Bone histology of dinocephalians (Therapsida, Dinocephalia): palaeobiological and palaeoecological inferences

Histological description of Anteosaurus

The femur (SAM-PK-K291) belongs to A. magnificus (Table 1; reference therein). It has a distinct, thick outer cortex and medullary region partially infilled by thick trabeculae (Fig. 1A). The cancellous bone is unevenly distributed inside the medulla and it progressively grades into cortical bone. The predominant bone tissue of the outer cortex is highly vascularized, uninterrupted fibrolamellar bone tissue (Fig. 1B, C) with a woven matrix as defined by fibre orientation, and globular and randomly distributed osteocyte lacunae (Fig. 1D). In the cross-section, fibrolamellar bone deposition continues up to the periphery, however, there is a thin layer of poorly vascularized lamellar bone tissue with multiple closely spaced rest lines (Fig. 1F), with isolated simple vascular channels. The perimedullary region shows extensive secondary reconstruction that leads to a number of large erosion cavities that give it a cancellous texture (Fig. 1A). It should be noted that the amount of cancellous bone varies in the cross-section. Despite extensive secondary reconstruction in the inner cortex, islands of primary bone tissue are present within the core of trabeculae. Secondary osteons are absent in the compacta. Circumferentially oriented longitudinal canals are predominant, but a large number of radial and reticular canals are present in the compacta (Fig. 1B–D).

Another femur (BP/1/5591) belonging to the genus Anteosaurus displays an open medullary cavity with thick struts of trabeculae extending into the medullary cavity (Fig. 2A). The medullary cavity is irregular and encircled by a thin layer of endosteal lamellar bone tissue (Fig. 2B). Similar to SAM-PK-K291, the outer cortex consists of highly vascularized fibrolamellar bone tissue with numerous primary osteons (Fig. 2C, D). The vascular channels are mostly consistent in size and density throughout the compacta. Vascular canals are mostly circumferential in orientation with a laminar (Fig. 2E) and reticular arrangement (Fig. 2C, D), with the former restricted to the inner cortex whereas the latter dominates the middle cortex and peripheral regions (Fig. 2A, E). However, radially oriented canals are present mostly in areas of muscle attachment such as the trochanter (Fig. 2F) and penetrate deep into the cortex, reaching the medullary cavity (Fig. 2A). Some radial canals occur locally in the compacta. Four widely spaced growth marks are present in the compacta (Fig. 2C, D), and it should be noted that there is no associated change in degree or pattern of vascularization prior to or after the growth marks (Fig. 2C, D).

Histological description of Keratocephalus moloch

The core sample of the ulna of K. moloch shows a laminar fibrolamellar bone tissue surrounding a medullary region that is infilled by bony trabeculae (Fig. 3A–C). The vascular canals near the medullary cavity are highly remodelled (Fig. 3A). The resorption between these canals is extensive and results in the large erosional spaces (Fig. 3D, E). On the ventral side, the cross-section is crushed (Fig. 3A), showing a mixture of circumferential and longitudinal vascular canals towards the inner region (Fig. 3G). The inner cortex has large erosional spaces (Fig. 3A) with islands of broken trabeculae with large erosion spaces (Fig. 3F), and some of them reach cancellous proportions towards the inner cortex (Fig. 3A). Lamellar bone tissue encircles these erosional spaces (Fig. 3D, E). Growth marks were not visible in the compacta. The periosteal surface is broken on the ventral side where the skeletal element was damaged, whereas it is smooth dorsally.

Histological description of Moschops

The core (Fig. 4A) of the humerus (BP/1/6987) shows the inner medullary cavity infilled with bony trabeculae and an outer layer of highly vascularized fibrolamellar bone that extends to the periosteal surface (Fig. 4B–E). Due to intense remodelling in the inner cortex, the transition between the medullary region and cortical bone is progressive. The inner cortex, formed during early ontogeny, consists of reticular fibrolamellar bone tissue (Fig. 4F, G). The subperiosteal and peripheral cortex consists of laminar fibrolamellar bone (Fig. 4A–E). The outer vascular canals are distinctly narrower than the inner vascular canals as the spacing between consecutive vascular canals decreases towards the periphery (Fig. 4B–E). However, the overall orientation of the vascular canals is circumferential, with the major axis parallel to the outer cortical margin (Fig. 4B–D). The inner cortex has large secondarily formed erosional spaces, and few primary osteons are preserved in the bony trabeculae (Fig. 4F, G). The open spaces in the bone tissue of the medullary region are mostly infilled with mineral sediment and most reach cancellous dimensions (Fig. 4A, F, G). These resorption cavities show varying degrees of centripetal deposition of lamellated bone (Fig. 4E–G), and some islands of primary bone consist of woven matrix towards the inner region (Fig. 4G).

The two ribs fused together are composed entirely of well-vascularized fibrolamellar bone (Fig. 5). Osteocyte lacunae are abundant and are arranged in a disorderly manner typical of fibrolamellar tissue (Fig. 5B). Circumferential vascular canals are predominant in a laminar arrangement (Fig. 5). The open spaces in the bone tissue are frequently infilled with mineral sediment (Fig. 5B, D). Growth marks are absent.

Histological description of Struthiocephalus whaitsi

The core of the femur (SAM-PK-11939a) has a distinct, thick cortex and a medullary region occupied by cancellous bone tissue (Fig. 6A). The outer cortex is highly vascularized and consists of fibrolamellar bone tissue (Fig. 6B) and numerous primary osteons within the woven bone matrix (Fig. 6C, D). Globular osteocyte lacunae are densely distributed in the matrix throughout the cortex. The inner cortex contains circumferential fibrolamellar bone (Fig. 6A) whereas the outer peripheral cortex has laminar bone tissue towards the dorsal and ventral margin of the skeletal element (Fig. 6B). The vascularization pattern is quite variable in the section given that the core was obtained near the femoral head. A laminar vascular pattern is seen in the middle of the section (Fig. 6A, E), which continues towards the ventral margin (Fig. 6A, H). In addition, in localized areas in the middle cortex few localized radial and longitudinal canals occur (Fig. 6C, D). The medullary region is highly remodelled (Fig. 6A) and many erosional cavities are visible (Fig. 6F, G). Some islands of primary bone tissue are preserved in the middle and inner cortex between resorption cavities and consist of a woven bone matrix (Fig. 6F).

The core sample of the humerus (SAM-PK-11939b; Fig. 7) shows an extremely thin cortex with an extensive amount of cancellous bone that extends from the middle cortex and completely infills the medullary cavity. Due to intense remodelling and endosteal deposition in the inner cortex (Fig. 7A–D), the transition between the medullary region and cortical bone is progressive (Fig. 7B). The dorsal and ventral periosteal bone margins are irregular and internal cracks developed during thin section preparation (Fig. 7A). The outer rim of cortical bone preserved is entirely composed of fibrolamellar bone tissue with longitudinal primary osteons (Fig. 7B). The medullary region is highly cancellous in nature with large erosional cavities (Fig. 7A), with many of them showing lamellar deposits along their endosteal surfaces (Fig. 7E). Three growth marks are present (Fig. 7C), two of which are closely spaced in the inner cortex, representing a double line of arrested growth (LAG) and the third is present toward the peripheral cortex.

Inter-elemental and intraspecies-level histological variation

Bone histology of the multiple skeletal elements of dinocephalians varied substantially in terms of degree of vascularization, tissue organization and amount of cancellous bone and appeared to be related to growth dynamics, palaeoecology and taxonomic differences. In a study on humeral microanatomy of terrestrial, semi-aquatic, and aquatic carnivorans, Nakajima & Endo (2013) reported that the degree of cancellousness around the ossification centres of the bone (i.e. the amount of resorption in periosteal compact bone) was the lowest in the semi-aquatic and the highest in terrestrial animals. Given that the degree of cancellous bone varies around ossification centres in animals of different modes of life (Nakajima & Endo 2013), it was also noted that the tissue type and vascularization pattern vary between species of different ecological and dietary niches (Nakajima et al. 2014). In the current study, histological variations in terms of degree of vascularization, tissue organization and amount of cancellous bone are visible in the bones studied. Furthermore, these variations were noted in the same diaphyseal section and between skeletal elements of the same or different taxa. In addition, there are distinctive differences in the vascular orientation of the tissue between an earlier formed bone and a later formed bone.

It appears that bone of carnivorous Anteosaurus had the most histological variations in terms of change in tissue type and growth strategy through ontogeny. In addition to differences in the amount of trabecular bone in the medullary region, the femur BP/1/5591 (Fig. 2A) has highly vascularized fibrolamellar bone in the peripheral region (Fig. 2C, D), suggesting that osteogenesis had not yet slowed down in this individual, whereas the femur SAM-PK-K291 (Fig. 1A) belonging to Anteosaurus magnificus has numerous peripheral rest lines (Enlow 1962; Fig. 1F). Although the early growth in BP/1/5591 (Fig. 2C, D) was periodic, as suggested by the presence of multiple growth marks in the fibrolamellar bone tissue, in SAM-PK-K291 (Fig. 1A) it is uninterrupted and continuous. The femur BP/1/5591 (Fig. 2A) belongs to a separate Anteosaurus species (Table 1). It is interesting to note that the diameter of the femur BP/1/5591 of Anteosaurus sp. is higher than the femur (SAM-PK-K291) of Anteosaurus magnificus, which had already attained skeletal maturity, as shown by the presence of peripheral rest lines in the outer cortex of the femur (Chinsamy-Turan 2005; Nacarino-Meneses et al. 2016; Calderón et al. 2019).

All the bones of the herbivorous dinocephalians (Keratocephalus, Moschops, Struthiocephalus) lack growth marks in the compacta, except for the humerus of Struthiocephalus, suggesting uninterrupted, rapid growth for the skeletal elements until this stage of their lives. The absence of LAGs and peripheral rest lines suggests that these elements belong to sub-adult individuals (e.g. Ray et al. 2005). In contrast, the femur of Struthiocephalus whaitsi shows changes in the organization of the vascular canals in different parts of the compacta, suggesting different rates of bone deposition in the fibrolamellar bone complex (de Margerie et al. 2004). Extensive remodelling of the original primary bone in the humerus of S. whaitsi leads to a completely remodelled inner and subperiosteal cortex. The LAGs in the peripheral margin (Fig. 6) and the decrease in vascularization indicate that the rate of bone deposition had slowed down substantially (Chinsamy-Turan 2005). It may be noted that the core of the femur (SAM-PK-11939a) was obtained from the proximal end of the bone and it is likely that the histological features of the humerus and femur may vary.

Histological variations and palaeobiology

In the current study, carnivorous (e.g. Anteosaurus) and herbivorous (e.g. Keratocephalus, Moschops, Struthiocephalus) dinocephalians were studied to assess for histological differences. To broaden the comparison, the histological information of the omnivorous Jonkeria (Bhat et al. in press) was included for comparative purposes. Overall, differences were seen in the multi-element comparison of dinocephalians with the skeletal elements of Anteosaurus, which showed changes in its growth strategy through ontogeny. Although the overall growth was fast, multiple growth marks were present in the femur (BP/1/5591) of Anteosaurus. The presence of multiple growth marks indicates a periodic or flexible growth strategy and directly suggests that this individual grew in an interrupted manner (Starck & Chinsamy 2002; Chinsamy-Turan 2005). Although the earlier growth marks may have been erased as a result of secondary reconstruction (Fig. 2A), we assume that it is unlikely that this individual was older than 4 years at the time of death, given that traces of growth marks are visible in the perimedullary region. Another femur (SAM-PK-K291) showed uninterrupted fibrolamellar bone. However, a few closely spaced peripheral rest lines are present, indicating that the growth has significantly slowed down prior to death (Francillon-Vieillot et al. 1990; Castanet et al. 1993; Chinsamy et al. 1995; Starck & Chinsamy 2002; Chinsamy-Turan 2005; Erickson 2005). In addition, the medullary regions of all of the bones except the femur (BP/1/5591) are secondarily infilled by bony trabeculae.

In contrast, the herbivorous dinocephalians (Keratocephalus and Moschops) had highly vascularized cortices without any interruptions in growth. However, the humerus of Struthiocephalus showed several LAGs in the outermost cortex, indicating a periodic ceasing of bone deposition. The femur of Moschops and the ulna of Keratocephalus have laminar fibrolamellar bone in the outer cortex surrounding a medullary cavity infilled with spongy bone. Both these animals are thought to be adapted to a semi-aquatic (hippo-like) lifestyle (von Huene 1931) and to have fed on fairly resistant food such as rhizomes or stems (Gregory 1926; King 1988). Similar histological features were noted in the femur of the herbivorous Struthiocephalus that is thought to have lived in marshy/littoral areas and fed on soft vegetation, possibly under water (Boonstra 1965; King 1988). The absence of LAGs in the femur of Struthiocephalus is similar to that in the femur of Moschops and the ulna of Keratocephalus, suggesting continuous and rapid deposition of bone tissue. The fused ribs of Moschops lack LAGs, which may indicate that they are from a young individual, or that, contrary to the observations made in sauropod dinosaurs (Stein & Sander 2009; Waskow & Sander 2014), ribs do not preserve a good track record of growth in this taxon. Further research on more ribs of different taxa through ontogeny is warranted to verify these deductions. The skeletal elements of the omnivorous Jonkeria had mixed histological features (Shelton et al. 2019; Bhat et al. in press), similar to both the herbivorous and carnivorous dinocephalians in the current study. Shelton (2015) reported different bone histology in herbivorous and carnivorous late Carboniferous and early Permian pelycosaur-grade synapsids. He showed that the carnivores grew much faster than the herbivores, which appeared to have secondarily reduced their growth rates despite having evolved to large sizes. In contrast, contemporaneous graviportal terrestrial pareiasaurs (Canoville et al. 2014) had a relatively rapid growth early in ontogeny but later slowed down (Canoville & Chinsamy 2017), with the primary bone matrix reported as parallel-fibred to lamellar. We note a predominance of fibrolamellar bone as the primary bone tissue in the compacta, suggesting rapid osteogenesis (Amprino 1947; Francillon-Vieillot et al. 1990; de Margerie et al. 2002, 2004; Starck & Chinsamy 2002; Chinsamy-Turan 2005) and an overall fast rate of bone deposition and growth for dinocephalians. Moreover, Haversian substitution with several generations of secondary osteons was prevalent in pareiasaurs (Canoville & Chinsamy 2017) but it was not present in our sample. However, histological variations were observed in the skeletal elements of the studied taxa, suggesting slight differences in growth trajectories.

Lifestyle adaptations

Cross-sectional geometry (cortical thickness or relative bone wall thickness) and bone microanatomy, as well as osteohistological characteristics have been used to deduce the lifestyle adaptations of a variety of tetrapods (Wall 1983; Chinsamy 1997; de Ricqlès & de Buffrénil 2001; Germain & Laurin 2005; Gray et al. 2007; Canoville & Laurin 2009, 2010; Houssaye 2009; Laurin et al. 2011; Hayashi et al. 2013; Nakajima et al. 2014; Houssaye et al. 2016; Canoville & Chinsamy 2017; Montoya-Sanhueza & Chinsamy 2018; Botha 2020). The bone of semi-aquatic and aquatic animals often shows an increase in bone compactness to reduce buoyancy in water (Chinsamy 1991; Gray et al. 2007; Kriloff et al. 2008; Hayashi et al. 2013), which often results in the loss of the medullary cavity and therefore increased bone wall thickness (Wall 1983). However, it should be noted that burrowing animals generally have thick bone walls with medullary spongiosa (Bramble 1982; Ultsch & Anderson 1988; Lips 1991; Magwene 1993; Botha & Chinsamy 2004, 2005; Ray & Chinsamy 2004; Chinsamy & Abdala 2008; Meier et al. 2013; Lyson et al. 2016; Montoya-Sanhueza & Chinsamy 2017; Legendre & Botha-Brink 2018; Bhat et al. 2019) as compared with terrestrial animals, which have lower bone wall thickness (relative bone wall thickness <30%; sensu Wall 1983) and open medullary regions. In such circumstances, it is essential to consider additional morphological adaptations for burrowing as well as other anatomical features that might aid in deducing their lifestyle habits (Nasterlack et al. 2012; Houssaye et al. 2016; Montoya-Sanhueza & Chinsamy 2017).

Tetrapods showing semi-aquatic adaptation achieve high bone density and compactness by losing medullary cavity through the extensive infilling of inner cavities by bony trabeculae (de Ricqlès & de Buffrénil 2001; Ray et al. 2005). The development of internal medullary spongiosa in the bones augments bone density to maximize the efficacy of buoyancy in water (Wall 1983; Taylor 1994; Gray et al. 2007), and flexural rigidity to increase the strength and toughness of the bone (Carter & Spengler 1978; Rogers & LaBarbera 1993). In the current study, all the long bones of herbivorous dinocephalians show extensive development of medullary spongiosa as compared with the carnivorous Anteosaurus. It was also noted that the resorption cavities in the long bones of Anteosaurus are extensively developed with wide open spaces between trabeculae. Similar features were noted in the skeletal elements of Jonkeria, which is an omnivore (Shelton et al. 2019; Bhat et al. in press). The medullary regions of the long bones of Jonkeria are completely infilled by spongiosa as compared with Anteosaurus. In contrast, the herbivorous animals in our sample (Keratocephalus, Moschops and Struthiocephalus) have medullary regions virtually filled with cancellous bone tissue.

Interestingly, the dicynodont Lystrosaurus murrayi has compact bone with spongy bone tissues and, based on this type of bone histology, Ray et al. (2005) proposed a semi-aquatic/aquatic lifestyle for Lystrosaurus, contradicting earlier work supporting a fully terrestrial lifestyle (King 1991; King & Cluver 1991). However, anatomical features such as a widened knee and flared scapula blade present in Lystrosaurus point towards a digging function. Botha-Brink & Angielczyk (2010), and later Botha-Brink et al. (2016), hypothesized that the rapid growth observed in Triassic Lystrosaurus is the result of a high degree of vascularization. They further suggested that the success of this taxon may also be attributed to similar life history adaptations including digging behaviour. However, more recently, Botha (2020) proposed a fully terrestrial mode of life for Lystrosaurus, using the same histological characteristic features noted by Ray et al. (2005). Botha (2020) further suggested biomechanical constraints as the cause of the extensive development of medullary spongiosa in Lystrosaurus to support the large body weight. Aquatic tetrapods have been shown to have compact bone walls with the medullary regions infilled with spongy bone tissue (Houssaye et al. 2016), which is considered to counterbalance buoyancy in water (Chinsamy 1991; Gray et al. 2007; Hayashi et al. 2013; Houssaye et al. 2016). In contrast, terrestrial graviportal animals such as elephants (Houssaye et al. 2016), giraffes (Smith 2020) and bison (Sander & Andrassy 2006; Canoville et al. 2016; Houssaye et al. 2016) do not develop medullary cavities with cancellous bone (Ray et al. 2010; Houssaye et al. 2016) as a biomechanical adaption for their bulk, contra Botha (2020). However, as in the current sample, the ribs of the known aquatic reptile, Claudiosaurus (de Buffrénil & Mazin 1989) show a complete infilling of the medullary region by cancellous bone and are considered pachyostotic as an adaptation for a semi-aquatic lifestyle (de Ricqlès & de Buffrénil 2001). The contemporaneous middle Permian pareiasaurs, which were similarly large bodied tetrapods, had similar histological characteristics to dinocephalians (i.e. spongy inner medullary regions, although not as extensive as those seen in dinocephalians; Canoville & Chinsamy 2017). Canoville et al. (2014) performed isotopic analysis of the teeth of dinocephalians, pareiasaurs and therocephalians and suggested that these large graviportal animals inhabited different ecological niches during the middle and late Permian times, with pareiasaurs and therocephalians sharing a terrestrial habitat. In addition, their data also suggest that both dinocephalians and pareiasaurs were terrestrial and that oxygen isotope differences reflect physiology and higher water turnover rates for dinocephalians.

On the basis of our histological findings, we cautiously propose that the histology of the herbivorous dinocephalians (e.g. Keratocephalus, Moschops and Struthiocephalus) and the previously described omnivorous Jonkeria (Bhat et al. in press) are indicative of a semi-aquatic habitat, while the apex predator Anteosaurus has histological features indicative of a more terrestrial lifestyle. It is, however, probable that Anteosaurus may have occasionally inhabited ephemeral water ponds, like the stereospondyl Lydekkerina (Canoville & Chinsamy 2015). Given our small sample size, we suggest that future more comprehensive sampling should be undertaken to verify our conclusions.