On taxonomic issues, ontogenetic series and tooth replacement. Comments on Diphyodont tooth replacement of Brasilodon—A late Triassic eucynodont that challenges the time of origin of mammals by Cabreira et al.
Recently, Cabreira et al. (2022) published a contribution analysing the dental replacement of Brasilodon, a cynodont from the Brazilian Norian, concluding that it had a mammalian (diphyodont) type of replacement. In this note, we will only deal tangentially with their main conclusion and instead will concentrate on criticism posed by Cabreira et al. (2022) to the contributions of Abdala et al. (2013) and Norton et al. (2020).
Cabreira et al. (2022) made the criticism to our research in their Introduction (section 1.1; p. 1425) and in section 4.6 Debate about the dental anatomy of the mammaliamorph Thrinaxodon in the Discussion (p. 1437).
The criticisms of our contributions presented by Cabreira et al. (2022) are reproduced below. Numbers in square brackets have been added to denote our responses to the different points raised by Cabreira et al. (2022). Text in italics is a direct quote from the paper of Cabreira et al. (2022).
Recently, micro-computed tomography (μCT) scanning techniques were employed to image and study the intra-mandibular teeth of Thrinaxodon and the nature of the dentition of extinct cynodonts (Abdala et al., 2013; Norton et al., 2020). [1] However, the rendered virtual sections provided do not permit the visualisation of the dentine-enamel junction and other critical histological enamel characters, also none of the details of bone resorption and rhizolysis. [2] Furthermore, the purported ‘ontogenetic series’ utilised by Abdala et al. (2013) or Norton et al. (2020) is made up of specimens that may reflect sexual dimorphism or even represent coeval species. [3] In palaeodontological studies, it is not the size of the skull that is the key factor determining the animal's ontogenetic stage, but the sequence of odontogenic development for each particular dental locus. [4] All the specimens of Thrinaxodon studied by Abdala et al. (2013), present a maximum number of eight post-canines and the replacement teeth are smaller. It is therefore clear that in Thrinaxodon, the dental replacement occurred under negative allometry, just the opposite of the ‘reptilian’ pattern proposed by those authors. [5] In other words, the radiographic evidence presented by Abdala et al. (2013) points to another diagnosis than the one with which they conclude. [6] Also, the number of post-canines (8) present in all the small, medium size and large specimens in their study curiously coincide with the hypothesis defended by Luckett (1993) and Butler (1995) for basal mammals. [7] Furthermore, in Thrinaxodon, teeth are substituted directly from underneath the deciduous ones and in a caudal-to-rostral direction, except for the upper canines (Abdala et al., 2013). Contrary to what is presented in that publication, the dental pattern presented by Thrinaxodon is in fact, typical of modern eutherians and diphyodont animals. [8] The exclusion or unavailability of very small and very large specimens of Thrinaxodon for their studies precludes information on neonatal and fully adult forms (Cabreira et al., 2022, p. 2).
In the Discussion (section 4.6) of the paper, Cabreira et al. (2022:1437) used the exact wording but added two final sentences: [9] Nievelt and van Smith (2005) provide a comprehensive discussion of the wide range of diphyodonty variations amongst marsupials and placental mammals. Brasilodon fits within that variation.
The following are our responses to the points raised by Cabreira et al. (2022):
[1] However, the rendered virtual sections provided do not permit the visualisation of the dentine-enamel junction and other critical histological enamel characters, also none of the details of bone resorption and rhizolysis.
That is correct, those X-ray μCT data do not permit the visualization of histological enamel characters. However, our μCT data show dental replacement teeth below functional ones, many of them with the typical crown of each species (see point [2] for a discussion related to species). We can assume that the replacement teeth “will replace” the functional ones, even though the resolution of the μCT data does not show the high-resolution histological features that Cabreira et al. (2022) consider necessary to be sure that a tooth will be replaced. In several instances, there is evidence of erosion on the lingual surface of the functional tooth, due to a replacement tooth developing lingual to the functional tooth and causing resorption of the root. Computed tomography techniques have become more frequently used in palaeontology because they are non-destructive (e.g. Chang et al., 2021; D'Emic et al., 2013; Duhamel et al., 2021; He et al., 2018; Olroyd et al., 2021; Pusch et al., 2019, 2020; Snyder et al., 2020), and we considered that the data necessary to analyse dental replacement in our material by using this technique was sufficient. It is important to mention here that we have high-resolution sub-micron synchrotron X-ray μCT data of the dentition of Thrinaxodon where the dentine-enamel junction, among other features, is clearly visible. We did not use the palaeohistological information because it was irrelevant for our purpose: the detailed description of the external morphology and variation of dentition in an ontogenetic series, and the dental replacement that was readily apparent without having to destroy the study specimens.
[2] Furthermore, the purported ‘ontogenetic series’ utilised by Abdala et al. (2013) or Norton et al. (2020) is made up of specimens that may reflect sexual dimorphism or even represent coeval species.
There is a possibility that Cabreira et al. (2022) are correct. We do not know much about the relationship between sexual dimorphism and dental replacement, but in general, there does not seem to be much difference in dental replacement between males and females in most species of mammals. Eventually, in dry skulls, dental replacement processes represented in small specimens are more likely to be of females, if sexual dimorphism is represented by size differences, but there is no alteration of the pattern of dental replacement except, of course, in teratological cases. This suggests that the replacement pattern will not be affected by sex, but at this stage we, as well as Cabreira et al. (2022), do not have access to that type of information from our fossils. In Abdala et al. (2013) we had to decide if we were dealing with one species or several (maximum of 54 different species, following table 2 of our paper), if we take into consideration the criticism of Cabreira et al. (2022) for coeval species. Unfortunately, there is no final objectivity in taxonomic decisions. However, for the past 50 years (following the seminal cynodont classification work by Hopson & Kitching, 1972), no researcher working on non-mammaliaform cynodont taxonomy and systematics has proposed the presence of more than one species of Thrinaxodon and Galesaurus in the South African Karoo. The concept of species in palaeontology is based on morphology and is decided by specialists working on the group. Thrinaxodon is one of the best-recorded, and most-studied cynodonts from the Karoo of South Africa with hundreds of specimens, and our sample included the smallest to the largest identified specimens housed in the collections of numerous institutions (see Abdala et al., 2013: figure 1). Cabreira et al. (2022) fail to recognize the thorough studies of cranial morphology and ontogeny in both Thrinaxodon (Jasinoski et al., 2015) and Galesaurus (Jasinoski & Abdala, 2017a) that have been undertaken recently. There are also several instances of mixed age/size aggregations of Thrinaxodon and Galesaurus (Jasinoski & Abdala, 2017b) which enable confident taxonomic identification of individuals across several ontogenetic stages.
Studies conducted by Botha and Chinsamy (2005) on the osteohistology of Thrinaxodon liorhinus, using individuals with different basal skull lengths, demonstrated that the material used formed an ontogenetic series. Juvenile stages are represented by a highly vascularized woven-parallel complex (woven bone associated with primary osteons) and the absence of growth marks (annual temporary decreases or cessations in growth). Subadults rarely contain a single growth mark indicating that at least 1 year had passed, and older (larger) individuals show an overall transition from the rapidly forming woven-parallel complex to the slower-growing, poorly vascularized parallel-fibred bone. This transition may indicate the onset of reproductive maturity. The material in the Botha and Chinsamy (2005) study covers all the sizes used in the samples of Abdala et al. (2013) and Jasinoski et al. (2015), indicating that these studies did indeed have an ontogenetic series for studying the cranial ontogeny and dental replacement in Thrinaxodon liorhinus. An analysis of the bone tissue patterns of Galesaurus planiceps by Butler et al. (2018) showed similar patterns to Thrinaxodon. In their study, the small specimen RC 845 represents an early subadult at 61% of the maximum known size for the taxon. The bone tissues of RC 845 reveal a highly vascularized woven-parallel complex with a possible growth mark at the periphery suggesting that this specimen may have just reached its first year at the time of death. Larger specimens indicate subadult status with the bone tissues showing slightly less vascularization (and thus slower growth) and a single growth mark. One of the largest specimens (NMQR 3542) which represents 89% of the maximum known size of this taxon, is considered to fall within the adult range based on the transition from the rapidly forming woven-parallel complex to the slower-forming parallel-fibred bone. Additionally, some of the elements in this specimen contain an External Fundamental System, which is a poorly vascularized or avascular peripheral region consisting of many closely spaced Lines of Arrested Growth (temporary cessations in growth). This feature indicates that this individual was an adult. The specimens used in the Butler et al. (2018) study cover the size range used in the Norton et al. (2020) study indicating that the latter work did indeed use a growth series to study the tooth replacement patterns in Galesaurus.
Currently, there are no Triassic cynodont specialists who recognise the presence of several species of Thrinaxodon or Galesaurus in South Africa. Of course, and again, this is a generally accepted hypothesis, and it is as subjective as considering that the three specimens of Brasilodon presented in Cabreira et al. (2022) represent only one species.
It is then important to remember, before making this kind of critique, that practical taxonomy with character analyses continue to be the method used to recognise fossil taxa (and also many living ones). The proposition of multiple species in a sample is always a possibility, but it is important to at least take recent taxonomic research into consideration to make an informed critique.
[3] In palaeodontological studies, it is not the size of the skull that is the key factor determining the animal's ontogenetic stage, but the sequence of odontogenic development for each particular dental locus.
This affirmation requires citation of at least some of those palaeodontological studies, but none are given. In palaeontological–neontological studies of ontogeny, which also include dentition (and we have expertise in this; e.g. Abdala & Giannini, 2000; Abdala et al., 2001; Flores et al., 2006; Jasinoski et al., 2015; Jasinoski & Abdala, 2017a; Flores et al., 2022), skull size is certainly a very important factor, as are also the changes in allometry perceived from juvenile to adults. Such changes are expected: for example, negative allometry for braincase and sense organs and positive allometry for the splanchnocranium. Furthermore, the ontogenetic stages of Thrinaxodon (Jasinoski et al., 2015) and Galesaurus (Jasinoski & Abdala, 2017a) are not defined only by the length of the skull, but rather by morphological features observed in the skull (also corroborated by osteohistological analyses; see point [2]). For example, see table 6 of Jasinoski et al. (2015), and table 5 of Jasinoski and Abdala (2017a), which list 12 craniomandibular characters used to differentiate the four ontogenetic stages in Thrinaxodon, and eight craniomandibular characters used to differentiate the three ontogenetic stages in Galesaurus, respectively.
[4] All the specimens of Thrinaxodon studied by Abdala et al. (2013), present a maximum number of eight post-canines and the replacement teeth are smaller. It is therefore clear that in Thrinaxodon, the dental replacement occurred under negative allometry, just the opposite of the ‘reptilian’ pattern proposed by those authors.
We are not sure why Cabreira et al. (2022) assume that replacement teeth in Thrinaxodon are smaller. We did not have many instances to observe this in our sample because the development of the replacements was at different stages, but in a couple of examples reported by Abdala et al. (2013: figure 3), it was clear that the replacement crown was larger than that of the functional tooth it was replacing. In addition, the information presented in figure 15 and table 5 of Abdala et al. (2013) clearly indicate an increase in the length of the postcanine crown from the smallest to largest specimens included in the μCT sample. Therefore, the assumption of dental negative allometry by Cabreira et al. (2022) is not substantiated.
[5] In other words, the radiographic evidence presented by Abdala et al. (2013) points to another diagnosis than the one with which they conclude.
We are not sure how to reply to this as it is a confusing statement. Perhaps Cabreira et al. (2022) can explain further what they mean by the term “diagnosis” in this context.
[6] Also, the number of post-canines (8) present in all the small, medium size and large specimens in their study curiously coincide with the hypothesis defended by Luckett (1993) and Butler (1995) for basal mammals.
We direct Cabreira et al. (2022) to table 3 of Abdala et al. (2013) which reflects great variation in the number of postcanines. Thus, to say eight postcanines are present “in all” the specimens is an oversimplification of the published data. In fact, the conclusion of Abdala et al. (2013) indicates that the average adult dental formula for Thrinaxodon is 4/3, 1/1, 6/7–8.
Cabreira et al. (2022) appear to base their argument on the section “Primitive number of premolars in Eutheria” of Luckett (1993, pp. 197–198). Notably, Luckett (1993) ascribes the “primitive eutherian dental formula” to be four premolars and three molars (i.e. seven postcanine teeth in total) in each jaw quadrant (not eight as cited in Cabreira et al., 2022), whereas the “ancestral eutherian pattern” included five premolars (i.e. a total of eight postcanine teeth per jaw quadrant). Considering the recent redescription of the dentition of the basal-most probainognathid cynodont Lumkuia (Benoit et al., 2022), we could argue that Lumkuia clearly exhibits Luckett's (1993) “primitive eutherian dental formula” as it has seven upper and six lower postcanines. We could extrapolate even further, and infer evidence of diphyodonty in Lumkuia, given the almost complete lack of evidence for postcanine tooth replacement preserved in the single known individual, which has been inferred to represent a subadult (Benoit et al., 2022). This would push back the purported origin of diphyodonty from the Norian (Late Triassic, ~227 to 208.5 Ma) of Brazil to the Anisian (Middle Triassic, ~247.2 to 242 Ma) of South Africa (and see also point [9]).
Cabreira et al. (2022) focus on the importance of there being eight postcanines in the smallest Thrinaxodon according to Butler (1995, p. 28). However, it must be noted that Butler (1995) cites only the studies of Parrington (1936) and Osborn and Crompton (1973), omitting numerous other studies of tooth replacement in Thrinaxodon that would have been available to him at the time. Most notable is the absence of citations to Crompton (1963) and Gow (1985) in Butler's (1995) work (these references are also absent from Cabreira et al., 2022). Gow's (1985) research in particular focused on several specimens of Thrinaxodon that were much smaller than those included in previous studies (see Abdala et al., 2013, table 1). In these smallest specimens (BSL ~30 mm), the postcanine count is 6–8 upper postcanines and 7–8 lower postcanines (Abdala et al., 2013, table 3). The sole use of the contributions of Luckett (1993) and Butler (1995) to discuss the primitive dental formula of eutherians is out of date. In recent years, the discovery of several new Mesozoic therian mammals has enabled more in-depth discussion on this issue (e.g. Bi et al., 2018; Hu et al., 2010; O'Leary et al., 2013; and references cited therein); but these are ignored by Cabreira et al. (2022).
[7] Furthermore, in Thrinaxodon, teeth are substituted directly from underneath the deciduous ones and in a caudal-to-rostral direction, except for the upper canines (Abdala et al., 2013). Contrary to what is presented in that publication, the dental pattern presented by Thrinaxodon is in fact, typical of modern eutherians and diphyodont animals.
Here Cabreira et al. (2022) are oversimplifying the process of dental replacement in Thrinaxodon and Galesaurus, and basing their observations on the two-dimensional reproductions of the three-dimensional reconstructions provided by Abdala et al. (2013). A cursory glance at the virtual thin sections presented by Abdala et al. (2013) clearly shows the lingual eruption of the replacement incisors and canines (e.g. figures 6 and 9). Similarly, the resorption pits evident on the lingual faces of several postcanines (e.g. figure 7a,d) demonstrate that the associated replacement postcanines were developing lingual to the functional tooth, as opposed to “directly from underneath” as claimed by Cabreira et al. (2022). This is also true for the positioning of the replacement dentition in Galesaurus, which developed lingual to the functional tooth (Norton et al., 2020). Considering the last phrase in this segment of the critique, the final conclusion of Cabreira et al. (2022) installed in the contribution title (Diphyodont tooth replacement of Brasilodon—a Late Triassic [Norian] eucynodont that challenges the time of origin of mammals), is inexact because they imply the presence of diphyodonty in the Early Triassic (Induan) Thrinaxodon, although we are completely unable to see how they will integrate the information of Abdala et al. (2013) to prove their claim. In addition, another feature not mentioned in Cabreira et al. (2022), that was studied in Thrinaxodon by X-ray μCT, is the type of postcanine tooth attachment that goes from gomphosis to ankylosis, differing significantly from the condition of mammaliaforms (LeBlanc et al., 2018) with truly diphyodont replacement (e.g. Luo et al., 2004; Newham et al., 2020), and that adds further support for the type of tooth replacement inferred for Thrinaxodon (Abdala et al., 2013).
[8] The exclusion or unavailability of very small and very large specimens of Thrinaxodon for their studies precludes information on neonatal and fully adult forms.
This is a remarkable thing to say, and quite limiting progress, especially considering that Abdala et al. (2013) assembled the largest quantity of data for the study of dental replacement and cranial ontogeny (Jasinoski et al., 2015) for any Triassic cynodont. The cranial skull length of the sample ranged from 30 mm to 96 mm, we imaged five specimens with X-ray μCT, representing at least four different size classes and studied first-hand 54 specimens, a quantity impossible to put together for similar studies of other species of Triassic cynodont. In addition to having a much larger sample size than Cabreira et al. (2022), the study of Abdala et al. (2013) included detailed descriptions of the replacement in the upper and lower dentition for all five individuals of Thrinaxodon imaged with X-ray μCT. Interestingly, this critique by Cabreira et al. (2022) can be applied to their own study, which only included three specimens. According to their caption of figure 4, the three specimens included in their study represent neonate (UFRGS-PV-0767-T), juvenile (UFRGS-PV-0825-T) and subadult (ULBRA-PVT0284) stages, suggesting that no “fully adult forms” were included in the study. In addition, the relationship of size (e.g. depth of dentary, length of postcanine tooth row, size of tooth at any position, etc.) among the three mentioned specimens is not explained in their work. Cabreira et al. (2022) also disregarded the results presented in the preliminary studies of tooth replacement in Brasilodon/Brasilitherium that included a larger sample of specimens (Martinelli & Bonaparte, 2011), as well as descriptions of multiple replacement teeth associated with a single locus observed through X-ray μCT (Martinelli, 2017; Martinelli et al., 2017, 2019).
[9] Nievelt and van Smith (2005) provide a comprehensive discussion of the wide range of diphyodonty variations amongst marsupials and placental mammals. Brasilodon fits within that variation.
These last two sentences added at the end of the Discussion (section 4.6), have no relationship whatsoever to the Thrinaxodon/Galesaurus dental replacement discussion. We cannot understand why Cabreira et al. (2022) used them to close this section of the Discussion.
We also must stress the title of the Discussion section, which incorrectly states that Thrinaxodon is member of Mammaliamorpha. This group was defined phylogenetically by Rowe (1988:249) as the clade including the last common ancestor of Tritylodontidae and Mammalia. Recently, Abdala (2021) used Mammaliamorpha as a monophyletic group composed of the most recent common ancestor of Homo and Pachygenelus and all its descendants (this lineage includes Brasilodontidae, Tritheledontidae, Tritylodontidae and Mammaliaformes). There is no, and never was, a phylogenetic hypothesis supporting Thrinaxodon as a mammaliamorph, and its placement as an early epicynodont is universally accepted (e.g. Benoit et al., 2022 and phylogenetic studies cited therein).
We celebrate studies that provide an advancement in the knowledge of dental replacement or ontogeny in Triassic cynodonts, particularly those directed to the understanding of the evolution of diphyodonty in mammals. This feature represents a milestone, having a historical use in the diagnosis of different groups of mammals (e.g. it was interpreted as present in the placental ancestor by Gregory, 1910). However, we always expect that such work is produced without misrepresenting the research of other authors, or by forcing interpretations that are not represented in those works.
