Stratigraphy of the Haut Var Paleogene continental series (Northeastern Provence, France): New insight on the age of the ‘Sables bleutés du Haut Var' Formation

The age of the Paleogene deposits of the Haut Var (Provence, France) has been the subject of debate. Particularly, the „Calcaire à Bithynies‟ and the „Sables bleutés‟ units were ascribed either to the early Eocene or to the Oligocene. A stratigraphical clarification is required in order to precise the paleogeographical relationships of the Haut Var Paleogene sedimentary series with coeval deposits in the neighbouring southern Provence and Subalpine regions and other European domains. The study area is characterized by tectonically separated synclines and grabens filled in by continental Paleogene deposits. Detailed mapping and lithostratigraphical


Introduction
Page 3 of 62 A c c e p t e d M a n u s c r i p t In Provence, the Paleogene is represented by continental deposits filling in graben and basinal areas (Fig. 1). The most widespread facies correspond to lacustrine, palustrine, fluviatile, and terrestrial environments Durand and Nury, 1984). Paleogene deposits from Provence recorded significant climatic (Cojan and Moreau, 2006) and tectonic (Leleu, 2005;Leleu et al., 2009) events. During Paleocene, alluvial fans derived from the incipient Basse-Provence E-W trending tectonic structures (Lutaud, 1957) spread away into flood-plain areas. In the Basse-Provence region, the Pyrenean-Provence major compressive tectonic phase occurred during late Eocene times and was responsible for emplacement of the main overthrusts (Lacombe and Jolivet, 2005;Andreani et al., 2010;Espurt et al., 2012;Bestani et al., 2015).
During the Oligocene, a distensive regime led to the opening of the main Manosque, Marseille and Aix-en-Provence grabens that were filled in either by coarse detritic or by carbonate and evaporitic facies (Gigot et al., 1977;Hippolyte et al., 1991).
In the Aix-en-Provence, Marseille and Manosque basins, the distinction between Paleocene, Eocene and Oligocene series has been the subject of a wide consensus Durand and Nury, 1984) and correlations were established with the Languedoc , the Pyrenean realm (Plaziat, 1984), and the southern Castellane subalpine grabens (Roux, 1974;Giannerini, 1980Giannerini, -1981. However, precisions are still required concerning the Paleogene stratigraphy of northeastern Provence (Haut Var regional area) that has been greatly debated Touraine, 1973a;Angelier and Aubouin, 1973;Durand, 1984;Philip, 2012Philip, , 2013Plaziat, 2013aPlaziat, , 2013b. A clarification of this regional stratigraphy is needed for a better understanding of tectonic and paleogeographical questions, as well as for establishing correlations with other Paleogene basins. Likewise, clarification is also required concerning the respective age range of some fossil groups used as biomarkers in the Haut Var Paleogene continental facies, mainly gastropods (Lapparent, 1938a;Rey, 1966) vs. mammals Godinot, 1981).
The aim of this paper is to revisit the Paleogene stratigraphy of the Haut Var area and to shed light on the age of the conspicuous but controversial "Sables bleutés du Haut Var" Formation. Hence, the new stratigraphical framework will lead to provide a better reconstruction of the paleoenvironmental and paleogeographical evolution of this area during the Paleogene and will contribute to highlight new perspectives for more accurate correlations with other Paleogene Provence and west European basins.

Page 4 of 62
A c c e p t e d M a n u s c r i p t

Geological setting and historical background
The Haut Var region represents the northern part of the French Var department, including the Barjols, Tavernes, Rians, Salernes and Aups administrative cantons ( Fig. 1: insert). It appears as an intermediate geological domain located between the Basse-Provence fold and north verging thrust systems to the south, and the Alpine fold and south verging thrust front to the north. In the Haut Var area, the Paleogene extends in synclines and grabens bounded by normal faults or thrusts (Fig. 2). Paleogene strata overlie either conformably the Maastrichtian (Rognacian local stage) continental series, or unconformably cover the Lower Cretaceous-Jurassic basement.
The first synthetic approach about the stratigraphy of the Haut Var Paleogene series was due to Lapparent (1938b), while Mennessier (1959Mennessier ( , 1970 mapped the Tavernes and Salernes geological sheets at the 1/50,000 scale. From 1966 to 1978, Touraine reported a number of stratigraphical observations, synthetized in two contributions (Touraine 1973a(Touraine , 1976. The Ph.D. thesis of Angelier (1970) dealt mainly with the tectonics, but also provided precisions on the structure and stratigraphy of Paleogene synclines and grabens. Also Cornet (1973b), Cojan (1993) and Cojan et al. (2000) provided accurate data on the sedimentology and stratigraphy of Paleogene deposits. These different studies however led to numerous discrepancies concerning the Paleogene biostratigraphy and the tectonic interpretation of the Haut Var area.
Following the first attempt by Lapparent (1938b), Mennessier (1959) and Ginsburg and Mennessier (1967) subdivided the Haut Var Eocene based on mammal biostratigraphy. Indeed, mammalian fossils dating from the early Ypresian have been discovered in the "Sables bleutés du Haut Var" Fm. Nevertheless, Touraine (1968a) claimed that the mammal specimens were reworked and established new subdivisions on the basis of gastropod associations (Rey, 1966) that he considered as early Oligocene in age. These different interpretations regarding the stratigraphy also resulted in discrepancies about the age of the Haut Var tectonic structures.

Page 5 of 62
A c c e p t e d M a n u s c r i p t

Material and methods
Mapping of the Paleogene series was undertaken at the 1/12,500 scale. Stratigraphical sections have been logged at metre-scale or less when necessary. Samples were collected with a regular spacing in order to document their lithological and micropaleontological components.
Thin sections and polished slabs were prepared using standard methods in order to examine textures, cements, and lithoclastic and bioclastic components of limestones. Soft sediments were disaggregated in water and oxygen peroxide. The fine sediment was removed using 1.0 and 0.063 mm mesh sieves. Ostracodes and charophytes were hand picked under a light binocular stereoscopic microscope at ×40 magnification. Both macrofossils (vertebrates and gastropods) and microfossils (charophytes, ostracodes, foraminifers) have been used for establishing stratigraphic subdivisions.
The paleontological and sedimentological materials are archived at the CEREGE, Institut Pytheas, Aix-Marseille University, Campus Saint-Charles, Marseille. The Touraine collection, housed at the same address, has provided many interesting paleontological materials (especially gastropods) originating from the Haut Var Paleogene strata, that have been very useful for complementary data and comparisons.

The Maastrichtian-Danian boundary
The Upper Cretaceous lithostratigraphical subdivisions of the Haut Var area have been established for a long time by Lapparent (1938b) and Angelier (1974) and ascribed to the Begudian-Rognacian local stages (middle Campanian-Maastrichtian). In most basins, the upper Maastrichtian deposits (upper Rognacian local stage) grade transitionnaly upwards into those of the Danian. The Rognacian generally ends with a reddish silty marl unit ("Ensemble marnogréseux sommital" or "Marnes terminales" sensu Angelier, 1974). Dinosaurian eggshells have been reported only from the lower part of this stratigraphical unit (Touraine, 1963;Angelier, 1974). Accordingly, in respect to the Arc syncline (Garcia and Vianey-Liaud, 2001;Cojan and Moreau, 2006), the K/Pg boundary could be located in the upper part of this reddish silty marls unit. Nevertheless, biostratigraphical and chemostratigraphical analysis are still needed for a more accurate determination of its precise location.

Page 6 of 62
A c c e p t e d M a n u s c r i p t

"Calcaire à Microcodium" Formation
The upper reddish silty marls of the K/Pg transition are conformably overlain by the "Calcaire à Microcodium" Fm. A first detailed attempt to describe the stratigraphy of this formation was made by Angelier (1974), while Cojan (1993) and Djurkovic-Colson (1996) have provided meaningful sedimentological data. In the southwestern border of the Rians syncline (L"Olivière, Bois des Taillades), the lower part of the formation consists of a 2 m-thick nodular mottled paleosol overlain by a 5 m-thick vadose calcrete characterized by silty and sandy limestone, with Microcodium fragments and fine (millimetre thick) irregular silica beds (Djurdjevic-Colson, 1996). Clay minerals (palygorskite associated with smectite and illite/smectite) have been identified in connection with these facies (Cornet, 1974;Cojan, 1993;Djurdjevic-Colson, 1996). This calcimorphic pedogenesis has been used as a regional key-marker event (Cojan, 1993). The "Calcaire à Microcodium" Fm. widely extends in the Haut Var area, except in the Moissac, St Julien-le-Montagné and Ampus synclines where the formation is missing (Appendix B1).
To the north of the Rians syncline, a reddish gravelly breccia (including Microdium clasts and weathered quartz grains) overlies the "Calcaire à Microcodium". This peculiar facies ("Marnes graveleuses" sensu Angelier, 1974) constitutes a 1 to 3 m-thick, characteristic stratigraphical level that also occurs in the southern part of the La Mourotte syncline ( Fig. 2; Angelier, 1971). By comparison with the Arc syncline, and given its close relationship with the Page 7 of 62 A c c e p t e d M a n u s c r i p t underlying "Calcaire à Microcodium", the "Brèche à Microcodium" could be tentatively ascribed to the Danian.
West of Rians (Bardouine locality), the upper part of the formation is formed by red claystones-fine sandstones-siltstones with interbeds of coarse sandstones, conglomerates and lenticular palustrine marls and limestones. According to Sirugue (1962), andAngelier (1974), thick bird eggshells belonging to the Ornitholithus arcuatus oospecies occur in these facies.
The "Marnes à oeufs d"oiseaux" Fm. spreads away in all basins of the study area, with only some variations of thickness and lithology such as, for instance, metre-thick interbeds of Microcodium-bearing paleosols or breccias in the Montmeyan, Salernes, Pontevès, and Moissac synclines. In the Salernes syncline (Les Cadenières), the uppermost part provides numerous and well preserved thick bird eggshell fragments. A particular emphasis must be placed on the occurrence of a rich charophyte association ( Fig. 3; Appendix A1) of early to middle Thanetian age in the red gypsiferous marls that crop out at the middle part of the formation in the Moissac syncline (Fig. 2).
Stratigraphical correlations can be established with the Arc Basin where bird eggshells have been reported (Corroy and Touraine, 1961;Dughi and Sirugue, 1968), correlated with other biostratigraphical markers such as gastropods (Dughi and Sirugue, 1968;Durand and Nury, 1984) or charophytes (Feist-Castel, 1975). On the basis of the stratigraphical framework established in the Arc Basin (Dughi and Sirugue, 1968;Cojan et al., 2000;Buffetaut and Angst, 2014), the thick-shelled eggs (O. arcuatus) occur above the "Calcaire de Saint-Marc" Fm., in red M a n u s c r i p t siltstones underlying a lacustrine unit ("Calcaire de Langesse" Fm.) that contains Physa columnaris, considered as early Ypresian in age (Dughi et al., 1969). In addition, according to Cojan et al. (2000), The "Calcaire de Saint-Marc" Fm. could encompass the Paleocene Eocene thermal maximum (PETM), which occurred at the boundary between the Paleocene and Eocene. Therefore, O. arcuatus eggshells have been ascribed to the earliest Eocene. This dating is somewhat supported by the recent attribution of O. arcuatus eggshells from southern France to Gastornis, a giant bird which is known in Europe from the Selandian to the Lutetian (Buffetaut and Angst, 2014;Angst et al. 2015).
The stratigraphical interval of the thin-shelled eggs (O. biroi) is less well constrained than that of the thick ones. However, according to Sirugue (1962, 1968), and referring to the Arc Basin, O. biroi would be present in the red siltstones underlying the "Calcaire de Saint-Marc" Fm., considered as Thanetian in age in its lower part, yielding Physa prisca (Dughi and Sirugue, 1968), and recently referred to the Selandian by Leleu (2005), but without any biostratigraphical evidence.
Accordingly, in the Rians syncline the lower part of the "Marnes à oeufs d"oiseaux" Fm.
should be tentatively referred to the Selandian(?)-Thanetian and correlated with the marls underlying the "Calcaire de Saint-Marc" Fm. in the Arc Basin, whereas the uppermost part could be ascribed to the earliest Ypresian and correlated to the red siltstones overlying the "Calcaire de Saint Marc" Fm. Charophyte data suggest that the Paleocene-Eocene transition would be located within this formation (Feist-Castel, 1975), but it has not been identified in the Haut Var due to the lack of a stratigraphical equivalent of the "Calcaire de Saint Marc" Fm. and moreover because no accurate biostratigraphical and/or chemostratigraphical data have been reported there so far. Lapparent (1938) emphasized the great development of fluvial bluish micaceous sand deposits in the Haut Var synclines. He proposed the name of "Sables bleutés" facies for these peculiar deposits. He indicated that the "Sables bleutés" facies generally overlies a gastropod bearing lacustrine limestone unit that he named the "Calcaire à Bithynies". Both lithostratigraphical units were ascribed by Lapparent to the lower Eocene ("Sparnacian").

"Sables bleutés du Haut Var" Formation
The same names were used by different authors Touraine, 1968aTouraine, , 1971Mennessier, 1970;Angelier, 1971;Cornet, 1973a), although they did not agree about the Page 9 of 62 A c c e p t e d M a n u s c r i p t ages of these units. However, Touraine (1971) created the denomination of "Sables bleutés" lato sensu Formation including two distinct facies: the fossiliferous Bithynia bearing limestones ("Calcaire à Bithynies"), and the bluish micaceous sand deposits ("Sables bleutés" stricto sensu).
Indeed, our stratigraphical investigations and mapping have shown that the "Calcaire à Bithynies" is represented by carbonate lenses interfingered with the "Sables bleutés" stricto sensu at different levels of the series (see below). Accordingly, taking into account this sedimentological feature, and in order to avoid any ambiguity of definitions, we propose to replace the name "Sables bleutés" lato sensu by "Sables bleutés du Haut Var" Formation, which includes two main facies: the "Calcaire à Bithynies" and the "Sables bleutés" stricto sensu.
In the Bauduen area, the "Calcaire à Bithynies"crops out in the western part of the syncline, to the north of Saint-Barthèlemy farm (Cornet, 1980;Fig. 4

(A)). It was considered as
Page 11 of 62 A c c e p t e d M a n u s c r i p t stratigraphically located either at the upper part of the "Sables bleutés" Fm. (Lapparent, 1938), or at its base (Cornet, 1978(Cornet, , 1980Durand, 1984). Actually, a detailed mapping of the St Barthèlemy area (Fig. 4(B)) has shown that the "Calcaire à Bithynies" occurs as limestone lenses interbedded at different levels in the lower part of the "Sables bleutés" facies, some metres above its boundary with the "Marnes à oeufs d"oiseaux" Fm. Chardin 1922, emend. Euromys aff. thaleri (Michaux, 1964) in Escarguel (1999; Fig. 7(C)), characteristic of the lower Ypresian in the Paris basin. Besides, Lapparent (1938) described two new gastropod species, Bithynia bauduensis Lapp. and Melania (Tarebia) cardinalis Lapp., that he considered as characteristic of this lithological unit in the whole Haut Var area. However, this chronostratigraphical assignment was later questioned by Touraine (1966a;1968a) andRey (1966), on the basis of gastropods they collected in the "Calcaire à Bithynies" and attributed to the lower Oligocene. Due to the discovery at Montmeyan and La Mourotte of specimens ascribed by Rey (1966) to Potamides lamarcki and P. laurae (species referred in Provence and in the Paris basin to the early (not earliest)-late Oligocene; Nury, 1988), Touraine (1971Touraine ( , 1976 ascribed the "Calcaire à Bithynies" to the early Rupelian (Sannoisian local stage) and compared it to the same stratigraphical interval represented at the Butte Iouton area (near Beaucaire town, Gard department) by typical gastropod-rich layers, attributed for a long time to the early Rupelian (Roman, 1910), but more recently  to the late Priabonian.
A c c e p t e d M a n u s c r i p t

Discussion
Gastropods are abundant but relatively poorly preserved in the "Calcaire à Bithynies".
Only gastropod (Bithynia sp.) shells are present in the lower sandy marls unit I (Fig. 7(D)). As mentioned above, they were considered by Lapparent (1938a) as belonging to a new species (B. bauduensis), while according to Rey (1966) they are close to B. ugernensis Roman from the Priabonian of the Butte Iouton. Other gastropods (preserved as external casts) occur in the unit II Bithynia-bearing limestones of Montmeyan ( Fig. 7(E-J)) and La Mourotte (Fig. 7(K)) synclines.
These gastropod casts have been ascribed to only one species [Melania (Tarebia) cardinalis] by Lapparent (1938a: fig. 9 and pl. I, fig. 13), whereas Rey (1966Rey ( , 1967 identified among them various genera (e.g., Potamides, Tympanotonos, Melanoides, Melanopsis, Brotia) and species, but did not provide any paleontological description or illustration. We have retrieved the type locality of Melania (Terebia) cardinalis Lapparent; the gastropod-bearing bed is situated to the east of Montmeyan (Les Aubarèdes); it has provided a lot of shell casts (also present in the Touraine collection), all fitting well with Lapparent"s species description (Fig. 7(E-J)). Touraine (1976: pl. I, fig. 1, and in collection) figured a shell cast from the same locality that he ascribed to Potamides lamarcki. However, it is worth mentioning that the shell cast is incomplete (peristomal aperture not fossilized), whereas the outer shell sculpture is similar to that of Melania (Tarebia) cardinalis type specimens figured by Lapparent (1938a).
Despite these unresolved complex and contradictory taxonomic questions, Rey (1966) and some authors after him (Touraine 1968a;Angelier 1970) maintained that the gastropod taxa present in the "Calcaire à Bithynies" were characteristic of the Oligocene. Contrary to this conclusion, the results presented here show that typical Oligocene gastropod taxa are missing in this facies. Lapparent"s systematic approach provides a specific singularity of the "Calcaire à Bithynies" gastropod association, in better agreement with other biostratigraphical elements and with the peculiar paleogeographical isolation of the Haut Var area during the Paleogene that possibly enhanced the occurrence of endemic species.
In order to resolve the apparent biostratigraphical discrepancy (Oligocene vs. Eocene) between gastropods and mammals, Touraine (1966a) proposed that the rodent half-mandible of Euromys aff. thaleri in the Bauduen "Calcaire à Bithynies" was reworked, disregarding any taphonomic features. However, the absence of any reworking indices on the rodent bone such as iron coating or abrasion traces on the surface of the half-mandible should be emphasized.

Page 13 of 62
A c c e p t e d M a n u s c r i p t Moreover, teeth are still in place in their respective sockets (cf. Lapparent, 1938: figs. 10, 11;pl. I, fig. 14;Escarguel, 1999: pl. 5, fig. m;Fig. 7(C)). The texture of the limestone matrix of the half-mandible (Bithynia sp. opercula bearing wackestone) is very similar to those that characterize the "Calcaire à Bithynies" lenses in the Bauduen (St Barthèlemy) area (Fig. 4). To sum up our opinion, the Euromys aff. thaleri half-mandible was buried and fossilized in soft lacustrine carbonate sediments, probably after a slight displacement from a neighbouring terrestrial biotope located a short distance from the "Calcaire à Bithynies" deposits in the Bauduen area. Accordingly, we support the intrinsic biostratigraphical value of this mammal fossil as a biomarker of the Ypresian age of the "Calcaire à Bithynies" lenses interbedded in the lower part of the "Sables bleutés" facies in the Bauduen area. It is worth remembering here that the genus Euromys is restricted to the early to late Ypresian (Mammalian Paleogene levels MP7 to MP10) but that Euromys thaleri is only known from a few localities in the Paris Basin (Pourcy, Mutigny, Condé-en-Brie and Avenay) dated from the MP 8+9 (early to ?middle Ypresian) (Escarguel 1999). Dughi and Sirugue (1968) reported thick bird eggshells (O. arcuatus) in the lower sandymarls (unit I) of Rians "Calcaires à Bithynies" and attributed them to the early Ypresian.

Other biostratigraphical data
However, reworking of this fossil material from the underlying "Marnes à oeufs d"oiseaux" Fm. cannot be totally excluded. In the same unit, Cornet (1973b) mentioned a charophyte association composed of several species (Appendix A2) characteristic of the early Eocene from both Languedoc and Aix-en-Provence Basins (Castel and Grambast 1969). In La Fabresse area, unit I provides a monospecific charophyte association of Lamprothamnium priscum that has no precise chronostratigraphical meaning, and was first reported in the early Eocene of the Corbières (France). This species (and the whole genus) constitutes a prominent biomarker of brackish water environment during the whole Cenozoic. West of Rians (Bardouine locality), unit II provides a charophyte assemblage characteristic of the early Ypresian (M. Feist, written comm.; list of taxa in Appendix A2).
The foraminifer Rosalina bractifera from the Rians and La Mourotte synclines (Fig. 7(A,   B)) is particularly frequent in the Eocene strata of the Paris basin (Bignot, 1988). It was also M a n u s c r i p t reported by Poignant (1968) in the Rupelian of the Aquitaine basin, although later this occurrence was not confirmed by the same author (Poignant, 1982(Poignant, , 1995. As shown by microfacies (Fig. 6), ostracodes are relatively abundant in the limestones (unit II) of the "Calcaire à Bithynies", but devoid of any biostratigraphical interest. Soft sediments sampled in the lower part (unit I), next to the La Fabresse farm (Fig. 8(A)), have provided to Apostolescu and Dellenbach (1999) a heterochronous assemblage of ostracodes that these authors attributed to the Late Cretaceous, early Eocene and Chattian. These varied ages suggest reworking, and consequently poor chronostratigraphical interest. To sum up, other fossils provide contradictory results, but an early Eocene age is clearly indicated by the charophyte assemblage.

Field data
Sedimentary continuity and strict conformity of the "Calcaire à Bithynies" upon the underlying "Marnes à oeufs d"oiseaux" Fm. (Fig. 5(A)) is noteworthy for all localities where this stratigraphical setting occurs. In the Rians and Montmeyan areas, the uppermost levels of the "Marnes à oeufs d"oiseaux" Fm. gradually evolves into the sandy marls of the Unit I, without any sedimentary gap or transgressive surface, as wrongly postulated by Touraine (1968). In the Bauduen syncline, the "Calcaire à Bithynies" lenses pass gradually to the underlying lower micaceous sands of the "Sables bleutés" facies (Fig. 4).
These results rebut the hypothesis of a stratigraphical gap of about 20 m.y.including the late Ypresian, middle and late Eocene, and earliest Oligocene (early Rupelian)that, according to Touraine (1971), would correspond to the "Marnes à oeufs d"oiseaux"/"Calcaire à Bithynies" boundary. This gap would have been around 24 m.y. if the Bithynia bearing sandy marls (Unit I) were ascribed to the Chattian, as postulated by Apostolescu and Dellenbach (1999). As a consequence, with the absence of such a gap there is no need to propose the hypothesis of tectonic movements in the Rians and Salernes synclines (Touraine, 1969;Angelier, 1970) in order to explain a sedimentation break lasting from the late Thanetian until the early Oligocene.
Our study shows there is a slight diachronism of the "Calcaire à Bithynies" corresponding basically to a lateral and coeval facies change with the lower part of the "Sables bleutés" facies.
Two stratigraphical patterns must be highlighted (Fig. 4) A c c e p t e d M a n u s c r i p t the "Sables bleutés" facies in the Bauduen syncline. If the Euromys aff. thaleri half-mandible found in the "Calcaire à Bithynies" at Bauduen is not reworked, as taphonomic features indicate, it would provide an early Ypresian age for the lower part of the "Sables bleutés du Haut Var" Fm.
This agrees with the mammalian association of Rians, also of early Ypresian age Godinot, 1981).

Paleoenvironmental interpretation
Sedimentary and paleontological data allow interpreting the "Calcaire à Bithynies" facies as formed in freshwater to brackish environmental conditions that probably prevailed in areas protected from fluvial outlets. In the southwestern part of the Haut Var (Rians syncline) these conditions have contributed to enhance deposition of alternating shallow lacustrine carbonate vs.
laminated clayey and cherty sediments and development of benthic lacustrine associations composed of gastropods, ostracodes, charophytes, and the epiphytic, euryhaline foraminifer Rosalina bractifera (Bignot, 1988). In the northwestern part of the Haut Var (Montmeyan, la Mourotte; Fig. 2), shallow brackish water conditions led to the deposition of alternating laminated carbonates, gypsum beds, ligniteous marls, and gastropod-bearing limestones. In La Mourotte and Bauduen areas, high-energy conditions (storms?) in a shallow water environment (shoreface?) prevailed and were responsible for the formation of lumachellic deposits. The "Calcaire à Bithynies" is missing particularly in the eastern part of the Haut Var (Salernes, Pontevès areas; Fig. 2), where there was a permanent input of siliciclastic fluviatile sands during the stratigraphical interval of its deposition.

Mineral composition
The bulk of the sands is composed of quartz and micas (muscovite and biotite), these latter amounting to 42% on average. According to Cornet (1976), the average composition of the heavy minerals fraction is: Garnet 50% to 25%; Staurotide 25 to 30%; Anatase 6 to 9%; Disthene 4%; Tourmaline 3 to 5%; Zircon 2 to 3%; Rutile 1 to 2%; Sphene 1 to 2%; Epidote 1%; Sillimanite 1%; and Chloritoid 1%. This mineralogical association is characteristic of the micaschist series that crop out in the metamorphic Hercynian Maures massif, located southeastward of the Haut Var area (Fig. 1). Accordingly, a southeastern origin is generally postulated for the detritic supply that built up the "Sables bleutés" facies. In support of this assumption, it is worth noting a high proportion of coarse sands and the frequence of conglomeratic beds (yielding granite, metamorphic rocks and quartz pebbles) in the eastern part of the Haut Var (Salernes syncline).

Stratigraphical subdivisions of the "Sables bleutés" facies in the Rians syncline
Owing to poor outcropping, no section of the "Sables bleutés" facies was logged until now in the Haut Var area. However, detailed observations and mapping, carried out in the Rians syncline, made it possible to provide a general stratigraphical overview of it.
Lower part. It is relatively well exposed to the east of Rians (Fig. 8(A)), between La Fabresse and La Neuve farms, and along the country lane of La Désidère farm. At la Fabresse, blue micaceous sands rest upon the "Calcaire à Bithynies". East of La Fabresse, the "Calcaire à Bithynies" pinches out, which results in the direct superimposition of the "Sables bleutés" facies on the "Marnes à oeufs d"oiseaux" Fm. in the whole eastern part of the Rians syncline, with gradual and conformable transition between both units. South of La Fabresse, the "Sables bleutés" facies contains the rich mammalian fauna of Rians , Godinot, 1981. The fossiliferous bed (43° 36" 36. 5" N; 5° 47" 34. 7" E) is located roughly 160 m above the top of the Page 17 of 62 A c c e p t e d M a n u s c r i p t "Calcaire à Bithynies" (Fig. 8(A, B)). Pinkish sandy marls and fine laminated sandy and silty beds overlying the mammal fossil layer have provided an ostracode association (Touraine, 1968b).
Eastwards, at 1.250 km (La Désidère; 43° 36" 20.87" N; 5° 48" 26.66" E; Fig. 8(A)) there are 10 m-thick grey or reddish micaceous sands overlain by 8 m-thick laminated silty beds of alternating grey, reddish or green colours that yielded a rich ostracode fauna (Fig. 9). The sequence is ended by a thin limestone bed (20 cm-thick) displaying a nodular and vuggy structure typical of palustrine paleosols. It is suggested that the peculiar La Désidère sequence may be correlated with the La Fabresse mammal and ostracode-bearing "Sables bleutés" beds. Overlying the la Désidère limestone bed, about 150 m of poorly stratified medium to coarse bluish micaceous sands crop out in bad conditions. The thickness of the lower part of the "Sables bleutés" facies is roughly estimated to be 275 m at the western Rians area.
Upper part. This part is poorly exposed, but displays some discontinuous exposures in

Lateral variations
Salernes syncline. The "Sables bleutés" facies widely crops out in this area. Contrary to the assertion of Touraine (1969), it conformably overlies without any gap the "Marnes à oeufs d"oiseaux" Fm. (Barbebelle, Les Esparus, La Cadenière) and is formed by cross-bedded bluish coarse micaceous sands, including numerous conglomeratic lenses (composed of quartz and metamorphick rock pebbles; Fig. 5(E, F)), and scarce fine sandy and silty beds (Les Esparus) yielding ostracodes. The "Sables bleutés" facies have been cored by several drillings for bauxite exploration (Durand et Mennessier, 1964) showing a thickness of about 100 m that could be representing only a part of its total thickness. A c c e p t e d M a n u s c r i p t fine-laminated bluish micaceous sands in which with a 1.5 m-thick laminated limestone lens interbedded 15 m above the boundary with the underlying "Calcaire à Bithynies" facies. The lens is composed of gastropod (Bithynia sp.) coquinas, alternating with ostracode-rich wackestones ( Fig. 6(A, B)). In the Montmeyan syncline, the "Sables bleutés" facies does not exceed 50 m in thickness.

Biostratigraphical data
Vertebrates. According to Ginsburg et al. (1967), the Rians (La Fabresse) fossiliferous locality contains a diverse and well-preserved vertebrate fauna composed of teeth, jaws and various bones belonging to mammals, reptiles (turtles and crocodilians) and amphibians. Godinot (1981) documented that the Rians mammalian fauna is unbiased, diverse (about 30 species and 25 genera), and rich in small condylarth Louisinidae, a family known from the mid-Paleocene to the early Eocene of Europe (Hooker and Russell 2012). Ginsburg et al. (1967) have pointed out the affinity of the mammalian association with the assemblages collected in the Sparnacian facies of England, Belgium and especially the Paris Basin. According to Godinot (1981) the Rians mammalian association occurs, in terms of evolutionary trends, near the earliest Eocene MP7 (Dormaal reference-level, Belgium). The most recent evaluation by Marandat et al. (2012), considers also that the Rians mammalian fauna is Ypresian in age and located within the MP7/MP8+9 interval.
A taphonomic study of the 6 to 7 cm-thick vertebrate-bearing bed was carried out by Godinot (1979) who attributed it to deposition in a fluvial environment characterized by Page 19 of 62 A c c e p t e d M a n u s c r i p t bioturbated and weathered red-brown marls and clays (riverbank deposits) and coarse or fine sands (channel deposits). Godinot concluded that there was no reworking of the vertebrate association, in contrast to Touraine"s (1968a)

assumption.
A second mammalian locality has been discovered in the "Sables bleutés" of Salernes (Ginsburg and Mennessier, 1973). It yielded two phalanges belonging to a phenacodontid "condylarth", a mammalian family which is known from the late Paleocene through the middle Eocene in North America and Europe (Thewissen, 1990). A third vertebrate locality, recently discovered near Esparron (Font Couverte locality), yielded fragmentary remains of crocodiles, turtles, mammals, and possibly birds (Buffetaut et al. 2016); some elements (notably a crocodile vertebra and the turtle association) suggest an Eocene age.
Ostracodes. In the Rians syncline, Touraine (1972)  and Cyprideis sp. were found, whereas sandy sediments sampled at the lower part of the "Sables bleutés", to the east of the Salernes syncline (Les Esparus), provided adult ostracode carapaces ascribed to Candona sp. and Ilyocypris sp. (Fig. 9). Unfortunately, no discriminant biostratigraphical data can be deduced from this micropaleontological association.

Spores and pollen.
A palynological investigation of the "Sables bleutés" facies was carried out in the central part of the Bauduen syncline (Cornet, 1973a) and provided a homogeneous palynological association investigated by J.J. Châteauneuf (Appendix A3).
According to the latter author, palynological spectrum supports the mammalian data from Rians and indicates without ambiguity an early Eocene age.
Nannoplancton. Nury and Touraine (1972) reported from the lower part of the "Sables bleutés" of Montmeyan (St Maime section) a nannoplancton assemblage (four species, determination Carla Müller, unfortunately not illustrated; Appendix A4) that they considered as M a n u s c r i p t autochtonous and ascribed to the early Rupelian. However, the autochtony of this nannoplancton assemblage can be questioned owing to the fluvial depositional setting of the "Sables bleutés" in the St Maime area (Fig. 5(D)). Moreover, three species of the nannoplancton assemblage have a large stratigraphical range including the early Eocene while only one of the species, Transversopontis zigzag, is of Oligocene age. Taking into account these contradictions, it seems difficult to draw any biostratigraphical conclusion from the results by Nury and Touraine (1972). Touraine (1968aTouraine ( , 1968b and Deschamps and Touraine (1969) interpreted the "Sables bleutés du Haut Var" Fm. as a transgressive deposit overlying a lithological unit (named "Argilites ferrugineuses") that they considered as an Oligocene paleosol deposit, unconformably resting on the "Marnes à oeufs d"oiseaux" Fm. Moreover, an angular unconformity was postulated by Touraine (1969Touraine ( , 1971 to occur between the "Sables bleutés du Haut Var" Fm. and the underlying Paleocene formations in some areas such as Salernes and the Pontevès-Sillans synclines. In Touraine"s hypothesis, the unconformity and the underlying paleosol were strong arguments supporting the Oligocene age of the "Sables bleutés" facies since they were interpreted as representing an important stratigraphical gap related to a pre-Rupelian tectonic phase. The "Argilites ferrugineuses" paleosol would correspond in their view to a long period of nondeposition and exposure.

Field data
Observations and mapping undertaken in the present study allow us to reject Touraine"s assumption. Indeed, in all the Haut Var synclines, the "Sables bleutés du Haut Var" Fm.
The angular unconformity reported by Touraine (1969Touraine ( , 1971) between the "Sables bleutés" facies and the underlying "Marnes à oeufs d"oiseaux" Fm. in the Salernes and Pontevès-Sillans synclines has not been confirmed in any further study. For instance, in the southern border Page 21 of 62 A c c e p t e d M a n u s c r i p t of the Salernes syncline (Les Esparus, Les Cadenières), the "Sables bleutés" facies conformably overlies the "Marnes à oeufs d"oiseaux" unit, both displaying the same 20° N dip (Fig. 5(F)).

Depositional environment
Since the studies of Lapparent (1938aLapparent ( , 1938b, most authors (Mennessier, 1959;Vogt-Di Poppa, 1968;Cornet, 1973aCornet, , 1977Godinot, 1979) have concluded that the "Sables bleutés" facies was of fluvial origin. Only Touraine (1973bTouraine ( , 1973cTouraine ( , 1976 related it with a marine environment. According to Vogt-Di Poppa (1968), the deposition of the micaceous sandy beds occurred as a result of large alluvial sheet floods and fluvial crevasse-splay processes. Indeed, new sedimentological investigations and mapping showed that this facies is linked to a fluvial system characterized by sandy deposits with channel cross-bedding that alternate with alluvial plain plane-beds ( Fig. 5(D)). This interpretation is further supported by the paleontological content of the "Sables bleutés". Godinot (1979) showed that the vertebrate assemblage found in the lower part of the "Sables bleutés" in the Rians syncline is composed of species living on riverbanks and along streams. The animals, belonging to different taxa and different environments (terrestrial and freshwater biotopes) have probably been transported post-mortem, mixed and concentrated by a fluvial flood in an oxbow pond. Afterwards, the bone remains were buried by riverine fine sand deposits. The ostracode association, composed of Neocyprideis sp., Cyprideis sp., Candona sp. and Ilyocypris sp., strengthens ecological conditions close to fresh-or brackish-water, compatible with a fluvial system encompassing lakes, ponds and lagoons with carbonate and evaporitic sedimentation ("Calcaire à Bithynies").
Otherwise, the homogeneous clay-mineralogy led Cornet (1973aCornet ( , 1977 to claim deposition of the "Sables bleutés" facies in a single alluvial sheet, spreading out on a flat paleosurface affecting the whole Haut Var area, which was secondarily broken up by tectonic movements and dissociated by erosion. Our data complement Cornet"s conclusion. Instead of a flat paleosurface the "Sables bleutés" facies would be rather deposited in a subsident foreland basin surrounded by incipient emergent paleostructures (Fig. 10) that afterwards evolved as northward or southward verging thrusts during the late Eocene Pyrenean-Provence compressional tectonic phase. The Durance High was inherited from the mid-Cretaceous paleogeography that led to the uplift of the so-called "Bombement durancien" (Masse and Philip, 1976) on which bauxite formed. The Greoux High was postulated by Morabito (1967) as an Upper Cretaceous-

Page 22 of 62
A c c e p t e d M a n u s c r i p t Eocene uplift, confirmed by deep petroleum drillings (Dubois and Curnelle, 1978). The Subalpine Land (the present Canjuers Jurassic plateau) was emerged as early as the Late Cretaceous (Lapparent, 1938;Kerckhove, 1978). The southern part of the foreland basin was probably bounded by the E-W trending Sambuc-Pallières and Bessillons-Salernes flexural fronts that acted as incipient paleostructures as early as the Maastrichtian and the Danian (Angelier, 1974). The main subsident foreland area would correspond to the Rians syncline where up to 400 m of micaceous sands and conglomerates settled in a narrow E-W trending depocenter located at the footwall of the Sambuc-Pallières thrust.

Location and historical background
In the Haut Var area, the Oligocene continental deposits are only represented in Bourdas, Plan d"Auron, Gigery, La Combe grabens, and locally in the Chargeaire, Pierrette, and Bellevue areas (Fig. 2). In contrast with the debate about the age of the "Sables bleutés du Haut Var" Fm., a general consensus arose about the Oligocene age of these deposits (the Bourdas facies sensu Touraine, 1971). However, a discussion took place concerning their stratigraphic correlations with the "Sables bleutés du Haut Var" Fm. Hence, according to Touraine (1971) andAngelier (1971), the Oligocene would present two coeval facies: the Bourdas facies and the Montmeyan facies (corresponding to the whole "Sables bleutés" facies and "Calcaire à Bithynies"), whereas Mennessier (1970) disagreed with this hypothesis and stated that the Bourdas and Montmeyan facies were diachronous. He assigned the first to the Oligocene, while the second would correspond to the lower Eocene. These results are discussed below.

Page 23 of 62
A c c e p t e d M a n u s c r i p t

Les Olagniers type-section
Unit I. It is formed by an argillaceous limestone bed unconformably resting on the Lower Cretaceous basement, passing upwards to coarse heterogeneous conglomerates (Fig. S1(A)) that contain a high proportion of pebbles and cobbles of local origin (Jurassic and Lower Cretaceous limestones), while quartz pebbles are relatively scarce.

Unit II.
Formed by about 100 m-thick sandy marls and siltstones that contain in the upper part gastropod shell fragments and gypsum crystals occurring in poor outcropping conditions. Unit III. Characterized by an alternation of gypsiferous and fossiliferous sandy marls, sandstones and finely laminated limestones (Fig. S1(B)). Fossil assemblages encompass gastropods, ostracodes (Neocyprideis sp.), and charophytes.

L"Américaine section
Unit I is represented by an alternation of conglomerates (similar to those of the Les Olagniers section), argillaceous limestones and fossiliferous limestone lenses containing gastropods and small bivalves (Pisidium sp.). Only the lower part of Unit II crops out in this section.

La Combe half graben
At the eastern end of the Rians syncline, the Oligocene deposits extend in the narrow N-S trending La Combe half graben parallel to the western faulted border of the Barjols Triassic uplift (Fig. 2). Initially ascribed to the Danian (Lapparent, 1938), the deposits were later attributed to the Eocene (Mennessier, 1966), then to the Oligocene (Touraine, 1967b;Angelier, 1971).
The Oligocene deposits are represented by Unit I and the lower part of Unit II only (Fig.   11). In the soutwestern part of the half graben, Unit I unconformably rests on the "Sables bleutés" facies while it lies on the Upper Cretaceous or Jurassic basement in the northwestern part (Angelier and Aubouin, 1976). A well exposed stratigraphical section crops out along the road D561 with complement along the old railway from Rians to Varages. Unit I is characterized by alternating beds of conglomerates and coarse red sandstones. Conglomeratic beds mainly encompass pebbles and cobbles made of Jurassic limestone (Fig. S1(C)) mixed with scarce M a n u s c r i p t pebbles of Triassic cargneules (Fig. S1(D)), quartz, and metamorphic rocks. Pebbles of gastropod-rich limestones ascribed to the Oligocene have been mentioned by Angelier (1971) at the upper part of the conglomeratic sequence.
At the Engaraude locality, close to the northeast flank of the La Combe half graben, the lower part of the conglomeratic sequence includes a limestone lens (Angelier, 1971;Angelier and Aubouin, 1973) that yields a rich gastropod association referred to as Striatella horizon, which was considered by Rey (1966) and Nury (1988) as characteristic of the earliest Rupelian.
However, Touraine (1971) interpreted the Engaraude limestone lens as collapsed from a hypothetic "Striatella limestone" resting on the nearby Triassic Barjols tectonic unit. This interpretation cannot be supported since such limestone does not exist in situ in the eastern part of the Rians syncline.
The Engaraude gastropod-bearing limestone is located east of the Montmayon domain (43° 36" 28.37"" N; 5°55" 15.58"" E; alt. 450 m). It is represented by a 2 m-thick, 100 m-wide limestone lens conformably underlying the conglomeratic Unit I (Fig. 11) that laterally overlaps the Jurassic basement. The carbonate lens is formed by alternating gastropod-bearing wackestones-floatstones and vuggy carbonate crusts ( Fig. S1(E, F)). Field data suggest shallow lacustrine-palustrine carbonate sedimentation followed by diagenetic processes leading to the formation of the Engaraude limestone, preceding the La Combe conglomerate deposition.

Gigery half graben
The Gigery half graben is located along the western Upper Jurassic-Triassic border of the Barjols Triassic uplift from which it is separated by a roughly N-S major fault (Fig. 2). The sedimentary infilling of the Gigery half graben was successively ascribed to the Danian (Lapparent, 1938), the Paleocene-early Eocene (Mennessier, 1966) and finally the Oligocene (Touraine, 1967b;Angelier, 1971). The Oligocene series overlies unconformably either on the Upper Jurassic basement in the northwestern part of the structure or on the uppermost Cretaceous one (Rognacian) in its southern part.
A detailed stratigraphical section can be studied in the northern part of the half graben, near the Gigery farm (Fig. 11). The lower part of the sequence displays lenticular beds of conglomerates alternating with coarse sandstone and red siltstone beds. These facies are referred to the Unit I of the Les Olagniers type-section. The conglomerates contain a high proportion of Page 25 of 62 A c c e p t e d M a n u s c r i p t pebbles and cobbles of Jurassic limestone, with minor quartz pebbles. Some conglomerate beds contain reworked boulders of gastropod limestones (Fig. S2(B)) referred to by Angelier (1971) as the "Calcaire à Striatelles" facies. The Unit I grades upwards to white-greenish marls probably corresponding to the lower part of the Unit II as observed in the Les Olagniers section. The upper part of the marly Unit II crops out poorly and forms the weathered and colluvium-covered topographic depression that extends to the east of the Gigery farm.

Stratigraphy
Gastropods are abundant in the limestone lenses linked to the conglomerates of the Bourdas and La Combe (Engaraude) areas. The gastropod association (Angelier, 1971;Touraine, 1967b and in collection; Philip, in collection; Appendix A5) is characteristic of the lowermost Rupelian, Striatella horizon, well represented in the Marseille Oligocene Basin (Nury, 1988). It is worth noting that the pebbles of gastropod-bearing limestone in the upper part of the La Combe or Gigery conglomerates contain the same fossils as the autochtonous carbonate lenses of Bourdas and La Combe areas (Angelier, 1971). Hence, they can be considered as reworked from underlying Striatella-bearing limestone deposits.
To sum up, our results confirm Mennessier"s stratigraphical interpretation. No correlation can be established between the Bourdas facies and the Montmeyan facies (sensu Touraine, 1971), since they are diachronous. The gastropod association that characterizes the limestone lenses interbedded in the Bourdas conglomerates is biostratigraphically younger that those of the "Calcaire à Bithynies". The Bourdas limestone lenses belong to the early Oligocene. Their deposition and that of the coeval conglomerates occurred later than the "Sables bleutés" facies, as attested in the La Combe half graben. Additionally, the Bourdas facies corresponds to a characteristic and well-dated conglomerate-dominant detrital series that can be assimilated to a stratigraphical formation of Oligocene (Rupelian) age in the Haut Var area.

Synthesis and conclusions
Our results allow to establish a new chronostratigraphical framework of the Haut Var Paleogene continental series and to make stratigraphical correlations at the regional scale (Fig.   12). The K/Pg boundary may lie in the red silty marls underlying the "Calcaire à Microcodium" Fm. ascribed to the Danian (Vitrollian local stage). Alluvial fans ("Brèches à Microcodium" Fm.) M a n u s c r i p t interbedded in the red siltstones of the upper Danian were originated during an incipient uplift affecting the southern part of the Haut Var area in relation with the northward verging Sainte-Victoire tectonic compressional system (Espurt et al., 2012). Later, the Selandian(?)-earliest Ypresian "Marnes à oeufs d"oiseaux" Fm. transgressively spread out in all Haut Var synclines, resting either conformably on the "Brèches à Microcodium", or unconformably on the Lower Cretaceous-Upper Jurassic basement.
Upward in the stratigraphic record, the "Marnes à oeufs d"Oiseaux" Fm. is covered conformably either by the "Calcaire à Bithynies" or the "Sables bleutés" facies, without any gap or tectonism contrary to what was wrongly postulated by Touraine (1971, 1976) and Angelier (1971. New sedimentological investigations have shown the absence of any reworked mammalian remains in the "Calcaire à Bithynies" and "Sables bleutés" facies, which supports a lower Ypresian age for these units, as previously proposed by Teilhard de Chardin and Lapparent (1933), Ginsburg et al. (1967), Godinot (1981), and Escarguel (1999). New analysis or reinterpretation of previous data concerning charophytes, foraminifera, ostracodes and palynoflora agree with this chronostratigraphical conclusion. The gastropod association present in the "Calcaire à Bithynies" dramatically differs from the one that characterizes the "Calcaire à Striatelles" horizon, typical of the lower Oligocene in both the Haut Var and Basse Provence basins.
Our results confirm the existence of a major stratigraphical gap, corresponding to the Lutetian-Priabonian interval, in the Haut Var area. This supports Lapparent"s (1938b) and Mennessier"s (1959)  A c c e p t e d M a n u s c r i p t roughly E-W trending compressional foreland basin surrounded by emerged areas and tectonic highs (Fig. 10). These paleogeographical features contributed to the relative isolation of the Haut Var area during early Ypresian times and probably enhanced episodes of brackish water or evaporitic sedimentation. Likewise, restricted paleoenvironmental conditions and relative biogeographical isolation could be invoked as responsible for the peculiar features shown by the Ypresian Haut Var gastropod associations. However, detailed systematic revision is needed in order to compare these latter with those of other coeval European sedimentary basins. Future studies will also need to concentrate on a chemostratigraphical framework of the "Sables bleutés du Haut Var" Fm. established on the hyperthermal events that occurred during the early Eocene.