Mandibular histology and growth of the nonmammaliaform cynodont Tritylodon

Abstract

An investigation of bone microstructure of nonmammalian therapsids has revealed distinctive signals pertaining to their ontogenetic growth and biology. Until now, histological studies of the nonmammaliaform cynodonts have focused only on postcranial material. Through the examination of micro-computed tomography (micro-CT) scans and serial thin sections, the current study provides a novel perspective on the structure and growth of the mandible of Tritylodon, a derived herbivorous cynodont from southern Africa. By tracking histological features across the serial thin sections, trends in relocation and modelling are documented for the growing Tritylodon mandible. For example, during growth, localized changes in the cross-sectional shape of the mandible occurred. Localized deposits of new lamellar and fibrolamellar bone on the lateral edge indicate widening of the mandible during different episodes of growth. The presence of radial channels indicates the deepening of the mandible at its anterior and posterior ends. The relocation of the paired mental foramina suggests that the mandibular body lengthened mainly in the posterior direction. The medial movement of a posterior postcanine tooth during growth and eruption is recorded in the histology. This histological assessment also documents the presence of Sharpey’s fibres in the cellular cementum of the first incisor, providing novel and unequivocal evidence that it was attached to the Tritylodon jaw by a periodontal ligament. This is the first comprehensive study that uses histological analysis to document the growth dynamics of the mandible of a nonmammalian therapsid, thus providing a unique perspective of localized mandibular growth in a fossil animal.

Introduction

Tritylodontids were nonmammaliaform cynodonts that originated in the Late Triassic and persisted until the Early Cretaceous (Botha-Brink et al., 2012). The family Tritylodontidae includes genera such as the basal tritylodontid Oligokyphus (e.g. Kühne, 1956; Luo & Sun, 1993), Tritylodon (Owen, 1884; Simpson, 1928), Kayentatherium (Sues, 1986), Bienotheroides (Sun, 1984), Bocatherium (Clarke & Hopson, 1985), and the recently described Yuanotherium (Hu et al., 2009). The tritylodontids are considered phylogenetically close to the ancestry of mammals, but cranial and dental features, such as the absence of a dentary-squamosal contact and the presence of complex multi-cusped postcanines, may rule out a direct ancestral relationship to mammals (see Luo et al., 2002). Their phylogenetic position remains contentious, with some studies indicating that they are the sister group to mammaliaformes or that they are more distantly related (see review by Kemp, 2012). Superficially, the tritylodontid skull resembles that of mammals because the postorbital bar is lost, causing their orbits to be confluent with the enlarged temporal fenestrae (Kemp, 2005).

Studies of the postcranial bone histology of nonmammaliaform cynodonts (de Ricqles, 1969; Botha & Chinsamy, 2000, 2004, 2005; Ray et al., 2004; Chinsamy & Abdala, 2008; Botha-Brink et al., 2012) have provided much information pertaining to their biology. For example, the thick bone walls of the limbs of some cynodonts, e.g. Tritylodon (Botha, 2002) and Trirachodon (Botha & Chinsamy, 2004), suggest that they had a fossorial lifestyle. Studies of tritylodontid postcranial histology indicate that they generally formed bone in a rapid, uninterrupted rate (de Ricqles, 1969; Ray et al., 2004; Botha-Brink et al., 2012). Chinsamy & Hurum (2006) showed the peripheral development of parallel-fibred bone with rest lines in a radius, indicating determinate growth in Tritylodon, and more recently this was corroborated by a fibula that had closely spaced peripheral lines of arrested growth (LAGs) (Botha-Brink et al., 2012). All of these earlier studies of nonmammaliaform cynodont bone histology have been restricted to analyses of the postcranial skeleton. The current study instead focuses on the microstructural organization and growth of the Tritylodon mandible, and compares the observed patterns with that of mammalian cranial growth.

Much of our understanding of mammalian cranial growth is derived from the extensive morphological and histological work on the development of the primate cranium and mandible done by Enlow (1963, 1990; Enlow & Harris, 1964). Smaller histological studies have also been conducted on the mandible of the rabbit (Bang & Enlow, 1967) and mink (de Buffrénil & Pascal, 1984). All these earlier studies documented local growth fields on the cranial elements and the progressive, sequential movement of component parts of the whole bone during growth, a process termed relocation (Enlow, 1990). Curiously, only a few nonmammalian vertebrate studies have documented the microstructural anatomy between different ontogenetic stages (e.g. Hua & de Buffrénil, 1996) and, until now, there has been no histological analysis of the cranial growth patterns in a fossil vertebrate.

The current study presents histological observations of a serially thin-sectioned mandible of Tritylodon. Using the histological data supplemented by micro-computed tomography (micro-CT) analyses, we investigated the microanatomy and histology of the mandible of tritylodontids. Our findings permitted a direct assessment of the growth processes of the mandible, and by tracking distinctive histological features, we determined the direction of its overall growth. This study therefore provides novel insights into the mandibular growth dynamics of tritylodontids.

Osteological review of the tritylodontid mandible and teeth

The anatomy of the tritylodontid mandible and teeth has been well documented in the published literature (e.g. Sues, 1986); as such, we provide only an overview here.

The tritylodontid mandible is dominated by an enlarged, dorsally expanded coronoid process with an extensive masseteric fossa. The insertions of the superficial and deep masseter muscles into the masseteric fossa are separated by a lateral ridge (Sues, 1986; Fig. 1A). A single mental foramen is usually observed on the lateral side of the dentary, near the anterior-most postcanine (Kühne, 1956; Sues, 1986; Luo & Sun, 1993; Watabe et al., 2007). The mental foramen is the exit point of the N. mentalis, a branch of the N. alveolaris inferior (Sues, 1986) that is housed within the larger, ventrally located mandibular canal (Fig. 1B). The Meckelian or internal groove, a long sulcus on the medial side of the dentary, most likely accommodated the Meckelian cartilage and mandibular nerve branches (Sues, 1986).

Fig. 1.

A comparison of the immature SAM-PK-K1411 (A,B) and adult BP/1/5288 (C,D) Tritylodon mandibles in lateral (A,C,F,G) and medial (B,D) views. The arrowhead in (B) marks the posterior opening of the mandibular canal. (E) Dorsal view of a coarse three-dimensional reconstruction of specimen SAM-PK-K1411 from micro-CT scans. Note the lateral position of the unerupted ultimate postcanine tooth (bracket) relative to the tooth row. (F–G) Three-dimensional micro-CT reconstructions of specimens SAM-PK-K1411 (F) and BP/1/5288 (G), showing the position of the mental foramen/foramina, which were not clearly visible on the lateral surface of the specimens (framed region). (F) The mandible of SAM-PK-K1411 is rendered slightly translucent to show the position of the mental foramen and anterior portion of the mandibular canal (both infilled with dark grey). (G) Specimen BP/1/5288 showing the positions of the two mental foramina, the anteroventral mental foramen and posterodorsal accessory mental foramen. (H,I) Oblique transverse micro-CT slices through the anteroventral mental foramen (H) and posterodorsal accessory mental foramen (I) of BP/1/5288. Scale bar: 0.5 cm. cb, coronoid boss for the attachment of the coronoid bone; d, diastema; I₁, first lower incisor; I₂, broken second lower incisor; lr, lateral ridge; maf, masseteric fossa on the coronoid process; mc, mandibular canal (anterior part); mf, mental foramen; PC₁, first postcanine with an intact crown; Sp, splenial; sym, symphysis; t, trough for the postdentary bones.

As in other derived cynodonts, the postdentary bones in tritylodontids are greatly reduced, forming a delicate compound rod that slots into a groove on the medial side of the mandible (Crompton, 1963; Sues, 1986; Fig. 1B). The craniomandibular joint in tritylodontids is formed by the quadrate and articular, and unlike the other advanced cynodont group tritheledontids there is no contact between the dentary and squamosal bones (see Luo & Crompton, 1994; Luo et al., 2002 and references therein).

Tritylodontids have one to three upper procumbent incisors, and one to two lower procumbent incisors. Although they lack canines, the upper second pair and lower first pair of incisors are enlarged and superficially resemble canines (Sues, 1986; Kemp, 2005; Fig. 1C). The small amount of wear evident on the incisors of North American tritylodontids has led to the deduction that their function was grasping rather than gnawing (Sues, 1986). A diastema separates the incisors from the multicusped postcanines (Fig. 1A). Unlike most other nonmammaliaform cynodonts, the postcanines have multiple roots in the plesiomorphic condition, but the roots can secondarily coalesce into double roots, or even a single root in derived tritylodontids (Kühne, 1956; Cui & Sun, 1987; Luo, 1994). The mandibular postcanines have two longitudinal rows of crescentic cusps that face posteriorly, whereas the maxillary postcanines have three rows of cusps that are oriented anteriorly (Kemp, 2005). During occlusion, both sides of the lower jaws equally occluded with the upper jaws, and the posterior (horizontal) movement of the mandible enabled the shredding of food material (Crompton, 1972, 1995; Kemp, 2005). Wear facets on the postcanines of Kayentatherium confirmed the unidirectional horizontal movement of the tritylodontid lower jaw (Sues, 1986). It appears that the cusps wore down quickly, leaving flat crowns with longitudinal grooves (Crompton, 1972).

Scanning electron microscopy studies of tritylodontid enamel revealed the absence of prismatic structure (Wood & Stern, 1997). However, a tritylodontid from the Lower Cretaceous of Japan showed a prismatic enamel structure similar to that of mammals (Kamiya et al., 2006).

Unlike more basal cynodonts, tritylodontid postcanine teeth were not vertically replaced. Instead, postcanines in the anterior tooth positions were progressively lost, and new postcanines were added to the posterior end of the tooth row (Kühne, 1956; Crompton, 1972; Sues, 1986). Thus in older individuals, the diastema became progressively larger due to the loss of the anterior postcanines that were not replaced (Sues, 1986). The roots of the anterior (older) postcanines were exposed through resorption of the alveolar bone (Kühne, 1956; Sues, 1986). In adult individuals, the empty alveoli anterior to the existing tooth row may sometimes contain roots of previously shed teeth (Kühne, 1956). The ultimate (posterior-most) postcanine tooth is generally unerupted (Sues, 1986) and is positioned lateral relative to the mandibular tooth row (Sun, 1984; Fig. 1E). Despite the lack of vertical replacement of postcanine teeth, there is evidence for the direct replacement of the upper incisors in tritylodontids [e.g. immature specimens of Kayentatherium and Dinnebitodon (Sues, 1986)]. All of these observations were made from gross osteological assessment of tritylodontid specimens and have not been corroborated with micro-CT, histology, or serial sections of the tooth-bearing skeletal elements.

Institutional abbreviations

The following abbreviations for institutions are used: BP, Bernard Price Institute for Palaeontological Research, University of the Witwatersrand, Johannesburg, South Africa; SAM, Iziko South African Museum, Cape Town, South Africa.

Materials and methods

Gross osteology

The mandibular anatomy of three different-sized specimens of Tritylodon (SAM-PK-K1411, SAM-PK-K1330, and BP/1/5288) were investigated (Table 1 and Fig. 1). Specimen SAM-PK-K1330 (consisting of only the posterior half of the dentary) and the left mandible of BP/1/5288 are not well preserved. Photographs of specimen BP/1/4778, a medium-sized Tritylodon specimen (Fig. 2B) on display at the BPI museum, were also examined. Unfortunately, no complete specimens smaller than SAM-PK-K1411 were available for this research.

Table 1.

Tritylodon specimens, listed in increasing size, examined in the present study.

	SAM-PK-K407	SAM-PK-K1411	SAM-PK-K1330	BP/1/5288
Material	Serial sectioned L. dentary	L. dentary	L. dentary, cranium	R. dentary, cranium
Formation	Elliot	Upper Elliot	Upper Elliot	Middle Elliot
Length of mandibular body	n/a	69.6	n/a	87.0*
Height of mandibular body	n/a	17.9	18.1	24.9
Maximum height of mandible (includes coronoid process)	52.7 (slide P9)	53.7	52.5*	72.0*
Length of diastema	n/a	11.0	n/a	38.9
Length of the base of the coronoid process	n/a	26.0	28.5	41.7
Basal skull length	n/a	n/a	111.1	145.5
Width of postcanines
PC1	n/a	3.4	n/a	5.5
PC2	n/a	4.2	3.8	5.5
PC3	n/a	4.5	4.3	5.3
PC4	n/a	4.8	4.5	4.2
PC5	n/a	5.2	5.1	un
PC6	n/a	5.4	4.4**	–
PC7	n/a	un	un	–

Measurements are in millimetres.

n/a, not applicable (element is not preserved, or cannot be measured); un, unerupted tooth.

^*Approximate because of incomplete/broken element.

^**Postcanine is only partially erupted.

Fig. 2.

Comparison of the mandible in lateral view across different ontogenetic stages of Tritylodon. The mandibles are aligned to the anterior tip of the dentary. The outline of the mandibles was traced from photographs and the positions of the mental foramina in specimens SAM-PK-K1411 (Fig 1F) and BP/1/5288 (Fig 1G-I) were determined from the micro-CT scans. A photograph of BP/1/4778, a specimen on display at the BPI museum, was provided by F. Abdala. (A) SAM-PK-K1411, (B) BP/1/4778, (C) BP/1/5288 (the right mandible is mirrored). (D) A schematic of the displacement of the mental foramen from its anterior position in SAM-PK-K1411 to its more posterior position in BP/1/5288 (infilled grey foramen). Note that the diastema becomes progressively larger in older individuals because of the loss of the anterior postcanines.

In addition to these specimens, published illustrations of mandibular specimens of other tritylodontid genera (e.g. Sun, 1984; Sues, 1986; Luo & Sun, 1993; Watabe et al., 2007) were examined.

Micro-CT scans

To observe the internal anatomy, three mandibular specimens of Tritylodon were micro-CT scanned using a XT H 225 system (X-Sight, Stellenbosch, South Africa). The SAM specimens were scanned with an isotropic voxel size of 39.3 μm, whereas BP/1/5288 had an isotropic voxel size of 49.2 μm. The resolution of the digital micro-CT slices allowed a gross comparison with the histological serial slices, although fine histological detail was not visible in the digital slices.

Three-dimensional reconstructions of the slice data were undertaken using mimics software (version 14.0; Materialise, Leuven, Belgium). Due to memory constraints of the software, the entire dataset could not be imported. Instead, every fourth transverse (coronal) micro-CT slice was used for the SAM specimens, and every eleventh slice was used for specimen BP/1/5288.

Serial thin sections

Seventy serial thin sections of the mandible of Tritylodon sp. (SAM-PK-K407) were examined and photographed using a Nikon E200 Eclipse microscope and a Nikon DS-Fi1 digital camera. The specimen was collected from Early Jurassic sediments of Lesotho during the British-South African Palaeontological Expedition in 1961 and was later thin-sectioned by a student of A. W. Crompton at Harvard University. Unfortunately, after its return to the SAM, no data were available regarding the size of the specimen, number of teeth, and the orientation and interval spacing of the thin-sections. The mandible had been serially sectioned into 94 thin sections, but many of the original slides were missing, incomplete or damaged (Table 2). In our analysis, we renumbered the thin sections, starting from the posterior-most slide, in order to take into account the numerous missing slides (Table 2). Only the height of the coronoid process (Table 1), measured directly from the serial sections, could be compared with the gross osteology specimens, and it was found to be comparable in size to specimens SAM-PK-K1411 and -K1330.

Table 2.

Cross-reference of the original slide labels with the labels used in the current study.

Current study number	Original slide number	Original other number	Identification of teeth on slide	Current study number	Original slide number	Original other number	Identification of teeth on slide
n/a		F0		n/a	47
P70		F1		n/a	48
P69		F2	I1	n/a	49
P68		F3	I1	P35	50	B16
P67		F4	I1	P34	51	B15
P66		F5	I1	P33	52	B14
P65		F6	I1	P32	53	B13
P64		F7	I1	P31	54	B12
P63		F8	I1	P30	55	B11
P62		F9	I1	P29	56	B10
P61		F10	I1	P28	57	B9
P60		F11	I1, I2	n/a	58	B8
n/a		F12		P27	59	B7
P59		F13	I1, I2	P26	60	B6
P58		F14	I1, I2	P25	61	B5
n/a		F15		P24	62	B4
P57		F16	I1, I2	P23	63	B3
n/a		F17		P22	64	B2
n/a		F18		P21	65	B1
P56		F19	I1, I2	P20	66	C17
n/a		F20		P19	67	C16
P55		F21	I1, I2	P18	68	C15
P54		F22	I1, I2	n/a	69	C14
P53		F23	I1, I2	P17	70	C13	Penultimate PC
n/a		F24		P16	71	C12	Penultimate PC
P52		F25	I1	n/a	72	C11
P51		F26	I1	P15	73	C10	Penultimate PC
n/a		F27		n/a	74	C9
P50		F28	I1	P14	75	C8	Penultimate PC
P49		F29	PC1	P13	76	C7	Penultimate PC
P48	30	A17	PC1	P12	77	C6
P47	31	A16	PC1	n/a	78	C5
P46	32	A15	PC1	P11	79	C4	Ultimate PC
n/a	33			P10	80	C3	Ultimate PC
n/a	34			P9	81	C2	Ultimate PC
P45	35	A12		n/a	82	C1
P44	36	A11		P8	83
P43	37	A10		P7	84	2	Ultimate PC
P42	38	A9		P6	85		Ultimate PC
n/a	39			P5	86	4
P41	40	A7		P4	87
P40		A6		n/a	88	6
P39	42	A5		n/a	89
P38	43	A4		P3	90
P37	44	A3		P2	91	9
n/a	45			P1	92	10
P36	46	A1		n/a	93

I, incisor; n/a, not applicable because slide is missing or blank; PC, postcanine.

The sections were digitally scanned using a flatbed scanner (CanonScan 9000F), and organized into a consecutive sequence to determine the orientation of the thin sections. Comparisons with the micro-CT digital slices permitted the deduction that the sections were made at an oblique transverse angle relative to the long axis of the mandible. This important finding allowed us to determine that the thin sections with separate teeth located ventral to the erupted teeth were in fact part of the roots of the erupted teeth and did not represent developing replacement teeth.

Results

Gross osteology and micro-CT data

Mandible

In all specimens studied, the general osteology of the Tritylodon mandible was similar to that described for other tritylodontids in the previous section. It is worth noting that the cross-sectional profile of the mandible changes along its length. At the posterior end of the tooth row near the base of the coronoid process, the lateroventral part of the mandible is depressed because of the ventral extension of the masseteric fossa (Fig. 1A). Below the tooth row, there is a slight rounded ridge midway along the lateral edge of the mandibular body. At the anterior tip of the mandible, there is a slight expansion along the ventromedial edge that forms the symphysis (Fig. 1B).

The mental foramen on the lateral surface of the dentary of SAM-PK-K1411 and BP/1/5288 is not clearly visible by simple gross osteological examination (Fig. 1A,C) but it is obvious in the micro-CT slices (e.g. Fig. 1H,I). In SAM-PK-K1411, the slightly elongated mental foramen is situated ventral to the anterior-most postcanine (Figs 1F and 2A). On the right dentary of BP/1/5288, a smaller accessory branch of the mental foramen occurs posterodorsal to the mental foramen (Figs 1G–I and 2C) and these two foramina are situated ventral to the diastema and anterior to the anterior-most postcanine tooth. In both Tritylodon specimens, the mental foramen and the accessory mental foramen (if present) branch from the ventrally situated mandibular canal that runs along the length of the mandibular body (Fig. 1B,F,H,I). The accessory branch of the mental foramen in specimen BP/1/5288 is interpreted to represent an individual variation.

The postdentary rod is preserved separately in specimen BP/1/5288, but it is not present in the other study specimens. An elongated splenial bone is clearly preserved in specimen SAM-PK-K1411, where it partially covers the Meckelian groove (Fig. 1B).

Teeth

In SAM-PK-K1411, there are six erupted postcanines, and one unerupted postcanine posterior to the tooth row (Table 1, Fig. 1B). The anterior root of the anterior-most postcanine is fully exposed (Fig. 1A), although over-preparation of this specimen cannot be ruled out. The crowns of the procumbent large first incisor and the small second incisor are broken. The micro-CT scan revealed that the long root of the first incisor is open and that the canal width remains relatively uniform along its length, whereas the root of the second incisor becomes constricted at its distal end.

The tooth row is incomplete and poorly preserved in SAM-PK-K1330. In gross view, four fully erupted postcanines and one partially erupted posterior postcanine tooth are present. As this mandible is comparable in size to specimen SAM-PK-K1411, it is assumed that these teeth represent postcanines 2–6 (Table 1). Possible fragments of an additional posterior unerupted postcanine are visible in the micro-CT scan.

In BP/1/5288, the largest individual in our sample, there are four erupted postcanines, and an unerupted ultimate postcanine is visible only on the right dentary. The area anterior to the tooth row has been prepared extensively, and there are no postcanine crowns present in this region (Fig. 1C). However, the micro-CT scan of the right dentary show remnants of the roots of two postcanines anterior to the tooth row. These roots are surrounded by matrix dorsally, suggesting that the crowns were lost prior to fossilization and probably before death. The loss of the crowns of the two anterior-most postcanines has lengthened the diastema, but their roots were not fully resorbed. On both dentaries, the first incisor is large (Fig. 1C) and has an open, procumbent root that slightly constricts near its distal end, and the small broken second incisor is visible in the micro-CT scan of the right mandible.

The diastema is the longest in the largest sampled Tritylodon, specimen BP/1/5288 (Table 1). The erupted postcanine teeth increase in lingual-buccal width from the anterior to the posterior end of the jaw in the smaller specimens (SAM-PK-K1330 and -K1411; Table 1). However, in the largest specimen, the postcanine teeth decrease in width posteriorly (Table 1). These trends in tooth size are similar to those observed in other tritylodontids of different ontogenetic age (e.g. Sues, 1986) and suggest that specimens SAM-PK-K1330 and SAM-PK-K1411 are immature individuals, whereas BP/1/5288 is an adult individual.

In the serially sectioned specimen SAM-PK-K407, the number of postcanine teeth cannot be determined due to the incomplete sequence of sections and the oblique angle of sectioning (Table 2). The ultimate postcanine tooth is unerupted, and the cusps are fully developed but no roots have yet formed. Similar to the gross osteological specimens of Tritylodon, the first incisor is much larger than the second one. The long open root of the procumbent first incisor extends far posteriorly along the ventral base of the mandible (Table 2). A diastema (slides P49–P57) separates the erupted crown of the second incisor from the first postcanine, which has double roots and multiple cusps.

Histological description (SAM-PK-K407)

Mandible

Overall, the dentary is a composite of different bone tissues reflecting the local patterns of bone formation and growth. The specific bone tissues observed in the mandibular body, coronoid process, and coronoid boss are discussed in the following sections.

Mandibular body

Changes in the bone histology of the mandibular body are described from the posterior (slide P1) to the anterior end (P70) of the mandible.

In the posterior part of the mandibular body (slides P1–P4), the channels [which housed blood vessels and other connective tissues (Starck & Chinsamy, 2002)] tend to be radial along the medial and ventral edges of the mandibular body (Fig. 3C). In the more anterior sections (e.g. slide P7), the bone becomes more compacted but the central region, situated closer to the lateral edge, becomes more cancellous and secondarily resorbed. The channels also tend to be smaller and are longitudinally arranged (oriented in the anteroposterior direction; e.g. Fig. 3A). This type of bone pattern in the ventral part of the mandibular body is maintained for most of the sections until slide P49, where the channels once again become radially oriented.

Fig. 3.

Mandibular bone histology of Tritylodon sp. (SAM-PK-K407). (A) Section through the ventral base of the mandible (slide P24) showing a high concentration of primary osteons that are oriented longitudinally, parallel to the length of the jaw. Scale bar: 500 μm. (B) Section through the ventral base of the mandible (slide P51) showing radially oriented channels. The tooth on the left is the root of the first incisor. Scale bar: 500 μm. (C) Section through the ventral base of the mandible (slide P3) showing radially oriented channels. Scale bar: 500 μm. (D) Slide P33 showing the successive lamellar and fibrolamellar layers, separated by a rest line (arrowhead), along the lateral edge of the mandibular body. Scale bar: 1 mm. (E) Slide P27 showing different orientations of the lamellar bone along the lateral edge of the mandibular body. Scale bar: 1 mm. (F) Lamellar bone with dorsoventrally elongated channels is prevalent on the lateral side of the coronoid process (slide P1). Scale bar: 200 μm. (G) In some parts of the coronoid process, the central region is secondarily resorbed and forms large cancellous spaces (slide P14). Scale bar: 500 μm. (H) Slide P4 showing the porous bone texture of the coronoid boss located on the medial side of the base of the coronoid process. Also note the large cancellous spaces in the central part of the base of the coronoid process, which contrasts with the more compacted bone further dorsally in the coronoid process. Scale bar: 1 mm. btr, bone just posterior to the tooth row; cb, coronoid boss for the attachment of the coronoid bone; cp, coronoid process; LB, lamellar bone.

In most of the sections posterior to the open anteroventral mental foramina (slide P46), there are deposits of lamellar bone on or near the lateral edge of the mandibular body (Fig. 3D,E), whereas the medial edge of most sections tends to consist only of fibrolamellar bone. In several sections, a distinctive rest line near the lateral edge of the mandibular body was observed (e.g. slides P16, P18, P20–P22), which marks the lateral deposition of a thin layer of fibrolamellar bone followed by lamellar bone. The lamellar bone layer becomes progressively thicker in slides anterior to slide P18, and in some sections these deposits have different orientations (e.g. Fig. 3E). Starting with slide P24, the lamellar bone is followed laterally by another layer of fibrolamellar bone, and a distinctive rest line separates the two tissues (e.g. Fig. 3D). This more recently deposited layer of fibrolamellar bone has an uneven thickness (Fig. 3D); its lateral edge also appears irregular, suggesting a growing surface.

The inner region of the mandibular body becomes progressively cancellous in the slides anterior to P24. In slide P38, most of the fibrolamellar bone on the medial margin of the lateral wall, which was present in more anterior sections, appears to have been resorbed, and the lateral wall is essentially composed of more recently formed lamellar bone, followed by a fibrolamellar layer.

Beginning with slide P43, there is a distinctive bulge on the medial side of the body that may correspond to the mandibular symphysis. This region is composed of fibrolamellar bone with reticular channels.

In slides P49–P53 (slides P47, P48, P54 are damaged) near the procumbent root of the first incisor, the ventral and lateroventral edges of the mandibular body have fibrolamellar bone with radially oriented channels (Fig. 3B). Anterior to slide P55, the root of the first incisor becomes increasingly larger, and the ventral and lateral walls of the mandibular body become correspondingly narrower. However, the channels in the base of the mandible are still oriented radially (slides P55–P59).

Coronoid process

The coronoid process is predominantly composed of lamellar bone, with dorsoventrally oriented channels and distinctive growth marks (Fig. 3F–H). In some parts of the coronoid process, especially in areas where the coronoid process becomes wider, fibrolamellar bone occurs medial to the lamellar bone (Fig. 3F). The bone tends to be fairly compacted; however, the central part of the base of the coronoid process is secondarily resorbed and forms large cancellous spaces (Fig. 3H). In the more anterior sections, the cancellous spaces extend further dorsally into the process (Fig. 3G) and become continuous ventrally with the cancellous spaces in the mandibular body (slides P11–P16).

Coronoid boss

The coronoid boss, medial to the base of the coronoid process (Fig. 1B), is composed mainly of porous woven bone with channels that have a reticular arrangement, a dramatically different bone texture from that of the adjacent coronoid process (Fig. 3H).

Mental foramina region

Two mental foramina (anteroventral and posterodorsal) occur in SAM-PK-K407. The open foramina and their previous positions during earlier growth stages of the mandible are recorded in the histology and can be traced across several serial thin sections (slides P42–P52; Table 2, Fig. 4A–C). The open (current) foramina represent the passageway of the nerve at the time of death, whereas the previous positions of the foramina have been infilled (closed) with newer bone deposits that have different orientations of channels and osteocyte lacunae than the surrounding mandibular bone (Fig. 4A–C,E). The infilled foramina also extinguish differently under polarized light than the surrounding bone (Fig. 4A–C). A faint, scalloped reversal line separates the infilled mental foramen from the surrounding mandibular bone (Fig. 4E).

Fig. 4.

Relocation of the mental foramina of Tritylodon sp. (SAM-PK-K407). Series of slides [from anterior to posterior: (A) slide P50; (B) slide P47; (C) slide P45] under polarized light showing the posterior relocation of the anteroventral mental foramen and posterodorsal accessory mental foramen. Scale bars: 500 μm. In (A) the arrow indicates the new fibrolamellar bone that covers the lateral exit point of the posterodorsal infilled foramen. In (C) the arrowhead indicates the edge of the current (open) anteroventral mental foramen canal. (D) Schematic of the posterior relocation of the anteroventral mental foramen and posterodorsal accessory foramen during growth of the mandibular body, with the approximate positions of the histological sections (A–C) superimposed. (The outline of the mandible of SAM-PK-K1411 is used here, although specimen SAM-PK-K407 has two mental foramina.) The open (current) mental foramen is black, the posterodorsal infilled foramen is dark grey, and the anteroventral infilled foramen is light grey. The anterior-most positions of the posterodorsal infilled foramen are not shown because fibrolamellar bone covers its lateral exit point (see A). The posterior relocation of the mental foramina indicates that the mandible was growing in the posterior direction. (E) Magnified view of the infilled anteroventral mental foramen of slide P52 showing a clear reversal line (arrowheads) between the infilled foramen and the surrounding bone of the mandible. Scale bar: 100 μm. (F) Arrowhead indicates a single case of osteonal drift, lateral to the first incisor. The osteon is drifting laterally, which indicates deposition of bone on the lateral edge of the mandibular body (slide P51). Scale bar: 100 μm. A, alveolus for the first postcanine; CV, current anteroventral mental foramen; D, infilled posterodorsal accessory mental foramen; M, mandibular foramen; R, root of the first lower incisor; V, infilled anteroventral mental foramen.

The arrangement of mental foramina in SAM-PK-K407 (Fig. 4A–D) is similar to that observed in the larger specimen BP/1/5288 (Figs 1G and 2C) and thus it is likely that the posterodorsal foramen represents an accessory foramen of the anteroventrally positioned mental foramen.

The first appearance of both of the mental foramina occurs ventral to the diastema. The current anteroventral mental foramen is situated ventral to the first postcanine (slides P45–P49; Fig. 4B,C). The posterodorsal mental foramen is partially open in slide P42 but the complete opening of the canal on the lateral edge of the mandible was not captured by the thin-sections and may have occurred between the posteriorly situated slide P42 and the anteriorly situated slide P43.

Other histological indicators of localized growth are found near the region of the mental foramina. At the anterior region of the posterodorsal infilled foramen, the former exit point of that foramen on the lateral edge of the mandible is overlain with newer deposits of fibrolamellar bone (Fig. 4A). A single occurrence of osteonal drift (Fig. 4F), lateral to the first incisor, indicates bone drift towards the lateral edge of the mandible. Thin layers of lamellar bone are visible on the lateral edge of the larger mandibular canal (e.g. slides P47–P50).

Anchorage of teeth in jaws

Two types of tissues surround the teeth: cellular cementum occurs around the tooth roots (Fig. 5B) and bone of attachment (= alveolar bone; sensuCaldwell et al., 2003) occurs within the tooth socket (Fig. 6B). Acellular cementum was not observed in any of the thin-sections. The alveolar bone has a more porous texture, and the osteocyte lacunae appear to be more globular than those of the mandibular bone. In the old infilled tooth sockets, vascular cementum with cementeons (= osteocementum, Caldwell et al., 2003) also occurs.

Fig. 5.

Microstructure of the incisors and surrounding tissues of Tritylodon sp. (SAM-PK-K407). (A) Thin section through the first (larger, right) and second (smaller, left) lower incisors (slide P59). Dorsal is to the left of the photomicrograph. Scale bar: 500 μm. (B) Higher magnification of the framed area in (A) showing Sharpey’s fibres in the cellular cementum surrounding the root of the first incisor. In this region of the first incisor, dentinal fibres, but not the incremental lines of von Ebner, are clearly visible. Scale bar: 100 μm. CC, cellular cementum; FB, fibrolamellar bone; R, root of the first lower incisor; SF, Sharpey’s fibres.

Fig. 6.

Evidence of the relocation of a postcanine tooth socket of Tritylodon sp. (SAM-PK-K407). (A) Broken line indicates the approximate location of slide P16 (SAM-PK-K407), overlain on the outline of the mandible of SAM-PK-K1411. (B) Section through the posterior part of the mandible (slide P16; for approximate location see A), showing the current medial position of the penultimate postcanine tooth relative to the old lateral position of the tooth socket. The red arrowhead indicates the reversal line between the osteocementum and the alveolar bone/mandibular bone. Scale bar: 500 μm. (C) Higher magnification of the region below the framed region in (B), indicating the reversal line between the osteocementum and alveolar bone/mandibular bone with red arrowheads. Scale bar: 100 μm. (D) Higher magnification of the framed region of (B) showing three layers from lateral to medial: (1) mandibular fibrolamellar bone; (2) cellular cementum (bracket); (3) osteocementum. Scale bar: 100 μm.

Here for the first time in a nonmammaliaform cynodont, Sharpey’s fibres associated with a tooth are documented, indicating that the tooth was anchored with a periodontal ligament. The Sharpey’s fibres occurred within the cellular cementum near the long root of the first incisor (Fig. 5A,B). It is assumed that the Sharpey’s fibres within the dental cementum represent the Sharpey’s fibres of a periodontal ligament (see Raspanti et al., 2000). However, the location of these fibres, i.e. close to but not directly in contact with the root surface, suggests that they indicate a previous position of the tooth during an earlier stage of mandibular development (M. Raspanti, personal communication, 2011). Sharpey’s fibres were not associated with any other tooth; however, the tissues surrounding the teeth were not always well preserved, and a black opaque mineral sometimes covered the edges of the teeth.

The position of a tooth socket from an earlier stage of ontogeny (= old tooth socket) is evident in the histology of the mandible. It is situated lateral to the penultimate postcanine tooth and can be traced across several slides (slides P12–P17; Fig. 6). Another old tooth socket was located further anteriorly along the jaw in slide P22 (the slides bracketing P22 were not well preserved). In all cases, only the lateral part of the cellular cementum layer that originally surrounded the tooth remains (Fig. 6B–D). Medial to this layer is a thicker deposit of osteocementum, characterized by smaller channels (= cementeons) than the surrounding bone, with a ground substance similar to that of the cellular cementum. The osteocementum appears to have infilled the old tooth socket. A reversal line cuts into the osteocementum layer and separates it from the more recently deposited mandibular bone or alveolar bone that surrounds the current position of the postcanine tooth (Fig. 6B,C). The reversal line is nearly continuous around the current tooth socket position.

Tooth microstructure

Dentine structure

Dentinal tubules were visible in most of the thin-sections of the teeth (Figs 5 and 7B,D). There was no evidence of secondary dentine within the pulp cavity of the incisors or the postcanine teeth.

In some thin sections, incremental lines were evident in the dentine (= von Ebner incremental lines), although regions with discernible sequential lines were difficult to find. In the first incisor (slides P59 and P63), the average spacing between the incremental lines was 13.7 μm (calculated from a total of 28 measurements in three different regions of the tooth). In the second incisor (slides P56 and P59), the average spacing was 13.6 μm (calculated from a total of 48 measurements in four different regions of the tooth).

Enamel structure

In contrast to the generally well preserved dentine, the enamel on the postcanines for the most part was poorly preserved (often covered by a black opaque mineral layer) or absent. The enamel was best preserved in the penultimate and ultimate postcanines, and was either absent or very thin on the more anterior postcanines. This may suggest that the enamel on the more anterior (older) teeth has been worn down due to mastication, but additional sampling of better preserved specimens is required to confirm this supposition.

On the crown of the unerupted ultimate postcanine tooth (slide P11; Fig. 7A,B), a fairly thick, though uneven, layer of enamel was present. The enamel was thickest on the tip of the cusps of this unerupted tooth, with an average thickness of 190.3 μm. The enamel on the lateral cusp of the unerupted ultimate postcanine measured 224.2 μm, whereas it was 200.7 μm (slide P15), 91.8 μm (P16), and 127.6 μm (P17) on the lateral cusp of the penultimate erupted tooth. The enamel on the crown of some postcanines appeared to consist of an extremely thin layer (e.g. 38.7 μm on slide P24) but its identification is equivocal due to its lack of microstructural detail. In some instances where the enamel is well preserved, distinctive incremental lines (= striae of Retzius) are visible (Fig. 7B,D).

Fig. 7.

Microstructure of the postcanine teeth of Tritylodon sp. (SAM-PK-K407). (A) Unerupted ultimate postcanine tooth within a crypt (slide P11). Scale bar: 250 μm. (B) Higher magnification of the lateral cusp (A, arrow) showing the incremental lines within the thick enamel layer (slide P11). The dentinal tubules in the dentine and the cellular cementum surrounding the unerupted postcanine tooth are also visible. Scale bar: 50 μm. (C) Erupted penultimate postcanine tooth (slide P15). Scale bar: 500 μm. (D) Higher magnification of framed area in (C) showing the incremental lines in the enamel of the penultimate postcanine tooth (slide P15). The dentinal tubules and the faint incremental lines of von Ebner are also visible in the dentine. Scale bar: 50 μm.

A thin layer of enamel occurs along the lateral and medial edges of the first incisor (slides P64–P69), and in some areas incremental lines are visible. The enamel is thicker on the lateral edge than on the medial edge of the first incisor. It is thickest at the laterodorsal corner of the incisor, with a maximum average thickness of 63.1 μm (measured from slides P65–P69). The enamel also becomes thicker towards the anterior tip of the incisor: it is thinnest in slide P64 (10.9 μm) and thickest in slide P69 (69.0 μm). The dorsal surface of the first incisor is broken, thus distribution of the enamel on that surface could not be described.

Discussion

Gross anatomy of the mandible

Our finding of a double mental foramen in specimens BP/1/5288 and SAM-PK-K407 was unexpected because previously only a single foramen was documented for tritylodontids. However, an additional mental foramen has been described for some individuals of the basal mammaliaform Morganucodon (Kermack et al. 1973) as well as for Homo sapiens (Katakami et al., 2008; Ramadhan et al., 2010 and references therein), suggesting that the number of mental foramina can vary within taxa. Through the examination of limited cone-beam CT images of the mandible of Homo sapiens, Katakami et al. (2008) found that if an accessory foramen is present, it was usually smaller and typically located posterior to the main mental foramen. A similar arrangement occurs in two specimens of Tritylodon (SAM-PK-K407, BP/1/5288), where the smaller accessory foramen is located posterodorsal to the main mental foramen.

The double mental foramina in BP/1/5288 (Fig. 2C) are situated further posteriorly along the jaw than the mental foramen of the smaller specimen SAM-PK-K1411 (Fig. 2A,D) and other tritylodontid specimens described in the literature (e.g. Sues, 1986; Luo & Sun, 1993; Watabe et al., 2007). The posterior shifting of the foramen/foramina in larger individuals indicates relocation of these structures during growth, a process that is described in detail in the following section.

General histology of the mandible (SAM-PK-K407)

The Tritylodon mandible was mainly composed of rapidly deposited fibrolamellar bone and more slowly deposited lamellar bone. Both tissue types are known to occur among basal synapsids (e.g. Huttenlocker & Rega, 2012). The distribution of fibrolamellar and lamellar bone in the mandible is discussed in the following section on growth dynamics.

The bone microstructure of SAM-PK-K407, with its prevalence of primary tissue, suggests that it was an immature individual. This interpretation is supported by gross anatomical features, such as no loss of the anterior postcanines and the relatively small size of the dentary (Table 1).

The size of the osteocyte lacunae appear to be more globular in the alveolar bone than the mandibular bone [in mosasaurs, the opposite pattern occurs (Luan et al., 2009)], although we are cautious in making this statement as there are many factors such as orientation of the lacunae and diagenesis that can affect their appearance and size.

Growth of the mandible

The processes of modelling and remodelling that occur in the skeleton are often conflated simply as ‘remodelling’ (e.g. Enlow’s studies). However, in studies of growth, it is pertinent to differentiate between these processes that involve both osteoblastic (depositional) and osteoclastic (removal) activity. Here we use Martin et al.’s (1998) definitions which characterize modelling as the processes that result in the change in shape and/or size of the skeletal element during developmental ontogeny, whereas remodelling is episodic, occurs throughout an individual’s lifetime, and is responsible for repairing microscopic damage, but does not necessarily involve changes in shape and/or size of the bone (Martin et al., 1998).

As a bone grows and increases in size, its shape and form are constantly maintained through structural (re)modelling (Enlow, 1963). This includes localized resorption or deposition on bone surfaces, which may change at different times during growth of the individual and area relocation (Enlow & Harris, 1964). An example of area relocation is when a region of bone that was originally located posteriorly in the jaw (e.g. part of the condyle) becomes incorporated into a more anterior part of the mandible as the jaw grows and lengthens posteriorly (Enlow & Harris, 1964). Thus the tissues that originally formed the condyle need to adjust to the new cross-sectional shape by adding or removing bone on its surfaces (Enlow & Harris, 1964). Evidence of both of these processes was observed in the serial sections of the Tritylodon mandible.

It is important to note that the histological sections only provide a ‘snapshot’ of the mandible at one particular time during the growth process. Enlow & Harris (1964) mainly focused on immature cranial specimens because they found that remodelling of bone in the adult specimens obscured earlier patterns of growth; our investigation of the immature specimen SAM-PK-K407 is therefore ideal to document the growth patterns in the Tritylodon mandible.

Although we did not observe mandibles smaller than specimens SAM-PK-K1411, we can infer that the growth trajectory and processes were similar to those that occurred between specimens SAM-PK-K1411 and BP/1/5288. From the sequence of gross anatomical specimens (Figs 1 and 2) it appears that during ontogeny, the mandibular body lengthened and increased in depth. However, only histological analysis can provide a direct assessment of the processes of how the mandible grew in size.

In the Tritylodon jaw, the infilled mental foramina are located anterior to the current (open) foramina (Fig. 4), which indicates that the foramina were shifted posteriorly as the mandible grew in length. Thus, growth of the mandibular body occurred mainly in a posterior direction (Fig. 8A), similar to that documented by Enlow (1963) for the human mandible. The posterior repositioning of the mental foramina represents a compensatory movement relative to the posterior direction of mandibular growth, which was a necessary process because the nerves these foramina house must continually innervate specific tissues and they cannot be passively carried along with the growth of the mandible. A similar pattern of posterior relocation was described (although not illustrated) for the large mandibular foramen in the human mandible, which maintained a constant position during the posterior and dorsal growth of the ascending ramus (Enlow, 1963; Enlow & Hans, 1996).

Fig. 8.

Schematic showing the growth dynamics of the Tritylodon mandible ascertained from the interpretation of bone histology of SAM-PK-K407. The jaw outline (A, lateral view) and reconstruction (B, dorsal view) are from a similar-sized specimen of Tritylodon (SAM-PK-K1411). The black arrows indicate areas of bone growth (see also Fig. 3), whereas the grey arrow indicates the medial movement of the ultimate postcanine tooth (see also Fig. 6). The large white arrow indicates that the overall direction of growth of the Tritylodon mandible is in the posterior direction. The positions of the arrows are approximate, as the exact location of sectioning is not known. Note that the localized growth of the mandible indicated here was not necessarily concurrent and/or sustained growth, but likely episodic.

The changes in bone texture in the mandibular body reflect a difference in the rate of bone formation and direction of growth. There were multiple episodes of bone deposition on the lateral edge of the mandibular body, posterior to the open anteroventral mental foramen (Fig. 8A). For instance, along the laterodorsal edge, the deposition of slowly forming lamellar bone was followed by more rapidly deposited fibrolamellar bone (e.g. Fig. 3D). In many cases, a rest line indicating a pause in growth separated the two layers, and the orientation of the layers of lamellar bone differed (Fig. 3E). In the mandibular body near the mental foramina, lamellar bone was absent from the lateral edge, but new fibrolamellar bone was deposited on the lateral edge of the mandibular body after the posterodorsal accessory mental foramen was relocated posteriorly (Fig. 4A). Also a single occurrence of osteonal drift (Fig. 4F) directly shows drift towards the lateral side of the mandible.

The rest lines evident in the Tritylodon mandible provide indications of localized changes in the rate of bone deposition as a consequence of local morphological changes in mandibular shape and size.

The different bone textures observed in the coronoid process reflect differences in the rate of growth and bone deposition. The lateral edge of the coronoid process, mainly composed of lamellar bone, indicates its relatively slow growth, whereas the fibrolamellar texture that occurs on the medial side of some sections suggests a relatively more rapid rate of bone deposition.

The deposition of newer lamellar and fibrolamellar bone on the lateral edge of the mandibular body, and the irregular, growing surface of the bone, indicates the widening of the mandibular body. This might correspond to the low boss that is positioned midway along the mandibular body (Fig. 1A). Also, the difference in orientation of the lamellar layers (Fig. 3E) may correspond to the curved contours of the mandibular body during an earlier stage of growth. The uneven thickness of the newer deposits of fibrolamellar bone lateral to the lamellar bone (Fig. 3D) increased the width of the mandibular body along its laterodorsal edge.

The orientation of the channels in different regions also provides pertinent information regarding mandibular growth. In two regions of the ventral base of the mandibular body [posterior (slides P1–P4) and anterior (slides P49–P59) parts; Fig. 8A], the channels tend to be radial along the medial and ventral edges of the mandibular body, which suggests faster bone deposition in these directions (see de Margerie et al., 2004). The deepening of the ventral base in the anterior part may correspond to the continued development of the mandibular symphysis during growth.

The channels in the narrow coronoid process are mainly oriented in a dorsoventral direction. The thin cross-sectional area of the bone may have restricted the orientation of the channels, a correlation that has been noted for some dicynodont cranial bones with narrow dimensions (Jasinoski & Chinsamy-Turan, 2012), and may have served as a strengthening feature. The orientation of the channels may also reflect the predominant direction of growth of the bony element (e.g. Lee, 2004).

Tooth attachment

This is the first time that definitive evidence for a periodontal ligament associated with the first incisor is documented in a derived cynodont. The attachment of Tritylodon teeth via a periodontal ligament, as indicated by Sharpey’s fibres, is unlike the condition in more basal cynodonts, where only the bone of attachment (cementum) anchored the teeth (see Fig. 4.7 in Crompton, 1995).

Crompton (1989; personal communication, 2010) proposed that some derived herbivorous cynodonts, such as the tritylodontids, had periodontal ligaments. His supposition was based on the structure of the tooth socket (the teeth were not ankylosed to the jaw bone) and the precise occlusion of the postcanines. Dental occlusion and a complex chewing mechanism require periodontal ligaments and their associated sensory nerves to provide mechanoreceptor (pressure receptor) feedback during occlusion (Crompton, personal communication, 2010). The discovery of Sharpey’s fibres near the tooth root of Tritylodon SAM-PK-K407 (Fig. 5B) supports his original hypothesis.

Tooth anatomy

In the histological sections, the enamel in the erupted postcanine teeth is thinner than that in the unerupted postcanine tooth, suggesting that the enamel wore down during mastication. The shaping of the crown through occlusion may have been important for the proper functioning of the tritylodontid postcanine tooth (Kühne, 1956). In the erupted lower postcanines of Kayentatherium, wear of the enamel along the edges of the cusps was noted, and dentine was exposed on at least one of the cusps surrounding the central longitudinal groove (Fig. 19B in Sues, 1986). Unfortunately, because of the poor surface preservation of many of the teeth, we were unable to verify this in our histological specimen.

It was reported in Kayentatherium that thin enamel covered only the anterolateral tip of the first incisor (Fig. 12 in Sues, 1986). In the histological sections of Tritylodon, the enamel on the lateral edge of the incisor progressively thickened from slides P64 to P69 (in the anterior direction). This observation, along with the lack of enamel in slides posterior to P64, may indicate that the enamel in Tritylodon was also confined to the anterior tip of the incisor. However, unlike in Kayentatherium, the enamel in Tritylodon was present on both the lateral and medial edges of the first incisor, although the enamel was much thinner along the medial edge than on the lateral edge. It was thickest on the laterodorsal corner of the first incisor.

The first incisor of Tritylodon has an open root, in contrast to the second, smaller incisor with a root that is constricted distally. This difference suggests that the first incisor was capable of continuous growth.

The width of the incremental lines of von Ebner in the dentine of the incisors of Tritylodon falls within the daily range reported for amniotes (1–30 μm per day; Erickson, 1996 and references therein). It is less than what has been reported for dicynodonts (average range of 17.4–22 μm; Thackeray, 1991; Green, 2012; Jasinoski & Chinsamy-Turan, 2012) but higher than reported in a mosasaur (average of 10.9 μm; Chinsamy et al., in press). It should be noted that the width of the von Ebner lines appears to depend on the location within the tooth: we found that close to the tip of the root of the second incisor (slide P53), they are less 6 μm wide. The incremental lines may also appear wavy in some regions of the tooth.

Tooth replacement and movement

The micro-CT scans of the largest individual (BP/1/5288) revealed that ventral to the diastema there are remnants of the roots of two shed anterior postcanine teeth. This is similar to that reported for a large specimen of Oligokyphus (R.7133, Kühne, 1956). These observations indicate that after the crowns were shed, the roots remained and were partially resorbed. In contrast, the CT-scans of the smallest individual (SAM-PK-K1411), which has a shorter diastema and a greater number of postcanines (Table 1), showed no vacant alveoli or root remnants of anterior postcanines. Remnants of the roots of the anterior postcanines were also absent from specimen SAM-PK-K407.

In SAM-PK-K407, traces of an old infilled tooth socket indicate the previous position of a postcanine tooth near the posterior part of the tooth row. The old tooth socket is not evidence of an older replaced postcanine tooth because tritylodontids do not directly replace pre-existing postcanines (e.g. Kühne, 1956). It is also unlikely that it is the socket of a shed anterior postcanine because the position of the old tooth socket is far posterior to the anterior end of the tooth row. A more likely explanation is that the old tooth socket represents the movement (includes drift, displacement and eruption; sensuEnlow & McNamara, 1973) of one of the posterior postcanines during growth of the mandible. As observed in gross osteological specimens of Tritylodon (personal observation) and other tritylodontids (Sun, 1984), the position of the posterior unerupted/partially erupted lower postcanine(s) is lateral relative to the tooth row (Figs 1E and 8B). Sues (1986) also noted that the crowns of newer (posterior) postcanines face medially compared with the older postcanines. During ontogeny, the lower postcanine moved medially to fall in line with the tooth row (Sun, 1984). Thus, the old tooth socket most likely indicates the original lateral-most position of a postcanine tooth. The lateral edge of the old tooth socket is demarcated by cellular cementum, and the reversal line and secondarily deposited bone define the lateral edge of the present tooth position after its migration medially. During growth of the mandible, the old tooth socket was also relocated posteriorly, thus the identification of exactly which postcanine tooth this old socket represents is not determinable.

Conclusions

Observations of the gross osteology and internal anatomy of Tritylodon, using micro-CT scans and bone histology, have enabled an assessment of the growth processes during ontogeny (Fig. 8). Although our research was conducted on a nonmammaliaform cynodont, our observations are similar to the findings made by Enlow on human mandibular growth and development.

The histological analysis of the bone and dental microstructure revealed the following for Tritylodon:

• The overall direction of growth of the mandibular body was posterior, based upon the relocation of the mental foramina in the posterior direction.

• During growth, there was also lateral widening of the mandible, as indicated by the deposition of newer bone along the lateral edges of the mandibular body. There were multiple episodes of bone deposition that occurred at different rates: the presence of lamellar bone indicates a slower rate of deposition, whereas the fibrolamellar bone was deposited more rapidly. Deepening of the ventral base, as indicated by radial channels, also occurred in the anterior and posterior ends of the mandible.

• During growth and the process of eruption, the posterior postcanines migrated medially. Histological traces consisting of at least one old tooth socket confirms the observations of gross osteology.

• The evidence of Sharpey’s fibres associated with the first incisor suggests that tritylodontid teeth were anchored via a periodontal ligament. It is likely that this type of tooth attachment also occurred in other derived cynodonts that had precise dental occlusion.

• The enamel is thinner on the erupted postcanines than on the unerupted ultimate postcanine, suggesting that enamel wear occurred during mastication. The enamel on the first incisor progressively thickens towards the tip of the tooth.

Although CT technology has advanced rapidly in the past few years, this study has shown that it is still necessary to thin-section specimens to determine their histology and their growth dynamics. Fine microstructural detail, such as subtle changes in bone texture, is not visible in the micro-CT slices. A synchrotron micro-CT scan may give this sort of fine detail but it is currently limited to small specimens such as a single tooth (e.g. Tafforeau & Smith, 2008). Without histological analysis, evidence of growth preserved in the microstructure could not have been detected in the mandible of Tritylodon.