The skull is not very well preserved, and is missing most of the region anterior to the frontal, much of the skull roof posterior to the frontal, the left cheek area, and most of the palate. Most of the mandible is preserved, though it is occluded to the skull.

The skull is sub-triangular in outline, and this is best indicated by the mostly intact mandible. The skull is approximately 2.5 cm in length. This is considerably smaller than the largest owenettid, Candelaria barbouri (Cisneros et al., 2004 and Price, 1947), though this smaller size is more in line with the rest of the owenettids. The orbits are large, primarily due to the posterior extension, and extend over half of the skull length. The interorbital portion of the skull roof is thin.

Only isolated portions of the premaxillae are preserved, and only the bases of some of the premaxillary teeth (Fig. 3A). The premaxillary teeth are much larger than the cheek (maxillary) teeth in cross section, and retain the labio-lingually expanded bases (Fig. 3C).

Neither of the maxillae is completely preserved, but a portion of the right can be seen in lateral view. The ventralmost portion of the dorsal process of the maxilla is visible in left lateral view (Fig. 2C), but not enough is preserved to determine whether the animal possessed the anterolateral maxillary foramen present in most parareptiles (Laurin and Reisz, 1995). As in ‘Owenetta’ kitchingorum and Candelaria, the suborbital ramus of the maxilla is short, only extending halfway under the orbit. It has been suggested that a shorter maxilla is a feature of Triassic owenettids, whereas Permian forms have a longer suborbital ramus that consequently contains a larger number of teeth (Cisneros et al., 2004). In CAMZM T997, the right maxilla has a minimum of 18 alveoli with space for at least four more (Fig. 3A). Some owenettids that branch off closer to the root of Owenettidae such as Owenetta rubidgei have as many as 30 maxillary teeth (Reisz and Scott, 2002) and the number decreases in taxa that branch off closer to the crown, with all known owenettids having more than 18 (Cisneros, 2008b). Amongst procolophonids, Coletta seca and Kitchingnathus untabeni have a maximum of 15-16 (Cisneros, 2008a and Gow, 2000), with more crownward taxa having as few as 6 (Cisneros, 2008b). See below for a more complete description of the dentition.

The lacrimal is visible on the right side (Fig. 2C). In lateral view it is tall and plate-like; however, the anterior and posterior process are not preserved, so its relationship to the naris anteriorly and the contact with the jugal posteriorly cannot be described.

Though the pre- and post-frontals are not completely preserved, it is evident that a lateral process of the frontal forms a large portion of the dorsal orbital margin (Fig. 2A, C). Portions of the right prefrontal can be seen, particularly in lateral view (Fig. 2A, C). The ventral process of the prefrontal can be seen in the anterior portion of the orbit, extending interior and slightly posterior to the posterior edge of the lacrimal. Although a portion of the prefrontal can be seen on the dorsal skull roof, it is not preserved well enough to determine the medial extent of the element. Though not completely preserved, there appears to be an orbitotemporal crest (sensu Cisneros et al., 2004) that is formed by the lateral margins of the prefrontal, frontal, and postfrontal (Fig. 2A). This feature is present in other owenettids and Nyctiphruretus acudens (Saila, 2010), as well as some procolophonids (Cisneros, 2008b). The existence of this feature was asserted for ‘nycteroleters’; however, in specimens of Macroleter poezicus this region is periodically upraised by a series of bosses rather than consisting of consistent, smooth crest (Tsuji, 2006).

The temporal region is not preserved in its entirety, and the glue binding the fragments of matrix could easily be mistaken for bone. It is easy to see why the fossil was provisionally called a diapsid given the state of the temporal region. The CT data is particularly helpful in this area (Fig. 2 and Fig. 3). The right postorbital is preserved almost in its entirety, and the posteroventral edge is emarginated (Fig. 2A), but it is not clear whether this emargination is for the reception of the jugal, in which case the postorbital would have been displaced significantly posteriorly, or whether it represents the anterior and dorsal margin of a potential upper temporal fenestra. The lack of skull elements posterior to this feature, both in the left and right sides makes it impossible to ascertain whether there was temporal fenestra in Ruhuhuaria based solely on CAMZM T997. Temporal fenestration is variable in procolophonoids, but in addition to a ventral temporal emargination, a large fenestra is present in Candelaria (Cisneros et al., 2004) and temporal fenestration is highly variable in parareptiles in general (Cisneros et al., 2004 and Tsuji, 2006). A portion of the jugal is present and visible in right lateral view (Figs. 2A, C). Although the jugal is somewhat obscured by the ventral process of the postorbital that has shifted downwards, it is possible to make out what looks like a finished posterior edge of the jugal, indicating that there was a ventral emargination of the postorbital region. The angle of the posterior edge of the jugal also indicates that the emargination would have been acute rather than broad as it is in procolophonids such as Procolophon trigoniceps (Carroll and Lindsay, 1985).

The mandible is partially preserved with the complete tooth row of the right dentary visible when segmented from the skull (Fig. 3B). The dentary contains a minimum of 24 tooth positions, which, like the maxillary teeth, have labio-lingually expanded bases (Fig. 2 and Fig. 3). Portions of the splenial can be seen but the nature of the symphysis cannot be determined. A small fragment of the angular is also visible along with parts of the surangular and what is likely a portion of the prearticular (Fig. 2B).

Tooth morphology is one of the more distinctive features of procolophonid anatomy. The tooth row of CAMZM T997 is flush with the maxillary surface (Fig. 2C), as it is in procolophonids that branch off closer to the root of Procolophonidae (such as Coletta, Pintosaurus magnidentis, and Sauropareion anoplus (Cisneros, 2008b)). As stated above, only 18 teeth and/or alveoli are preserved on the right maxilla; however, it is clear from the CT scan that the bases of these teeth are labio-lingually expanded, as much as two times as wide labio-lingually as mesio-distally (Figs. 3A, C). The largest and anteriormost teeth, both in the dentary and mandible, show this morphology most clearly, whereas the more posterior teeth have less expanded bases. The transverse expansion of the base of the maxillary teeth is a recognized synapomorphy of Procolophonidae (Cisneros, 2008b). The presence of this trait in Ruhuhuaria is therefore unexpected given its owenettid-like appearance. The dentary teeth are likewise expanded and seem to increase in size anteriorly (Fig. 3C), a feature that has not been observed in other taxa and is therefore an autapomorphy of Ruhuhuaria. No complete crowns are preserved; however, the teeth appear to bear single cusps, and they are simple, conical, and not recurved (Fig. 2). There is also no evidence of any infolding of the dentine (Fig. 3C).

There is little evidence in owenettids of lateral expansion of the cheek teeth, but rarely are the skulls preserved in a way that the base of the teeth is visible, often with the mandible occluded to the skull (i.e. BP/1/4105, Reisz and Scott, 2002); thus, it must be noted that the shape of the tooth bases are not known for all owenettid taxa. Indeed, it is only the CT scan that permitted us to recognize the lateral expansion of the base of the teeth in Ruhuhuaria.

At first glance this fossil is ambiguous in appearance. CAMZM T997 was tentatively identified as a diapsid reptile on the specimen label in the collection, and indeed, it is obvious that this small skull likely belongs to a small reptile. An assignment to a known taxon is difficult, but close examination has revealed four aspects of the morphology that can be identified as characters of a known clade, and through this an assignment can be made: a)

lateral expansion of the base of the teeth. As shown by the CT scan, all of the teeth, particularly those more anterior, are labio-lingually expanded at their base while remaining peg-like and sub circular in cross section for the remainder of their length with the anteriormost teeth the most expanded;

Based on the above determination as a procolophonoid, a phylogenetic analysis of this clade was conducted in order to determine the relationships of the new taxon. The matrix was modified from Cisneros et al. (2004), which was in turn based on that of Modesto et al. (2003). These are the only matrices that scored every known owenettid species separately. Other taxa were added to the matrix: the procolophonoid Kitchingnathus was included, and Procolophon was considered monospecific, as per Cisneros (2008c). In total, 10 ingroup taxa were included, with Macroleter poezicus as the outgroup (Character List, Appendix A; Character Matrix, Appendix B). All characters were weighted equally. One character, #21, maxillary tooth count, was ordered as it was in Cisneros (2008b), and Character #16 of Cisneros et al. (2004) was split into two separate characters in this analysis. Due to the incompleteness of the fossil, only six out of 21 characters could be scored for Ruhuhuaria. The phylogenetic analysis was completed using TNT Version 1.1, downloaded January 2013 (Goloboff et al., 2008), using an exhaustive search, which examines all possible combinations. Tree collapsing was set to max. length = 0. In addition, a bootstrap resampling analysis of 5000 replicates was run and the decay indices were calculated, all using TNT.

Five most parsimonious trees were recovered, consisting of 34 steps (strict consensus; Fig. 4). The strict consensus tree has a Consistency Index (CI) of 0.7353, and a Retention Index (RI) of 0.7955. Owenettidae continues to be monophyletic but the clade is not supported by more than 50% of the boostrap replicates, and the relationships within the clade are not well resolved. As in previous analyses (Cisneros et al., 2004 and Modesto et al., 2003), the two Owenetta species are not monophyletic, with O. rubidgei the sister taxon to Barasaurus and ‘O.’ kitchingorum forming a sister group relationship with Candelaria. Thus, ‘O.’ kitchingorum continues to be only provisionally assigned to the genus. The remaining members of the Owenettidae including Ruhuhuaria and Saurodektes fall within an unresolved polytomy (Fig. 4). The analysis did not support the clade of Saurodektes + (O. kitchingorum + Candelaria) of Cisneros et al. (2004), with Saurodektes instead falling within an unresolved polytomy with (O. kitchingorum + Candelaria), Ruhuhuaria, and (O. rubidgei + Barasaurus). Procolophonidae is monophyletic and well supported (75% of bootstrap replicates and decay index of 2).

The retention of a monophyletic Owenettidae is somewhat surprising given that Ruhuhuaria possesses a suite of characters that appears to be transitional between owenettids and procolophonids such as the expanded bases of the maxillary teeth (generally a procolophonid character), and a relatively high tooth count (an owenettid character). Owenettidae is supported by 4 unambiguous synapomorphies: #9, supratemporal lateral margin straight (0 => 1); #14, entepicondylar foramen present (0 => 1); #18, presence of orbitotemporal crests (0 => 1); and #20, posteriorly-expanded naris (0 => 1). Of these, Ruhuhuaria can only be scored for character #18.

The identification of this taxon as a procolophonoid has interesting implications for the evolution of procolophonians (pareiasauromorphs + procolophonoids). Owenettidae contains six species–Barasaurus besairei, Candelaria barbouri, ‘Owenetta’ kitchingorum, Owenetta rubidgei, and Saurodektes rogersorum (Cisneros et al., 2004, Modesto et al., 2003 and Modesto et al., 2004), and now Ruhuhuaria reiszi. O. rubidgei is exclusively Late Permian, present in the Cistecephalus and Dicynodon Assemblage Zones of the South African Karoo Basin (Broom, 1939, Cisneros et al., 2004 and Modesto and Damiani, 2007). Barasaurus besairei is found in the Sakamena Group of Madagascar, and fossils have been found in both Permian and Triassic Formations within this Group, indicating that the species survived the end-Permian mass extinction event (Ketchum and Barrett, 2004). Apart from Ruhuhuaria, there is only one other Middle Triassic owenettid: Candelaria barbouri, from the Ladinian of Brazil (Cisneros et al., 2004 and Price, 1947). The discovery of Ruhuhuaria provides additional evidence of the existence of Middle Triassic owenettids as predicted by Cisneros and colleagues (2004), and that consequently, two separate clades of procolophonians (owenettids and procolophonids) co-existed well into the Triassic.

Ruhuhuaria is the first Triassic parareptile described from Tanzania, but its presence here is not unexpected given the distribution and stratigraphic occurrence of its relatives. Procolophonoids are fairly speciose, found in similarly aged faunas of South Africa, and are also found in the Triassic of Europe, North and South America, Antarctica, Australia (Bartholomai, 1979), and China (Cisneros, 2008b). Recent fieldwork in Tanzania has also produced what is a yet-to-be described procolophonid from the Manda Formation (C. Sidor, pers. comm.), indicating that in addition to a remarkable diversity of archosauriforms, the Lifua Member of the Manda Formation may yet reveal a number of additional parareptile taxa.