Were the synapsids primitively endotherms? A palaeohistological approach using phylogenetic eigenvector maps

1. Introduction

One of the greatest challenges of current research in palaeobiology is the inference of physiological features of extinct vertebrates. Among them, endothermy is particularly relevant because this feature is linked to a wide array of anatomical, physiological and behavioural features. Endothermy has been defined as the presence of any mechanism of non-shivering thermogenesis that increases both body temperature and resting metabolic rate (RMR) [1]. Mammals, the only extant synapsids, are endotherms [2–5]. The origin of mammalian endothermy is poorly constrained both phylogenetically and temporally, in spite of the fact that the acquisition of this ability is a key innovation (endotherms are less dependent on climatic conditions and can occupy more ecological niches [6]). Many approaches have been used to constrain the phylogenetic and temporal frames of this acquisition. One of them involves identifying unequivocal anatomical correlates of endothermy such as respiratory turbinates or evidence for an insulative pelage [7]. Respiratory turbinates are rarely preserved in fossils but bone ridges of turbinate attachment are first recognizable in therocephalians and in cynodonts [7], so they may have been acquired by the eutheriodonts. Oldest evidence for an insulative pelage (fossilized fur impressions) has been found in the Middle Jurassic non-mammalian therapsids (Castorocauda lutrasimilis [8], Megaconus mammaliaformis [9], Agilodocodon scansorius [10]). Thermal modelling [11], the isotopic composition of mineralized remains [12] and bone palaeohistology [1,13–15] have also been used to infer the thermophysiology of non-mammalian synapsids. Here, we will use this last approach.

Traditionally, qualitative bone histology has been used to estimate the bone growth rate (BGR) [16–18], indirectly linked to the metabolic rate, and to the thermometabolism [19,20]. This framework is based on (A) theoretical grounds ‘the types of tissue deposited in the bones of extinct animals are the most direct evidence of basal metabolic rates, because they directly reflect growth rates […]. The sustained deposition of fast-growing bone tissues, as displayed by mammals, birds and other dinosaurs, must reflect sustained high basal metabolic rates [21, p. 123] and (B) empirical evidence: the variation of BGR significantly explained the variation of RMR in a sample of extant amniotes [19]. More recently, quantitative histology and phylogenetic eigenvector maps (PEMs) [13,22] allowed direct inference of the RMRs of extinct Archosauromorpha [1,13], Sauropterygia [15] and extinct Therapsida [14].

Olivier et al. [14] used this last approach and provided evidence for an ancestral acquisition of endothermy at the node Neotherapsida. However, the sample composition of this study [14] did not allow testing of the hypothesis of an acquisition of endothermy in a more inclusive node. Thus, the present study is aimed at further constraining the temporal range and the phylogenetic frame of the acquisition of endothermy in synapsids. To do so, we will infer the metabolic rates of a sample of extinct synapsids (including Carboniferous non-neotherapsid synapsids) using a recently defined bone histological variable, the relative primary osteon area (RPOA) (named previously ‘primary osteon density’ in [15]), and PEMs [13,22].

2. Material and methods

3. Results

4. Discussion

The presence of endothermy in extinct synapsids has been inferred using different proxies. Using proxies is obviously necessary because the thermometabolic regime of extinct taxa is not accessible through direct observation of fossilized features. The presence of anatomical features linked to endothermy in extant mammals, such as the presence of fur [53] or of respiratory turbinates [7], has been widely used. The former has traditionally been seen as evidence for, at least, a near-endothermic condition, because of its homeothermic function by retaining the generated heat [7,53]. The earliest known occurrences of fossilized fur impressions have been dated to the Middle Jurassic (Castorocauda [8], Megaconus [9], Agilodocodon [10]). Considering that endothermy has been defined by Cubo & Jalil [1] as the presence of any mechanism of non-shivering thermogenesis (e.g. [5]) that increases both body temperature and RMR, the presence of fur is not definitive evidence for endothermy because, at best, it can indicate a homeothermic condition.

The same reasoning can be applied to respiratory turbinates: these are osseous or cartilaginous convoluted pieces in the nasal cavity. The oldest presence of respiratory turbinates in synapsids dates from the Lopingian (Permian) [54]. Respiratory turbinates warm and moisten the incoming air during inhalation and do the opposite during expiration [54,55]. Doing so, they help to keep a stable temperature and limit the loss of moisture through breathing. According to Ruben, endothermy cannot be sustained without these structures [56], leading them to be considered the best evidence for an endothermic condition. However, the obligate presence of respiratory turbinates to achieve homeothermy has been challenged recently [57].

Another proxy is geochemistry. The isotopic ratio between ¹⁶O, the common form of oxygen, and ¹⁸O, is used to infer body temperature. This ratio, measured in calcium phosphate from bones and, especially, dentary enamel, is dependent on environmental and body temperatures during the formation of the sample [12,58]. We can infer the relative internal body temperatures from calcium phosphate using fractionation equations [12,58–61]. Lots of studies have been conducted in different organisms, such as ichthyosaurs and plesiosaurs [61]. Recently, Rey et al. [12] studied a large sample of Neotherapsida using this approach and discussed two competing hypotheses: either a single acquisition of endothermy at the Neotherapsida node, or two independent acquisitions at the node Eucynodontia and the node Lystrosauridae + Kannemeyeriiformes. According to these results, the endothermy was acquired in the synapsid clade, during the Guadalupian (Permian) with the single acquisition scenario. With the evolutionary scenario suggesting convergent origin, endothermy may have appeared during the Lopingian (Permian), as for respiratory turbinates.

On the specific case of Dimetrodon, some studies tried to estimate the effect of the sail in thermal regulation [11,62]. Authors created models in order to simulate the heat exchanges and the specific role of the sail. They concluded that the sail allowed Dimetrodon to be active earlier than similar but sail-less predators. However, this proxy deals with external heat income and not with endogenous mechanisms of non-shivering thermogenesis (endothermy).

Bone histology, the last proxy, has largely been based on Amprino's rule [63]: the organization of the collagen matrix records bone growth rate [63]. In other words, fast growth leads to a poorly organized type of tissue (i.e. woven bone), whereas a low growth rate produces a highly organized type of tissue (lamellar or non-lamellar parallel-fibred bone) [25]. Thus the presence of woven bone has been interpreted as evidence for fast bone growth rate, linked to endothermy [64–66]. However, many counterexamples exist: many ectotherms, e.g. Alligator [67] or Varanus [68], are able to form woven bone. By contrast, RPOA, the histological feature recently proposed by Fleischle et al. [15] under the designation ‘primary osteon density’, is tightly linked to bone growth rate and to RMR. Most endothermic amniotes are able to form an initial scaffold composed of woven bone that includes large cavities and produces a rapid volume expansion of the cortex during early ontogenetic stages (e.g. in the long bones of ratites [69] and in the long bones of the king penguin [70]). On the contrary, in most ectothermic amniotes, periosteal bone is composed of parallel-fibred bone including small cavities, if any, and producing a slow volume expansion of the cortex (e.g. in squamates [71]). Consistently, RPOA was used in this study to infer the thermometabolic regime (ectothermic or endothermic) of our sample of extinct synapsids.

RPOA explains a significant fraction of the variation of RMR (quantified using respirometry) in the sample of extant tetrapods (results obtained using PGLS regressions). Thus, assuming the actualism principle, a priori the former can be used to infer the latter in extinct taxa. Inferred RMR values for our sample of extinct synapsids using PEMs are shown in figures 1 and 2. The value inferred for Mycterosaurus should be analysed with caution because the branch length of this taxon (42.3 Myr) is bigger than 10% of the total height of the tree (358.9 Myr) and, according to [50], this fact leads to a loss of accuracy when performing inferences. Our inferences for Lystrosaurus sp. (1.936 ml O₂ h⁻¹ g^−0.67; 1.491–2.515) and Oudenodon bainii (2.397 ml O₂ h⁻¹ g^−0.67; 1.828–3.142) are congruent with those obtained by Olivier et al. [14], who used a different histological feature (osteocyte lacunae density). We performed maximum-likelihood ancestral states reconstructions using these values and those measured in extant species (figure 2). As expected, RMRs inferred for nodes 49 (birds) and 36 (mammals) in figure 2 are significantly higher than the threshold separating endotherms from ectotherms (1.5 ml O₂ h⁻¹ g^−0.67). Likewise, RMR inferred for node 35 (Anomodontia: (Lystrosaurus-Oudenodon)) is also significantly higher than the quoted threshold. However, the value inferred for node 34 (Therapsida): (Lystrosaurus-Oudenodon) Mammals) does not exclude ectothermy. Finally, we performed a parsimony optimization of the presence of endothermy (figure 3). For this, we assigned to each extinct taxon a thermometabolic regime (ectothermy or endothermy) depending on whether the inferred values were significantly higher, lower or not significantly different from the RMR value separating ectotherms from endotherms (1.5 ml O₂ h⁻¹ g^−0.67), using the results obtained in the present study and those obtained in previous studies using PEMs [1,13–15]. According to the reconstructed ancestral thermometabolic regime, endothermy arose three independent times in amniotes: in Archosauromorpha, in Sauropterygia and in Neotherapsida (figure 3).

The thermometabolism of other groups of tetrapods has been analysed: ichthyosaurs, mosasaurs, pterosaurs and Thalattoshuchia. A study using geochemistry suggested a high inner temperature and a fully developed endothermic condition for ichthyosaurs and a medium inner temperature and an incipient endothermy for mosasaurs [61]. This result is congruent with that obtained by de Buffrénil & Mazin, who found the presence of woven bone, suggesting a high growth rate and, so, a high metabolic rate in ichthyosaurs [72]. The case of mosasaurs is more complex. Houssaye et al. [73] analysed the bone histology of mosasaur limb bones and found a predominance of parallel-fibred bone, which doesnot need a high metabolic rate to occur. But they also described unusual parallel-fibred bone, which could reflect a growth rate intermediate between that typical of parallel-fibred bones and that associated with woven bone. The thermometabolic condition of mosasaurs is ambiguous, and gigantothermy, a heat conservation mechanism, linked to high body mass, is not excluded [73]. For pterosaurs, de Ricqlès et al. [74]) analysed the histology of limb bones and found the presence of woven bone, especially in juvenile specimens, leading them to the conclusion that these animals had a growth curve similar to those found in extant birds. A high growth rate suggests an endothermic condition for pterosaurs. This condition would be correlated with the presence of air sacs detected in some species [75]. Such organs are very important in modern birds to sustain a unidirectional airflow in the lungs and a high metabolic rate associated with flight. The finding of unidirectional airflow in the lungs of alligators [76] suggests that this condition would be primitive for archosaurs. Finally, Séon et al. [77] have shown that Thalattosuchia (marine crocodylomorphs) had body temperatures intermediate between those of typical ectotherms and those of typical endotherms. Among them, metriorhynchids (pelagic active predators) may have been slightly warmer than teleosaurids (coastal ambush predators) [77].

Supplementary Material

Supplementary Material

Supplementary Material

Perspectives

Further research using quantitative histology and PEMs will help to elucidate the thermometabolic condition of ichthyosaurs, pterosaurs, mosasaurs, Thalattosuchia and early synapsids. The predictive power of inference models will increase by including additional explanatory factors such as life-history traits, as well as new osteohistological features correlated to thermometabolism, such as for instance, minimum diameter of vascular canals [78].

Data accessibility

The electronic supplementary material contains new data used in the analyses performed in this study

Authors' contributions

J.C. designed research; M.G.F.-B. performed histological quantifications, adapted the script from sources and made statistical analyses; J.C. and M.G.F.-B. wrote the paper

Competing interests

We declare we have no competing interests