Lesothosaurus diagnosticus (Sereno, 1991)

DIAGNOSIS OF LESOTHOSAURUS DIAGNOSTICUS

Sereno (1991) proposed six autapomorphies for Lesothosaurus diagnosticus , using both cranial and postcranial features. Butler (2005) reviewed this list, surveyed the anatomy of other early ornithischians and concluded that none of these could be considered as valid autapomorphies as they were either more widely or ambiguously distributed within Ornithischia . We discuss these characters, below:

1. Slot in maxilla for lacrimal

The maxilla and lacrimal articulate via an unusual tongue-in-groove joint in Lesothosaurus , with the anterior process of the lacrimal fitting into a slot on the dorsal margin of the maxillary ascending process ( Sereno, 1991). A similar articulation has also been claimed to be present in Agilisaurus louderbacki ( Peng, 1997) and Jeholosaurus shangyuanensis ( Xu, Wang & You, 2000) but the nature of this contact cannot be confirmed in either of the latter taxa ( Butler, 2005; Barrett & Han, 2009). Damage or incompleteness prevents the assessment of this feature in many other early ornithischian taxa (such as Eocursor and Laquintasaura : see Butler, 2010; Barrett et al., 2014) but it is clearly absent in Tianyulong ( Zheng et al., 2009) , Heterodontosaurus (SAM-PK-K1332) and Scelidosaurus (NHMUK PV R1111) , three other taxa that have this region preserved. Butler (2005) argued that because this feature is also present in NM QR 3076 (a specimen that he excluded from Lesothosaurus on other grounds) it is more widely distributed among basal ornithischians and cannot be considered an autapomorphy of L. diagnosticus .

However, we posit that the shared possession of this feature in Lesothosaurus and NM QR 3076 supports referral of the latter specimen to Lesothosaurus and that it should potentially be regarded as a valid autapomorphy of L. diagnosticus . Further evidence for the assignment of NM QR 3076 to L. diagnosticus is provided below. Dryosaurid ornithopods also possess a maxillary/ lacrimal articulation similar to that seen in Lesothosaurus ( Galton, 1983; Butler et al., 2008a, b), but as this feature is absent in other ornithopods (e.g. Hypsilophodon, NHMUK PV R 197, R2477: Galton, 1974) it seems more likely that this represents a convergent acquisition of this feature, rather than a shared plesiomorphy.

2. Anterior premaxillary foramen

Butler (2005) proposed that the presence of this feature in NM QR 3076 prevented this feature from being autapomorphic for Lesothosaurus . However, we regard it as a potential autapomorphy that further supports the referral of NM QR 3076 to Lesothosaurus (see above and further Discussion, below). When discussing this feature as a potential autapomorphy of L. diagnosticus, Sereno (1991: 172) used the term ‘anterior premaxillary foramen’ to describe ‘an additional foramen...[that] pierces the premaxilla near the alveolar margin’. He went on to describe this feature further noting that ‘A groove passes anteroventrally from the premaxillary foramen to a second larger foramen, here termed the anterior premaxillary foramen’ ( Sereno, 1991: 177). However, the term anterior premaxillary foramen has been used in subsequent works to describe a feature seen in several heterodontosaurids such as Echinodon and Fruitadens ( Butler et al., 2008b, Butler et al., 2011; Sereno, 2012). Whether the feature in heterodontosaurids is homologous with the premaxillary foramen or the anterior premaxillary foramen in Lesothosaurus diagnosticus is hard to ascertain. However, the presence of two premaxillary foramina connected by an anteroventrally orientated groove appears to be unique to Lesothosaurus diagnosticus and is not present in any other early ornithischian. A similar groove and second foramen are present on a specimen of Hypsilophodon foxii (NHMUK PV R2477) and this was also noted by Sereno (1991). However, although this specimen does appear to have a second ‘anterior’ premaxillary foramen damage to this specimen makes it impossible to determine if this feature is both genuine and homologous to that in L. diagnosticus .

3. Short forelimb

Sereno (1991: 172) defined a ‘short forelimb’ as ‘less than 40% of [the] hindlimb length’ (defined as the sum of propodial, epipodials and third metapodials length in the fore- and hind limbs). Butler (2005) argued that this feature was also present in several other basal ornithischian taxa, but did not provide comparative values. A reassessment of this feature found that the forelimb length of Lesothosaurus diagnosticus is 41% of the hind limb length, based upon the syntype material. In comparison, forelimb/hind limb ratios in other taxa are: Abrictosaurus consors 47%, Tianyulong confuciusi 36%, and Heterodontosaurus tucki and Hypsilophodon foxii both 53%. Unfortunately, few basal ornithischian taxa preserve all of the elements required to make this calculation. Nevertheless, given the range of values available we follow Butler (2005) in rejecting the use of a short forelimb as an autapomorphy of L. diagnosticus .

4. Lateral exposure of the brevis surface on the postacetabular process

This feature is also present in several other ornithischian taxa, such as Agilisaurus louderbacki and Scelidosaurus harrisonii , and is most likely an ornithischian symplesiomorphy ( Barrett et al., 2005; Butler, 2005).

5. Dorsal groove on the ischial shaft ( Fig. 13 View Figure 13 )

This feature is also present in Agilisaurus louderbacki and Eocursor parvus ( Peng, 1990, 1992, 1997; Butler, 2005, 2010). Given its distribution among other early ornithischians it cannot be considered as an autapomorphy of Lesothosaurus , nor a feature that can be used to unite Lesothosaurus and Stormbergia to the exclusion of other taxa.

6. Pedal digit I reduced ( Figs 17 View Figure 17 , 18 View Figure 18 )

Pedal digit I is reduced to varying extents in many other ornithischian taxa, including Hypsilophodon foxii , Agilisaurus louderbacki , Hexinlusaurus multidens and Jeholosaurus shangyuanensis ( Galton, 1974; He & Cai, 1984; Barrett et al., 2005; Han et al., 2012) and so cannot be regarded as an autapomorphy of L. diagnosticus ( Butler, 2005) .

As features 1 and 2 are not present in any other early ornithischians, we provisionally regard them as valid autapomorphies of L. diagnosticus . The presence of these features in NM QR 3076 allows us to refer this specimen to this taxon. This makes NM QR 3076 the largest known definitive individual of L. diagnosticus , larger even than the specimens referred to Stormbergia ( Fig. 23 View Figure 23 ). This material provides us with further evidence about the postcranial anatomy of L. diagnosticus , particularly in larger individuals, and has wider implications for our interpretation of the ontogenetic status of the syntype material of this taxon as well as the taxonomic status of S. dangershoeki , which this study considers to be a junior subjective synonym of L. diagnosticus (see above).

In addition to the aforementioned features, other workers have suggested potential postcranial autapomorphies for L. diagnosticus . Norman et al. (2004) listed the presence of a prominent obturator process on the ischium and the absence/strong reduction of metatarsal V as autapomorphies. An obturator process may or may not be present in one of the syntypes, NHMUK PV RU B17, as there is some debate as to whether fragments previously attached to the ventral edge of the left ischium represent a true feature of the bone, and whether the remaining ventral expansion of the ischial shaft is a remnant of a now broken process. Thulborn (1972) and Norman et al. (2004) described this specimen of Lesothosaurus as having a small but distinct obturator process, whereas Butler (2005) regarded it as absent, as did Sereno (1991). Butler (2005) used the proposed absence of the obturator process in L. diagnosticus and the clear presence of this feature in S. dangershoeki to distinguish the two taxa. However, there is now evidence for this feature being variably expressed between individuals and its distribution among specimens is complex. It seems likely that the syntype specimen had a small ‘ventral corner’, as in Eocursor parvus , which developed into a larger, more distinct obturator process during maturity in L. diagnosticus (see below). Regardless, the presence of an obturator process in many other ornithischian taxa means that this cannot be regarded as a valid autapomorphy for L. diagnosticus .

The loss of metatarsal V occurs in all specimens of L. diagnosticus that preserve articulated metatarsals, as well as SAM-PK-K1105 (‘ S. dangershoeki ’). Although Santa Luca (1984) reported the presence of an articular surface for metatatsal V in specimen SAM-PK-K401, this study has not found any evidence of such feature in SAM-PK-K401, or in any of the other specimens that preserve articulated pedes. Reduction of metatarsal V does occur in other ornithischian taxa, such as Heterodontosaurus tucki , Scelidosaurus harrisonii , Agilisaurus louderbacki and Jeholosaurus shangyuanensis but complete loss of metatarsal V does not appear to occur in any other basal ornithischian ( Santa Luca, 1980; He & Cai, 1984; Peng, 1990, 1992, 1997; Han et al., 2012). The holotype of Tianyulong confuciusi, STMN 26-3, does not preserve a metatarsal V, but Sereno (2012) attributes this absence to post-mortem loss. Although Sereno (2012) does not report it, a highly reduced, splint-like metatarsal V is present in Abrictosaurus consors (NHMUK PV RU B54). Metatarsal V is lost in more derived taxa such as Hypsilophodon foxii ( Galton, 1974) , Gasparinisaura cincosaltensis ( Coria & Salgado, 1996) and many other members of Iguanodontia ( Norman et al., 2004), but we consider this a homoplastic loss, given the plesiomorphic retention of metatarsal V in other more basal ornithischians. Thus, given that S. dangershoeki and L. diagnosticus are considered synonymous herein, the loss of metatarsal V becomes a valid autapomorphy of L. diagnosticus . The only specimen that Butler (2005) referred to Stormbergia that includes any of the tarsus is the holotype SAM-PK-K1105. In this specimen, metatarsals I – IV are preserved together, along with an astragalus and calcaneum and a distal fibula, but metatarsal V is absent. Close inspection of this cluster of bones found no traces of the fifth metatarsal and so its absence is provisionally considered a genuine feature and not as a result of taphonomic loss. As metatarsal V is extremely reduced in several heterodontosaurids, Agilisaurus louderbacki and Jeholosaurus shangyuanesis, it seems likely that this reduction was widespread in early ornithischians, with L. diagnosticus having the unique and extreme condition of complete absence, at least among those taxa close to the base of the tree.

A further potential autapomorphy of L. diagnosticus was proposed by Butler (2010), who noted the presence of a shallow rostrocaudally extending groove on the dorsal margin of the surangular. This feature is present in NHMUK PV RU B17 ( Porro et al., 2015: fig. 2F; Barrett et al., 2016) and NHMUK PV RU B23 and MNHN LES18. This feature is not present in Eocursor parvus , Heterodontosaurus tucki or Manidens condorensis ( Butler, 2010; Pol et al., 2011; Sereno, 2012). The absence of this feature in other basal ornithischian taxa and its presence in L. diagnosticus suggests that it is a valid autapomorphy of L. diagnosticus ( Butler, 2010) .

ONTOGENY

All of the specimens mentioned in this study are from the upper Elliot Formation, have the same general morphology for the majority of their preserved skeletal elements and share the same unique combination of characters (once certain ambiguous characters are discounted). Some of the largest specimens (e.g. NM QR 3076: femur length = 234.5 mm) are almost 230% larger than the smallest specimens (e.g. NHMUK PV RU B17: femur length = 103.5 mm). Many other specimens fill the size range between these extremes. Minor differences in morphology can be observed in several of these specimens, in a number of different skeletal elements, and seem to correlate with the size of the individual (see above). For example, the relative expansion of the distal end of the humerus is greater in the larger individuals and the level of medial bowing in the femur is smaller. It seems plausible that these differences are most likely the products of differing maturity within the available sample, rather than reflecting specific or generic separation.

Knoll (2002) and Knoll et al. (2010) suggested that the synytpe material of L. diagnosticus represents juvenile or immature individuals, based on histological analyses of referred specimens from the upper Elliot Formation. The individuals included within these studies showed growth patterns indicative of juvenile status, such as the absence of external fundamental systems and evidence of sustained high growth rates. On this basis, Knoll et al. (2010) suggested that Stormbergia should be regarded as an adult morph of Lesothosaurus and that the former taxon was a junior subjective synonym of the latter. However, the material used in this study did not include any material that could be definitively referred to either Lesothosaurus or Stormbergia , but was sourced from fragmentary specimens lacking the proposed diagnostic features of either taxon. Consequently, the results of these analyses cannot be considered conclusive. Subsequent phylogenetic studies overlooked the synonymy suggested by Knoll et al. (2010) and continued to include Stormbergia as a valid taxon ( Butler, 2010; Han et al., 2012; Osi} et al., 2012; Barrett et al., 2014). Nevertheless, Knoll et al. (2010) did provide evidence that at least some ornithischian specimens, of comparable size to those securely referred to Lesothosaurus , were definitely immature.

Butler (2005: 185) also discussed the possibility that the syntype material of Lesothosaurus represents juvenile individuals, citing the lack of neurocentral fusion in the vertebral column and lack of sacral fusion as evidence. This lack of fusion was also reported as present in a specimen referred to Stormbergia dangershoeki ( Butler, 2005: 185) . However, this specimen (BP/1/4885) is the smallest specimen that Butler (2005) considered referable to S. dangershoeki . Here, this specimen is considered to be a medium-sized L. diagnosticus , one that exhibits a transitional form between the L. diagnosticus and S. dangershoeki morphotypes in terms of both its size and its ischial morphology (see above). If this specimen is a mid-sized, immature individual of L. diagnosticus then the lack of neurocentral fusion would still be expected and could not be used as evidence for the separation of S. dangershoeki and L. diagnosticus . Further to this, the short snout length, relatively large orbits, arched skull roof and unfused neurocranial sutures of the syntype skull material (NHMUK PV RU B17 and NHMUK RU B23) also indicate that these individuals are juveniles ( Galton, 1974; Carpenter & McIntosh, 1994).

The assignment of NM QR 3076 to L. diagnosticus , based on the presence of three autapomorphies, shows that the maximum body size of the species is much greater than that achieved by the syntypes, again implying that the latter are juvenile individuals. Further to this, NM QR 3076 lies well within the size range of those specimens referred to S. dangershoeki by Butler (2005) and, given that they all share a very similar combination of characters (identical if we disregard the ambiguous ischial characters discussed above), as well as a single L. diagnosticus autapomorphy (absence of metatarsal V), we consider the specimens of Stormbergia dangershoieki to represent older individuals of L. diagnosticus , whose distinguishing features are a result of ontogeny and not taxonomic difference.

The most apparent differences between the specimens are in the presence/absence and form of the obturator process of the ischium. Given the proposed synonymy of Lesothosaurus diagnosticus and Stormbergia dangershoeki , we propose that the differences observed in the ischia could be a result of ontogeny and that, in L. diagnosticus , ischial ontogeny follows the following sequence:

1. Juveniles: short, partially twisted ischia with a symphysis that contributes a relatively large proportion of the total ischial shaft length, with an extended ventral edge (exemplified by NHMUK PV RU B17 right ischium).

2. Sub-adult: longer, straighter ischia with the symphysis forming a smaller proportion of the total length following elongation of the shaft proximal to the ischial symphysis, and the presence of a continuous flange-like obturator process (as in BP/1/4885 left ischium; NHMUK PV RU B17, left ischium?).

3. Adult: long, straighter shaft with only a distal symphysis and a more tab-like obturator process (NHMUK PV R11000, SAM-PK-K1105).

In addition to changes in ischial morphology several other features appear to covary with individual size. The femur of NHMUK PV RU B17 has a greater degree of medial bowing than larger specimens of L. diagnosticus (SAM-PK-K401, SAM-PKK1107). This bowing is present to a lesser extent in other smaller specimens and is completely absent in the largest specimens (including Stormbergia dangershoeki ). Further to this, and occurring in parallel, the angle of offset of the femoral head to the femoral shaft increases with increased femoral length. This is also the case in other ornithischian taxa with partial ontogenetic sequences, such as Hypsilophodon foxii ( Galton, 1974) . This change in the medial bowing of the femur could be related to the increased mechanical loading associated with increased body size: the effect of increased loading has been suggested as the cause of ontogenetic changes among the femora of Dryosaurus lettowvorbecki ( Heinrich, Ruff & Weishampel, 1993) . Another explanation for this change is that it could be related to a potential shift from quadrupedality in juveniles to bipedality in adults. This idea of a changing stance through ontogeny has been suggested for the ornithischian taxon Dryosaurus lettowvorbecki , and also for other basal dinosaurs such as Massospondylus carinatus ( Heinrich et al., 1993; Reisz et al., 2005, 2010). However, there is currently no evidence for obligate quadrupedality in juvenile individuals of Lesothosaurus diagnosticus , or any other basal ornithischian. More work is needed to elucidate the biomechanical underpinning of this ontogenetic change.

The morphology of the proximal humerus also appears to change ontogentically (compare Figs 7 View Figure 7 , 24 View Figure 24 ). NHMUK PV RU B17 has a humerus in which the proximal expansion is approximately the same width as the distal expansion and that is symmetrical with respect to the long axis of the bone. By contrast, larger specimens, such as SAM-PK-K1106 and SAM-PK-K1107, have a proximal expansion that is wider than the distal expansion and which is asymmetrical across the long axis of the bone. This asymmetry in the larger specimens is a result of there being a greater degree of medial expansion than of lateral expansion ( Fig. 24 View Figure 24 ). The best example of the condition in larger individuals is seen in SAM-PK-K1107. Butler (2005: 199) listed SAM-PK-K1107 as ‘cf. S. dangershoeki ’ and argued that this difference in humeral morphology was sufficient to distinguish it from Lesothosaurus diagnosticus . However, given its larger size, and the range of morphology present among upper Elliot Formation ornithischians, it seems more likely that this humeral morphology represents one end of a continuum, which appears to be directly related to the size of the individual. Changes in the relative size of the proximal expansion of the humerus that covary with body size, and particularly the change in long-axis symmetry, is also seen in other basal taxa that have well-documented ontogenetic sequences. Another relatively small ornithischian, Hypsilophodon foxii , shows proximal humerus asymmetry with increasing size (NHMUK PV R5829 vs. NHMUK PV R5830: Galton, 1974).

The scapulae of L. diagnosticus show a small range of distal morphologies too. As with the humerus, there is a clear symmetry across the long axis of the scapulae of smaller individuals but asymmetry in the larger individuals (see above). Given the covariation with body size, this plausibly represents ontogenetic rather than taxonomic variation.

OTHER INDIVIDUAL VARIATION AND SKELETAL

ASYMMETRY

The morphology of the obturator foramen also varies among specimens of Lesothosaurus diagnosticus , with some exhibiting a fully enclosed foramen while in others it is partially open, forming a variably open obturator notch. However, unlike the other variations observed, this does not appear to be related to individual size. NHMUK PV RU B17, one of the smallest specimens, preserves a proximal pubis with an open obturator notch. SAM-PK-K401, on the other hand, appears to have a fully enclosed obturator foramen although it is only slightly larger than NHMUK PV RU B17. Larger specimens also vary in the nature of the obturator notch (SAM-PK-K1106, closed; BP/1/4885, open). In the specimens with an open obturator notch the posterior section of the proximal pubis, which forms a posteroventrally projecting process, is not broken distally, as would be expected if the open notch was the result of damage. SAM-PK-K1106 provides a possible explanation for this type of variation. In this specimen the obturator foramen is closed and a possible junction between the posteroventrally projecting posterior section and a separate, posterior projection from the rod-like pubic shaft can be seen on the posterior margin of the rim enclosing the obturator foramen. If genuine, and not the result of damage, the presence of this junction may explain why there is inconsistency in the opening/closure of the obturator notch in other Lesothosaurus specimens. The two processes that merge to form the rim of the foramen may ossify at different ontogenetic stages or at different rates. If the distally projecting posterior section of the proximal pubis ossified first and the posterior process on the shaft ossified later it could be the case that those specimens that do not have a fully closed obturator foramen simply died before the two processes could meet to form this combined structure. It would also explain why the distally projecting posterior section of the proximal pubis appears unbroken in those specimens.

Whatever the underlying cause, the variation in the morphology of the proximal pubis is considered to be intraspecific variation. A similar case of intraspecific variation in the proximal pubis of Hypsilophodon foxii was reported by Galton (1974), a taxon in which some specimens possess a closed obturator foramen (NHMUK PV R193 and NHMUK PV R196) and others an open obturator notch (NHMUK PV R195, NHM UK PV R5829).

Specimens like BP/1/4885 suggest that possible asymmetry occurs between the left and right ischia of some smaller specimens. Skeletal asymmetry is rare but has been documented in dinosaurs, including the theropod Mirischia asymmetrica ( Martill et al., 2000) . Asymmetry of the skeleton could be sexually dimorphic, or may simply reflect a common tetratological lack of ossification either on one or both sides of the pelvis. It is also plausible that the process is prefigured as cartilage in juveniles, remaining so until later stages of maturity, and that the distribution that we see in the fossil record is as a result of this developmental process. These differences in the ischia and pubes.

suggest that the whole pelvic region might have undergone substantial modification during growth. These changes may have occurred to enable oviposition in adults.