Prototherium cf. ausetanum Balaguer & Alba, 2016

Voss, Manja, Hampe, Oliver, Mahlow, Kristin & Vilanova, Joan C., 2023, New findings of Prototherium ausetanum (Mammalia, Pan-Sirenia) from paving stones in Girona (Catalonia, Spain)?, Fossil Record 26 (1), pp. 135-149 : 135

Prototherium cf. ausetanum Balaguer & Alba, 2016

Nomenclatural remark.

The present systematic palaeontology considers the quite recently established clade Pan-Sirenia , O’Leary et al., 2013. Thereby, O’Leary et al. (2013: supplementary material, table 4) apply the total group concept ( Jefferies 1979) and distinguish between Pan-Sirenia as representing the total clade of Sirenia Illiger, 1811, whereas Sirenia is redefined and restricted to the node-based crown group as the least inclusive clade containing all living sirenians, their last common ancestor, and all of its descendants. The same concept is introduced later again by Vélez-Juarbe and Wood (2018). Therefore, priority, and hence reference for Pan-Sirenia , is given here to O’Leary et al. (2013).

Material.

MAC GIR-042375, partial skull and rib remains documented in urban paving stones of Girona, Catalonia, Spain.

Locality.

Limestone quarry south of Sant Vicenç de Castellet, central Catalonia, Spain.

Stratigraphic position.

Late Bartonian, middle Eocene. The age of the find is corrected from early Bartonian ( Voss et al. 2019) to late Bartonian ( Galindo et al. 2003).

Description

The sirenian remains are embedded in a shallow carbonate platform sediment consisting of an alternation of biomicritic limestones, lutites and bluish-grey claystones (Figs 1 , 2 ), and are distributed on two slabs, each measuring 2 cm in thickness. The preserved elements mainly belong to the skull of a single specimen and include parts of the rostrum (premaxilla and maxilla), frontal, vomer, basicranium, and fragments of the zygomatic arch (jugal and squamosal). From the teeth, the upper molars M1-M3 including crowns and roots are preserved. Otherwise, the dentition is indicated by the presence of alveoli. In addition to the macroscopic analysis, CT-scanning and post-processing allow more detailed morphological examination of the preserved skull parts. As already performed in Voss et al. (2019), the slab preserving a relative upper region of the skull roof and snout is designated as slab 1, the other representing a relatively deeper level within the skull is designated as slab 2 (Fig. 3 ).

The first new observation refers to the orientation of the single fossil-bearing slabs. With the help of an adult skull of the extant dugong ( Dugong dugon , ZMB_MAM 69306), the partial skull was aligned in an anatomically correct position mainly considering the slope of both, the jugal and the rostrum. As a result, each slab shows oblique angles in the transversal plane, hence do not lie one-to-one at that specific level. For slab 1, the angles are 28.22° in lateral view, and 4.3° in frontal view; for slab 2, the angles are 28.22° in lateral view, and 6.0° in frontal view, respectively (Fig. 3a, b ). This explains, amongst others, why each slab preserves skull elements from a vertically higher level on the right side and from a somewhat lower level on the left side. The same applies to the antero-posterior orientation of the slabs, which accordingly show skull elements from a lower level anteriorly and cranial parts from a higher level posteriorly. As it will be outlined below, the oblique orientation sometimes impedes the correlation of specific structures between the slabs, which resulted in amendments as to some observations made in Voss et al. (2019).

The distance between both slabs could be ascertained to be 11.4 mm (Fig. 3 ). As such, a total of 50 mm, i.e., 40 mm fossil-bearing stone plus circa 10 mm non-preserved interspace, is the basis for this study. This equals about 50% to 77% of the height of a typical Eocene-Oligocene sirenian skull ( Voss 2013: appendix 3, Measurements of crania: top of supraoccipital to ventral sides of occipital condyles (de)).

Furthermore, the preliminary results by Voss et al. (2019) on the individual age and the taxonomic assignment of the Girona find can be verified.

Voss et al. (2019) determine the Girona specimen as representing at least a young adult based on the advanced state of wear of the M3 and its position at the level of the posterior margin of the zygomatic orbital bridge ( Mitchell 1973). At this point, an amendment to the preceding description in Voss et al. (2019) is necessary. While the position of the M3 is confirmed in macroscopic view and in the CT-scans ( Voss et al. 2019: fig. 6b, c), the position of M1-2 as shown in slab 1 ( Voss et al. 2019: fig. 6a) has to be adjusted. The two visible alveoli in the latter figure belong most certainly to P4-P3 considering the higher anatomical level of slab 1 combined with the transversally non-planar cut of the slabs (Fig. 4a ).

For relative age determination, it is also highly indicative whether sutures between certain bones are open, fused or even obliterated ( Pocock 1940; Mitchell 1973). The contact between the basisphenoid and basioccipital is especially important for the identification of adulthood. Elements of the basicranium are clearly and best visible on the unpolished bottom side of slab 1 (Fig. 4a ) showing parts of the basisphenoid, the spheno-occipital eminences, and the basioccipital that is mainly preserved with its left extension. The anterior ends of the spheno-occipital eminences mark the approximate contact between basisphenoid and basioccipital. Both bones are considered as being firmly fused, since there is no indication of a suture. This is also confirmed in the respective CT-Scan (Fig. 4b ) despite interferences that reduce the scan quality. Consequently, the Girona specimen definitively belongs to an adult individual.

With respect to the taxonomic assignment, Voss et al. (2019) identified the Girona sea cow as a representative of the Eocene genus Prototherium (and finally P. ausetanum ) based on the antemolar dental formula. This is corroborated also in this study and observable in slab 2 (Fig. 5 ). Thereby, Fig. 4a reflects mainly alveoli in macroscopic dorsal view of slab 2 and Fig. 4b shows the corresponding CT-scan with teeth in ventral view of slab 2. As a result, four alveoli for single-rooted premolars (P1-4) can be unambiguously ascertained (Fig. 5a ). The complete number of premolars reflects plesiomorphic conditions and is characteristic for Eocene sirenians (e.g., Voss 2013). The most important feature is the absence of a persistent DP5, a molariform upper deciduous fifth premolar, which is a peculiarity in the dental formula of Prototherium (e.g., Domning 1994; Bizzotto 2005; Balaguer and Alba 2016). While also the lack of DP5 can be deduced directly from the preserved maxillary parts on the slabs, the presence and absence of alveoli is hardly observable in the corresponding CT-Scan even after careful post-processing and evaluation of the ventral side of the slab (Fig. 5b ). This is primarily based on the fact that the contrast between sirenian bone and surrounding matrix is relatively low. Additionally, inclusions of iron-rich minerals cause harmful interferences.

Nevertheless, the present results support the Girona specimen as belonging to the genus Prototherium . As such, its identification on the species level focusses on comparisons with three Prototherium species generally considered valid: P. veronense De Zigno, 1875 from the late middle Eocene (Bartonian) of Verona (Italy), P. intermedium Bizzotto, 1983 from the late Eocene (Priabonian) of Possagno (north-east from Verona, Italy), and the recently established P. ausetanum Balaguer & Alba, 2016 from the Bartonian of Tona (Catalonia, Spain). Additionally, the taxonomic-morphological comparisons are complemented by two further Prototherium species from the Bartonian of Catalonia, P. solei Pilleri et al., 1989 and P. montserratense Pilleri et al., 1989. As it will be discussed later, their status and affinities are uncertain (e.g., Domning 1996; Bizzotto 2005; Balaguer and Alba 2016). However, the geographic and stratigraphic occurrence of both species raise particular relevance for the comparison with specimen MAC GIR-042375 from Girona.

The subsequent new results on the morphology and taxonomy of the Girona specimen include additional data on the premaxilla, the maxilla (zygomatic-orbital bridge), the basicranium, and the dentition. All comparative descriptions rely on the preserved features only.

For the premaxilla, plesiomorphic conditions of the dental formula have been assumed by Voss et al. (2019). However, analyses of the CT-scans do not allow drawing further conclusions on this. While one pair of upper incisor tusks (I1) is clearly indicated by distinct alveoli and verified by both, macroscopic and CT-data (Fig. 5a, b ), further alveoli, i.e., for a second and third upper incisor (I2-3), cannot be ascertained from the CT-scans. Although the premaxillary dental formula is known to be complete in P. veronense and P. intermedium and hence would not provide a distinctive feature on species level here, the morphological information of the Girona specimen remains limited in this regard. As such, the available data of the holotype of P. ausetanum cannot be complemented assuming that the Girona sea cow is a representative of that species. The same applies to P. solei and P. montserratense . While P. solei shows an incomplete premaxilla that does not reveal informative characters relevant here, the premaxilla of P. montserratense is temporarily not accessible, because the holotype specimen is still embedded in sediment.

Further preserved diagnostic features of the premaxilla refer to the length of the alveolus for the first upper incisor (I1). Slab 2 reveals the alveoli of both I1 (Fig. 5a, b ). While the alveolus on the right side is distorted by inclusions, the one on the left side is already shown to be tapered dorsally indicating the uppermost part of the alveolus (Fig. 5c ). Slab 2 represents the lower 2 cm of the preserved partial skull with the left side reflecting a more ventral level than the right side. Hence, it is plausible to conclude that the left I1 alveolus definitively did not exceed half the length of the premaxillary symphysis considering slab 1 and the interspace between slab 1 and 2, both commencing at least 3 cm (Fig. 5c ). Moreover, it is likely that the I1 alveolus extends even less than half the length of the symphysis similar to P. veronense and P. intermedium (re-scored by Voss 2013: character 157), even though it is not exactly clear how much of the dorsalmost and ventralmost parts of the premaxilla are missing in MAC GIR-042375.

Another morphological feature with diagnostic value is related to the position of the zygomatic-orbital bridge of the maxilla with respect to the palate (e.g., Domning 1994: character 11; Voss 2013: character 21; Springer et al. 2015). According to Balaguer and Alba (2016), it is more or less level with the alveolar plane in P. ausetanum (about 6 mm above the tooth arcade). This is also the case with the Girona specimen as can be inferred from the CT-Scan shown in Fig. 4c . In fact, the zygomatic-orbital bridge in MAC GIR-042375 is even on the same level with the tooth arcade. While this condition is also observable in P. solei , this is a strong contrast compared to P. veronense and, according to personal observations also to P. intermedium , where the zygomatic-orbital bridge is elevated above the palate, with its ventral surface lying more than 15 mm above the alveolar margin ( Voss 2013). In P. montserratense , the zygomatic-orbital bridge is not accessible.

The morphology of the basicranium may have very likely diagnostic value. As stated above, the basisphenoidal-basioccipital region is preserved, which is best visible in macroscopic view of slab 1 (Fig. 4a ). Amongst sirenians, the shape of the spheno-occipital eminences varies in either being concave or convex, and therefore, may provide differences at species level ( Voss 2013). While these are convex in P. veronense and P. montserratense , they are concave in P. intermedium . This is a similarity with the Girona sea cow MAC GIR-042375. The postulated concavities at this level in macroscopic view are supported by the respective CT-scan (Fig. 4b ). In P. ausetanum , this character cannot be assessed, because the respective area is incomplete, covered by sediment and deformed insofar that bony material from the right side overlay the posteroventral region of the skull (Jordi Balaguer 2023, pers. comm.). In P. solei , the basicranium is not preserved.

At this point also a correction as to the presence of a hypoglossal foramen is made ( Voss et al. 2019: fig. 6a). CT-scanning and post-processing do not confirm a hypoglossal foramen on the left side of the basioccipital. Instead, the visible “foramen” in Fig. 4 is either the result of an inclusion or it could be a ventral excavation in the basioccipital, a pharyngeal fossa (PF). However, a hypoglossal foramen, if present in that species, would be positioned more caudad within the basioccipital.

Moreover, our examination of the preserved M1-3 crowns in deeper levels of slab 2 further helps to substantiate the taxonomic identity of the Girona specimen (Fig. 5a ). Compared with the species of Prototherium , it shows differences in the mesiodistal length of the molars (Table 1 ). The M1 in MAC GIR-042375 is generally smaller than in the three Prototherium taxa hitherto considered valid with a deviation of about 16% and 21% in P. veronense and P. intermedium , respectively, but only 2% in P. ausetanum if the M1 from the left side is compared. The M2 is also distinctly smaller than in P. veronense and P. intermedium having a difference of 18% and 9.5%, respectively, but it is about 5% larger than in P. ausetanum . With the exception of P. intermedium , which lacks respective information (presumably, because both M3 in the holotype are not fully measurable according to personal observations), the mesiodistal length of the M3 in the Girona sea cow is 11% and 5% smaller than in P. veronense and P. ausetanum , respectively.

Extending the comparisons upon the preserved M2-3 in P. solei and P. montserratense from Spain unveils interesting results. The M2-3 in MAC GIR-042375 (and in P. ausetanum ) is about 13% and 12% larger than in P. solei , respectively, making the latter species the smallest in terms of the molar dimensions. However, compared with P. montserratense , the M2-3 are almost as large as in MAC GIR-042375 (Table 1 ). Taken together, these measurements point to a sirenian that is rather small compared with the Prototherium taxa from Italy, but that is very similar in size compared with P. ausetanum and P. montserratense from Spain.

Another molar feature to distinguish MAC GIR-042375 refers to the shape of the M2-3. Except for P. ausetanum , all taxa for comparison show a trapezoid M2, with the protoloph each wider than the metaloph, whereas in both, MAC GIR-042375 (Fig. 5b ) and P. ausetanum ( Balaguer and Alba 2016: fig. 4), the proportions of the lophs are similar resulting in a rather rectangular outline of the M2. Similarly, the M3 is trapezoid in all Prototherium species from Italy and Spain, but not in MAC GIR-042375 (Fig. 5b ) and P. ausetanum ( Balaguer and Alba 2016: fig. 4) where the M3 is rather oval and tapering distad.

In the following, the description of specimen MAC GIR-042375 is complemented with regard to the preserved morphology visible in slab 1 (Fig. 6 ) in order to provide overall insights to the reader.

The CT-scan of the slab represents a relatively deep level of the upper part of the skull (Fig. 6b ).

After post-processing, the nasal area could be made well visible by reconstructing the extension of the vomer. The vomer is a mediolaterally thin element and triangular in cross section as in other sirenians. It passes through most of the internal narial passage, contacts the maxilla laterally (and most likely the palatine, that, however, cannot be distinguished here) and extends in a longitudinal groove on the dorsal surface of the maxilla slightly forward into the mesorostral fossa (external nasal opening). Anteriorly, the vomer builds up the nasal septum. Posteriorly, the vomer runs indistinctly out into the basicranium.

On the left side, parts of the frontal are visible with the slightly downturned supraorbital process. A complex of remnants of the left jugal anteriorly and the left zygomatic process posteriorly indicate the zygomatic arch of the squamosal.

Discussion

The analysis of the CT-Scans resulted in a genuine added value in terms of supplemented morphological data that allow for a more coherent taxonomic assignment of the Girona sea cow.

First of all, this specimen is undoubtedly considered as representing a Prototherium species. As mentioned above, the key characteristic for the genus Prototherium is the lack of a persistent DP5/ DP5 (e.g., Domning 1994; Bizzotto 2005; Balaguer and Alba 2016) that is also supported in this study.

Taking into account observations in more derived sea cows, i.e. members of the Dugonginae that show mesial drift of their molars and, as a consequence, lose preceding teeth in lifetime, one can argue that there may have been an unreplaced DP5 that was worn out and shed given the advanced age of the Girona specimen. The gap between P4 and M1 would then have been closed by mesial drift. Although this is an interesting perspective, it is likewise a difficult one. For example, there are two specimens known from Eocene strata representing adults of " Halitherium " Halitherium taulannense that show irregular resorption of DP5/DP5 (holotype skull RGHP D040 and paratype mandible RGHP E.7.096a ( Voss 2013)). The respective specimens reveal traces of resorption in so far that the area between P4/P4 and M1/M1 is roughened and spongy, but still retracing the former alveoli. Mesial drift however is not indicated in those specimens.

Traces of resorption are commonly present in specimens that are known to have lost teeth due to mesial drift, especially amongst members of the subfamily Dugonginae that are characterized by the reduction of the antemolar dentition (e.g., Domning 1989). According to Luckett (1993), resorption can take place in different degrees. For example, this is observable in the maxilla of Dioplotherium manigaulti Cope, 1883 in which the antemolar area is characterised by bone-filled, degenerated alveoli indicating the former presence of deciduous premolars; the only distinct socket remaining in this roughened area is for the lingual root of the originally three-rooted DP5 ( Domning 1989). Exactly such traces of resorption are missing in the Girona specimen. Instead, the area between P4 and M1 appears to be unsuspicious as are the roots of the preserved molars that do not indicate mesial drift (Fig. 5a, b ). Consequently, the lack of a persistent DP5/DP5 as a generic distinguishing feature for Prototherium is not questioned here.

Although the data basis remains limited not at least due to the lack of material between and in each of the two fossil-bearing slabs, the present study provides a combination of characters that argues with the utmost probability for the Girona specimen as being a representative of P. ausetanum . Two characters corroborate that hypothesis on the species level and refer to the shape of the M2 and M3. As outlined above, only MAC GIR-042375 and P. ausetanum share a rectangular M2 as well as an oval M3. Three further characters that do not argue unambiguously for a taxonomic assignment to P. ausetanum , but can be put into perspective, refer to the mesiodistal length of the molar crowns, the position of the zygomatic-orbital bridge, and the shape of the spheno-occipital eminences. Of all taxa for comparison, the mesiodistal length especially of M2-3 is most similar between MAC GIR-042375, P. ausetanum , and P. montserratense . A zygomatic-orbital bridge nearly level with the alveolar margin is congruent only in MAC GIR-042375, P. ausetanum , and P. solei . While P. montserratense differs from MAC GIR-042375 in having convex spheno-occipital eminences (a morphological feature that is also present in P. veronense ), personal observations of the holotype specimen also reveal that it is not fully grown. This is indicated by the non-obliterated spheno-occipital suture and the only slightly worn M3. As such, the subadult age may qualify the current results on the mesiodistal length of M2-3 in P. montserratense . The material basis of P. solei does not provide further morphological characters that can be compared here. However, three differing dental characters as mentioned above versus one shared cranial character, i.e. the level of the zygomatic-orbital bridge, question a taxonomic assignment of MAC GIR-042375 to P. solei . Against that background, the feature of concave spheno-occipital eminences in the Girona specimen may potentially supplement the morphological information of the holotype of P. ausetanum , which lacks respective data. In fact, these eminences are likewise concave in P. intermedium , but the latter differs from MAC GIR-042375 in the dental features outlined above and in that the zygomatic-orbital bridge is clearly elevated above the alveolar margin. As such, the fossil sea cow remains from Girona resp. Sant Vicenç de Castellet contribute to more and new information that enhance and supplement our knowledge on the morphology and, considering the similar age and area, distribution of P. ausetanum .

In addition to the morphology, the stratigraphy and palaeogeography implicate, at least in part, new hypotheses on the taxonomic identity of the species for comparison, which, conversely, support the assignment of MAC GIR-042375 to P. ausetanum . In the Bartonian, potentially four Prototherium taxa are present in southern Europe: P. ausetanum (including MAC GIR-042375), P. montserratense and P. solei , each from Spain, and P. veronense from Italy. For the latter species, different and sometimes contradictory information on its geological age exist throughout the literature, and therefore, this issue shall be clarified first. Apart from Bizzotto (1983: 106) and Balgauer and Alba (2016), for example, who provided a Bartonian age for P. veronense , it is also indicated as coming from late Eocene (Priabonian) strata ( Domning 1994, 1996; Zalmout and Gingerich 2012: table 1). Meanwhile, the age of the find locality of P. veronense , Monte Duello in the Italian Province of Verona, is considered to pertain to the middle Eocene, i.e. Bartonian ( Zorzin et al. 2012; Zorzin and Frisone 2013), making this taxon synchronous with the sirenians from Spain. The only Prototherium species coming from stratigraphically younger sediments, i.e. Priabonian, is P. intermedium from Italy ( Bizzotto 1983; Domning 1996). Interestingly, the stratigraphically older and palaeo-geographically more distant species, P. montserratense and P. solei ( Pilleri et al. 1989), are considered synonymous with P. intermedium , either both ( Domning 1996) or only P. solei ( Sagne 2001; Balgauer and Alba 2016). New hypotheses about the validity of the synonymously treated species are discussed as follows.

As outlined above, the present study provides morphological information from the not yet fully prepared holotype of P. montserratense that is even now valuable for comparison. For the time being, the morphological characters of it presented here argue for a species distinct from P. intermedium , hence, neither supporting a synonymy ( Domning 1996: 388) nor the status of a subspecies ( Bizzotto 2005) with the latter taxon. While Sagne (2001) already refuted the synonymy of P. montserratense with P. intermedium on a morphological basis, the distinction of both is also reinforced by their stratigraphic and palaeogeographic data. Moreover, P. montserratense is not considered a nomen dubium (Balguer and Alba 2016) though this treatment is well noted as being only provisional pending further investigations. Instead, similarities with P. veronense (and differences with P. intermedium ) like convex spheno-occipital eminences, may argue for the hypothesis that P. montserratense is a subadult representative of the former, hence potentially synonymous with P. veronense . Such synonymy that was also pondered by Balaguer and Alba (2016) would be additionally in accordance with the stratigraphic occurrence of both and the consideration of sirenians as migrating species. This hypothesis finds support through a number of other morphological features that are, however, not present in the object of this study. Taking into account that a taxonomic revision of the genus Prototherium is not the scope of the present study, this shall nevertheless be shortly outlined here. For example, P. veronense and P. montserratense show a ventral border of the horizontal mandibular ramus that is tangent to the angle as well as a posterior border of the mandible that has a distinct steplike process below the condyle, both character states contrasting with P. intermedium ( Voss 2013: appendix 4). Therefore, it shall be emphasized here that it is highly relevant to examine the holotype specimen of P. montserratense completely in order to clarify its status, either as a distinct, and hence valid species, or a synonym of P. veronense .

Similarly, P. solei is questioned here to be either a junior synonym ( Domning 1996; Sagne 2001; Balaguer and Alba 2016) or a subspecies ( Bizzotto 2005) of P. intermedium . On the one hand, the stratigraphic and palaeogeographic data provide arguments for a distinction of both species. On the other hand, morphological differences like the position of the zygomatic orbital bridge argue against an assignment of P. solei to P. intermedium . Considering again characters beyond those preserved in the object of this study, the mandibular morphology in P. solei reveals an interesting combination of characters. The course of the ventral border of the horizontal ramus that is most likely tangent with the angle resembles that in P. montserratense and P. veronense , whereas the posterior border of the mandible seems to lack a distinct process below the condyle similar to P. intermedium . In summary, P. solei demonstrates similarities and, at the same time, differences with all taxa for comparison, including P. ausetanum (and MAC GIR-042375). These incongruences illustrate that P. solei needs to be carefully reinvestigated in consideration of the other Prototherium species aiming at a thorough revision of the genus in total. As such, and for the time being, P. solei is provisionally treated as a distinct species according to Pilleri et al (1989).

Further studies could also help to fix the issue of the controversially discussed interrelationships of the genus Prototherium , which is shown as para- and polyphyletic in Balaguer and Alba (2016) as well as in the first ( Domning 1994) and the most recent ( Díaz-Berenguer et al. 2022; Heritage and Seiffert 2022) cladistic analyses of sirenians. In Vélez-Juarbe and Wood (2018), the interrelationships of Eocene sirenians, including Prototherium , are not resolved at all, showing a large polytomy. Domning (1996: 388) already assumed that P. intermedium is probably generically distinct from P. veronense , a hypothesis that has not further developed until to date. Therefore, this issue should be rather addressed after a thorough taxonomic-morphological revision of the genus, including the clarification of the status and affinities of P. solei and P. montserratense as outlined above.

This approach is particularly suggested in light of an alternative phylogenetic hypothesis that supports the monophyly of Prototherium by showing a sister group relationship between P. veronense and P. intermedium ( Voss 2013). This sister grouping is confirmed by a unique combination of homoplasies and a good agreement between the phylogeny and stratigraphy ( Voss 2013: fig. 75). Except character 21 referring to the level of the zygomatic-orbital bridge ( Voss 2013), none of the features supporting the sister grouping are preserved in MAC GIR-042375. Therefore, Voss’ (2013) data set and cladistic analysis shall be updated and tested for stability first before further conclusions on the phylogenetic interrelationships are drawn.

Considering the bigger picture, better-resolved intra- and intergeneric relationships based on new informative fossils such as MAC GIR-042375 have the potential to contribute to a better understanding of the interrelationships of Eocene taxa and sirenians in total. Although this task is beyond the scope of this paper, it shall be outlined here that the classification of Sirenia has long been controversial. In other words: the traditional classification concept of Sirenia , which distinguishes four families (e.g., Domning 1996), is incongruent with the results of phylogenetic analyses of the order. This refers especially to the Dugongidae , the family that P. ausetanum is traditionally referred to according to Balaguer and Alba (2016). While only two dugongid groups are consistently revealed as monophyletic, the derived Dugonginae and Hydrodamalinae , Eocene dugongids are conspicuously paraphyletic as well as most of the genera (e.g., Domning 1994; Springer et al. 2015). Additionally, some approaches (e.g. Vélez-Juarbe et al. 2012; Springer et al. 2015) reveal Dugongidae as a clade, whereas other studies, including one of the latest ( Domning 1994; Voss 2013; Díaz-Berenguer et al. 2018), do not affirm this hypothesis. Consequently, Eocene sirenians, and Eocene dugongids from the traditional point of view, represent one of the most controversially discussed taxa.

In order to solve this drastic paraphyly problem, Voss (2013) proposed the total group concept ( Jefferies 1979) and distinguishes a stem group, which, by definition, is consistently paraphyletic, and a monophyletic crown group within the clade Sirenia , hence setting aside the traditional classification concept of four sirenian families. In the same year, and as already outlined in the Systematic palaeontology chapter, O’Leary et al. (2013) introduced Pan-Sirenia as the most inclusive group comprising all species more closely related to crown- Sirenia than to any other living species. This finds also consideration in the very latest and total evidence phylogenetic analysis by Heritage and Seiffert (2022). As such, Eocene dugongids in the traditional sense (compare previous studies cited above) fall into the stem group of Pan-Sirenia . Most of the more derived dugongids are part of crown- Sirenia and form the more exclusive, but monophyletic Dugongidae as defined by Heritage and Seiffert (2022). In terms of P. ausetanum , this species has to be considered now as a stem-group representative, i.e., a pan-sirenian that falls outside crown- Sirenia .

Figures

Figure 1. Panoramic view of the SW side of the limestone quarry southeast of Sant Vicenc de Castellet showing the massive blocks of Bartonian carbonates bound by micritic matrix, and containing besides macroforaminifers, such as nummulites and discocyclinids, bryozoans, corals, and bivalves vertebrate remains (here described: MAC GIR- 042375).

Figure 2. Modified extract from the geological map (Mapa geologic de Catalunya 1: 25 000, Monistrol de Montserrat 392 - 1 - 1 (71 - 29 )) showing the position of the limestone quarry (circle: SE Sant Vicenc de Castellet, E of Llobregat river) from which the pavement slabs containing the MAC GIR- 042375 Prototherium cf. ausetanum remains came from. All sediments from this area are dated to the late Bartonian.

Figure 3. Spatial orientation of the fossil-bearing slabs in A. Lateral view; B. Frontal view. CT-scans of the upper slab 1 and the lower slab 2 (both in color) are aligned in an anatomically correct position using the extant dugong as reference (grey-scaled CT-scan of Dugong dugon ZMB _ MAM 69306).

Figure 4. Prototherium cf. ausetanum, MAC GIR- 042375. A. Photograph; B. Corresponding CT-scan of slab 1 with focus on the basicranium. The unpolished bottom side of slab 1 (A) shows parts of the basisphenoid, the spheno-occipital eminences, and the basioccipital, all of which being firmly fused indicating adulthood. The CT-Scan of the basicranium (B) shows interferences, however, the suture between basisphenoid and basioccipital is unequivocally obliterated and estimated to be more or less level with the vertical interference (indicated by arrows). Abbreviations: BO, basioccipital; BS, basisphenoid; J, jugal; M, maxilla; PF, pharyngeal fossa; PM, premaxilla; P 3 - 4, third and fourth upper permanent premolars; SOE, spheno-occipital eminences; ZOB, zygomatic-orbital bridge of maxilla.

Figure 5. Prototherium cf. ausetanum, MAC GIR- 042375. A. Photograph showing the dorsal side of slab 2; B. Corresponding CT-scan showing the ventral side of slab 2 focusing on the dentition; C. CT-scans of anatomically correct combined slab 1 and 2 in lateral view. The CT-scan of the lower side of slab 2 (B) reveals the crowns of the upper molars 1 - 3 (M 1 - 3) with the M 3 fully erupted and clearly in wear. Abbreviations: BS, basisphenoid; C, upper canine; FI, foramen incisivum; I 1, first upper incisor; J, jugal; M, maxilla; M 1 - 3, first to third upper molars; P 1 - 4, first to fourth permanent upper premolars; PM, premaxilla; ZOB, zygomatic-orbital bridge of maxilla.

Figure 6. Prototherium cf. ausetanum, MAC GIR- 042375. A. Photograph; B. Corresponding CT-scan of slab 1 providing data on the overall morphology of the specimen. Abbreviations: BC, basicranium; F, frontal; M, maxilla; PM, premaxilla; V, vomer.