Mandibular and dental characteristics of the Late Jurassic mammal Henkelotherium guimarotae (Paurodontidae, Dryolestida)

Luo, Zhe-Xi; Martin, Thomas

doi:10.1007/s12542-023-00651-z

Mandibular and dental characteristics of the Late Jurassic mammal Henkelotherium guimarotae (Paurodontidae, Dryolestida)

Research Paper
Published: 06 April 2023

Volume 97, pages 569–619, (2023)
Cite this article

Download PDF

Access provided by Autonomous University of Puebla

PalZ Aims and scope Submit manuscript

Mandibular and dental characteristics of the Late Jurassic mammal Henkelotherium guimarotae (Paurodontidae, Dryolestida)

Download PDF

426 Accesses
3 Citations
10 Altmetric
Explore all metrics

Abstract

Henkelotherium guimarotae Krebs 1991 is an important Jurassic mammal for understanding therian evolution. We are presenting a new study of extensive, previously undescribed, mandibles and dentitions. The revised dental formula is: I4? or 5?/i4, C1/c1, P4/p4, M6/m7. The canine and premolars show an alternate replacement that ends with M4/m4 eruption, and is followed by a late sequential eruption of the last three lower (m5-7) and last two upper (M5-6) molars. The lower premolars erupted in the following order: p1 → p3 → p2 → p4, and the canine erupted most probably shortly before p4. The timing of the premolar replacement before the late molar eruption is similar to that of Dryolestes leiriensis, and is a characteristic of dryolestidans. Henkelotherium lower molars have subequal roots, a plesiomorphy of non-dryolestidan mammals, and the upper molars are supported by a strong, curved lingual root; a derived character. In the upper molars, the postvallum wear surface is contiguous to the parastyle wear surface of the succeeding molar, which differs from dryolestids. The parastylar lobe of the succeeding molar, and the postvallum of the preceding molar, are imbricated, and can develop strong, continuous wear surfaces, matching the prevallid crest of the lower molar. Henkelotherium differs from dryolestids in having an inflected, shelf-like mandibular angular process with a foramen. This large sample of Henkelotherium shows a significant variation gradient along the molar series, with the strongest wear occurring only in two to three consecutive molars. The extraordinarily long molar row is correlated with the late growth of jaws; and the jaw with late addition of molars sustained an effective mastication, much longer in older adults of dryolestidans than in other Mesozoic stem therians. The late eruption of several more molars after completion of antemolar replacement suggests that dryolestidans had either a longer-lived life, or slower life-history traits, or a combination of both, than crown therians.

Distinct mandibular premolar crown morphology in Homo naledi and its implications for the evolution of Homo species in southern Africa

Article Open access 06 August 2020

New Specimens of Reigitherium bunodontum from the Late Cretaceous La Colonia Formation, Patagonia, Argentina and Meridiolestidan Diversity in South America

Article 13 December 2021

New information on the maxilla, dentary, and dentition of Maotherium sinense, with comments on the zhangheotheriid dental formulae

Article 15 June 2019

Use our pre-submission checklist

Avoid common mistakes on your manuscript.

Introduction

Dryolestidans are the most stem-ward group of the cladotherian clade that includes living marsupials and placentals known as crown therians. They are fossil relatives of crown therians such as tribosphenid mammals (boreosphenidans) (Crompton 1971; McKenna 1975; Prothero 1981; Martin 1999; Luo et al. 2002; Rougier et al 2011; Wible and Rougier 2017). By current consensus, dryolestidans represent the proximal relatives to the common ancestor of “peramuran” and tribosphenid mammals known collectively as Zatheria (Kielan-Jaworowska et al. 2004; Martin 2018). In comparative morphology of Mesozoic mammals, dryolestidans are more derived than the spalacotherioid “symmetrodontans” in cranial and molar features, but more plesiomorphic than zatherians (including the paraphyletic “peramurans”) in dental, mandibular and petrosal characteristics (Krebs 1991; Sigogneau-Russell 1999; Luo and Wible 2005; Martin and Rauhut 2005; Rougier et al. 2007; Luo et al. 2007a, b, 2012; Ruf et al. 2009; Schultz and Martin 2014; Hughes et al. 2015; Grossnickle et al. 2022).

Dryolestidans have an extensive fossil record. The two diverse and abundant families are the Paurodontidae and Dryolestidae. The earliest fossil of the Paurodontidae, from the Balabansai Formation of Kyrgyzstan, is of Callovian age, Middle Jurassic (Martin and Averianov 2010), and the family extends to the Late Jurassic (Simpson 1929; Krebs 1991, 2000; Averianov and Martin 2015). It should be noted that in some recent analyses, the paurodontids have turned out to be paraphyletic (Rougier et al. 2012; Averianov et al. 2013; Wible and Rougier 2017; although see Martinelli et al. 2021). The earliest fossils of the family Dryolestidae are from the Middle Jurassic Berezovsk site of Siberia (Averianov et al. 2014) and the Late Bathonian, Middle Jurassic Forest Marble Formation of Great Britain (Freeman 1976, 1979).

The peak taxic diversity and specimen abundance, and the widest geographic distribution, of dryolestidans occurred in the Late Jurassic and Early Cretaceous, during which time the group is represented by relatively abundant fossils from the Morrison Formation of North America (Simpson 1929; Prothero 1981; Kielan-Jaworowska et al. 2004; Averianov and Martin 2015), the Late Jurassic Guimarota coal mine of Portugal (Martin and Krebs 2000) and the Late Jurassic Langenberg quarry of Germany (Martin et al. 2021), with a single putative record of Brancatherulum from the Late Jurassic in Africa (Heinrich 1991). Dryolestidans have a moderate diversity in the Lower Cretaceous of Great Britain, Spain, Germany, and Northern Africa (Simpson 1928; Henkel and Krebs 1969; Krebs 1971, 1985, 1993; Sigogneau-Russell 1991; Ensom and Sigogneau-Russell 1998; Martin 1998, 1999; Martin et al. 2022b; Lasseron et al. 2022), plus a putative dryolestidan fossil from Australia (Clemens et al. 2003). From the late Early Cretaceous and onward, dryolestidans are unknown among the faunas of the Laurasian continents, and have presumably disappeared. There are putative records from the Late Cretaceous of North America (Lillegraven and McKenna 1986), however, and from South America (Martin et al. 2022a).

Recent discoveries of meridiolestidan mammals from the Cretaceous of South America, and their inclusion in phylogenetic analyses, have altered the interpretation if paurodontids are still monophyletic, as previously assumed (e.g., Kielan-Jaworowska et al. 2004). Meridiolestidans were considered to be related to dryolestidans, when these fossils were first discovered and described (Bonaparte 1986, 1990, 1994, 2002; Kielan-Jaworowska et al. 2004; Chornogubsky 2011; Rougier et al. 2009, 2011, 2012, 2021b). However, a majority of the recent re-analyses of their relationships placed meridiolestidans and some paurodontids (with exception of Henkelotherium) as an independent clade from dryolestids (Rougier et al. 2012; Averianov et al. 2013; O’Meara and Thompson 2014; Wible and Rougier 2017), rendering paurodontids as a paraphyletic grouping. Other phylogenetic analyses with a smaller taxonomic sample still recognize the Paurodontidae as a clade (e.g., Martinelli et al. 2021), or recognize meridiolestidans as a group nested in the spalacotherioid clade, not related to dryolestidans (Averianov et al. 2013).

Eight endemic genera of meridiolestidans are known from the Cretaceous and Cenozoic of South America (Martinelli et al. 2021). The genus Peligrotherium is known from the Paleocene of Argentina, and the genus Necrolestes from the Miocene of Argentina is also considered to be a meridiolestidan (Rougier et al. 2012; Chimento et al. 2012; Wible and Rougier 2017). These fossils provide some interesting examples of a Mesozoic mammal group that survived the end of the Cretaceous mass extinction (Gelfo and Pascual 2001; Rougier et al. 2021a, b), well into the Neogene (Rougier et al. 2012). Pending a resolution of the competing hypotheses to place meridiolestidans with regards to paurodontids (Chimento et al. 2012; Rougier et al. 2012), or to spalacotherioids (Averianov et al. 2013), dryolestidans may be the longest-living Mesozoic mammal clade that survived well into the Neogene.

Henkelotherium guimarotae (Krebs 1991) is so far the best-preserved taxon (Fig. 1) of dryolestidans (Martin and Krebs 2000). It yielded significant information for the understanding of early mammal evolution, for the postcranial locomotor function, and for interpreting the ear region of therians (Krebs 1991; Vázquez-Molinero et al. 2001; Ruf et al. 2009; Luo et al. 2012; Hughes et al. 2015; Jäger et al. 2013a, b; 2019a). It is a key comparative taxon for the overall mammal phylogeny (Hu et al. 1997, 1998; Rougier et al. 1998; Ji et al. 1999; Luo et al. 2002), and for studying the ecological diversity of Mesozoic mammals (Krebs 2000; Luo 2007; Martin 2018; Jäger et al. 2019a).

On the basis of a larger, and previously unpublished sample of fossils of Henkelotherium guimarotae, we provide a detailed description of the mandibles and teeth, and document the character variation of the teeth, to augment the information on this important fossil mammal (Fig. 1). We also attempt to interpret the tooth replacement and eruption, and to characterize the late eruption of molars during the prolonged growth of the mandible, for the understanding of the paleobiology of Henkelotherium.

Materials and methods

Fossil specimens for this study (Tables 1 and 2; Appendix 1 in Supplementary Information) are from the Late Jurassic (Kimmeridgian) Guimarota coal mine, which has yielded a rich assemblage of fossil mammals, with over 32 species (Martin and Krebs 2000). Five species of dryolestidans have been recognized: the dryolestids Dryolestes leiriensis, Krebsotherium lusitanicum and Guimarotodus inflatus (Martin 1999), and the paurodontids Henkelotherium guimarotae (Krebs 1991) and Drescheratherium acutum (Krebs 1998). In abundance of specimens, Henkelotherium guimarotae is the fourth most common mammal species after the docodontan Haldanodon exspectatus and the dryolestids Dryolestes leiriensis and Krebsotherium lusitanicum from this site (Martin and Krebs 2000). For this study, we have available six specimens of the upper tooth series, some of which show replacement pattern in the antemolar (incisor, canine, premolar) positions (Table 1), and 21 mandibular specimens with lower teeth that cover the entire range of growth from juvenile to adult, and show the replacement of the teeth in the antemolar positions (Table 2; Appendix 1).

Table 1 Henkelotherium guimarotae upper dental specimens

Full size table

Table 2 Henkelotherium guimarotae lower dentition and mandible specimens

Full size table

For a description of the molars (Figs. 2, 3, 4, 5), we followed the character definition of Prothero (1981) and Martin (1999), as modified by Kielan-Jaworowska et al. (2004: Fig. 10.2) and by Schultz and Martin (2011: Fig. 1). For stages of dental wear in molars, we follow the criteria of Schultz and Martin (2011). To avoid potential ambiguity of terms, terminology of molars is illustrated in Figs. 2, 3, 4, 5. For the morphology of the mandible, we follow the terminology of Krebs (1971) and Martin (1999), as augmented by this study (Figs. 1, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16).

Institutional Abbreviations: AMNH—American Museum of Natural History (New York, USA); Gui Mam—Guimarota Mammal Collection of the Geological Museum, National Laboratory of Energy and Geology (MG/LNEG), Lisbon, Portugal, currently under study at Universität Bonn (Germany). NLMH—Niedersächsisches Landesmuseum, Hannover, Germany. WMNM, LWL—Museum für Naturkunde/Westfälisches Landesmuseum mit Planetarium, Münster, Germany. For Geological Museum (MG/LNEG) repository specimen numbers see Supplementary Information.

Results

The body of the mandible (Corpus mandibulare) (Figs. 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16) is relatively deep, to accommodate the very long roots of the premolars and molars (Figs. 6, 7, 8, 9, 10, 11, 12, 13, 14, 15), as previously noted for Dryolestes and Krebsotherium (Martin 1999). The medial surface of the mandibular body is flatter, and has a greater vertical depth from the ventral margin to the tooth alveoli than the more convex lateral surface of the mandibular body, which has a shorter vertical depth. This was first observed in the H. guimarotae holotype by Krebs (1991) and can now be confirmed by additional mandibles examined in this study. We note that this difference between the medial depth and lateral depth of the mandible is complementary to the difference in the crown height difference on the lingual and labial sides of the tooth crown. Crowns of the lower molars are much taller on the labial side and shorter on the lingual side. The different depths of the medial and lateral surfaces of the mandibular body compensate for these differential heights of the lingual and labial sides of the molar crowns, such that the occlusal surface across the molar trigonids is transversely leveled, along the molar series. This feature is present in all dryolestidans (Krebs 1971; Martin 1999). It is also a common feature of therian mammals with zalambdodont molars (Ungar 2010; Kelt and Patton 2020)—the molars with steeply v-shaped trigonids that are also antero-posteriorly shortened.

The lingual alveolar margin of the mandible shows a slightly concave curvature in the premolar region (Figs. 1 and 6), such that the alveolar line along the anterior part of the tooth row shows a series of sigmoid curvatures (Figs. 1 and 12d). The lingual alveolar line is slightly raised (convex) dorsally under i4 and the canine, and then lower down to the lowest point under p1 and p2. The lingual alveolar line is raised from p3 and p4 through m1 and m2. The alveolar line is straight and leveled along most of the molar row (m2–m6) (Krebs 1971) (Fig. 12d).

The ventral margin of the mandible shows a rounded outline in cross-section (Figs. 1 and 12). Longitudinally, the posteroventral margin of the mandible is concave ventrally, from the level of the mandibular foramen to the apex of the angular process. This concave mandibular outline is related to the slight downward and medial orientation of the angular process. From the level of the mandibular foramen to the level of the posterior premolars, the mandibular ventral margin is gently convex. Further anteriorly from the premolar region, the ventral margin is straight. However, the anterior (mesial) end of the mandible is strongly convex under i1 and i2. Correspondingly, i1 and i2 have more vertically oriented crowns but long, posteriorly curved roots that are bent to conform to the symphyseal margin. The roots of i1 and i2 extend posteriorly to the medial side of the i3 and i4 roots. The canine has two roots that are also strongly curved posteriorly (Figs. 9, 10, 13, 14).

Anterior part of mandible

The mandible appears to be more constricted dorso-ventrally immediately posterior to the canine in both medial and lateral views, due to the concave curvature of the alveolar margin in the premolar region (Figs. 1 and 2). In the majority of the specimens, there are two mental foramina on the labial side of the incisor to premolar region: a circular mental foramen under i4, or between i4 and the canine, and a second, larger and oval mental foramen under the p2–p3 junction, or under p3. However, the number of foramina can be variable: two juvenile mandibles have three mental foramina (Gui Mam 44/80 and 25/81) (Figs. 9 and 11).

The variability of the mental foramina in Henkelotherium is similar in scope and in number to the variation pattern already known in Dryolestes (Martin 1999), most specimens of which have two mental foramina, one each under the canine, and under p2 or the p2–p3 junction. However, in one specimen (“Guimarota 6”), there are two foramina under p2 and p3 in a region where a single mental foramen is expected in other Dryolestes specimens (Martin 1999). In all dryolestids, the posterior-most position of the mental foramen is limited to the anterior premolar region, and usually anterior to the p3 locus and within the rostral 20% of the mandibular length. The relatively more anterior placement of the posterior-most mental foramen in both Henkelotherium and Dryolestes is similar to the anterior mandibular region in the mammaliaforms Morganucodon and Haldanodon, australosphenidan mammals, in spalacotherioids, and in meridiolestidans (Kermack et al. 1973; Krusat 1980; Crompton and Luo 1993; Rich et al. 1999, 2001, 2016; Asher et al. 2007; Ji et al. 2009; Rougier et al. 2011).

In Henkelotherium, the three foramina for the mental nerves are connected to the mandibular canal that carried the inferior alveolar nerve (Figs. 6, 7, 9). The more anterior placement of the mental foramina of dryolestidans is in contrast to the more derived zatherian mammals, as defined by the common ancestor of the Mesozoic “peramurans”, metatherians and eutherians. In zatherians, the posterior-most mental foramen is under the first lower molar (m1) for metatherians (Rougier et al. 1998, 2004; Luo et al. 2003), or under the ultimate premolar for the “peramuran” Peramus (Clemens and Mills 1971; Davis 2012), and in early eutherians such as Eomaia (Ji et al. 2002), Maelestes (Wible et al. 2009) and Juramaia (Luo et al. 2011), or in eutherians in general (Novacek 1986). Because “peramuran” mammals have a shorter molar row of only three-to-four molars (far fewer molars than most dryolestidans that have six-to-nine molars), a placement of the posterior-most mental foramen below the ultimate premolar or the first molar represents a more significant posterior shift of the mental foramina in “peramurans”, than the more anterior placement of this foramen in Dryolestes and Henkelotherium and other mammaliaforms more stem-wards in the phylogeny. The mental foramina are conduits for mental nerves (branches of the inferior alveolar nerve) that innervate the chin area and the anterior portion of the lower jaw in extant mammals (Evans 1993; Dyce et al. 1995). It is likely that the innervation area on the anterolabial surface of the lower jaw on the head surface is smaller in the more stem-ward dryolestidans, than in the phylogenetically more derived zatherians.

Symphysis

The symphysis is the contact between the two hemi-mandibles in mammals (Scapino 1965, 1981; Scott et al. 2012), and an unfused symphysis is a typical feature of mammaliaforms (Kermack et al. 1973; Crompton and Luo 1993; Luo and Martin 2007; Panciroli et al. 2019, 2021). The symphysis in Henkelotherium is an area of slightly rugose texture on the medial surface of the anterior part of the mandible from i1 to p2 (Figs. 1, 6, 7, 8, 9, 12, 13, 14, 15). The depth of the symphyseal contact between the two mandibles (hemi-mandibles) is about the half of the mandibular depth at the level of the canine. A symphyseal foramen, which is connected to the mandibular canal in the interior of the jaw, is present in the symphysis below i4 or just anterior to the canine root (Gui Mam 44/88, 133/77, 48/78).

The extent of the symphysis is variable with mandibular growth. It extends from the anterior (mesial-most) end, posteriorly to the level below the junction of p2 and p3 in juvenile and young adult specimens (Fig. 8). For example, in a juvenile mandible (Gui Mam 44/80) in which this structure is well preserved, the symphysis is relatively shorter, and ends below p1 (Fig. 9d). However, in the larger mandibles of presumably older individuals, the symphysis becomes longer proportionally and in absolute size, extending further posteriorly to below the junction of p2 and p3 in the next larger mandible (Gui Mam 133/77) (Fig. 6d). The symphysis was not fully exposed in the holotype specimen (Krebs 1991), but the visualization of CT scans shows that it appears to have a relatively longer symphysis, as in Gui Mam 48/78 (Fig. 17). In some of the largest available mandibles, such as Gui Mam 48/78, the symphysis extends further to the posterior edge of p3 (Figs. 8, 13). From the growth series of the mandible, it is clear that the extent of the symphysis has become longer with older growth stages of the mandible.

The symphysis is not fused in any of the mandibles in the large sample of this study. This suggests that the symphysis was mobile in life in Henkelotherium. Our interpretation is that the two mandibles were connected in an amphiarthrosis condition (sensu Scott et al. 2012), more likely by ligaments than by fibrocartilage. Both types of tissue structure are known in the symphysis of extant mammals (Scapino 1965, 1981). The two unfused mandibles were mobile relative to each other during jaw movement, as generally is the case in extant mammals with an unfused symphysis (Crompton and Hylander 1986; Scott et al. 2012). This is a prerequisite for the hemi-mandible to rotate on its long axis, for better matching of the occlusal surfaces of the tooth crowns during occlusion (Crompton 1974; Schultz and Martin 2011). The unfused symphysis of Henkelotherium (Figs. 8, 9, 12, 13, 14, 15) and Dryolestes (Martin 1999) is a general feature and plesiomorphy of other Mesozoic mammaliaforms (Kermack et al. 1973; Cifelli and Madsen 1999; Krebs 1991; Martin 1999).

Meckel’s sulcus

A distinctive Meckel’s sulcus is present on the medial surface of the mandibular body in the majority of the examined mandibles. The anterior part of the sulcus appears to extend into the rugose area of the mandibular symphysis in juveniles and some adults (Gui Mam 44/80; Gui Mam 133/77) (Figs. 6d and 9d). The sulcus extends posteriorly from the symphysis towards the margin of the pterygoid fossa, and either ends just below the posterior opening of the mandibular canal, or fades out below m5 without reaching the pterygoid fossa margin. However, the extent of this sulcus is dependent on growth stages, and is more prominent in juvenile and subadult specimens. It can be reduced or absent in older individuals with late erupting fifth to seventh molars, and with a longer mandibular body. When present, the sulcus is a shallow but very distinct groove, wider (relative to the mandibular depth) and fully continuous along the mandible in juvenile and young adult specimens (Gui Mam 44/80; Gui Mam 133/77). However, we note that the sulcus is variable in other individuals; it is fully visible in the left and right partial mandibles of Gui Mam 78/76 (a mandible of five molars). However, in Gui Mam 78/76 (with six molars), the anterior part of Meckel’s sulcus appears to be a weak and faint line from the symphysis, and the posterior part of the sulcus is absent (Fig. 15). In the longer jaws of the examined sample, with more late-erupted molars (Gui Mam 48/78, 65/77; 138/76 holotype) (Figs. 1, 12, 13, 14), the sulcus is absent, which represents the extreme of a wide range of variability of the Meckel’s sulcus.

In specimens of Henkelotherium in which Meckel’s sulcus is fully present, the sulcus is parallel to the ventral margin of the mandible, as in the dryolestids Dryolestes leiriensis and Krebsotherium (Martin 1999). Among mammaliaforms, the Meckel’s sulcus can either be parallel to the mandibular margin, such as in Sinoconodon and many eutriconodontans and spalacotherioids (Simpson 1928; Luo et al. 2007a, b; Gao et al. 2009; Ji et al. 2009; Meng et al. 2011; Mao et al. 2020), or intersecting with the ventral margin, as seen in Morganucodon and several docodontans (Kermack et al. 1973; Lillegraven and Krusat 1991; Luo 1994; Schultz et al. 2019; Panciroli et al. 2019). Henkelotherium shows the former condition (Fig. 8) (Krebs 1991).

The dryolestidan taxa show interspecific variation of Meckel’s sulcus. In Dryolestes leiriensis, the Meckel’s sulcus is present in all specimens, without exception (Krebs 1971; Martin 1999). The sulcus is also present in Krebsotherium (Martin 1999, 2000). Previously the sulcus was interpreted to be present only in the posterior-most part of the mandible in the Cretaceous dryolestid Crusafontia (Henkel and Krebs 1969; Krebs 1971), but a new CT visualization shows this sulcus to be very weak, if present at all (Martin et al. 2022b: figs. 11 and 12). The sulcus is absent in the dryolestid Beckumia (Martin et al. 2022b). Meckel’s sulcus is also well developed in the Late Cretaceous meridiolestid Cronopio (Rougier et al. 2011), but can be variable among other meridiolestidans, an endemic clade of South America (Rougier et al. 2021a).

Phylogenetic variation of Meckel’s sulcus is not uncommon for other Mesozoic stem therian groups (Luo 2011; Meng et al. 2011). At least two taxa of spalacotherioid “symmetrodontans” have a fully developed Meckel’s sulcus, linked with ossified Meckel’s element (Ji et al. 2009; Zhou et al. 2019; Mao et al. 2020), which is connected to the middle ear (Zhou et al. 2019; Luo and Manley 2020). By comparison, the more derived taxa of the same spalacotherioid clade lack this sulcus altogether (Cifelli and Madsen 1999; Tsubamoto et al. 2004). More recent discoveries show that the ossified Meckel’s element is variably retained in Mesozoic eutherian mammals (Wang et al. 2022), leaving a definitive sulcus in some specimens, but not in other taxa of the same group (Urban et al. 2017; Wang et al. 2022). The variability of Meckel’s sulcus here in Henkelotherium is consistent with the recent findings in eutherians (Wang et al. 2022).

The Meckel’s sulcus undergoes ontogenetic change in extant mammals. The sulcus is the interim contact site for the Meckel’s element, which is fully cartilaginous and associated with the mandible during embryonic and fetal stages in extant mammals (Anthwal et al. 2017; Urban et al. 2017). The sulcus on the surface of the mandible disappears after the bone remodeling in the aftermath of the incorporation of the middle part of Meckel’s cartilage into the mandible, and the resorption of the posterior part of the cartilage, in neonates of extant mammals (Zeller 1989; Anthwal et al. 2017; Urban et al. 2017).

Variation of the Meckel’s sulcus among individuals of Henkelotherium, and the systematic variation among taxa of dryolestidans may be interpreted in one of two scenarios. The first and more likely scenario is that the groove remained on the mandible after a cartilaginous Meckel’s element was re-absorbed in adults, and the sulcus has remained as a vestigial structure. The variability of this sulcus would be the consequence of uneven remodeling of the mandible bone post-resorption of the Meckel’s cartilage in neonate stages (Ramirez-Chavez et al. 2016; Anthwal et al. 2017).

An alternative scenario is also possible (although not probable) that Meckel’s element had been ossified, as in the rare preservation of this element in the eutherian Cokotherium (Wang et al. 2022), but the ossified Meckel’s element was so flimsy that it was not preserved in mandibular fossils. Several eutriconodontan mammals are known to have preserved an ossified Meckel’s cartilage with its original connection to the mandible (Wang et al. 2001; Luo et al. 2007a, b; Meng et al. 2011; Luo 2011; Mao et al. 2020). Among trechnotherian mammals (inclusive of dryolestidans), there are two spalacotherioid species, Maotherium asiaticum and Origolestes lii (Ji et al. 2009; Zhou et al. 2019; Plogschties and Martin 2020; Mao et al. 2020) in which an ossified Meckel’s element is connected to the relatively wide Meckel’s sulcus, and fully connected to the middle ear (Luo and Manley 2020). However, the relatively wide Meckel’s sulcus of spalacotherioids and eutriconodontans is quite different from the thin Meckel’s sulci in dryolestidans examined here. By this conspicuous difference, we argue that it is not probable that Meckel’s cartilage was ossified in life in Henkelotherium, because its Meckel’s sulcus is so thin and different from the broader sulcus of spalacotherioids and eutriconodontans known to have accommodated the stout Meckel’s element in full connection to the middle ear.

In earlier studies, which did not have the benefit of extensive new fossils of Meckel’s element in Mesozoic mammals, it was hypothesized that the sulcus would hold the mylohyoid nerve and artery, in addition to the association of Meckel’s cartilage (Krebs 1971), but given the strong fossil evidence in the last two decades, this earlier interpretation (Krebs 1971) is no longer tenable (see the discussion by Anthwal et al. 2017).

Comment on “splenial scar”

The splenial scar, as interpreted for Dryolestes leiriensis (Krebs 1971; Martin 1995, 1999), is a narrow triangular area anterior to the pterygoid fossa and between two faint grooves flanking the posterior-most part of the Meckel’s sulcus. This “scar” is interpreted to be the contact area for the vestigial splenial bone, although the splenial bone itself is not preserved in dryolestids (Krebs 1971, 1991; Martin 1995: Fig. 1, 1999, 2000). In specimens of Henkelotherium, this “scar” can be identified in one juvenile specimen (Gui Mam 44/80), where there is a part of the Meckel’s sulcus, which is wider in young individuals. However, this splenial “scar” is absent in the five well-preserved mandibles (including the holotype: Figs. 12, 13, 17) in which the periosteal surface of the mandible is intact and does not show any trace of a “scar” anterior to the pterygoid fossa. Other mandibles (more than seven specimens) are not well preserved enough, due to fractures and damages in this location, to be informative about the “splenial scar.” Given the wider sampling of more specimens, we interpret that the “splenial scar” is a variable feature of juveniles, and a part of the wider Meckel’s sulcus of early growth stages. However, it is absent in fully adult mandibles. Further, the splenial bone itself is not present in Henkelotherium, as the bone is also unknown in Dryolestes leiriensis (Krebs 1971; Martin 1999).

Coronoid scar

The scar of the coronoid fossa or facet is located at the corner between the crest-like anterior margin of the coronoid plate and the mandibular body, and the scar borders on the pterygoid fossa. The coronoid scar is a slightly rugose area on a platform slightly elevated from the surrounding bone. When well preserved, the outline of the rugose surface of the scar shows a roughly triangular shape in medial view (Gui Mam 133/77, 48/78) (Figs. 6, 13), but its outline can be variable, with a half-circle outline, or a narrow strap (Gui Mam 15/77), and even irregular among specimens in differing conditions of preservation. Despite the preservational variation, it is always present in all specimens in which the relevant region is preserved, regardless of the growth stage of a mandible. We have confirmed the coronoid fossa in eight mandibles in which this region is preserved (Appendix). The coronoid scar is a consistent feature of all dryolestids, as noted first by Krebs (1969, 1971) and corroborated by observation of Dryolestes leiriensis and Krebsotherium lusitanicum (Martin 1995: Fig. 1, 1999; Figs. 8a, 9, and 18a).

Pterygoid fossa

The pterygoid fossa is a broad and shallow depression on the medial surface of the posterior region of the mandible (Figs. 1, 7, and 8). The dorsal part of the pterygoid fossa is bound anteriorly by a crest-like anterior margin on the coronoid process. The surface of the fossa is overall flat, but slightly convex medially in its dorsal part and more concave in the ventral part. Below the raised area of the coronoid fossa, it is bound by a curved margin that continues to the ventral border of the mandible; more posteriorly, the pterygoid fossa margin curves postero-dorsally and extends further to the articular condyle. The ventral and posterior margins of the pterygoid fossa are ridge-like and very distinctive in most specimens. The slightly convex dorsal part of the pterygoid fossa is likely for the insertion of the deep temporalis muscle which inserts on both the medial and lateral sides of the coronoid process. The distinctive ridge-like ventral and posterior margins of the pterygoid fossa are likely for insertion of the deep pterygoid muscle (M. pterygoideus internus) which may have further expanded to the inflected angular process (Turnbull 1970; Lautenschlager et al. 2017; Grossnickle 2017). The mandibular foramen (“Foramen dentale” of Martin 1999) is located in the deepest part in the ventral area of the fossa, just opposite to the posterior end of Meckel’s sulcus in juvenile specimens. The mandibular foramen is the posterior opening of the mandibular canal for entrance of the inferior alveolar nerve and its associated vessels into the mandible (Figs. 6, 7, 9, and 13).

Coronoid process and masseteric fossa

The coronoid process is plate-like, and dorsally curved to a posterior angle (preserved in Gui Mam 133/77) (Fig. 6). In side view, the lower part of the anterior border of the coronoid process is straight, but the top part of the coronoid margin curves posteriorly and forms a recurved apex of this process (Fig. 1; Appendix). The apophysis or the posterodorsal apex of the coronoid process is present, at least in one specimen (Gui Mam 133/77).

The straight part of the anterior margin forms an angle to the molar alveolar line from 50° to 65° (measured in lateral view on Gui Mam 48/78 and the holotype, reported by Krebs 1991). This angle is much smaller than the 80° to 90° angle in Dryolestes and Crusafontia, but overlaps the 60° angle in Krebsotherium (Martin 1999).

The masseteric fossa is well developed on the lateral aspect of the coronoid process. The fossa is bound anteriorly by the sharp anterior crest on the coronoid process, and the anterior margin of the fossa is flaring antero-laterally. The fossa is bound posteriorly by the gently convex lateral crest (= lateral ridge) (Figs. 1 and 6). The masseter fossa shows a gradational transition to the lateral side of the mandibular body, without a crest to demarcate its margin. Similar to other dryolestidans especially dryolestids (Krebs 1971; Martin 1999), the masseter fossa does not extend anteriorly onto the mandibular body, as in multituberculate mammals (Hahn and Hahn 2000; Kielan-Jaworowska and Hurum 2001), and it has no lateral mandibular foramen in the deep part of the masseteric fossa as in the Cretaceous stem zatherian Peramus, and the eutherians Prokennalestes and Juramaia (Kielan-Jaworowska and Dashzeveg 1989; Rougier et al. 1998; Luo et al. 2011).

The coronoid process shows an antecoronoid depression in its front base. Flanked by the lateral flaring of the anterior masseter margin and medially by the platform of the coronoid fossa, this depression is best seen in dorsal view (Figs. 1, 6, and 12). The antecoronoid depression is also present in Dryolestes leiriensis (Martin 1999) and Beckumia (Martin et al. 2022b), although not well developed in Crusafontia. This is a derived feature associated with the strong development of the crest for the anterior margin of the masseter fossa.

Below the recurved angle (= the apex) of the coronoid process, the posterior notch of the coronoid is strongly concave, but it has no crest in medial or lateral aspects. On the lateral aspect of the mandible, a low, broad crest arises from the ventral margin of the mandible to the articular condyle. The lateral crest is interpreted here to be the equivalent of the lateral ridge of the articular process of mammaliaforms (Crompton and Luo 1993; Schultz et al. 2019; Panciroli et al. 2019; Luo et al. 2022) (Figs. 1 and 6).

Articular process (“dentary condyle”)

The articular process is oriented postero-dorsally and it terminates in the dentary condyle. The condyle is raised well above the molar alveolar line. The dentary peduncle (sensu Luo et al. 2002), a neck-like constriction between the condyle and the rest of the coronoid part of mandible, is noticeable, although short and stout. The condyle and the peduncle are reinforced by a lateral crest forming the boundary of the masseter fossa on the lateral side, and are supported on the medial side by the crest-like posterior margin of the pterygoid fossa (Figs. 1 and 7). The condyle is cylindrical in dorsal view but shows a triangular outline in posterior view (Figs. 1 and 7) (Krebs 1991). The condyle appears to have a larger and more rounded lateral end, than medial end, due to being more expanded and rounded laterally than medially. Below the medial rim of the articular condyle, there is a small and slightly depressed area, likely to be for the insertion of the lateral pterygoid muscle or attachment of its associated tendon. A similar depression on the medial side of the articular process is present in Dryolestes. The articular process is similar in most characteristics to that of Dryolestes (Martin 1999).

Angular process

The angular process reaches posterior to the level of the dentary condyle but is separated from the dentary condyle by a wide, angular embayment, formed by the concave posterior margin of the dentary (sensu Martin 1999) (Figs. 1 and 6). The main part of the angular process is medially inflected and this inflection is best seen in ventral view, and also in posterior view (Figs. 1, 7, and 12). The ventral mandibular margin anterior to the angular process is noticeably curved (concave) in side view (Fig. 1).

The lateral face of the base of the angular process shows a flat area, which is bound by the posterior margin of the masseteric fossa, and by the lateral crest to the dentary condyle (Fig. 6a). The medial aspect of the inflected angular process is continuous with the prominent ventral margin of the pterygoid fossa. Near the base of the angular process is a triangular depression between the sharp crest of the pterygoid fossa posterior margin and the angular process. This depression houses the angular foramen, a vascular conduit to the trabecular space in the interior of the bone (Figs. 6, 7, 12, and 13). The angular foramen is present in all specimens of Henkelotherium where this region is preserved. This feature is generally absent in other dryolestidans. Among other Jurassic mammals, this foramen is present in the stem zatherian Nanolestes from the Guimarota mammal assemblage (Martin 2002), an exception and the only other example of this structure outside dryolestidans.

Henkelotherium differs from dryolestids in that the angular process of the latter group is not medially inflected, and has a straight and leveled ventral margin (Krebs 1971; Martin 1999). Henkelotherium is similar to meridiolestidans in the inflection of the angular process (Rougier et al. 2011, 2021a). Both the meridiolestidan Cronopio, and Necrolestes that is hypothesized to be related to meridiolestidans, have a prominent, medially inflected angular process (Asher et al. 2007; Rougier et al. 2012), although the degree of inflection and size of the angular process in Henkelotherium are much less than those of meridiolestidans. Averianov et al. (2013) re-interpreted that the inflected angular process was not present in meridiolestidans. Based on a review of meridiolestid specimens by Luo, we disagree with this interpretation. The degree of inflection of the angular process in Henkelotherium is similar to that of the mammaliaform Hadrocodium (Luo et al. 2001, 2022). One spalacotherioid mammal, Spalacolestes (Cifelli and Madsen 1999), has a medially flaring ventral margin of the pterygoid fossa, but the tip of the angular region of this fossil is not preserved to suggest a continuity of this margin to an inflected angle (Cifelli and Madsen 1999: Fig. 8), as in Henkelotherium. An inflected angular process continuous with the angular process is well known for metatherians (Sánchez-Villagra and Smith 1997), and the angular process and its related pterygoid shelf are attachment site for the medial pterygoid muscles in extant therians (Turnbull 1970; Lautenschlager et al. 2017; Grossnickle 2017). Here, Henkelotherium is interpreted similarly.

Interior structure of mandible

CT scanning and visualization have revealed the interior trabecular structure and spaces for periodontal attachment of the teeth, nerve conduit, and vascular channels. These hollow spaces are present inside the dentary condyle (Figs. 6 and 13: condylar plexus), and along the ventral part of the mandible through the entire length of the mandibular body to the symphysis, and in the angular region (angular plexus) (Figs. 6 and 7).

The roots of the lower teeth are held by their respective root alveoli that open ventrally into the shared space of the periodontal ligament (PDL) and inferior alveolar nerve (AIN) along the entire length of the mandibular body (Figs. 6 and 7: AIN-PDL spaces). The ventral one-third to one-half of the tooth roots are in the hollow space for the inferior alveolar nerve and its associated blood vessels, and the top one-half of the alveolar tubes are for soft tissues, holding the periodontal ligament (Figs. 6 and 7: PDL space) that is confluent with the mandibular canal.

In extant mammals, the mandibular canal houses the inferior alveolar nerve (IAN), a branch of the mandibular nerve of the trigeminus (cranial nerve V-3), and the major nerve that innervates the mandible and the lower teeth (Evans 1993; Dyce et al. 1995). We interpret that the IAN nerve entered the mandible through the mandibular foramen (“Foramen dentale”), and traversed the mandibular canal space along the root tips of the lower teeth (Fig. 7a: mandibular “canal”), through to the symphyseal region of the mandible, in Henkelotherium as in extant mammals. This nerve gave rise to the mental nerves that exited through mental foramina on the lateral surfaces of the mandible (as described earlier), and one branch of the nerve entered the symphyseal region through the symphyseal foramen. The hollowed space, beside the distal half and one-third of the roots (Fig. 7c), was filled with blood vessels, periodontal ligaments, and the associated soft tissues. The angular plexus inside the angular process is connected to the mandibular canal, and to the outside of the mandible via a vascular channel through the angular foramen.

The trabecular spaces associated with the mandibular canal and root alveoli are now documented in the dryolestids Crusafontia and Beckumia (Martin et al. 2022b), and are also newly visualized in Dryolestes here. Dryolestes is similar to Henkelotherium to the extent that the interior trabecular space, in the angular plexus inside the angular process, is anteriorly connected to the confluent space of the mandibular canal for the inferior alveolar nerve, its companion vessels with the periodontal ligament and related soft tissues. However, there are some minor differences: in Henkelotherium, this space is posteriorly connected to the angular foramen, a vascular conduit (Figs. 6 and 7: IAN-PDL space). In dryolestids, the angular plexus is present, but smaller relative to the size of the jaw, and not connected to the outside of the mandible, as the dryolestids examined so far lack an angular foramen.

Antemolar replacement/molar eruption

The several young and juvenile mandibles have made it feasible to establish a growth series and to interpret tooth replacement and eruption, in the context of mandibular growth (Figs. 7, 8, 9, and 11). The youngest specimen (Gui Mam 25/81; Figs. 8e and 11) is represented by the shortest of jaws (preserved length 11 mm) available for this study. This young individual has preserved the deciduous canine and deciduous premolars dp1 through dp4, with a cusp of permanent p1 and cap of permanent p3 (Figs. 8e and 11). This youngest specimen (Gui Mam 25/81) has preserved the two-rooted alveoli for m1 and m2—two root sockets for each tooth (although m1 and m2 are lost), plus an unerupted m3 crown at bell stage inside its crypt (Figs. 8e and 11).

The next juvenile specimen, represented by Gui Mam 44/80 (Fig. 8), suggests that canine, p1 and p3 have already replaced the deciduous canine, dp1 and dp3. However, this specimen has a developing (permanent) p4 in a crypt near the roots of the dp4 (Figs. 7a and 9), suggesting that dp4/p4 replacement had not been completed. We interpret the empty alveoli as for dp2. Under this interpretation, the replacements of dp2/p2 and dp4/p4 would occur in alternate canine/premolar loci of canine, p1 and p3. This suggests that Henkelotherium has a pattern of replacement at alternating premolar positions (p1 → p3, then p2 → p4), similar to that of Dryolestes, and of Zhangheotherium and Maotherium (Martin 1997; Luo and Ji 2005; Mao et al. 2020). Replacement of the canine and premolars at alternate loci appears to be a consistent feature of trechnotherian mammals through to Cretaceous eutherians (Kielan-Jaworowska 1981; Luckett 1993; Martin 1997; Luo et al. 2004). This alternating replacement is in contrast to the sequential replacement of premolars of mammaliaforms, which is interpreted to be plesiomorphic for mammaliaforms (Luo et al. 2004; Schultz et al. 2019).

As revealed by CT scans of the Gui Mam 44/80 specimen (Figs. 7, 8, 9), there is a developing m4 in the thickened bone in the posterior part of the mandibular body. The developing m4 crown has fully formed its trigonid at the bell stage of development (Figs. 7, 8, 9). CT scans have also revealed an empty space interpreted to be the m5 crypt, although an m5 crown is not detected, in this juvenile mandible. The m5 crypt is located inside the coronoid platform and posterior to the m4 bell crown (Fig. 9).

The two juvenile specimens here provide definitive evidence that the replacement sequence of premolars coincides with the successive eruption of m1 through m3; the replacement of dp4 by permanent p4 occurs approximately in synchrony with the eruption of m4. After replacement of all antemolars (incisors, canine and premolars) and eruption of m1–4, the later and successive eruption of m5 through m7 would further add to the molar row, concurrent with a prolonged growth of the mandible along the mandibular length (Fig. 8). The ramification of this mandibular growth pattern and similar pattern in the upper tooth row will be further explored in the “Discussion” section.

Upper dentition

Incisors

The ultimate right upper incisor is the only upper incisor preserved in the holotype specimen (Krebs 1991: 31) (Fig. 17). No other upper incisors are preserved in our study sample. The ultimate incisor is small, and shows a simple main cusp and a small mesial accessorial cuspule. The incisor has a long root (more than four times the crown height), which is curved distally and somewhat linguo-labially compressed (Krebs 1991: 36f). The tooth is associated with a bone fragment, here interpreted to be an incomplete premaxilla (Fig. 17, see also Krebs 1991). However, it cannot be ruled out that this incisor is in the maxilla, because the premaxilla–maxilla suture is not preserved in the type specimen. For the closely related Drescheratherium acutum, one, or possibly two, maxillary incisors have been reported (Krebs 1998: Figs. 1 and 2). Also, in the dryolestid Dryolestes leiriensis, the ultimate upper incisor (I5) is implanted in the maxilla (Martin 1999). The presence of maxillary incisors is a common feature for dryolestidans.

Canine

In the holotype specimen (Gui Mam 138/76), the upper canine crown is partially exposed from the lingual side (Fig. 17a); the labial side is visible through the transparent plastic matrix (Krebs 1991: 37). Our CT visualization of fully adult specimens of H. guimarotae shows that the upper canine has two long and fully divided and divergent roots, as exemplified by Gui Mam 114/77, or a two-rooted canine was indicated by two divided alveolar grooves as visualized for Gui Mam 17/76 (Figs. 19, 20, 21). The distal root is slightly stronger and straighter than the mesial one. In a juvenile (Gui Mam 91/75, with four molars), the crown of the canine has erupted, but the roots are not yet fully formed—this indicates that the full development of roots of the permanent canine occurs in late growth stages when M5 and M6 erupt (Fig. 18).

The main cusp is acutely triangular, and the crown is tall and laterally compressed with a convex labial side and slightly convex lingual side, as described previously (Krebs 1991). The crown is recurved distally, and its mesial edge is convex and rounded, and the distal edge is slightly concave with a weak crest. The canine bears a small drop shaped apical wear facet that faces distally. On the mesial side and above the crown base, there is an accessorial cuspule which continues into a short cingular ledge onto the labial side. On the distal side sits a tiny accessory cuspule at the cervical end of the vertical distal crest. This cuspule is lingually placed, and slightly more elevated than the mesial cuspule.

Premolars

Krebs (1991: 37f) reported four upper premolars for the holotype specimen. This is now corroborated by new CT visualization (Fig. 17). Three premolars are preserved in situ (P1, P3, and P4), while P2 fell off during preparation and is conserved separately from the main specimen. The P2 position was interpreted to be represented by the gap between P1 and P3 (Krebs 1991: Fig. 3). According to Krebs (1991), P2, as the smallest upper premolar, is single rooted. Krebs’s (1991) interpretation can now be corroborated by our CT visualization revealing the P2 alveolus invisible on the surface, apparently fractured and shifted by a slight fault (Fig. 17c). P1, P3, and P4 have two roots in the holotype specimen, and can be seen from all aspects in the CT visualization.

In the holotype specimen, P1 is exposed only from the labial side. Here, we present supplementary information from Gui Mam 17/76 where all four premolars are preserved and are visible from all aspects. The main cusp of P1 is broad triangular and bilaterally compressed with the labial aspect somewhat more convex than the lingual aspect. The mesial margin is slightly convex with a small mesial accessorial cuspule near the crown base. A short cingular line is extending from the basal mesial cuspule posteriorly onto the lingual side until it fades in the middle. The distal margin of the crown is straight with an indistinct crest that ends in a distal accessorial cuspule at the crown base. The main cusp is somewhat pointing distally, but less so as illustrated by Krebs (1991: Fig. 3). The P1 bears a small apical wear facet which is oriented distally.

In Gui Mam 17/76, P2 is not yet fully erupted and corresponds in size and morphology to P1 except for a more pronounced distal heel that supports the distal cuspule. Differing from Krebs’s (1991) observation on the holotype specimen, P2 of Gui Mam 17/76 has two roots (Fig. 19).

The P3 main cusp is broadly triangular (Figs. 19 and 21). The crown is labio-lingually compressed with the labial side slightly more convex than the lingual side. The mesial margin of the crown is slightly convex and rounded, and the distal margin is concave with a crest which ends in a distal accessory cuspule sitting on a broad base. The main cusp is slightly pointing backwards and bears a small apical wear facet. On the mesial side, there is a small ledge near the crown base with an accessory cuspule. Although the crown of P3 is not yet fully erupted, its morphology corresponds to that of P1 except for a more pronounced distal cuspule. The P3 is twice the size of P1 and the estimated size of P2.

The P4 is the largest upper premolar and is fully erupted in Gui Mam 17/76 and 138/76 (Figs. 17, 19, and 21). The crown is only slightly compressed linguo-labially, and its main cusp is conical and is slightly pointing backwards and bears a small apical wear facet. The mesial margin of the crown is slightly convex in lateral view, and the distal margin is slightly concave with a distinct crest that ends in an accessory cuspule. The distal crest of P4 of Gui Mam 17/76 is not crenulated as suggested by (Krebs 1991). The labial aspect of the main cusp is convex, and the lingual aspect is mostly convex, except that the distolingual side is somewhat concave, but this concave curvature is less pronounced than was previously illustrated for Gui Mam 138/76 by Krebs (1991: Fig. 3). On the mesial margin of the crown base, there is a small ledge with a mesial accessory cuspule and an additional smaller cuspule on the lingual side. On the distal end of the crown base, there is a distinct heel bearing the distal cuspule and additional tiny knots.

Molars

Here, we provide new visual information from the new specimens of our study sample (Figs. 18, 19, 20, 21, 22, 23, 24, 25, 26, 27), beyond the previous illustrations and photographs on lingual or occlusolingual aspect of the teeth for the holotype specimen (Krebs 1991: Fig. 3). In several specimens (Gui Mam 91/75, 138/76, 17/76, 79/77, 9/82, and 114/77) (Figs. 17, 18, 19, 23, 24, 25, 26, 27), the size of the molars increases from M1 to M2; M3 and M4 are the largest, M5 corresponds in size to M2, and M6 would be the smallest as indicated by its small socket in the holotype and very small root in Gui Mam 17/76. In proportions, M1 is relatively the narrowest (labio-lingually) and longest (mesio-distally); M2 to M5 are increasingly wider and shorter. In five of the six examined specimens with upper molars, the parastylar corner of each succeeding molar overlaps the distolabial corner of the preceding molar, such that the upper molars form staggered contacts between the adjacent molars (Figs. 18, 23, 24, 25, 26, 27). The paracone of the more posterior molars (especially M4 and M5) appears to be slightly inclined anteriorly (Figs. 15, 20, 21, 23, and 25). We note that this staggered contact between upper molars and the changing orientation of the paracone constitute a morphological gradient along upper molars in many dryolestidans (Prothero 1981; Martin 1999), a distinctive feature of dryolestidans (Fig. 27).

Of the primary trigon, the paracone is the largest and highest cusp, followed by the stylocone and metacone (Figs. 2 and 22). The paracone is robust, and has a smooth, rounded lingual side that bears a slight bulge near the crown base on M2 and M3 (Figs. 18, 19, 22, 23, 24, 25), but this bulging can be indistinct or absent in other molars. The stylocone sits near the labial end of the preparacrista in the anterior molars (M1-2), and along the labial side, it marks the anterior end of the labial cingulum (Figs. 2a–d and 22a). In the more posterior molars, beginning with either M3 or M4, the stylocone is slightly shifted in lingual direction and no longer directly connected to the labial cingulum (Figs. 2h–i, 23, 24, 25, 26). In M4, the stylocone may develop a short cingular crest reaching towards the anterolingual aspect of the crown. In M5-6, the stylocone is shifted further in lingual direction, and becomes more separated by a curved groove around the labial side of the stylocone from the periphery of the tooth crown. In M4 and M5 of Gui Mam 79/77, the stylocone is strongly worn (Fig. 23). The preparacrista connects the tip of the paracone with the stylocone and borders the primary trigon mesially (Fig. 2a–i). It is strongly worn with dentine exposure in M1 (e.g., Gui Mam 9/82) and successively less worn in the more posterior molars. The postparacrista connects the paracone with the metacone which is more of a swelling on the postparacrista than a cusp and is mesio-distally compressed (Figs. 2a–h and 22a). The crest continues as metacrista to the metastyle which is shifted labially. Within the primary trigon basin, a bulge-like median ridge is running from the paracone in labial direction, and fades before reaching the labial edge of the crown (Fig. 2a–h). The median ridge varies somewhat in expression among different specimens, and is most strongly developed—almost crest-like—in the M4 of Gui Mam 79/77 (Fig. 22). The labial border of the crown is straight, except for the M5 of Gui Mam 79/77 where it exhibits a weakly developed ectoflexus (Fig. 22a). The labial cingulum bears a tiny mesostyle in the middle between stylocone and metastyle (Fig. 2a–h), and this small cusp can vary in size and expression between specimens. Close to the metastyle, a minute stylar cusp D can be variably present (e.g., in M3) (Fig. 2g). The parastylar lobe is well developed and separated from the labial cingulum in M1 (Fig. 2a, b). However, in the more posterior molars, it becomes more fully connected to the labial cingulum (Fig. 2c–h). The parastylar lobe and its cusp (if present) are larger and can bear strong apical wear in M1 through M3, but become much smaller in M4 and M5 (Figs. 23, 24, 25, 26).

The upper molars have three roots (Figs. 2, 4a, d, and 5), a lingual root supporting the paracone, and a mesiolabial root supporting the parastyle, as well as a distolabial root supporting the metastyle. The lingual root is the strongest, and the mesiolabial and distolabial roots are somewhat smaller (Figs. 2, 4a, d, and 5). The roots of M1 and M2 are separated from each other. However, on M3–M5 of Gui Mam 17/76, the three roots are connected by Y-shaped dentine ridges in the cervical region. The Y-shaped dentine ridges are variable in strength in different molars, and are best developed in the middle molars (M3, M4, and perhaps also M5), and also in older specimens with more fully erupted molars. By comparison, in Dryolestidae, these Y-shaped ridges between the upper molar roots are more strongly developed and more uniformly present along the molar series. In more derived taxa of the Dryolestidae, the labial roots can even be fused in posterior upper molars (Fig. 5), but in all Henkelotherium specimens, the mesiolabial and distolabial roots are always fully separated (Figs. 5a–d and 19). In Henkelotherium and several dryolestids (including Dryolestes), the lingual root is curved mesially, and this curvature is related to the successive rotation of the posterior upper molars (Fig. 27).

Molar wear pattern

In Gui Mam 9/82, the upper molars are strongly worn (Figs. 24 and 26). The paracone bears a well-developed apical wear facet with dentin exposure on M1, whereas the paracone apices of M2-4 are only slightly worn on M2 and hardly worn on M4. The preparacrista is strongly worn only in M1; in M2–M4, it exhibits only slight-to-moderate wear. Wear on the postparacrista/metacrista is much stronger with dentin exposure than on the preparacrista in the molars on Gui Mam 9/82 (Figs. 24 and 26). The wear of the postparacrista and the metacrista extends along their entire length (metacone and metastyle completely worn off) in M1 and M2. This wear facet on M1 is so much developed that it further extends onto the parastyle of M2, shearing off the M2 parastylar cusp (Figs. 24 and 26). The preparacrista and stylocone of M2 are only moderately worn. Although the postparacrista/metacrista of M2 are strongly worn, the parastylar cusp of M3 exhibits only a small apical wear facet. The same wear pattern occurs between the distal side of M3 and parastyle of M4.

Gui Mam 114/77 and 79/77 (Figs. 22, 23, 25, and 26) are two relatively younger maxillary specimens as indicated by their molar eruptions (Fig. 26), with the preparacrista slightly-to-moderately worn on M1, and to a lesser extent also on M2. On the more posterior molars, which presumably erupted later than M1–2, the preparacrista is unworn. The wear of the stylocone is stronger in M3 and M4 of Gui Mam 79/77, whereas in Gui Mam 114/77, it has only a tiny apical wear facet in the same molar positions. On the postvallum wear pattern in the molars of Gui Mam 114/77 (Fig. 26), the metacone is moderately worn in M1–2. This cusp is worn enough that it becomes connected to the postparacrista/metacrista in M1-2, and the wear is more developed in M1-2, than in M3-4. The paracone is worn in the M1 of both Gui Mam 79/77 and in Gui Mam 114/77 (Figs. 2, 22, 23, 25, and 26), but this cusp is less worn on M2-4 (or at all on M4) that would follow much later in tooth eruption (only tiny apical wear facet in M4). The parastyle exhibits wear in the M1 in both specimens, although less worn in Gui Mam 79/77 than in Gui Mam 114/77. In the more posterior molars, it is unworn. The metastyle exhibits slight wear on the M1 of Gui Mam 114/77, whereas in the other molars of both specimens, it is without noticeable wear. This suggests that the molar wear can have intraspecific variation.

In Henkelotherium (as in dryolestids), the upper molars form an imbricating pattern, such that the parastylar lobe of the succeeding molar overlaps the distolabial side of the preceding molar (Figs. 18 and 27). The wear continues to develop on the postvallum of the preceding molar, and the postvallum can form a continuous wear facet with the parastyle, and even the entire parastylar lobe (e.g., between M1-M2 of Gui Mam 9/82) (Fig. 24). This extensive wear surface contiguous across adjacent upper molars can match the prevallid crest of the lower molar. Therefore, we infer that the imbricating interlock of the upper molars has a masticatory function (Fig. 27).

Maxillary alveolar structure for upper teeth

The maxilla of Gui Mam 17/76 (Figs. 19 and 20a) contains the infraorbital canal system. The infraorbital canal system in the maxilla is here interpreted to have two external (facial) openings: the well-preserved anterior opening shows a circular outline and is located above P3 and P4 and between these two premolars. The posterior opening is broken but is here interpreted to be located above M2. Both these openings are identified by comparison to a well-preserved maxilla specimen of Dryolestes leiriensis (Gui Mam 43/77) (Fig. 20b). Although the maxilla (Gui Mam 17/76) is incomplete and distorted by fractures, we interpret that the infraorbital canal in the maxilla has two facial openings for Henkelotherium, similar to that of Dryolestes (Fig. 20).

Of the two external openings of the maxilla in Dryolestes, one opening is posteriorly positioned above M2, which we identify to be the posterior opening of the infraorbital canal (Fig. 20b). This is equivalent to the single infraorbital foramen in therians, as it corresponds to the common location of this foramen in extant therians (Evans 1993; Dyce et al. 1995; Benoit et al. 2020). This is equivalent to the “infraorbital foramen 2” as previously interpreted for Morganucodon (Kermack et al. 1981), and more recently re-interpreted by Benoit et al. (2020) to be the only opening that carried the infraorbital nerve, as seen in extant therians. The other facial opening in the maxilla of Dryolestes is more anteriorly positioned, above P4 at the level of the P3-P4 junction (Fig. 20b). This foramen is much larger than the aforementioned “posterior” infraorbital opening, and almost twice the size of the latter. This large foramen indicates that a major nerve branch through this foramen innervated the facial part of the skull, in addition to the infraorbital nerve branch through the posterior opening. The anterior opening corresponds to the “infraorbital foramen 1” of Morganucodon as identified by Kermack et al. (1981). But more recently, this foramen has been re-interpreted by Benoit et al. (2020) to be the foramen of the rostral alveolar nerve (= superior anterior alveolar nerve in Fig. 20b). The large openings for the infraorbital nerve have been interpreted to indicate tactile sensory acuity for placental mammals (Muchlinski and Kirk 2017).

We have visualized the connecting network of canaliculi of the rostral alveolar nerve canal (= superior anterior alveolar canal in Fig. 20b). These are clearly connected to tiny canalicular network around the roots of the anterior molars and posterior premolars, likely for the branches of the superior middle alveolar nerve (and its associated vessels) to these upper teeth. The channel of the superior middle alveolar nerve (and its associated vessels) for the posterior premolars is separated from the channel for the superior anterior alveolar nerve (and its vessels) for the incisors, canine, and anterior premolars (Fig. 20b). Each premolar has two alveoli in the maxilla, and each three-rooted upper molar has three alveoli in the maxilla. The alveolar sockets of the more posterior molars are successively deeper, corresponding to successively longer roots in these posterior molars—M4–6 have the longest roots. The roots of the last two molars are twisted in both Henkelotherium (Fig. 19) and Dryolestes (Fig. 20b), with the larger lingual root bent anteriorly. The curvature of bending increases from the penultimate to the ultimate molar (Figs. 20b and 27b).

In the well-preserved maxilla of Dryolestes (Gui Mam 43/77) (Figs. 20b, 27a, and b), the bony walls of the alveolar sockets in the maxilla are trabeculated and associated with a network of interconnected small channels that are dorsally connected to the larger canals of the infraorbital canal system in the maxilla (Fig. 20b). The nerves and vessels in these trabecular bones provide innervation and vascular supplies for the roots of the upper teeth. The two maxillary specimens of Henkelotherium (Gui Mam 91/75 and 17/76) are incomplete, but enough of their maxillary tooth alveoli are preserved to show some resemblance to the trabecular characteristics in a better-preserved maxilla of Dryolestes (Gui Mam 43/77). We interpret that Henkelotherium has a similar network of trabecular bone and channel network in the maxilla to innervate and supply the upper teeth.