Reptile-like physiology in Early Jurassic stem-mammals

Despite considerable advances in knowledge of the anatomy, ecology and evolution of early mammals, far less is known about their physiology. Evidence is contradictory concerning the timing and fossil groups in which mammalian endothermy arose. To determine the state of metabolic evolution in two of the earliest stem-mammals, the Early Jurassic Morganucodon and Kuehneotherium, we use separate proxies for basal and maximum metabolic rate. Here we report, using synchrotron X-ray tomographic imaging of incremental tooth cementum, that they had maximum lifespans considerably longer than comparably sized living mammals, but similar to those of reptiles, and so they likely had reptilian-level basal metabolic rates. Measurements of femoral nutrient foramina show Morganucodon had blood flow rates intermediate between living mammals and reptiles, suggesting maximum metabolic rates increased evolutionarily before basal metabolic rates. Stem mammals lacked the elevated endothermic metabolism of living mammals, highlighting the mosaic nature of mammalian physiological evolution.


Supplementary Tables
Supplementary Table 1 | Results of quantitative analyses of precision between three independent observers for the cementum increment counts of fossil mammals studied here, Morganucodon and Kuehneotherium, and those of ten previous studies of extant mammals with comparable age ranges. CV -coefficient of variation.  1 Procyon lotor 54 9 20.6 Gasawey et al. 2 Alces alces 72 9 14.2 Klevezal and Pucek 3 Bison bonasus 45 21 20.57 Kay and Cant 4 Macaca mulatta 65 24 29.34 Cederlund et al. 5 Capreolus capreolus 74 9 30.5 Bodkin et al. 6 Enhydra lutris 14 14 26.24 London et al. 7 Canis lupus 12 6.8 25.92 Christensen-Dalsgaard et al. 8 Ursus maritimus 32 15 15.2 Pasda 9 Rangifer tarandus 63 16 19.4 Perez-Barberia et al. 10 Cervus elaphus 164 17 16.22 extant mammal taxa and Morganucodon, with resulting estimates of blood flow index Qi. The phylogenetic relationship of Morganucodon and Kuehneotherium is shown in Fig The fissure material is disarticulated, and individual elements are usually broken or cracked during deposition and preservation. However, the individual preservation is generally good in Pontalun 3, and the material has been prepared without acid. The first stage of the project, therefore, focused on Morganucodon watsoni from this fissure, particularly the lower second molars, as they have large roots and are easily identified and scanned. The project was then expanded to include other Morganucodon isolated teeth and dentulous specimens from Pontalun 3. Morganucodon dentary specimens are more likely to retain teeth than those of Kuehneotherium, due to the expanded root apices in the Morganucodon ultimate premolars and the molars. The Morganucodon dentulous specimens proved to have the best preservation of the cementum in general, especially when the crowns were intact at the time of fossilisation, possibly due to the cementum and root canal not being exposed to taphonomic processes. Although all the Morganucodon specimens are from the same fissure, they cannot be regarded as in any way a cohort population, as the specimens occurred scattered throughout the large fissure in small pockets, and were probably washed in from the surrounding area during rain storms following forest fires 17,18 .

Supplementary Note 3: Eruption sequence and timing
As noted in the main text, establishing individual ages from several teeth along Morganucodon lower jaws provides information on the tooth eruption sequence and timing.
Eight scanned dentaries provided information (Table 1). There are several specimens with tooth row increment counts covering p1 to m2: For example, NHMUK PV M 95790 for p1 to p3; NHMUK PV M96413 for p3 to m1; NHMUK PV M 96396 for p4 to m2 (Table 1) One specimen (NHMUK PV M 95790) has cementum increment counts indicating that the permanent final incisor i4, and the canine, erupted during the year after the eruption of the first three permanent premolars. We therefore conclude that the ultimate incisor, canine and the third molar erupted during the year following the eruption of p1 to m2 in Morganucodon.
We do not have information on eruption timing of more anterior incisors, or the fourth molar.
The fifth molar is only very occasionally present.
The eruption of the final incisor and canine after the premolars may be a plesiomorphic feature in Morganucodon and does not follow the general antero -posterior pattern seen in later therians. In the Late Jurassic dryolestids 22 , tooth replacement takes place in two waves; with i2, i4, p1, and p3 in the first series and i1, i3, c, p2, and p4 in the second.
However, this is also different from the pattern seen here in Morganucodon, where the ultimate incisor and canine come in after the permanent premolars.
Unfortunately, there are no available tooth rows of Kuehneotherium with cementum increment counts. Dentulous specimens are rare and the three specimens in the UMZC collection are not suitable for scanning for cementum annuli, as they were previously mounted in plaster and dissected to reveal the tooth roots. Kuehneotherium has a longer tooth row than Morganucodon with six premolars and six molars 23

. A difference from
Morganucodon is that there is evidence from specimens of juvenile Kuehneotherium dentaries 23 that the canine, ultimate premolar, p6, and third molar all erupted at about the same time, which suggests that the canine erupted relatively earlier in Kuehneotherium than in Morganucodon.
The loss of the anterior permanent postcanines has been noted in Morganucodon jaws 24 , and is assumed to proceed along the tooth row with age. In some dentaries, the first premolar appears to be lost relatively early and is only represented by a small, indistinct, resorbed root just posterior to the canine. Accordingly, it is not certain if the timing, and degree, of loss of the anterior premolars is variable in individuals, but this question is part of a current study. Specifically, the specimen NHMUK PV M 95790 with cementum annuli counts for i4 to p3, obviously has not shed the anterior premolars, despite a minimum age of eight years.

Supplementary Note 4: Comparisons between fossil lifespan estimates and captive lifespans of extant taxa
In our primary study, we compare the lifespan estimates for our fossil taxa with maximum known wild lifespans for extant mammals and reptiles. Environmental and ecological pressures upon wild populations mean that wild individuals rarely live until the maximum possible age for a taxon, and this is shown when we compare the maximum known wild lifespan and the maximum known captive lifespan for individual mammal (n = 244) and reptile (n = 68) taxa ( Supplementary Fig. 4). ANCOVA comparisons show that the wild and captive regression slopes are significantly different for both mammals (p<0.001) and reptiles (p = 0.048). Similar to previous studies 25, 26 , we find that small bodied taxa, both mammals and reptiles, show larger relative per-taxon lifespan increases in captivity compared to larger taxa ( Supplementary Fig. 4 Supplementary Fig. 5b).When this data is PGLS regressed against body mass, the fossil mammaliaforms plot outside the living mammal confidence and prediction intervals, within the living reptile prediction intervals, and are distributed either side of the reptilian regression line, close to the confidence interval lines (Supplementary Fig. 5b).
When this data is PGLS regressed against body mass, all estimates for the fossil mammaliaforms fall outside of prediction intervals for the the mammal data, the majority fall within the reptile prediction intervals, and all are further below the mammal regression line than any extant mammal below approximately 8kg ( Supplementary Fig. 5c).
In summary, whether using their raw/wild or adjusted 'captive' lifespan estimates, our fossil mammaliaforms are consistently predicted to have msSMR and K values lower than comparable extant endothermic mammals but within the range of values of extant ectothermic reptiles.