4.1. Burial History of the Granite Massifs
 Except for the NV7 sample, 40Ar/39Ar data on biotite and K-feldspar present intermediate ages between the Hercynian and Pyrenean deformation episodes, which thus cannot be interpreted as cooling ages. This suggests that after the post-Hercynian erosion, temperature never exceeded about 300°C in the two massifs, the temperature required for a major argon resetting in the two chronometers. Considering a normal geothermal gradient, this would indicate that the maximum burial depth of the two massifs before and during the Pyrenean orogeny was less than 10 km, most probably in the range 6–8 km. The 40Ar/39Ar thermochronological data also suggest that the more internal Néouvielle massif was buried deeper than the external Bielsa one.
 The 40Ar/39Ar dating of NV7 K-feldspar provides a complex age spectrum with total age of circa 53 Ma that contrasts with the 236 Ma biotite age from the same sample. NV7 was sampled at a few tens of meters from one of the mylonite corridors belonging to the eastern termination of the Eaux-Chaudes thrust. A Rb-Sr age of 48 ± 2 Ma obtained by Wayne and McCaig  on mylonite veins (whole rock and vein minerals: amphibole, epidote, calcite, and albite) confirms the occurrence of Pyrenean movements along those shear zones. Petrographic observations indicate that important fluid circulation occurred within the mylonite corridors [Henderson and McCaig, 1996] with temperatures of 300 to 350°C deduced from fluid inclusions and petrography [Henderson and McCaig, 1996; Ingles et al., 1999; M. Campani, unpublished data, 2005]. In NV7 sample, it is likely that K-feldspar overgrowths formed around the magmatic K-feldspar are coeval with this fluid activity as suggested by the pseudoplateau age of about 50 Ma obtained at intermediate temperature (Figure 5). This reinforces the idea of tectonic activity along the mylonitic zones that form the eastern termination of the Eaux-Chaudes thrust and thus early Eocene activity along the thrust itself. Younger ages would reflect cooling of these overgrowths and older ones, inherited argon in the magmatic K-feldspar core. The fluid activity in this sample is also attested by the advanced chloritization of the biotite population, the dated grain at 236 Ma being one of the less affected by this process, thus having best preserved its Hercynian argon signature.
 Fission track ages indicate that all the samples have been exhumed from temperatures higher than 110°C during the Pyrenean episode. Considering a geothermal gradient of 25 to 30°C km−1, the granite massifs were thus buried, before circa 35 Ma for Néouvielle and Bordère-Louron, and circa 20 Ma for Bielsa to depths of at least 4 km (below the fission track annealing zone) and likely less than 6–8 km as suggested by 40Ar/39Ar thermochronological data, depending on the considered massif.
 Except thin local remnants of Permian sediments, the Bielsa massif is directly covered by Upper Cretaceous limestones. The sedimentary cover of the Néouvielle is unknown due to erosion, but farther west, the Balaïtous granite which has the same altitude and structural position as the Néouvielle massif, is also capped by Upper Cretaceous limestones. We thus assume that both the Néouvielle and Bielsa massifs were at, or close to, the surface before the Late Cretaceous transgression, and were then covered by a sedimentary succession similar to that presently preserved along the southern border of, and above the western termination of, the Axial Zone (Figure 3). This implies ∼1000 m of Upper Cretaceous–Paleocene shelf deposits, followed by the lower part of the lower Eocene turbidite succession, that corresponds to the first stages of subsidence of the south Pyrenean foredeep [Labaume et al., 1985; Teixell, 1996]. Although the original thickness of these turbidites is not known, it probably not exceeded 2000 to 3000 m. The sedimentary burial of the granite massifs may have thus been close to, or even reached, the 3–4 km necessary for the reset of the apatite fission track system as a result of the overlying foredeep fill. Sedimentary burial was then followed by tectonic burial when the deformation front propagated southward, thus leading the granites to their maximum temperature during the Pyrenean orogenic cycle (Figure 7). On the basis of the tectonic-sedimentation relationships in the Jaca and Ainsa basins, this probably occurred during the early middle Eocene for the Néouvielle massif (emplacement of the Eaux-Chaudes thrust, passing southward to the Monte Perdido thrust in the Jaca and Ainsa basins) and the late Eocene–early Oligocene for the Bielsa massif (emplacement of the Gavarnie thrust) [Labaume et al., 1985; Teixell, 1996].
4.2. Exhumation History of the Granite Massifs
 Both central fission track ages and track lengths modeling of summit samples NV1 and BS1 indicate that the Néouvielle massif has been exhumed prior to the Bielsa massif.
 In the Néouvielle massif, sample NV1 crossed the 110°C isotherm around 35 Ma which is consistent with a late Eocene age of activity on the Gavarnie thrust. Cooling of sample NV1 stopped around 25 to 20 Ma suggesting that the Gavarnie thrust was not active anymore.
 Samples NV13, NV7, and PE2, as well as sample LA7 from Morris et al. , were collected along a mostly horizontal profile around 2000 m of elevation and parallel to the direction of compression. These samples display, within the error margins, the same central fission track age of about 25 ± 2 Ma. The Néouvielle massif thus appears to have been exhumed as a single, rigid block on the hanging wall of the Gavarnie thrust. Sample PE2, in the footwall of the Pic-Long thrust does not reveal any major movement on that fault during or after the exhumation of the massif indicating that this fault was active before circa 30 Ma but did not participate to the exhumation of the Néouvielle massif. Furthermore, there is no indication of tilting of the massif during its exhumation.
 The central age of 31.8 ± 2.2 Ma obtained for sample BL1 in the Bordère-Louron granite (close to the 35.0 ± 2.8 Ma age obtained by Morris et al.  on the same massif) is more difficult to explain. This age is close to the 35.1 ± 2.3 Ma obtained for sample NV1 situated 2200 m higher in the Néouvielle massif. Furthermore, the thermal histories derived from fission track lengths modeling of both samples are similar, implying that they have been exhumed at the same time and in the same way between 110 and 60°C. Differential exhumation probably not happened after 5 Ma (outside of the apatite partial annealing zone) because (1) there is no evidence of a major fault between the two granites that could explain similar ages at very different altitudes, and (2) by that time the tectonic activity in the Axial Zone had stopped. The favored hypothesis is the occurrence at depth of an east dipping lateral footwall ramp of the Gavarnie thrust, subparallel to the direction of movement, and located between the Néouvielle and Bordère-Louron massifs. This resulted in thrusting of the Néouvielle massif higher than the Bodère-Louron massif, inducing folding and tilting of the isotherms following a paleotopography inclined toward the east above the lateral ramp. This resulted in a similar cooling path for samples NV1 and BL1 not correlated to their difference in altitude.
 Samples NV1 and NV7 from the Néouvielle subvertical profile show relatively discordant thermal histories. NV7 entered the PAZ at the end of the initial cooling episode of NV1 which is consistent with its position deeper in the crust. However, while NV1 reached the very slow cooling stage, NV7 keep cooling at a higher rate until circa 15 Ma when its thermal history can be superimposed to the one of NV1. One possible explanation for this discrepancy between the two thermal histories could be the formation through erosion between 20 and 15 Ma of a paleovalley similar to the actual one. Localized erosion above sample NV7 would have induced cooling of sample NV7 compared to a stable NV1. Vertical incision rates would have been of the order of 0.1 mm yr−1, similar to the Miocene and Quaternary incision rates reported by Vergés et al.  farther east.
 In the Bielsa massif, samples BS1 and BS7 with their central fission track ages of 18 ± 1.9 and 19.4 ± 2.7 Ma, respectively, imply that the Bielsa massif was first uplifted vertically before being tilted southward between 18 and 11 Ma when BS6 (in the core of the massif) was exhumed. The sample BA3 from Morris et al. , at an elevation of 1020 m and also located at the top of the granite (Figure 3), yielded a central age of 16.0 ± 1.5 Ma consistent with this interpretation. Casas et al.  showed that the Bielsa granite was deformed by folding during its exhumation when passing above a footwall ramp of the Bielsa thrust. Two scenarios can explain the relationship between exhumation of the granite and the formation of this hanging wall ramp anticline: a) from circa 22 to 17 Ma (Figure 6), the Bielsa massif was exhumed without tilting or folding along the ramp of the Bielsa thrust. Subsequently folding and thus tilting of the post-Variscan surface occurred when the granite body passed from above the ramp to the footwall flat; b) the initial exhumation was driven by the more external Guarga thrust (Figure 2) and later, out-of-sequence movement along the Bielsa thrust induced folding and tilting. The second hypothesis implies that movement on the Guarga thrust continued until the Burdigalian, as it was proposed by some authors [e.g., Millán Garrido et al., 2000] but is not confirmed by the dating of synorogenic sediments (see discussion below). On the other hand, we rule out the possibility that the similar ages obtained along the tilted top surface of the granite may result from posttilting (i.e., postthrust) cooling across inclined isotherms. Indeed, this would imply that isotherms inclination was of the same order as the tilting of the granite top surface, i.e., a value of 35–40° unlikely at a 3.5–4 km depth.
 Finally, the last cooling event around 5 Ma recorded by sample NV7 from Néouvielle is also well expressed in sample BS1 from the Bielsa massif and could also be recorded in sample BL1 from Bordère-Louron although in this last sample it is much less constrained and will not be discussed further (Figure 6). Because it can be related to other geochronological and geological data that are discussed below, we cannot consider that this last cooling corresponds to the widely recognized artifact in modeling (the “late Miocene” cooling) generally associated with the use of the Laslett et al.  annealing model [e.g., Dempster and Persano, 2006].
 The present morphology of the Pyrenees is characterized by the occurrence within the Axial Zone of remnants of high-elevation, low-relief erosion surfaces [Babault et al., 2005]. When, by late Eocene, the Ebro basin became endoreic [e.g., Vergés et al., 2002] it was filled by Oligo-Miocene continental deposits up to about 1000 m thicker than the presently preserved succession and that onlapped the southern edge of the Axial Zone [e.g., Babault et al., 2005, and references therein]. Many authors proposed that reexcavation of the southern Pyrenees and Ebro basin was induced by the capture of the Ebro river by the Mediterranean Sea through a combination of the Miocene extensional tectonics within the Catalan chain and the Messinian desiccation crisis [e.g., Nelson and Maldonado, 1990; Coney et al., 1996; Garcia-Castellanos et al., 2003]. Using forward modeling of apatite fission track analysis data from the Maladeta massif, Fitzgerald et al.  described a final cooling event starting between 10 and 5 Ma with a magnitude corresponding to a total denudation of 2 to 3 km. They relate this cooling to reexcavation of the southern Pyrenees by a proto-Ebro river during the Messinian. However, recent work by Babault  and Babault et al.  pointed out that the Miocene sediment accumulation and recent reexcavation also occurred on the northern flank of the belt. Considering that the base level of a chain is set by the river capacity to transport sediments, they defined the “efficient” base level of a chain as “the limit of the most proximal, extensive detrital sedimentation” and proposed that the recent lowering of this efficient base level, responsible for the reexcavation of both flanks of the belt, was due to the frequent and abrupt variations in climate since the Pliocene [Babault et al., 2005].
 The last cooling event observed in our samples corresponds in age with the final cooling event observed by Fitzgerald et al.  in the Maladeta and can reasonably be attributed to erosion during reexcavation of both the northern (Néouvielle and Bordère-Louron massifs) and southern (Bielsa massif) flanks of the Pyrenees. However, our data set is too small to discuss the differences between the models presented above.
4.3. Implications for the South Pyrenean Wedge Propagation
 The combination of our results with geological data is thus consistent with the propagation of basement thrusts toward the south from the Eaux-Chaudes thrust during the early middle Eocene to the Gavarnie thrust during the late Eocene–early Oligocene and the Guarga thrust during the late Oligocene–early Miocene (Figure 8). This propagation caused the migration of the depocenters in the basin. The Hecho Group turbidites were deposited in the foreland of the Eaux-Chaudes thrust during the early middle Eocene and were affected by deformation when movement on the Eaux-Chaudes thrust propagated in decollement levels within cover sediments, resulting in the emplacement of the Monte Perdido thrust during the late Lutetian-Bartonian [Labaume et al., 1985; Teixell, 1996]. When the Gavarnie thrust was activated, deformation and uplifting of the turbidite succession continued, and the depocenter migrated southward in the Guarga syncline filled by shallow marine and continental upper Eocene–lower Oligocene deposits. Activation of the Guarga thrust during the late Oligocene resulted in a new migration of the depocenter in the Ebro basin [Teixell, 1996]. The upper part of the synorogenic sediments deposited along the south Pyrenean thrust yielded ages up to 22 Ma [Arenas et al., 2001], indicating that most of the movement along the Guarga thrust occurred before the middle part of Aquitanian. Minor latest movements are not dated and may have continued during the upper part of Aquitanian and possibly during the Burdigalian [Millán Garrido et al., 2000].
 Our data on the Bielsa massif also show that out-of-sequence basement thrusting occurred during the Miocene along the southern edge of the Axial Zone (Figure 8). Depending on the kinematic interpretation adopted for the Bielsa massif exhumation (see discussion in previous section), activity of the Bielsa thrust may have begun during the Burdigalian around 19–18 Ma (central fission track ages on samples BS1 and BS7), but in any case continued more recently in the Miocene (southward tilting of the granite and overlying Gavarnie thrust sheet). We cannot exclude that this late tectonic activity at the southern edge of the Axial Zone was coeval with the youngest, undated movements along the south Pyrenean frontal thrust. However, it seems kinematically difficult that movement of the Bielsa thrust was transmitted to the frontal thrust along the Triassic décollement level that was already severely deformed by the Guarga thrust (Figure 2). It seems more likely that the activity of the Bielsa thrust resulted in internal deformation of the Jaca and Ainsa basins. Such a young deformation has not been recognized in the basin fill up to now and may correspond to out-of-sequence thrusting, back thrusting, distributed deformation or any combination of these.
 In the central Pyrenees, apatite fission track analyses by Fitzgerald et al.  and Sinclair et al.  showed that exhumation of the highest massifs followed a piggyback sequence of thrusting, i.e., middle-late Eocene for the Riberot massif in the northern part of the Axial Zone, and late Eocene–Oligocene for the Maladetta and the Marimanya massifs in the central part (see location of the massifs in Figure 1). Exhumation of the Maladetta-Marimanya, the highest massif of the central Pyrenees, was thus coeval to that of the Néouvielle massif, the highest massif on the section studied here, but the thrusts involved are not the same. Indeed, the eastward extension of the Gavarnie thrust passes between the Maladetta and Marimanya massifs without being associated with a significant difference in exhumation, which is consistent with a dramatic decrease of offset across the thrust when going eastward [Soler et al., 1998]. The thrust involved in the exhumation of the Maladetta-Marimanya is thus located below and to the south of the Gavarnie thrust. At the south of the Maladetta massif, the small Barruera massif (Figure 1) yielded an exhumation age around 20 Ma, younger than that of the samples located at the same elevation in the Maladetta massif, thus attesting Miocene basement thrusting along the southern edge of the Axial Zone [Sinclair et al., 2005]. Our results on the Bielsa massif thus demonstrate that this late tectonic activity affected a significant length of the Axial Zone.