The early/middle Eocene transition at the Ésera valley (South-Central Pyrenees): Implications in Shallow Benthic Zones (SBZ)

Y/L boundary to be recognised, the taxa found are enough to support the chronostratigraphic attribution. Data obtained in the Ésera valley section has improved the knowledge of larger benthic foraminifera ( Nummulites and Assilina ) distribution through chron C21. SBZ 11 to SBZ 12 transition took place at the lowermost C21r, as shown in previous works. SBZ 12 assemblages extend into C21n, where the SBZ 12 to SBZ 13 boundary occurs. These data, obtained in shallow marine siliciclastic facies, with in situ fauna, results in a shift of the SBZ 12/SBZ 13 boundary to the Lower Lutetian, younger than previously believed. Accordingly, the Ypresian/Lutetian boundary occurs in SBZ 12.


INTRODUCTION
The Ypresian/Lutetian (early/middle Eocene) boundary is defined by the Global Stratotype Section and Point (GSSP) of the base of the Lutetian Stage in the Gorrondatxe section (Molina et al., 2011). This section is located in the Gorrondatxe Beach (Biscay synclinorium, Basque Country, Spain), within the Basque-Cantabrian Basin (Fig. 1) that has been generally referred to as the Western Pyrenees (see Barnolas and Pujalte, 2004). Previous works of the Ypresian/ Lutetian Boundary Working Group (ICS) carried out on the Paleogene Pyrenean realm (South-Central Pyrenees and the Basque-Cantabrian Basin) preceded the selection of the Gorrondatxe section for its nomination Orue-Etxebarria et al., 2006, 2009Payros et al., , 2007Payros et al., , 2009aPayros et al., , b, 2011. The Gorrondatxe section includes a deep marine sequence of marls and pelagic limestones alternating with thin-bedded siliciclastic and calciclastic turbidites (Molina et al., 2011;Payros et al., 2009a).
The GSSP for the base of the Lutetian Stage was defined in a dark marly level characterized by the First Occurrence (FO) of Blackites inflatus (CP12a/b boundary) in the An integrated study including magnetostratigraphy, larger benthic foraminifera and calcareous nannofossil biostratigraphy is presented herein. This work was performed in shallow marine siliciclastics rich in larger foraminifera, around the Ypresian/Lutetian boundary in the Ésera valley (South-Central Pyrenees). Although the calcareous nannofossil content in the studied interval is low, not allowing a precise Y/L boundary to be recognised, the taxa found are enough to support the chronostratigraphic attribution.
Data obtained in the Ésera valley section has improved the knowledge of larger benthic foraminifera (Nummulites and Assilina) distribution through chron C21. SBZ 11 to SBZ 12 transition took place at the lowermost C21r, as shown in previous works. SBZ 12 assemblages extend into C21n, where the SBZ 12 to SBZ 13 boundary occurs. These data, obtained in shallow marine siliciclastic facies, with in situ fauna, results in a shift of the SBZ 12/SBZ 13 boundary to the Lower Lutetian, younger than previously believed. Accordingly, the Ypresian/Lutetian boundary occurs in SBZ 12. A xi a l Z o n e C a t a la n C o a s t a l R a n g e s FIGURE 1. Simplified geological map of the Pyrenees including locations cited in text. 1) Besians and La Puebla de Fantova sections (this work); 2) San Pelegrín section (Rodríguez-Pintó et al., 2013); 3) Gorrondatxe section Molina et al., 2011;Orue-Etxebarria et al., 2006, 2009Payros et al., 2007Payros et al., , 2009a; 4) Otsakar section  Gorrondatxe section, located in the middle of Chron 21r and represents a maximum flooding surface with a presumed global extent (Molina et al., 2011). However, B. inflatus shows up at the base of C21n and very close to Nannotetrina cristata in some oceanic drills in the Pacific Ocean (Agnini et al., 2006;Backman, 1986;Norris et al., 2014).

Ripoll
In the Eocene shallow marine sequences, larger foraminifera are key forms for biostratigraphy. Their usage, since the pioneering studies in the 19 th century, increased after the contributions made by Hottinger (1960) and Schaub (1981), and the definition of the Shallow Benthic Zones (SBZ) by Serra-Kiel et al. (1998). Nevertheless, most of the GSSPs for the Paleogene System have been defined after Serra-Kiel et al. (1998) work. That is the case of the Ypresian/Lutetian boundary that shifted from its previous position at the C22/C21 boundary (see Luterbacher et al., 2005) to its newly defined GSSP location within C21r (Molina et al., 2011). For this reason, it is necessary to review how the SBZs confront the new definition of Stage boundaries.
In the North Atlantic Ocean and western European successions, the Ypresian/Lutetian boundary is usually hampered by stratigraphic gaps (Aubry, 1995;Molina et al., 2011;Payros et al., 2009b). In the South Pyrenean realm, the occurrence of erosional gaps around the Y/L transition was described first by Payros et al. (2009b) and evidenced later in new sections Rodríguez-Pintó et al., 2013. In the San Pelegrín section (Southern Pyrenees) (Fig. 1), which belongs to the foreland shallow carbonate platform, the Ypresian/Lutetian boundary is absent due to an erosional gap that includes at least all C21r (Rodríguez-Pintó et al., 2013).
In the Gorrondatxe section, larger foraminifera appear reworked and transported from shallow water zones, which makes it difficult to assign an accurate zone attribution using the SBZ scale. Around the Y/L boundary, larger foraminifera are scarce and undoubtedly SBZ 13 forms occur within C21n Molina et al., 2011;Orue-Etxebarria et al., 2006), which also happens in the Otsakar section  located as well in the Western Pyrenees (see Fig. 1). In the last two decades, the SBZ 12/SBZ 13 transition has been used as the Ypresian/Lutetian boundary in shallow water facies with larger foraminifera and absence of pelagic markers . In the Ésera valley in the South-Central Pyrenees, where the Cuisian (upper Ypresian) parastratotype was defined (Schaub, 1992), there is a continuous shallowmarine siliciclastic sequence rich in shallow benthic faunal assemblages of Cuisian and Luterian age (Schaub, 1981;Tosquella, 1995;Tosquella and Serra-Kiel, 1998). This section appears as an alternative locality to study the Y/L transition in shallow marine settings with abundant larger foraminiferal faunas. The results of this study, including magnetostratigraphy, larger foraminifera and calcareous nannofossils, are presented in this work.

GEOLOGICAL SETTING
The studied area is located in the Ésera valley in the south-central Pyrenees (Fig. 1). The Eocene succession in the Ésera valley belongs to the Tremp -Graus Basin. At the studied time interval, the Tremp-Graus Basin was carried piggyback (Atkinson, 1986;Atkinson and Elliot, 1985) on the South-Central Pyrenean Unit (Muñoz et al., 2013;Séguret, 1972).
Shallow marine, transitional and terrestrial siliciclastic systems represent the main facies assemblages filling the Tremp-Graus Basin (Barnolas et al., , 2019Chanvry et al., 2018;van Eden, 1970;Nijman, 1998;Nijman and Nio, 1975, and references therein). Shallow marine carbonates are scarce in the piggyback sequence, reduced to some transgressive horizons in the Ypresian and early Lutetian successions out of the studied interval (Fig. 2).
The Puebla de Fantova and Besians sections are located in the middle part of the Ésera valley (Eocene) succession ( Fig. 3), which remained unsampled or poorly sampled for magnetostratigraphy, as exposed by Payros et al. (2009b) (Morillo de Liena -Santa Liestra section). They are stratigraphically younger than the Navarri section (Bentham and Burbank, 1996) (BB-1 in Fig. 3), in the lowermost Eocene units outcropping in the upper Ésera valley, below the Campanúe conglomerates. Compared to the Santa Liestra, Mesón de Pascual, and Grustán sections of Bentham and Burbank (1996) (BB-2, BB-3 and BB-4, respectively in Fig. 3), located in the lower Ésera valley, the Puebla de Fantova and Besians sections are stratigraphically older except for the Santa Liestra section that is equivalent to the uppermost part of the La Puebla de Fantova section and the lowermost part of the Besians sections.
To the south, between the Isábena and Ésera valleys, deltaic facies of the Perarrúa Fm., with dominant paleocurrents to the west and northwest, overlie the Lower Campanúe conglomerate (López-Olmedo and Ardèvol, 2016;Teixell et al., 2016). These deltaic facies laterally grade southwards to and are overlain by prodeltaic and open marine marls (Teixell et al., 2016). This deepening sequence, developed over the Lower Campanúe conglomerate, appears as Middle Perarrúa unit in this work (Fig. 2).
The upper boundary of the Perarrúa Fm. consists of an abrupt transition to the fluvial red bed facies of the Capella Fm. (Cuevas-Gozalo, 1989;Nijman and Nio, 1975) (Fig. 2). To the north (Ésera valley outcrops), this transition follows a sharp irruption of coarse alluvial fan conglomerates (upper prograding wedge of Santa Liestra-2 of Crumeyrolle, 1987) (Fig. 2), which results in a complex sequence of conglomerates, sand bars and bioclastic sands that include coralgal limestones, interpreted as barrier-bar complexes (Nijman and Nio, 1975).

MATERIALS AND METHODS
The Puebla de Fantova and Besians sections were logged and sampled for this study. The Puebla de Fantova section is located along the local road to Centenera village (Fig.  3). The base of this section corresponds to the delta front facies of the Middle Perarrúa unit at the Casa del Molino bridge (cartographic unit 22 of Teixell et al., 2016;12 in Fig. 3) and continues close to the road between the shallow marine marls and sandstones of this unit (cartographic unit 23 of Teixell et al., 2016;13 in Fig. 3). The Puebla de Fantova section ends within the lowest sandstone levels of the Upper Perarrúa unit in the Ermita hill, close to La Puebla de Fantova village (base of cartographic unit 24 of Teixell et al., 2016;14 in Fig. 3). The contact between the Middle and Upper Perarrúa units is sharp, marked by the base of a regional cartographic reference level (Teixell et al., 2016), which works as the correlation level between the Puebla de Fantova and Besians sections. This contact is interpreted as the result of an abrupt input of fan delta deposits. The base of the second section (Besians section), near the Besians village (Fig. 3), is located in the upper part of the Middle Perarrúa marls (cartographic unit 23 of Teixell et al., 2016;13 in Fig. 3) and includes the lower part of the Upper Perarrúa unit (cartographic unit 24 of Teixell et al., 2016;14 in Fig. 3) where several levels with lower Lutetian shallow benthic larger foraminifera where recognised (Tosquella, 1995). The lower part of the Besians section (first 75m) is made up of fan-delta sandstones and conglomerates of northern provenance (Santa Liestra 1-2 transition of Crumeyrolle and Mutti, 1986, which are interpreted as the product of reworking of northern derived coarse-grained sediment, resulting in a succession of barrier-bar complexes (Nijman and Nio, 1975). The upper part of this stratigraphic profile corresponds to more open marine facies, rich in larger foraminifera, including marls, siltstones and fine-grained sandstones. The upper boundary of this sequence is the Besians gully truncation, which marks also the top of the studied succession.

Biostratigraphy: Sampling and procedures
A total of sixteen samples were collected to study the larger foraminiferal assemblages (Nummulites, Assilina and Alveolina) in the studied stratigraphic interval, nine along the Puebla de Fantova section ( Fig. 4) and seven along the Besians section (Fig. 5). In all the samples the specimens were isolated and prepared accordingly to study its surface features and equatorial and axial sections.
The samples were disaggregated in an oxygen peroxide and Na 2 CO 3 solution, and the fine sediment removed using 1.0-, 0.5-and 0.2-mm mesh sieves. The larger foraminifera were picked from the washed residues, each specimen split by its equatorial section, and studied and measured under a light microscope at 40x magnification. The systematic approach in this work follows the taxonomic criteria used by Schaub (1951Schaub ( , 1966Schaub ( , 1981, Tosquella (1995), Tosquella and Serra-Kiel (1998) and Hottinger (1960 Calcareous nannofossil assemblages were analysed in 18 samples; thirteen were collected in the Puebla de Fantova section and five in the Besians section ( Fig. 4; 5, for sample location). All the samples were prepared from unprocessed material as smear slide using standard procedure (Bown and Young, 1998). The nannofossil biostratigraphy presented here is based on the examination of the samples with a Zeiss Axioplan petrographic microscope at varying magnification (x 1000-x 1600), using plain and cross-polarized light. Due to the scarcity of calcareous nannofossils in the studied samples, more than 5 traverses were analysed to detect rare species with key biostratigraphic value. The standard schemes of Martini (1971) and Okada and Bukry (1980) and the standard taxonomy for Cenozoic calcareous nannofossils (Agnini et al., 2014;Bown, 2005;Perch-Nielsen, 1985;Young and Bown, 1997) have been adopted for this study. Table 1 summarizes the obtained results.

Magnetostratigraphy: Sampling and procedures
A total of 99 paleomagnetic samples were drilled at the two sections: a standard core was taken at 4-6m stratigraphic intervals at Puebla de Fantova (50 cores along the 220m-thick stratigraphic section), and 2-3m at Besians (49 cores along 125m-thick stratigraphic section). For these tasks, a water-cooled gasoline-powered drilling machine was used to avoid disturbance on the paleomagnetic record. Cores were in-situ oriented using a clinometer attached to a magnetic compass.
Once in the laboratory, cores were cut in standard specimens of 10.4cm 3 and relabelled for further analyses. Magnetic measurements were taken in the Laboratory of Paleomagnetism of the Montanuniversität Leoben (Austria). An SRM-755 cryogenic magnetometer (2G) and an MMTD60 (by Magnetic Measurements Ltd) thermal demagnetizer were used to carry out the stepwise thermal (TH) demagnetization of samples. A 2G Pulse magnetizer was used to impart the samples with an Isothermal Remanent Magnetization (3 axes-IRM). Finally, an MFK1 Multifunction Bridge (by AGICO) was used for bulk susceptibility monitoring during the thermal treatment.
The magnetic Local Polarity Sequence (LPS) in the studied profiles was built from more than 100 stepwise thermal demagnetizations. In every specimen, up to 11 steps were studied in the thermal demagnetization  Early/middle Eocene, Ésera valley 8 Martini (1971) Okada and Bukry (1980)   Kirschvink, 1980) using the VPD software (Ramón et al., 2017).
Paleomagnetic directions were grouped into three different types according to their quality and reliability. Type 1 vectors display a good definition (usually estimation errors below 15°) and address undoubtedly to the origin. Type 2 vectors may be poorer (errors ≥15°), but display an unambiguous polarity. Type 3 vectors are weaker and/or noisier directions. The LPS profile was only characterized by type 1 and 2 samples to ensure the reliability of the magnetostratigraphic record. Profile directional means were calculated using Fisher's (1953) and Bingham's (1974) statistics under the software Stereonet version 9.9.1 Cardozo and Allmendinger, 2013).
Three axis-IRM acquisition and its TH demagnetization (Lowrie's 1990 test) were applied for a set of samples aiming to obtain information about the coercivity of the magnetic minerals present in the rocks. For this, a peak field was applied in each of the three main axes. Applied fields were: Z: 2.0; Y: 0.4 and X: 0.012 Tesla, consecutively. Then, the thermal demagnetization routine was performed in thirteen steps from 20°C up to 680°C.

Calcareous nannofossils
The studied samples were of poor quality for the calcareous nannofossil study due to the poor preservation and low abundance of the autochthonous specimens and the presence of abundant reworked Cretaceous specimens, which represent the 60-95% of the total assemblage in all the samples (Table 1).
The lower samples of La Puebla de Fantova section (F 1n-F 6n) (Fig. 4) are almost barren of calcareous nannofossils.
Only the sample F 1n shows more than 1 specimen per field of view (spp./f.v.). The observed specimens are usually broken and show traces of dissolution. In this interval, the reworked, dissolution resistant, Cretaceous species of the genera Eiffellithus, Watznaueria and Prediscosphaera dominate the assemblages. The more prevalent autochthonous species are Coccolithus pelagicus and representatives of the genera Cyclicargolithus and Reticulofenestra including the species Reticulofenestra dictyoda. The calcareous nannofossil abundance and preservation improves up-section. Sample F 7n shows 21 spp/f.v. but almost all the identified specimens are reworked Cretaceous species (Table 1). From sample F 8n up to the top of the section (sample F 13n), the calcareous nannofossil abundance ranges from 1 to 8 spp./f.v. In this samples the reworked Cretaceous specimens are more abundant than the autochthonous specimens and the preservation varies from moderate to poor. The autochthonous calcareous nannofossil assemblage is still dominated by Coccolithus pelagicus and species of Reticulofenestra and Cyclicargolithus. The presence of 5-rayed Discoaster sublodoensis in sample F 10n (Table 1) is noteworthy.
The five samples of the Besians section (Fig. 5) are very poor. In samples B 1n to B 3n only 1 spp/f.v. was recorded. Reworked Cretaceous specimens dominate the  Early/middle Eocene, Ésera valley 14 assemblages and only a few representatives of the genera Reticulofenestra, Coccolithus and Discoaster (fragments) were found. Samples B 42n and B 49n were almost barren of calcareous nannofossils and the few specimens found were reworked Cretaceous specimens.
Paleomagnetic and rock magnetism data Lowrie's (1990) tests performed on several samples from both profiles (Fig. 10) reveal the presence of soft, medium and high coercivity range, although soft and medium coercivity curves dominate quantitatively at diagrams. The soft coercivity curves decay completely about 580°C clearly indicating the presence of magnetite as the main magnetic carrier. Medium coercivity curves often fall around 300-350°C and suggest the occurrence of undifferentiated sulphides (likely greigite). Moreover the BE02 sample shows a final decay at a higher temperature (about 675°C), showing a medium-hard coercivity mineral behaviour (likely hematite). Finally, the high coercivity curve, generally decays about 675°C, but also shows decay at very low temperatures in very few cases (decay before 200°C, in some cases, such as in FA22) due to the presence of some goethite. All these data together with demagnetization features, seems to indicate the presence of magnetite and undifferentiated sulphides as the main carriers, accompanied by the presence of hard coercivity minerals in some few cases. These results are in agreement with early diagenetic re-equilibrium of magnetite and greigite that may result, despite some noise, in reliable primary records of the magnetic field (Larrasoaña et al., 2003) as it has been demonstrated in many magnetostratigraphic studies of the South Pyrenean basin (Garcés et al., 2020 and references therein).
Stepwise thermal demagnetizations as displayed in Zijderveld's (1967) diagrams (Fig. 10) allowed fitting reliable directions from 82 samples. Data from remaining samples were ruled out because of weak Natural Remanent Magnetization (NRM) intensity or erratic demagnetization (type 3). Characteristic Remanent Magnetization (ChRM) directions were fitted with 6-7 steps on average. Most of samples were fully demagnetized up to 500°C, and in some samples authigenic magnetic minerals (magnetite) grew up during their heating in the oven beyond 400°C, as revealed by significant increments of susceptibility (as in many similar Pyrenean rocks; Pueyo et al., 2002;Larrasoaña et al., 2003). Some samples showed very weak magnetizations, and higher demagnetization steps were not considered in further calculations.
As a general rule, Zijderveld diagrams (Fig. 10) show two magnetic components: a viscous low temperature component up to 180-240°C, in average, very similar to the present geomagnetic field (DEC: 358, INC: 53); and a high temperature component from ~200°C up to 420 -570°C (when possible). This component displayed normal and reversed directions with linear trajectories very often decaying to the origin (type 1) and it has been considered the ChRM used to build the LPS's.
From a directional point of view (only type 1 and 2 samples), the ChRMs in the lower hemisphere and merging both profiles together, are well-grouped After Bedding Correction (ABC) and display a robust Fisher distribution and expected inclinations; Dec: 020, Inc: 61 (α95: 8°, k; 4.8 and R: 0.7945) in agreement to data of the South Pyrenean Central Unit (Beamud et al., 2004;Dinarés-Turell, 1992) and to the closer Ainsa oblique zone (Muñoz et al., 2013; Oliva-Urcia and Pueyo, 2019 and references therein).
Once the data are restored to the paleohorizontal, the normal polarity dominates the dataset. Mean Fisher statistics (Table 2) by polarity are Dec: 012, Inc: 64 (α95: 9.3°, k: 4.8) for the normal polarity directions (61points) and 215, -52 (α95: 15.6° and k: 5.1) for the reverse ones (21 points). The fold and reversal tests yield non-significant results if our own dataset is used alone; partially (fold test) because of the lack of enough bedding dip differences and also (reversal test) because of the small number of reverse vectors. In any case, the pseudo-antiparallel character of both polarities (Fig. 11), together with several previous works with well-proven primary attributes in the southern Pyrenees (see recent overviews by Garcés et al., 2020;Oliva-Urcia and Pueyo, 2019), allow us to be confident about the primary character of the dataset.
Studied sections were based on a detailed stratigraphic study, measured sections with enough sampling density to characterize the LPS, complete TH demagnetization and ChRMs fitted with standard methods and fully published (dec, inc, Virtual Geomagnetic Poles (VGP) latitude) apart from the polarity log, rock magnetism analyses to determine the magnetic carriers, pseudo-antipodal means (after restoration) with expected inclinations and, despite the lack of our own fold test, plenty data in the region supporting the primary character of the magnetization. Therefore, and following the van der Voo (1990) and Opdyke and Channel (1996) criteria, the results obtained in La Puebla de Fantova and Besians are reliable to propose a chronostratigraphic age model together with the biostratigraphic data introduced in this paper.
In the samples from La Puebla de Fantova section ( Fig.  4; Table 1), markers of calcareous nannofossil zones are almost absent due to the poor fossil preservation; therefore, the assignment to a specific zone has been done with auxiliary markers, so that the zonal assignment may not be accurate.
The calcareous nannofossil assemblages recorded in the Besians section ( Fig. 5; Table 1) are not good enough to assign them to a specific zone. Sample B 2n is stratigrafically equivalent to samples 102-105 of the Campo section (Perarrúa) studied by Kapellos and Schaub (1973). These authors pointed out that these samples contained, among other species, Chiphragmalithus aff. quadratus. But, the specimen depicted by these authors in their plate 7, fig. 15, is closer to Nannotetrina cristata, which first occurrence is slightly preceded by the first occurrence of Blackites inflatus, the marker species of subzone CP12b (Agnini et al., 2014;Bernaola et al., 2006;Tori and Monechi, 2013;Westerhold et al., 2017). Therefore, the upper part of the Besians section could be assigned to subzone CP12b.

Magnetic local polarity sequence and correlation
We have calculated the VGPs from the ChRMs after bedding restoration along the studied profiles (only type 1 and 2 samples were used for such purpose). The VGP latitude log has allowed building the LPS for each profile; La Puebla de Fantova (Fig. 12) and Besians (Fig. 13). At La Puebla de Fantova section, from bottom to top, we can identify a reverse and a normal polarity magnetozones; R1 and N1. Magnetozones R1 and N1 are built with 8 and 18 sites respectively and expand along the section from the base to meter 58 (R1) and then up to meter 162 (N1). At Besians section (Fig. 13), all the samples display normal polarity, as it was identified in 47 different stratigraphic levels (90% of type 1-catalogued quality). This relatively simple pattern of magnetozones is robust and clear, even in magnetozone R1 (La Puebla de Fantova section) that is characterized by more than 20 consecutive levels where reverse polarity was identified.
The sampling was initially designed to avoid any possible stratigraphic gap between the two sampled sections and thus we used the sharp transition between the cartographic units 23 and 24 of Teixell et al. (2016) as the correlation level; the correlation follows the lowest first prominent coarse levels located at the top of La Puebla de Fantova section (about meter 220), to the equivalent levels at the Besians section (about meter 30 from the base) (Fig. 14) in the same stratigraphic unit. As a whole (Fig. 14), only two magnetozones are found. The lower part of the Puebla de Fantova section shows reverse polarity (R1 magnetozone) whereas the rest of the Puebla de Fantova and the Besians sections show a normal polarity signal (N1).

Data integration and interpretation
The correlation between the Puebla de Fantova and Besians sections (Fig. 14) allows clarifying the chronostratigraphy along the Ésera valley section, which includes the Y/L boundary. The Puebla de Fantova section (Figs. 3;4;12;14) is above the lower Ésera section (Navarri section of Bentham and Burbank, 1996, or Morillo de Liena section of Payros et al., 2009b), with an interval unsampled for paleomagnetism in-between, which corresponds to the Lower Campanúe conglomerate (Santa Liestra-1; Fig. 2), as explained above. The lower Ésera section ends, in terms of paleomagnetic data (Bentham and Burbank, 1996), at the base of Lower Campanúe conglomerates (BB-1 in Fig. 3) (Fig. 2), below this conglomerate, contain SBZ 11 (middle Cuisian) larger foraminifera (Kapellos and Schaub, 1973;Schaub, 1981;Tosquella, 1995), calcareous nannofossils of the NP13 to NP14 transition (Payros et al., 2009b following data of Kapellos and Schaub, 1973) and include the uppermost C22r, the entire C22n, and the base of C21r (Bentham and Burbank, 1996). The upper limit of C21r in the Ésera valley section was unsolved. The Santa Liestra section (Bentham and Burbank, 1996;Payros et al., 2009b) (BB-2 in Fig. 3), which follows the Navarri section, starts in normal polarity magnetic values with an unsampled intermediate stratigraphic gap. The Santa Liestra section is lateral-equivalent to the upper part of the Puebla de Fantova section and the lower part of the Besians section (Figs. 5;13) showing normal polarity. After the data and results presented in this study, only the Lower Campanúe conglomerates remains partially un-sampled for magnetostratigraphy.
According to Speijer et al. (2020), the NP13/NP14 boundary appears near the C22n/C21r transition, in the upper part of SBZ 11, in agreement with its occurrence in the Lower Perarrua unit (Payros et al., 2009b). The lower reverse magnetic polarity (R1), found in the Puebla de Fantova section (Fig. 14), which contains SBZ 12 larger foraminifera, is interpreted as belonging to C21r. The inverse/normal polarity boundary, which occurs in the lower part of this section, is interpreted as the C21r/ C21n boundary. The rest of the Puebla de Fantova and the complete Besians sections (Figs. 3;5;13;14), with a normal-polarity signal (N1), are interpreted as C21n (Fig.  14). Below the correlation point between the Puebla de Fantova and Besians sections, C21n contains SBZ 12 larger foraminifera and NP14 (CP12a) calcareous nannofossils. Above the correlation point, after a stretch without fossil data, it contains SBZ 13 larger foraminifera and CP12b calcareous nannofossil forms in the upper part of the Besians section (Figs. 5;14).

Discussion
The scarcity and poor preservation of calcareous nannofossils precluded the possibility of fixing a precise location for the Ypresian/Lutetian boundary in the Ésera valley section. Nevertheless, the existing data (from samples in this study and previous works: Kapellos and Schaub, 1973;Payros et al., 2009b) is enough to confirm the robustness of the interpretation, despite the calcareous nannofossil zone markers appear in a younger stratigraphic position than expected.
In the Puebla de Fantova section, the FO of D. sublodoensis, the marker of the base of zone NP14 (CP12a), is registered in sample F 10n, with normal polarity (C21n) and SBZ12 larger foraminifera. One of the two D. sublodoensis specimens found in this sample was a 5-rayed D. sublodoensis ( Table 1) that according to Agnini et al. (2014) its First Common Occurrence (FCO) is at the base of chon C21r, slightly after the FO of rare D. sublodoensis in chron C22n.
Thus, the finding of two specimens of D. sublodoensis in sample F 10n confirms the attribution to chon C21n of the normal polarity signal N1 and suggests that the absence of zone NP14 (P12a) marker species at the base of the Puebla de Fantova section is due to poor preservation of the calcareous nannofossil assemblages in this section.
The base of the Subzone CP12b is recorded by stratigraphic correlation in the lower third of the Besians section and is given by the FO of N. cristata in samples CK 102-105 in the Campo (Perarrúa) section (Kapellos and Schaub, 1973). In their work, these authors provide a list of species present for a given range of samples but do not detail the association of calcareous nannofossils for each of the studied samples. This prevents us from knowing if some of the studied samples had very poor preserved calcareous nannofossil assemblages or even if calcareous nannofossils were absent in some of the samples. Several attempts to reproduce the results obtained by Kapellos and Schaub (1973) in the samples CK 23-24 from the Campo section (equivalent to F13n) have shown that the calcareous nannofossil assemblages of these samples are exclusively composed of reworked Cretaceous species or even that in some of the samples calcareous nannofossils are absent.
The above observations suggest that the absence of subzone CP12b markers in CK19-25 samples at Campo section and in the upper half of the Puebla de Fantova section is due to the poor preservation of calcareous nannofossil assemblages.
Outside of the Pyrenees, the Agost section (Betic Chain, western Tethys) provides a chronostratigraphic dataset including calcareous nanofossils, larger foraminifera and magnetic polarity (Larrasoaña et al., 2008). In the Agost section, larger foraminifera also appear as reworked components outside their original habitat, with apparent no mixture of different age specimens (Larrasoaña et al., 2008). Larger foraminifera found around the CP12a/CP12b boundary (Y/L boundary), within C21r, still belong to the SBZ 11 (Larrasoaña et al., 2008). The SBZ 12 and SBZ 13 forms are found in C21n. In the Agost section, larger foraminifera forms appear in a younger stratigraphic position than in the Ésera valley, which expects to reworked faunas.
With the obtained magnetic polarity dataset for the Y/L transition, data from the shallow marine Ésera valley sections (La Puebla de Fantova and Besians) results in a robust stratigraphic chart for the larger foraminifera distribution around the C21. These results allow shifting the SBZ 12-SBZ 13 boundary to the lower part of C21n Early/middle Eocene, Ésera valley 20 (Fig. 15). Accordingly, the Y/L boundary occurs within SBZ 12.
The calcareous nannofossil assemblages are very poorly preserved and showed low abundance of autochthonous specimens. According to these assemblages the lower half of La Puebla de Fantova section belongs to Zone NP13 (CP11), and the upper half of this section and the lower part of the Besians section belong to Zone NP14 (CP12a). The rest of the Besians section belongs to Zone NP14 (CP12b). However, due to the poor preservation of the assemblages, the zone markers are almost absent and thus their first occurrences in the studied sections occur in a younger position than expected.
A robust local polarity sequence was built and based on 87 reliable paleomagnetic determinations in the two studied sections (La Puebla de Fantova and Besians) with  an unambiguous stratigraphic correlation. Characteristic directions are pseudo-antiparallel and considered as a primary record of the magnetic field supported by several paleomagnetic studies in the region. Two local chrons (R1, N1) together with the biostratigraphic information (larger foraminifera and calcareous nannofossils) allow us proposing a new age model for the Ésera valley section, the most complete section in the Southern Pyrenees across the Y/L transition. There, the SBZ 12 faunal assemblages, identified since now as late Cuisian (uppermost Ypresian) (Speijer et al., 2020 and references herein), extend across the Y/L boundary, ranging into the lower part of C21n. o l o g i c a A c t a , 2 0 . 6 , 1 -2 5 ( 2 0 2 2