As originally described by Cabrera (1947), the tooth crowns of Amygdalodon are spatulate, being mesiodistally broad, buccolingually compressed, and their apex is slightly curved lingually (Fig. 1). The mesial and distal margins are asymmetrical (Rauhut, 2003), the mesial margin being more convex and the distal margin more straight (and with a small bulge at the base of the crown in some teeth). This asymmetric profile of the tooth crowns seems to be present in most (if not all) sauropod taxa with broad spatulated tooth crowns (e.g., Tazoudasaurus [Allain and Aquesbi, 2008], Shunosaurus [Chatterjee and Zheng, 2002], Camarasaurus [Madsen et al., 1995], Astrodon [Carpenter and Tidwell, 2005]). The crowns are relatively low in comparison to their maximum mesiodistal width, having an SI index (Upchurch, 1998) that varies between 1.34 and 1.50. These values are usually higher in Eusauropoda (as well as in more basal saurpodomorphs), so that Amygdalodon teeth are comparatively lower and broader than in most sauropodomorph taxa. The breadth of Amygdalodon crowns is also remarkable in comparison with eusauropods when the root width is taken into account, as the maximal crown mesiodistal width is 130–150% of the maximum root mesiodistal width (resembling the condition of the basal gravisaurian Tazoudasaurus). The crown therefore abruptly expands mesiodistally above with crown-root limit (Fig. 1). All teeth have a D-shaped cross-section with their buccal surface is mesiodistally convex and their lingual surface relatively flat (except for MLP 46-VIII-21-1/13, see Fig. 1 and below). Based on these features the teeth of Amygdalodon clearly fall within the broad-crowned (BC) category as defined by Barrett and Upchurch (2005).

There are two remarkable features of Amygdalodon dentition in comparison with those of most eusauropods with spatulate dentition. The first one is the absence of denticles on the distal and mesial margins (Fig. 1). This feature, noted by previous authors (Cabrera, 1947 and Rauhut, 2003) seems to be rare among non-neosauropods and is only shared with the teeth of Shunosaurus (Chatterjee and Zheng, 2002) and with those referred to Kotasaurus (Yadagiri, 2001). The second feature is the presence of a particular wrinkling pattern on the enamel outer surface. Although there is some variation in the wrinkling pattern among the preserved teeth (see below), the lingual and buccal enamel surface of all Amygdalodon teeth share the presence of apicobasally aligned pits, which are subovoid, deep, and relatively small (their major diameter ranges between 100 and 200 μm; Fig. 2). Each set of apicobasally-aligned pits are usually joined to each other by a continuous narrow sulcus (Fig. 2). Besides the presence of these pits and sulci, the buccal and lingual enamel surface of Amygdalodon teeth is mostly smooth. This pattern contrasts with the generalized condition of eusauropods and related forms (Allain and Aquesbi, 2008 and Wilson and Sereno, 1998) that have a wrinkled enamel surface with much more numerous and well developed anastomized grooves and ridges. Thus, to our knowledge, the particular pattern of Amygdalodon (with well delimited pits joined by apicobasal sulci) seems to be different from all gravisaurians. Non-gravisaurian sauropodomorphs also differ from the condition of Amygdalodon, having either a smooth enamel surface (Wilson and Sereno, 1998) or a pattern of fine irregular wrinkles (Yates, 2004).

As noted by previous authors (Cabrera, 1947 and Rauhut, 2003), the crowns of Amygdalodon have deep major grooves that extend along the mesial and distal margins on the buccal and lingual surfaces (Fig. 1). There is variation in the development of these grooves (see below) but all crowns have, at least, a well-developed distal buccal groove. The enamel of Amygdalodon becomes progressively thinner and gradually disappears at the crown-root limit, in contrast with the condition of several basal eusauropods in which the enamel ends abruptly at the crown-root limit (e.g., Patagosaurus MACN-CH 2008, BMNH 3377 [Barrett, 2006]).

The three most complete crowns of Amygdalodon have wear facets that only extend over one of the crown's edges. In the upper teeth the wear facets extends from the apex along the mesial edge (mostly on their lingual surface; Fig. 1A–B), whereas in the lower teeth the wear facets extends from the apex along the distal edge (mostly on their buccal surface; Fig. 1G). The most damaged tooth crown (MLP 46-VIII-21-1/17; Fig. 1D), however, has a horizontal wear facet at its apical region, suggesting that Amygdalodon may display the V-shaped wear facet (sensu Wilson and Sereno (1998)) in extensively worn teeth. Thus, the presence of wear facets on only one margin may simply reflect an early stage of tooth wear in the other three teeth (or a different position in the toothrow). Irrespective of the underlying cause for this difference, the wear facets in Amygdalodon are unusually asymmetrical in comparison with those of eusauropods with broad spoon-shaped crowns.

Two of the preserved elements (MLP 46-VIII-21-1/12 and MLP 46-VIII-21-1/15) are interpreted as upper teeth because both have wear facets that extends over the lingual surface, as commonly found in maxillary teeth of eusauropods (e.g., Shunosaurus [Chatterjee and Zheng, 2002]). One of the teeth (MLP 46-VIII-21-1/12) is more extensively worn (exposing a broad surface of dentine) than the other, but both share the location, shape, and orientation of the wear facet. This facet is ovoid shaped, having its major axis directed from the crown's apex to the mesial edge (Fig. 3). The flat surface of the wear facets of these teeth bears numerous striae but no pits (Fig. 3B).

These two teeth, however, differ in several aspects that suggest they may belong to different sections of the toothrow. The crown of MLP 46-VIII-21-1/15 is apicobasally higher and more symmetrical (i.e., lacking the basal bulge on the distal margin) than that of MLP 46-VIII-21-1/12 (or any other teeth). This suggests the former may be more anteriorly located than the latter, as in most eusauropods the tooth size decreases continuously along the toothrow (e.g., Shunosaurus [Chatterjee and Zheng, 2002], Omeisaurus [Feng et al., 2001], Mamenchisaurus [Russell and Zheng, 1993]); a trend that is also present in non-eusauropod sauropodomorphs (e.g., Melanorosaurus [Yates, 2007], Mussaurus [Pol and Powell, 2007a], Massospondylus [Sues et al., 2004]).

The crown of MLP 46-VIII-21-1/15 also differs from the other crowns of Amygdalodon in several characters. This crown shows two well-developed grooves: the distal buccal groove present in all preserved Amygdalodon crowns plus a mesial buccal groove (Fig. 1E). In addition to these two grooves, the enamel buccal surface is relatively smooth. The lingual surface of MLP 46-VIII-21-1/15 lacks well-developed grooves but bears two major sulci (Fig. 3A) and several minor sulci between them. As in the other crowns these sulci join apicobasally-aligned pits.

The crown of MLP 46-VIII-21-1/12 has a single (distal) buccal groove (Fig. 1F) and a unique but well-developed (mesial) lingual groove (Fig. 1B). The morphology of these grooves differs from the ones previously described for MLP 46-VIII-21-1/15 due to the presence of a minor groove that runs parallel to the main (deeper) groove, enclosing a slightly convex area between them (Fig. 4). Both buccal and lingual surfaces have the wrinkling pattern of pits and sulci, although a different wrinkled pattern is observed close the basal region of the distal and mesial edges (Fig. 4). The sulci in these basal areas of the crown are not apicobasally directed and form an anastomized network without well-delimited pits (see below in MLP 46-VIII-21-1/13), resembling in these restricted areas the enamel wrinkling pattern of eusauropods. In addition to the apical wear facet, there are two small areas at the base of this crown in which the enamel is worn. These surfaces are flat and lack striation (Fig. 4C). The basal lingual worn surface is located near the distal edge, whereas the basal labial worn surface is close to the mesial edge. The flat morphology and the position of these surfaces suggest they were produced by tooth-tooth contact between adjacent crowns, rather than by abrasion. Similar basal worn surfaces are present at least in some eusauropod taxa (e.g., cf. Patagosaurus MACN-CH 2008, MPEF-PV 3060, MPEF-PV 3055; Camarasaurus [Wilson and Sereno, 1998: Fig. 10]) in which the tooth crowns are disposed in an en-echelon tooth arrangement. Although not all taxa with an en-echelon tooth arrangement have these worn surfaces, we interpret their presence in Amygdalodon as an indication that the mesial margin of one crown was (buccally) overlapped by the distal margin of the preceding tooth, creating an en-echelon tooth arrangement for Amygdalodon, a character that has been considered as synapomorphic of Eusauropoda (Wilson, 2002) or Gravisauria (Allain and Aquesbi, 2008).

The tooth MLP 46-VIII-21-1/13 is interpreted as a dentary element because the location of the wear facet is opposite to those of the upper teeth. First, the wear facet extends mostly over the buccal surface, as noted by Rauhut (Rauhut, 2003). Second, the facet extends from the tooth apex toward the base of the crown along its distal margin (Fig. 1G). This tooth is the most extensively worn and has a broad surface of dentine exposed. The worn surface of this tooth is actually divided in two different areas of the distal margin, divided by an abrupt step. The apical area of the wear facet faces apicodistally whereas the basal area of the wear facet is vertically oriented and faces distally (Fig. 5 A–B). The apical area is flat and much broader than the basal one and extends mostly on the buccal surface of the crown forming an angle of approximately 60 degrees with the transversal plane (Fig. 5). In this region the apical wear facet also extends onto the lingual surface of the crown, forming a small and slightly convex surface that contrasts with the flat and extensive surface that extends on the buccal side of the apical region of the crown. These differences suggest the reduced lingual surface was produced by abrasion whereas the extensive and flat worn surface on the buccal side was produced by tooth-tooth occlusion. The basal area of the wear facet is much narrower (buccolingually) than the apical area. The worn surface is flat and faces distally, resembling the condition of the wear facets described for Camarasaurus (Calvo, 1994). Numerous striae and pits are observed on the apical and basal region of the wear facet.

Two deep and well-developed grooves are present on the buccal surface of this tooth (as in MLP 46-VIII-21-1/15), although much of the distal groove has been worn (Fig. 5). Below the basal end of the mesial buccal groove the crown has a large rounded depression, which may be a pathological feature of this particular tooth (Fig. 5B). The lingual surface bears a deep groove on its mesial side, which is deeper near the base of the crown and becomes progressively shallower and broader toward the apex. At the apicobasal midpoint of the crown, this groove bifurcates in two sulci that join closely spaced and well-delimited pits (Fig. 1C, 2A). This wrinkling pattern of pits and sulci is particularly well developed on the lingual surface of this crown (Fig. 1C), whereas the buccal surface is smoother, probably due to abrasive wear (except on its basal most region; Fig. 5B). As was noted for MLP 46-VIII-21-1/12, the mesial surface bears an anastomized wrinkling pattern near the base of the crown (Fig. 5C). This pattern is also present in the distal edge of this tooth, although much of this region has been worn. This anastomized wrinkling pattern resembles the condition of most basal eusauropod teeth (e.g., Chebsaurus [Mahammed et al., 2005], Patagosaurus [MACN-CH 2008], ‘Cetiosaurus’ [Barrett, 2006]; see Discussion).

An unusual characteristic of this crown is that the lingual surface is slightly convex mesiodistally, a feature noted by Rauhut (2003) to resemble the crown morphology of Cardiodon (Upchurch and Martin, 2003). This morphology contrasts with the generalized spoon shaped crowns of most basal eusauropods that have a concave lingual surface. The spoon-shaped morphology (with a concave lingual surface) of the other teeth described above shows this feature may actually vary along the toothrow in A. patagonicus.

Up to date no published cladistic analysis of Sauropodomorpha has included A. patagonicus, although this taxon has been regarded as a member of Eusauropoda by Wilson (2002), Rauhut (2003), and Upchurch et al. (2004). Wilson (2002) considered this taxon as part of the clade formed by Patagosaurus and more derived eusauropods based on the presence of cervical ribs positioned ventrolateral to the centrum (inferred from the position of the parapophysis). Rauhut (2003) noted, however, that this feature might have been artificially produced by preservational causes. Rauhut (2003) also argued for the eusauropod affinities of Amygdalodon based on the presence of features then considered as eusauropod synapomorphies (e.g., high neural arch in the dorsal vertebra, cervical vertebra opisthocoelous and with centrodiapophyseal laminae), but he suggested a basal position within this clade given the absence of some derived characters present in most eusauropods (e.g., dorsal neural arches with deep depressions on cranial surface and internal cavities above neural canal, cervical vertebra with low neural arch, lacking depression above parapophysis, and lacking pleurocoels).

The morphology of the Amygdalodon tooth (MLP 46-VIII-21-1/13) studied by Rauhut (2003) also prompted this author to refer this taxon to Eusauropoda, based on its spatulate morphology and the presence of wrinkled enamel, buccal grooves, and V-shaped wear facets. In order to discuss the significance of the dental anatomy of Amygdalodon described here, we have tested its position through a cladistic analysis based on the data matrix of Wilson (2002), increasing the taxon and character sampling for non-eusauropod sauropodomorphs based on recent contributions (Allain and Aquesbi, 2008, Upchurch, 1998, Upchurch et al., 2004, Upchurch et al., 2007b and Yates and Kitching, 2003) (Appendix A). The phylogenetic analysis was conducted using equally weighted parsimony analysis through a heuristic search in TNT (Goloboff, 2008a and Goloboff et al., 2008b) which retrieved 23 most parsimonious trees of 473 steps (CI = 0.58; RI = 0.79); the abbreviated strict consensus of which is shown in Fig. 6 (Appendix A). This analysis depicts Amygdalodon as a non-eusauropod sauropod (Fig. 6), forming a polytomy with Gongxianosaurus, Isanosaurus, and the clade formed by Vulcanodontidae and Eusauropoda (Gravisauria sensu Allain and Aquesbi [2008]) The present analysis, however, depicts Amygdalodon as more derived than Chinshakiangosaurus or the Lessemsaurus + Antetonitrus clade (Fig. 6; Appendix A). Amygdalodon, as well as other basal sauropods such as Chinshakiangosaurus and Isanosaurus, are currently known from fragmentary material (with more than 90% of missing data). Their incompleteness creates some degree of uncertainty regarding their phylogenetic placement and consequently the basal nodes of Sauropoda and Eusauropoda have low support values (e.g., with only one extra step these three fragmentary forms can be placed in alternative positions within the basalmost nodes of Gravisauria and Eusauropoda). Nevertheless, given the available information the more parsimonious hypotheses provide a phylogenetic placement for these taxa as eusauropod outgroups, which bears implications for the early evolution of some dental features of Sauropoda that are commented in the following section.

Within this phylogenetic context, the total absence of mesial and distal denticles in Amygdalodon is shared with some other basal sauropods (e.g., Gongxianosaurus [He et al., 1998], Shunosaurus [Chatterjee and Zheng, 2002]), but differs from the serrated condition of other non-neosauropod sauropods (e.g., Tazoudasaurus [Allain and Aquesbi, 2008], Patagosaurus MPEF-PV 1670). The optimisation of this character in the present phylogenetic analysis shows homoplastic transformations implying either the loss of denticles in Amygdalodon and more derived sauropods (with reversals to the serrated condition in Tazoudasaurus and some basal eusauropods such as Patagosaurus) or the independent loss of denticles in Amygdalodon, Gongxianosaurus, Shunosaurus, and Neosauropoda. Irrespective of these alternative scenarios, the available evidence suggests that this feature was a labile character during the early phases of sauropod evolution and lacks a clear phylogenetic signal, as recently noted by other authors (Barrett and Upchurch, 2005, Buffetaut, 2005 and Upchurch et al., 2007a).

The peculiar tooth morphology of Amygdalodon and the phylogenetic position retrieved for this taxon within Sauropoda has implications for the early evolution of several dental characters in basal sauropods. The most remarkable of these characters is the presence of tooth-tooth occlusion and extensive wear facets. Amygdalodon is one of the oldest sauropod taxa in which extensive wear facets are known (together with Tazoudasaurus [Allain and Aquesbi, 2008]). This feature has previously been interpreted either as synapomorphic of Eusauropoda (Upchurch et al., 2004, Upchurch et al., 2007a and Wilson, 2002) or as an ambiguous synapomorphy of a more inclusive clade (Allain and Aquesbi, 2008): Gravisauria (Vulcanodontidae + Eusauropoda). The basal position retrieved for Amygdalodon in our analysis and the presence of these extensive wear facets imply these features (potentially indicating an increase in oral processing of plants (Barrett and Upchurch, 2005 and Upchurch and Barret, 2000)) appeared earlier than previously thought in the evolutionary history of Sauropoda, diagnosing a more inclusive clade composed (at least) by Amygdalodon and more derived sauropods. Interestingly, the wear facets of Amygdalodon are highly asymmetrical and extend almost exclusively on the mesial edge of the upper teeth and on the distal edge of the lower tooth. This differs from the characteristic wear facets of basal eusauropods (and Tazoudasaurus [Allain and Aquesbi, 2008]) that have a more symmetrical V-shaped facet produced by interlocking tooth-tooth occlusion, which extend from the apex along both the mesial and distal margins (Calvo, 1994, Chatterjee and Zheng, 2002 and Wilson and Sereno, 1998). The presence of asymmetrical wear facets in Amygdalodon may imply an early stage in the evolution of sauropod jaw mechanics, before the onset of the extensive interlocking tooth-tooth occlusion of eusauropods.

This change in the optimisation of the presence of extensive wear facets in sauropods is paralleled by the case of other dental characters (e.g., crown cross-section, enamel wrinkling, labial grooves, en-echelon arrangement) originally regarded as eusauropod synapomorphies (Upchurch et al., 2004, Upchurch et al., 2007a and Wilson, 2002), but recently reinterpreted as synapomorphies of more inclusive clades of Sauropodomorpha (Allain and Aquesbi, 2008, Upchurch et al., 2007b, Yates, 2004 and Yates, 2007). Their presence in Amygdalodon confirms in this study the same optimisation pattern, although their distribution deserve some comments.

Tooth crown shape. All the teeth of Amygdalodon are greatly expanded mesiodistally with respect to the root and three of the four teeth of Amygdalodon have a concave lingual surface (i.e., creating a D-shaped cross-section of the crown). The former character represents the strongly spatulate condition of sauropod teeth (Upchurch, 1998 and Yates and Kitching, 2003) and is currently optimised as appearing in Amygdalodon and more derived sauropods. The crowns of more basal sauropodomorphs (including Chinshakiangosaurus [Upchurch et al., 2007b]), instead, are lanceolate and not greatly expanded relative to the root. On the other hand, the strongly convex buccal surface and slightly concave lingual surface of Amygdalodon creates a D-shaped cross-sectional shape of the crown, a character shared with Gongxianosaurus (He et al., 1998), Tazoudasaurus (Allain and Aquesbi, 2008), and eusauropods (Wilson, 2002). As noted by Upchurch et al. (2007b), this feature was previously considered synapomorphic of Eusauropoda but may have appeared gradually starting with an incipient concavity in Chinshakiangosaurus (at the base of Sauropoda) and reaching the full D-shaped cross-section in Amygdalodon and more derived sauropods. It must be noted that, as described above (and as was noted [Rauhut, 2003]), one of the teeth of Amygdalodon has a slightly convex lingual surface. This implies that, at least in basal sauropods, this character can vary along the toothrow and therefore its use for taxonomic purposes should be taken with caution when is applied to fragmentary material.

Buccal grooves. The presence of buccal grooves along the mesial and distal edges was previously regarded as a synapomorphy of Eusauropoda, and was recorded in Shunosaurus and more derived eusauropods (Upchurch, 1998 and Upchurch et al., 2004). Recent studies on the basal sauropod Chinshakiangosaurus showed that this taxon has only a distal groove, suggesting that this feature appeared earlier than the mesial groove (Upchurch et al., 2007b). The presence of both grooves in the more derived Amygdalodon expands the known distribution of the mesial groove and lends support to the step-wise appearance of these two grooves along the early evolutionary history of Sauropoda (before Eusauropoda). It is interesting to note that the absence of a mesial groove in at least one tooth of Amygdalodon may imply that this feature did not simultaneously appear along the entire toothrow.

Enamel wrinkling. As noted by Rauhut (2003) Amygdalodon shares with eusauropods the presence of wrinkled enamel outer surface, a character first considered as synapomorphic of Eusauropoda by Wilson and Sereno (1998). This derived condition is also present in other non-eusauropod sauropods such as Tazoudasaurus, Gongxianosaurus, and Chinshakiangosaurus (Allain and Aquesbi, 2008, Upchurch et al., 2007a and Upchurch et al., 2007b) and therefore diagnose a more inclusive group than Eusauropoda. Therefore, the evolutionary history of this character is probably more complex than currently depicted. For instance, a faintly developed wrinkling pattern has been noted for some teeth of sauropod outgroups (e.g., Mussaurus, Anchisaurus, Melanorosaurus [Pol and Powell, 2007a, Yates, 2004 and Yates, 2007]). The wrinkling of these taxa is certainly not as developed as in Amygdalodon and eusauropods but differs from the plesiomorphic smooth surface present in the most basal sauropodomorphs or ‘prosauropods’ (e.g., Thecodontosaurus, Plateosaurus, Massospondylus). Within this context, the character may have gradually appeared during the evolution of Sauropodormorpha and the autapomorphic pattern of pits and sulci of Amygdalodon may represent an incipient condition in comparison with the more notoriously developed anastomized pattern observed in eusauropods. Although the wrinkling pattern described above for Amygdalodon might have been affected by wear and/or preservational causes, we consider this pattern reflects an autapomorphic feature for this taxon because: a) the abrasion on teeth with different degrees of wear does not alter significatively the wrinkling pattern in other basal sauropods (e.g., Patagosaurus MPEF-PV 3006, MPEF-PV 3060, MACN-CH 2008); b) if the pattern of pits and sulci were related to preservational causes, it remains to be explained why such process affected only the lingual and buccal surfaces but not the mesial and distal edges of the same teeth (e.g., MLP 46-VIII-21-1/12 and MLP 46-VIII-21-1/13) that have an anastomized wrinkling pattern.

En-echelon tooth arrangement. Another character previously regarded as synapomorphic of Eusauropoda (Wilson, 2002) or the more inclusive clade Gravisauria (Allain and Aquesbi, 2008) is the presence of an en-echelon arrangement of the tooth crowns. The basal wear facets present in some teeth are interpreted here as the product of this type of arrangement in Amygdalodon. Such interpretation, coupled with the similar disposition of tooth crowns in the more basal sauropodomorph Mussaurus patagonicus (Pol and Powell, 2007a), shows that this character has a markedly broader distribution and appeared before the origin of Sauropoda. In fact, Yates (2007) scored this condition as present in most sauropodomorphs. As in the previous character, the degree of overlapping between adjacent crowns may have increased along the evolution of Sauropodomorpha, with sauropods showing a more extensive overlap and intimate contact between adjacent crowns.

The large morphological gap that existed until recently between the dental anatomy of eusauropods and more basal sauropodomorphs have led previous authors to infer that the origins of Eusauropoda was signed by a marked specialization toward herbivory (Upchurch and Barret, 2000 and Wilson, 2005). These authors, however, recognized that the absence of knowledge on dental anatomy of basal sauropods precluded establishing if some of these features may have appeared earlier in the evolution of the group (potentially enlarging the morphological gap between eusauropods and other forms due to the lack of adequate fossil record). The remains described here of Amygdalodon interpreted within the context of the present phylogenetic analysis, coupled with other recent studies on basal sauropods (e.g., Tazoudasaurus [Allain and Aquesbi, 2008], Chinshakiangosaurus [Upchurch et al., 2007b]), confirm that this is in fact the case for many of the derived dental characters previously regarded as eusauropod synapomorphies. As discussed above, these features must now be interpreted as diagnostic of more inclusive groups, indicating gradual appearance of these derived dental features at different nodes below Eusauropoda and underscoring the importance of feeding adaptations in the early evolution of Sauropoda. Interestingly, within the context of the phylogenetic analysis presented here, most of the dental characters discussed above do not involve multiple instances of homoplasy, as predicted by Upchurch et al. (2007b) due to the putative plasticity of teeth in response to changes in diet and feeding habits. The major exception, however, is the optimisation of multiple transformations in the presence/absence of tooth serrations among non-neosauropod sauropods.

It must be noted that, despite this new evidence, our understanding on the early sauropod dental evolution is still fragmentary due to the incomplete nature of most of basal sauropods, including A. patagonicus. The incompleteness of these remains creates a considerable degree of uncertainty on the phylogenetic interrelationships of these taxa, as shown by the low values of nodal support (see above). This precludes the establishment of a solid evolutionary scenario for the dentition of basal sauropods. Furthermore, the lack of teeth in the basal sauropods Lessemsaurus (Pol and Powell, 2007b) and Antetonitrus (Fig. 6) creates ambiguous optimisations for dental characters and therefore many of them may indeed diagnose even more inclusive clades. Many of the characters discussed above, however, are absent in Mussaurus (and Melanorosaurus [Yates, 2007]) and therefore are likely to have evolved at the base of Sauropoda (or at a slightly less inclusive node).