Bone microstructure and the evolution of growth patterns in Permo-Triassic therocephalians (Amniota, Therapsida) of South Africa

Bone microstructure in therapsids

Bone histology has offered insights into the lifestyles and growth patterns of many of the major subclades of nonmammalian therapsids. Recent examples include investigations of feeding and locomotion, habitat use, and especially growth dynamics (e.g., Ray, Chinsamy & Bandyopadhyay, 2005; Jasinoski, Rayfield & Chinsamy, 2010; Chinsamy-Turan, 2012). Earlier surveys of bone histology emphasized differences between basal (pelycosaurian-grade) synapsid and therapsid tissue composition, matrix organization, and degree of vasculature of the limb bones (Enlow & Brown, 1957; Enlow, 1969; Ricqlès, 1969; Ricqlès, 1974a; Ricqlès, 1974b; Ricqlès, 1976). Particularly, fibrolamellar tissue complexes (vascularized bone tissues that incorporate primary osteons within a woven-fibered matrix) were found to be near ubiquitous among limb elements of sampled therapsids, suggesting that this tissue complex appeared early during therapsid evolution prior to the origination of mammals bearing this tissue-type (Ricqlès, 1969; Ricqlès, 1974a; Ricqlès, 1974b; Ray, Botha & Chinsamy, 2004; Chinsamy & Hurum, 2006; Ray, Bandyopadhyay & Bhawal, 2009). Indeed, fibrolamellar bone has been reported in basal therapsids (e.g., Biarmosuchus: Ricqlès, 1974b), as well as in some immature pelycosaurian-grade synapsids (e.g., Sphenacodon and Dimetrodon juveniles) and in fast-growing portions of the skeleton (e.g., the elongated neural spines of Dimetrodon and crossbars on the neural spines of Edaphosaurus) (Huttenlocker, Rega & Sumida, 2010; Huttenlocker, Mazierski & Reisz, 2011; Huttenlocker & Rega, 2012; Shelton et al., 2013). However, there is great histovariability in the organization of fibrolamellar bone even within major subclades of therapsids. Fibrolamellar bone may be formed by varying degrees of woven- and parallel-fibered interstitial matrix and incorporates a variety of vascular motifs, and may be zonal (punctuated by cyclic growth marks) or azonal. Ray, Botha & Chinsamy (2004) reported the presence of zonal fibrolamellar bone in many Permian and Middle Triassic taxa, but suggested that sustained (non-cyclic) growth patterns might have arisen occasionally in a number of phylogenetically disparate taxa that do not encompass the immediate ancestors of mammals (gorgonopsian Aelurognathus; eucynodont Cynognathus; and some bidentalian dicynodonts). The abundance of dicynodont fossils in Permian and Triassic rocks and recent advances in their systematic relationships have permitted more detailed comparisons of growth patterns in this diverse subclade (Chinsamy & Rubidge, 1993; Botha, 2003; Ray & Chinsamy, 2004; Ray, Chinsamy & Bandyopadhyay, 2005; Botha & Angielczyk, 2007; Ray, Bandyopadhyay & Bhawal, 2009; Botha-Brink & Angielczyk, 2010; Green, Schweitzer & Lamm, 2010; Nasterlack, Canoville & Chinsamy-Turan, 2012; Ray, Botha-Brink & Chinsamy-Turan, 2012). Phylogenetic comparative surveys have revealed patterns of increasing tissue vascularity during the evolutionary history of bidentalian dicynodonts (especially in Triassic forms like Lystrosaurus), and determinate growth patterns with peripheral rest lines and systematic cortical remodeling in large kannemeyeriiforms (Botha-Brink & Angielczyk, 2010; Green, Schweitzer & Lamm, 2010; Ray, Botha-Brink & Chinsamy-Turan, 2012).

The diversity of growth patterns in other nonmammalian therapsid groups, as well as their phylogenetic and temporal distributions, is incompletely known. A body of literature on nonmammalian cynodont histology has accrued in recent years (e.g., Ricqlès, 1969; Botha & Chinsamy, 2000; Botha & Chinsamy, 2004; Botha & Chinsamy, 2005; Ray, Botha & Chinsamy, 2004; Chinsamy & Abdala, 2008; Botha-Brink, Abdala & Chinsamy, 2012), but sampling has been more limited in other theriodonts, such as the gorgonopsians and therocephalians (Ray, Botha & Chinsamy, 2004; Chinsamy-Turan & Ray, 2012; Huttenlocker & Botha-Brink, 2013). Ricqlès (1969) suggested differential rates of growth between a basal therocephalian from the Middle Permian of South Africa and the Late Permian whaitsiid ‘Notosollasia’ (= Theriognathus). Given the comparatively more vascularized cortical bone in the radius of the whaitsiid (1969: plate IV), Ricqlès suggested that therocephalians might have exhibited accelerated growth rates later in their evolutionary history, paralleling the aforementioned temporal pattern of increasing growth rates in some dicynodonts. More recently, Ray, Botha & Chinsamy (2004) and Chinsamy-Turan & Ray (2012) analyzed additional material from an indeterminate scylacosaurid (erroneously identified as ‘Pristerognathus’), in addition to other therapsid material, and argued for similar ‘flexible’ growth patterns in gorgonopsians, basal therocephalians, and most early cynodonts. The authors suggested that more rigorous taxonomic sampling would better substantiate parallel trends toward a loss in developmental plasticity and acceleration of growth rates as in dicynodonts. Inadequate sampling of eutherocephalians before and after the end-Permian mass extinction limits our understanding of evolutionary patterns in therapsid histomorphology and skeletal growth during this important geologic transition.

Present study

Although eutherocephalians have not been sampled histologically for such comparisons, recent revisions to Permo-Triassic boundary-crossing taxa have necessitated cursory descriptions of eutherocephalian histology for its ontogenetic and paleobiological implications (e.g., Tetracynodon: Sigurdsen et al., 2012; Moschorhinus: Huttenlocker & Botha-Brink, 2013). During the course of this work, we developed a database of histological data and images with the goal of addressing features of life history evolution in Permian and Triassic eutheriodonts (therocephalians and cynodonts), particularly in the context of the end-Permian extinction. Here, we present a reappraisal of limb bone microstructure in Permian and Triassic therocephalians based on new histologic sampling, and offer a hypothesis of the evolution of their growth patterns. We suggest that therocephalians provide a robust study system for investigating the evolution of growth strategies during the Permian-Triassic transition, and a useful point of comparison and contrast to other groups that lived during this time (e.g., dicynodonts, cynodonts).

Specimen selection and histological processing

Specimens were selected based on completeness and availability for histological processing, but a broad sample of the major representatives of South African therocephalians was desired in order to recognize long-term patterns (if present) or clade specific histomorphology. Specimens that were semi-articulated and included diagnostic cranial material were preferred for accuracy of taxonomic identifications. Some specimens were not diagnosable to genus, but were resolved to their respective higher taxon as in the case of five indeterminate scylacosaurids described here. Scylacosaurids are generally difficult to identify unless a complete and accurate antecanine tooth count can be made, and some authors have suggested that the diversity of scylacosaurids is over-split because variations in tooth count may be ontogenetically variable (e.g., Abdala, Rubidge & van den Heever, 2008). The sample therefore included 80 limb elements from 33 individuals in 11 genera: Lycosuchus, Glanosuchus, Moschorhinus, Olivierosuchus, Hofmeyria, Mirotenthes, Theriognathus, Ictidosuchoides, Tetracynodon, Scaloposaurus, and Microgomphodon (Fig. 1), plus additional scylacosaurid material not diagnosed to genus. Thin-sections were prepared using standard histological techniques modified from Chinsamy & Raath (1992) and Wilson (1994). Limb bone midshafts were sampled cross-sectionally and imaged using Nikon Eclipse 50i and LV100 POL petrographic microscopes with a digital image capture system. Histomorphometric variables (discussed below) were measured using NIS-Elements and NIH ImageJ v. 1.42q software (Rasband, 1997–2012). Moschorhinus histomorphology was excluded from present description as it has been discussed elsewhere (Huttenlocker & Botha-Brink, 2013), but data from Moschorhinus were included in the quantitative analyses (Table 1).

Bone tissue typology: definitions and selection of growth proxies

Bone tissue texture exhibits marked variation in therocephalians and other therapsids, varying from highly organized and lamellar to disorganized and woven. Only a few recent studies have integrated qualitative and quantitative assessments of tissue texture and vascular proxies of growth in Permo-Triassic therapsids (Botha & Chinsamy, 2000; Botha & Chinsamy, 2004; Botha & Chinsamy, 2005; Ray & Chinsamy, 2004; Ray, Botha & Chinsamy, 2004; Ray, Chinsamy & Bandyopadhyay, 2005; Ray, Bandyopadhyay & Appana, 2010; Botha-Brink & Angielczyk, 2010; Huttenlocker & Botha-Brink, 2013). Generally, cellular bone forms as osteoblasts become incorporated into the extracelluar (or interstitial) matrix (ECM) forming quiescent osteocytes. The overall bone apposition rate affects the texture of the mineralized ECM, with collagen fibers and crystallites bearing a more lamellar organization under slower growth and a nonlamellar (woven-fibered) texture under faster growth. Parallel-fibered bone, an intermediate tissue-type, can be identified by its ‘streaky’ appearance under polarized light, with the predominant fiber orientation being parallel to the surface of the bone and forming a woven-basket texture in most cases. In cases in which birefringent properties have been disrupted by diagenetic processes, it is possible to approximate the relative organization of mineralized fibers with reference to the organization of the lacunocanalicular network within the ECM (Stein & Prondvai, 2013). In contrast to parallel-fibered bone, woven-fibered bone includes large, globular osteocyte lacunae that are usually densely packed within the mineralized ECM. Nonlamellar tissues (parallel- and woven-fibered) may also frequently incorporate large vascular canals that later become infilled with one or two concentric lamellae forming primary osteons (diagnosed by their ‘Maltese cross’ pattern of birefringence under polarized light). These osteons house passageways for blood vessels and nerves while also contributing to the structural integrity of the bone by providing added bone mass and helping to blunt microcracks (Currey, 2002). The result is a fibrolamellar bone complex (herein ‘FLB’), in which a disorganized, fibrous or nonlamellar interstitial matrix incorporates an anastomosing network of centripetally lamellated primary osteons. Currey (1987) and Currey (2002) defined FLB broadly as a tissue complex formed by parallel- (or woven-) fibered bone with primary osteons (2002: p. 18). By contrast, Ricqlès (1974a) originally restricted the term ‘fibrolamellar’ to tissues formed largely by woven-fibered bone with primary osteons, excluding parallel-fibered bone from this category. Herein, we follow the traditional usage of Ricqlès, but temper this strict definition by noting that parallel- and woven-fibered bone form a continuum that is often ill-defined (and may be present simultaneously in many therapsid bony tissues, even within the same section). Bone cortices formed primarily by lamellar tissue, which forms at relatively slower apposition rates (∼1 µm/day or less) (de Margerie, Cubo & Castanet, 2002; Lee et al., 2013), do not typically incorporate primary osteons, instead bearing simple vascular canals or being avascular. In lamellar bone, the lacunocanalicular network is ordered, the osteocyte lacunae being small and more lenticular in appearance with the long axis oriented parallel to the surface of the bone. Both tissue complexes can be zonal (periodically interrupted by growth marks) or azonal. Growth marks in zonal bone may be present in the form of lines of arrested growth or ‘LAGs’ (denoted by an opaque cement line, traceable around the entire cortex, and indicative of a temporary cessation of growth) or annuli (thin bands of dense, annular tissue, usually parallel-fibered or lamellar, deposited during periods of slowed growth).

For quantitative histomorphometric analysis, two vascular proxies of skeletal growth were selected: cortical vascularity and mean primary osteon diameter. These proxies were selected in order to evaluate the extent to which histological correlatives of growth varied across phylogeny, and whether their evolution was tied to body size or other biological factors. Vascular proxies have offered useful indicators of skeletal growth in extant and extinct tetrapods (Castanet et al., 2000; de Margerie, Cubo & Castanet, 2002; de Margerie et al., 2004; Buffrénil, Houssaye & Böhme, 2007; Cubo et al., 2012), and their utility here allows comparisons with other histological studies of therapsids in which similar measures were used (e.g., Botha-Brink & Angielczyk, 2010; Huttenlocker & Botha-Brink, 2013). Notably, ontogenetic variation in growth may introduce a lesser degree of cortical vascularity in adult bones that exhibited decreasing apposition rates prior to death and burial, and may therefore introduce variation in a single cross-section where this transition is recorded. Nonmammalian therapsids, however, are well suited to relative growth rate estimation based on tissue texture and vascularity, due to their generally thick bone walls (preserving the early record of primary growth) and limited secondary remodeling (Botha-Brink & Angielczyk, 2010). Measurement of cortical vascularity (%CV, the relative area of the cortex that is occupied by porous, vascular spaces) follows Lee et al. (2013). Measurements were restricted to the inner two-thirds to three-quarters of the cortex where the bone formed at high, sustained growth rates (Cubo et al., 2012), and were averaged from ten quadrants sampled circularly around each midshaft cross-section. The subsampled quadrants excluded areas of secondary reconstruction and outer regions of simple canals. We also estimated mean primary osteon diameter (POD) by measuring in microns the transverse (or minimum) widths of 15 primary osteons visible in the subsampled regions and averaging them across all regions within a given midshaft cross-section.

To examine potential effects of size and robusticity on measured histomorphometric variables, we also estimated two proxies of overall bone robusticity: (1) K (the proportional diameter of the medullary region relative to the total diameter of the cross-section; Currey & Alexander, 1985) and (2) relative bone wall thickness or ‘RBT’ (the percentage of the average cross-sectional thickness of the bone wall relative to the total diameter of the cross-section; Chinsamy, 1993). These variables play a frequent role in histological studies of fossil tetrapod bone, as bone robusticity may correspond to habitat preferences or mechanical loading (Wall, 1983; Currey & Alexander, 1985; Currey, 2002; Laurin, Girondot & Loth, 2004; Germain & Laurin, 2005; Kriloff et al., 2008). However, their relationships to size, growth, and vascularity have been underexplored in a comparative framework. These measurements, as well as cross-sectional area at midshaft, were attained using NIH ImageJ, and were tested for correlations with vascular growth proxies using Pearson’s product-moment correlation tests. Histomorphometric data are recorded in Table 1.

Correlation tests

Results 1: qualitative histomorphology

Due to space limitations, the present description is accompanied by a more detailed account in the Supplemental Data file. In general, limb bone cortices in studied therocephalians were thick and formed primarily by FLB with varying degrees of vascularization, depending on size and phylogenetic position. For example, large basal therocephalians from the Middle Permian exhibited subplexiform FLB in the propodial elements with a comparatively high degree of vascularization (%CV ∼13–16%; POD ∼75–168 µm). By contrast, some smaller-bodied taxa in both the Permian and Triassic tended to incorporate more parallel-fibered bone or showed evidence of increased parallel-fibered and lamellar bone deposition with less vasculature toward the outer cortex, indicative of growth attenuation and attainment of somatic maturity (de Margerie, Cubo & Castanet, 2002). This pattern is marked in Mirotenthes, which preserves an external fundamental system (‘EFS’) in some elements (an outer collar of avascular, generally acellular, lamellar bone indicative of virtual cessation of circumferential growth in adult individuals; Cormack (1987); Huttenlocker, Woodward & Hall (2013)). Systematic cortical remodeling by Haversian bone formation was rare and only a few sparse secondary osteons were identified in some Permian scylacosaurian taxa [e.g., an indeterminate scylacosaurid and Moschorhinus, contrary to reports of extensive skeletal remodeling by Chinsamy-Turan & Ray (2012): p. 203)]. As in dicynodonts and other therapsids in which histology is well known, most basal therocephalians and eutherocephalians exhibited remarkably thick bone walls (∼20–35% RBT in forelimb elements of lycosuchids, scylacosaurids, and akidnognathids; ∼40% in forelimbs of hofmeyriids). This condition, however, was lost in baurioids, which were instead characterized by a much thinner cortical bone wall (<25% in the forelimb and propodials) and more slender limb bones.

Permian basal therocephalians and eutherocephalians (Figs. 2A–2C, 2F, 2G; 3A, 3B) generally showed multiple growth zones as well, demarcated by cyclic growth marks indicative of periodic pauses in growth as in other Permian therapsids (Ray, Botha & Chinsamy, 2004; Ray, Botha-Brink & Chinsamy-Turan, 2012; Botha-Brink & Angielczyk, 2010; Botha-Brink, Abdala & Chinsamy, 2012). Although Triassic therocephalians (Figs. 2D, 2E; 3C–3F) showed a variety of vascular motifs and tissue-types (e.g., richly vascularized and disorganized bone tissues in akidnognathids; less vascularized parallel-fibered bone tissue in baurioids), fewer growth marks were recorded in all Triassic species. Small Triassic baurioids in particular showed thinner bone walls with few growth marks, less vascularized limb bone cortices, and smaller, more sparsely distributed primary osteons on average.

Result 2: quantitative histometrics

Pearson’s correlation and phylogeny-independent contrasts

Results of raw and phylogeny-corrected correlation tests are presented in Table 2. Correlation tests on raw data found that vascular growth proxies were strongly positively correlated with size [i.e., the natural log of midshaft cross-sectional area, underscoring greater overall tissue vascularity in large bones from larger taxa (e.g., Lycosuchus, scylacosaurids, Moschorhinus)] (Fig. 4; Table 2). Both %CV and POD also exhibited a strong positive correlation with each other, indicating that %CV and POD represent equivalent proxies for understanding relationships between bone tissue vascularization and growth across the therocephalian clade. Whereas the selected vascular growth proxies shared a consistent positive relationship with size for all data partitions, putative associations between robusticity and size were less clear. Correlation tests on raw size data and RBT were non-significant. Likewise, correlation tests on raw vascular growth proxies and RBT either yielded non-significant p-values or, in the case of the propodial-only data partition, had low correlation coefficients with only a weakly positive association. This result is likely due to the fact that even some small-bodied taxa exhibited unexpectedly thick bone walls, as in the case of the hofmeyriids Hofmeyria and Mirotenthes. Bone wall thickness correlates poorly with size and growth proxies, suggesting that bone robusticity is not necessarily tied directly to growth (perhaps being constrained by ecology, habitat, or mechanical regimen in different groups of therocephalians).

Table 2:

Pearson’s product-moment correlation statistics (Pearson’s r and p) for size, robusticity, and vascular growth proxies in therocephalians.

	Propodial-only		Epipodial-only		Pooled
	r	p *	r	p	r	p
All therocephalians, raw data
ln midshaft area vs. RBT	0.309	0.109	−0.186	0.267	−0.126	0.301
ln midshaft area vs. %CV	0.744	<0.001	0.714	<0.001	0.746	<0.001
ln midshaft area vs. POD	0.853	<0.001	0.847	<0.001	0.850	<0.001
RBT vs. %CV	0.452	0.012	0.034	0.839	0.149	0.217
RBT vs. POD	0.360	0.050	0.111	0.512	0.143	0.237
%CV vs. POD	0.760	<0.001	0.802	<0.001	0.796	<0.001
All therocephalians, independent contrasts
ln midshaft area vs. RBT	−0.466	0.174	−0.476	0.232	–	–
ln midshaft area vs. %CV	0.765	0.006	0.807	0.015	–	–
ln midshaft area vs. POD	0.656	0.028	0.859	0.006	–	–
RBT vs. %CV	−0.077	0.831	−0.224	0.592	–	–
RBT vs. POD	−0.472	0.168	−0.010	0.979	–	–
%CV vs. POD	0.639	0.034	0.877	0.004	–	–

DOI: 10.7717/peerj.325/table-2

Notes:

* r and p-values in boldface are significant at α = 0.05.

Results of phylogeny-corrected correlations further indicated a strong relationship between size and some histometrics (but not with limb bone robusticity) (Table 2). Evolutionary decreases in cross-sectional area of limb bones were generally associated with decreases in average %CV and POD. Eutherocephalians and particularly baurioids demonstrated a noteworthy pattern in which reconstructed ancestor-descendant size reductions of Late Permian and Triassic lineages were associated with decreases in the average level of tissue vascularization. Patterns in limb bone robusticity were less clear. As in the analyses of the raw data, phylogeny-corrected correlations between RBT and size or growth proxies were non-significant.

General histological patterns

Phylogenetic patterns

Previous interpretations of growth patterns in early therocephalians were based on limited information from incomplete specimens (Ricqlès, 1969; Ray, Botha & Chinsamy, 2004; Chinsamy-Turan & Ray, 2012). Additional specimens described here suggest that at least some large-bodied predators from the Middle Permian, including Lycosuchus and some scylacosaurids, exhibited subplexiform FLB in propodial elements. Some of the larger-bodied predators in the sample also showed the highest degree of cortical vascularization. This is remarkable, as subplexiform FLB is one of the most rapidly deposited tissue-types in archosaurian and mammalian limb bones, and similarly vascularized bone in birds and mammals forms periosteally at a rate generally greater than 15 µm/day (Castanet et al., 2000; de Margerie, Cubo & Castanet, 2002; Cubo et al., 2012). It is noteworthy that the subplexiform tissue complex, present in Lycosuchus and scylacosaurids, represents the prototype upon which earlier workers first described FLB [see Stein & Prondvai (2013) for a review]. This subplexiform condition is conspicuously lacking in the whaitsiid Theriognathus(Fig. 5) and some later eutherocephalians. Consequently, the evolutionary scenario suggested by Ricqlès (1969) and critically re-examined by Chinsamy-Turan & Ray (2012) should be revised: basal therocephalians exhibited highly vascularized FLB and grew relatively rapidly over many growing seasons but with frequent interuptions, in contrast to some later eutherocephalians (with the exception of the akidnognathid Moschorhinus, which maintained relatively fast growth).

Moreover, elevated vascularity and rapid growth may be a characteristic of larger bodied taxa, an interpretation that is supported by observations in other large predatory theriodonts such as the cynodont Cynognathus and some large gorgonopsians (Botha-Brink, Abdala & Chinsamy, 2012; Chinsamy-Turan & Ray, 2012). Importantly, evolutionary decreases in body size during the Permian (Huttenlocker, 2013; Huttenlocker, 2014) were associated with decreases in overall degree of skeletal vascularization. Both hofmeyriids and Theriognathus (in spite of the latter’s substantial size) showed modest cortical vascularity and smaller mean primary osteon diameters than in basal lycosuchids, scylacosaurids or akidnognathids, but the baurioids (which are deeply nested in the tree and generally smaller-bodied) had the least vascularized bone tissues (Fig. 6). These clade-level patterns suggest that reductions in body size of some Late Permian eutherocephalians were coupled with decreased rates of skeletal apposition leading up to the end-Permian extinction, particularly in the baurioid lineage.

Growth and the end-permian extinction

The end-Permian extinction was associated with evidence of temporary body size reductions, restricted primarily to faunas of the earliest Triassic (Induan) (although size reductions in some therapsids may reflect larger-scale patterns of body size evolution tracing back to the Permian; Huttenlocker, 2013; Huttenlocker, 2014). Such size shifts have been documented previously based on invertebrate burrows, foraminifera, brachiopods, gastropods, bivalves, conodonts, and fish (Twitchett & Barras, 2004; Payne, 2005; Twitchett, 2007; Luo et al., 2008; Mutter & Neuman, 2009; Metcalfe, Twitchett & Price-Lloyd, 2011; Song, Tong & Chen, 2011; Rego et al., 2012). Nevertheless, questions remain regarding the evolution of growth patterns and their underlying influence on size shifts (Twitchett, 2007; Harries & Knorr, 2009). For example, organisms may have experienced slower overall growth rates in response to more limited resources or environmental degradation during the Permian-Triassic transition. Growth mark analyses on marine brachiopod shells (i.e., ‘Lingula’) have indicated slow growth rates with frequent interruptions to growth in earliest Triassic shells (Metcalfe, Twitchett & Price-Lloyd, 2011). The authors attributed their results to suboptimal environmental conditions, including episodes of benthic hypoxia, hypercapnia, ocean acidification, and/or disruptions to primary productivity. Such environmental factors would place strong physiological limits on shell formation. However, identifications of lingulid specimens were tenuous given conservatism in external morphology of the group, and growth marks were only studied in shells from Triassic survival and recovery faunas without being compared to Permian specimens. Alternatively, surviving lineages could have exhibited heterochronic shifts shortening time to maturity (e.g., progenesis) and thus rapid, sustained growth over a brief growth period. Progenesis is a classic example of an r-selection strategy in perturbed or unstable environments (Gould, 1977) and has been identified as a potential mechanism of some Lilliput patterns (Harries, Kauffman & Hansen, 1996). Finally, if shifts in body size distributions were influenced primarily by differential extinction of large- versus small-bodied forms, then there may have been no resulting changes in growth patterns at all (that is to say, changes in growth dynamics are not necessary to explain post-extinction body size distributions).

Growth patterns in Permo-Triassic therapsids compared

New data on growth patterns in nonmammalian therapsids, as well as other Permo-Triassic tetrapods, offer the potential to further evaluate patterns of selectivity during mass extinctions (Botha-Brink & Angielczyk, 2010; Botha-Brink & Smith, 2012). Bone histology has been studied in numerous genera of Permian and Triassic dicynodont therapsids (Chinsamy & Rubidge, 1993; Ray, Botha & Chinsamy, 2004; Ray, Bandyopadhyay & Bhawal, 2009; Ray, Botha-Brink & Chinsamy-Turan, 2012) and recent progress in dicynodont paleobiology has permitted evolutionary investigations of their growth patterns (Botha-Brink & Angielczyk, 2010). Increased rates of skeletal growth originated relatively early, either within or prior to the divergence of bidentalian dicynodonts (e.g., Dicynodon, Lystrosaurus) by the early-Late Permian. Medium-to-large Late Permian dicynodonts continued to show patterns of increased cortical vascularity, especially within the Permo-Triassic boundary-crossing genus Lystrosaurus, which demonstrated some of the highest levels of tissue vascularity (∼20%). Triassic specimens of Lystrosaurus showed relatively higher vascularity and fewer growth marks. In the therocephalian Moschorhinus, the only large therapsid predator to cross the Permian-Triassic boundary, within-lineage size reduction was associated with the maintenance of rapid but attenuating growth over a short period (Huttenlocker & Botha-Brink, 2013). This pattern suggests that Moschorhinus may have been under selection for reaching its adult size relatively more quickly than in other Permian therocephalians. Notably, Moschorhinus exhibited comparably high levels of tissue vascularity as Lystrosaurus during the Triassic (∼20–25%).

Whereas anecdotal evidence in Lystrosaurus and Moschorhinus is suggestive of within-lineage heterochronic shifts, clade-level patterns introduce a more complex explanation for observed body size reductions in therocephalians. In particular, body size reductions occurred early during the evolution of eutherocephalians and were associated with a lesser degree of cortical vascularity in medium-to-small-bodied Permian and Triassic forms (e.g., hofmeyriids and especially baurioids). The two major subclades of therocephalians that persisted into the earliest Triassic, Akidnognathidae and Baurioidea, revealed distinctly different growth patterns from each other as interpreted through their tissue texture and degree of vascularity. Their histology suggests a bimodality of life history strategies in earliest Triassic therocephalians: small-to-medium akidnognathids with well-vascularized (fast-growing) bone and smaller-bodied baurioids with less vascularized (slower-growing) bone. However, both groups shared a reduced number of growth marks compared to their Permian relatives in addition to their generally smaller sizes. This nuance is not evident from the quantitative analysis based on vascular proxies of growth rate alone, but is evident from growth mark counts in surveyed specimens. Permian theriodonts that have been sampled histologically, including some gorgonopsians, the cynodont Procynosuchus, basal therocephalians, Permian (but not Triassic) specimens of Moschorhinus, hofmeyriids, and Theriognathus, typically showed evidence of prolonged, multi-year growth often to larger body sizes, a pattern that is not represented in earliest Triassic therapsids sampled to date (Botha-Brink & Angielczyk, 2010; Botha-Brink, Abdala & Chinsamy, 2012).

The present discussion of associations between micro- and macrostructure provides a more functional and wholistic context for the evolution of histological features in Permian and Triassic therocephalians. However, it is important to note that the patterns discussed here have not been evaluated quantitatively and therefore merit future investigation. Recent large-scale phylogenetic studies in other therapsids have addressed how life history and functional morphology might have contributed to the success of some groups during the Late Permian and Triassic (e.g., bidentalian dicynodonts). For example, the degree of development of the secondary palate has been linked anecdotally to new environmental conditions with the onset of the Early Triassic, particularly a dramatic decline in atmospheric pO₂ from the Late Permian and continuing through the Middle and Late Triassic (Retallack, 2003). However, a rigorous collections-based study found no difference between secondary palate length in Permian and Triassic dicynodonts when corrected for size and phylogeny (Angielczyk & Walsh, 2008). Similarly, a bony secondary palate can be found in a number of therocephalian subgroups, both in the Permian and Triassic, and Triassic therocephalian faunas consisted of species both with (e.g., baurioids) and without (e.g., akidnognathids) a secondary palate (Fig. 7). No clear association with global hypoxia and respiratory efficiency can be made based on this character, and, as a result, other factors may have been more important in maintaining and shaping the evolution of the secondary palate (e.g., feeding mechanics; Thomason & Russell, 1986).

Similar studies addressing the possible effects of hypoxia on cortical tissue vascularity have found few differences between Permian and Triassic therapsids, instead demonstrating that highly vascularized tissues with enlarged canals evolved early in bidentalian dicynodonts (Botha-Brink & Angielczyk, 2010). In therocephalians, variation in the degree of vascularization of limb cortices is best explained by size variation observed across clades (Fig. 7). Most smaller-bodied groups exhibited less vascularized limb bone cortices, a character that evolved relatively early in the evolutionary history of eutherocephalians and held over in some small Triassic taxa. The smallest Triassic forms, derived bauroids, typically had lighter, more gracile skeletons with an open medullary cavity, thinner bone walls, and few to no growth marks. In addition to longer, more slender limb bones, they also had an elongate hind foot with a calcaneal tuber on the heel, maxillary bridge forming a bony secondary palate (discussed above), increasingly specialized multi-cusped teeth, and often lacked a parietal foramen (or reduced the pineal body altogether as in Tetracynodon; Sigurdsen et al., 2012). The selective value of maintaining a parietal eye for temperature regulation and modulating melatonin production may have been diminished in some small, nocturnal or crepuscular baurioid therocephalians, or in short-lived animals less dependent on seasonal cues in photoperiodicity (Roth, Roth & Hotton, 1986). The latter scenario is consistent with the lack of cyclic bone deposition and paucity of growth marks in small Triassic baurioids (although one genus, Scaloposaurus, is distinguished from other small Triassic baurioids in its retention of the parietal foramen and pineal body).