Provenance analysis and thermochronology of the Chivillas Formation, Mexico: a record of basin formation and inversion in the southern Sierra Madre Oriental during the Early Cretaceous–Palaeogene

ABSTRACT Sandstone petrography, zircon and apatite U-Pb geochronology, and apatite geochemistry of Early Cretaceous siliciclastic turbidites interbedded with basalt (Chivillas Formation) in the southern Sierra Madre Oriental, Mexico, indicate local provenance from westerly basement sources. Three main detrital populations were observed in both the zircon and apatite geochronology: Meso–Neoproterozoic, Carboniferous–Permian, and Early Cretaceous. Apatite Sr/Y vs light rare earth element discrimination diagrams indicate that most Precambrian grains have affinity with high-grade metamorphic rocks and most Phanerozoic grains with igneous rocks. These results are consistent with derivation from exposed crystalline sources within the Acatlán-Oaxacan block including the Grenvillian Oaxacan Complex, the East Mexico Arc, and the Early Cretaceous continental arc. The compositional disparities observed in individual samples reflect differences in source of individual turbiditic flows comprising the Chivillas Formation. We surmise that various rivers from the northern and southern Acatlán-Oaxacan block drained into the basin during the Early Cretaceous. Furthermore, detrital apatite fission-track analyses of Chivillas sandstones yielded mostly 42–40 Ma ages (Eocene) that post-date the shortening of the Mexican orogen in the southern Sierra Madre Oriental. Thermal models calculated based on track lengths indicate rapid exhumation of Chivillas rocks at 46–36 Ma, which is the time of the development of the Tehuacán Valley and reactivation of the Oaxaca Fault System as a normal fault.


Introduction
Sedimentary provenance analysis allows linking basin stratigraphic records with their potential detrital sources (e.g. Pettijohn et al. 1987;Weltje and von Eynatten 2004). Provenance studies that combine conventional sandstone petrography with detrital zircon geochronology have been abundantly used to identify the types, ages, and tectonic evolution of detrital source rocks (e.g. Gehrels 2008b, 2009a;Lawton et al. 2018). Such kind of analyses have been successfully applied in palaeogeographic and tectonic reconstructions of Mexico. For example, they have allowed identifying western equatorial Pangea as the source of the Triassic successions deposited before the opening of the Gulf of Mexico in northern Mexico (e.g. Barboza-Gudiño et al. 2010;Ortega-Flores et al. 2016), and they have provided evidence for a sedimentological connection between the southwestern Laurentia Colorado Plateau erg systems with Jurassic strata in northern Mexico (e.g. Lawton et al. 2018). Recently, Martin et al. (2022) interpreted a mainly local sediment source for the Upper Triassic-Jurassic strata from northeastern Mexico with a transport distance no longer than 100 km.
Provenance studies have also allowed to recognize metamorphic and igneous basement and older sedimentary successions as the sources for Upper Jurassic-Early Cretaceous strata that were the fill of basins formed by back-arc extension in southern mainland Mexico (Sierra-Rojas et al. 2016). The stratigraphic records, facies, and depositional models of these basins have been widely studied (Mendoza-Rosales 2010; Mendoza-Rosales et al. 2010, 2013Sierra-Rojas and Molina-Garza 2014;Sierra-Rojas et al. 2016). However, there are cases of exposed inverted basins with intricate structures, complex thermal histories, and metamorphism, features which complicate provenance studies. Such is the case of the Early Cretaceous continental Chivillas basin in the southern Sierra Madre Oriental (Figures 1 and 2), which was contracted and slightly metamorphosed during the Late Cretaceous-Palaeocene Mexican orogen event    Angeles-Moreno 2006;Fitz-Díaz et al. 2018;Graham et al. 2020): disentangling its geologic history requires integrating sedimentary provenance techniques and thermochronology.
In this contribution, we provide a multiproxy provenance analysis to elucidate the detrital sources of the Early Cretaceous Chivillas Formation. The multiproxy approach integrates detrital zircon and apatite U-Pb geochronology and geochemistry data, as well as petrographic information. The new data allow us to propose palaeogeographic constraints for the Barremian-Aptian of southern Mexico. Additionally, we present apatite fission-track data from the northern part of the southern Sierra Madre Oriental. The new apatite fission-track ages and the thermal modelling, combined with available geologic constraints, provide new insights into the Cretaceous-Palaeogene inversion of the Chivillas basin.

Geologic setting
The Mexican territory is characterized by a series of crustal blocks with disparate geologic evolutions separated by large-scale structures; this feature reflects an intricate Mesoproterozoic-Cretaceous tectonic evolution. According to Sedlock et al. (1993), these blocks or tectonostratigraphic terranes in southern Mexico include the Zapoteco, Mixteco, Maya, and Cuicateco as depicted in Figure 1 A and B.
The Oaxacan Complex (Fries and Schmitter 1962) is the basement of the Zapoteco terrane (Figure 1 B and C) and it is the oldest of the various crystalline complexes that crop out in southern Mexico. This complex is composed of granulite-facies rocks including orthogneiss, paragneiss, amphibolite, charnockite, anorthosite, migmatite, and calcsilicate rock (Ortega-Gutiérrez 1977;Keppie et al. 2003;Solari et al. 2003;Ortega-Gutiérrez et al. 2018). The ages of the protoliths range between ca. 1.3 and 1.0 Ga (Keppie et al. 2001Solari et al. 2003;Keppie and Dostal 2007).
The granulite-facies metamorphic event in the Oaxacan Complex occurred at ca. 1000-990 Ma, with retrogression through amphibolite facies at ca. 980 Ma . After the metamorphic event, rocks cooled through ca. 400-500°C between ca. 980 and 940 Ma, as indicated by 40 Ar-39 Ar ages obtained on hornblende and phlogopite Keppie et al. 2004b). Additionally, a group of post-tectonic pegmatites yielded U-Pb zircon ages in the ca. 980-930 Ma range (Shchepetilnikova et al. 2015;Shchepetilnikova 2018). The complex is covered discordantly by the Palaeozoic sedimentary Tiñu, Santiago, and Ixtaltepec Formations, which include limestone, siltstone, shale, and sandstone (Pantoja-Alor and Robison 1967). The detrital zircon ages in these formations are mainly Proterozoic with subordinate clusters of Palaeozoic ages (Gillis et al. 2005).
The Acatlán Complex (Ortega-Gutiérrez 1978) is the basement of the Mixteco terrane ( Figure 1, B and C). The complex is composed of Palaeozoic rocks with various metamorphic grades and a convoluted tectonothermal history (Nance et al. 2006). The low-and medium-grade metamorphic rocks consist of Cambrian-Ordovician metasedimentary rocks intruded by mafic and felsic igneous bodies between ca. 480 and 440 Ma, Devonian-Permian shallow marine successions, and mafic volcanic rocks (Ortega-Gutiérrez et al. 1999;Keppie et al. 2008b). The complex includes Mississippian high-pressure metamorphic rocks (Estrada-Carmona et al. 2016), whose protoliths were metasedimentary rocks and maficultramafic units intruded by bimodal magmatic rocks (Keppie et al. 2008b).
In the Mixteco and Zapoteco terranes, a series of late Palaeozoic plutons (Figure 1, B and C) have been considered part of the continental East Mexico Arc developed across Mexico and Central America (Torres et al. 1999;Dickinson and Lawton 2001;Kirsch et al. 2012;Ortega-Obregón et al. 2014). The igneous bodies range in size between stock and batholith and are mainly made up of granite, granodiorite, diorite, and tonalite (Torres et al. 1999;Kirsch et al. 2012;Ortega-Obregón et al. 2014).
The Matzitzi Formation (Calderón-García 1956) unconformably overlies the Oaxacan and Acatlán complexes (Elías-Herrera and Ortega-Gutiérrez 2002). The formation is composed of interbedded sandstone, scarce conglomerate, dark shale, and siltstone (Centeno-García et al. 2009). In addition, some beds contain late Permian paleoflora (Flores-Barragán et al. 2019). Bedoya et al. (2021) proposed that the deposits defined as Matzitzi Formation are two different units: (1) Permian fluvial deposits with provenance from the Oaxacan and Acatlán complexes, and subordinately from the East Mexico Arc and (2) an informal unit named Agua de Mezquite formation (post-177 Ma) derived from the Oaxacan Complex, the East Mexico Arc, Lower-Middle Jurassic volcano-sedimentary rocks, and carbonate successions. For the first unit, Bedoya et al. (2021) suggest keeping the name Matzitzi Formation because it comprises the outcrops on which the formation was originally defined.
The Acatlán and Oaxacan complexes were amalgamated across the Caltepec fault zone and welded by the Cozahuico granite during the early Permian (Elías-Herrera and Ortega-Gutiérrez 2002). Since the Cretaceous, both complexes have formed a single basement unit with similar stratigraphic and tectonic records: The Acatlán-Oaxacan block (Nieto-Samaniego et al. 2006).
The Maya terrane is a block in southeastern Mexico (Isthmus of Tehuantepec to Yucatan Peninsula), northern Guatemala, and Belize (Sedlock et al. 1993). In Mexico, this terrane is composed of crystalline rocks from the Grenvillian Guichicovi Complex (Murillo-Muñetón 1996), the Ordovician El Triunfo Complex , the Permian-Triassic Chiapas Massif (Weber et al. 2005), and the Permian-Jurassic La Mixtequita suite (Murillo-Muñetón 1996). The terrane includes low-grade Carboniferous metasedimentary rocks from the Lower Santa Rosa Formation (López-Ramos 1979). The latter is unconformably overlain by sandstone and shale of the Upper Santa Rosa Formation (Pennsylvanian?; López-Ramos 1979).
The basement rocks of the Maya terrane are also exposed in northwest and central Guatemala. These include the low-to high-grade metamorphic rocks of the Chuacús Complex (Ortega-Gutiérrez et al. 2004), and amphibolite, schist, migmatite, and granitoids in the Altos Cuchumatanes (Anderson et al. 1973). Permian strata in northern Guatemala are composed of shale (Tactic Formation;Walper 1960) and limestone (Esperanza and Chóchal formations; Vachard et al. 1997Vachard et al. , 2000. The basement of the Maya terrane is locally covered by Jurassic to Early Cretaceous siliciclastic and minor volcanic rocks of the Todos Santos Formation (Anderson et al. 1973;Godínez-Urban et al. 2011;SGM, Servicio Geológico Mexicano 2014 and references therein).
The Cuicateco terrane includes an inverted sedimentary basin with NW-SE orientation and northward wedge-shaped tapering (Ortega-Gutiérrez et al. 1995;Pérez-Gutiérrez et al. 2009) (Figure 1 A and B). In the west, the Cuicateco terrane is in fault contact with the Zapoteco terrane through the Oaxaca Fault System and in the east with the Maya block along the Vista Hermosa fault ( Figure 1B).
The western part of the Cuicateco terrane ( Figure 1C) includes the Sierra de Juárez Mylonitic Complex (Alaniz-Álvarez et al. 1994 The Mazateco complex is exposed in the eastern part of the Cuicateco terrane ( Figure 1C). The complex comprises the amphibolite-facies La Nopalera schist and the greenschist-facies Mazatlán de las Flores schist (Angeles-Moreno 2006). The contact between these two units is cryptic because they are separated by a tectonic wedge in which the Chivillas Formation is located (Angeles-Moreno 2006).
The Chivillas Formation (Pano 1973), which is the focus of this paper, is exposed on the highest parts of the Mazateca range in the northern Cuicateco terrane ( Figure 1C). At the type locality (San Antonio Cañadas, Puebla; Pano 1973; Figure 1C Figure 1C).
The shortening in the Chivillas formation was produced by the Mexican orogen (Angeles-Moreno 2006), as is  below. That deformation is characteristic of the Mazateca range; therefore, the ca. 5 km thickness estimated by Mendoza-Rosales et al. (2013) may be an overestimation.
The Chivillas Formation has been interpreted as the accumulation of turbidite deposits in a basin associated with (a) a dextral pull-apart basin related to the separation between North and South America (Angeles-Moreno 2006); (b) a marine rift basin linked with the Gulf of Mexico ridgetransform system (Mendoza-Rosales et al. 2010); a back-arc basin associated with roll-back of the Arperos slab (Coombs 2016;Sierra-Rojas et al. 2016), or an oblique hyper-stretched intra-arc basin between continental crust blocks .
The Chivillas basin paleobathymetry is unknown, but two arguments allow concluding that the Chivillas Formation was deposited in a water column less than 500 m: (a) Chivillas pillow lavas have vesicles with diameter >1 mm ( Figure 2E), indicating that the water column depth ranged 330-550 m (Moore 1965;Jones 1969;Merle et al. 2005); (b) the presence of ammonites, which is indicative of epicontinental seas reaching ca. 500 m depth (Naglik et al. 2015).
The Jaltepetongo Formation has been correlated with the Chivillas Formation, although the fossil record indicates that it may be somewhat older . The Jaltepetongo is a marine succession (Ortega-González and Lambarria-Silva 1991) exposed south of Teotitlán ( Figure 1C) that comprises a lower conglomerate and coarse-grained sandstone unit and an upper very finegrained sandstone and shale unit . The thickness of the Jaltepetongo Formation is uncertain due to strong deformation. The age of the Jaltepetongo has been determined as Berriasian-Barremian based on fossils such as Protocardium hillanum and Acteonella sp. (Ortega-González and Lambarria-Silva 1991). The detrital zircon U-Pb geochronology in sandstones of Jaltepetongo indicates it was sourced from Grenville rocks within the Oaxacan Complex .
During the Late Cretaceous and Early Cenozoic, the Mexican orogen produced compressive structures that episodically migrated from western to eastern Mexico (Cuéllar-Cárdenas et al. 2012;Garduño-Martínez et al. 2015;Fitz-Díaz et al. 2018). In southern Mexico, the Mexican orogen deformation took place between Santonian-Campanian and middle Eocene; this process resulted from the collision of the Acatlán-Oaxacan block with the Sierra de Juárez Complex (Nieto-Samaniego et al. 2006). The shortening produced regional thrust faults, tectonic wedges, and inversion of the Chivillas basin (1) onset of extension constrained by the accumulation of early-middle Eocene Tilapa red beds ( Figure 1C) discordantly deposited over pre-Cenozoic units (e.g. Chivillas Formation); (2) development of relay ramps associated with the propagation of deformation during middle Eocene-middle Oligocene; (3) a pulse represented by a disconformity between the late Eocene and late Oligocene units; (4) a pulse associated with progressive deformation within the fault system that ended with the accumulation of the last lacustrine deposits affected by faulting during the Pliocene-Pleistocene.

Methods
Previous work has focused on outcrops along Las Salinas creek (Mendoza-Rosales 2010; Mendoza-Rosales et al. 2010, 2013 because of the preservation of sedimentary features and the lack of outcrop-scale deformation. Nevertheless, we sampled the Chivillas Formation in a broader area because our goal was to conduct a more representative study. We sampled turbidites, specifically medium-to coarse-grained sandstone beds, in well-exposed outcrops with minimum weathering, in five localities ( Figure 1C): LS-2 northeast of Tehuacán in Las Salinas creek, RN-3 and RN-4 northwest of Santa Maria, V-1 and V-2 northeast of Teotitlán de Flores and southeast of Coxcatlán. Details about Chivillas Formation outcrop descriptions and lithofacies can be found in supplemental online material 1.
Because of the structural complexity of the unit and possible repetitions of sections due to thrusting, it was only possible to determine the stratigraphic position of sample LS-2 with certainty ( Figure 3).
Petrographic analysis of the samples included textural parameters and framework grain identification. For sample LS-2, the sandstone modal composition was analysed according to the Gazzi-Dickinson method (Ingersoll et al. 1984), counting 500 framework grains. For the remaining samples, in which deformation and very low-grade metamorphism partially modified the primary features of the rocks, the framework grain content was visually estimated. Overall, considering the slightly metamorphic character of the Chivillas Formation, we deemed conventional petrography as of limited utility.
Zircon and apatite separation for geochronology consisted of crushing, sieving using mesh sizes 450 and 180 µm, magnetic separation using a Frantz®, and density separation using bromoform (CHBr 3 ). The separated zircon and apatite grains were randomly mounted under a binocular microscope and a 1-inch disk was made using epoxy resin. Afterwards, the mounts were polished to expose the zircon and apatite crystals. Additionally, the polished apatite grains were etched in 5.5 M nitric acid (HNO 3 ) at 21°C for 20 s to reveal spontaneous fission tracks (Donelick et al. 2005).
The U-Pb isotopic analyses were performed using Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) at Laboratorio de Estudios Isotópicos, Centro de Geociencias, UNAM, utilizing a Resonetics Resolution M50 193-nm excimer laser workstation coupled with a Thermo ICap Qc quadrupole mass spectrometer. We followed the methodologies reported in Solari et al. (2018) for zircon and Ortega-Obregón et al. (2019) for apatite. We used a 23 µm-diameter spot, 5 Hz repetition rate, and 6 J/cm 2 energy density for zircon analyses and a 60 µm-diameter spot, 4 Hz repetition rate, and 4 J/cm 2 energy density for apatite. During the analytical sessions, signals of U and Pb isotopes and trace elements of both mineral phases were measured.
Zircon ages were determined from the 206 Pb/ 238 U isotopic ratio for those grains with an apparent age ≤800 Ma and from the 207 Pb/ 206 Pb for those grains with an apparent age >800 Ma. Analyses were deemed unreliable if the following criteria were met: (1) a discordance greater than 5% or 206 Pb/ 238 U error >4% for analyses with an apparent age >400 Ma and (2) 207 Pb/ 206 Pb error >20% or 206 Pb/ 238 U error >10% for analyses with an apparent age <400 Ma.
The Chivillas Formation maximum depositional age was calculated using the youngest zircon analyses set whose ages overlapped at the 2σ-error level (Dickinson and Gehrels 2009b). The Wetherill and Tera-Wasserburg Concordia diagrams were plotted using IsoplotR (Vermeesch 2018) and the kernel density models were produced using RadialPlotter (Vermeesch 2009).
Apatite U-Pb ages were obtained from samples LS-2, RN-3, and V-1. Most analysed grains were subroundedrounded and >60 µm along their longest dimension. Each analysis was performed in individual apatite grains. The apatite U-Pb analyses form linear arrangements in Tera-Wasserburg Concordia plots due to mixing between a nonradiogenic Pb component and a radiogenic Pb component. These discordia arrays were used to discriminate grain populations with similar ages (Mark et al. 2016;Gillespie et al. 2018). The lower intercept ages of each array were calculated by free fit to the data, without anchoring to an initial 207 Pb/ 206 Pb value, using IsoplotR (Vermeesch 2018).
Previous to laser ablation, we selected those etched apatite grains greater than 60 µm that showed homogeneous and quantifiable track distributions to measure fission-track densities and lengths using a Zeiss Axio Scope-A1 microscope at Laboratorio de Termocronología (LaTer), Centro de Geociencias, UNAM. The 238 U content was measured in-situ by LA-ICP-MS (Hasebe et al. 2004;Donelick et al. 2005), following the methodology reported in Abdullin et al. (2018); (2021) and Ramírez-Calderón et al. (2021), during the same analytical session used for the U-Pb geochronology. The results of fission-track analyses are presented in radial plots obtained using RadialPlotter (Vermeesch 2009).
The time-temperature thermal models were built in HeFTy (Ketcham 2005) using the apatite fission-track ages, track length distributions, and the chlorine content (Cl wt.%) as kinetic parameter (Ketcham et al. 2007). The constraints used for thermal modelling include the apatite fission-track ages and a temperature higher than the partial annealing zone upper limit of the fission-track system (>120°C).

Sandstone petrography
The samples are medium-to coarse-grained sandstones with moderate to very poor sorting ( Figure 4). The framework grains are mainly angular to subrounded monocrystalline and polycrystalline quartz ( Figure 4, A and B), foliated and non-foliated quartz grains (Figure 4, B and E), K-feldspar (some with micrographic texture), plagioclase ( Figure 4E), felsitic, microlithic and vitric volcanic lithics, metamorphic lithics with strong cleavage and schistosity ( Figure 4C), and very scarce sedimentary lithics. Contacts are chiefly concave-convex and sutured, the plagioclase content is greater than K-feldspar, and the felsitic lithic content is larger than that of the microlithic grains.
The quartz grains show evidence of dissolution, polygonization, and fractures mostly filled with calcite. The feldspars are partially altered to sericite. The sandstone cement is calcite ( Figure 4C), and this mineral commonly replaces detrital feldspar, quartz, and rock fragments. The pseudomatrix is composed of deformed labile grains of muscovite and feldspar. The sandstone accessory minerals include zircon, apatite, muscovite, titanite, rutile, tourmaline, chlorite, garnet, allanite, and haematite.
Most samples, except LS-2, show primary textural features somewhat modified by incipient metamorphism, including neoformation of white mica associated with a poorly developed cleavage ( Figure 4C) and stylolites made up of haematite and clay ( Figure 4F). Therefore, a modal analysis was only conducted for LS-2 sample, whereas in other samples the grain content was visually estimated. The sandstones were classified according to Folk (1980) as lithic arkose (LS-2, V-1), lithic arkose to feldspathic litharenite (V-2), and arkose (RN-3, RN-4) ( Figure 5). We did not use the ternary compositional plots of Dickinson (1985) or Garzanti (2019) because the classification could be misleading due to the slightly metamorphosed nature of most samples.

Detrital zircon U-Pb geochronology
A total of 136 analyses were conducted in sample LS-2. One hundred twenty-three analyses were deemed reliable, of which 119 correspond to individual grains and four analyses to the cores and rims of two grains. Three main age groups were recognized ( Figure 6A); they have distinct grain shapes and features under cathodoluminescence (CL). First, a Meso-Neoproterozoic group (69 analyses) has ages ranging ca. 1360-870 Ma with most Th/U ratios between 0.12 and 0.91. The grain size of this group ranges 150-400 µm along the longest dimension, with rounded to sub-rounded shapes and aspect ratio between 1:1 and 3:1. Under CL, these grains are mainly dark with homogeneous texture or black cores bordered by a bright rim. According to this CL, we interpret this group as produced by a metamorphic event. A second, Permian-Triassic group (38 analyses) ranges ca. 281-216 Ma with Th/U ratios between 0.1 and 1.2. The zircon crystals are prismatic, their aspect ratio is 2:1-4:1, and the grain size ranges 130-380 µm. Under CL, the grains show concentric zoning typical of igneous growth (Corfu et al. 2003). A third, Jurassic group (11 analyses) ranges ca. 201-159 Ma and most of their Th/U ratios are between 0.2 and 1.5. The zircon grains are long prismatic with bipyramidal terminations and an aspect ratio of 3:1-5:1. The grain size ranges 160-400 µm along the longest dimension. Under CL, the grains display concentric zonation patterns. We interpret an igneous origin for this detrital grain group. Furthermore, a Palaeoproterozoic grain (1687 Ma), Calymmian grains (1478 and 1408 Ma), and Early Cretaceous grains (141 and 139 Ma) were identified. The latter are prismatic, bipyramidal, and euhedral with concentric CL zoning, and we interpret them as of volcanic origin. In this sample, we did not recognize zircon grains with depletion in heavy rare earth elements (HREE) that would suggest growth concomitant with garnet.
In sample RN-3 ( Figure 6B), 133 analyses were performed. One hundred twenty-eight analyses were accepted; they were performed in individual grains but in one case, in which the core and the rim of an individual grain were analysed. We identified two chief age populations. The oldest is a Meso-Neoproterozoic group (ca. 1375-959 Ma; 33 analyses) that has Th/U ratios varying over a broad range (0.02-0.57) and rather disparate REE patterns. The grain size of this zircon population ranges 100-450 µm along the longest dimension with an aspect  (Folk, 1980) for detrital modes of sandstones from Chivillas Formation. For sample LS-2, the sandstone modal compositional was analysed according to the Gazzi-Dickinson method (Ingersoll et al. 1984). For the remainder samples, the framework grain content was visually estimated. ratio between 1:1 and 3:1. The grains are ovoid or prismatic with rounded pyramids. Under CL, the former are dark with a homogenous texture or a dark internal zone mantled by a bright rim. The least rounded grains show homogeneous textures, patchy zoning, core and rim texture, or concentric zoning. We interpret that most of these detrital grains possibly have a metamorphic origin, and a few may be of igneous origin. The second is a Permian-Triassic group with ages that range from ca. 294 to 219 Ma (90 analyses) yielding a median of 273 +2 / −4 Ma. Most analyses have Th/ U = 0.3-1.2 and REE spectra with HREE enrichment relative to light rare earth elements (LREE). The grain size of this group ranges 150-580 µm along the longest dimension with an aspect ratio between 1:2 and 1:4. The zircon shapes are prismatic with pyramidal terminations and concentric zoning under CL. We infer that these zircon grains are likely of igneous origin. Additionally, we identified in the sample individual grains that yielded Middle Jurassic (168 Ma), Ordovician (451 Ma), Cambrian (507 and 520 Ma), and Ediacaran (576 Ma) ages.
One hundred three analyses were performed in zircon of sample RN-4 ( Figure 6C). Ninety-two analyses were accepted; they correspond to individual grains and to two grains in which both cores and rims were analysed. The largest population is Carboniferous-Permian with ages ranging ca. 308-252 Ma (n = 85), with a median of 279 ± 3 Ma. The Th/U ratios of this grain group range 0.1-1.4. The REE patterns show enriched HREE and depleted LREE. The zircon grains are prismatic with bipyramidal terminations, an aspect ratio between 2:1 and 4:1, and their size ranges 140-340 µm along the longest dimension. Under CL, the zircon grains show internal zoning typical of igneous growth (Corfu et al. 2003). Based on the above features, we interpret an igneous origin for this detrital population. In addition to a Carboniferous-Permian population, individual grains in the sample yielded 1032, 938, 505, 367, 244, 232, and 214 Ma.
In sample V-1 ( Figure 6D), 117 analyses were performed, of which 108 were deemed reliable. Two main age populations were identified, and they correlate with zircon grain shapes and CL features. The Meso-Neoproterozoic group ranges from ca. 1220 to 890 Ma (37 analyses) with most of their Th/U ratios between 0.1 and 0.6 and diverse REE patterns. The zircon grains of this population are mostly ovoid to prismatic with rounded pyramids. Under CL, the ovoid grains show homogeneous texture or core-rim textures. The less rounded grains mostly display homogeneous texture or cores mantled by rims. Most of these detrital zircon grains may be of metamorphic origin. A second, Permian-Triassic group (53 analyses) ranges in age from ca. 275 to 207 Ma. The Th/U ratio is 0.12-1.20 and the REE patterns show HREE enrichment compared to the LREE. The zircon grains are long prismatic to stubby with bipyramidal terminations. Grain size ranges 110-470 µm along the longest dimension with an aspect ratio between 2:1 and 4:1. Under CL, the grains display zoned textures, a feature typical of magmatic zircon (Corfu et al. 2003). Additionally, we recognized the following groups: Calymmian (Mesoproterozoic; 1590-1540 Ma; n = 4), Cryogenian-Ediacaran (Neoproterozoic; 730-550 Ma; n = 6), Cambrian-Devonian (508-419 Ma; n = 5), and Lower Jurassic (194,187,and 185 Ma). One Ediacaran analysis is characterized by its depressed HREE, suggesting concomitant zircon and garnet growth (Rubatto 2002). Under CL, the Calymmian grains are rounded with homogeneous textures, the Cryogenian-Ediacaran grains are mainly dark with various shapes, the Cambrian-Devonian grains are mostly prismatic with homogeneous textures, and the Lower Jurassic grains are prismatic fragments with internal zoning.
Eighty-one analyses were performed in zircon grains from sample V-2 ( Figure 6E). Seventy-nine analyses correspond to individual grains and two analyses to the core and rim of one grain. Seven analyses were not further considered, based on the filtering criteria. The largest and youngest population in the sample is Upper Jurassic (Tithonian)-Early Cretaceous, 146-118 Ma (48 analyses). The maximum depositional age was calculated by the mean of the six youngest overlapping ages of the latter group (Dickinson and Gehrels 2009b), which yielded 124.2 ± 0.8 Ma (MSWD = 1.4; Figure 6F). This maximum depositional age is analogous to the 126 Ma reported by Mendoza-Rosales et al. (2013). The Early Cretaceous population has Th/U ratios ranging 0.20-1.20. This population is characterized by prismatic crystals with bipyramidal terminations or by euhedral crystal fragments. The grain size of complete crystals ranges from 150 to 360 µm with an aspect ratio of 2:1 to 3:1. Under CL, all grains exhibit concentric zoning, a characteristic of igneous zircon grains (Corfu et al. 2003). We interpret these grains to be of igneous origin. Pre-Cretaceous analyses in the sample are Paleoproterozoic (1781 Ma
Overall, based on all recognized linear arrays, we inferred three main sources of detrital apatite: Tonian (Neoproterozoic), Permian, and Early Cretaceous.

Detrital apatite geochemistry
Apatite geochemistry was obtained on the same grains analysed for U-Pb geochronology. Apatite geochemistry data were plotted on a principal component analysis (PCA) diagram, using the Provenance package for R (Vermeesch et al. 2016), to identify clusters based on the following geochemical features: total rare earth element concentrations (ΣREE in wt.%), light rare earth elements (LREE) defined as La-Nd (O'Sullivan et al. 2020), chondrite-normalized contents of La and Lu, Y (ppm), Sr (ppm), Eu anomaly (Eu/Eu*), Th/U ratio, and Ce/Yb ratio as a proxy to REE slope. These variables have been used to distinguish between apatite from igneous and metamorphic rocks and in provenance studies for linking the apatite grains with their potential sources The lines across the linear arrays are discordia lines, whose lower intercepts were calculated by free fit to the data, without anchoring to an initial 207 Pb/ 206 Pb value. The lower intercept for each linear array was regarded as the age of the group.
The Eu anomaly is very variable in the analysed apatite grains ( Figure 8C). In Tonian analyses, the Eu anomaly ranges from strongly negative to absent, with Eu/Eu* values between 0.04 and 1.00. In this group, 77% of analyses have an Eu/Eu* ratio <0.5 and 85% have a ΣREE >0.4%. Permian grains have an Eu anomaly between very negative to positive, with Eu/Eu* between 0.10 and 1.40. In most of these analyses, the Eu/Eu* ratio is >0.5 (81%) and the ΣREE <0.4% (85% of the analyses) ( Figure 8C). In the Early Cretaceous group, the Eu anomaly is negative with Eu/Eu* values ranging 0.08-0.80 ( Figure 8C).
The Sr/Y vs LREE biplot of O'Sullivan et al. (2020) has been used to infer the source rock of detrital apatite grains. The Sr/Y vs LREE biplot ( Figure 8D) classify most Chivillas apatite grains in four fields: (1) felsic granitoids, S; (2) mafic granitoids-mafic igneous rocks, IM; (3) highgrade metamorphic and anatectic rocks, HM; and (4) low-to medium-grade metamorphic, LM. Most Neoproterozoic grains plot within the IM, S, and HM fields; most Permian grains plot in the IM and LM fields; and most Early Cretaceous grains plot in the IM field.

Detrital apatite fission-track thermochronology
Apatite fission-track ages were obtained for samples LS-2, RN-3, and V-1. The number of dated grains ranged from 40 to 90 per sample. The fission-track lengths were not measured in sample LS-2 because the grains had too low track densities.
Overall, the single-grain apatite fission-track ages range from 147 ± 23 to 10 ± 4 Ma. Samples LS-2 and RN-3 yield a central age of 21 ± 1 and 40 ± 1 Ma, respectively (Figure 9, A and B). Both samples passed the chisquared probability test [P(χ 2 ) ≥5%], indicating that the grains belong to a single age population (Galbraith 1981). Sample V-1 failed the chi-squared probability test, suggesting that the central age represents a mixedage ( Figure 9C). We used RadialPlotter (Vermeesch 2009) to decompose the mixed ages and obtain the minimum age model, as most ages are overlapping in the youngest component. The minimum age calculated by sample V-1 is ca. 42 ± 1 Ma ( Figure 9C).
In samples V-1 and RN-3, Cl concentrations of apatite grains are between 0.14 and 1.70 wt.%. Fission tracks in apatite grains with high chlorine levels (Cl-rich apatite) are more resistant to thermal annealing than those in fluorapatite crystals (Donelick et al. 2005). This is due to Cl content being correlated with the fission-track annealing characteristics. Thus, thermal models were performed for the apatite grains with Cl <0.9 wt.% in sample V-1 and Cl <1.0 wt.% in RN-3. In these ranges, both samples have a unimodal track length distribution (Figure 10), with a mean track length value of 13.08 ± 1.27 (SD) µm and 13.30 ± 1.35 (SD) µm, respectively. We considered a partial annealing zone between 60°C and 120°C, typical of F-apatite grains (Gleadow and Duddy 1981;Green et al. 1986;Gleadow et al. 2002;Donelick et al. 2005;Fitzgerald and Malusà 2019). The time-temperature (t-T) models indicate rapid cooling through the partial annealing zone during the Eocene, with a temperature decrease from 120°C to 60°C during ca. 46-36 Ma (Figure 10).

Pre-Cretaceous provenance of the Chivillas Formation
Most of the Chivillas samples, except LS-2, are very lowgrade metamorphic rocks containing stylolites and neoformed white mica associated with a poorly developed cleavage. We therefore have not used sandstone tectonic discrimination diagrams for most thin sections, as these would lead to dubious conclusions.
Nonetheless, we were able to establish that the primary composition of most analysed Chivillas samples corresponds to arkoses ( Figure 5). Overall, petrography indicates that the detrital source of Chivillas are mainly granitoids, high-grade metamorphic rocks, and felsicintermediate volcanic rocks. This contention is consistent with the observed accessory mineral content, which includes tourmaline, garnet, rutile, and allanite.
The detrital zircon age spectra of the Chivillas Formation show three main age populations: Meso-Neoproterozoic, Carboniferous-Triassic, and Early Cretaceous ( Figure 6) The Meso-Neoproterozoic population (1400-910 Ma) is present in all studied samples and is characteristic of the Grenville Orogeny (e.g. Rino et al. 2008;Dickinson and Gehrels 2009a). Most zircon grains in this age range indeed have metamorphic features and could have been derived from the high-grade metamorphic rocks of the Oaxacan Complex (e.g. Keppie et al. 2003;Solari et al. 2003) (Figures 6 and 11). However, Grenvillian detrital zircon grains are widely distributed in pre-Cretaceous strata in many southern Mexico units (e.g. Martens and Molina-Garza 2021). For example, they are present in the Sample V-1. Gray color scale represents Cl content in wt%. The left side of the plots depict the 2σ uncertainty. The dotted lines represent the maximum depositional age of the Chivillas Formation. Sample V-1 does not pass the P(χ2) test, so the solid line depicts the minimum age of the mixed ages model. Symbols: n, grain number; P(χ 2 ), chi-squared test; σ/t, relative standard error; t/σ, precision. The plots were built using RadialPlotter (Vermeesch 2009). Palaeozoic sedimentary cover of the Oaxacan Complex (Tiñu, Santiago, and Ixtaltepec Formations; Gillis et al. 2005), the metasedimentary rocks of the Acatlán Complex and its sedimentary cover (Talavera-Mendoza et al. 2005;Kirsch et al. 2012), the late Palaeozoic Matzitzi Formation Martini et al. 2022), and the Late Triassic-Middle Jurassic Ayú Complex (Helbig et al. 2012). Therefore, it is unsound to propose links to a unique detrital source, i.e. the Oaxacan Complex, solely based on Grenvillian zircon grains.
On the other hand, partial derivation of Chivillas detritus from the Oaxacan Complex is strongly suggested by the concordant Tonian ages of apatite in sample RN-3 (993-901 Ma; Figure 7B). Apatite grains in detrital rocks are very likely derived from first-cycle detritus and their U-Pb ages indicate cooling through amphibolite-or high greenschist-facies conditions (Morton and Hallsworth 1999;Chew et al. 2011). Indeed, the Tonian cooling ages and the isochrons calculated for the discordant apatite analyses (933, 932, and 920 Ma; Figure 7) are consistent with the ca. 980 and ca. 940 Ma cooling ages obtained by other geochronologic systems with similar closure temperatures in northern Oaxacan Complex rocks Keppie et al. 2004b). Further support for Tonian apatite having been derived from the Oaxacan complex comes from the Sr/Y vs LREE diagram by O'Sullivan et al. (2020), which indicates that apatite grains with such ages were derived from highgrade metamorphic, anatectic, or granitic rocks ( Figure 8D). Such rock types are widely exposed in the Oaxacan Complex (Ortega-Gutiérrez 1977;Keppie et al. 2003;Solari et al. 2003). Modelling was carried out for the detrital apatite grains with Cl <0.9 wt.% for sample V-1 and Cl <1.0 wt.% for RN-3. The models were developed using HeFty (Ketcham 2005). The t-T models are valid for the partial annealing zone (PAZ) of the apatite fission-tracks. The black lines represent the best fit cooling path and dotted lines represent weighted mean fit cooling paths. The dark gray and light gray areas represent good fits (GOF>0.5) and acceptable fits (GOF>0.05), respectively (Ketcham 2005). Symbols: GOF, goodness-of-fit between the measured value and the model value; Ngr, number of apatite grains dated; Ntr, number of confined tracks measured; K, Cretaceous; P, Palaeogene; N, Neogene; Q, Quaternary.  (Rino et Tonian apatite derived from the Oaxacan Complex shows distinct geochemical features when compared with other apatite populations in the Chivillas Formation. We therefore propose that apatite chemistry alone may be useful as a provenance tracer. For instance, Tonian apatite in Chivillas is distinctly high in ΣREE, LREE, Y, and Eu/Eu* <1 ( Figure 8A-C). Furthermore, we propose that apatite in the field Y > 600 ppm and ΣREE >0.4 wt.% ( Figure 8B) is highly likely derived from the Oaxacan Complex. Consequently, the Y and ΣREE contents of detrital apatite could be used to distinguish Tonian sources from of other populations, but they are not helpful to distinguish among Phanerozoic groups.
The dated Chivillas samples contain abundant zircon with ages ranging latest Carboniferous to Triassic, indicative of derivation from the East Mexican Arc (Torres et al. 1999;Dickinson and Lawton 2001). Two distinct sets of samples can be recognized in terms of late Palaeozoic-Triassic zircon: whereas samples RN-3 and RN-4 have a main population with 310-255 Ma grains (medians of ca. 276 Ma), samples LS-2 and V-1 contain a younger population with grains yielding 275-220 Ma (Figures 6 and 11). This is a significant difference in terms of provenance. The 310-255 Ma magmatism in the Acatlán-Oaxacan block (Kirsch et al. 2012;Ortega-Obregón et al. 2014) has been subdivided into two geographically distinctive groups (Martens and Molina-Garza 2021): plutons in the south of the block near the Chacalapa and Caltepec faults are more extensive ( Figure 1B), have more juvenile Hf isotopes, and yield ages in the ca. 310-280 Ma range; plutons in the northeastern part of the block are smaller ( Figure 1B) and yield ages ranging ca. 272-255 Ma . Additionally, the Atolotitlán tuff (or sill), located northwest of the latter plutons, has yielded zircon ages of ca. 245-235 Ma (Bedoya 2018).
We conclude that samples RN-3 and RN-4 had contributions from detrital sources in the southern Acatlán-Oaxacan block, whereas samples LS-2 and V-1 had contributions from detrital sources in the northern part of the block. We cannot preclude that other Permian-Early Triassic source may have been involved, such as the Chiapas Massif (Weber et al. 2005(Weber et al. , 2007, La Mixtequita batholith (Murillo-Muñetón 1996), or central Guatemala (Ratschbacher et al. 2009;Solari et al. 2011;Milián de la Cruz 2013). But these sources are further away and there is no evidence of their exposure in the Early Cretaceous.
Chivillas sandstone derivation from igneous bodies in the Acatlán-Oaxacan block is also supported by the Palaeozoic U-Pb isochronic age of detrital apatite (289-260 Ma; Figure 7). Additionally, these apatite analyses also suggest that the Permian contribution is not recycled detritus from units such as the Matzitzi Formation , the Jurassic siliciclastic successions (e.g. Piedra Hueca or Otlaltepec Formations; Martini et al. 2016), or the Ayú Complex (Helbig et al. 2012).
The geochemical features of Permian detrital apatite support derivation from igneous bodies due to the following reasons: (1) the seemingly direct trend between the Y and ΣREE contents ( Figure 8B) that in igneous apatite have been associated with the increase in the degree of magmatic fractionation (Belousova et al. 2002;Morton and Yaxley 2007;Mercer et al. 2020); (2) their Eu anomaly variation between strongly negative to positive, which in igneous apatite has been interpreted as indicative of the degree of magmatic differentiation (Chu et al. 2009;Cao et al. 2012); and (3) most of the apatite grains overlap in composition with apatite of igneous affinity in Figure 8D. The East Mexico Arc bodies in the Acatlán-Oaxacan block are compositionally diverse, including granite, granodiorite, diorite, tonalite, and minor ultramafic rocks (Kirsch et al. 2012;Ortega-Obregón et al. 2014). Thus, the Chivillas apatite grains with the highest Y and REE content and negative Euanomaly can be correlated with the most felsic rocks while the lowest Y and REE contents and weak or absent Eu-anomaly with mafic compositions. Oxidized magma conditions are suggested by the presence of titanite and magnetite in igneous rocks (Takagi and Tsukimura 1997;Valencia-Moreno et al. 2006). Both mineral phases have been identified in the Etla granite (Ortega-Obregón et al. 2010), Totoltepec pluton (Kirsch et al. 2013), and Zanitza batholith . Thus, the presence of these minerals in plutons of the East Mexican Arc could explain the positive Eu-anomaly in some of the Chivillas apatite grains.
On the other hand, some Permian apatite grains plot far from the chief cluster in Figure 8D, especially those with LREE <500 ppm. This may suggest a subordinate metamorphic source for these apatite grains. Apatite with such chemical signature may be derived from the al. 2008; Dickinson and Gehrels 2009a); Oaxacan Complex ; Pan-African/Brazilian orogens (da Silva et al. 2005;Cordani and Teixeira 2007); granitoids in the Acatlán Complex (Keppie et al. 2008a); Carboniferous-Triassic East Mexican arc (Torres et al. 1999;Dickinson and Lawton 2001;Kirsch et al. 2012 (Dickinson and Gehrels 2009a). Additionally, these ages have been identified in Oaxacan Complex rocks  and metasedimentary units of the Acatlán Complex (Talavera-Mendoza et al. 2005). The late Paleoproterozoic-early Mesoproterozoic grains within Chivillas samples could have been derived from these basement provinces.
Cryogenian-Cambrian ages were identified in most samples, except LS-2 (Figures 6 and 11). Given that basement rocks of such age are nearly absent in Mexico (e.g. Martens and Molina-Garza 2021), these grains were likely recycled. Potential candidates are within the Mixteco terrane: The Cosoltepec, Magdalena, Chazumba, or Tecomate formations, the Ayú Complex, or sedimentary succession recycled from them, all of which contain Pan-African/Brasiliano zircon (Talavera-Mendoza et al. 2005;Kirsch et al. 2012).
An Ordovician-Silurian zircon group was identified in samples RN-3, V-1, and V-2 ( Figures 6 and 11). These grains were possibly derived from the abundant Ordovician-early Silurian granitoids in the Acatlán Complex (Miller et al. 2007;Keppie et al. 2008a). Similar ages have been reported in granitoids from the southern Chiapas Massif (Estrada-Carmona et al. 2012) andGuatemala (Ortega-Obregón et al. 2008;Solari et al. 2010). However, we discard these sources as there is no evidence of their exposure in the Early Cretaceous (e.g. Witt et al. 2012).
The Jurassic detrital zircons in our samples (Figures 6  and 11) were possibly derived from volcanic rocks related to the Early-Middle Jurassic Nazas igneous province (Dickinson and Lawton 2001;Parolari et al. 2022). These volcanic rocks were widely distributed in northern, eastern, and southeastern Mexico (Bartolini et al. 2003;Barboza-Gudiño et al. 2008;Lawton and Molina-Garza 2014). Jurassic volcanism has been also recognized in the Acatlán-Oaxacan block, such as the Diquiyú volcanics (Zepeda-Martínez 2013) and the Las Lluvias ignimbrite (Campa-Uranga et al. 2004), so they are not ruled out as possible sources. Recently, Sierra-Rojas et al. (2022) reported Jurassic ages in volcanic clasts from the Caltepec Formation (Mendoza-Rosales 2010). Therefore, these authors suggest magmatism contemporaneous with the Nazas igneous province in the Zapoteco terrane. The Ayú Complex could have been another source of the Jurassic detrital grains in Chivillas samples. The former has ages ranging 171-160 Ma associated with a migmatization event and intrusion of igneous bodies (Helbig et al. 2012).
Additionally, we conclude that samples RN-3 and RN-4 were derived from similar sources, in contrast to LS-2 and V-1, for the following reasons: (1) the more narrowly constrained age of the Carboniferous-Permian population; (2) the median ages of the latter population are the same within the error; (3) Triassic grains are almost absent in samples RN-3 and RN-4 but are abundant in samples LS-2 and V-1; and (4) RN-3 and RN-4 have higher feldspar content, especially plagioclase, and lesser lithic content than other Chivillas samples.

Influence of the Early Cretaceous continental arcs
The youngest age component in Chivillas sandstones is Early Cretaceous (139-118 Ma), which was identified in samples V-2 and LS-2 (Figures 6 and 11). The former sample is different from the rest, being characterized by a high content of volcanic lithic grains, relatively abundant Cretaceous zircon, and a low proportion of pre-Cretaceous zircon compared to the other samples ( Figures 6 and 11). Early Cretaceous zircon grains in sample V-2 provide a maximum depositional age of ca. 125 Ma for the Chivillas Formation ( Figure 6F), which is virtually alike to a 126 Ma reported by Mendoza-Rosales et al. (2010);(2013) and similar to the palaeontological ages in Alzaga and Pano (1989). The Early Cretaceous age component was also identified in detrital apatite grains (132 ± 12 Ma; Figure 7A). Although the isochron age is a little older, it overlaps within error with the maximum depositional age determined by detrital zircon U-Pb ages.
Taking into account that Chivillas bears Early Cretaceous fauna, the magmatic origin of the Early Cretaceous detrital zircons, and the igneous affinity of similarly aged detrital apatite grains (e.g. good correlation between Y and ΣREE contents; Figure 8B), we conclude that volcanic activity was coeval with Chivillas deposition. The above is also evidenced by the field relations of Chivillas lava flows (Coombs 2016) and the thin ashfall layers interbedded between the siliciclastic rocks ( Figure 2D). Mendoza-Rosales (2010) proposed that Cretaceous zircon in Chivillas sandstones were directly derived from interbedded Chivillas basalt. However, this interpretation is unlikely because (1) mafic rocks have low zircon fertility (Moecher and Samson 2006), (2) Zr concentration in Chivillas basalt is low (75-300 ppm; Mendoza-Rosales et al. 2010;Coombs 2016), and (3) we do not envision a mechanism by which zircon from within the interbedded lavas could become isolated and migrate into the adjacent sandstones. A better explanation for the source of Cretaceous zircon was given by Coombs (2016), who suggests a detrital derivation from the relatively local Xonamanca Formation (Carrasco et al. 1975) and/or the Teotitlán Complex. Nevertheless, the youngest rocks in these units have ages of ca. 133 Ma (Coombs 2016), which is somewhat older than the Chivillas Formation maximum depositional age. We therefore propose that the Early Cretaceous continental arc was too a source of the Chivillas basin. This arc is a belt of igneous rocks with ages that range from 137 to 130 Ma, located in the western margin of the Mixteca terrane. The magmatism can be extended to central Honduras (Chortis block; Rogers et al. 2007;Ratschbacher et al. 2009). Such rocks include intrusive bodies in the Xolapa terrane (Pozuelo Granite, ca. 129 Ma;Solari et al. 2007), and volcanosedimentary units such as Chapolapa Formation (maximum depositional age ca. 126 Ma; Hernández-Treviño et al. 2004), Taxco and Taxco Viejo (ca. 137-130 Ma;Campa-Uranga et al. 2012), and Zicapa Formation (maximum depositional age ~133 Ma and crystallization ages of ~126 Ma; Sierra-Rojas and Molina-Garza 2014). The continental arc includes felsic to intermediate deposits that could explain the kind of lithics observed in sample V-2, the thin ash layers observed interbedded within sandstone ( Figure 2D), and the maximum depositional ages calculated for the Chivillas Formation ( Figure 6F).

Significance of sample heterogeneity
The petrographic and geochronological disparity herein observed in the Chivillas samples has also been reported by Coombs (2016) and Sierra-Rojas et al. (2020). Such heterogeneity can be explained by the interbedding of individual submarine fans derived from disparate local sources, as suggested by Sierra-Rojas et al. (2020). We propose that there were various feeder rivers draining the Acatlán-Oaxacan block instead of a large river system, which agrees with the westerly sources proposed by Mendoza-Rosales et al. (2013). Some of these rivers drained detritus from southerly and others from northerly sources within the block (Figure 12). Southerly sources were mainly recognized in samples RN-3 and RN-4, while northerly sources in samples LS-2 and V-1. The contrasting detrital zircon age signature in nearby samples of Chivillas could also have been produced by the juxtaposition of tectonic slivers of Chivillas Formation during the Mexican orogen. Ultimately, the contrasting composition of Chivillas sandstones sampled close to each other may reflect a combination of both. Finally, a third, and perhaps less likely possibility, is that the Mexican orogen has mixed beds of the Early Cretaceous Chivillas Formation (samples LS-2 and V-1) with a pre-Cretaceous but post-Permian unit composed of analogous lithofacies (samples RN-3 and RN-4), hence very hard to distinguish in the field from the Chivillas Formation in the Mazateca range. We are not aware of such a unit having been recognized in the area.

Early Cretaceous geologic model
Our geologic model for the Chivillas basin is summarized in Figure 12. The maximum depositional age of the Chivillas Formation is the Barremian-Aptian (ca. 125 Ma; Figure 6F). Overall, Early Cretaceous tectonics of Mexico was influenced by subduction along its paleo-Pacific margin and the end stages of Gulf of Mexico opening (Martini and Ortega-Gutiérrez 2018;Pindell et al. 2020 and references therein). In southern Mexico, an isolated internal carbonate platform (Córdoba) had developed along the Mexican passive margin (Ortuño-Arzate et al. 2003). The Acatlán-Oaxacan block had been exhumed and formed a basement high, as established by apatite fission-track ages (140-130 Ma; Bedoya et al. 2017) and thermal models of the Oaxacan Complex .
East of the block, the Chivillas extensional basin was developing in a marine environment that probably did not exceed 500 m depth. The basin was being filled by turbidity currents that formed a set of submarine fans that were locally interbedded with basaltic lava flows and pillow basalt. Our provenance results indicate that the sources of sediment were shed chiefly from the west, in the Acatlán-Oaxacan block, confirming previous ideas by Mendoza-Rosales et al. (2010); (2013). The great disparity in terms of provenance among nearby samples strongly suggests derivation from diverse sources within the block (north and south), and that the submarine fans were probably being fed by discrete hydrologic basins, mainly along the eastern part of the block. Lithologically, the sources included the Oaxacan Complex and associated late Palaeozoic granitoids, and the Early Cretaceous continental arc. Recycled detritus from units covering the crystalline basement of the Acatlán-Oaxacan block were probably a minor component. Overall, and considering that detrital apatite mainly represents first-cycle detritus (Morton and Hallsworth 1999;Chew et al. 2011), we conclude that sediment transport from the eastern part of the Acatlán-Oaxacan block to the Chivillas basin was relatively short.
Our sampling was in an inverted basin produced by the Mexican orogen. This precluded analysing the evolution of the infill of the Chivillas basin, as this would have required knowing the stratigraphic position of the samples in a column. Such an approach could be attempted in future work, if undeformed areas beyond the well-studied Las Salinas creek , 2013 are found.

Cenozoic tectonic model for inversion and exhumation of the Chivillas basin
The apatite fission-track ages obtained in Chivillas samples indicate a postdepositional thermal reset between the Early Cretaceous and Miocene (Figure 9). The length of fission tracks and the t-T models suggest a fastcooling phase between ca. 46 and 36 Ma ( Figure 10).
As shown in Figure 13, the Chivillas basin was inverted by the shortening produced by the Mexican orogen in the Late Cretaceous and earliest Palaeogene (Fitz-Díaz et al. 2018) ( Figure 13A). In the Mazateca range, the orogen developed regional thrusts, uplifting and juxtaposition rocks produced at disparate structural levels (Angeles-Moreno 2006;Nieto-Samaniego et al. 2006) (Figure 1). In this context, slivers of the Oaxacan and Sierra de Juárez/Teotitlán complexes were thrust over the Chivillas Formation ( Figure 13A), heating the sedimentary rocks above the annealing temperature of apatite fission tracks, even producing local low-grade metamorphism.
During the Eocene, shortening had ceased and had been followed by the first extensional pulses of the (normal) Oaxaca Fault System (Nieto-Samaniego et al. 2006;Dávalos-Álvarez et al. 2007). This developed the Tehuacán Valley and the Mazateca range (Figures 1 and  13B). The minimum age for the formation of the valley is constrained by the deposition of Tilapa red beds over the Early Cretaceous rocks (e.g. Chivillas Formation), which is early-middle Eocene (Dávalos-Álvarez et al. 2007). This is also the cooling age of the apatite fissiontrack system (ca. 42-40 Ma; Figure 9, B and C) in Chivillas rocks, post-dating the Mexican orogen event in the southern Sierra Madre Oriental. Although the apatite fission-track central value of sample LS-2 is ca. 21 Ma ( Figure 9A), younger than the first pulse of extension of the Oaxaca Fault System, it coincides with the timing of the Tehuacán relay ramp in the early Miocene (pulse 4;Dávalos-Álvarez et al. 2007). Therefore, we surmise that the Tehuacán ramp's development was the event responsible for the apatite fission-track cooling age recorded in sample LS-2.
The thermal models herein obtained indicate that Chivillas exhumation took place between ca. 46 and 36 Ma (Figure 10). The fast Lutetian-Priabonian (Eocene) cooling may be related to the first extensional pulses of the Oaxaca Fault System (Figures 10 and 13B). On the other hand, the Tehuacán Valley has an early Eocene-early Oligocene infill of a minimum of 700-1000 m (Centeno-García 1988;Dávalos-Álvarez et al. 2007). The units representing the base of the infill are richer in metamorphic clasts (gneiss, schist, and amphibolite), but these become scarcer in the upper levels of the sedimentary infill (Dávalos-Álvarez 2006). These features suggest that erosion of the Mazateca range contributed to the exhumation of the Chivillas Formation, continuously removing the overburden and supplying sediments to the Tehuacán basin ( Figure 13B).
We propose that erosion of the Mazateca range that exhumed the Chivillas Formation was, therefore, driven by the reactivation of the Oaxaca Fault System as a normal fault ( Figure 13B).

A word on the benefits of integrating zircon ages with apatite geochemistry and thermochronology as a provenance tool
Sedimentary provenance studies have made ample use of U-Pb zircon geochronology by comparing detrital zircon population ages with those of potential source rocks (e.g. Dickinson and Gehrels 2003;Chen et al. 2019). However, this technique provides an incomplete picture of provenance because of differential zircon fertility (Moecher and Samson 2006;Dickinson 2008a) and its refractory character through multiple sedimentary cycles (Fedo et al. 2003;Andersen et al. 2016). The use of other detrital mineral phases as a complement to elucidate provenance has increased over the years to avoid this shortcoming (e.g. Jafarzadeh et al. 2014;Mark et al. 2016;Fairey et al. 2018;O'Sullivan et al. 2018O'Sullivan et al. , 2020Gaschnig 2019;Mandal et al. 2019;Mulder et al. 2019;Chew et al. 2020).
Apatite is a mineral phase with several advantages as provenance tracer: (1) it is more susceptible to chemical and physical weathering and therefore more likely to represent first-cycle sediment (e.g. Morton and Hallsworth 1999;Gillespie et al. 2018); (2) it is a powerful U-Pb and fissiontrack thermochronometer, with closure temperatures between 375-550°C and 60-120°C, respectively, with the ability to produce double-dates on the same crystal (e.g. Cochrane et al. 2014;Kirkland et al. 2018;Danišík 2019); (3) its trace element abundances are sensitive to the chemical environment in which the grains were formed (e.g. Morton and Yaxley 2007;O'Sullivan et al. 2020).
Here, we successfully used apatite as a provenance and thermochronology tool. It was useful to preclude pre-Cretaceous strata in southern Mexico as the source of Precambrian and late Palaeozoic material in the Chivillas Formation. It was also useful to distinguishing the source lithology of detrital apatite, coupling the U-Pb ages and the geochemical fingerprint following the O'Sullivan et al. (2020) methodology. Finally, the fission track work allowed dating the inversion of the basin where the Chivillas Formation was deposited during the Mexican orogen and the reactivation of the Oaxaca Fault System.

Acknowledgments
This work was supported by PAPIIT-DGAPA-UNAM grant IN101520 to L. Solari and PAPIIT-DGAPA-UNAM grant IA105521 to U. Martens. The first author acknowledges the Consejo Nacional de Ciencia y Tecnología (CONACyT) for providing a master's degree scholarship and the Sistema Nacional de Investigadores-CONACyT for a research assistant scholarship for the accomplishment of this work. We thank Juan Tomás Vásquez (Centro de Geociencias, Universidad Nacional Autónoma de México) for staining thin sections; Carlos Ortega-Obregón and Ofelia Pérez Arvizu (Centro de Geociencias, Universidad Nacional Autónoma de México) for providing technical assistance during sample preparation and isotopic dating. The Editorial handling of Dr Robert Stern and the careful reviews of Dr Aaron J. Martin and an anonymous reviewer are acknowledged. Their comments significantly improved the manuscript.

Disclosure statement
No potential conflict of interest was reported by the author(s).