa Maria Supersequence of Rio Grande do Sul, Brazil. One specimen, GSN EN-3, from the upper Omingonde Formation of Namibia is also tentatively referred to this species. Aleodon cromptoni is clearly differentiated from A. brachyrhamphus from the Manda Beds of Tanzania [20] on the basis of its upper and lower postcanine dentition, and includes individuals that reached larger skull sizes (up to 31cm skull length) than the known sample of the Tanzanian species. Inclusion of this taxon in a phylogenetic analysis provides further evidence for close relationship between Aleodon and Chiniquodon, as originally proposed by Hopson and Kitching [25], and we include both taxa in the family Chiniquodontidae. Both genera share a very similar cranial morphology despite conspicuous differences in their dentition. The recognition of Aleodon in Brazil, hidden in most collections with cynodonts from the Dinodontosaurus AZ, and the comparisons with the known sample of Chiniquodon from Argentina and Brazil brings to light taxonomic problems with the latter genus that require additional study. The presence of Aleodon cromptoni in the Dinodontosaurus AZ reinforces its similarity to the upper Omingonde Formation of Namibia, which also share the dicynodont Stahleckeria potens [46] and the cynodonts Luangwa sp. and Chiniquodon sp. [13], making this one of the most strongly-correlated Gondwanan Triassic faunal associations. Biostratigraphic issues in South America are still controversial and further taxonomic refinements and new radiometric and geological data are required to resolve the biostratigraphy for the Middle-Late Triassic of Gondwana.