There are some controversies regarding the different centers/elements that build the basal area of the skull from back to front (including those settled along the sagittal plane and those lateral to them) in mammals, either in terms/names or in their presence/absence.
The basioccipital develops from the parachordal [4] and is the most posterior element of the floor of the skull. Laterally, it is flanked by the auditory region [6], and posterolaterally, it contacts the exoccipitals. The basisphenoid ossifies from the hypophyseal cartilage [4], anterior to the basioccipital at the midline base of the skull (= basisphenoid body sensu [27]), and laterally contacts the alisphenoids (basisphenoid wings, sensu [29 − 27]). Anteriorly, the basisphenoid will contact the orbitosphenoids whenever they join at the midline.
For the smallest prenatal specimen of Dasypus hybridus (TCL = 15.7 mm), the elements of the skull base (i.e., basioccipital, exoccipitals, basisphenoid, aliesphenoids, or orbitosphenoids) were not ossified, unlike the entire anterior and palatal portions of the skull (rostral elements). In the next prenatal specimen in our series (LTC = 19.9 mm), the basioccipital, basisphenoid, alisphenoid, and orbitosphenoid regions were ossified (but not necessarily in contact). In C. villosus and C. vellerosus, the sequence of ossification for these four bones is as follows: the alisphenoids ossify first, followed by the basioccipital and basisphenoid, and finally the orbitosphenoids. Although our ossification sequence on D. hybridus for these bones is poor, Hautier et al. [30] showed a more complete sequence for Dasypus novemcinctus, where the basioccipital ossified first, followed by the alisphenoids and orbitosphenoids, and finally the basisphenoid. The variability in the sequence of ossification among different clades of armadillos reported here suggests greater variability in the skull development of Xenarthra than previously hypothesized by Hautier et al. [30].
The presphenoid vs. orbitosphenoid alternatives
In mammals, the presphenoid appears as an ossification that arises from a single center and is placed in the midline of the neurocranial base, anterior to the basisphenoid, and between the orbitosphenoids [6]. The homology of the presphenoid is controversial in the case of Cingulata; according to Broom [23], a presphenoid is well ossified in Dasypus novemcinctus and uninterruptedly continuous into the mesethmoid region, so a suture between the presphenoid and mesethmoid bones is absent. Furthermore, the relationship (sutures, fusion) between the presphenoid and the orbitosphenoids is not described by this author. Ferigolo [8], studying two young and one adult specimen of Euphractus sexcinctus, reported that the contralateral orbitosphenoids fuse in the sagittal plane and fill the space left by the absence of the ossification of the presphenoid (Fig. S7). In the older adult specimen analyzed by him, the skull base showed that all the axial elements were fused. He concluded that the cranial base of the armadillos consists of four elements, from caudal to rostral: the basioccipital, the basisphenoid, a “presphenoid” (formed by the medial union of the orbitosphenoids), and the mesethmoid. Finally, Wible and Gaudin [6] described the skull of E. sexcinctus in detail using an ontogenetic series that included a fetus, an immature specimen, and several adult specimens. They named “presphenoid” to the bone anterior to the basisphenoid in the adult skull and with a visible suture between them in some specimens. Nevertheless, they failed to identify an ossification center in the midline of the skull base, anterior to the basephenoid and agreed with Ferigolo’s statement on the replacement of the presphenoid by the orbitosphenoids.
Our findings indicate that the presphenoid is not ossified in D. hybridus, C. villosus or C. vellerosus; moreover, there is no clear evidence that this bone ossifies in other cingulates. The absence of this bone, which was replaced by a medial fusion of the expanded orbitosphenoids, agrees with Ferigolo´s observations [8] on E. sexcinctus. In this regard, the supposed basisphenoid-presphenoid suture, traditionally used to estimate age in several groups of mammals, including armadillos [31], is absent in this group, and a basisphenoid-orbitosphenoid suture should replace it since this suture bears no homology with the basisphenoid-presphenoid suture present in other mammals with a true presphenoid bone.
Mesethmoid
This bone is described as the perpendicular plate to the cribriform plate and is also called the perpendicular ethmoid [14, 16, 8] or ethmoid [6]. In a detailed description of a series of embryos of D. hybridus, Parker [14] reported a cartilaginous structure as a perpendicular ethmoid (= mesethmoid) and, more posteriorly, a large mass of cartilage at the root of the presumptive orbitosphenoids, corresponding to the presumptive presphenoid. In its schemes, the two elements, perpendicular ethmoid (= mesethmoid) and presphenoid, are pointed out with no clear boundary between them (Fig. S8). Based on the presence/absence of the mesethmoid, Broom [23, 32] proposed a classification of mammals. On one hand, the Neotherida has the mesethmoid as the most anterior part of the basicranial axis, hence the skull base is formed by four bones, caudal to rostral: the basioccipital, the basisphenoid, the presphenoid, and the mesethmoid. Tubulidentata, Hyracoidea, Macroscelidea, Afrosoricida, Pholidota, Carnivora, Chiroptera, Rodentia, and Primates are all members of this group. The Paleotherida, on the other hand, have skulls without a mesethmoid, and the base is completed by three bones: the basioccipital, the basisphenoid, and the presphenoid. Monotremata, Marsupialia, Xenarthra, Proboscidea, Sirenia, Artiodactyla, and Perissodactyla [32] belong to this group. In the case of Sirenia, Broom [23] included them within Paleotherida, but when describing the condition observed in the specimens, he stated: “In the Royal College of Surgeons’ Museum there is the skull of a young Dugong No. 2695 skull of a young male. Concerning this skull, Mr. Burne writes, “The bones are too much consolidated to show any original separations. As it is there is no indication of separation between the presphenoid and mesethmoid”. According to his description, he cannot assert whether the mesethmoid is present or absent. Regarding Xenarthra, as Broom failed to identify an independent ossification center that can be considered a mesethmoid bone, he suggested that the presphenoid bone occupied the topographic place of the mesethmoid (Fig. S9). In a dry skull of a young E. sexcintus, Ferigolo [8] observed an independent anterior ossification to the orbitosphenoids and perpendicular to the cribriform plate, which he regarded as a mesethmoid. He considered that the specimens observed by Broom [23, 32] were more advanced and that the mesethmoid was already fused at the base of the skull, making it difficult to individualize. No ossification center equivalent to the mesethmoid bone was present in our prenatal specimens of Dasypus hybridus, Chaetophractus villosus or C. vellerosus (Fig. 1a, and S6a). Nevertheless, in postnatal specimens of D. hybridus and C. vellerosus, we can identify a mesethmoid bone, which ossifies in the most posterior part of the nasal septum medial to the ethmoturbinals. We probably lack an adequate stage of C. villosus in our series to confirm the presence of a mesethmoid in this species. Squarcia et al. [29], when examining three litters (8, 9, and 60 days old), reported a mesethmoid bone as a single ossification center placed rostral to the two orbitosphenoids that fuse in the midline and are separated from them by cartilage. The mesethmoid we can see in later stages (postnatal) was observed as a medial ossification posterior to the nasal septum and medial to the cribriform plate and flanked by the ethmoturbinals (Fig. 1b, 1c, 2a, 2b, S6a; S6b, S10, S11). Our observations agree with those of Ferigolo [8] in Euphractus sexcintus and extend the presence of this characteristic to other members of Cingulata from both clades, Dasypodidae (D. hybridus) and Chlamyphoridae (C. villosus and C. vellerosus).
The Xenartha in the context of Eutheria
The last comprehensive classifications of extant Eutherians proposed four large clades: Afrotherians, Xenarthrans, Euarchontoglires, and Laurasiatherians [33, 34, 35]. The Afrotherians and Xenarthrans are considered sister groups and together form the clade Atlantogenata [36], which is the sister clade to Boreoeutheria (Laurasiatheria plus Euarchontoglires). Currently, the Afrotherians are divided into Paenungulata with the proboscideans, sirenids, and hyracoids and the Afroinsectiphilia that group the Tubulidentata, Marcroselidea, and Afrosoricida [35].
According to Broom’s studies, several now considered Afrotherians to have a mesethmoid (Tubulidentata, Hyracoidea, Afrosoricida, except Chrysochloris, and Macroscelidea), while only Proboscideans, Sirenids and Chrysochloris lack this bone. As stated above, the information provided by Broom [23] on Sirenids is not clear, and we were not able to find more recent information on the presence/absence of a mesethmoid in this group. By analyzing two cingulates (the armadillos Euphractus sexcinctus and Dasypus novemcinctus) and one pilosan (Pigmy Anteater Cyclopes didactylus), Broom suggested the absence of the mesethmoid in the clade. Ferigolo [8], and our contribution, show a mesethmoid present among xenarthran armadillos (Cingulata), implying that a mesethmoid is widely present in the Atlantogenata clade, as in most groups of Boreoeutheria (the rest of Eutheria). This leads to the assumption that the mesethmoid is a basal characteristic of all eutherians and that secondary loss occurs in some clades. Furthermore, in marsupials, a mesethmoid is present in Monodelphis, and the phylogenetic reconstruction suggests an independent acquisition of this character in this taxon. Nevertheless, considering that Monodelphis is a living representative of more primitive marsupials and that the number of marsupial species analyzed is relatively low, the presence of a mesethmoid within the clade might be underestimated. Further information on an increasing number of marsupial taxa might indicate that the presence of a mesethmoid is basal to all therian mammals, and secondary loss also occurs in most marsupials. Further studies are also needed to confirm the presence/absence of a mesethmoid in some less explored eutherian groups, and those clades considered to lack a mesethmoid (e.g., other pilosans among Xenarthra) might need reassessment by adding more stages in their ontogenetic series of both prenatal and postnatal stages, especially the latter. In this regard, the mesethmoid may appear late in ontogeny, as may the ossification of the presphenoid or the fusion of the orbitosphenoids. Therefore, if a taxon is analyzed only based on few early stages, this might wrongly lead to the assumption that ossification centers are absent, while they might be present (or substituted by others in position) in advanced ontogenetic ones.
The environments in which mammals inhabit model or induce the way of life and the strategies necessary to survive. In addition, different strategies can be observed at birth since the most important morphological changes occur in this period [37]. In mammals, the degree of morphological and neonatal maturity [38, 39, 37] is defined as an altricial-precocial spectrum, where the altricial end is represented by offspring that are born hairless, with eyes closed, without body temperature control, and that need to be fed; for this reason, they are sheltered in caves. On the contrary, the precocial ones are provided with hair, open eyes and are more independent, capable of rapidly facing the environment on their own. In xenarthran cingulates, the Dasypodidae, represented by Dasypus, are presumably precocial due to the advanced development of their postcranium at the time of birth and the complete ossification of the osteoderms of the integument [40]. In contrast, Chaetophractus (Chlamyphoridae) is considered altricial when compared to Dasypus (Dasypodidae), as the postcranium and integument are less advanced at birth [40]. The early appearance of the mesethmoid in newborns of Dasypus hybridus and the subsequent ossification of this bone in advanced postnatal stages of Chaetophractus vellerosus is consistent with the assumption of a precocial strategy for Dasypus and an altricial strategy for Chaetophractus. This suggests that when looking for an ossification as the mesethmoid, which appears late in the cranial ossification sequence, the specimens examined should be in later, more mature stages, especially in those that present altricial-type strategies.
In summary, our study supports the existence of a postnatal mesethmoid ossification in armadillos; the presence of a “presphenoid” formed from the bilateral sagittal integration of the orbitosphenoids and not from an ossification center of its own; reinforces the idea of a different ontogenetic trajectory in armadillos at the cranial level, precocial for Dasypus and altricial for Chaetophractus; proposes that the presence of the mesetmoid is primitive for Eutheria and that it was lost independently in minor clades; and highlights that most likely the different interpretations about the presence or absence of certain centers/bones of the skull base are the result of analyzing specimens in different ontogenetic states and an analogous assignment of elements.