1. Ecology

Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triassic therapsid clades

  1. Kévin Rey  Is a corresponding author
  2. Romain Amiot
  3. François Fourel
  4. Fernando Abdala
  5. Frédéric Fluteau
  6. Nour-Eddine Jalil
  7. Jun Liu
  8. Bruce S Rubidge
  9. Roger MH Smith
  10. J Sébastien Steyer
  11. Pia A Viglietti
  12. Xu Wang
  13. Christophe Lécuyer
  1. Univ Lyon, Université Lyon 1, Ens de Lyon, CNRS, UMR 5276 LGL-TPE, France, France
  2. University of the Witwatersrand, South Africa
  3. Univ Lyon, Université Lyon 1, CNRS, UMR 5023 LEHNA, France, France
  4. Institut de Physique du Globe de Paris, France
  5. Centre de Recherches en Paléobiodiversité et Paléoenvironnements, UMR 7207 CNRS-MNHN-UPMC, Museum National d’Histoire Naturelle, France
  6. Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, China
  7. Iziko South African Museum, South Africa
  8. Institute of Geology and Geophysics, Chinese Academy of Sciences, China
  9. Institut Universitaire de France, France
https://doi.org/10.7554/eLife.28589
Share this article
Cite this article
  1. Kévin Rey
  2. Romain Amiot
  3. François Fourel
  4. Fernando Abdala
  5. Frédéric Fluteau
  6. Nour-Eddine Jalil
  7. Jun Liu
  8. Bruce S Rubidge
  9. Roger MH Smith
  10. J Sébastien Steyer
  11. Pia A Viglietti
  12. Xu Wang
  13. Christophe Lécuyer
(2017)
Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triassic therapsid clades
eLife 6:e28589.
https://doi.org/10.7554/eLife.28589
  2. Download BibTeX
  3. Download .RIS

Abstract

The only true living endothermic vertebrates are birds and mammals, which produce and regulate their internal temperature quite independently from their surroundings. For mammal ancestors, anatomical clues suggest that endothermy originated during the Permian or Triassic. Here we investigate the origin of mammalian thermoregulation by analysing apatite stable oxygen isotope compositions (δ18Op) of some of their Permo-Triassic therapsid relatives. Comparing of the δ18Op values of therapsid bone and tooth apatites to those of co-existing non-therapsid tetrapods, demonstrates different body temperatures and thermoregulatory strategies. It is proposed that cynodonts and dicynodonts independently acquired constant elevated thermometabolism, respectively within the Eucynodontia and Lystrosauridae + Kannemeyeriiformes clades. We conclude that mammalian endothermy originated in the Epicynodontia during the middle-late Permian. Major global climatic and environmental fluctuations were the most likely selective pressures on the success of such elevated thermometabolism.

https://doi.org/10.7554/eLife.28589.001

eLife digest

School textbooks often refer to “cold-blooded” and “warm-blooded” animals, but these terms are misleading. Rather than being cold, animals like reptiles have body temperatures that are mostly determined by their external environment and can actually achieve high body temperatures, for example, by basking in the sun. By contrast, “warm-blooded” mammals produce their own heat and typically maintain a body temperature that is warmer than their environment. As such, so-called warm-blooded animals are more accurately referred to as “endotherms” and cold-blooded animals as “ectotherms”.

Endothermic animals share several characteristics, including insulating layers – like fur or feathers – that keep the body warm, and a secondary palate that separates the mouth and nose for continuous breathing, even while eating. Many of these traits are seen in fossils belonging to a group of animals called the therapsids. Also known as the “mammal-like reptiles”, these animals are descended from ectothermic reptiles but are the ancestors of the endothermic mammals. They dominated the land between 270 and 220 million years ago, during periods of time called the Permian and the Triassic. They also survived two major mass extinction events, including the most devastating mass extinction in all of Earth’s history. However, when the ancestors of mammals became truly endothermic remains an open question. Previous studies that have tried to determine this by focusing on the physical characteristics of therapsids have not yet given a consistent date.

Rey et al. took a new approach to answer when endothermy first evolved in the mammal-like reptiles, and instead looked at the chemical makeup of minerals in over 100 fossils. Oxygen can exist in different forms called stable isotopes: oxygen-16 and the rarer and heavier oxygen-18. The ratio of these two isotopes in a fossil will depend on, among other things, where the animal lived and, importantly, its body temperature. Therefore, Rey et al. compared oxygen-containing minerals in the bones and teeth of therapsids to those of other animals that lived alongside them to look for signatures that indicated differences in body temperature and how it was regulated.

It appears that two different branches of the therapsid’s family tree independently became endothermic. One branch includes the mammals and their direct ancestors, while the second is more distantly related to mammals. Both became endothermic towards the end of the Permian Period, between about 259 and 252 million years ago. Based on these findings, Rey et al. suggest that endothermy allowed these animals to better cope with fluctuating climates, which helped them to be among the few species that survived the mass extinction event at the end of the Permian.

Going forward, these new findings can help scientists to understand which physical characteristics were necessary for endothermy to first develop and which helped to optimize it afterwards. Furthermore, they also suggest that endothermic animals are more able to survive fluctuations in climate, which could guide efforts to protect modern-day endangered species that are most at risk from the ongoing effects of climate change.

https://doi.org/10.7554/eLife.28589.002

Introduction

One key adaptation enabling tetrapods to cope with fluctuating climatic conditions was the acquisition of endothermy (Paaijmans et al., 2013). This character is defined here as the ability to actively produce body heat through metabolic activity (Cannon and Nedergaard, 2004). Its development and anchoring in populations constitutes a major step in vertebrate evolution because it modified the energy relationships between organisms and their environments. By actively raising and maintaining body temperature within a narrow range that allows optimal physiological and biochemical functioning, endothermic vertebrates are able to colonise environments with extreme thermal conditions, for example freezing at high latitudes and altitudes (Day et al., 2015). Endothermy is commonly associated with homeothermy, being the capacity to regulate the body heat through metabolic activity as well. This combination corresponds to one end of a gradient of thermoregulatory strategies observed in living animals. The other end of the spectrum is ectothermy combined with poïkilothermy which animals use as a thermoregulatory strategy to increase their body temperature toward optimal levels by using external heat sources. Their body temperature therefore traces that of their surroundings and is the most commonly occurring energy saving strategy. Amongst modern vertebrates, various thermoregulatory strategies have been adopted between these two end-members, such as regional endothermy (Bernal et al., 2001; Katz, 2002) or inertial homeothermy (McNab and Auffenberg, 1976), and only birds and mammals fall within the endothermy end of the spectrum. It has been proposed that bird thermoregulation originated within non-avian dinosaurs (Seebacher, 2003; Amiot et al., 2006; Grady et al., 2014), or even earlier within basal archosauriforms (Farmer and Carrier, 2000; de Ricqlès et al., 2003; Seymour et al., 2004; Summers, 2005; Gower et al., 2014; Legendre et al., 2016). Various approaches have been tried by many researchers to assess the origin of mammalian endothermy (McNab, 1978; Bennett and Ruben, 1986; Farmer, 2000; Hillenius, 1992, 1994; Kemp, 2006a; Khaliq et al., 2014; Owerkowicz et al., 2015; Benoit et al., 2016b; Crompton et al., 2017). Some consider the appearance of endothermy to have either occurred during the transition from basal synapsid ‘pelycosaurs’ to therapsids, and to be either due to a shift in foraging ecology (Hopson, 2012) or due to a response to the availability of a seasonally arid, savanna-like biome by the end of the Early Permian (Kemp, 2006b).

How and why endothermy evolved in mammals remains a contentious issue, mostly because of the very low fossilization potential of anatomical and behavioural features associated with thermoregulation. Amongst the latter features, the presence of hair as an insulating integument is unequivocally associated with endothermy in all extant mammals. The oldest synapsid fossils preserved with fur imprints are Castorocauda (Ji et al., 2006) and Megaconus (Zhou et al., 2013). These early relatives of mammals were recovered from the Middle-Late Jurassic of China, implying that hair and fur appeared before ~165 Ma. The occurrence of retracted, fully ossified and non-ramified infraorbital canals (a structure associated with the presence of maxillary vibrissae) within non-mammaliaform Prozostrodontia, implies an older age of approximately 240 to 246 Ma for the occurrence of fur and hair (Benoit et al., 2016b).

Another anatomical character interpreted as associated with endothermy is the bony secondary palate. This is a feature associated with efficient respiratory capabilities considered to be linked to the high energy required for elevated metabolic rates. In some Triassic non-mammaliamorph therapsids, bauriid therocephalians and cynodonts, a bony secondary palate is fully developed (Abdala et al., 2014). It is noteworthy that a complete bony secondary palate is also present in dicynodonts, however it is primarily formed by the premaxilla (King, 1988) and not the maxilla as documented in therocephalians, cynodonts and extant mammals. Although a secondary osseous palate is ubiquitous in mammals, it also occurs in a few ectotherms (crocodiles, scincid lizards), thus questioning its direct link to endothermy (Bennett and Ruben, 1986).

Almost all extant endotherms possess nasal turbinate bones covered with mucosa that reduce heat loss and moisten air during respiration (Owerkowicz et al., 2015). This feature, absent in extant ectotherms (Witmer, 1995), may have been present in therocephalian, cynodont and dicynodont therapsids, as postulated from bony ridges in the nasal cavities interpreted as supports for the turbinate complex (Hillenius, 1992, 1994; Crompton et al., 2017).

A peculiar histological structure of fast-growing bone associated with highly vascularised woven-fibred matrix and primary osteons known as fibrolamellar bone (FLB), is another feature often used as evidence of a high metabolic activity (Montes et al., 2010; Legendre et al., 2016). Accordingly, several bone palaeohistological studies have addressed the quest for the presence of FLB in therapsids (de Ricqlès, 1972, 1979; Botha, 2003; Botha and Chinsamy, 2001, 2004; Ray et al., 2004; Olivier et al., 2017). Ray et al. (2004) and Olivier et al. (2017) analysed several therapsid groups (anomodont, gorgonopsian, therocephalian, cynodont) and found FLB in some genera (Aelurognathus, Pristerognathus, Tritylodon, Oudenodon, Lystrosaurus, Moghreberia), suggesting sustained fast growth, and thus elevated metabolic activity. The presence of FLB has also been demonstrated in some earlier non-therapsid synapsids such as Sphenacodon, Dimetrodon or even Ophiacodon (Huttenlocker et al., 2010; Shelton et al., 2012; Shelton and Sander, 2017). However, FLB also occurs in a few ectotherms such as in some turtles and crocodilians, and is absent in small mammals and passerine birds (Bouvier, 1977), showing that FLB is mostly correlated with high growth rates, which does not always correlate to high metabolic rates. Therefore, these characters alone cannot be considered as definitive evidence of endothermy, leaving the question of the timing and selection pressure for the origin of mammal endothermy still heavily debated.

Because the oxygen isotope fractionation between bone or tooth phosphate and body fluids is temperature dependent, and phosphate has a strong resistance to diagenetic alteration, oxygen isotope compositions of therapsid apatite phosphate (δ18Op) has been used in this pilot study to investigate the origin of mammalian endothermy. Indeed the δ18Op value of vertebrate apatite (bone, tooth) reflects both the oxygen isotope composition of the animal body water (δ18Obw) and its body temperature (Tb). Body water derives mainly from drinking meteoric water or plant water (D’Angela and Longinelli, 1990; Kohn, 1996a), and the δ18O value of this water in turn depends on climatic parameters such as air temperature, hygrometry, and amount of precipitation (Dansgaard, 1964; von Grafenstein et al., 1996; Fricke and O'Neil, 1999).

Variations in the δ18O values of ectotherm apatite, along with increasing latitude, are expected to reflect decreasing air temperatures as their body temperatures follow those of the environment. In contrast endotherms, which have a constant body temperature, should not be affected by environmental temperatures changes. Moreover, physiological adaptation to specific habitat use (aquatic, semi-aquatic or terrestrial) affects the δ18Obw value by controlling the magnitude of body input and output oxygen fluxes, some of them being associated with oxygen isotopic fractionations (Amiot et al., 2010). Therefore, co-existing endotherms and ectotherms should have distinct apatite δ18Op values reflecting their body temperature and ecological differences. By comparing apatite δ18Op values of therapsids with those of co-existing ectotherms of known ecologies at various palaeolatitudes, it should be possible to infer therapsid thermophysiology, a methodology that has previously been applied to non-avian dinosaurs (Fricke and Rogers, 2000; Amiot et al., 2006).

Following the protocol of previous research undertaken to establish the Permo-Triassic climatic conditions that prevailed during which South African tetrapods, including therapsids, radiated (Rey et al., 2016), this study aims to investigate thermophysiological strategies developed by various Permo-Triassic therapsid groups using the stable oxygen isotope compositions of their phosphatic remains. Our results add new data to the discussion of the origin of mammalian endothermy and its link to global climatic change.

Results

Permian therapsids

The 13 sampled South African Permian therapsids come from three different assemblage zones (AZ) of the Beaufort Group: the lower Tapinocephalus AZ, the Tropidostoma AZ and the lower Daptocephalus AZ (Viglietti et al., 2016).

Oxygen isotope compositions of three therapsid genera (Dicynodon, Diictodon and Oudenodon) from the youngest assemblage zone (lower Daptocephalus AZ) and seven therapsid genera (Aelurosaurus, Diictodon, Ictidosuchoides, Oudenodon, Rhachiocephalus, Tropidostoma and a basal cynodont) from the Tropidostoma AZ were respectively compared with one co-occuring Rhinesuchus and one co-occuring rhinesuchid. Differences between all the Permian therapsids and ectothermic temnospondyl range from +1.1 ± 0.6‰ to +8.0 ± 0.9‰ (Figure 1A), encompassing the expected range for which therapsids are considered ectothermic.

Figure 1
Download asset Open asset
δ18Op differences between Permian therapsids and other tetrapods.

Differences in δ18Op values between therapsids and stereospondyls (white symbols) and between therapsids and parareptiles (black symbols) from the same localities are plotted against their corresponding palaeolatitude. A theoretical framework based on modern temperature gradient (0.6 ± 0.1°C/°Lat; see Appendix 1) and phosphate-water-temperature oxygen isotope fractionation (Lécuyer et al., 2013) predicts various δ18Op value differences. The lighter orange and red areas correspond to the uncertainty of the temperature gradient. Dap.: Daptocephalus; Tap.: Tapinocephalus.

https://doi.org/10.7554/eLife.28589.003

In addition, δ18Op values of the therapsids Dicynodon, Diictodon and Oudenodon are compared to those of the supposedly semi-aquatic parareptile Pareiasaurus (Ivakhnenko, 2001; Kriloff et al., 2008) (see Appendix 1), with an observed range of +4.3 ± 0.4‰ to +6.8 ± 0.5‰ (Figure 1A) which is similar to that measured between therapsids and amphibians. This also supports the ectothermic status of Dicynodon, Diictodon and Oudenodon.

Anteosaurus, Criocephalosaurus, Struthiocephalus, Glanosuchus and a titanosuchid from the lower Tapinocephalus AZ have also been compared to two co-occuring basal pareiasaurs which are attributed to either Embrithosaurus, Nochelosaurus or Bradysaurus (Lee, 1997) and are considered to have been terrestrial animals (Canoville et al., 2014). From Figure 1B, the δ18Op differences range from −1.4 ± 0.6‰ to 0.7 ± 1.0‰, also supporting the ectothermic status of these therapsids.

From the Middle Permian of China, one anteosaurid Sinophoneus yumenensis from the low palaeolatitude locality of Dashankou has a δ18Op value 4.4 ± 0.3‰ lower than the co-existing bolosaurid parareptile Belebey chengi, which is considered to have been a terrestrial ectotherm (Berman et al., 2000; Müller et al., 2008). This difference between only two values would suggest that Sinophoneus was endothermic, but it is also very close to the expected ranges for ectothermic therapsids (Figure 1B). Considering Sinophoneus as semi-aquatic, as has been suggested for some anteosaurids (Boonstra, 1955, Boonstra, 1962), the δ18Op difference would imply an ectothermic thermophysiology for this therapsid. This hypothesis needs to be tested with a larger number of samples, which are not yet available.

Early to Middle Triassic therapsids

From the Cynognathus AZ (subzone B) of South Africa, differences between the therapsids Kannemeyeria, Cynognathus and Diademodon and the temnospondyl amphibians Xenotosuchus and Microposaurus range from −1.5 ± 1.1‰ to +0.9 ± 1.5‰ (Figure 2A), which fit within the range predicting endothermic therapsids. Interestingly, these therapsids have values ranging from 0.0 ± 1.6‰ to +1.8 ± 1.6‰ higher than the coexisting terrestrial archosauriform Erythrosuchus (Botha-Brink and Smith, 2011), a range suggesting that they shared a similar thermophysiology (Figure 2B). Therefore δ18Op values imply that, as in the case of the therapsids, Erythrosuchus was also endothermic which is consistent with the elevated growth rates implied by its palaeohistology (de Ricqlès et al., 2008; Botha-Brink and Angielczyk, 2010).

Figure 2
Download asset Open asset
δ18Op differences between Early to Middle Triassic therapsids and other tetrapods.

Differences in δ18Op values between therapsids and stereospondyls (white symbols) and between therapsids and archosauriforms (black symbols) from the same localities are plotted against their corresponding palaeolatitude. A theoretical framework based on a lower-than-today thermal gradient (0.4 ± 0.1°C/°Lat; see Appendix 1) and phosphate-water-temperature oxygen isotope fractionation (Lécuyer et al., 2013) predicts various δ18Op value differences. The lighter orange and red areas correspond to the uncertainty of the temperature gradient.

https://doi.org/10.7554/eLife.28589.004

Also from South Africa, five Lystrosaurus specimens from the lower Lystrosaurus AZ have δ18Op values similar to those of the co-existing semi-aquatic stereospondyl Lydekkerina (Schoch, 2008; Canoville and Chinsamy, 2015). In addition, an indeterminate lystrosaurid from the Induan Jiucaiyuan Formation of the Xinjiang Province has a δ18Op value similar (with a difference of −0.1 ± 0.6‰; Figure 2B) to that of the proterosuchid ‘Chasmatosaurus’ yuani, a basal archosauriform considered terrestrial and possessing an intermediate thermometabolsim based on a palaeohistological study (Botha-Brink and Smith, 2011). The combined results from South Africa and China suggest that the analysed lystrosaurids were terrestrial endotherms (Figure 2; see Appendix 1).

From the Ermaying Formation of the Shanxi Province (China), the therapsids Shansiodon wangi and Parakannemeyeria youngi have respectively δ18Op values of 2.0 ± 0.7‰ and 1.7 ± 0.7‰. These are both lower than the sampled erythrosuchid archosauriform Shansisuchus shansisuchus, which fall within two theoretical overlapping ranges (Figure 2B). As for the South African erythrosuchids, if we consider Shansisuchus as a terrestrial endotherm-like animal and the low palaeolatitude of this part of China region, then the two therapsids also fall within the range of endotherms.

Middle to Late Triassic therapsids

The late Anisian cynodont Diademodon and the kannemeyeriiform Kannemeyeria, from the Cynognathus AZ (subzone C), have both lower δ18Op values than those of the contemporary semi-aquatic stereospondyls Paracyclotosaurus and Xenotosuchus with differences ranging from −3.9 ± 2.7‰ to −0.5 ± 0.6‰ (Figure 3A). This pattern fits within the main range predicting endothermic therapsids.

Figure 3
Download asset Open asset
δ18Op differences between Middle to latest Triassic therapsids and other tetrapods.

Differences in δ18Op values between therapsids and stereospondyls (white symbols) and between therapsids and archosauriforms (black symbols) from the same localities are plotted against their corresponding palaeolatitude. A theoretical framework based on a lower-than-today thermal gradient (0.5 ± 0.1°C/°Lat; see Appendix 1) and phosphate-water-temperature oxygen isotope fractionation (Lécuyer et al., 2013) predicts various δ18Op value differences. The lighter orange and red areas correspond to the uncertainty of the temperature gradient.

https://doi.org/10.7554/eLife.28589.005

The Moroccan kannemeyeriiform Moghreberia nmachouensis from the early middle Carnian of the Argana Basin has a mean δ18Op value 2.0 ± 0.5‰ higher than the co-existing aquatic stereospondyl Almasaurus habbazi (Figure 3A), thus implying that Moghreberia nmachouensis was also endothermic.

An indeterminate cynodont from the Rhaetian Lower Elliot Formation of Lesotho has a δ18Op value 2.1 ± 0.3‰, higher than that of an indeterminate basal sauropodomorph. The suspected endothermy and terrestriality of both dinosaurs (Amiot et al., 2006; D'Emic, 2015) and cynodonts are in agreement with their δ18Op difference that falls within the expected range predicting similar thermophysiology between the two (Figure 3B).

Discussion

According to the δ18Op value differences observed between therapsids and co-existing non-therapsid tetrapods, elevated thermometabolism seems to have been acquired by at least two therapsid clades: the unnamed dicynodont clade comprising Lystrosauridae + Kannemeyeriiformes, abbreviated the ‘L+K’ clade, and the Eucynodontia (Figure 4).

Figure 4
Download asset Open asset
Phylogeny of sampled therapsids.

Phylogeny of the sampled therapsids plotted alongside a stratigraphic scale, based on proposed therapsid phylogenies (Ruben and Jones, 2000; Hillenius and Ruben, 2004; Gebauer, 2007; Cisneros et al., 2012; Liu, 2013; Ruta et al., 2013) and their biostratigraphic ranges (Kammerer et al., 2011; Kemp, 2012; Ruta et al., 2013; Huttenlocker, 2014; Day et al., 2015; Viglietti et al., 2016). The thickest parts of the bold lines represent the age range uncertainty of the localities where the samples come from. Species identified as endotherm-like are written in bold and red. Node numbers refer to clades quoted in the text: 1: Neotherapsida; 2: Dicynodontoidea; 3: Lystrosauridae; 4: Kannemeyeriiformes; 5: Epicynodontia; 6: Eucynodontia.

https://doi.org/10.7554/eLife.28589.006

Among the interpreted endothermic therapsids, six belong to the L+K clade (Figure 4): Moghreberia nmachouensis, Parakannemeyeria youngi, Kannemeyeria simocephalus and Shansiodon wangi belong to the Kannemeyeriiformes clade, whereas the Lystrosauridae clade comprises Lystrosaurus and the Chinese indeterminate lystrosaurid. The interpretation of these taxa as endothermic-like animals is better supported for the African Kannemeyeriiformes, M. nmachouensis and K. simocephalus, where more individuals were sampled, than for the Chinese species, S. wangi and P. youngi. Concerning the lystrosaurids, Viglietti et al. (2013) demonstrated aggregating behaviour in the Early Triassic L. declivis and interpreted this as a means to keep warm under extreme climatic conditions. This is in agreement with our endothermic interpretation for the genus.

Based on the above interpretations, Dicynodon seems to have been ectothermic, a fact which would suggests the rise of endothermy amongst the dicynodont L+K clade during the latest Permian (Lopingian). This can be investigated in the future through the isotopic study of a basal dicynodontoid such as Daptocephalus from the Lopingian of South Africa (Kammerer et al., 2011; Viglietti et al., 2016) or Peramodon from the Salarevo Formation of Russia (Sushkin, 1926).

Based on our interpretations, the monophyletic group Eucynodontia (Ruta et al., 2013) (including Cynognathus, Diademodon, an indeterminate cynodont from Lesotho, and extant mammals) possessed endothermic-like metabolism. A parsimonious interpretation would imply rooting the origin of mammal endothermy within non-Eucynodontia Epicynodontia, between the end-Permian and the earliest Triassic. According to our results, the closest sampled relative of Eucynodontia, the late Permian basal cynodont (SAM-PK-K05339), was probably ectothermic. Therefore the origin of mammal endothermy could have taken place among ‘intermediate’ groups belonging to the Epicynodontia (such as Cynosaurus from the latest Permian or Thrinaxodon from the Early Triassic of South Africa) that have in the past been considered to have been endothermic, based on anatomical features (Hillenius and Ruben, 2004; Benoit et al., 2015, 2016a).

In agreement with recent phylogenies (Kemp, 2012; Ruta et al., 2013), endothermic-like body temperature regulations seem to root sometime during the late Permian (Lopingian) independently within the L+K and Epicynodontia clades, the latter being at the origin of mammal endothermy.

An alternative hypothesis considers that both the L+K and Epicynodontia clades possessed a homologous endothermy inherited from their direct common ancestors, the basal Neotherapsida (Figure 4). This suggests that biochemical and physiological mechanisms enabling mammal endothermy, appeared at the base of the neotherapsids during the middle Permian, which is a conclusion recently published based on the paleohistology of some dicynodonts (Olivier et al., 2017). In our case, the absence of an endothermic signal in the sampled Permian therapsids could be due to an endothermy being only seasonal, linked to the presence of a cold season or to the reproduction period (as observed today in some reptile species; Tattersall et al., 2016), which would not be visible in a bulk signal. Therefore, effective acquisition of mammal ‘true endothermy’ was expressed independently within these two lineages, possibly as a result of extrinsic factors.

Global and regional palaeoclimate reconstructions show a cooling trend toward the end-Permian, followed by an abrupt and intense warming at the Permian-Triassic Boundary (Chen et al., 2013; Rey et al., 2016). Interestingly, most of the therapsid clades which survived the end-Permian mass extinction were supposedly endothermic. It thus appears that climatic fluctuations may have acted as selective pressures which favoured or ‘activated’ elevated thermometabolic capabilities within therapsids, at the origins of mammal endothermy. A possible explanation could be the acquisition of a fast growth rate due to the high metabolic rate of the endothermy. According to a recent palaeohistology study (Botha-Brink et al., 2016), Early Triassic therapsids, such as Lystrosaurus or even therocephalians and cynodonts, had a high growth rate allowing them to reach reproductive maturity within a few seasons and compensate their shortened life expectancies. This adaptation might have enabled certain therapsids to survive the intense climatic change of that time and conquer the newly vacant niches.

Concluding remarks

In order to investigate the origin of mammal endothermy amongst the Permo-Triassic therapsids, stable oxygen isotope compositions of apatite phosphate and carbonate from therapsids and associated taxa recovered from several palaeolatitudes were analysed. The following results are highlighted:

  • Assuming that analysed samples have preserved their original isotope composition of phosphate, all the Permian therapsids analysed appear to have ectotherm-like thermoregulation and representatives of two Triassic therapsid clades are considered to have had endotherm-like thermoregulation: the Lystrosauridae + Kannemeyeriiformes and the Eucynodontia.

  • It is proposed that constant elevated thermometabolism appeared independently, at least twice during therapsid evolution. Following the principles of parsimony and phylogenetic systematics, both evolutionary events occured during the late Permian.

  • It seems that the timing of the acquisition of elevated thermometabolism among amniotes coincides with major global climatic and environmental fluctuations at the Permo-Triassic boundary and may had a selective advantage to survive the extinction event and result ultimately in mammalian endothermy.

Material and methods

Sample collection

Request a detailed protocol

Nineteen new fossil apatite samples were analysed to determine stable oxygen isotope compositions of apatite phosphate and carbonate, along with 89 samples for which oxygen isotope compositions have already been published (Rey et al., 2016; Supplementary file 1). This sample total comprises 41 teeth and 65 bones of 90 individual tetrapods (Therapsida, Archosauriformes, Parareptilia and Stereospondyli) recovered from Permian and Triassic deposits of South Africa, Lesotho, Morocco and China. All the sample localities are correlated to the marine biostratigraphic stages using the absolute ages accepted by the International Commission on Stratigraphy (Cohen et al., 2013); updated 12/2016), with the Permo-Triassic and Guadalupian-Lopingian boundaries now respectively considered to be at 251.90 ± 0.02 Ma (Burgess et al., 2014) and 259.1 ± 0.5 (Zhong et al., 2014) Ma.

South African samples comprise Permian and Triassic bones and teeth of therapsids, pareiasaurs, archosauriforms and stereospondyls recovered from 10 localities in the Beaufort Group (Karoo Supergroup), and housed in the collections of the Iziko South African Museum, Cape Town (SAM, Supplementary file 1) and at the Evolutionary Studies Institute, University of the Witwatersrand, Johannesburg (ESI, Supplementary file 1). Permian biozone ages of South African localities were taken from (Rubidge et al., 2013; Day et al., 2015), whereas Triassic age determination has been achieved by biostratigraphic correlation with Laurasian sequences (Hancox et al., 1995; Rubidge, 2005; Abdala and Ribeiro, 2010).

Lesotho samples comprise a cynodont therapsid and a basal sauropodomorph dinosaur from a Triassic locality near the town of Pokane, and are part of the Paul Ellenberger Collection at the Institut des Sciences de l’Evolution, University Montpellier, France (ISEM, Supplementary file 1). The locality belongs to the ‘Red Beds inférieurs a or b’ of the lower Elliot Formation which is currently regarded as latest Triassic (late Rhaetian) (Knoll, 2004).

Moroccan samples comprise therapsid and stereospondyl bones recovered from the ‘Locality 11’ of the Argana Group (Jalil, 1999) near the village of Alma, and housed at the Museum National d’Histoire Naturelle, Paris, France (MNHN, Supplementary file 1). The locality is biostratigraphically correlated to the upper Timezgadiouine Formation, considered to be Middle to early Late Carnian (Jalil, 1999).

Chinese samples are from Permian and Triassic localities situated in Gansu, Shanxi and Xinjiang provinces and comprise therapsids found in association with archosauriforms or parareptiles. These remains are curated at the Institute of Vertebrate Paleontology and Paleoanthropology in Beijing, China (IVPP, Supplementary file 1). The Dashankou locality, from Gansu Province, is biostratigraphically dated as Early Roadian (Liu et al., 2009; Liu, 2010). From Shanxi Province, sampled fossils originate from three localities in the Ermaying Formation which is considered to be Anisian (Liu et al., 2013). From Xinjiang Province, two localities in the Jiucaiyuan Formation have been sampled and are considered Early Triassic (Metcalfe et al., 2009).

Calculation of palaeogeographic coordinates of the sampling sites was performed after careful selection of the magnetic poles of West Gondwana (Muttoni et al., 2001), North China ( et al., 1992), South Jungar (Choulet et al., 2013) and the Alashan terrane (Meng, 1992; Yuan and Yang, 2015). The Apparent Polar Wander Path (APWP) of South Africa (Torsvik et al., 2012) was used to constrain the palaeolatitudinal position of the South African and Lesotho fossil sites. Palaeolatitudes and associated uncertainties (A95) are shown in Supplementary file 1.

Analytical techniques

Request a detailed protocol

To measure the oxygen isotope composition of the apatite phosphate group, the phosphate ions were isolated using acid dissolution and anion-exchange resin applying a standard protocol (Lécuyer, 2004). Silver phosphate was quantitatively precipitated in a thermostatic bath set at a temperature of 70°C. After filtration, washing with double deionized water and drying at 50°C, an aliquot of 300 μg of Ag3PO4 was mixed with 300 μg of nickelised carbon in a silver reaction capsule. Silver phosphate was then reduced into CO to measure its 18O/16O ratio (Lécuyer et al., 2007; Fourel et al., 2011). Each sample was heated at 1450°C by pyrolysis using a VarioPYROcube EA system (Elementar) interfaced to an IsoPrime isotope ratio mass spectrometer working in continuous flow mode at the UMR CNRS 5276 LGLTPE, University Claude Bernard Lyon 1.

Isotopic compositions are quoted in the standard δ notation relative to V-SMOW. Silver phosphate precipitated from standard NBS120c (natural Miocene phosphorite from Florida) was repeatedly analysed (δ18O = 21.71 ± 0.20‰; n = 30) along with the silver phosphate samples derived from the tetrapod remains. For the oxygen isotope analysis of apatite carbonate, about 10 mg of tooth or bone powder was pre-treated (Koch et al., 1997). Powders were washed with a 2% NaOCl solution to remove organic matter, then rinsed five times with double deionized water and air-dried at 40°C for 24 hr. Potential secondary carbonate was removed by adding 0.1 M acetic acid and leaving for 24 hr, after which the powder was again rinsed five times with double deionized water and air-dried at 40°C for 24 hr. The powder/solution ratio was kept constant at 0.04 g mL−1 for both treatments. Stable isotope ratios were determined by using a Thermo Finnigan Gasbench II at the geochemistry laboratory of the Institute of Geology and Geophysics (Chinese Academy of Sciences, China). For each sample, an aliquot of 2 mg of pre-treated apatite was reacted with 5 drops of supersaturated orthophosphoric acid at 72°C for one hour under a He atmosphere before starting 10 measurement cycles of the isotopic composition of the CO2 produced with a Finnigan MAT 253 continuous flow isotope ratio mass spectrometer. The measured oxygen isotopic compositions were normalized relative to the NBS-19 calcite standard and have a reproducibility index better than ±0.2‰. Isotopic compositions are quoted in the standard δ notation relative to V-SMOW.

Robustness of the stable isotope record

Request a detailed protocol

Analysed materials consist of bone or tooth dentine, which is more porous than enamel with small and less densely inter-grown apatite crystals (Mills, 1967). Thus, their original stable isotope compositions are more prone to diagenetic alteration that may have taken place through precipitation of secondary minerals within and at the surface of bioapatite crystals, adsorption of ions on the surface of apatite crystals, or dissolution and recrystallization with isotopic exchange. The samples from South Africa have been previously tested for primary preservation through comparison between their δ18Op values, δ18Oc values and carbonate content on the basis of the following considerations: (1) the carbonate content in apatite of modern vertebrates typically ranges from less than 1% up to 13.4%. Thus, samples that have a carbonate content exceeding 13.4 wt% likely contain additional inorganic carbonate precipitated from diagenetic fluids, and would result in potentially biased δ18Oc values of apatite carbonate (Figure 5); (2) In modern vertebrates, the oxygen isotope composition of apatite carbonate is higher than that of co-occurring apatite phosphate (7–9 ‰ in mammals), and up to 14.7‰ in sharks (Vennemann et al., 2001). Experimental ( et al., 1967) and empirical studies (Zazzo et al., 2004b) have shown that microbially-mediated diagenetic alteration of apatite phosphate results in a greater difference between δ18Oc and δ18Op values. Therefore, fossil samples exhibiting δ18Oc-δ18Op differences larger than 14.7‰ are most likely altered and can be disregarded (Figure 5). Inorganic alteration at low temperature has little effect on the δ18Op values of phosphates, even at geological time scales (Lecuyer et al., 1999), so samples affected by inorganic diagenetic alteration of carbonates, (resulting either in a high overall carbonate content or anomalous δ18Oc-δ18Op differences), may still preserve the original oxygen isotope composition of their phosphate (Figure 5). Using these two assessments, newly measured δ18Op values are considered to have preserved their original isotopic signatures and can be interpreted in terms of ecologies and physiologies.

Figure 5
Download asset Open asset
Isotopic preservation assessment.

δ18Oc-δ18Op differences between teeth and bones plotted against the structural carbonate content (wt%) of apatite. Samples that have δ18Oc-δ18Op differences higher than 14.7‰ are considered doubtful as regards potential diagenetic alteration (see text). For carbonate contents (wt%) higher than 13.4%, the δ18Oc values are considered to be inherited from inorganic diagenetic processes. A high difference between δ18Oc and δ18Op is interpreted as the result of a microbially-mediated alteration of the apatite phosphate or too high δ18Oc values resulting from the addition of inorganic carbonate or isotopic exchange with an external source of inorganic carbon. The grey crosses refer to previously published South African bone and tooth samples (Rey et al., 2016).

https://doi.org/10.7554/eLife.28589.007

Assessment of therapsid thermophysiology

Request a detailed protocol

For all localities, average δ18Op values were calculated for each tetrapod species. Differences in δ18Op values between therapsid species and co-occurring non-therapsid tetrapods (amphibians, parareptiles or archosauriforms) were calculated and plotted against their corresponding palaeolatitude for three time intervals: the middle to late Permian (Figure 2), the Early to Middle Triassic (Figure 3) and the Middle Triassic to latest Triassic (Figure 4). These differences were compared to the following four theoretical areas of values represented as coloured areas in Figures 1–3. To construct those theoretical areas, both the phosphate-water temperature scale from Lécuyer et al., 2013 and the differences of stable oxygen compositions between mammals of various ecologies from Cerling et al. (2008) have been used. (see Appendix 1 for their construction details).

Orange and green areas in Figures 1A, 2A and 3A represent expected δ18Op value differences between terrestrial therapsids and semi-aquatic stereospondyls (white symbols) or parareptiles (black symbols); red and blue areas in Figures 1B, 2B and 3B represent expected δ18Op value differences between terrestrial therapsids and terrestrial Permian parareptiles or Triassic archosauriforms (black symbols). Oblique orange and red areas in Figures 1–3 represent expected δ18Op value differences between an endotherm and an ectotherm. Vertical green and blue areas in Figures 1–3 represent expected δ18Op value differences between animals having similar thermophysiology.

Appendix 1

Theoretical curves construction

In order to infer therapsid thermophysiologies based on the δ18Op differences between therapsids and the associated fauna, the following theoretical framework is proposed and represented in Figures 1, 2 and 3 as colored areas. These areas are based on the two main factors influencing animal δ18Op values: their thermometabolism (here simplified as ectotherm and endotherm) and their lifestyle (here terrestrial and semi-aquatic).

The δ18Op value ranges are estimated based on the phosphate-water temperature scale (Lécuyer et al., 2013):

(1) Tb=117.4−4.5(δ18Ob−δ18Obw)

Where Tb corresponds to body temperature, δ18Op correspond to the oxygen isotope composition of apatite phosphate, and δ18Obw correspond to the oxygen isotope composition of body fluids. For endothermic vertebrates, Tb is assumed to be at 37°C, the average Tb of most extant placentals and possibly of the common ancestor of all extant mammals (Watson and Graves, 2013). For ectothermic vertebrates, Tb is assumed here to reflect immediate environmental temperature (Te). According to Equation 1, the δ18Op difference between an endotherm and an ectotherm can be expressed as:

δ18Op−endotherm−δ18Op−ectotherm=(Te−37)/4.5+(δ18Obw−endotherm−δ18Obw−ectotherm)

Because vertebrate δ18Obw value depend on their ecology that affects the input-output balance of body water (Luz et al., 1984; Bryant and Froelich, 1995; Kohn, 1996a), the difference δ18Obw−endotherm−δ18Obw−ectotherm will mainly reflects that of their lifestyle (terrestrial, semi aquatic or aquatic), as well as their dependency on the surface water they ingest. Cerling et al., 2008, the δ18Op difference between water-dependent and water-independent terrestrial mammals can be up to 4‰. This is illustrated by the ranges 1 and 3 in Appendix 1—Figure 1 and by the red and blue ranges in the main text Figures 1A, 2A and 3A).

Appendix 1—figure 1
Download asset Open asset
Expected latitudinal variation of the δ18Op difference between vertebrate taxa of various physiologies and ecologies.

Based on modern relationships between climate and phosphate-water-temperature oxygen isotope fractionation (Lécuyer et al., 2013) the following δ18Op values differences are predicted: (1) corresponds to a terrestrial endotherm compared to a terrestrial ectotherm; (2) corresponds to a terrestrial endotherm compared to a semi-aquatic ectotherm; The vertical range (3) corresponds to terrestrial animals having similar thermometabolism; (4) corresponds to the difference between a terrestrial and a semi-aquatic animals having a similar thermometabolism.

https://doi.org/10.7554/eLife.28589.009

This variation in δ18Op values can be overprinted by the animal lifestyle, a semi aquatic animal having δ18Op value 3‰ lower than that of co-existing water-dependent terrestrial species, and 7‰ lower than that of co-existing water-independent terrestrial one (Appendix 1—figure 1, ranges 2 and 4, and main text Figures 1B, 2B and 3B, ranges orange and green).

The environmental temperature (Te) used to approximate ectotherm body temperature (Tb) is estimated based on the present-day relationship between mean air temperature and latitude. Because this latitude-temperature relationship is not valid for low latitudes below about 10° (corresponding to the thermal equators of present-day Earth), we assume a constant temperature between 0° and 10° of latitude.

As a simplification, three periods are considered:

  1. The Middle to Late Permian with equatorial sea surface temperatures close to the modern ones (Chen et al., 2013) (~25°C) for which we assume a present-day thermal gradient of 0.6 °C/°Latitude (Amiot et al., 2004; Williams et al., 2007) (Main text Figure 1).

  2. The Early to Middle Triassic having globally warmer temperatures and a flatter temperature gradient. Based on (Trotter et al., 2015), mean equatorial sea surface temperatures were about 10°C higher than Late Permian ones (Trotter et al., 2015) (~35°C). For this time period, we assume a flatter thermal gradient that we arbitrarily set at 0.4 °C/°Latitude (Main text Figure 2).

  3. The Middle Triassic to Early Jurassic having globally intermediates temperatures (Trotter et al., 2015) that we set to 30°C with an intermediate gradient of 0.5 °C/°Latitude (Main text Figure 3).

In order to take into account the possible variations of the thermal gradient within the selected periods, an interval of ±0.1 °C/°L is added for each Main text figures.

Consequently, latitudinal thermal gradients will lead to δ18Op-endotherm - δ18Op-ectotherm differences varying along with latitude (Appendix 1—figure 1, ranges 1 and 2). This simplified framework is used in this study to predict the following scenarios based on δ18Op differences between therapsids and co-existing other tetrapods (Appendix 1—figure 1): Endothermic and terrestrial therapsid vs ectothermic and terrestrial tetrapod (range 1); Endothermic and terrestrial therapsid vs ectothermic and semi-aquatic tetrapod (range 2); Therapsids and other tetrapods having similar thermophysiologies and lifestyle (range 3); Terrestrial therapsid and semi-aquatic tetrapods having similar thermophysiologies (range 4).

Ecology of sampled tetrapods and its impact on stable oxygen isotope compositions

Prior to interpreting differences between therapsid and associated non-therapsid tetrapod δ18Op values in terms of differences in thermophysiologies, ecological traits must be considered as they also affect the oxygen isotope compositions of apatite phosphate. Indeed, δ18Op values recorded in phosphatic tissues depend on the animal body temperature, as well as on the oxygen isotope composition of body water, the latter being affected by water turnover rate and isotopic fractionations associated with water loss (Bryant and Froelich, 1995). Indeed, water loss through transcutaneous evaporation, sweat and exhaled water vapour tends to 18O-enrich the remaining body water (Kohn, 1996a). This isotopic enrichment is amplified or mitigated depending on the rate of water turnover, which depends itself on the animal ecology. Aquatic and semi-aquatic animals regularly ingest and release large amounts of their environmental water compared to terrestrial ones, which in turn reduces the magnitude of body water 18O-enrichment relative to that of environmental water (Kohn, 1996a).

Stereospondyl amphibians

The sampled stereospondyl amphibian clade (Almasaurus habbazi, Lydekkerina, Microposaurus, Paracyclotosaurus, Rhinesuchus and Xenotosuchus) includes semi-aquatic to aquatic animals (Schoch, 2008). A study dedicated to Rhinesuchus palaeohistology concluded that it had a fully aquatic lifestyle (McHugh, 2014), whereas Lydekkerina, a basal stereospondyl, was amphibious with a tendency to be terrestrial (Canoville and Chinsamy, 2015).

Pareiasaurid and bolosaurid parareptiles

The sampled parareptiles include the Chinese Bolosauridae, Belebey chengi, and the South African Pareiasauria, Pareiasaurus and basal pareiasaurs (Embrithosaurus, Nochelosaurus or Bradysaurus; (Lee, 1997). The Chinese taxon is considered to have been terrestrial (Berman et al., 2000; Müller et al., 2008), but interpretations of the life habits and habitats of the South Africa pareiasaurs still lack a consensus. According to authors, they are variously considered fully aquatic (Ivakhnenko, 2001), semi-aquatic (Kriloff et al., 2008) or fully terrestrial (Voigt et al., 2010; Canoville et al., 2014). The low δ18Op values measured in both Pareiasaurus and Rhinesuchus are similar to each other, Pareiasaurus having slightly higher δ18Op values than Rhinesuchus (~1‰), but significantly lower than those measured in therapsids (about 4‰ to 8‰). Considering Rhinesuchus as an aquatic ectothermic amphibian (McHugh, 2014), it is suggested that the ectothermic Pareiasaurus was also aquatic or semi-aquatic as previously suggested (Ivakhnenko, 2001; Kriloff et al., 2008). In contrast to the semi-aquatic Pareiasaurus from the Lower Daptocephalus AZ, the Tapinocephalus AZ pareiasaurs may have been terrestrial as recently proposed in a study on pareiasaur ecology and based on oxygen and carbon isotope compositions of apatite carbonate(Canoville et al., 2014). Moreover, those authors found no significant differences between therocephalians and pareiasaurs, agreeing with our dataset, and a ~ 3‰ overlap between pareiasaur and dinocephalian δ18Oc values, even though most dinocephalian values are several per mil lower than those of pareiasaurs.

Archosauriforms

Archosauriformes are represented by the Erythrosuchidae Erythrosuchus and Shansisuchus shansisuchus, by the Proterosuchidae ‘Chasmatosaurus’ yuani, as well as by an indeterminate basal sauropodomorph dinosaur. Proterosuchid lifestyle is still unresolved, but Botha-Brink and Smith, 2011 favoured a terrestrial rather than an aquatic lifestyle, whereas the erythrosuchids are considered as the largest terrestrial predators of their time (Botha-Brink and Smith, 2011).

Lystrosaurid therapsids

Therapsids are generally considered as terrestrial dwellers (Kemp, 2012) with a few uncertainties such as Anteosaurus considered as riparian (Boonstra, 1955, Boonstra, 1962) and Lystrosaurus considered by some as semi-aquatic (Germain and Laurin, 2005; Ray et al., 2005; Botha-Brink and Angielczyk, 2010; Canoville and Laurin, 2010).

Because the Lystrosaurus-Lydekkerina δ18Op difference falls within the same range as the Kannemeyeria-Xenotosuchus and Kannemeyeria-Microposaurus values from the Cynognathus subzone B assemblage, Lystrosaurus could be considered as having the same thermophysiology and lifestyle as Kannemeyeria, i.e. a terrestrial endotherm (Figure 2A). Because Lystrosaurus may have either been terrestrial (Botha and Smith, 2006) or semi-aquatic (Canoville and Laurin, 2010), it cannot be totally excluded that Lystrosaurus shared a similar lifestyle and thermometabolism with the amphibian Lydekkerina. If Lystrosaurus was indeed semi aquatic, then it could also be predicted to have been an ectotherm.

Given that the thermophysiology of the terrestrial archosauriform ‘Chasmatosaurus’ remains unclear, the two similar δ18Op values observed in the lower Triassic Jiucaiyuan Fm. of Xinjiang province of China may indicate either that both the lystrosaurid and the proterosuchid shared similar thermometabolism and lifestyle (terrestrial endotherms or terrestrial ectotherms; Figure 2), or that one of them may have had a lower body temperature. Indeed, compared to low latitudes where ectotherms have slightly lower δ18Op values than co-existing endotherms due to differences in metabolic activity and body temperatures, mid latitude ectotherms have even lower body temperatures (reflecting the environmental thermal gradient) resulting in higher δ18Op values that mimic those of co-existing endotherms (Amiot et al., 2004). It is worth noting that based on histological features, Proterosuchus, had intermediate growth rates, suggesting an intermediate thermometabolism (Botha-Brink and Smith, 2011) that could apply to the Chinese ‘Chasmatosaurus’. Considering a terrestrial lifestyle for the Chinese lystrosaurids (see above), and the conflicting hypotheses of Lystrosaurus from South Africa (terrestrial endotherm vs. semi-aquatic ectotherm), the hypothesis for terrestrial endothermy for the lystrosaurids is favoured.

References

  1. Book
    1. Abdala F
    2. Jashashvili T
    3. Rubidge BS
    4. Heever J
    (2014)
    New Material of Microgomphodon oligocynus (Eutherapsida, Therocephalia) and the Taxonomy of Southern African Bauriidae
    In: Kammerer C. F, Angielczyk K. D, Fröbisch J, editors. Early Evolutionary History of the Synapsida, Vertebrate Paleobiology and Paleoanthropology. Netherlands: Springer. pp. 209–231.
  6. Book
    1. Bennett AF
    2. Ruben JA
    (1986)
    The Metabolic and Thermoregulatory Status of Therapsids. the Ecology and Biology of Mammal-Like Reptiles
    Washington, DC: Smithsonian Institution Press.
    1. Boonstra LD
    (1955)
    The girdles and limbs of the South African Deinocephalia
    Annals of the South African Museum 42:185–326.
    1. Boonstra LD
    (1962)
    The dentition of the titanosuchian dinocephalians
    Annals of the South African Museum 46:57–112.
    1. Botha J
    2. Chinsamy A
    (2004)
    Growth and life habits of the triassic cynodont Trirachodon, inferred from bone histology
    Acta Palaeontologica Polonica 49:619–627.
    1. Botha J
    (2003)
    Biological aspects of the Permian dicynodont Oudenodon (Therapsida: dicynodontia) deduced from bone histology and cross-sectional geometry
    Palaeontologia Africana 39:37–44.
    1. Cohen KM
    2. Finney SC
    3. Gibbard PL
    4. Fan J-X
    (2013)
    The ICS International Chronostratigraphic Chart
    Episodes 36:199–204.
    1. D’Angela D
    2. Longinelli A
    (1990)
    Oxygen isotopes in living mammal’s bone phosphate: Further results
    Chemical Geology: Isotope Geoscience Section 86:75–82.
  39. Book
    1. de Ricqlès A
    (1972)
    Recherches Paléohistologiques Sur Les Os Longs Des Tétrapodes: Titanosuchiens, Dinocéphales Et Dicynodontes
    Masson.
    1. de Ricqlès A
    (1979)
    Quelques remarques sur l’histoire évolutive des tissus squelettiques chez les Vertébrés et plus particulièrement chez les Tétrapodes
    Année Biologique 18:1–35.
    1. Farmer CG
    2. Carrier DR
    (2000)
    Ventilation and gas exchange during treadmill locomotion in the american Alligator (Alligator mississippiensis)
    The Journal of Experimental Biology 203:1671–1678.
    1. Gebauer EV
    (2007)
    Phylogeny and evolution of the Gorgonopsia with a Special Reference to the Skull and Skeleton of GPIT/RE/7113 ('Aelurognathus?’parringtoni)
    Universität Tübingen.
    1. Hancox PJ
    2. Shishkin MA
    3. Rubidge BS
    4. Kitching JW
    (1995)
    A threefold subdivision of the Cynognathus Assemblage Zone (Beaufort Group, South Africa) and its palaeogeographical implications
    South African Journal of Science 91:143–144.
    1. Ivakhnenko MF
    (2001)
    Tetrapods from the east european placket—Late Paleozoic natural territorial complex
    Tr. Paleontol. Inst. Ross. Akad. Nauk 283:1–200.
    1. Katz SL
    (2002)
    Design of heterothermic muscle in fish
    The Journal of Experimental Biology 205:2251–2266.
  65. Book
    1. Kemp TS
    (2012)
    The origin and radiation of Therapsids
    In: Chinsamy-Turan A, editors. Forerunners of Mammals: Radiation, Histology, Biology. Bloomington: Indiana University Press. pp. 3–28.
  67. Book
    1. King GM
    (1988)
    Anomodontia
    Stuttgart: Gustav Fischer Verlag GmbH & Company KG.
    1. Lee MSY
    (1997)
    A taxonomic revision of pareiasaurian reptiles: implications for Permian terrestrial palaeoecology
    Modern Geology 21:231–298.
    1. Lécuyer C
    (2004)
    Oxygen isotope analysis of phosphate
    Handbook of Stable Isotope Analytical Techniques 1:482–499.
    1. Liu J
    2. Li L
    3. Li XW
    (2013)
    SHRIMP U-Pb zircon dating of the Triassic Ermaying and Tongchuan formations in Shanxi, China and its stratigraphic implications
    Vert PalAs 51:162–168.
    1. Liu J
    (2010)
    Dashankou Fauna: a unique window on the early evolution of therapsids
    Bulletin of the Chinese Academy of Sciences 24:83–85.
    1. Meng ZF
    (1992)
    Paleomagnetic results from the Late Permian redbeds along the Hexi Corridor, China
    Chinese Science Bulletin = Kexue Tongbao pp. 1452–1456.
  88. Book
    1. Mills A
    (1967)
    Structural and Chemical Organization of Teeth
    Elsevier.
    1. Owerkowicz T
    2. Musinsky C
    3. Middleton KM
    4. Crompton AW
    (2015)
    Great Transformations in Vertebrate Evolution
    143–165, Respiratory turbinates and the evolution of endothermy in mammals and birds, Great Transformations in Vertebrate Evolution, Chicago, University of Chicago Press.
  98. Book
    1. Ruben J
    2. Hillenius WJ
    3. Kemp TS
    4. Quick DE
    5. The evolution of mammalian endothermy
    (2012)
    Forerunners of Mammals: Radiation, Histology, Biology
    Chinsamy-Turan A, editors. Bloomington: Indiana University Press.
    1. Schoch RR
    (2008)
    The Capitosauria (Amphibia): characters, phylogeny, and stratigraphy
    Palaeodiversity 1:189–226.
    1. Sushkin PP
    (1926)
    Notes on the pre-Jurassic tetrapoda from Russia
    Palaeontologia Hungarica 1:323–344.
  119. Book
    1. Williams M
    2. Haywood AM
    3. Gregory FJ
    4. Schmidt DN
    (2007)
    Deep-Time Perspectives on Climate Change: Marrying the Signal From Computer Models and Biological Proxies
    Geological Society of London.
    1. Zhong Y-T
    2. He B
    3. Mundil R
    4. Xu Y-G
    (2014)
    CA-TIMS zircon U–Pb dating of felsic ignimbrite from the Binchuan section: Implications for the termination age of Emeishan large igneous province
    Lithos, Special Issue Permian Large Igneous Provinces: Characteristics, Mineralization and Paleo-Environment Effects 204:14–19.

Decision letter

  1. Diethard Tautz
    Reviewing Editor; Max-Planck Institute for Evolutionary Biology, Germany

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triassic therapsid clades" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

You will see from the reviews that they were generally very positive about the question and the approach. However, there is still a considerable list of uncertainties that are listed in the reviews, especially in review #3. Resolving these would likely require more than the three-month that we allow for revisions. But we will happy to look at a new submission, in case you address all relevant points at a later time.

Reviewer #1:

This is a potentially important paper, presenting a new geochemical means to identify thermoreguylatory principles in early reptiles, and identifying evidence that endothermy may have arisen twice in the synapsids.

However, it is not ready for review in a journal yet as it reads like a thesis chapter. First, the different sections of the paper must be discriminated – 'Methods' are mixed through the Materials and methods and Results sections, and must be separated. By reorganizing the text and focusing on a single (rather than multiple) Materials and methods sections, the paper can become simpler, shorter, and easier to read. Make sure the Results section reports simply the results.

I'm not sure that eLife format has been strictly followed either – the authors should check section numbering and styles are correctly followed.

In the figures, try to minimize abbreviations (e.g. ‘Terr. Vs. semiaq. Similar thermophy.'; this is not necessary. Use pure abbreviations (e.g. AZ) or spell out in full.

Results subsection “Robustness of the stable isotope record”: This reads pretty much like 'Materials and methods', so move it back to that section.

Results section second paragraph: This paragraph is a curious mixture of Methods and Figure caption. Return 'methods' stuff to the Materials and methods section – 'were calculated' 'were compared' – all methods. Then the description of symbols and shading in Figures 2–4 should be placed in the figure captions. Then, under Results, we need a plain writing description of what the plots show.

I think 'Results' begins at subsection “Therapsid thermophysiology” – so move text, combine sensibly, and re-label sections.

No, not really: the first paragraph of subsection “Therapsid thermophysiology” is more 'Methods' – telling us which taxa were studied and from where. Please apply these general principles of scientific paper writing properly and reorganize and rationalize the paper.

Reviewer #2:

General comments: This paper uses comparisons of stable oxygen isotope ratios of Permian-Jurassic non-mammalian synapsids to their contemporaries to try to infer when an endothermic metabolism arose in the lineage leading to mammals. The results are thought provoking, but I think a number of changes and additions will need to be made before the paper is ready to accept. I've made a number of comments and suggestions below, but a few general issues that must be addressed:

1) I am not an isotope geochemist, and I think it will be important for one to examine the paper. I'm not so worried about the analytic methods, which seem fairly standard, but I think it would be good to get a more informed perspective on the method the authors use to predict the expected magnitude of differences between ectotherms and endotherms. This is the foundation of everything else in the paper, so it is important to make sure that this is acceptable.

2) Assuming the method is acceptable, I strongly recommend that the authors conduct a sensitivity analysis using a range of plausible average temperatures and latitudinal gradients for each of their time periods of interest. Their results will be stronger if they are consistent across the plausible range of values, and it will remove any appearance of cherry-picking temperatures/gradients to get the results that they want.

3) The authors seem to play pretty fast and loose with the terms "endothermy" and "ecothermy." A particular manifestation of this is the frequency with which they use bone histology (particularly the presence of fibrolamellar bone) as a correlate of endothermy. Although this was more common in earlier work on paleohistology, more recently the field has been much more circumspect about this relationship, with much more attention placed (rightly, in my opinion) on the implications of histology for growth rates/patterns. Elevated growth rates often are correlated with "endothermy" but this is not universally true. I think the authors need to be more thoughtful in their presentation of this relationship, and they also should provide more explicit, detailed definitions of "endothermy" and "ecothermy" at the start of the paper.

4) I recommended some changes to the discussion that would explore the link between endothermy and survivorship during the Permo-Triassic Mass Extinction (PTME) that would increase the broad interest and impact of the paper. Alternatively, the authors could provide a more detailed discussion of their results in the context of previous work and current ideas on the evolution of endothermy. Either way, much of what's in the current disucssion isn't very interesting (and I say this as someone who loves the intricacies of dicynodont taxonomy) and it undersells the paper.

5) There are a couple places in the main text and supplement that could be strengthened in terms of the references that are cited.

6) The writing style of the paper is good, but it could be improved upon. I recommend having Rubidge, Smith, and Viglietti go over the final version of the paper to clean things up and make sure the idiom is appropriate.

Abstract: mammals didn't exist in the Permian, so you should say something along the lines of "endothermy evolved in the Permian ancestors of mammals."

Abstract: Cynognathia is a subclade of Eucynodontia, so I'm not sure what you mean here.

Abstract: change to "originated in Epicynodontia during the…". Also, change to middle-late Permian.

Introduction opening sentence: this sentence is quite vague and is kind of meaningless as a result. Please be more specific about particular ways in which environmental changes affect vertebrates if you want to use this kind of example.

Introduction, first paragraph: as written, this paragraph provides something of a definition of endothermy. However, because endothermy is kind of a loaded term, and has been used in different ways by different authors, I suggest that you explicitly define what you mean by it so as to avoid any confusion.

Introduction paragraph two: again, defining endothermy will help to avoid confusion about what you mean by "true endotherms."

In the same paragraph: I think you're kind of downplaying the amount of work on this subject. For example, McNab, 1978, Hillenius, 1992; 1994, Ruben, Hillenius, Kemp and Quick, 2012, Owerkowicz, Musinsky, Middleton and Crompton, 2015, all would be appropriate to cite here.

Introduction paragraph four: bony secondary palate would be a better term to use.

In the same paragraph: Dicynodonts also have a complete secondary palate, although it is formed primarily by the maxilla.

Introduction paragraph five: Owerkowicz, Musinsky, Middleton and Crompton, 2015, would be especially good to cite in the context of whether turbinates are necessary in the context of mammalian endothermy. Also Laaß (2011) presents some evidence of turbinals in Lystrosaurus, although I'm kind of skeptical as to whether that's actually what's visible in the scan.

Reports of fibrolamellar bone in synapsids go back farther than these studies. For example, de Ricqles reports fibroilamellar bone in some therapsids is in papers on bone histology from the 1970's.

Fibrolamellar bone is a lot more ubiquitous among therapsids than what you suggest here; if anything it's more common than not. There's even pelycosaur-grade synapsids like Sphenacodon and Ophiacodon that produce fibrolamellar bone for at least part of their ontogenies.

In addition to the fact that some non-ectothermic teterapods can produce fibrolamellar bone, and important point to make here is that the presence of fibrolamellar bone really is more a reflection of the rate at which bone tissue is being formed than endothermy in a strict sense. While the high growth rates required to produce fibrolamellar bone often are correlated with high metabolic rates, this is not always the case.

Introduction paragraph six: although the degree of resistance to diagenesis varies among tissues, with enamel being the more resistant by far.

Materials and methods section, subsection “Sample collection”: you should provide justification for why you're using these alternate dates. Also, looking at your supplementary table, it seems like some of your dates imply more precision than is actually the case. For example, as far as I know there aren't any published radiometric dates for the Burgersdorp Fm. If you're making the correlation to marine stages (and their dates) using tetrapod biostratigraphy, I would suggest caution in this case, given the recent dates from South America that have been published by Ottone et al., 2014 and Marsicano et al., 2016. In any case more information on the source of your dates should be provided.

Subsection “Robustness of the stable isotope record”: “contain” instead of “contains”.

Figure 1: the colouration of the different parts of the plot doesn't seem to be explained. I understand that the dark orange area at the top right is the area that represents diagenesis, but do the other coloured regions have some meaning?

Subsection “Robustness of the stable isotope record”: does partial preservation of the isotopic signal require different interpretation than pristine preservation?

In the same subsection: Figure 2A is repeated here. Do you mean 2A, 3A, and 4A?

Would it be better to refer to the colours instead of light and dark gray? Since the journal is primarily an online publication the colour online/grayscale printed version seems like it isn't really a concern, and referring to the colours would help readers understand the figures.

Permian Therapsids general comment: I recommend reorganizing this section a little. I think it would be most logical to present the results subdivided by time (i.e., all middle Permian things first, then early late Permian, then latest Permian). Within each time bin, you can subdivide further by geographic area.

Early-Middle Triassic Therapsids general comment: as with the Permian taxa, I think it would be most useful to subdivide the results first by time and then by geography.

Subsection “Early to Middle Triassic therapsids”: because bone histology is really more a record of growth rates, it might be better to say something along the lines of "…which is consistent with the elevated growth rates implied by the bone histology of Erythrosuchus."

I think the paragraph describing the results on the Chinese Lystrosaurid should be combined with the previous paragraph on the South African specimens. Also, Lystrosaurus is the only Lystrosaurid known from China, so I would say its safe to assume that it is Lystrosaurus.

In the same subsection and in Figure 3 you make it seem like it is pretty certain that Lystrosaurus is an endotherm. However, in the supplement, you make it seem like there's actually a fair amount of uncertainty based on questions about the ecology of Lystrosaurus and the metaboloic status 'Chasmatosaurus' (which is inferred from information on growth rates inferred from bone histological data from related taxa in South Africa). I think you need to better portray this uncertainty here, and also be sure to take it into account in the Discussion section of the paper.

Subsection “Middle to Late Triassic therapsids”: contemporary instead of co-existing

Discussion section and elsewhere: although wordier, something like "the unnamed clade comprising Lystrosauridae + Kannemeyeriiformes" would make more sense.

Discussion paragraph three: "basal dicynodontoid" would work better.

Discussion paragraph five: I don't really agree that this is a valid hypothesis. You have a fair number of Permian therapsids from different clades that are reconstructed as being ecothermic, and it seems implausible that you would have so many reversals to ecothermy. In the previous paragraph, you discuss a hypothesis based on a parsimony optimization of the character on the phylogeny. That is justified: you have a clear-cut optimality criterion and you are discussing the best hypothesis under that criterion. I don't think there's any method/optimality criterion that would support the number of reversals needed to account for all so many of your Permian therapsids being reconstructed as endotherms, so the argument is much less sound. If you're doing that, then you can pretty much make up whatever hypothesis you want because there's no constraint on what could be considered plausible.

In the same paragraph: biochemical instead of biochemicals

Discussion paragraph seven: assuming that Lystrosaurus was indeed an endotherm (see comments above and below), these observations are especially interesting in the context of the changes in growth and life history at the PTB noted by Botha-Brink et al., 2016, among others. This would be something interesting to explore further in the discussion, as well as a more general consideration of the role endothermy played in determining survivorship during the PTME. There's a couple obvious ways to get the space needed to do this. One would be to get rid of the previous paragraph, which I don't think is needed. Likewise, you could get rid of everything from paragraph two, and almost everything in paragraph three.

Concluding remarks: this is a really stilted way to say that most of your samples preserved an original signal that had not been diagenetically altered. I think you can get rid of most of this (see comments above).

General comment on the supplement: you should reorganize so that the first section is the one describing how you came up with your theoretical ranges of isotope values for taxa with different ecologies and thermophysiologies. That's the foundation for all of the interpretation of the fossil data you present in the paper and the supplement, so it deserves to go first. If space (or editorial largesse) allows, I would even suggest moving the information on how you calculated your expected ranges into the main text.

Appendix paragraph two: change heading to "Stereospondyl amphibians." Also make this change in the first line of this paragraph. Also, does the fact it was amphibious complicate your comparisons between it and Lystrosaurus?

Appendix subsection “Pareiasaurid and bolosaurid parareptiles”: would the fact that Pareisaurus was an herbivore whereas Rhinesuchus was a faunivore make any difference here? I suppose the results for the older pareiasaurus would suggest not.

Appendix subsection “Lystrosaurid therapsids”: going along with my comments above about the identity of the Chinese material, I recommend changing the title of this section to "Lystrosaurus."

In the same section: there is a much deeper history of debate in the literature about Lystrosaurus potentially being semi-aquatic that would be worth citing here: Broom 1903b, 1932; Watson 1912, 1913; Williston 1914; Brink 1951; Camp 1956; Cluver 1971; Kemp 1982; Hotton 1986; King 1991; King and Cluver 1991; Germain & Laurin 2005; Ray et al. 2005; Botha-Brink and Angielczyk 2010

Your interpretation of Lystrosaurus as an endotherm seems a lot more uncertain here than you portray it in the text. This uncertainty has important implications for your the evolutionary scenarios you present in the Discussion section and considerations of the potential relationships between endothermy and the PTME. Therefore, I think you need to present this uncertainty more clearly in the main text (there you make it seem like Lystrosaurus is definitely an endotherm, which seems kind of misleading in the context of what's in the supplement).

Appendix subsection “Theroretical curves construction”: again, there doesn't seem to be any gray areas in the figures in the main text, so I would refer to the colours you use in those figures.

It would be good to have a citation for 37C being the average body temperature of terrestrial mammals.

In the same subsection: refer to colours for the main text figures.

General comment on the three time periods: estimates of global temperature and latitudinal thermal gradients obviously are not exact (especially when you're arbitrarily picking values for the latter in at least some cases). Because of this, it would be good to have more citations for where you're getting your values from, and justification for why you chose particular values/references. Likewise, it seems like it would be good to do a sensitivity analysis using a range of different (plausible) average temperatures and thermal gradients. Your results will seem a lot stronger if you consistently reconstruct your proposed endothermic taxa as endotherms across the range of plausible values than would be the case if you you have them come out as endotherms for certain combinations of values. Doing this will also avert criticisms that you are cherry-picking values to get the results you want.

Reviewer #3:

I have completed my review of Rey et al.'s manuscript, "Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triasic therapsid clades'. This is a somewhat challenging paper to evaluate: The manuscript is exceptionally well written – clear, concise, scholarly, rationally organized; really a pleasure to read. And I see no reason to question the significance of the samples studied or the quality of the isotopic measurements. So, on these bases alone I would say the paper should be published with modest revisions. However, I do not believe several of the assumptions that underlie the interpretation of the isotopic data set are well justified, and therefore I don't consider the study to offer a strong new constraint on the question of therapsid physiology. On the third hand (!?!), the authors have looked at one of the few geochemical properties that potentially constrain body temperatures of these organisms, and I'm a strong believer that it is always worthwhile to publish possibly significant observations. So, I would like to see this paper published but encourage the authors to do some soul searching about some of their interpretations.

The two most important issues, in my mind, are the state of preservation of the samples, and the robustness of the models that are used as a reference frame for interpreting their findings.

First, there is evidence indicating that geochemical properties of ancient bone or dentin do not consistently constrain body temperature, except in rare cases of extraordinary preservation. Both are composed of fine-grained phosphate intermixed with organics, and demonstrably undergo diagenetic changes in oxygen isotope composition on geologically short time scales. It is asserted that phosphate resists such changes, but there is plenty of data showing this rule of thumb is often violated (see Stolper et al., 2016 for an example).

The authors have tried to avoid this by filtering the database for signs of alteration, but they do so with relatively simple, indirect metrics (percent carbonate and the phosphate-carbonate fractionation). I would have found it more compelling if they could have shown through microscopy and/or minor element mapping that they were studying well-preserved fabrics. Or if they had focused on only tooth enamel, which is generally much more resistant to alteration.

Second, the models of the oxygen isotope compositions of endotherm and ectotherm d18O values are based on two variables that strike me as poorly known or known to be false:

1) The assignments of taxa to terrestrial vs. semi-aquatic ecologies are essential to interpretations here, but my reading of the appendix suggests they are often debated and the choices made here involve an element of circular reasoning (i.e., they chose the one that leads to some simpler, preferred interpretation of the oxygen isotope data). This seems to me like a case where you really need to focus on unambiguous cases (e.g., marine reptiles, frogs, etc.).

2) It is assumed that ectotherms have growth temperatures equal to mean annual air temperatures, which are approximated with a linear dependence on latitude. These are both contraindicated by environmental data. Ectothermic tetropods generally have body temperatures well in excess of mean annual air temperature, particularly during their active seasons when they likely do most of their growth. For example, a recent survey of field-measured lizard body temperatures yielded a range of ~30-35 C across a wide range of latitudes and altitudes. And the earth's atmospheric temperature gradient is relatively gentle up to 60 ˚C, after which it really drops like a shot. The combination of these factors suggests to me that the authors have over estimated the contrast between ectotherm and endotherm d18O values, by perhaps a factor of two or more. These could have been improved on using modern field data (i.e., compare similar organisms in modern environments).

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triassic therapsid clades" for further consideration at eLife. Your revised article has been favorably evaluated by Diethard Tautz (Senior editor), a Reviewing editor, and one reviewer.

The manuscript has been improved but there are some small remaining issues that need to be addressed before acceptance, as outlined below:

Reviewer #1:

The paper presents an unexpected result, that endothermy arose more than once among lineages of synapsids. The paper is novel in basing this result on new isotopic data measured from fossil bones. The methods are difficult to get right, but the Lécuyer laboratory at Lyons leads the world in these methods, which adds credibility to the paper.

The Abstract is not entirely clear about the key claim, that endothermy arose twice, independently, among Synapsida. If that is the claim, say so clearly. The last sentence needs revision.

The Introduction starts with a single-sentence first paragraph about environmental changes – this could be cut. Then go straight into definition and explanation of endothermy – the first sentences about endothermy need some references to justify all the claims.

Results section: thirteen = 13. Check throughout that numbers follow the usual convention, of one to ten in full, and 11 upwards as digits, except when the number is the first word in a sentence, when it is given in full.

Discussion section paragraph two: cut – pointless sentence.

Discussion paragraph seven: inherited by = inherited from.

Discussion paragraph seven: what is the purpose of this paragraph? – it wanders everywhere. Focus simply on the key point: could endothermy have evolved once only in the ancestors of the L+K clade and Epicynodontia? Discuss and reject. Move to new paragraph.

Figure 5: make the arrows larger and brighter – maybe red. This is the key of the paper to argue the phylogenetic point. I'd also mark all sampled synapsids with a coded symbol for definitely endothermic and definitely ectothermic – this will prove that ancestors shared by the K+L clade and Epicynodontia were ectothermic, so confirming the likelihood that endothermy arose twice, independently.

https://doi.org/10.7554/eLife.28589.010

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

[…] Reviewer #1:

This is a potentially important paper, presenting a new geochemical means to identify thermoreguylatory principles in early reptiles, and identifying evidence that endothermy may have arisen twice in the synapsids.

However, it is not ready for review in a journal yet as it reads like a thesis chapter. First, the different sections of the paper must be discriminated – 'Methods' are mixed through the Methods and Results sections, and must be separated. By reorganizing the text and focusing on a single (rather than multiple) Methods sections, the paper can become simpler, shorter, and easier to read. Make sure the Results section reports simply the results.

The ‘Methods’ elements from the ‘Results’ section have been moved to the corresponding section. Several sections in Materials and methods have been selected to divide several independent parts (samples, analysis, diagenetic process, model construction).

I'm not sure that eLife format has been strictly followed either – the authors should check section numbering and styles are correctly followed.

There is no imposed format from eLife. Therefore, we chose one constant format for the manuscript and the Appendix.

In the figures, try to minimize abbreviations (e.g. ‘Terr. Vs. semiaq. Similar thermophy.'; this is not necessary. Use pure abbreviations (e.g. AZ) or spell out in full.

The abbreviations to describe the areas have been spelled out in full in the Figures 2, 3 and 4.

Results subsection “Robustness of the stable isotope record”: This reads pretty much like 'Methods', so move it back to that section.

This section has been moved into ‘Materials and methods’.

Results section second paragraph: This paragraph is a curious mixture of Methods and Figure caption. Return 'methods' stuff to the Materials and methods section – 'were calculated' 'were compared' – all methods.

The ‘Methods’ parts have been moved to the Materials and method section.

Then the description of symbols and shading in Figures 2–4 should be placed in the figure captions.

In order to avoid very long, and identical, figure captions for Figures 2, 3 and 4, the description of the different areas is left in its own Methods section. This information is presented in the Figures with embedded texts for each area.

Then, under Results, we need a plain writing description of what the plots show.

The Results section describes the range of values obtained, country by country, for each time period, and the areas where the vales were obtained.

I think 'Results' begins at subsection “Therapsid thermophysiology” – so move text, combine sensibly, and re-label sections.

Section moved and re-labelled.

No, not really: the first paragraph of subsection “Therapsid thermophysiology” is more 'Methods' – telling us which taxa were studied and from where. Please apply these general principles of scientific paper writing properly and reorganize and rationalize the paper.

This paragraph has been rewritten as:

“The thirteen sampled South African Permian therapsids come from three different assemblage zones (AZ) of the Beaufort Group: the lower Tapinocephalus AZ, the Tropidostoma AZ and the lower Daptocephalus AZ (Viglietti et al., 2016).

Oxygen isotope compositions of three therapsid genera (Dicynodon, Diictodon and Oudenodon) from the youngest assemblage zone (lower Daptocephalus AZ) and seven therapsid genera (Aelurosaurus, Diictodon, Ictidosuchoides, Oudenodon, Rhachiocephalus, Tropidostoma and a basal cynodont) from the Tropidostoma AZ were respectively compared with one co-existing Rhinesuchus and one co-existing rhinesuchid.”

In addition: Before giving the values of the isotopic differences between the therapsids and the co-existing tetrapods, it is pertinent for us to present which of these are therapsids and their locality information. This information could be included in Materials or Methods, but we consider that it will be easier to read and comprehend if the genus name and isotopic values were placed in the same paragraph instead of separated by several pages. Moreover, some of the taxa (such as Dicynodon, Diictodon and Oudenodon) are present at several localities, and need to be cited in results. For this reason we have kept this section under results.

Reviewer #2:

General comments: This paper uses comparisons of stable oxygen isotope ratios of Permian-Jurassic non-mammalian synapsids to their contemporaries to try to infer when an endothermic metabolism arose in the lineage leading to mammals. The results are thought provoking, but I think a number of changes and additions will need to be made before the paper is ready to accept. I've made a number of comments and suggestions below, but a few general issues that must be addressed:

1) I am not an isotope geochemist, and I think it will be important for one to examine the paper. I'm not so worried about the analytic methods, which seem fairly standard, but I think it would be good to get a more informed perspective on the method the authors use to predict the expected magnitude of differences between ectotherms and endotherms. This is the foundation of everything else in the paper, so it is important to make sure that this is acceptable.

We have now explained in the section “2.4” how the expected magnitude of differences between ectotherms and endotherms have been defined. We added the following sentence to provide more information in the text (a detailed explanation is provided in the Appendix 1): “To construct those theoretical areas, both the phosphate-water temperature scale from Lécuyer, Amoit, Touzeau and Trott, 2013 and the differences of stable oxygen compositions between mammals of various ecologies from Cerling et al., 2008 have been used.”

2) Assuming the method is acceptable, I strongly recommend that the authors conduct a sensitivity analysis using a range of plausible average temperatures and latitudinal gradients for each of their time periods of interest. Their results will be stronger if they are consistent across the plausible range of values, and it will remove any appearance of cherry-picking temperatures/gradients to get the results that they want.

Following the reviewer’s request, we recalculated the areas of theoretical values for the difference between an endotherm and an ectotherm by using a range of plausible thermal latitudinal gradients. All three different gradients, 0.6°C\°L, 0.4°C\°L and 0.5°C\°L, are now with an uncertainty range of 0.1°C\°L to take into account the possible variation of this thermal latitudinal gradient within the periods selected. Therefore, we added in the Appendix 1 the following sentence: “In order to take into account the possible variations of the thermal gradient within the selected periods, an interval of ± 0.1°C/°L is added for each Main text figures.”

3) The authors seem to play pretty fast and loose with the terms "endothermy" and "ecothermy." A particular manifestation of this is the frequency with which they use bone histology (particularly the presence of fibrolamellar bone) as a correlate of endothermy. Although this was more common in earlier work on paleohistology, more recently the field has been much more circumspect about this relationship, with much more attention placed (rightly, in my opinion) on the implications of histology for growth rates/patterns. Elevated growth rates often are correlated with "endothermy" but this is not universally true.

We acknowledged the fact that this relationship is indirect and have thus added the following sentence in the Introduction: “However, FLB also occurs in a few ectotherms such as in some turtles and crocodilians, and is absent in small mammals and passerine birds(Bouvier, 1977), showing that FLB is mostly correlated with high growth rates, which does not always correlate to high metabolic rates.”

I think the authors need to be more thoughtful in their presentation of this relationship, and they also should provide more explicit, detailed definitions of "endothermy" and "ecothermy" at the start of the paper.

We added the following paragraph in the Introduction:

“Endothermy is commonly associated with homeothermy, being the capacity to regulate the body heat through metabolic activity as well. This combination corresponds to one end of a gradient of thermoregulatory strategies observed in living animals. The other end of the spectrum is ectothermy combined with poïkilothermy which animals use as a thermoregulatory strategy to increase their body temperature toward optimal levels by using external heat sources. Their body temperature therefore traces that of their surroundings and is the most occurring saving strategy.”

4) I recommended some changes to the discussion that would explore the link between endothermy and survivorship during the Permo-Triassic Mass Extinction (PTME) that would increase the broad interest and impact of the paper. Alternatively, the authors could provide a more detailed discussion of their results in the context of previous work and current ideas on the evolution of endothermy. Either way, much of what's in the current discussion isn't very interesting (and I say this as someone who loves the intricacies of dicynodont taxonomy) and it undersells the paper.

In order to make this important part of the paper more exciting we have extensively modified it and also added the following paragraph: “A possible explanation could be the acquisition of a fast growth rate due to the high metabolic rate of the endothermy. According to a recent palaeohistology study (Botha-Brink et al., 2016) Early Triassic therapsids, such as Lystrosaurus, therocephalians and cynodonts, had a high growth rate allowing them to reach reproductive maturity within a few seasons and compensate their shortened life expectancies. This adaptation might have enabled certain therapsids to survive the intense climatic change of that time and conquer the newly vacant niches.”

5) There are a couple places in the main text and supplement that could be strengthened in terms of the references that are cited.

Numerous references have been added to address this comment.

6) The writing style of the paper is good, but it could be improved upon. I recommend having Rubidge, Smith, and Viglietti go over the final version of the paper to clean things up and make sure the idiom is appropriate.

All three of them went through the manuscript after all the comments from the authors were applied.

Abstract: mammals didn’t exist in the Permian, so you should say something along the lines of “endothermy evolved in the Permian ancestors of mammals.”

“[M]ammal” has been changed to “mammal ancestors”.

Abstract: Cynognathia is a subclade of Eucynodontia, so I’m not sure what you mean here.

“Cynognathia and Eucynodontia” has been changed to “Eucynodontia”.

Abstract: change to “originated in Epicynodontia during the…”. Also, change to middle-late Permian.

Changes have been made.

Introduction opening sentence: this sentence is quite vague and is kind of meaningless as a result. Please be more specific about particular ways in which environmental changes affect vertebrates if you want to use this kind of example.

We have modified the sentence as follows: “Rapid environmental changes have played a major role in shaping the evolutionary history of terrestrial vertebrates not only as the major cause of mass extinctions but also favouring species with biological adaptations that allowed them to survive and oikilothe the altered ecosystems..”

Introduction, first paragraph: as written, this paragraph provides something of a definition of endothermy. However, because endothermy is kind of a loaded term, and has been used in different ways by different authors, I suggest that you explicitly define what you mean by it so as to avoid any confusion.

We added the following paragraph in the Introduction: “Endothermy is commonly associated with homeothermy, being the capacity to regulate the body heat through metabolic activity as well. This combination corresponds to one end of a gradient of thermoregulatory strategies observed in living animals. The other end of the spectrum is ectothermy combined with oikilothermy which animals use as a thermoregulatory strategy to increase their body temperature toward optimal levels by using external heat sources. Their body temperature therefore traces that of their surroundings and is the most occurring saving strategy.”

Introduction paragraph two: again, defining endothermy will help to avoid confusion about what you mean by “true endotherms.”

The sentence has been modified to: “Amongst modern vertebrates, various thermoregulatory strategies between these two end-members have been adopted, such as regional endothermy (Bernal, Dickson, Shadwick and Graham, 2001; Katz, 2002) or inertial homeothermy (McNab and Auffenberg, 1976), and only birds and mammals fall within the endothermy-end of the spectrum.”

In the same paragraph: I think you're kind of downplaying the amount of work on this subject. For example, McNab, 1978, Hillenius, 1992; 1994, Ruben, Hillenius, Kemp and Quick, 2012, Owerkowicz, Musinsky, Middleton and Crompton, 2015, all would be appropriate to cite here.

This section has been rewritten and all the suggested references have been added.

Introduction paragraph four: bony secondary palate would be a better term to use.

The term has been changed.

In the same paragraph: Dicynodonts also have a complete secondary palate, although it is formed primarily by the maxilla.

We added the following sentence: “It is noteworthy that complete bony secondary palate is present in dicynodonts, the difference being that it is primarily formed by the premaxilla (King, 1988) whereas it is mostly the maxilla in therocephalians, cynodonts and extant mammals.”

Introduction paragraph five: Owerkowicz, Musinsky, Middleton and Crompton, 2015 would be especially good to cite in the context of whether turbinates are necessary in the context of mammalian endothermy. Also Laaß (2011) presents some evidence of turbinals in Lystrosaurus, although I'm kind of skeptical as to whether that's actually what's visible in the scan.

The first reference has been added.

Reports of fibrolamellar bone in synapsids go back farther than these studies. For example, de Ricqles reports fibroilamellar bone in some therapsids is in papers on bone histology from the 1970's.

We modified accordingly and added the following references: “(de Ricqlès, 1972, 1979; Botha, 2003; Botha and Chinsamy, 2001, 2004; Ray, Botha and Chinsamy, 2004; Olivier, Houssaye, Jalil and Cubo, 2017)”.

Fibrolamellar bone is a lot more ubiquitous among therapsids than what you suggest here; if anything it's more common than not. There's even pelycosaur-grade synapsids like Sphenacodon and Ophiacodon that produce fibrolamellar bone for at least part of their ontogenies.

We modified the paragraph to: “Accordingly, several bone palaeohistological studies have have addressed the quest for the presence of FLB in therapsids (de Ricqlès, 1972, 1979; Botha, 2003; Botha and Chinsamy, 2001, 2004; Ray, Notha and Chinsamy, 2004; Olivier, Houssaye, Jalil and Cubo, 2017). Ray, Notha and Chinsamy, 2004 and Olivier, Houssaye, Jalil and Cubo, 2017 analysed several therapsid groups (anomodont, gorgonopsian, therocephalian, cynodont) and found FLB in some genera (Aelurognathus, Pristerognathus, Tritylodon, Oudenodon, Lystrosaurus, Moghreberia), suggesting sustained fast growth and thus elevated metabolic activity. The presence of FLB has also been demonstrated in some earlier non-therapsid synapsids such as Sphenacodon, Dimetrodon or even Ophiacodon (Huttenlocker, Rega and Sumida, 2010; Shelton, Sander, Stein and Winkelhorst, 2012; Shelton and Sander, 2015).”

In addition to the fact that some non-ectothermic teterapods can produce fibrolamellar bone, and important point to make here is that the presence of fibrolamellar bone really is more a reflection of the rate at which bone tissue is being formed than endothermy in a strict sense. While the high growth rates required to produce fibrolamellar bone often are correlated with high metabolic rates, this is not always the case.

We have added the following sentence: “However, FLB also occurs in a few ectotherms such as in some turtles and crocodilians, and is absent in small mammals and passerine birds(Bouvier, 1977), showing that FLB is mostly correlated with high growth rates, which does not always correlate to high metabolic rates.”

Introduction paragraph six: although the degree of resistance to diagenesis varies among tissues, with enamel being the more resistant by far.

The paragraph here explains the reason why the oxygen isotopic composition of the phosphate group is used. The difference between tissues is described later in “Materials and methods” as follows: “Analysed materials consist of bone or tooth dentine, which is more porous than enamel with small and less densely inter-grown apatite crystals (Mills, 1967)”.

Materials and methods section, subsection “Sample collection”: you should provide justification for why you're using these alternate dates.

Since the submission of this manuscript, the alternative dates have become the official ones. We have modified the sentence to read: “All the sample localities have been correlated to the marine biostratigraphic stages using the absolute ages accepted by the International Commission on Stratigraphy (Cohen et al., 2013; updated 12/2016), with the Permo-Triassic and Guadalupian-Lopingian boundaries now considered to be at 251.90 ± 0.02 Ma (Burgess, Bowring and Shen, 2014) and 259.1 ± 0.5 (Zhong, He, Mundil and Xu, 2014) Ma, respectively.”

Also, looking at your supplementary table, it seems like some of your dates imply more precision than is actually the case. For example, as far as I know there aren't any published radiometric dates for the Burgersdorp Fm. If you're making the correlation to marine stages (and their dates) using tetrapod biostratigraphy, I would suggest caution in this case, given the recent dates from South America that have been published by Ottone et al. (2014) and Marsicano et al. (2016). In any case more information on the source of your dates should be provided.

We intended to give an age value to every studied stratigraphic interval. However, as pointed out by the reviewer, not all formations have radiometric ages and in these instances we used biostratrigraphic correlation (and references) as explained in the text:

“Triassic age determination has been achieved by biostratigraphic correlation with Laurasian sequences (Hancox, Shishkin, Rubidge and Kitching, 1995; Rubidge, 2005; Abdala and Ribeiro, 2010).”

“The locality belongs to the “Red Beds inférieurs a or b” of the lower Elliot Formation which is currently regarded as latest Triassic (late Rhaetian) (Knoll, 2004).” “The locality is biostratigraphically correlated to the last segment of the Timezgadiouine Formation, considered to be Middle to early Late Carnian(Jalil, 1999).”

“The Dashankou locality, from Gansu Province, is biostratigraphically dated as Early Roadian (Liu and Rubidge, 2009; Liu, 2010). From Shanxi Province, sampled fossils originate from three localities in the Ermaying Formation which is considered to be Anisian (Liu, Li and Li, 2013). From Xinjiang Province, two localities in the Jiucaiyuan Formation have been sampled and are considered Early Triassic (Metcalfe et al., 2009).”

Subsection “Robustness of the stable isotope record”: “contain” instead of “contains”.

The modification has been made.

Figure 1: The colouration of the different parts of the plot doesn't seem to be explained. I understand that the dark orange area at the top right is the area that represents diagenesis, but do the other coloured regions have some meaning?

The arrows on the side describe the figure with all the upper section corresponding to phosphate diagenesis, and the part on the right corresponding to carbonate diagenesis. Therefore, the upper right part corresponds to both phosphate and carbonate diagenesis.

Subsection “Robustness of the stable isotope record”: does partial preservation of the isotopic signal require different interpretation than pristine preservation?

Partial preservation means that some of the samples might have their isotopic compositions slightly modified by diagenesis, meaning that it is impossible to be 100% certain if the signal is completely original. With no way to quantify it, we interpret our data as preserved, acknowledging the possible slight alteration. Therefore, we modified the sentence as follows: “Using these two assessments, newly measured δ18Op values are considered to have preserved their original isotopic signatures and can be interpreted in terms of ecologies and physiologies”.

In the same subsection: Figure 2A is repeated here. Do you mean 2A, 3A, and 4A?

The modification has been made to 2A, 3A and 4A.

Would it be better to refer to the colours instead of light and dark gray? Since the journal is primarily an online publication the colour online/grayscale printed version seems like it isn't really a concern, and referring to the colours would help readers understand the figures.

We have now included colour as suggested and all the shades of grey have been changed to the corresponding colours.

Permian Therapsids general comment: I recommend reorganizing this section a little. I think it would be most logical to present the results subdivided by time (i.e., all middle Permian things first, then early late Permian, then latest Permian). Within each time bin, you can subdivide further by geographic area.

The chosen subdivision has been made in order to avoid repetition and make reading easier. Accordingly we first listed the South African fossils because of their greater numbers and their possible occurrences in several AZ (eg. Diictodon and Oudenodon in both lower Daptocephalus AZ and Tropidostoma AZ). Some AZ are also linked together with the groups used for comparison (eg. stereospondyls).

Early-Middle Triassic Therapsids general comment: as with the Permian taxa, I think it would be most useful to subdivide the results first by time and then by geography.

In this case we preferred to subdivide the results first by country with a link between South Africa and China, which included Lystrosaurus which is common to both. Division by time would cause the text to alternate between the South African and Chines fossils.

Subsection “Early to Middle Triassic therapsids”: because bone histology is really more a record of growth rates, it might be better to say something along the lines of "…which is consistent with the elevated growth rates implied by the bone histology of Erythrosuchus."

The sentence has been changed to: “Erythrosuchus was also endothermic which is consistent with the elevated growth rates implied by its palaeohistology”.

I think the paragraph describing the results on the Chinese lystrosaurid should be combined with the previous paragraph on the South African specimens. Also, Lystrosaurus is the only Lystrosaurid known from China, so I would say its safe to assume that it is Lystrosaurus.

As suggested the paragraphs have been combined. As there is no more precise official identification for the Chinese fossil, we preferred to leave it as it is.

In the same subsection and in Figure 3 you make it seem like it is pretty certain that Lystrosaurus is an endotherm. However, in the supplement, you make it seem like there's actually a fair amount of uncertainty based on questions about the ecology of Lystrosaurus and the metaboloic status 'Chasmatosaurus' (which is inferred from information on growth rates inferred from bone histological data from related taxa in South Africa). I think you need to better portray this uncertainty here, and also be sure to take it into account in the Discussion section of the paper.

The Appendix paragraph mentions that the South African Lystrosaurus is either terrestrial-endotherm or semiaquatic-ectotherm, whereas the Chinese one is either terrestrial endotherm or terrestrial ectotherm. Considering that the South African and Chinese Lystrosaurus probably occupied the same ecological niche, the only common possibility is terrestrial endotherm.

Subsection “Middle to Late Triassic therapsids”: contemporary instead of co-existing

Modification has been made.

Discussion section and elsewhere: although wordier, something like "the unnamed clade comprising Lystrosauridae + Kannemeyeriiformes" would make more sense.

The sentence has been modified to: “the unnamed dicynodont clade comprising Lystrosauridae + Kannemeyeriiformes, abbreviated the “L+K” clade later,”

Discussion paragraph three: "basal dicynodontoid" would work better.

Modification has been made.

Discussion paragraph five: I don't really agree that this is a valid hypothesis. You have a fair number of Permian therapsids from different clades that are reconstructed as being ecothermic, and it seems implausible that you would have so many reversals to ecothermy. In the previous paragraph, you discuss a hypothesis based on a parsimony optimization of the character on the phylogeny. That is justified: you have a clear-cut optimality criterion and you are discussing the best hypothesis under that criterion. I don't think there's any method/optimality criterion that would support the number of reversals needed to account for all so many of your Permian therapsids being reconstructed as endotherms, so the argument is much less sound. If you're doing that, then you can pretty much make up whatever hypothesis you want because there's no constraint on what could be considered plausible.

We consider this second hypothesis still valid, as it does not involve reversal from endothermy to ectothermy, but that the characteristics for endothermy were present but not a full endothermy. Therefore, we added the following sentences: “This suggests that biochemical and physiological mechanisms allowing mammal endothermy, appeared at the base of the neotherapsids at any given time during the middle Permian which is a conclusion recently published based on the paleohistology of some dicynodonts (Olivier, Houssaye, Jalil and Cubo, 2017). In our case, the absence of an endothermic signal in the sampled Permian therapsids could be due to an endothermy being only seasonal, linked to a cold season or to the reproduction period (as observed today in some reptile species; Tattersall et al., 2016), which would not be visible in a bulk signal. Therefore, effective acquisition of mammal “true endothermy” was expressed within these two lineages, possibly as a result of extrinsic factors.”

In the same paragraph: biochemical instead of biochemicals

Modification has been made.

Discussion paragraph seven: assuming that Lystrosaurus was indeed an endotherm (see comments above and below), these observations are especially interesting in the context of the changes in growth and life history at the PTB noted by Botha-Brink et al., 2016, among others. This would be something interesting to explore further in the discussion, as well as a more general consideration of the role endothermy played in determining survivorship during the PTME. There's a couple obvious ways to get the space needed to do this. One would be to get rid of the previous paragraph, which I don't think is needed. Likewise, you could get rid of everything from paragraph two, and almost everything in paragraph three.

We included this hypothesis at the end of the Discussion: “A possible explanation could be the acquisition of a fast growth rate due to the high metabolic rate of the endothermy. According to a recent palaeohistology study (Botha-Brink et al., 2016) Early Triassic therapsids, such as Lystrosaurus, therocephalians and cynodonts, had a high growth rate allowing them to reach reproductive maturity within a few seasons and compensate their shortened life expectancies. This adaptation might have enabled certain therapsids to survive the intense climatic change of that time and conquer the newly vacant niches.”

Concluding remarks: this is a really stilted way to say that most of your samples preserved an original signal that had not been diagenetically altered.

We deleted the offensive sentence and modified the second one as it follows: “Assuming that analysed samples have preserved their original isotope composition of phosphate, all the Permian therapsids analysed appear to have ectotherm-like thermoregulation and representatives of two Triassic therapsid clades are considered to have had endotherm-like thermoregulation: the Lystrosauridae + Kannemeyeriiformes and the Eucynodontia.”

I think you can get rid of most of this (see comments above).

We think that this hypothesis is still valid (see above).

General comment on the supplement: you should reorganize so that the first section is the one describing how you came up with your theoretical ranges of isotope values for taxa with different ecologies and thermophysiologies. That's the foundation for all of the interpretation of the fossil data you present in the paper and the supplement, so it deserves to go first. If space (or editorial largesse) allows, I would even suggest moving the information on how you calculated your expected ranges into the main text.

The section “Theoretical curve construction” has been moved to the first section. However, to keep the “Materials and methods” section as simple as possible, we deliberately chose to keep the calculation of the theoretical range in the Appendix. We have now added the following sentence in the main text: “To construct those theoretical areas, both the phosphate-water temperature scale from Lécuyer, Amoit, Touzeau and Trotter, 2013 and the differences of stable oxygen compositions between mammals of various ecologies from Cerling et al., 2008, have been used.”

Appendix paragraph two: change heading to "Stereospondyl amphibians." Also make this change in the first line of this paragraph.

Changes have been made.

Also, does the fact it was amphibious complicate your comparisons between it and Lystrosaurus?

Not really, the Figures 2A, 3A and 4A are the comparison between a therapsid and an amphibious animal.

Appendix subsection “Pareiasaurid and bolosaurid parareptiles”: would the fact that Pareisaurus was an herbivore whereas Rhinesuchus was a faunivore make any difference here? I suppose the results for the older pareiasaurus would suggest not.

The effect of diet on the oxygen isotope ratio is not overwhelming with regard to the importance of the water budget. Differences in diet may produce a much more significant difference for carbon and nitrogen isotope ratios but nothing significant for the oxygen.

Appendix subsection “Lystrosaurid therapsids”: going along with my comments above about the identity of the Chinese material, I recommend changing the title of this section to "Lystrosaurus."

As mentioned above: Even though we agree with the comment of the reviewer, we prefer to retain the term “Chinese lystrosaurid” until an official new identification has been made.

In the same section: there is a much deeper history of debate in the literature about Lystrosaurus potentially being semi-aquatic that would be worth citing here: Broom 1903b, 1932; Watson 1912, 1913; Williston 1914; Brink 1951; Camp 1956; Cluver 1971; Kemp 1982; Hotton 1986; King 1991; King and Cluver 1991; Germain & Laurin 2005; Ray et al. 2005; Botha-Brink and Angielczyk 2010

The references from 2005 to present have been added in order to present that this point of view. Several authors are still debating the subject.

Your interpretation of Lystrosaurus as an endotherm seems a lot more uncertain here than you portray it in the text. This uncertainty has important implications for your the evolutionary scenarios you present in the Discussion section and considerations of the potential relationships between endothermy and the PTME. Therefore, I think you need to present this uncertainty more clearly in the main text (there you make it seem like Lystrosaurus is definitely an endotherm, which seems kind of misleading in the context of what's in the supplement).

As mentioned above: The Appendix paragraph mentions that the South African Lystrosaurus is either terrestrial-endotherm or semiaquatic-ectotherm, whereas the Chinese one is either terrestrial endotherm or terrestrial ectotherm. Considering that the South African and Chinese Lystrosaurus probably occupied the same ecological niche, the only common possibility is terrestrial endotherm. We do not think that the text is misleading our interpretation of the ecology of Lystrosaurus.

Appendix subsection “Theroretical curves construction”: again, there doesn't seem to be any gray areas in the figures in the main text, so I would refer to the colours you use in those figures.

We replaced all the shades with their respective colours.

It would be good to have a citation for 37C being the average body temperature of terrestrial mammals.

A reference has been added as follows: “For endothermic vertebrates, Tb is assumed to be 37°C, the average Tb of most extant placentals and possibly of the common ancestor of all extant mammals (Watson and Graves, 1988).

In the same subsection: refer to colours for the main text figures.

We changed all the shades to their respective colours.

General comment on the three time periods: estimates of global temperature and latitudinal thermal gradients obviously are not exact (especially when you're arbitrarily picking values for the latter in at least some cases). Because of this, it would be good to have more citations for where you're getting your values from, and justification for why you chose particular values/references. Likewise, it seems like it would be good to do a sensitivity analysis using a range of different (plausible) average temperatures and thermal gradients. Your results will seem a lot stronger if you consistently reconstruct your proposed endothermic taxa as endotherms across the range of plausible values than would be the case if you you have them come out as endotherms for certain combinations of values. Doing this will also avert criticisms that you are cherry-picking values to get the results you want.

In order to reinforce our interpretations, we extended our theoretical ranges by adding an uncertainty of ± 0.1°C/°L for each thermal gradient we used. Uncertainties may reflect the possible variations of temperature during each chosen timespan. No interpretations were affected by this increase in those theoretical ranges.

Reviewer #3:

I have completed my review of Rey et al.'s manuscript, "Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triasic therapsid clades'. This is a somewhat challenging paper to evaluate: The manuscript is exceptionally well written – clear, concise, scholarly, rationally organized; really a pleasure to read. And I see no reason to question the significance of the samples studied or the quality of the isotopic measurements. So, on these bases alone I would say the paper should be published with modest revisions. However, I do not believe several of the assumptions that underlie the interpretation of the isotopic data set are well justified, and therefore I don't consider the study to offer a strong new constraint on the question of therapsid physiology. On the third hand (!?!), the authors have looked at one of the few geochemical properties that potentially constrain body temperatures of these organisms, and I'm a strong believer that it is always worth while to publish possibly significant observations. So, I would like to see this paper published but encourage the authors to do some soul searching about some of their interpretations.

The two most important issues, in my mind, are the state of preservation of the samples, and the robustness of the models that are used as a reference frame for interpreting their findings.

First, there is evidence indicating that geochemical properties of ancient bone or dentin do not consistently constrain body temperature, except in rare cases of extraordinary preservation. Both are composed of fine-grained phosphate intermixed with organics, and demonstrably undergo diagenetic changes in oxygen isotope composition on geologically short time scales. It is asserted that phosphate resists such changes, but there is plenty of data showing this rule of thumb is often violated (see Stolper et al., 2016 for an example).

The authors have tried to avoid this by filtering the database for signs of alteration, but they do so with relatively simple, indirect metrics (percent carbonate and the phosphate-carbonate fractionation). I would have found it more compelling if they could have shown through microscopy and/or minor element mapping that they were studying well-preserved fabrics. Or if they had focused on only tooth enamel, which is generally much more resistant to alteration.

Admittedly enamel would be the best apatite to sample, but before its development in the Mammalia, enamel was just a thin layer and was possibly not sufficient to sample for the analysis. Moreover, some species did not have enamel at all (eg. Diictodon), or even teeth (eg. Oudenodon), therefore dentine and bone offer the best alternative. Much of the data presented in this manuscript has already been published (Rey et al., 2016) and showed a correlation with marine data from South China through the studied interval. Currently there is no method to demonstrate if diagenetic processes might have significantly altered the oxygen isotope compositions of the phosphate group. In their study, Pucéat et al., 2004 showed that recrystallization and Rare Earth Element fractionations are possible without altering oxygen isotope compositions.

Second, the models of the oxygen isotope compositions of endotherm and ectotherm d18O values are based on two variables that strike me as poorly known or known to be false:

1) The assignments of taxa to terrestrial vs. semi-aquatic ecologies are essential to interpretations here, but my reading of the appendix suggests they are often debated and the choices made here involve an element of circular reasoning (i.e., they chose the one that leads to some simpler, preferred interpretation of the oxygen isotope data). This seems to me like a case where you really need to focus on unambiguous cases (e.g., marine reptiles, frogs, etc.).

2) It is assumed that ectotherms have growth temperatures equal to mean annual air temperatures, which are approximated with a linear dependence on latitude. These are both contraindicated by environmental data. Ectothermic tetropods generally have body temperatures well in excess of mean annual air temperature, particularly during their active seasons when they likely do most of their growth. For example, a recent survey of field-measured lizard body temperatures yielded a range of ~30-35 C across a wide range of latitudes and altitudes. And the earth's atmospheric temperature gradient is relatively gentle up to 60 ˚C, after which it really drops like a shot. The combination of these factors suggests to me that the authors have over estimated the contrast between ectotherm and endotherm d18O values, by perhaps a factor of two or more. These could have been improved on using modern field data (i.e., compare similar organisms in modern environments).

Comparing isotope compositions to understand thermophysiology of fossil species have been already published a few times (Amiot et al., 2006; Barrick et al., 1994; 1996). Here, we used a theoretical framework based on extant species to increase the validity of our result. Nevertheless, we added a paragraph in the Discussion where we acknowledge the possible limits of this theoretical framework: “Such interpretations are based on a theoretical framework in which estimations had been made as accurately as possible, but may not entirely represent the complex ecologies or thermophysiologies of each analysed tetrapod or the environmental conditions where they lived. Therefore, even though we have confidence in these results, it is important to compare them with other palaeobiological proxies to increase their validity.”

[Editors' note: the author responses to the re-review follow.]

Reviewer #1:

The paper presents an unexpected result, that endothermy arose more than once among lineages of synapsids. The paper is novel in basing this result on new isotopic data measured from fossil bones. The methods are difficult to get right, but the Lécuyer laboratory at Lyons leads the world in these methods, which adds credibility to the paper.

The Abstract is not entirely clear about the key claim, that endothermy arose twice, independently, among Synapsida. If that is the claim, say so clearly. The last sentence needs revision.

We clarified this as follows:

“It is proposed that cynodonts and dicynodonts independently acquired constant elevated thermometabolism, respectively within the Eucynodontia and Lystrosauridae + Kannemeyeriiformes clades”.

The sentence has been changed to:

“We conclude that mammalian endothermy originated in the Epicynodontia during the middle-late Permian. Major global climatic and environmental fluctuations were the most likely selective pressures on the success of such elevated thermometabolism.”

The Introduction starts with a single-sentence first paragraph about environmental changes – this could be cut.

The sentence has been removed.

Then go straight into definition and explanation of endothermy – the first sentences about endothermy need some references to justify all the claims.

Several sentences had references added:

“One key adaptation enabling tetrapods to cope with fluctuating climatic conditions was the acquisition of endothermy (Paaijmans et al., 2013). This character is defined here as the ability to actively produce body heat through metabolic activity (Cannon and Nedergaard, 2004).”

“…endothermic vertebrates are able to colonise environments with extreme thermal conditions, for example freezing at high latitudes and altitudes (Khaliq et al., 2014).”

Results section: thirteen = 13. Check throughout that numbers follow the usual convention, of one to ten in full, and 11 upwards as digits, except when the number is the first word in a sentence, when it is given in full.

The modifications have been applied.

Discussion section paragraph two: cut – pointless sentence.

The sentence has been removed.

Discussion paragraph seven: inherited by = inherited from.

The modification has been made.

Discussion paragraph seven: what is the purpose of this paragraph? – it wanders everywhere. Focus simply on the key point: could endothermy have evolved once only in the ancestors of the L+K clade and Epicynodontia? Discuss and reject. Move to new paragraph.

This paragraph shows that even with an independent acquisition of endothermy in the two clades, we cannot discard a possible homologous origin of the metabolic process leading to constant endothermy in some clades. Accordingly we modified the last sentence of that paragraph, as follows:

“Therefore, effective acquisition of mammal “true endothermy” was expressed independently within these two lineages, possibly as a result of extrinsic factors.”

Figure 5: make the arrows larger and brighter – maybe red. This is the key of the paper to argue the phylogenetic point. I'd also mark all sampled synapsids with a coded symbol for definitely endothermic and definitely ectothermic – this will prove that ancestors shared by the K+L clade and Epicynodontia were ectothermic, so confirming the likelihood that endothermy arose twice, independently.

The arrows have been modified to be larger and red. The endothermic species were already in bold, and are now also in red.

https://doi.org/10.7554/eLife.28589.011

Article and author information

Author details

  1. Kévin Rey

    1. Univ Lyon, Université Lyon 1, Ens de Lyon, CNRS, UMR 5276 LGL-TPE, Villeurbanne, France, France
    2. Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, Johannesburg, South Africa
    Contribution
    KR, Conceptualization, Data curation, Formal analysis, Investigation, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    kevin.rey@wits.ac.za
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6788-2453

  2. Romain Amiot

    Univ Lyon, Université Lyon 1, Ens de Lyon, CNRS, UMR 5276 LGL-TPE, Villeurbanne, France, France
    Contribution
    RA, Conceptualization, Data curation, Formal analysis, Supervision, Project administration, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  3. François Fourel

    1. Univ Lyon, Université Lyon 1, Ens de Lyon, CNRS, UMR 5276 LGL-TPE, Villeurbanne, France, France
    2. Univ Lyon, Université Lyon 1, CNRS, UMR 5023 LEHNA, Villeurbanne, France, France
    Contribution
    FFo, Conceptualization, Data curation, Project administration, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  4. Fernando Abdala

    Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, Johannesburg, South Africa
    Contribution
    FA, Resources, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  5. Frédéric Fluteau

    Institut de Physique du Globe de Paris, Paris, France
    Contribution
    FFl, Formal analysis, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  6. Nour-Eddine Jalil

    Centre de Recherches en Paléobiodiversité et Paléoenvironnements, UMR 7207 CNRS-MNHN-UPMC, Museum National d’Histoire Naturelle, Paris, France
    Contribution
    N-EJ, Resources, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  7. Jun Liu

    Key Laboratory of Vertebrate Evolution and Human Origins of Chinese Academy of Sciences, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China
    Contribution
    JL, Resources, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  8. Bruce S Rubidge

    Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, Johannesburg, South Africa
    Contribution
    BSR, Resources, Funding acquisition, Project administration, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  9. Roger MH Smith

    1. Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, Johannesburg, South Africa
    2. Iziko South African Museum, Cape Town, South Africa
    Contribution
    RMHS, Resources, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  10. J Sébastien Steyer

    Centre de Recherches en Paléobiodiversité et Paléoenvironnements, UMR 7207 CNRS-MNHN-UPMC, Museum National d’Histoire Naturelle, Paris, France
    Contribution
    JSS, Resources, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  11. Pia A Viglietti

    Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, Johannesburg, South Africa
    Contribution
    PAV, Resources, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  12. Xu Wang

    Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
    Contribution
    XW, Resources, Data curation, Formal analysis, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  13. Christophe Lécuyer

    1. Univ Lyon, Université Lyon 1, Ens de Lyon, CNRS, UMR 5276 LGL-TPE, Villeurbanne, France, France
    2. Institut Universitaire de France, Paris, France
    Contribution
    CL, Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Project administration, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

Funding

Institut National des Sciences de l'Univers, Centre National de la Recherche Scientifique

  • Romain Amiot

Palaeontological Scientific Trust

  • Bruce S Rubidge

National Research Foundation

  • Bruce S Rubidge

Centre of Excellence in Palaeosciences

  • Bruce S Rubidge

National Basic Research Program of China (2012CB821900)

  • Romain Amiot
  • Xu Wang

National Basic Research Program of China (2012CB821902)

  • Jun Liu

Institut Universitaire de France

  • Christophe Lécuyer

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors thank the MNHN, the IVPP, Iziko SA Museum and the ESI for granting access to the fossils, and the South African Heritage Resources Agency (SAHRA) for their authorizations to sample and export fossils for stable isotope analysis (PermitID: 1858). We also acknowledge S Jiquel (ISEM) for the access to the Ellenberger’s collection of the University Montpellier 2, J Falconnet (MNHN) for his identification, L. Cui (IGGCAS) for laboratory help and J Cubo (UPMC) for constructed discussions that improved the manuscript. This work was supported by a French project INSU of the CNRS, the Palaeontological Scientific Trust (PAST) and its Scatterlings of Africa programmes, National Research Foundation (NRF) and DST/NRF Centre of Excellence in Palaeosciences. AR and WX were supported by the National Basic Research Program of China (2012CB821900), and LJ by the National Basic Research Program of China (2012CB821902). We thank the four reviewers, K Angielczyk, J Eiler and two anonymous ones, for their constructive comments.

Reviewing Editor

  1. Diethard Tautz, Max-Planck Institute for Evolutionary Biology, Germany

Publication history

  1. Received: May 12, 2017
  2. Accepted: June 20, 2017
  3. Version of Record published: July 18, 2017 (version 1)

Copyright

© 2017, Rey et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,938
    Page views
  • 422
    Downloads
  • 30
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Kévin Rey
  2. Romain Amiot
  3. François Fourel
  4. Fernando Abdala
  5. Frédéric Fluteau
  6. Nour-Eddine Jalil
  7. Jun Liu
  8. Bruce S Rubidge
  9. Roger MH Smith
  10. J Sébastien Steyer
  11. Pia A Viglietti
  12. Xu Wang
  13. Christophe Lécuyer
(2017)
Oxygen isotopes suggest elevated thermometabolism within multiple Permo-Triassic therapsid clades
eLife 6:e28589.
https://doi.org/10.7554/eLife.28589

Categories and tags