MISTRALESTES ARCENSIS
SP. NOV.
( FIGS 8–10
View Figure 8
View Figure 9
View Figure 10
)
Holotype: MHNAix-PV. 2008.1.1, right dentary fragment with p5 to m3 and roots of p4.
Type locality: La Cairanne-Highway (village of Rousset, Aix-en-Provence Basin, Bouches-du-Rhône, France), Upper Campanian.
Etymology: From the coastal river Arc, which flows south of the La Cairanne-Highway locality.
Diagnosis:
Mistralestes
differs from other known Late Cretaceous eutherian mammals by the unique combination of the following characters: a robust premolariform p5 without cingulid, paraconid, or metaconid; a gradual compression of the trigonid from m1 to m3; a transverse protocristid; and a paracristid probably fused with the protocristid on m3.
Measurements (in mm): MHNAix-PV. 2008.1.1 (p5: L = 2.24, W = 1.16, LTr = 1.40; m1: L = 2.30, LTr = 1.05, WTr = 1.38, WTa = 1.34; m2: L = 2.84, LTr = 1.47, WTr = 1.75, WTa = 1.64; m3: L ≥ 2.30, LTr = 1.15, WTr> 1.54, WTa ≥ 1.42).
Description: The horizontal ramus of the dentary is robust but moderately deep (height = 3.9 mm under m2). A virtual deconstruction and reconstruction of the holotype, using 3D μCT scan data ( Figs 9
View Figure 9
, 10
View Figure 10
), allowed us to replace manually some parts of the specimen such as the trigonid of m2 and the anteriormost part of the dentary. Four teeth are preserved, the first one is premolariform and the other three are molariform. According to our interpretation, MHNAix-PV. 2008.1.1 contains p5 to m3 (see below). A bony sulcus on the dentary extends ventrolingually from the mesial root of p4 to the mesial root of p5; this sulcus does not correspond to the mesial end of the mandibular symphysis but to an artefact of preservation (the bone being clearly altered dorsally to this sulcus). The dentary has two mental foramina. The mesial one is located ventral to the mesial root of p4. The second one is situated beneath the distal root of m1 (the position of these mental foramina is discussed below). The dentary is labiolingually enlarged at the level of m3. The μCT scan data show that the roots of p4 are smaller and shorter than those of p5, suggesting that p4 was somewhat reduced ( Fig. 9
View Figure 9
). The p5 is as long as m1. The m2 is larger than m1, and also probably than m3. From p4 to m2, both roots have a similar morphology; the anterior root is shorter and less robust than the posterior one. In contrast, the anterior root of m3 is longer and more robust than the posterior root.
The p5 is a robust premolariform tooth without a cingulid ( Fig. 8D
View Figure 8
); the protoconid is the dominant cusp and is apically truncated by breakage. A weak paracristid is faintly visible; this cristid does not reach the mesial base of the crown. There is no trace of a swelling of the crown in either the metaconid or paraconid regions. The talonid is nearly as wide as the protoconid and long (about one third of the total mesiodistal length of the tooth). The cristid obliqua is long and very robust; it reaches the distal flank of the protoconid. The hypoconid is partly broken in its mesiolabial part, labially offset from the protoconid, and connected to a tiny cuspid (incipient hypoconulid?). This cuspid is distally protruding and has a short lingual cristid that partially closes the talonid basin.
The m1 is the best preserved molar ( Figs 8E
View Figure 8
– 10
View Figure 10
). The trigonid is less than twice as high as the talonid and has bulbously constructed cusps. The paraconid is cuspate and rounded; it is located only slightly labial to the metaconid. The base of the paraconid is high on the crown. There is a prominent cingulid cuspule f (see Kielan-Jaworowska, Cifelli & Luo, 2004), mesial to the protoconid. This cuspule f is joined to the paracristid notch by a short crest. The protoconid is larger than the metaconid; both cusps are transversally aligned relative to the long axis of the crown. Although the height of the protoconid and metaconid cannot be accurately estimated because they are worn, the protoconid was surely the highest cusp of the trigonid. The paracristid is rectilinear from the protoconid to the paraconid (not L-shaped). The protocristid notch is narrow and rather shallow. The distal flank of the trigonid has a concave vertical wear surface. A short postmetacristid is visible. On the talonid, the cusps and crest are largely obliterated by wear. However, the cusps were evidently peripheral and the hypoconid was certainly the most voluminous cusp on the talonid. The orientation of the cristid obliqua cannot be adequately determined. The lack of a labial valley in the hypoflexid area may suggest that the cristid obliqua was labially orientated. The talonid is longer than and as wide as the trigonid. Lingually to the hypoconid, the distal outline of the crown is concave; this concavity receives the cuspule f of the m2. Judging from this concavity, the entoconid and hypoconulid, which are not clearly visible on the distolingual border of the crown, were probably quite close or even twinned, and distally salient relative to the hypoconid. There is no postcingulid.
fluent, the protoconid is more mesiodistally pinched, and the cuspule f appears smaller in labial view but remains mesiodistally extended. The cristid obliqua is labial and short; there is no hypoflexid. The distal part of the crown suggests that the talonid was more elongated relative to the trigonid.
The m2 is rather similar to the m1; m2 differs from it in being longer and larger, in having a more crestiform cuspule f, and a paraconid more tightly appressed to the metaconid ( Fig. 10
View Figure 10
). During the preparation of the specimen, the trigonid was broken and improperly glued, so it appears higher than that of m1 ( Fig. 8A, B
View Figure 8
). The μCT-scan reconstruction ( Fig. 9B, C
View Figure 9
) exemplifies that the height of the trigonid relative to the talonid is similar in both m1 and m2. On m2, the apices of the closely approximated entoconid and hypoconulid are still visible despite extensive wear ( Fig. 10
View Figure 10
).
Although its distal part is broken off, m3 seems shorter than but as wide as the m2 ( Fig 10
View Figure 10
). The lingual margins of the paraconid, metaconid, and talonid are broken off. The apex of the protoconid is unworn; distolabial to this cusp, a small ectostylid is present. By comparison with the two other molars, the paracristid and the protocristid are probably con- Comparisons and discussion: The first question is to determine whether p5-m3 or p4-m2 (with p5 or dp5) is preserved in the holotype of
Mistralestes
. The higher degree of molarization of the three posterior teeth, relative to the premolariform aspect of the first tooth, speaks for the first hypothesis. However, the slight reduction of the second tooth, relative to the third one, may alternatively suggest that these teeth are p5 (or dp5) and m1, respectively. Moreover, the position of the posterior-most mental foramen under the distal root of the second tooth also favours the hypothesis that this tooth is p5 (or dp5). In most Cretaceous eutherians, the posterior-most mental foramen is indeed under the ultimate premolar, rarely at the mesial root of m1. Consequently, the holotype of
Mistralestes
may preserve p4−m2 (with p5 or dp5) instead of p5-m3. This uncertainty can be partly resolved using our µ CT analysis. Our reconstruction reveals that (1) there is no developing premolar beneath the erupted teeth; (2) there is no trace of a root for a supplementary tooth distal to the ultimate preserved molar; and (3) the distal root of the last preserved molar is shorter and slightly compressed laterally and more posteriorly inclined than that of the other molars, suggesting that this last preserved molar is the ultimate locus of the tooth row. These data support the hypothesis that
Mistralestes
is documented by p5−m3. As for the position of the posteriormost mental foramen under the distal root of m1, which may appear problematic, we do not attach decisive importance to this character as it is well known that the position and number of mental foramina is variable amongst mammals. For example, amongst zhelestids, the anterior foramen is under i1, i2, or the canine and the posterior foramen between p4 and p5. A third foramen can be present under p1. In the zalambdalestoid
Bobolestes
, the posterior mental foramen is under the posterior root of p4, or between p4 and p5. A third small mental foramen can be also present under the anterior root of m1. As a consequence, we cannot exclude the possibility that the posterior-most mental foramen in
Mistralestes
is a supernumerary one and that another foramen was present under the p5 where the dentary is broken. Finally, the reduction of the m1 relative to the m2, although not characteristic of most early eutherians, occurs in numerous taxa, including the earliest ‘condylarths’ (e.g.
Protungulatum, Baiconodon
,
Oxyprimus
) (see also Halliday & Goswami, 2013). Hence, we are confident that the holotype of
Mistralestes
preserves p5−m3. Consequently, the absence of m4 from the dental formula, together with the reduction and labial position of the paraconid relative to the metaconid and the lack of a prominent vertical keel on the mesiolingual edge of the paraconid, strongly indicate that
Mistralestes
is not a metatherian but a eutherian mammal.
Amongst Late Cretaceous eutherian mammals,
Mistralestes
superficially resembles
Zalambdalestidae
. They share lower molars with a mesiodistally compressed trigonid, a relatively low trigonid, and a large and wide talonid. Zalambdalestids however are distinct from
Mistralestes
by the increase in height of the talonid from m1 to m3, their less bulbously constructed cusps, and their high-crowned, large, molariform ultimate premolar (with paraconid and metaconid well developed).
Mistralestes
is peculiar because of its simple p5 without a cingulid, paraconid, or metaconid. Other unusual characters of
Mistralestes
are, at least on m3, the apparent lack of a hypoflexid and the labial orientation of the cristid obliqua, the lingually displaced hypoconid, and the gradual compression of the trigonid from m1 to m3. These traits excludes any relationships with the vast majority of Late Cretaceous to early Palaeogene eutherians described so far (Asioryctheria,
Cimolestidae
, plesiadapiforms, ‘condylarths’, Gypsonyctops,
Deccanolestes
).
Comparisons with Late Cretaceous eutherians from Europe are limited to
Valentinella
,
Labes
, and
Lainodon
.
Valentinella
and
Mistralestes
share a robust dental morphology and a reduced m3. However,
Mistralestes
differs from
Valentinella
by being half the size, and considering the smaller size of the roots of its p4, we can easily suppose that
Mistralestes
has a relatively smaller p4 than
Valentinella
.
Valentinella
also differs from
Mistralestes
by a molarized p5 with a hypoflexid and a talonid as wide as the trigonid.
Labes
and
Lainodon
share with
Mistralestes
numerous characters that are also diagnostic features of zhelestids, suggesting that
Mistralestes
could be a zhelestid. According to Gheerbrant & Astibia (2012), these characters include a robust dental morphology with specialized crushing-grinding function, as illustrated by their basally inflated cusps, a horizontal wear pattern, lower molars with low trigonid, a paraconid more or less lingually located, a large and wide talonid, and a hypoconulid more or less appressed to the entoconid. Further similarities between Lainodontinae and
Mistralestes
include the lack of a postcingulid, the protoconid being taller than the metaconid, and the occurrence of both a postmetacristid and robust cuspule f.
Labes
is known by
Lab. quintanillensis
from Quintanilla del Coco (Burgos Province, Spain) and by
Lab. garimondi
from Champ-Garimond (Languedoc- Roussillon Province, France) ( Pol et al., 1992). These two species are documented by noncomparable material,
Lab. garimondi
by two m1 or 2 and
Lab. quintanillensis
by only one m3. Pol et al. (1992) and then Archibald & Averianov (2012) argued that these two species essentially differ by their size,
Lab. quintanillensis
being the smaller. The m1−2 of
Lab. garimondi
is intermediate in size between the m1 and the m2 of
Mistralestes
. Moreover, in addition to the zhelestid characters listed above,
Mistralestes
evokes
Labes
by a transverse protocristid (metaconid and protoconid set at the same transversal level). However, contrary to
Labes
, which is characterized by a trigonid on m3 that is moderately compressed mesiodistally,
Mistralestes
has a fully compressed trigonid on m2−3.
Labes
also differs from
Mistralestes
by a deeper hypoflexid, a wider and deeper V-shaped protocristid and, on m3, by a more centrally positioned hypoconulid and the fusion of the two roots.
Lainodon
is known by at least three species:
Lainodon orueetxebarriai
,
Lainodon ragei
, and
Lainodon sp.
from Laño ( Spain), as well as by
Lainodon sp.
from Taveiro ( Portugal), which may be conspecific with
Lai. ragei ( Gheerbrant & Astibia, 2012)
. The only known lower premolar of
Lainodon
that allows comparisons with
Mistralestes
is the p4 or p5 L1AT 20 attributed to
Lai. cf. orueetxebarriai
. It shares with
Mistralestes
a similar size, the robustness of the structures, the lack of a distinct cingulid, the crown dominated by a large and high protoconid without a metaconid, the long and robust cristid obliqua that rises high and labially on the trigonid, and the small but complex talonid with an enclosed basin, and a distally salient (and laterally compressed) hypoconulid. However,
Mistralestes
differs by a talonid as large as the protoconid and the absence of a postmetacristid, which is continuous with the entocristid in
Lainodon
. The lower molars of
Lainodon
are only known by two well-preserved probable m1 (L1AT 14 and L1AT 6), a damaged and worn putative m2 (L1AT 1), and a few fragmentary trigonids and talonids. The m1 of
Lai. ragei
is comparable in size with that of
Mistralestes
, whereas it is much larger in
Lai. orueetxebarriai
. Unfortunately, further comparisons in teeth dimensions are not possible with
Lai. ragei
because of the lack of both m2 and m 3 in this species. In
Lai. orueetxebarriai
, the m1 is as large as the m2, whereas the m1 is significantly smaller than the m 2 in
Mistralestes
.
Mistralestes
also differs from
Lai. orueetxebarriai
(specimen L1AT 1) by a lower horizontal ramus. Comparisons are impossible for m2, because it is damaged in the only
Lai. orueetxebarriai
specimen. The m1 of
Lai. orueetxebarriai
also differs by slightly stronger bunodonty, a mesially canted trigonid bearing a larger protoconid relative to the metaconid, the presence of a paraconid along its lingual margin, the distal position of the metaconid (protocristid oblique), a more developed postmetacristid, a shorter talonid, and the lower crown base below the hypoconid.
Labes
and
Lainodon
were recently gathered in their own subfamily by Gheerbrant & Astibia (2012), whereas Archibald & Averianov (2012) regarded these two genera as
Zhelestidae
incertae sedis along with
Sheikhdzheilia
( Uzbekistan; early Cenomanian),
Borisodon
( Kazakhstan; early Turonian),
Gallolestes
( Mexico and Texas, USA; Campanian), and
Avitotherium
(Utah and Montana, USA; Campanian).
Amongst these genera,
Mistralestes
shares with
Sheikhdzheilia
the lack of a metaconid on p5, a protoconid taller than the metaconid (as in
Labes
and
Lainodon
), and the absence of a postcingulid on the molars. However,
Mistralestes
differs from
Sheikhdzheilia
by numerous characters, including a p5 without precingulid, lingual cingulid, or paraconid; a transverse protocristid, a more closed trigonid with gradual appression of the paraconid and metaconid; and a labially orientated cristid obliqua on the molars.
Mistralestes
differs from
Avitotherium
by the same characters.
Mistralestes
more closely resembles
Borisodon
and
Gallolestes
by its transverse protocristid between the protoconid and the metaconid. However, it differs from them by the gradual appression of the paraconids and metaconids from m1 to m3, the labial orientation of the cristid obliqua, and the lack of both postcingulid and hypoflexid (at least on m3). On p5,
Mistralestes
and
Borisodon
share the absence of a metaconid, but
Borisodon
differs by a longer talonid, a complete lingual cingulid, and a less robust, distinctly curved, and hook-like protoconid, with its apex pointed distally.
Borisodon
has a faint Meckelian groove, which is clearly lacking in
Mistralestes
.
Mistralestes
also differs from
Borisodon
by the m1 being much smaller than the m2.
The other
Zhelestidae
, which are grouped within the subfamily
Zhelestinae
, belong to the genera
Zhelestes
,
Aspanlestes
,
Parazhelestes
, and
Eoungulatum
from the Turonian to?Coniacian of Uzbekistan. Compared with the other zheslestids, these genera are more reminiscent of
Mistralestes
by the significant reduction of m3 and the apparent mesiodistal constriction of the trigonid. The wear pattern of the talonid seen in
Mistralestes
also evokes that of zhelestines (i.e. CCMGE 36/12000, the holotype of
Parazhelestes mynbulakensis
). However, p5 of the zhelestines is molariform with a separate metaconid and a cuspate paraconid (crestiform or absent in
Zhelestes
), and the metaconid of the molars is as high and as wide as the protoconid. More importantly, the mesiodistal constriction of the trigonid of m1 to m3 of zhelestids is mainly due to the reduced and crestiform aspect of the paraconid from m1 to m3 and the L-shaped aspect of the paracristid. In
Mistralestes
, the paracristid is rectilinear in m1−2 (certainly fused with the protocristid on m3) and the paraconid is twinned with the metaconid on m2 (and also probably on m3).
