Burial in the western Central Andes through Oligocene to Miocene ignimbrite flare‐ups recorded by low‐temperature thermochronology in the Cañete Canyon, Peru

Thermochronological data are essential to constrain thermal and exhumation histories in active mountain ranges. In the Central Andes, bedrock outcrops are rare, being blanketed by widespread late Palaeogene–Neogene and younger volcanic formations. For this reason, the exhumation history of the Western Cordillera (WC) in the Peruvian Andes has only been investigated locally along the mountain range. Dense thermochronological data are only available in canyons of the Arequipa (16° S) and Cordillera Negra regions (10° S). We present new apatite (U‐Th)/He and fission‐track data from the 1 km deep Cañete Canyon (13° S), where the Oligo‐Miocene deposits are preserved lying conformably on an Eocene palaeo‐topographic surface. Thermal modelling of thermochronological data indicate that the 30–20 Ma ignimbrite deposits overlying the bedrock were thick enough to cause burial reheating. We demonstrate that burial associated with thick volcanic formations should be taken into account when interpreting thermochronological data from the WC or in similar volcanic‐arc settings.

thermal impact of the major Oligocene-early Miocene ignimbrite flare-ups (Figure 1).To address these questions, we provide new apatite fission-track (AFT) and (U-Th)/He (AHe) data from an altitudinal profile located in bedrock below the ignimbrites in the upper part of the 1 km deep Cañete Canyon.

| G EOLOG IC AL AND G EODYNAMIC S E T TING
The WC in Chile and Peru constitutes the Pacific flank of the Andean Altiplano (Figure 1).It is characterized by a 4-km-high topographic step separating the Coastal Cordillera to the west and the discontinuous volcanic arc to the east (Mamani et al., 2010).
Along strike of the Peruvian Andes, changes in the dip of the subducting slab are argued to control volcanic activity and rock uplift in the Cordillera Negra region (Figure 1; 10° S; Margirier et al., 2015) and in the Ocoña-Colca region (16° S; Gunnell et al., 2010;Schildgen et al., 2007); no major thermal impact on the crust was recorded in either case (Margirier et al., 2016;Schildgen et al., 2009).In the Cañete region, slab flattening caused landward migration of volcanic activity during the Palaeogene (~50-30 Ma; Mamani et al., 2010).This time frame coincides with the initiation of counterclockwise oroclinal bending in the Central Andes (Roperch et al., 2011), leading to a stronger obliquity of convergence and partitioning of deformation in southern Peru with respect to northern Chile.Several studies combining structural geology and thermochronology suggested that this flat slab period is contemporaneous with rock uplift of the WC (e.g.Gérard et al., 2021;Noury et al., 2017).Following this phase, the slab steepened at ca. 30 Ma, triggering the ignimbrite flare-up in the Central Andes, before westward arc migration towards the present-day location at 15 Ma (Decou et al., 2011).Since the Miocene, low rates of exhumation and rock uplift have been proposed for the WC of Peru and Chile, based on low-temperature thermochronology, erosion (Gunnell et al., 2010;Schildgen et al., 2009;Van Zalinge et al., 2017) and tectonic studies (Benavente et al., 2017(Benavente et al., , 2021;;Hall et al., 2012;Wörner et al., 2002).

| Sample locations and sampling strategy
We collected six samples along a profile spanning ~2 km in elevation in the Cañete Canyon (Figures 1 and 2).The samples were selected from different igneous bedrock lithologies below the thick Miocene ignimbrite layers, between the valley bottom and the base of the ignimbrites (Figure 1; Table 1).We followed the approach of Schildgen et al. (2007Schildgen et al. ( , 2009) ) in representing the palaeo-elevation of the samples using their reconstructed depth beneath the base of the volcanic rocks (Table 1; Figure 4).

Significance Statement
We provide apatite fission-track and AHe data along an altitudinal bedrock profile from the Cañete canyon in the western Andes of southern Peru.We demonstrate that it is essential (1) to use geological constraints in thermochronological models to take into account the exposure of the bedrock before Oligo-Miocene ignimbrite deposition and (2) to constrain a meaningful thermal history.Burial related to ignimbrite emplacement should be incorporated when interpreting thermochronological data in this area.Our

| Thermal-history modelling
Thermal-history modelling was performed jointly for all the data with the QTQt code (Gallagher, 2012).The inversion code incorporates a radiation damage-dependent kinetic model for He diffusion (Flowers et al., 2009), as well as a multi-kinetic AFT annealing model (Ketcham et al., 2007).According to the oldest thermochronological record and independent geological constraints on the timing of batholith emplacement (~60 Ma), the time span explored starts at 60-52 Ma (Figure 5).We designed two models, both starting with an initial temporal and thermal constraint corresponding to pluton emplacement.The first model is free from any other constraints.In the second model, we constrained samples to be close to the surface at ~28 ± 2 Ma, based on the field evidence discussed above (Figure 5; Table 2 for parameter values).For both models, we did not impose any reheating event, but we allow the code to explore T-t paths that include reheating.To test the robustness of thermal histories, we compared predicted ages from the most likely T-t paths derived by QTQt to the observed data (Figure 5).

| AFT and AHe data
Five and four samples were analysed for AHe and AFT respectively.
Data from this study and from Wipf et al. (2008) are reported along a south-west-north-east profile (Figure 3) and as a function of palaeodepth beneath the Oligocene-early Miocene ignimbrites (Figure 4 and Data S1).The spatial pattern of AFT and AHe ages forms a 'boomerang' shape, with the youngest AHe ages encountered ~60 km from the coast at the valley bottom (Figure 3).Two AFT ages, corresponding to the highest sample (3002 m) and to the lowest sample (295 m; Wipf et al., 2008), are similar to the crystallization age of the batholith (Figure 3).AFT ages of samples from the lower part of the profile (1019-2459 m) are the same within error (24.1 ± 5.5, 22.7 ± 3.7 and 20.9 ± 3.7 Ma) and are partially annealed, as confirmed by limited track-length data (Table 3).AHe ages are younger than the corresponding AFT ages, with corrected ages ranging from 1.8 ± 0.2 to 2.9 ± 0.3 Ma for the lowermost sample (1019 m) to similar ages of  4).The highest-elevation sample (08, 3002 m) shows a significantly older AHe age (13.8 ± 1.2 Ma).

| Thermal-history modelling
AHe and AFT ages predicted by QTQt reproduce the observed ages well for both thermal histories (Figure 5), with or without a t-T constraint related to the erosion surface at the base of the ignimbrite.F I G U R E 3 AFT and AHe ages as a function of distance to the coast, overlain on a topographic swath profile of the Cañete Canyon (refer to Figure 1c for location).Ar-Ar crystallisation ages of the plutonic rocks (Noble et al., 2005)   Both time-temperature histories require reheating of the Cañete samples at 30 Ma before final cooling, but the most recent part of the cooling history is significantly different for the two models (Figure 5).

| DISCUSS ION
The difference between the two thermal histories indicates that it is essential to implement the t-T constraints related to the documented Eocene-Oligocene palaeo-erosion surface in the WC (Figure 5d).
The thermal history including this geological constraint reproduces the observed AHe and AFT data well (Figure 5e,f).We favour this model, as it explicitly includes the formation of the Eocene erosion surface and corresponding near-surface position of the rocks prior to ignimbrite eruption.Some AFT ages (samples 08 of this study and MW34 of Wipf et al., 2008; Figure 3) overlap with the crystallisation ages and both thermal models reveal that our samples experienced cooling during the Palaeocene-Eocene.We suggest that this initial cooling phase corresponds to post-magmatic cooling after Palaeocene intrusion (Figure 5).
Our preferred thermal history shows slow cooling from 45 to 30 Ma (Figure 5d-f), which we link to moderate erosion and development of the extensive erosion surface prior to 30 Ma (Noble et al., 1979) on which ignimbrites were deposited in the region between Cañete and Arequipa (Noble et al., 2005).This model is consistent with previous studies showing the WC to be eroded to a low-altitude topography in a semi-arid context, unroofing the plutonic complexes (Margirier et al., 2015;Quang et al., 2005).
Thermochronological data and modelling indicate that burial related to ignimbrite deposition during volcanic flare-ups impacted the thermal history of the Cañete bedrock samples (Figure 5).Both our models show reheating, which reaches a peak at ~30-25 Ma, synchronous with the timing of thick ignimbrite deposition (Figure 5a,d).
These major igneous episodes may also have contributed to an increase in the local geothermal gradient at that time.However, such an increase in the geothermal gradient is not recorded in the regional data (Gérard et al., 2021;Schildgen et al., 2009).We suggest that heating was related to burial caused by the emplacement of a large volume of volcanic deposits rather than magmatic heating, because the AFT age of the highest-elevation sample ( 08) is not reset.
For a geothermal gradient between 30 and 40°C/km (Springer & Förster, 1998), the recorded ∼50°C heating would correspond to an ignimbrite thickness between 1.7 and 1.3 km.Whereas the model without constraint suggests continuous final cooling since 25 Ma, the model including a surface constraint shows a two-stage cooling history, with final cooling starting at ~14 Ma (Figure 5d) that we related to major canyon incision, as observed southwards in the Ocoña and Colca Canyons (Gunnell et al., 2010;Schildgen et al., 2007Schildgen et al., , 2009;;Thouret et al., 2016).
We propose a palaeo-geographic model for the evolution of the WC, summarizing the geodynamic implications of our data integrated into previous regional models of exhumation (Figure 6).

Geochronological and thermochronological data indicate intrusion
emplacement and post-magmatic cooling during the Palaeocene (Figure 6a) followed by erosion of the WC forming an extensive Eocene erosional surface.Previous studies proposed regional rock uplift to occur from 50 to 14 Ma, less than 300 km from our study area south in the WC (Evenstar et al., 2020) and east in the Altiplano (Gérard et al., 2021).Rock uplift resumed at ca. 30-20 Ma in both Cordilleras and in the Altiplano (e.g.Sundell et al., 2019).Before ~30 Ma, the topography was at 1-2 km above the sea level, developing a low-relief plateau, which was subsequently capped by the extensive regional ignimbrites between ~30 and 20 Ma (Figure 6b; Thouret et al., 2016).In this region of the Central Andes, δ 2 H stable isotopic analyses of volcanic glass yield an elevation of ~2 km at ~20 Ma (Sundell et al., 2019).The present-day elevation of up to 3 km was only reached at ~15 Ma (Figure 6c), similar to inferred topographic histories in southern Peru (Ocoña-Colca region; Roperch TA B L E 4 Apatite (U-Th)/He data; AHe ages are corrected from alpha ejection using the F T factor (Ketcham et al., 2011).et al., 2021;Victor et al., 2004) accompanied by regional monoclinal tilting (Schildgen et al., 2009;Van Zalinge et al., 2017).There are no marked changes in regional tectonic uplift from North to South.

ID
Valley incision was a long-term response to plateau uplift and initiated after 15 Ma in Peru (Schildgen et al., 2009), after monoclinal new thermochronological data from the Cañete Canyon constrain the exhumation and the uplift history of the Western Cordillera on the Peruvian Pacific margin during the Oligo-Miocene Andean mountain building phase.This flat slab region (Nazca Plate subduction flattening initiated north of our site 15 Myr ago) corresponds to a transition area, and we document the evolution of the Cañete region since Oligocene-Miocene times, before Nazca Ridge arrival 4 Myr ago (addressed by Wipf et al., 2008).The significance of volcanic burial (by ignimbrites covering the bedrock and filling the early canyons) and its role in the thermal evolution of the upper crust has rarely been addressed before in a volcanic-arc setting.Numerous studies relying on thermochronological data to model exhumation or incision histories of the Andes do not consider this process.To our knowledge, only Colleps et al. (2021) applied apatite (U-Th)/He thermochronology to test the extent of the Deccan Traps on the Indian craton.
Location of the study area in South America.The white line shows the extent of the Andes along the western flank of the continent.(b) Topographic map centred on the Peruvian Andes showing the location of the Western Cordillera, Altiplano, Eastern Cordillera and Subandean region.Location of the low-temperature thermochronological data from this study (orange star in Cañete) and from the literature [white stars; 1: Margirier et al. (2015); 2: Ruiz et al. (2009); Gérard et al. (2021); 3: Schildgen et al. (2009, 2010); 4: Gunnell et al. (2010)] and incision data [white star 5: Evenstar et al. (2020)] are marked on the map.(c) Geological map of the study area showing the Cañete sampling sites and Nazca Ignimbrite in the Western Cordillera.Simplified sample numbers from this study are also reported.The Moquegua formation includes Cenozoic intrusive (int.) and volcanic (ign.: ignimbrite) rocks, Miocene Pisco sediments and Mesozoic Cenozoic intrusive (int.) and volcanic rocks.Palaeozoic, Proterozoic and Quaternary rocks occur mainly along the Coastal Cordillera.Along the Western Cordillera, the Cenozoic Nazca ignimbrite unconformably covers the outcropping Mesozoic (intrusive and volcanics) and Cenozoic intrusive rocks.Positions of palaeo-calderas and main faults are indicated.The white square locates the aerial photo of Figure 2. The black outline comprising the Cañete Canyon data and the dashed black line shows the location of the swath profile of Figure 3. [Colour figure can be viewed at wileyonlinelibrary.com] 7.8 ± 0.5, 7.1 ± 1.4 and 7.8 ± 1.7 Ma for the samples with elevations between 1485 and 2153 m (Table F I G U R E 2 Aerial photograph close to the Western Cordillera front (white line) near the city of Nazca, showing the bestpreserved ignimbrite deposits (extent outline by black dashed lines) in the Cañete-Nazca region.Maximum (3020 m asl) and minimum (1180 m asl) elevations are indicated.Deep canyons (>780 m), cutting through both Nazca ignimbrite and Cretaceous rocks, indicate an episode of incision after deposition of the ignimbrites.[Colour figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 4 AFT and AHe ages as a function of palaeo-depth below the ignimbrite for the Cañete profile samples.The AFT age of sample 08 reflects post-crystallization and pre-ignimbrite cooling of the bedrock.The AHe age of sample 08 was reset by burial heating below the ignimbrites but not its AFT age.The AHe age reflects post-ignimbrite exhumation cooling from about 15 Ma.AFT, apatite fission-track.[Colour figure can be viewed at wileyonlinelibrary.com]Both thermal histories indicate an initial cooling episode from ~55 to ~40 Ma, constrained by the crystallization age and the oldest AFT age of the profile (Figure 5a-d), but the later thermal histories differ.The thermal history without constraint indicates reheating from ~40 to ~30 Ma followed by renewed cooling from ~30 Ma to the presentday.The thermal history with an additional t-T constraint (samples close to the surface at 28 ± 2 Ma) indicates slow cooling from ~40 to 30 Ma, followed by a short reheating phase from 30 to 25 Ma.Subsequently, the model indicates two cooling phases, a first one from 25 to 20 Ma and a second cooling phase starting at ~14 Ma.

F
I G U R E 5 Time-temperature (T-t) modelling results for the Cañete Canyon profile using QTQt modelling, predicted (Pred) and observed (Obs) AFT and AHe ages.(a) T-t history modelled without additional constraints; associated predicted ages are shown in (b) (age − elevation plot) and (c) (predicted vs. observed age).(d) T-t history modelled with geological constraint (samples close to the surface at 30-26 Ma); associated predicted ages are shown in (e) (age-elevation plot) and (f) (predicted vs. observed age).Green boxes in (a) and (d) correspond to the pluton emplacement constraints (initial constraint in Table 4).Lines in (a) and (d) represent the best-fit T-t model for each sample of the profile, with the red and blue lines corresponding to the lowest and highest sample respectively.Red and blue shaded areas represent the uncertainty envelopes of the best-fit models for the lowest and highest sample respectively.The vertical light blue bars in (b, c, e, f) represent the mean 95% credible range for the predictions from all thermal-history models.In some cases, the best model age prediction is out of the 95% credible range.AFT, apatite fission-track.[Colour figure can be viewed at wileyonlinelibrary.com] correction factor(Ketcham et al., 2011).'Rs' is the equivalent-sphere radius calculated using the procedure ofGautheron and Tassan-Got (2010).'Age c' is the age corrected by the crystal geometry and ejection factor F T .The 1σ error of the AHe age is considered at 8%, reflecting the sum of errors in the ejection factor correction and age dispersion of the standards.a Corrected, AHe age from alpha ejection using the F T factor (fromKetcham et al., 2009).et al., 2011; Schildgen et al., 2009) or Chile (18° S to 22° S; Van Zalinge et al., 2017).The western margin of the Andes has been undergoing uplift since the early Miocene(Evenstar et al., 2020) or even earlier since the Oligocene(Barnes & Ehlers, 2009;Van Zalinge et al., 2017).Jeffery et al. (2013) argued that the evolution of the Ocoña Canyon is consistent with local plateau elevations of 1-3 km at 16 Ma and either steady or punctuated uplift of 1.5-3.5 km since.The uplift of this region from Peru to Chile is accommodated by reverse fault systems (Figure1; Benavente

F
Schematic models of the topographic and volcanic evolution of the Western Cordillera at 60 Ma, 30-20 Ma, 15-12 Ma and the present day.The orange line on the maps shows the location of the corresponding schematic cross-section.[Colour figure can be viewed at wileyonlinelibrary.com] deformation (from Nazca, Tacna in Peru to Oxaya in Chile; Figure 1) and oroclinal bending of the WC, as observed by palaeomagnetic data in southern Peru (Roperch et al., 2011).At ~8 Ma, the WC experienced a shift to more humid climatic conditions, with the development of large fluvial systems producing re-incision or development of erosional surfaces in the forearc (Figure 6d; Evenstar et al., 2020).6 | CON CLUS ION Thermal modelling of our low-temperature thermochronology data, together with geological data, demonstrates the influence of the Oligocene-early Miocene ignimbrite flare-up on the thermochronological record in the Western Central Andes.This study provides new constraints on landscape and relief evolution since ~55 Ma in the WC of southern Peru.Our results are compatible with previous data suggesting slow Palaeogene exhumation to the east in the Altiplano, but highlight this pattern in the forearc at this latitude for the first time.Additionally, they also demonstrate that slow exhumation of the WC extended latitudinally from 10° S to 17° S along the Western Andes in Peru, apparently independent of Nazca Ridge migration.Finally, the post-15 Ma exhumation of the WC in the Central Andes is captured in thermochronological records spanning the forearc from southern Peru to northern Chile.This tectonic episode induced major incision of canyons despite regionally contemporaneous low denudation rates in the forearc.
are reported in Table1.AFT, apatite fission-track.[Colour figure can be viewed at wileyonlinelibrary.com]Samples location and description.
Apatite fission-track (AFT) data.Fission tracks were counted by M. Balvay at ISterre GTC using a zeta value of 247.2 ± 16.1.The uncertainty of the AFT ages is given as a 2σ range.
TA B L E 2 Inversion parameters for QTQt modelling.TA B L E 3 a Not covered by ignimbrite.