Terastiodontosaurus Georgalis & Smith, 2024

Georgalis, Georgios L., T., Krister, Marivaux, Laurent, Herrel, Anthony, Essid, El Mabrouk, Ammar, Hayet Khayati, Marzougui, Wissem, Temani, Rim & Tabuce, Rodolphe, 2024, The world’s largest worm lizard: a new giant trogonophid (ºquamata: Amphisbaenia) with extreme dental adaptations from the Eocene of Chambi * Tunisia, Zoological Journal of the Linnean Society 202 (3) : -

Terastiodontosaurus Georgalis & Smith gen. nov.

Zoobank registration: urn:lsid:zoobank.org:act:DC23B781-B109-4EFB-9C0B-DA42DC09B838

Etymology: The genus name derives from the Greek words ‘τεΡάστιος’ (‘terastios’)* meaning ‘huge’/‘enormous’* ‘ὀδούς’ [in genitive: ‘ὀδόντος’ (‘odontos’)]* meaning ‘tooth’* and ‘σΑύΡΑ’ (‘saura’)* meaning ‘lizard’. The gender of the new genus name is masculine.

Type and only known species: Terastiodontosaurus marcelosanchezi Georgalis & ºmith gen. et sp. nov.

Diagnosis: As for the type and only known species.

Note on the proper authorship and spelling of Trogonophidae : Although authorship of Trogonophidae is generally attributed to Gray (1865) (e.g. Estes 1983 * Bailon 2000 * Kearney 2003)* it should be noted that versions of that name had also appeared earlier. These are the Trogonophidina of Bonaparte (1838a) * Trogonophina of Bonaparte (1838b * 1839 * 1840a * 1840b)* Trogonophidae of Gray (1840) * and Trogonophes of Fitzinger (1843). Even within the works of Gray* that author earlier misspelled this group as Trigonophes and Trigonophidae ( Gray 1844)* with Kuhn (1966 * 1967) subsequently assigning authorship to Gray (1844). Vanzolini (1951) supposedly created Trogonophinae as a new subfamily; however* obviously this cannot be the case following the Principle of Coordination of ICZN (1999: Article 36)* which dictates that ‘A name established for a taxon at any rank in the family group is deemed to have been simultaneously established for nominal taxa at all other ranks in the family group’. This being said* it thus appears that Trogonophidina of Bonaparte (1838a: 392) is the first introduction of the name* although it was not accompanied by a diagnosis or an explicit mention of a type genus. The same applies to the second usage of the name* again by the same author and again in the same year* Trogonophina of Bonaparte (1838b: 124) * which also was not accompanied by a diagnosis or a type genus. The following year* Bonaparte (1839: 10) applied* for the first time* a (rather brief) diagnosis for Trogonophina* simply stating ‘Dentes cum maxillis concreti’* but again still no explicit mention of a type genus was made; this is also exactly the case for his subsequent works ( Bonaparte 1840a: 286 * 1840b: 99)* where he applied the name with the same exactly diagnosis but again with no explicit mention of a type genus. Later on* the same author added the number of species he included in that group (i.e. one) and its geographical distribution as ‘Africa’ ( Bonaparte 1850 * 1852). Duméril and Bibron (1839) used the name Trogonophides* also providing a thorough description of Trogonophis wiegmanni ; however* it is evident in their text that this name was simply an informal plural name for the genus Trogonophis * for which they were also using the informal singular term ‘Le Trogonophide’. Gray (1840: 42) was the first explicitly to mention a genus (‘ Trogonophis * Kaup’) associated with his family group name Trogonophidae * and Fitzinger (1843) was the first explicitly to mention both a genus (‘ Trogonophis . Kaup’) and a species (‘ Trogonoph. Wiegmanni . Kaup’) associated with his family group name Trogonophes. Nevertheless* an explicit mention of a type genus is not a formal requirement for family-group names that were established before 1999* but instead it is enough that there is an indirect inference of the genus from the stem of the family-group name (see ICZN 1999: Article 11.7). Accordingly* authorship of Trogonophidae should be attributed to Bonaparte (1838a).

This being said and now that the proper authorship of the family-group name is clarified* a further comment on the proper spelling of the name is also required. Taking into consideration that the proper authorship of the family-group name is Bonaparte (1838a) * it follows that his Trogonophidina would be transformed into Trogonophididae* a spelling that has not appeared the literature. It should be noted that some* mostly recent* authors have used the spelling Trogonophiidae (e.g. Pyron et al. 2013 * Čerňanský et al. 2015a * Zheng and Wiens 2016* Burbrink et al. 2020). Nevertheless* the spelling that has made the most frequent appearance in the literature is Trogonophidae (e.g. Gray 1840 * Cope 1887 * Taylor 1951 * El-Assy and Al-Nassar 1976 * Gans 1978 * Estes 1983 * Charig and Gans 1990 * Bailon 2000 * Kearney 2002 * 2003 * Augé 2005 * 2012 * Vidal and Hedges 2005* Maisano et al. 2006 * Gans and Montero 2008* Vidal et al. 2008 * Wiens et al. 2010 * 2012 * Longrich et al. 2015 * Baeckens et al. 2017* Hawkins et al. 2022 * Araújo ºalvino et al. 2024* Bell et al. 2024). Following ICZN (1999: Article 29.5)* the spelling of a family-group name that is in prevailing usage should be maintained* even if this spelling is not the original spelling and even if its derivation from the name of the type genus is not formed in a grammatically correct manner. Accordingly* the proper spelling of this family-group name is Trogonophidae .

Terastiodontosaurus marcelosanchezi Georgalis & Smith sp. nov.

( Figs 2–17, 23A; Supporting Information, Figs S1–S 6)

Zoobank registration: urn:lsid:zoobank.org:act:881978AE-4954-4D2C-8D6E-12CD56CB4C20

Etymology: The species epithet is named after Professor Marcelo ºánchez-Villagra* director of the Palaeontological Institute of the University of Zurich * as an honour for his major contributions to palaeontology* zoology* and evolutionary biology* in addition to the kind friendship and his great support to the first author (G.L.G.) for many years.

Holotype: A right maxilla (ONM CBI-1-645) ( Figs 2 * 3).

Paratype: A left dentary (ONM CBI-1-646) ( Figs 4 * 5* 23A).

Referred specimens: Four premaxillae (ONM CBI-1-658* ONM CBI-1-672* ONM CBI-1-711* and ONM CBI-1-1021)* five right maxillae (ONM CBI-1-649* ONM CBI-1-651* ONM CBI-1-654* ONM CBI-1-667* and ONM CBI-1-1017)* six left maxillae (ONM CBI-1-648* ONM CBI-1-653* ONM CBI-1- 1012* ONM CBI-1-1016* ONM CBI-1-1018* and ONM CBI- 1-1022)* one right maxilla fragment (ONM CBI-1-650)* five right dentaries (ONM CBI-1-656* ONM CBI-1-660* ONM CBI-1-666* ONM CBI-1-1013* and ONM CBI-1-1020)* 10 left dentaries (ONM CBI-1-647* ONM CBI-1-655* ONM CBI-1- 657* ONM CBI-1-659* ONM CBI-1-661* ONM CBI-1-662* ONM CBI-1-668* ONM CBI-1-670* ONM CBI-1-1014* and ONM CBI-1-1015)* a fragment of the coronoid process of a left dentary (ONM CBI-1-664)* and 17 tooth-bearing bone fragments (ONM CBI-1-652 and ONM CBI-1-671 [16 elements]). Tentatively also: numerous presacral vertebrae (ONM CBI-1- 673* ONM CBI-1-682* ONM CBI-1-685* ONM CBI-1-687* ONM CBI-1-691 [‘lot / batch’ with numerous vertebrae]* ONM CBI-1-706 [‘lot / batch’ with numerous vertebrae]* ONM CBI- 1-820* ONM CBI-1-833* and ONM CBI-1-860) and two caudal vertebrae (ONM CBI-1-686 and ONM CBI-1-689).

Diagnosis: Terastiodontosaurus marcelosanchezi can be referred to Amphisbaenia based on the prominent and enlarged median premaxillary tooth* the large anterior premaxillary foramina* the low tooth count on the maxilla and dentary* the ventral extension of the mandibular symphysis below Meckel’s groove* the broad insertion fossa for mandibular adductor muscles on the posterolateral surface of the dentary* and the strong and elevated coronoid process of the dentary. Terastiodontosaurus marcelosanchezi can be referred to Trogonophidae based on the presence of acrodont dentition* closely appressed (‘fused’) teeth* the interdigitating suture between the frontal and the facial process of the maxilla* and ectopterygoid abutting the posteromedial corner of the maxilla.

Terastiodontosaurus marcelosanchezi is united with Trogonophis wiegmanni by: thick enamel on marginal teeth* and ‘twinning’ of paired premaxillary teeth* with median tooth separated by a diastema from paired teeth. Terastiodontosaurus marcelosanchezi is united with Todrasaurus gheerbranti by: thick enamel on marginal teeth* extremely enlarged (>60% longer than adjacent teeth) dentary tooth 3–4 positions from the rear* and small ‘hills’ on posterior dentary teeth. Terastiodontosaurus marcelosanchezi can be differentiated from all other amphisbaenians by the combination of the following features: very large size (with maximum known maxilla length> 16 mm and maximum dentary length of 17 mm)* the flat apical surface of the cheek teeth* the position of the largest tooth on the maxilla* the number of maxillary teeth (usually three maxillary teeth)* the ratio of the largest maxillary tooth length to the total maxillary tooth row length (between 0.5 and 0.7)* and the position of the largest tooth on the dentary (fourth from posterior).

Type locality and horizon: Chambi 1 (CBI-1)* Djebel Chambi* Kassérine region * western part of Central Tunisia * Tunisia; late early to early middle Eocene (late Ypresian to early Lutetian).

Geographical and stratigraphical range: Taxon known exclusively from its type locality.

Nomenclatural remark on the new taxon: The authorship of this new genus and species should be referred to as Terastiodontosaurus Georgalis & ºmith gen. nov. (for the new genus) and Terastiodontosaurus marcelosanchezi Georgalis & ºmith gen. et sp. nov. (for the new species)* following Article 50.1 and the ‘recommendation 50A concerning multiple authors’ of the International Code of Zoological Nomenclature (ICZN 1999).

Description

Holotype ( Figs 2, 3) Theholotyperightmaxilla (ONMCBI-1-645)isalmostcomplete ( Figs 2 * 3). The premaxillary process curves up anterodorsally* forming part of the medial border of the external naris. The sharp rim weakens anteriorly* and the anterior portion of the process as a whole slopes ventrolaterally. The dorsal surface is slightly striated* probably where it was overlapped by the septomaxilla. The ventral surface of the premaxillary process curves upwards and is more coarsely striated where it overlaps the premaxilla. The superior alveolar canal opens anteriorly through two foramina: a dorsally directed foramen on the medial side of the facial process* and an anteriorly directed foramen immediately anterior to the facial process; between them the canal is roofed* but there appears to be a narrow groove* as if two growing folds of perichondral bone met over a channel but did not fuse fully. Anterior to the anterior opening of the superior alveolar canal the dorsal surface of the premaxillary process is coarsely striated; here it was probably overlain by the septomaxilla* into which the neurovascular structure(s) of the superior alveolar nerve continued. It is unclear whether the division of the anterior opening of the superior alveolar canal marks a separation of the course of the subnarial artery and superior alveolar nerve* as in Iguanidae ( Oelrich 1956 * ºmith 2009).

On the medial side of the premaxillary process is a strong* dorsomedially facing facet for the vomer. Behind this the palatal shelf is restricted and its dorsal surface hollows out* where Jacobson’s organ sat; together with the vomer* here the maxilla formed the fenestra vomeronasalis externa. Posterior to the cavity for Jacobson’s organ the maxilla possesses a distinct medial process that contacts the horizontal wing of the vomer. The process continues laterally across the palatal shelf and extends posterodorsally as a weak ridge on the medial surface of the facial process of the maxilla. In other squamates* ºmith and Gauthier (2013) identified this as the ‘nasolacrimal ridge’ after ascertaining a relationship with the lacrimal duct.

Behind this process the palatal shelf is medially extensive. A distinct palatine process is absent* but an extensive* striated facet for the palatine articulation is present. The (posterior) superior alveolar foramen opens at the level of the anterior end of the palatine facet. A distinct facet for the ectopterygoid cannot be discerned* but it must have been an abutting one* and it might be confluent with the palatine facet.

The posterior process of the maxilla is broad and flares sharply outwards (or laterad fide Gans and Montero 2008). Although striated surfaces of this maxilla commonly show tiny holes* the entire posterior surface of the posterior process is heavily porous and coarsely striated. Dorsally* the posterior process is thick and also coarsely longitudinally striated; there is no evidence that a distinct element articulated on this suborbital part of the maxilla.

The facial process is thick. A rough facet for the nasal bone is found on the anterior margin of the facial process and turns posterodorsally as it ascends the process. There is a distinct change in angulation below the peak of the facial process* and the facet behind it (now more on the medial than the anterior surface of the facial process) is smoother; probably* the change marks the boundary between the nasal and part of the frontal facet. Immediately behind its apex the facial process is deeply notched* almost certainly to receive a tongue-like process of the frontal bone. Additionally* there is a number of rough* anteroventrally trending ridges and grooves posterior to the frontal notch* which possibly indicate a more extensive overlap of the maxilla on the prefrontal (like in lizards) than is observed today in Trogonophis wiegmanni .

The lateral surface of the maxilla is porous and weakly rugose in its dorsal part* with some longitudinal grooves and ridges on the posterior process. There are three distinct labial foramina* all of them more or less above the level of the largest tooth. The two anterior ones are more closely spaced* while there is a considerable distance between the middle and posterior one* the latter also being the largest. These foramina communicate internally* as expected* with the superior alveolar canal* which runs longitudinally through the maxilla immediately above the tooth row. It conveys the superior alveolar nerve* which passes from the orbit into the maxillary canal through the superior alveolar foramen* in addition to the maxillary artery ( Oelrich 1956).

The dentition is acrodont. The maxilla possesses only three teeth* of which the second is by far the largest (some 75% longer than the next-largest tooth) and* in fact* covers much of the ventral portion of the bone; this tooth is followed in size by the anteriormost tooth* while the posteriormost tooth is considerably tiny. The teeth are closely appressed* with no interdental gaps between them. All three teeth are much flattened* particularly the two largest ones: their dorsoventral height is extremely low. Moreover* the μCT scan reveals that the enamel is extremely thick (ºupporting Information* Fig. º1A). ºeveral nutritive foramina are present at the base of the teeth.

Paratype ( Figs 4, 5)

The paratype left dentary (ONM CBI-1-646) is relatively complete ( Figs 4 * 5). It is referred to the same species on the basis of co-occurrence and the following derived morphological features: great enamel thickness* reduced tooth row length and tooth count* presence of a greatly enlarged tooth* the ‘hill’ on the small* posteriormost teeth* and the apically flat teeth in the middle.

The symphysis is broad and anterodorsally inclined. As in other amphisbaenians ( Longrich et al. 2015)* it is not restricted to the area above the Meckelian groove but instead curves anteroventrally around it and terminates posteroventrally in a sharp corner. It appears as though the Meckelian groove is closed and fused immediately behind the symphysis and that the Meckelian cartilage itself might be ossified as an anterodorsally– posteroventrally extensive wedge in the centre of the symphysis* but this interpretation is not entirely clear.

Meckel’s groove is open ( Figs 4 * 5B* C). Above it is the indistinct supra-Meckelian lip (of Bhullar and ºmith 2008)* above which is the subdental shelf (sensu Rage and Augé 2010) that extends medial to the tooth row. The subdental shelf commences below the posterior portion of the first tooth. It is narrow anteriorly and becomes gradually wider in the posterior half of the dentary* and behind the tooth row it grades into the coronoid process. The part of the groove in which Meckel’s cartilage was situated is relatively narrow throughout most of its length* but it widens a bit posteriorly* immediately in front of the mandibular foramen. That groove is straight for most of its length but turns up strongly near the symphysis* giving it a hockey-stick shape* as in other amphisbaenians ( Longrich et al. 2015). Posteriorly in the dentary there are two strong* elongate facets below the groove for Meckel’s cartilage* for the angular and the splenial. The presence of two distinct facets strongly suggests that those mandibular elements were discrete* not fused* as in Trogonophis wiegmanni * where only one facet is present; of course* without the remainder of the mandible this cannot be taken as certain. The angular facet extends as far anteriorly as the anterior end of the enlarged tooth (fourth from the back)* whereas the splenial facet only just passes the posterior end of the enlarged tooth. The ventral margin of the dentary behind the symphysis is straight.

The dentary possesses a posterodorsally ascending coronoid process immediately behind the posteriormost tooth. The dorsal tip of the process is incomplete* but even the preserved portion extends high above the tooth row* as in other amphisbaenians. The process is prominent and thick. There is a strong diagonal ridge on its lateral surface that delineates the adductor fossa dorsally. The ridge is narrower at its base but grows thick and porous as it curves dorsally and then diminishes; it might have served as an attachment site for other jaw adductors* probably via a bodenaponeurosis.

The mandibular canal runs across the length of the dentary* transmitting the inferior alveolar nerve and mandibular artery; it communicates with the labial foramina present in the labial surface of the dentary and with the nutritive foramina at the bases of the teeth. Its entrance is the mandibular foramen* which is located well behind the tooth row at the level of the anterior edge of the coronoid process. In labial view ( Figs 4A * 5A)* four labial foramina are present in the anterior half of the dentary.

The dentary bears eight acrodont teeth* of which the largest is the fourth from posterior. The posteriormost tooth is the smallest one. The first two teeth are bulbous* and the anteriormost tooth is procumbent* extending beyond the anterior margin of the symphysis. As in the holotype maxilla described above* the teeth are closely appressed* with almost no interdental gaps between them (at maximum* only tiny empty spaces between some but not all teeth). All teeth except for the first two are much flattened (particularly the two largest ones)* with the exception of the two anteriormost ones* which are bulbous and tall. The posterior teeth have tiny central cusps. Multiple nutritive foramina are situated above the subdental shelf ventrally to the each of the tooth bases.

Referred specimens

Premaxillae ( Figs 6–8)

The most complete premaxillae are ONM CBI-1-672 and ONM CBI-1-711* which include a large portion of the nasal process* whereas this structure is mostly broken in ONM CBI-1-658 and ONM CBI-1-1021 ( Figs 6–8; ºupporting Information* Fig. º2). The nasal process is large* dorsoventrally elongated and moderately wide; it gradually narrows in width apically. Given its preserved extent* it is likely that the nasal process reached the frontals* a condition unique to Trogonophidae among amphisbaenians ( Kearney 2003)* but more complete specimens are needed to verify this. A pair of large anterior premaxillary foramina is developed at the base of the nasal process. These are the anterior openings for the ethmoidal nerve that enters the premaxilla through the posterior premaxillary foramina on the posterior side of the nasal process. No rostral process or rostral blade is present. The palatal shelf or alveolar plate is most complete in ONM CBI-1-672 ( Figs 7C * D* 8P– R)* but it is ONM CBI-1-658 ( Figs 6F * G* 8J–L) that most clearly shows it to be bipartite.

The dentition is acrodont. On the alveolar plate* there are five teeth* all of bulbous morphology. The central median tooth is the most robust and prominent (a synapomorphy of amphisbaenians; see Gans 1978 * ºmith 2009). The μCT scans reveal that the median tooth possesses great apical enamel thickness (ºupporting Information* Fig. º1C). There is a diastema between the median tooth and the lateral ones that contrasts strongly with the otherwise close appression of the teeth generally and gives the impression of the lateral two teeth being ‘twinned’. In fact* in the two pairs of lateral teeth* the dental gaps between the teeth are almost absent. A slight vertical striation is observable on all teeth* being more distinct in the central tooth.

Maxillae ( Figs 9–12)

The available maxillae pertain to different-sized individuals* as can be attested by the drastic size range between the smallest and the largest specimens (see ºupporting Information* Figs º3* º4). The most complete maxillae are the holotype (ONM CBI-1-645) and the specimens ONM CBI-1-649 ( Figs 9E * F* 11J–Q) and ONM CBI-1-654 ( Fig. 11A–I)* which are both significantly smaller. In fact* the holotype ONM CBI-1-645 ( Figs 2 * 3) represents the largest known individual of Terastiodontosaurus marcelosanchezi * whereas ONM CBI-1-649 represents one of the smallest among our sample. Nevertheless* the morphology of the maxillae overall is very similar. Here* we focus on comparisons.

The anterodorsally trending premaxillary process is best preserved or complete in ONM CBI-1-654 ( Fig. 11A–I)* ONM CBI-1-649 ( Figs 9E * F* 11J–Q)* the holotype ONM CBI-1-645 ( Figs 2 * 3)* and the fragmentary specimens ONM CBI-1-1012 ( Fig. 12G–I)* ONM CBI-1-1016 ( Fig. 12J–L)* ONM CBI- 1-1018 ( Fig. 12M–O)* and ONM CBI-1-1022 ( Fig. 12P–R). The outwardly flaring posterior process of the maxilla is most complete in ONM CBI-1-654 ( Fig. 11A–I)* followed in completeness by the holotype ONM CBI-1-645 ( Figs 2 * 3)* ONM CBI-1-649 ( Figs 9E * F* 11J–Q)* ONM CBI-1-651 ( Fig. 10E–J)* ONM CBI-1-653 ( Fig. 11R–U)* ONM CBI-1-667 ( Fig. 12A– C)* and ONM CBI-1-1017 ( Fig. 12D–F). A medial process that contacts the horizontal wing of the vomer is clearest in the holotype (ONM CBI-1-645) and especially in ONM CBI-1-648

( Figs 9B * C* 10B* C) and ONM CBI-1-649 ( Figs 9E * F* 11K–Q). The porosity and ridge-like form of the labial surface* observed in the holotype ONM CBI-1-645 ( Figs 2A * 3A* F)* is otherwise evident only in the second largest specimen (ONM CBI-1-651; Fig. 10E) and is absent in all other smaller specimens* in which this surface is almost completely smooth. This suggests that this feature is subject to size/ontogenetic variation.

There are three labial foramina* placed almost in a row* in the holotype ONM CBI-1-645 ( Figs 2A * 3A) and in ONM CBI-1- 649 ( Figs 9G * 11J); in both these specimens* the two anterior foramina are more closely spaced. ONM CBI-1-648 also has three labial foramina ( Fig. 9A)* but the posterior two foramina are more closely spaced. In ONM CBI-1-649* there is also a further foramen situated dorsal to the row of the three foramina* situated approximately above the first foramen. Even more* in ONM CBI-1-654* there are two small foramina above the two of three foramina ( Fig. 11A). The number of foramina cannot be assessed fully in the remaining incomplete maxillae. The superior alveolar foramen is usually relatively large.

ºimilar to the holotype (ONM CBI-1-645)* maxillae almost always bear three teeth* of which the second is by far the largest and* in fact* covers much of the ventral portion of the bone; this tooth is followed in size by the anteriormost tooth* while the posteriormost tooth is tiny. However* there are two notable exceptions* denoting some degree of variability: in ONM CBI-1-651* there is a fourth tiny tooth located posteromedially to the third tooth* not in line with it ( Fig. 10F)* while in ONM CBI-1-649 the tiniest tooth is absent* meaning that specimen bears only two teeth ( Figs 9G * 11M–O). A further* interesting variation occurs also in the right maxilla fragment ONM CBI- 1-650* where the anterior (smaller) preserved tooth is different from the same tooth in other specimens* being extremely narrow ( Figs 9D * 10K–M). Tentatively* we attribute this variation as being intraspecific (or even ontogenetic). This could indeed be the case* taking into consideration that especially if the more anterior teeth are being ‘de-emphasized’ in favour of the huge tooth (which serves as the prey ‘cracker’)* and thus the anterior teeth perhaps become more of a remnant/vestige* then one would expect a greater variation. Moreover* again in the same specimen* there seems to be a considerable gap between the two preserved teeth ( Figs 9D * 10K–M); this is unusual* because in most remaining maxillae and dentaries* the teeth are closely appressed* with no interdental gaps between them (but in ONM CBI-1-653 the third tooth is somewhat more widely separated; Fig. 11T).

Notably also* in some cases* there is a distinct ‘hill’ forming on the smallest tooth or teeth. This is the case with ONM CBI- 1-651 (where the ‘hill’ is prominent in both the small third and fourth teeth; Fig. 10E * I* J) and ONM CBI-1-653 ( Fig. 11 º–U).

ºeveral nutritive foramina are always present at the base of the teeth in all specimens* in the remnants of the subdental gutter; however* their number and size are variable. Usually* these foramina are mostly at the medial side of the maxilla; however* there is also some variation: in ONM CBI-1-650* they are equally present in both labial and medial aspects of the bone ( Figs 9D * 10K* L).

In order to facilitate quantitative investigation* we introduce the ratio of largest tooth length on the maxilla to total tooth row length of the maxilla. This was complete in only four specimens:

Holotype * ONM CBI-1-645: largest tooth length* 4.9 mm / total tooth row length* 9.5 mm; ratio* 0.52.

ONM CBI-1-649: largest tooth length* 2.6 mm /total tooth row length* 3.7 mm; ratio* 0.70.

ONM CBI-1-651: largest tooth length* 3.7 mm /total tooth row length* 7.3 mm; ratio* 0.51.

ONM CBI-1-654: largest tooth length* 2.3 mm /total tooth row length* 4.1 mm; ratio* 0.56.

Besides* some information on this ratio can also be gleaned tentatively from some incomplete maxillae:

ONM CBI-1-650: largest tooth length* 4.8 mm /preserved tooth row length (incomplete)* 7.8 mm; estimated ratio* <0.62.

ONM CBI-1-648: largest tooth length* 3.2 mm /preserved tooth row length (incomplete)* 5.6 mm; estimated ratio* <0.57.

Dentaries ( Figs 13–16)

Apart from the paratype dentary ONM CBI-1-646* all remaining dentaries are rather incomplete. The available sample denotes a range of sizes* but substantially less disparate than the maxillary sample (cf. ºupporting Information* Figs º3–º6). The paratype ONM CBI-1-646 represents one of the largest individuals* with the fragmentary dentary ONM CBI-1-659 pertaining to a more or less similar size. All dentaries closely approach in overall morphology the paratype ONM CBI-1-646 described above. ºome specimens are nevertheless highly incomplete* sometimes preserving only the anterior portion of the dentary (e.g. ONM CBI-1-1014* ONM CBI-1-1015* and ONM CBI-1-1020)* while ONM CBI-1-664 is only a fragment of the coronoid process of a left dentary.

The tooth row is complete only in the paratype ONM CBI-1- 646* where it comprises eight acrodont teeth* the largest being the fourth one (counting from posteriorly). Otherwise* the tooth row is almost complete in ONM CBI-1-657 ( Figs 14A * 15A–C)* which preserves all but the seventh tooth (counting from posteriorly). In that specimen also* the fourth tooth is the largest one and the posteriormost tooth is the tiniest one (both counting from posteriorly). One important difference in ONM CBI-1-657 is that there appears to be a dental gap between the anteriormost tooth and the succeeding tooth position (although we cannot be certain that this is not an artefact). Otherwise* in all specimens* all teeth are almost adjoined* with almost no interdental gaps between them (at maximum only tiny empty spaces between some* but not all teeth* do exist). As in the paratype ONM CBI-1-646* also in ONM CBI-1-647* ONM CBI-1-655* ONM CBI-1-657* ONM CBI-1-662* ONM CBI-1-666* ONM CBI-1-1014* ONM CBI-1-1015* and ONM CBI-1-1020* the anteriormost tooth is more bulbous and dorsoventrally high and projects beyond the anterior surface of the symphysis. This is also the case for the second anteriormost tooth in the paratype ONM CBI-1-646* as in ONM CBI-1-647* ONM CBI-1-655* ONM CBI-1-662* ONM CBI-1-666* ONM CBI-1-1014* ONM CBI-1-1015* and ONM CBI-1-1020* which is also relatively bulbous and dorsoventrally tall. In dentary ONM CBI-1-670* the teeth are not that flattened but apparently represent the anterior teeth [second* third* or fourth (counting from anteriorly)]* near the symphysis.

As in the maxillae* there is a distinct ‘hill’ on the tiny teeth (e.g. ONM CBI-1-646 and ONM CBI-1-657). Great enamel thickness is also found on the dentary teeth.

Nutritive foramina are open at the base of various teeth above the subdental shelf in all dentaries; their number is not consistent* and it can vary between tooth positions and individuals at the same position. In ONM CBI-1-657* these are poorly developed.

Meckel’s groove is open in all specimens where this can be studied* most notably ONM CBI-1-660. Owing to their incompleteness* besides the paratype dentary ONM CBI-1-646* which bears four* the exact number of labial foramina cannot be assessed in the remaining dentaries. Interestingly* however* ONM CBI-1-659 is pierced by several (at least nine) tiny foramina in its labial surface* most of which are closely spaced.

As in the case of the maxillae above* in order to facilitate further quantitative investigation* we introduce the ratio of largest dentary tooth length to the total tooth row length of the maxilla. This was complete in only two specimens:

Paratype ONM CBI-1-646: largest tooth length* 2.8 mm /total tooth row length* 11.9 mm; ratio* 0.24.

ONM CBI-1-657: largest tooth length* 1.6 mm /total tooth row length* 6.4 mm; ratio* 0.25.

Besides these specimens for which ratios could be calculated* the following dentaries provide data on the length of the largest tooth in further specimens:

ONM CBI-1-647: largest tooth length* 2.9 mm.

ONM CBI-1-659: largest tooth length* 2.3 mm.

ONM CBI-1-655: largest tooth length* 1.9 mm.

ONM CBI-1-656: largest tooth length* 1.5 mm.

Vertebrae ( Fig.17)

Vertebrae are referred tentatively to the same taxon on the basis of co-occurrence and the fact that large numbers of jaws have yielded only a single species of amphisbaenian thus far from the locality.

Presacral vertebrae range in size between ~1 and 5 mm ( Fig. 17A–U). They are procoelous and dorsoventrally compressed. In anterior view* the prezygapophyses are strongly inclined* much exceeding in height the anterodorsal edge of the neural canal. There is no zygosphene. The cotyle is elliptical and strongly depressed. In posterior view* the condyle is also elliptical and strongly depressed. The neural arch is depressed. The lateral walls of the neural arch form moderately robust centropostzygapophyseal laminae (sensu Georgalis et al. 2018b). There is no zygantrum. In dorsal view* the prezygapophyses extend anterolaterally. The prezygapophyseal articular facets are large and broad; in some specimens* there are prominent prezygapophyseal accessory processes. There is no neural spine. The interzygapophyseal constriction is deep. There is practically no posterior median notch of the neural arch. In ventral view* the centrum is flattened* with only slightly concave lateral margins. Two usually large* occasionally asymmetrical subcentral foramina are present* one at each lateral side of the ventral surface of the centrum. The synapophyses are robust and more or less rounded. In lateral view* the neural arch rises distinctly* with a gentle curve towards its posterior end. Each prezygapophysis is connected to the related postzygapophysis by a relatively low interzygapophyseal ridge.

Caudal vertebrae have haemapophyses fused to the centrum. A short anterior caudal vertebra has forked lymphapophyses ( Fig. 17V * W)* whereas more elongate posterior caudal vertebrae have unitary pleurapophyses ( Fig. 17X * Y). If the number of caudal to presacral vertebrae can be determined* the methodology of ºmith (2013) might be used to estimate the proportion of caudal vertebrae and thus to constrain the relative length of the tail.

Results of the phylogenetic analysis

Our main analysis used only two topological constraints: Afrobaenia and ºouth American Amphisbaenidae . Fully consistent with all recent phylogenetic analyses (e.g. Kearney 2003 * Müller et al. 2011 * Gauthier et al. 2012 * Jones et al. 2013 * Čerňanský et al. 2015a * Longrich et al. 2015 * Zheng and Wiens 2016* ºtreicher and Wiens 2017 * ºimões et al. 2018* Burbrink et al. 2020 * ºinghal et al. 2021* Tałanda et al. 2022 * Brownstein et al. 2023 * Čerňanský and Vasilyan 2024)* our analysis finds strong support for amphisbaenian monophyly ( Fig. 18). The position of Cryptolacerta hassiaca Müller * Hipsley* Head* Kardjilov* Hilger* Wuttke & Reisz* 2011* from the early to middle Eocene of Messel* Germany * originally described as a link between lacertids and amphisbaenians by Müller et al. (2011) * is unresolved (but see also Longrich et al. 2015 * Brownstein et al. 2023 * Čerňanský and Vasilyan 2024). In general* higher-level relationships within Amphisbaenia are poorly supported* but our analysis shares the basal position of Rhineuridae with Gauthier et al. (2012) * which was the first purely morphological analysis to recover this topology. A surprise was the close relationship between the unnamed amphisbaenian from Adrar-Mgorn 1 of Augé and Rage (2006) to Blanidae * although with poor support (Bº <0.50); furthermore* if all higher-level relationships within Afrobaenia (e.g. Graboski et al. 2022) are enforced* its position becomes unresolved. This unnamed pleurodont form represents the second amphisbaenian from the locality of Adrar-Mgorn 1 in Morocco (the other being Todrasaurus gheerbranti ) and was originally described by Augé and Rage (2006) as bearing some resemblance to both blanids and amphisbaenids* whereas the phylogenetic analysis of Longrich et al. (2015) recovered it as an Amphisbaenia incertae sedis. The cadeid Cadea palirostrata Dickerson * 1916 was assessed with relatively strong support as the sister taxon of Afrobaenia* i.e. the group encompassing Cadeidae * Trogonophidae * and Amphisbaenidae (Bº = 0.80). Terastiodontosaurus marcelosanchezi was inferred with moderate support (Bº = 0.65* one unambiguous character state change* Bremer support 1; Table 1) to be the sister taxon of Todrasaurus gheerbranti * and the two together were inferred with strong support to be the sister taxon of Trogonophis wiegmanni (Bº = 0.95* five unambiguous character state changes* Bremer support 3)* a novel result. The herein novel topology of Todrasaurus differs from that of Longrich et al. (2015) * who had tentatively recovered this Moroccan taxon on the stem of Afrobaenia ( Longrich et al. 2015). Moreover* Trogonophidae (comprising Trogonophis and its stem plus Diplometopon zarudnyi Nikolskyi * 1907) was inferred to be monophyletic with strong support (Bº = 0.96* 26 unambiguous character state changes* Bremer support 5). Enforcing all major topological constraints within Afrobaenia (fide Graboski et al. 2022) did not affect the relationships within Trogonophidae (including fossil taxa) or their basal position in Afrobaenia.