Valentinella vitrollense

Tabuce, Rodolphe, Tortosa, Thierry, Vianey-Liaud, Monique, Garcia, Géraldine, Lebrun, Renaud, Godefroit, Pascal, Dutour, Yves, Berton, Sévérine, Valentin, Xavier & Cheylan, Gilles, 2013, New eutherian mammals from the Late Cretaceous of Aix-en-Provence Basin, south-eastern France, Zoological Journal of the Linnean Society 169 (3), pp. 653-672 : 658-663

VALENTINELLA VITROLLENSE TABUCE, VIANEY- LIAUD & GARCIA, 2004

Holotype: UM-VLP-2, damaged right dentary with p4-m3.

Referred material: UP-VLP-10-01, a left dentary fragment with roots of m1−m3; UM-VLP-4, a dentary with remnants of right and left molariform teeth; UM-VLP-3, fragment of a probable right PX; UP-VLP- 07-04, fragment of a left upper molar.

Remarks: Despite the poor preservation of all these specimens, their attribution to Valentinella vitrollense is supported by their large size. Lower molars of Valentinella are at least twice as large as those of most Late Cretaceous eutherian mammals ( Asioryctitheria , Zalambdalestidae , Zhelestidae , Gypsonictops , Deccanolestes , Kharmerungulatum ) and similar in size to those of Cimolestes stirtoni , which is, along with two larger species, Cimolestes magnus and Protungulatum coombsi , one of the largest Late Cretaceous eutherian mammals. Additionally, all specimens described herein were found concentrated in the same excavation area ( Valentin et al., 2012).

Revised diagnosis: Valentinella differs from all the other Late Cretaceous eutherian mammals by a unique combination of characters: the jaw angle is elongated with a convex ventral border indicative of a relatively elevated angular process; simple and bulbous p4; p5 with a hypoflexid and a molarifom talonid; robust lower molars with labiolingually expanded talonid; labially inclined protocone with a long lingual slope; and development of an incipient hypocone on the cingulum.

Measurements (in mm): UM-VLP-2 (p4: L = 3.11, W = 1.91; p5: L> 3.2, W = 2.77, LTa = 1.5; m1: L = 4.04, W = 2.8, LTa = 2.45; m3: L = 3.48, W> 2.33); UM-VLP-3 (Px: L ≥ 3.33, W> 2.71).

Description: The newly discovered dentary fragment UP-VLP-10-01, although having only preserved roots and/or alveoli of m1−m3, is informative in showing a robust and deep mandibular body (8.95 mm below m2). Ventrodistally to the m3, the mandibular body displays the ventral extremity of a blunt coronoid crest ( Fig. 3 ). The close proximity between this structure and the posteriorly inclined large distal root of m3 (mesially indented by a thin bony lamina) indicates that the retromolar space was reduced or even absent. The mandibular ramus is ventrodistally elongated behind m3 and has a very ventral convex border. This morphology indicates the occurrence of a rounded angle with the development of a relatively elevated angular process (near the alveolar margin). On the holotype UM-VLP-2, the mandibular body revealed by 3D reconstruction is too severely crushed and displaced labially to be described; only the mandibular canal is preserved ( Fig. 4 ).

The holotype has two premolars and three molars; all teeth are damaged, with the exception of the relatively well-preserved penultimate premolar ( Figs 5 , 6 ). The premolars are numbered as p4 and p5 based on a growing consensus that the primitive dental formula of early eutherians and placentals comprises five premolars ( Cifelli, 2000; Wible et al., 2009; Archibald & Averianov, 2012). The two-rooted p4, oval in occlusal outline, is a simple tooth dominated by a large and bulbous protoconid; distally, there is a low, small unbasined heel with a small cingular cusp. The p5, more poorly preserved, is longer than p4 and shorter than m1. There are two labial and one lingual patches of enamel and a remnant of the hypoflexid. Enamel elements on the distal wall of the trigonid indicate also that the talonid has a similar length as the trigonid. The talonid is however narrower compared with the trigonid. The hypoconulid is distolingually displaced. The potential presence of a metaconid, proposed by Tabuce et al. (2004) based on the believed labial position of the protoconid, is possibly an erroneous interpretation as most of the crown is missing. However, the hypothesis that, as seen in p4, the trigonid was unicuspid with a large and bulbous protoconid is unlikely regarding the molariform and basined talonid. The combination of a single-cusped bulbous trigonid and a basined talonid is indeed unknown in Cretaceous or Palaeogene eutherians. As a result, the trigonid of p5 was certainly molariform, with three cusps.

The molars are preserved on the holotype as well as on UM-VLP-4, which was initially interpreted as bearing?p4−?m2 by Tabuce et al. (2004). Subsequent preparation of this specimen revealed that this dentary fragment has poorly preserved right and left teeth. On the labial side of the talonid of a molariform tooth, a small portion of wrinkled enamel is preserved. On the holotype, characters of the molars are also mainly obliterated by wear and distortions caused by tectonic deformation. These distortions are obvious in μCT reconstructions ( Fig. 4 ); the trigonids are distally and vertically shifted relative to the talonids. As a result, we cannot define exactly the height and degree of compression of the trigonid (i.e. the position of the paraconid relative to the metaconid). Likewise, we cannot determine whether the hypoconulid is twinned with the entoconid or more centrally positioned on the postcristid. On the labial side of m1, some enamel remnants can be distinguished on both the protoconid and the hypoconid. The base of both of these cusps is bulbously constructed. The hypoflexid is also preserved and appears relatively deep. All the cusps show considerable attrition; dentine is well exposed in both the distal part of the trigonid and the whole talonid. The worn occlusal surfaces are flat, so the trigonid and talonid appear to be the same height. A wear facet on the labial side of the hypoconid is still visible. The talonid appears shorter and wider than the trigonid. The m2 is so crushed and worn that no unquestionable morphological observations can be made. Judging from the robustness of their roots (also obvious on UP-VLP-10-01), it seems that the relative molar size is m1 <m2> m3. Only the labial half of m3 is not crushed. Like the occlusal surface of m1, m3 is strongly worn, suggesting a crushing function of the molars during dental attrition. On the talonid, which is shorter than the trigonid, the remaining area of the postcristid does not suggest any strong distal development of a hypoconulid.

The upper dentition is represented by two specimens. UM-VLP-3 is a two-rooted tooth that could be, according to its size, a P3 or a P4 ( Fig. 7A, B ). This premolar was initially thought to be preserved along with possible remnants of canines on a maxillary ( Tabuce et al., 2004). However, a better a preparation of the specimen and μCT-scan analysis indicate that both the supposed maxillary and canines are merely isolated fragments of bone without an obvious direct relationship with the premolar. On the latter, the main cusp is attributed to the paracone as it is centrally positioned above the two roots. The central position of the paracone, relative to the roots, also suggests that there is no metacone. The paracone is laterally narrow and slightly expanded along its lingual margin, but, as the crown is broken lingually, the presence or absence of a small protocone is unknown. The paracone exhibits a sharp distal crest ending in the incipient metastyle; there is no labial cingulum. The second specimen, UP-VLP-07-04, is a fragmentary left upper molar, with only the distolingual part of the crown preserved ( Fig. 7C, D ). This fragment displays the hypocone and part of the protocone. Although the protocone is too fragmentary to estimate its height and size, this cusp appears somewhat mesiodistally inflated, inclined labially and has a long lingual slope. The hypocone is an incipient cusp resulting in elevation of the lingual end of the postcingulum. The hypocone is affected to a minor degree by occlusal wear.

Comparisons and discussion: The holotype of Valentinella is regarded as having two premolars (p4 and p5) and three molars. However, considering that the p4 is a simple premolariform premolar and that the p5 is interpreted as a molariform tooth, a combination that is not characteristic of eutherians, two other alternative interpretations of the dental formula of the holotype need to be discussed: p4-dp5- m1-m2-m3 or alternatively p3-m1-m2-m3-m4. The first hypothesis, which claims the presence of a dp5 on the holotype, appears unlikely as it would imply that this deciduous last premolar is retained in old adults (the m3 is very worn on the holotype). In addition, μCT-scan analysis indicates that there is no trace of p5 beneath the presumed deciduous tooth. The second interpretation, the presence of four molars, would imply that Valentinella is a metatherian, an assumption that is clearly also refutable considering the presence of a hypocone on the distal cingulum of the upper molar UP-VLP-07-04 and the morphology of the dentary UP-VLP-10-01, which does not indicate the presence of a medial inflected angle. As a result, our identification of the preserved loci in the holotype of Valentinella (p4-p5- m1-m2-m3) appears to be the most logical hypothesis. Finally, the lack of gradual molarization of the premolars (combination of a simple p4 with a molariform p5) in Valentinella can be observed in other Late Cretaceous eutherians (e.g. Zalambdalestes ).

Valentinella was initially compared to both Zhelestidae and some ‘condylarths’, based on the robust dental morphology of the holotype with specialized crushing-grinding function as illustrated by its bulbously constructed cusps that are severely worn horizontally (see Tabuce et al., 2004). Valentinella was subsequently regarded as closely related to the zhelestids because of ‘the lingual or sublingual position of the paraconid with a clear appression to the metaconid, the entoconid and the hypoconulid twinned, the expanded talonid, and by the indication of a quite similar height of both trigonid and talonid’ ( Tabuce et al., 2004: 352). However, based on both μCT analysis and an advanced preparation of the holotype, it appears that it is not possible to determine confidently the position of the paraconid and hypoconulid relative to the metaconid and entoconid, respectively. Therefore, we refute our initial assumption that Valentinella is partly diagnosed by an appression of the trigonid and an entoconid twinned with hypoconulid. Similarly, Archibald & Averianov (2012) recently considered V. vitrollense as a nomen dubium because of the putative lack of diagnostic features on the holotype.

However, we estimate that some characters on the Valentinella hypodigm are relevant to maintain this genus as a valid taxon. Firstly, Valentinella is characterized by a robust dental morphology with specialized crushing-grinding function as illustrated by the horizontal apical wear facet of the hypocone and by the bulbously constructed cusps of the holotype that are severely worn horizontally. Valentinella is also characterized by the rounded and enlarged angle of its lower jaw with an assumed relatively elevated angular process, a bulbous protoconid and an unbasined heel on p4, a molariform p5 having a wide talonid and a hypoflexid, a large talonid on the molars, a somewhat reduced m3 relative to m2, a premolariform?P3 or?P4 lacking a metacone. Finally, the upper molar exhibits few, but important, characters such as the long lingual slope of the labially inclined protocone and the development of an incipient hypocone.

These characters on the upper molar are important to compare Valentinella . Amongst zhelestids, a swollen and lingually elevated postcingulum, without true hypocone, developed as a distinct cusp, is present only in the Zhelestinae Aspanlestes aptap , Parazhelestes robustus , and Eoungulatum kudukensis . A developed hypocone, as observed in Valentinella , has not been described in zhelestids so far. Amongst other Cretaceous mammals, the postcingulum is slightly enlarged in the asioryctitherian Kennalestes and the cimolestids Batodon and Maelestes , but a hypocone is not developed. The only known Cretaceous mammal in which a hypocone is incipiently developed as in Valentinella is Gypsonictops . A lingually high postcingulum with a hypocone shelf showing both a lingual slope and an occlusal wear facet is present in some M2 of this North American genus (AMNH 21984; Clemens, 1973: 20). However, the hypocone of Gypsonictops is less developed in occlusal view than in Valentinella . The lower dentition of Gypsonictops and Valentinella share the molarization of p5. Gypsonictops has however a more molarized p4 with constantly present paraconid and metaconid, and a basined talonid.

Amongst latest Cretaceous−early Palaeocene mammals, Luo (1991) observed a morphocline on the lingual part of the upper molars. This morphocline arises with the cimolestid Procerberus , in which the postcingulum is low lingually and the protocone has a short lingual slope. In the asioryctitherian Kennalestes , the postcingulum rises lingually but does not present a lingual slope. The next grade is exemplified by the leptictid Gypsonictops (or to a lesser degree by the zhelestids and the plesiadapiform Purgatorius ), in which a small lingual slope of the postcingulum occurs and the lingual slope of the protocone is longer. The ultimate grade is represented by Protungulatum , in which a true incipient hypocone occurs and the lingual slope of the protocone is very long relative to the tooth size.

Along with Valentinella , the first mammals that developed an incipient hypocone are therefore taxa that have been included within the nebulous paraphyletic ‘condylarths’ ( Protungulatum included). Whether these ‘condylarth’ lineages occurred in the latest Cretaceous or just after the Cretaceous/Tertiary (K/T) boundary is still a matter of debate ( Archibald et al., 2011). In the current state of our knowledge, the first ascertained latest Cretaceous ‘condylarth’ species are Protungulatum coombsi from Montana, USA ( Archibald et al., 2011) and Kharmerungulatum vanvaleni from India ( Prasad et al., 2007). The presence or absence of a hypocone on the upper molars of these species could not be determined as they are only known by an upper premolar and a lower molar, respectively. The earliest North American ‘condylarths’ with a developed hypocone are members of various genera that appeared during the earliest Palaeocene (Puercan 1, NALMA), namely the?arctocyonid Protungulatum , the arctocyonids Oxyprimus and Baiconodon, and several periptychids such as Mimatuta and Oxyacodon ( Archibald, 1982; Lofgren et al., 2004).

The hypocone of Valentinella therefore appears to be a significant character because it limits the comparisons to just a few genera. However, from a phylogenetic point of view, there is no clear evidence that this character supports ‘condylarth’ affinities for Valentinella . Indeed, the hypocone convergently evolved more than 20 times amongst placentals during the Cenozoic ( Hunter & Jernvall, 1995), and the same is probably true for the latest Mesozoic (the pre-hypocone grade seen in Gypsonictops and zhelestids supports this hypothesis, see above). Further detailed comparisons between Valentinella and earliest ‘condylarths’ are limited because of the poor preservation of the lower dentition of the former. Valentinella resembles Mimatuta in having a molariform p5 with talonid and trigonid of equal width, and a long, basined talonid; but it differs by the reduced m3 relative to m2. This character also distinguishes Valentinella from Protungulatum , Oxyprimus, Baiconodon , and Oxyacodon . How- ever, some genera from the Puercan 2, such as Mimotricentes , Litomylus , and Eoconodon , are characterized by a reduced m3 as in Valentinella . However, considering that these taxa belong to various families (namely the arctocyonids, hyopsodontids, and triisodontids, respectively), the reduction of m3 is a probable homoplastic trait amongst ‘condylarths’. Some Puercan ‘condylarths’, such as Protungulatum donnae (UCMP 121782, the holotype of Protungulatum ‘ mckeeveri ’) and Eoconodon coryphaeus (AMNH 16329), also resemble Valentinella by their elongated mandibular ramus with a rounded angle and an elevated angular process. Finally, based on enamel microstructure, Tabuce et al. (2004) observed several characters that could indicate ‘condylarth’ affinities for Valentinella : closely packed prisms, with a circular cross-section, and a large size (6–8 μm in diameter), single-layered Schmelzmuster with radial enamel, reduction of the IPM (interprismatic matrix), and a similar angle between the crystallites of the IPM and prism’s long axes (40–45°). Interestingly, Protungulatum shares with Valentinella the three latter traits, the two latter being derived amongst eutherians.

Based on enamel microstructure alone, comparisons between Valentinella and zhelestids are unfortunately impossible because enamel microstructure of the latter was neither described, nor illustrated. A study of the enamel microstructure of stem eutherians is clearly needed. In any case, Valentinella differs from all zhelestids by being twice as large in size, an enlargement of the angle of the jaw with a higher elevated angular process, a more bulbous protoconid on p4, and a more advanced condition of the hypocone (see above). Besides these differences, Valentinella shares with zhelestids a submolariform p5 having a wide talonid ( Paranyctoides , the supposed sister-taxon of zhelestids, presents a less molarized p5), a robust molar morphology with potential specialized crushing-grinding function (bulbously constructed cusps of the holotype that are severely worn horizontally, horizontal apical wear facet of the hypocone, and large talonid), a somewhat reduced m3 relative to m2 ( Aspanlestes , Parazhelestes , Zhalmouzia , but not Borisodon ), simple?P3 or?P4 lacking a metacone.

Figures

Figure 3. Valentinella vitrollense from Vitrolles-La Plaine, UP-VLP-10-01, left dentary fragment with roots of m1−m3. A, occlusal; B, labial views. Note the posteriorly inclined distal root of m3, mesially indented by a thin bony lamina.

Figure 4. Valentinella vitrollense from Vitrolles-La Plaine, holotype UM-VLP-2, right dentary with p4−m3. A, twodimensional virtual slice images acquired by X-ray computed microtomography (μCT); three-dimensional reconstruction using μCT scan images; B, labial; C, mesial views.

Figure 5. Valentinella vitrollense from Vitrolles-La Plaine, holotype UM-VLP-2, right dentary with p4−m3. A, labial; B, occlusal views; C, interpretative drawing in occlusal view, salmon-coloured areas indicate preserved patches of enamel.

Figure 6. Valentinella vitrollense from Vitrolles-La Plaine, holotype UM-VLP-2, occlusal stereoview of p4−m1.

Figure 7. Valentinella vitrollense from Vitrolles-La Plaine. UM-VLP-3, probable right?P3 or?P4; A, occlusal; B, labial stereoviews. UP-VLP-07-04, fragment of a left upper molar; C, occlusal; D, distal stereoviews. Yellow, white, and red dotted lines indicate, respectively, the broken parts of the crown, the slope of the protocone, and the lingual slope of the hypocone on the postcingulum.