Timing of Magmatism and Skarn Formation at the Limon, Guajes, and Media Luna Gold ± Copper Skarn Deposits at Morelos, Guerrero State, Mexico

The Guerrero gold belt in Guerrero State, southern Mexico (Fig. 1), is an emerging gold province with a total endowment of more than 30 Moz of gold (Belanger, 2012; Neff et al., 2017; Torex Gold Resources Inc., 2021; Davidson et al., 2022). Major gold resources of the Guerrero belt are related to magnetite- or pyrrhotite-rich skarns with significant gold and silver grades as well as, in some cases, economic copper mineralization. Gold ± copper skarns in the Guerrero gold belt are intimately associated with Upper Cretaceous-Paleocene granodiorite stocks that intruded carbonate rocks of the Morelos-Guerrero basin and form a well-defined trend (Fig. 1) that extends more than 35 km in a northwest-southeast direction (Meza-Figueroa et al., 2003; Levresse et al., 2004; Belanger, 2012). From northwest to southeast, the trend includes the Ana Paula, Limon, Guajes, Media Luna, Nukay, Los Filos, Bermejal, and Xochipala skarn deposits and/or prospects (Fig. 1; de la Garza et al., 1996; Jones and Jackson, 1999; Belanger, 2012). Previous studies mainly focused on mineralogical descriptions of the southern Nukay, Los Filos, and Bermejal skarns (de la Garza et al., 1996; Jones and Jackson, 1999) and geochronology of associated intrusions (Meza-Figueroa et al., 2003; Levresse et al., 2004), whereas direct dating of skarn mineralization was previously not conducted. The present study is the first published description of the Limon, Guajes, and Media Luna skarns in the Morelos District as well as the first detailed geochronologic investigation of intrusions and skarns in this important new gold ± copper district.

The Torex Gold Resources Inc.-owned 290-km² Morelos project occupies a central position in the Guerrero gold belt (Fig. 1). To date, three skarn deposits with a total resource of 8.5 Moz Au (production plus resources; as of October 31, 2021, for Media Luna and December 31, 2021, for Limon-Guajes) have been discovered within the Morelos district and are associated with the Upper Cretaceous-Paleocene Limon granodiorite (Belanger, 2012; Davidson et al., 2022), which is exposed in the center of the district (Fig 1). The neighboring Limon and Guajes skarn deposits are situated north of the exposed Limon granodiorite. About 2.29 Moz of gold have been produced since 2016 from open-pit and underground mining of the Limon and Guajes deposits, with remaining resources of 17.6 Mt at 2.84 g/t Au (measured + indicated + inferred) from the open pits and 5.9 Mt at 5.93 g/t Au (measured + indicated + inferred) from the underground mine (Torex Gold Resources Inc., 2021; Davidson et al., 2022). The Media Luna skarn deposit is still under exploration, is situated south of the exposed El Limon granodiorite, and contains resources (measured + indicated + inferred) of 39.4 Mt at 2.77 g/t Au, 30.5 g/t Ag, and 1.08 wt % Cu (Davidson et al., 2022). High copper concentrations in the Media Luna skarn distinguish it from other known, generally copper-poor deposits in the Guerrero gold belt. Detailed geologic and mineralogical descriptions of skarns in the Morelos district are not available, and geochronological data of related plutons are limited to two ages provided by Meza-Figueroa et al. (2003) and Belanger (2012).

The difficulty of directly dating skarn mineralization is a problem for many skarn provinces because the main skarn mineral associations typically do not contain phases suitable for conventional methods of absolute age dating (Meinert et al., 2005; Chiaradia et al., 2014). However, recent advances in U-Pb LA-ICP-MS geochronology of garnet (Seman et al., 2017; Gevedon et al., 2018; Wafforn et al., 2018; Burisch et al., 2019a; Millonig et al., 2020) provide a time- and cost-efficient way to date skarn-related mineralization, since garnet is a ubiquitous skarn mineral, usually occurring early in the paragenetic sequence (Meinert et al., 2005). The precision of U-Pb laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) garnet ages is highly dependent on the composition (U, Pb, and U/Pb) of the analyzed garnet, but also continues to improve with progressive development of the method (Millonig et al., 2020). For example, Reinhardt et al. (2022) reported several U-Pb LA-ICP-MS garnet ages with analytical errors below 1%. Although the lifetime of individual magmatic-hydrothermal systems often prevails for only ~50,000 to several hundred thousand years (Chiaradia et al., 2014; Moncada et al., 2019; Large et al., 2020), further improvement of the analytical errors and favorable compositions of analyzed garnet samples could potentially provide a direct constraint on the duration of individual skarn events.

In order to constrain the magmatic and magmatic-hydrothermal evolution of the El Limon system and skarn formation in the Guerrero gold belt, we applied U-Pb LA-ICP-MS to zircon related to the main granodiorite and to a series of late-stage porphyry dikes, which was complemented by Re-Os geochronology of molybdenite, U-Pb chemical abrasion-thermal ionization mass spectrometry (CA-TIMS) geochronology of titanite, and Ar-Ar geochronology of phlogopite related to hydrothermal vein and skarn mineralization. Moreover, we applied innovative in situ U-Pb LA-ICP-MS geochronology of skarn-related garnet to directly date prograde skarn formation. Analyzed garnet samples represent both contact metamorphic (skarnoid) and metasomatic textures as well as proximal, intermediate, and most distal portions of the Limon and Media Luna skarns to test if systematic age differences in the spatial and temporal evolution of the mineral system can be revealed by U-Pb LA-ICP-MS. As neither intrusive rocks nor skarns of the Morelos district were described previously in the literature, we present detailed geologic observations to provide a sound framework for the subsequent multimethod geochronological study of the El Limon magmatic-hydrothermal system.

The Morelos Project is situated in the Guerrero gold belt (Sierra Madre del Sur) in the State of Guerrero in southern Mexico, about 200 km south-southwest of Mexico City (Fig. 1). This location is at the western edge of the Mixteca terrane (Fig. 1 inset), which is composed of a pre-Carboniferous basement of metamorphic rocks of the Acatlán Complex (Ortega-Gutiérrez, 1981; Centeno-Garcia et al., 2008) unconformably overlain by a thick succession of Cretaceous sedimentary rocks - the Morelos, Cuautla, and Mezcala Formations (Fig. 2), which together form the Morelos-Guerrero basin (Fries, 1960). The Mixteca terrane is overlain to the north by Neogene-Quaternary rocks of the Trans-Mexican volcanic belt. To the west, the Mixteca terrane is separated from the Guerrero Composite terrane (Campa and Coney, 1983) by the Teloloapan thrust, whereas the strike-slip fault contact with the Xolapa metamorphic terrane represents its southern boundary. To the east, the Mixteca terrane is separated from the Acatlán Complex by the NE-SW–trending Papalutla thrust (Centeno-Garcia et al., 2008).

The Cretaceous sedimentary successions of the Morelos-Guerrero basin are the major hosts for skarn mineralization (Figs. 1, 2) in the area and consist of 600- to 1,000-m-thick shallow marine carbonate rocks of the Morelos Formation (Fries, 1960; Aguilera-Franco et al., 2001). The Morelos Formation is mainly composed of calcareous limestones with some dolomite in lower units of Albian to Upper Cenomanian age that formed as a carbonate platform (Fries, 1960). Clastic components are generally very low, and the degree of dolomitization is variable (Aguilera-Franco and Hernández Romano, 2004). The occurrence of dolomite is recognized in the deeper parts of the Morelos Formation, whereas the uppermost sequences of the Morelos Formation are invariably composed of calcareous limestone (Aguilera-Franco et al., 2001). The limestones of the Morelos Formation are bioclastic, peloidal, and intraclastic packstones as well as wackestones, which are rich in inner shelf organisms (Aguilera-Franco and Hernández Romano, 2004). The Morelos Formation is conformably overlain by the Cuautla Formation (Late Cenomanian-Coniacian), which varies from several 10s to 100s of meters in thickness (Aguilera-Franco and Hernández Romano, 2004). The base of the Cuautla Formation is characterized by similar limestones as the top of the Morelos Formation but contains a different biota (Aguilera-Franco et al., 2001). The upper parts of the Cuautla Formation are composed of thin successions of laminated bioclastic and intraclastic limestones that gradually transition upward to pelagic limestones and siliciclastic rocks of the latest Cenomanian to probably Maastrichtian Mezcala Formation. The base of the Mezcala Formation is dominated by gray argillaceous limestones that gradually transition to shales, siltstones and sandstones, and conglomerates at the very top (Fries, 1960; Aguilera-Franco and Hernández Romano, 2004).

The change in the depositional environment from shallow marine carbonate platform (Morelos Formation) to flysch sedimentation (Mezcala Formation) marks the transition from extension to shortening in the region during this period and was associated with the development of a foreland basin on top of the Cretaceous carbonate platform (Fries, 1960; Aguilera-Franco and Hernández Romano, 2004). This shortening was caused by the final accretion of the Guerrero composite terrane with the continental margin and emplacement against the Mixteca terrane in the Late Cretaceous (Campa and Coney, 1983). Cretaceous sedimentary rocks were folded into N-S–striking anticlines and synclines as a result of the accretion of the Guerrero composite terrane against the Mixteca terrane (Centeno-Garcia et al., 2008), which occurred synchronously with the Sevier and Laramide orogenies in western North America (Dickinson et al., 1988; Bird, 1998). Although these events occurred at the same time, it is unclear if the same geodynamic forces that drove the Laramide orogeny in North America were also responsible for orogenesis at the end of the Cretaceous in southern Mexico (Meza-Figueroa et al., 2003; Centeno-Garcia et al., 2008). Nevertheless, subduction of the Farallon Plate beneath the North American craton caused magmatism that resulted in the formation of Late Cretaceous-Paleocene diorites and granodiorites that intruded the folded and thrusted sedimentary units of the Morelos-Guerrero basin within the Mixteca Terrane (Cerca et al., 2007). These intrusions form the linear northwest-trending Guerrero gold belt, which is approximately parallel to the inferred former subduction zone (Cerca et al., 2007). In addition to the intrusive and volcanic rocks related to the Guerrero gold belt in Guerrero and Mexico States, extensive batholiths and minor volcanic units of similar age occur farther to the west in Jalisco, Colima, and Michoacán and delineate another zone of activity related to the same regional tectonic event (Morán-Zenteno et al., 2007).

Late Cretaceous to Paleocene intrusive rocks in the Guerrero gold belt are moderate- to high-K calc-alkaline series plutons that contain 56 to 71 wt % SiO₂ (de la Garza et al., 1996; González-Partida et al., 2003a; Meza-Figueroa et al., 2003). Available geochronological data for intrusive rocks in the Guerrero gold belt are mainly restricted to ⁴⁰Ar/³⁹Ar biotite, U-Pb LA-ICP-MS zircon, and U-Pb ion probe zircon analyses of intrusive rock types at Los Filos, Bermejal, and Nukay. However, they also include two ages of Morelos plutons; together ranging from ~67 to 63 Ma (Jones and Jackson, 1999; Meza-Figueroa et al., 2003; Levresse et al., 2004; Belanger, 2012).

Folded and thrusted Cretaceous sedimentary rocks of the Morelos-Guerrero basin are overlain by clastic sedimentary rocks and felsic volcanic rocks. The felsic volcanic rocks consist of andesite to rhyolite lavas that belong to the Tetelcingo Formation, having reported K-Ar ages of 68.8 ± 2.4 and 66 ± 2.3 Ma, respectively (Ortega-Gutiérrez, 1980). This relates these volcanic rocks to the same magmatic event as the fertile intrusions of the Guerrero gold belt.

Cretaceous units are partly covered by fluvial and alluvial sedimentary rocks of the Balsas Group. From the Eocene to middle Miocene, there was a general migration of volcanism from the Michoacán region to southeastern Oaxaca (Morán-Zenteno et al., 2007). Late Eocene to Miocene volcanic activity is widespread in the Morelos, Guerrero, and western Oaxaca regions and represents one of the major accumulations of volcanic rocks in southwestern Mexico. However, only a small cluster of intrusive stocks of Eocene-Oligocene age have been recognized in the middle of the Guerrero gold belt, where the Morelos district is located (Meza-Figueroa et al., 2003).

Mining in the Guerrero gold belt (also known as the Mezcala district) of southern Mexico dates back to the colonial period (Diaz-Salgado, 2000). The first recorded production is from small but relatively high-grade oxidized skarn and iron oxide bodies along the contact between granodioritic intrusions and limestone (de la Garza et al., 1996; González-Partida et al., 2003b). Mining activity was restricted to artisanal and small-scale mining until the discovery of the large but low-grade Bermejal deposit in 1989 (de la Garza et al., 1996) and the Los Filos gold deposit in 1995 (Belanger, 2012). Since then, more than 30 Moz of gold resources have been discovered in six separate mineralizing centers (Fig. 1) in a belt stretching more than 35 km (Belanger, 2012; Neff et al., 2017; Torex Gold Resources Inc., 2021).

Skarns affiliated with Late Cretaceous-Paleocene intrusive stocks are the major host for the gold mineralization in the Guerrero gold belt (Figs. 1–3). Skarn alteration in the belt occurs as well-developed endoskarn (skarn replacing igneous rock types) and exoskarn (skarn replacing sedimentary rock types) along the contacts between Late Cretaceous-Paleocene intrusive stocks and their sedimentary host rocks (Fig. 4) of the Morelos-Guerrero basin (Belanger, 2012). Prograde skarn mineralogy is characterized by garnet, pyroxene, and phlogopite, which are locally overprinted by retrograde magnetite, epidote, amphibole, chlorite, sulfides, hematite, and quartz. Economic gold mineralization may be spatially associated with all mineralogical varieties including endoskarn and exoskarn, as well as with prograde and retrograde mineral assemblages (de la Garza et al., 1996; Jones and Jackson, 1999; Belanger, 2012). A distinct concentration of gold with late-stage quartz and hematite alteration is reported for some Guerrero gold belt skarn deposits (Levresse et al., 2004; Belanger, 2012), which is, however, not the case at Morelos. Detailed descriptions of the skarn alteration and associated mineralization are limited to the southern Bermejal, Los Filos, and Nukay deposits (de la Garza et al., 1996; Levresse et al., 2004; Belanger, 2012), which have a gold endowment (production + reserves + resources) of more than 20 Moz (Belanger, 2012). The inactive Xochipala mine is a small occurrence at the south end of the belt that had 0.36 Moz of gold production (Lunceford, 2010). The Limon and Guajes deposits have been in production since 2016 and, together with the 2012-discovered Media Luna deposit (Suchomel et al., 2013), are located within the Morelos project (Fig. 3) with a gold endowment of 8.5 Moz (production from Torex Gold Resources Inc. [2021]; measured + indicated + inferred resources from Davidson et al. [2022]). The Ana Paula project at the north end of the belt is reported to contain 1.5 Moz of gold in an irregular subvertical breccia body (Neff et al., 2017).

At least 900 m of the Morelos Formation are exposed in the canyon of the Balsas River in the east-central part of the project area (Figs. 2, 3). The thin-bedded dark gray limestones of the Cuautla Formation discontinuously occur around the Morelos project and are a maximum of 30 m thick. At least 2 km of the Mezcala Formation occur just west of Media Luna (Fig. 3). The thicknesses of the Morelos, Cuautla, and Mezcala Formations were affected by folding and faulting; hence, the exposed thicknesses are commonly not representative of the true thicknesses. Additionally, the presence of abundant Paleocene sills in the Mezcala Formation may have inflated the section in the Media Luna area significantly. None of the younger sedimentary sequences, such as the Tetelcingo Formation or Balsas Group, were mapped in the Morelos district, but both are preserved in the Early Tertiary Upper Rio Balsas basin located 15 km east of Media Luna (Cerca et al., 2007). Conglomerates that are thought to be part of the Balsas Formation occur 2 km south of the Nukay and Los Filos deposits south of Media Luna. Quaternary sands and gravels occur along the banks of the Balsas River. Colluvium and locally thick caliche soils were mapped across the Morelos project (Fig. 3).

Regional structure at Morelos is characterized by N-S–trending, E-verging folds and low-angle thrusts associated with strike-slip faults (Figs. 3, 4). Thrusting occurred in the late Cretaceous during accretion of the Guerrero composite terrane to the western margin of North America (Campa and Coney, 1983). The Teloloapan subterrane, the easternmost part of the Guerrero composite terrane, is exposed on the western margin of the property (Fig. 1). It consists of low-grade metamorphic volcano-sedimentary rocks that were thrusted on top of the Mixteca terrane along the low-angle, north-south–trending, eastward-verging Teloloapan thrust (Salinas-Prieto et al., 2000). A secondary, subparallel thrust runs just east of the Morelos project boundary from Xochipala to the north, putting limestones of the Morelos Formation on top of flysch sediments of the Mezcala Formation.

Several granodioritic stocks are exposed at surface within the Morelos district (Fig. 3). The Limon granodiorite stock is the largest in the district and is situated in the center of the property. Intrusion of the Limon granodiorite stock created a contact metamorphic halo of mainly hornfels in the Mezcala Formation and marble in the Morelos Formation (Figs. 2, 3). This contact metamorphic halo typically extends 400 to 800 m from the granodiorite contact and may locally extend up to 1.5 km away from the igneous contact. The Mezcala Formation was metamorphosed to biotite-bearing hornfels, which were subsequently and more proximally partly replaced by pyroxene hornfels. Bedding and other sedimentary structures are well preserved in the fine-grained hornfels. The Morelos Formation was metamorphosed to marble, which around and above the Limon granodiorite exhibits a distinct foliation parallel to the igneous-sedimentary contact. Locally, well-developed boudins of chert layers, skarnoid (layered sedimentary or metamorphic texture preserved) garnet nodules, ptygmatic calcite veins, and deformed fossils can be recognized in the marble. This ductile deformation is most likely related to the emplacement of the El Limon granodiorite. Two orientations of almost perpendicular normal faults, which strike SW-NE and SE-NW, occur at the periphery of the mapped extent of the Limon granodiorite (Figs. 1, 3, 4).

Rock classification is based on whole-rock geochemistry (App. Fig. A1) and the R1R2 plutonic rock classification of de la Roche et al. (1980). Based on crosscutting relationships, the oldest rock type of the composite El Limon magmatic system is the large, mostly equigranular El Limon granodiorite that is exposed over an area of about 15 km² (Fig. 3). The El Limon granodiorite also includes several sills of equigranular granodiorite to porphyritic feldspar tonalite composition that occur in both the Morelos and Mezcala Formations in a wide area surrounding the outcropping portion of the main El Limon granodiorite stock. The outermost zones of the El Limon stock are locally composed of porphyritic feldspar tonalite with identical texture and composition to some of the previously mentioned sills. On the north side of the intrusion, in the vicinity of the Limon and Guajes deposits, the geometry of the El Limon stock is characterized by multiple sill-like intrusion fingers. The Limon deposit is situated above such a >100-m-thick sill-like intrusion (Fig. 4A), whereas the Guajes deposit is situated along the NW margin of the Limon granodiorite, where several (up to 30-m-thick) sills emanate from the Limon granodiorite into the sedimentary host rock (Fig. 4A). To the south, the Limon granodiorite has the shape of two domes along a NW trend with a small saddle in between. One dome is centered on the Naranjo exploration area (Fig. 3) and the other on the Media Luna area (Figs. 3, 4B).

Nomenclature of intrusive rock types is based on the mineralogy of phenocrysts (if present) and bulk rock composition (Fig. 5). The Limon granodiorite is equigranular, phaneritic, and consists mainly of plagioclase, biotite, hornblende, K-feldspar, and quartz (Figs. 5A, B). Accessory phases include apatite, titanite, and zircon. The rock contains 2 to 3 vol % hornblende phenocrysts that exhibit a preferred orientation resulting in a weak magmatic foliation. Tonalite porphyry sills with feldspar phenocrysts (Fig. 5C, D) have the same overall mineral assemblage as the granodiorite (Fig. 5A, B), although containing more biotite and less quartz. Both the Limon granodiorite and feldspar tonalite porphyry intrusions contain rounded to irregularly shaped mafic enclaves that range in size from several centimeters to meters in diameter. The enclaves are fine grained with abundant biotite, hornblende, and plagioclase phenocrysts and have a monzodiorite to gabbro-diorite composition. Gneiss xenoliths of the Acatlán metamorphic basement are rarely observed in the El Limon granodiorite. Where the Limon granodiorite and feldspar tonalite porphyry are in contact with limestone, they are usually altered to endoskarn.

Dikes and sills of feldspar-biotite-hornblende-quartz granodiorite porphyry cut both the Limon granodiorite and the feldspar tonalite porphyry sills. The porphyritic feldspar-biotite-hornblende-quartz granodiorite rock types were mapped in two areas: as a swarm of NE-trending subvertical dikes at the Limon deposit and as a series of approximately radial dikes with subvertical dips centered just NW of the Media Luna deposit (Fig. 3). The mineralogy of feldspar-biotite-hornblende-quartz dikes is nearly identical to the Limon granodiorite but with a distinct crowded porphyritic texture reflecting a high abundance of phenocrysts (Fig. 5E, F). Quartz phenocrysts are often rounded. Various feldspar-biotite-hornblende-quartz porphyritic intrusions may vary slightly in terms of mineralogy and texture, but all seem to predate skarn alteration. In the southern part (Media Luna) of the El Limon magmatic system, feldspar-biotite-hornblende-quartz granodioritic rock types vary from almost equigranular to porphyry texture containing 1 to 2 vol % potassium feldspar megacrysts, locally up to 5 cm in length. In the northern part of the Limon deposit, the dikes lack K-feldspar megacrysts but may locally show mafic enclaves with a composition similar to those recognized in the Limon granodiorite. A >600-m-wide body of feldspar-biotite-hornblende-quartz granodiorite porphyry is shown in the very north of the geologic map (Fig. 3) but was not mapped in detail. Reconnaissance mapping indicates this unit also has multiple intrusive phases. Although its overall composition and texture are similar to the feldspar-biotite-hornblende-quartz porphyry dikes, its paragenetic relationship remains unclear. Feldspar-biotite-hornblende-quartz dikes can be affected by endoskarn alteration where they are in contact with limestone.

Less abundant but widely distributed across the Morelos property are feldspar-biotite-hornblende monzodiorite to tonalite porphyritic dikes (Fig. 5G, H). Feldspar-biotite-hornblende dikes contain plagioclase, biotite, and hornblende, which constitute both the groundmass and the phenocrysts (Fig. 5G, H). Quartz is rare, and accessory phases include apatite, titanite, and zircon. Some feldspar-biotite-hornblende dikes in contact with limestone are fairly unaltered, whereas others are altered to skarn. Crosscutting relationships and the variable presence of skarn alteration suggest that feldspar-biotite-hornblende dikes were emplaced both prior to and after the main magmatic-hydrothermal stage.

Quartz-feldspar-hornblende granite porphyry dikes cut all the previously mentioned intrusions; however, crosscutting relationships between feldspar-biotite-hornblende monzodiorite-tonalite and quartz-feldspar-hornblende granite porphyry dikes could not be observed in the field, and the relative timing of the two is hence not directly constrained. Quartz-feldspar-hornblende granite porphyry dikes consist of phenocrysts of plagioclase, K-feldspar, quartz, biotite, and hornblende in a groundmass of the same minerals (Fig. 5I, J). Accessory minerals include apatite, titanite, and zircon. Quartz phenocrysts range from rounded to euhedral (beta quartz) in shape. Quartz-feldspar-hornblende granite porphyry dikes are most common at the southern part of the El Limon granodiorite and in the area of the Media Luna deposit (Fig. 3). Only one individual N-S–trending quartz-feldspar-hornblende granite porphyry dike cuts through the Limon deposit (Fig. 3), indicating that it postdates skarn alteration. The dikes generally exhibit pervasive but very weak alteration with illite, chlorite, or epidote but do not exhibit typical endoskarn alteration. Quartz-feldspar-hornblende granite porphyry dikes, as well as all previously mentioned intrusions, may be affected by late argillic alteration with local replacement by kaolinite, pyrite, and calcite.

Other intrusions present at Morelos, but not specifically studied in this contribution, include dikes and sills with phenocrysts of feldspar, hornblende, and biotite in a groundmass of plagioclase and fine-grained biotite. They appear fairly similar to feldspar-biotite-hornblende dikes but typically are less than 5 cm in thickness, are finer grained, and have an andesite-trachyte composition. Accessory minerals include zircon with abundant apatite and titanite. These intrusions were affected by early hornfels and endoskarn alteration and thus were emplaced prior to skarn alteration, although clear crosscutting relationships are lacking. Unaltered basalt and andesite dikes are the youngest intrusions in the district and postdate the Paleocene magmatic event.

The earliest hydrothermal mineralization predates skarn alteration and consists of widely-spaced, irregularly oriented molybdenite-quartz veins that are mainly recognized within the Limon granodiorite and porphyritic feldspar tonalite sills and dikes (Fig. 6). These veins may extend several meters into the hornfels, whereas they usually pinch out within the intrusive rocks where the intrusion is in contact with marble. The veins may have laminated potassium feldspar along their margins accompanied by patches of molybdenite and minor chalcopyrite. Therefore, they share many similarities with B-veins in a porphyry copper system (Sillitoe, 2010). Molybdenite-quartz veins are commonly overprinted by a prograde pyroxene and/or garnet alteration assemblage.

There are two separate but related gold ± copper skarn deposits in the Morelos District: the Limon-Guajes deposit at the northern margin and the Media Luna deposit at the southern margin of the El Limon granodiorite stock (Figs. 3, 4). The Limon-Guajes deposit formed in the fairly pure limestone (typically less than 1 wt % Mg) of the Cuautla Formation near the stratigraphic contact with the Mezcala Formation (Figs. 2, 3). Conversely, the Media Luna skarn formed deeper in the stratigraphic section, in the Morelos Formation where the rocks tend to be more dolomitic (up to 11.4 wt % Mg). At both deposits, skarn replaces igneous (endoskarn; Fig. 7A-C) as well as carbonate protoliths (exoskarn; Fig. 7D-F). Limon-Guajes endoskarn and exoskarn combined form a 0.1 to 0.6- × ~2.5-km irregularly shaped lens along the igneous-sedimentary contact that extends 15 to 80 m into the sedimentary rocks (Figs. 3, 4). Skarn at the Media Luna deposit forms a 1.5- × 0.4- to 1.5-km lens with a thickness of 20 to 100 m along the SW-dipping igneous-sedimentary contact.

The contact between endoskarn and exoskarn is often difficult to define, and the different igneous protoliths of endoskarn are difficult or impossible to distinguish due to intense metasomatism. The prograde skarn stage at Limon-Guajes and Media Luna is mainly characterized by garnet and pyroxene, with a systematic decrease of garnet/pyroxene ratio from proximal to distal relative to the igneous-sedimentary contact, a feature common to most skarn systems (Meinert, 1997). Moreover, there is a zonation from proximal red-brown garnet (Fig. 7B, D) to intermediate brown (-green) garnet (Fig. 7E), to distal green garnet (Fig. 7F) in both deposits. In addition, there is very distal garnet in skarnoid layers (Fig. 7G) and nodules (Fig. 7H) in marble beyond the main skarn boundary. Semiquantitative analyses (scanning electron microscopy-energy dispersive X-ray [SEM-EDX]) indicate grossular to andradite garnet compositions with textures ranging from lobate poikilitic shapes to complexly zoned euhedral crystals (Fig. 7E, F). Pyroxene varies from early, proximal diopside to later, distal hedenbergite-rich varieties (Fig. 7A-C, I, J).

Retrograde skarn minerals include scapolite, amphibole, and chlorite (Fig. 7I) accompanied by locally abundant early magnetite (Fig. 7K, L) and later sulfides, tellurides, and native metals (Fig. 7C, D, I-K, M). Magnetite postdates prograde pyroxene, garnet, and phlogopite (Fig. 8A. B) but predates retrograde amphibole. Late-stage opaque minerals postdate or are coeval with retrograde amphibole and extensively replace prograde skarn minerals (Fig. 8C-F). Late-stage opaque minerals include hedleyite (Bi₇Te₃), tetradymite (Bi₂Te₂S), cosalite (Pb₂Bi₂S₅), bismuthinite (Bi₂S₃) pyrrhotite, cobaltite (CoAsS), chalcopyrite, molybdenite, sphalerite, arsenopyrite, native gold, native bismuth, and electrum (Fig. 8).

Retrograde alteration of earlier skarn minerals continued with formation of serpentine, talc, and chlorite along with pyrite. Calcite occurs in all retrograde assemblages. Late-stage veins and breccias of kaolinite-pyrite-calcite ± hematite can be recognized in crosscutting intrusions, skarn, and marble.

Outside the limit of calc-silicate alteration, fluid escape structures (Fig. 7N) can be recognized that record the passage of hydrothermal fluids beyond the T-P-X limits of calcsilicate minerals. Furthermore, lead-zinc-silver mantos and vein mineralization are recognized, up to 1.5 km away from the Limon granodiorite contact, that are composed of galena, iron-poor sphalerite, pyrrhotite, pyrite, stibnite, fluorite, and calcite. Based on their mineralogy (Meinert, 2000; Bonsall et al., 2011; Fontboté et al., 2017; Burisch et al., 2019b), they are considered to be a distal manifestation of the magmatic-hydrothermal El Limon system or other, as yet undiscovered buried intrusions.

Although both the Limon-Guajes and Media Luna deposits have similar overall mineralogical characteristics, there are also distinct differences. The Media Luna deposit is characterized by a significantly higher abundance of magnetite and chalcopyrite (Figs. 7, 8), whereas pyrrhotite is less abundant compared to the Guajes-Limon skarn. Nevertheless, both Limon-Guajes and Media Luna are strong magnetic highs due to the abundance of monoclinic pyrrhotite and magnetite, respectively (Fig. 9). Furthermore, the Media Luna deposit is characterized by a more magnesian mineralogy including forsterite, chondrodite ((Mg, Fe²⁺)₅(SiO₄)₂ (F, OH)₂), humite ((Mg, Fe₂²⁺)₇(SiO₄)₃(F, OH)₂), clinohumite ((Mg, Fe²⁺)₉(SiO₄)₄(F, OH)₂), ludwigite (Mg₂Fe³⁺(BO₃)O₂), periclase, tremolite, phlogopite, and serpentine (Fig. 7L, M).

Another observation specific to the Media Luna deposit is textures associated with some high-grade mineralization in the form of finely bedded (geopetal) sediments and chaotic blocks that were replaced or rimmed by skarn minerals. Although they look like ordinary breccias, they could be the result of replacement of primary or secondary sedimentary geopetal, doline, and karst structures developed in the near-surface environment. Similar karst features were documented for other skarn deposits on the Guerrero gold belt (Belanger, 2012).

Furthermore, several skarn occurrences were mapped on surface along the contact of the El Limon granodiorite and the sedimentary host rock. The largest of those exposures is a V-shaped skarn that is located at the northeastern margin of the Naranjo and Media Luna areas and cut by the valley of the Balsas River (Fig. 3). The N-S–trending leg of the skarn body is unmineralized, and the SW-NE–trending eastern leg of the skarn (Todos Santos) contains copper and minor gold, which has been explored for in the past.

Sample preparation and analytical work for zircon U-Pb LA-ICP-MS geochronology were conducted at the Pacific Centre for Isotopic and Geochemical Research (PCIGR) at the Department of Earth and Ocean Sciences, University of British Columbia. Two samples of each of the five principal intrusions (Limon granodiorite, feldspar tonalite porphyry, feldspar-biotite-hornblende-quartz granodiorite porphyry, feldspar-biotite-hornblende tonalite-monzodiorite porphyry, and quartz-feldspar-hornblende porphyry) at Morelos were analyzed (Table 1). One sample per lithology was selected from the northern margin of the El Limon granodiorite, near the Limon-Guajes deposit (LG), whereas the second stems from the southern margin, near the Media Luna deposit (ML; Figs. 3, 4).

Zircons were analyzed following the procedure of Tafti et al. (2009) using a New Wave UP-213 laser ablation system and a ThermoFinnigan Element2 single collector, double-focusing, magnetic sector ICP-MS. Zircons with >50 µm in diameter were handpicked and prepared as epoxy bedded grain mounts with a high-quality polish. The surface of the mounts was washed for 10 minutes with dilute nitric acid and rinsed in ultraclean water prior to analysis. The highest quality portions of each grain, free of alteration, inclusions, or possible inherited cores, were selected for analysis.

A laser power level of 40% and a spot diameter of 30 µm were used to ablate line scans rather than spot analyses in order to minimize elemental fractionation during analyses. Backgrounds were measured with closed laser shutter for ten seconds, followed by 35 seconds of ablation for data collection. The time-integrated signals were processed with the GLITTER software (Griffin, 2008). Corrections for mass and elemental fractionation were made based on regular analyses of the Plešovice zircon standard (Sláma et al., 2008) that were analyzed between Morelos zircon samples. The Temora 2 reference zircon was analyzed as an unknown in order to monitor the reproducibility of the age determinations on a run-to-run basis. Final interpretation and plotting of the analytical results employed the ISOPLOT software of Ludwig (2007).

Six samples of paragenetically early quartz-molybdenite veins were selected from drill holes across the Morelos project. One sample is from the east part of the El Limon deposit (TMP-1040-77), two samples are from Guajes (DLIM-338-171 and DLIM-516-79A), one sample is from a reconnaissance drill hole on the west side of the project (TMP-1064-369), and two samples are from the Media Luna area (ML-77-485 and ML-175-452) on the south side of the project (Figs. 3, 6). Analyzed samples are all located within the Limon granodiorite and hornfels (Table 2), with some overprinted by skarn alteration. Regardless of high-temperature overprinting and alteration, the molybdenite Re-Os chronometer is known to remain intact through high-grade contact metamorphism (Bingen and Stein, 2003; Stein, 2014); Morelos molybdenite shows no visible evidence for alteration (Fig. 6).

Rhenium-osmium dating of molybdenite has grown into a well-established method to date ore deposits (Stein et al., 1997, 2001). In this study, mineral separates were acquired for six molybdenite-bearing samples from the six above noted drill core intersections. Separates were made using a small handheld diamond-tipped slow speed drill, targeting selected molybdenite occurrences in each sample. The resulting molybdenite powders were transferred to small vials, where an amount between 10 and 106 mg was subsequently used for the Re-Os analysis. Each molybdenite powder was placed in a Carius tube together with inverse aqua regia and a known amount of mixed double Os spike (¹⁸⁵Re-¹⁸⁸Os-¹⁹⁰Os). Sample dissolution and sample-spike equilibration were acquired after heating at 250°C for 12 hours. As these Re-Os data were acquired in 2013, at that time osmium was separated from the solution by solvent extraction into chloroform followed by back-extraction into hydrobromic acid (HBr). Subsequent microdistillation provides a high-purity Os separation ready for mass spectrometry. Rhenium is extracted from the remaining inverse aqua regia solution by anion exchange chromatography and ready for mass spectrometry without further processing. Rhenium and Os isotope ratios are measured on Triton thermal ionization mass spectrometers (TIMS) after being loaded onto outgassed platinum filaments. Barium activators enhance ion yield, resulting in Re intensities allowing simultaneous Faraday cup detection, whereas osmium is measured by peak-hopping using a secondary electron multiplier or by simultaneous Faraday cup collection, depending on signal strength. Rhenium and Os are measured by NTIMS (negative thermal ion mass spectrometry) on instruments at the AIRIE Program at Colorado State University (USA).

All data are blank corrected, with blanks at the time of analysis of these samples (2013) being Re = 3.184 ± 0.039 pg and Os = 0.174 ± 0.012 pg with a ¹⁸⁷Os/¹⁸⁸Os composition of 0.313 ± 0.017. All reported data are at the 2-sigma level of uncertainty. The uncertainties reported for ages include the ¹⁸⁷Re decay constant uncertainty.

Rhenium concentrations are maximums, as all samples had some silicate dilution on drilling the separates. In one sample (MD-1411), the lack of visibly crystalline molybdenite resulted in drilling a bluish strand in a quartz vein and a notably lower Re concentration, as significant quartz was part of the mineral separate. Minor and age-insignificant common osmium was measured in some samples (App. Table A1).

Sixteen samples with abundant garnet from the El Limon (eight samples) and Media Luna (eight samples) skarns were selected for geochronology from six drill cores (Table 2). These samples cover the entire range from proximal, intermediate, and distal portions of the skarns at both the El Limon and Media Luna deposits and include metasomatic garnet related to endo- and exoskarn as well as most distal skarnoid garnet (Table 2; App. Table A2; App. Fig. A2). Samples selected for garnet U-Pb LA-ICP-MS geochronology exhibit a variable degree of retrograde overprint. Hence, garnet may be accompanied by amphibole, magnetite, and sulfides. Areas that were not affected by retrograde overprint were selected for U-Pb analyses.

In total, 16 skarn samples (only 11 were dated successfully) with abundant garnet were cut to size and mounted in 1-inch epoxy rounds. Uranium-Pb data were subsequently collected on a ThermoScientific ElementXrsector field ICP-MS coupled with a RESOlution (COMpex Pro 102) 193-nm ArF Excimer laser equipped with a two-volume (Laurin Technic S155) ablation cell at the Frankfurt Isotope and Element Research Center (FIERCE) of Goethe-Universität Frankfurt (Germany) following the methodology of Burisch et al. (2019a) and Reinhardt et al. (2022). Measurement points for each polished section derive from a small area (<1 cm²) and were set after careful screening to identify growth zones with higher ²³⁸U/²⁰⁶Pb ratios before each analytical session and to avoid superficial exposure of mineral inclusions by monitoring trace element ratios during screening. Samples were then analyzed in situ in fully automated mode overnight. During the measurements, the signals of ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³²Th, and ²³⁸U were acquired by peak jumping in pulse counting mode with a total integration time of ~0.1 s, resulting in 400 mass scans. The raw data were corrected offline deploying an inhouse Microsoft Excel spreadsheet program (Gerdes and Zeh, 2006, 2009). Mali grandite (dated by TIMS at 202.0 ± 1.2 Ma; Seman et al. [2017]) was used as a primary reference material (RM); garnet RM Lake Jaco (Seman et al., 2017) and Mal-iGUF (in-house reference material) were employed as secondary quality control RMs. Common Pb correction has not been applied to the data due to its high variability, although common Pb content was determined using the ²⁰⁸Pb signal after subtracting the radiogenic ²⁰⁸Pb (Millonig et al., 2012). Garnet U-Pb ages are calculated using linear regression in Tera-Wasserburg concordia diagrams (Tera and Wasserburg, 1972) since the andradite-grossular solid solution incorporates a mixture of nonradiogenic and radiogenic Pb formed due to in situ decay of uranium. Garnet U-Pb dates are defined as the lower intercept with the concordia curve as determined by linear regression of discordant arrays. Reported ages are based on multiple concordant analyses from one or more individual garnet grains and were calculated using Isoplot 3.71 (Ludwig, 2008). All uncertainties are reported at the 2σ level.

One sample (WML-31-545m; Table 2; App. Fig. A3) with abundant titanite was collected from drill core WML-31 at 545 m, within the west side of the Media Luna deposit. Titanite is accompanied by quartz, calcite, and amphibole that overprint massive pyroxene skarn.

Sample preparation, geochemical separation, and mass spectrometry were done at the Pacific Centre for Isotopic and Geochemical Research in the Department of Earth and Ocean Sciences, University of British Columbia. Minerals were separated from samples using conventional crushing, grinding, and Wilfley table techniques, followed by final concentration using heavy liquids and magnetic separations. Mineral fractions for analysis were selected according to grain quality, size, magnetic susceptibility, and morphology. Grains were transferred into 300-µL PFA microcapsules (crucibles), and 200 µL subboiled 29N HF and 14N HNO₃ in a 10:1 mixture was added. Each fraction was spiked with a ^233-235U-²⁰⁵Pb tracer solution (EARTHTIME ET535), capped, and placed in a 125-mL Teflon liner (8–13 microcapsules per liner), which in turn went into a Parr-style stainless steel high-pressure dissolution device for 40 hours at 220°C. Sample solutions were then evaporated to dryness at 130°C, and residues were redissolved in 200 µL of subboiled 6.2 M HCl in high-pressure devices for 12 hours at 180°C. These solutions were evaporated to dryness and were redissolved in 3.1 N HCl in a high-pressure device for 12 hours at 210°C. Purification of Pb and U employed ion exchange column techniques modified slightly from those described by Parrish (1987). Lead was eluted into PFA beakers and U into a second set of beakers and further purified by passing through columns a second time. U was then eluted into beakers containing Pb. Elutants were dried in 7-mL screw top PFA beakers on a hotplate at 120°C in the presence of 2 µL of ultrapure 0.2 N phosphoric acid (H₃PO₄). Samples were subsequently loaded onto degassed, zone-refined Re filaments in 2 µL of silicic acid emitter (Gerstenberger and Haase, 1997). Isotopic ratios were measured using a modified single collector VG-54R TIMS equipped with an analogue Daly photomultiplier. Measurements were done in peak-switching mode on the Daly detector. Analytical blanks during the course of this study were 0.2 pg for U and 4.4 pg for Pb. Fractionation was determined directly on individual runs using the ^233-235U tracer, and Pb isotope ratios were corrected for fractionation of 0.30%/amu, based on replicate analyses of the NBS-982 Pb standard and the values recommended by Thirlwall (2000). Data reduction was conducted with the Excel-based program of Schmitz and Schoene (2007). Isochron diagrams and regression intercepts were calculated with Isoplot (Ludwig, 2003). Unless otherwise noted, all errors are quoted at the 2σ or 95% level of confidence. Isotopic dates are calculated using the decay constants λ²³⁸ = 1.55125 × 10^–10 and λ²³⁵ = 9.8485 × 10^–10 (Jaffey et al., 1971).

Samples WZML-30-588 and SS-6-720 are both related to the Media Luna deposit and contain abundant phlogopite (Table 2; App. Fig. A4), feldspar, pyroxene, quartz, and titanite. Sample WZML-30-588 is from a drill hole located along the western edge of the Media Luna deposit. The sample was collected about 4 m below the bottom of a magnetite-sulfide horizon from a patchy network of massive phlogopite (Fig. A4). Sample SS-6-720 is from endoskarn-altered Limon granodiorite encountered in a drill hole located 1 km west of the Media Luna deposit. The sample is from a 3-cm-wide vein of coarse-grained phlogopite cutting pyroxene-altered Limon granodiorite about 25 m below a mineralized exoskarn contact (the term mineralized is used to distinguish sulfide-bearing from sulfide-free skarn).

The samples were broken by hand, and vein material was separated from the host rock. Sample material was washed in distilled water and ethanol and sieved when dry to 0.25- to 0.42-mm fraction. Phlogopite was separated from the rest of the vein material by magnetic separation and tapping on a glass plate. The mineral separates were wrapped in aluminum foil and stacked in an irradiation capsule with similar-aged samples and neutron flux monitors (Fish Canyon Tuff sanidine [FCs], 28.201±0.046 Ma [Kuiper et al., 2008]). The samples were irradiated at the McMaster Nuclear Reactor in Hamilton, Ontario, for 160 MWh in medium flux site 8E. Analyses (n = 85) of 19 neutron flux monitor positions produced errors of <0.5% in the J value.

The samples were then analyzed at the Noble Gas Laboratory, Pacific Centre for Isotopic and Geochemical Research, University of British Columbia. The mineral separates were step-heated at incrementally higher powers in the defocused beam of a 10-W CO₂ laser (New Wave Research MIR10) until fused. The gas evolved from each step was analyzed by a VG5400 mass spectrometer equipped with an ion-counting electron multiplier. All measurements were corrected for total system blank, mass spectrometer sensitivity, mass discrimination, and radioactive decay during and subsequent to irradiation, as well as interfering Ar from atmospheric contamination and the irradiation of Ca, Cl, and K—isotope production ratios are as follows: (⁴⁰Ar/³⁹Ar)K = 0.0302±0.00006, (³⁷Ar/³⁹Ar) Ca = 1416.4 ± 0.5, (³⁶Ar/³⁹Ar)Ca = 0.3952 ± 0.0004, Ca/K = 1.83 ± 0.01(³⁷ArCa/³⁹ArK). Decay constants are from Min et al. (2000); air argon composition is from Lee et al. (2006).

Initial data entry and calculations were carried out using the software ArArCalc (Koppers, 2002). The plateau and correlation ages were calculated using Isoplot version 3.09 (Ludwig, 2003). Errors are quoted at the 2-sigma (95% confidence) level and are propagated from all sources except mass spectrometer sensitivity.

In the following, we present geochronological data of intrusive rock types (Table 1), early-stage molybdenite-quartz veins, and skarn-related garnet, titanite, and phlogopite (Table 2). The detailed background data set and additional documentation are provided in the Appendix (Tables A1-A4; Figs. A2-A4).

For each lithology, two samples of zircon were analyzed by U-Pb LA-ICP-MS: one sample from the northern margin (LG) and one sample from the southern margin (ML) of the El Limon magmatic complex. The age data are presented in the order of their relative age inferred from crosscutting relationships.

Main intrusive rock types: Samples LG1 from the northern margin and ML1 from the southern margin of the El Limon granodiorite (Figs. 3, 4) yield ages of 66.86 ± 0.40 and 65.63 ± 0.69 Ma, respectively (Fig. 10A-B). Two samples of feldspar tonalite porphyry rocks yield ages of 65.88 ± 0.79 Ma (LG2) and 66.32 ± 0.64 Ma (ML2), respectively (Fig. 10C-D), and are therefore, within error, coeval to the emplacement of the El Limon granodiorite. The ages are thus consistent with field observations, supporting that feldspar porphyry sills are, together with the main granodiorite, the earliest intrusions of the system.

Late-stage dikes and sills: Two samples of feldspar-biotite-hornblende-quartz dikes were analyzed, with sample LG4 having an error-weighted mean age of 63.9 ± 1.3 Ma. The single grain ages of sample LG4 vary significantly from 69.6 ± 10.2 to 58.4 ± 3.8 Ma, which indicates that the suite of analyzed zircons is related to several magmatic events, or some grains were affected by postcrystallization isotopic modification (Fig. 10 E). As a consequence, this age has been excluded from the interpretation. Sample ML4 (feldspar-biotite-hornblende-quartz dike) shows individual zircon ages ranging from 67.4 ± 1.4 to 64.0 ± 2.1 Ma and yields an error-weighted mean age of 65.66 ± 0.43 Ma, which is not in conflict with field observations. Samples LG3 and ML3 of feldspar-biotite-hornblende dikes yield error-weighted mean ages of 64.68 ± 0.38 and 65.9 ± 1.1 Ma, respectively (Fig. 10G-H). Samples LG5 and ML5 result in error-weighted mean ages of 64.88 ± 0.75 and 65.29 ± 0.49 Ma, respectively (Fig. 10I-J).

All six samples of molybdenite yield Re-Os ages that lie within a narrow range of 66.63 ± 0.22 to 65.55 ± 0.25 Ma, a span of about 700,000 years (Fig. 11). Rhenium concentrations are moderately high, with lower values attributable to silicate dilution of the mineral separate (App. Table 1A) for several Morelos samples. Molybdenite from sample TMP-1040-77 is related to the Limon granodiorite sill that underlies the eastern part of the Limon deposit (Fig. 4) and has the oldest age at 66.63 ± 0.22 Ma. The two samples from the Guajes deposit (DLIM-338-171 and DLIM-516-79B) are related to molybdenite-quartz veins hosted by the northeastern part of the El Limon granodiorite and yield the same age at the 2-sigma level of uncertainty (66.27 ± 0.23 and 66.24 ± 0.23 Ma, respectively). Sample TMP-1064-369 is from the Naranjo area (Fig. 3) and is a molybdenite-quartz vein hosted by granodiorite, yielding an age of 65.86 ± 0.24 Ma. Samples ML-175-452 and ML-77-485 from the Media Luna area yield the same age (65.58 ± 0.22 and 65.55 ± 0.25 Ma, respectively). Because model ages for the different localities are analytically distinguishable, an isochron approach cannot be employed (Stein, 2014). There are four analytically distinct ages considering only the analytical uncertainty on Re-Os ages (Fig. 11). Spatially, the ages appear to be systematically younger along a counterclockwise trend (From N to W to S) from Limon to Guajes to Naranjo to the Media Luna area (Table 2; Fig. 11).

Garnet U-Pb LA-ICP-MS ages are reported with an internal and an expanded uncertainty (i.e., age ± internal/expanded; Fig. 12; App. Fig A2). Internal uncertainties are calculated taking into account internal standard errors, excess of scatter of the primary reference material, background, counting statistics, and excess of variance from the offset RM (Horstwood et al., 2016). Expanded uncertainties take into consideration long-term variance (1.5%, 2-sigma) and decay constant uncertainties and should be used if reported garnet ages are compared against other data sets and methods. Eleven of the 16 analyzed skarn-related garnet samples were dated successfully by U-Pb LA-ICP-MS (Table 2; Fig. 13; App. Table A2; App. Fig. A2) and yield lower intercept ages in the range of 63.12 ± 0.61/1.13 to 64.63 ± 0.79/1.25 Ma (Table 2). The five unsuccessful samples either result in no age at all or an age with unsatisfactory high uncertainty due to insignificant variation in the 238U/206Pb ratio or a too-high overall common lead.

Five samples from the Limon skarn have lower intercept ages ranging between 63.12 ± 0.61/1.13 and 64.63 ± 0.79/1.25 Ma (Fig. 13). Almost the exact same range of ages is recognized for the six samples from the Media Luna skarn, ranging from 63.2 ± 0.44/1.05 to 64.39 ± 0.57/1.13 Ma (Fig. 13). For both deposits, analyzed sample suites include proximal skarn (red-brown garnet), intermediate skarn (green garnet), and distal skarnoid skarn. Although all eleven garnet ages overlap in their errors, the oldest ages are recognized for garnet related to the most distal skarnoid garnet (beyond the metasomatic skarn front), whereas ages below 64 Ma are restricted to samples from proximal and intermediate endo- and exoskarn.

Five individual titanite grains from sample WML-31-545 (Media Luna deposit) yield similar U-Pb CA-TIMS ages within their analytical error, resulting in a mean age of 65.40 ± 0.74 Ma. Although the titanite age seems to be slightly older than the garnet ages, it overlaps within error with garnet, zircon, and most molybdenite ages. Phlogopite of samples SS-6-720 and WML-30-588 from the Media Luna deposit yields Ar-Ar ages of 65.3 ± 1.9 Ma (inverse isochron) and 65.31 ± 2.05 Ma (integrated age), respectively. These Ar-Ar ages overlap within uncertainties with garnet, zircon, molybdenite, and titanite ages.

Crosscutting relationships of different intrusive stages well constrain the relative timing of the magmatic evolution of the El Limon intrusive system. The evolution starts with the voluminous El Limon granodiorite, which likely consists of several substages. The U-Pb zircon age of the El Limon granodiorite at the north side of the system (Guajes-Limon) is distinctly older than the U-Pb zircon age of the El Limon granodiorite at Media Luna, even accounting for 2-sigma uncertainty. Although most of the U-Pb LA-ICP-MS zircon data are not in conflict with field observations, the large associated errors do not allow a precise determination of the different stages of the magmatic evolution. However, a statistically unsupported trend from older ages associated with the main granodiorite body and the early sills with younger ages associated with the late-stage dikes is recognized (Fig. 13). The oldest high-precision Re-Os age of molybdenite (66.63 ± 0.22 Ma at Limon and 65.58 ± 0.22 Ma at Media Luna) defines the minimum age for the main stage of emplacement (El Limon granodiorite and feldspar tonalite porphyry sills), as field observations indicate that those veins postdate the early magmatic stage. The oldest zircon age of 66.86 ± 0.40 (LG1) thus provides the best estimate for the age of emplacement of the northern portion of the El Limon granodiorite intrusion, as it is not in conflict with Re-Os ages and crosscutting relationships (Limon-Guajes; Fig 13). According to the same argument, the 65.88 ± 0.79 Ma age for the feldspar tonalite sill (LG2) seems to be too young, yet this age is not in conflict with field observations due to its relatively large analytical error. Samples of the Limon granodiorite (ML1), feldspar porphyry (ML2), and feldspar-biotite-hornblende-quartz granodiorite porphyry (ML4) from the southern portion of the magmatic system range in age between 66.63 ± 0.69 and 65.66 ± 0.4 Ma and are hence older than ages obtained for molybdenite veins at Media Luna (65.58 ± 0.22 and 65.55 ± 0.25 Ma), which is consistent with field observations.

Neglecting the analytical uncertainties, there is a weak trend from older zircon ages related to the northern margin (Limon and Guajes area) to slightly younger ages related to the southern margin (Media Luna area) of the intrusive stages of the El Limon magmatic system. As this trend is not statistically significant, this statement has to be regarded with caution; however, the same trend is recognized in the Re-Os molybdenite ages that are invariably younger at the southern Media Luna and Naranjo areas compared to the northern Limon-Guajes area (Fig. 12). Therefore, this trend could be related to protracted magmatic activity during the emplacement of the El Limon composite intrusive stock. The oldest zircon age of 67.3 Ma (66.89 + 0.40 Ma, LG1) provides an estimate for the upper limit, whereas the oldest molybdenite ages at Media Luna provide a rough estimate for the lower limit 65.8 Ma (65.58 + 0.22 Ma, ML-175-452) of the ~1.5-m.y. time span in which the main emplacement of the El Limon granodiorite occurred.

The 63.9 ± 1.3 Ma zircon age of feldspar-biotite-hornblende-quartz granodiorite porphyry dike LG4 is in conflict with crosscutting relationships and is significantly too young. Possible reasons could be due to lead loss or some other unknown isotopic disturbance. The other sample (ML4) of feldspar-biotite-hornblende-quartz porphyry dikes yields 65.66 ± 0.43 Ma, which relates the formation of feldspar-biotite-hornblende-quartz dikes, in agreement with field observations, to the early stage of the emplacement of the El Limon granodiorite.

Based on crosscutting relationships, the youngest Re-Os molybdenite age (65.55 ± 0.25 Ma) indirectly defines the maximum age of feldspar-biotite-hornblende porphyry dikes. Although one U-Pb zircon age of a feldspar-biotite-hornblende dike is older (65.9 ± 1.1 Ma) and the other younger (64.68 ± 0.38 Ma) than Re-Os molybdenite ages, they are not in conflict with field observations due to the large analytical error associated with the U-Pb ages. Although statistically insignificant, this age range is consistent with the observation that feldspar-biotite-hornblende dikes had a protracted episode of emplacement and formed both pre- and post-skarn.

Quartz-feldspar-hornblende porphyry dikes range in age (without uncertainty) from ~65.3 to 64.9 Ma, making them the youngest intrusions, as a group, consistent with crosscutting relationships. Neglecting the errors, quartz-feldspar-hornblende ages are older than one of the feldspar-biotite-hornblende dikes, which would be in conflict with field observations if the errors were not as large. Furthermore, quartz-feldspar-hornblende dikes should be clearly younger than garnet U-Pb LA-ICP-MS ages, since they postdate skarn formation based on crosscutting relationships. Again, the large analytical errors associated with the U-Pb LA-ICP-MS method do not allow resolution of this conflict.

In comparison to other intrusions of the Guerrero gold belt, the estimated timing of the main emplacement stage of the El Limon stock is within error of reported ages of the Ana Paula, Nukay, Los Filos, and Bermejal deposits (Meza-Figueroa et al., 2003; Levresse et al., 2004; Belanger, 2012). Nevertheless, the new geochronological data reported here further support the idea that emplacement of fertile intrusions was synchronous and occurred in a relatively narrow time window from ~67 to 63 Ma in the Guerrero gold belt.

Molybdenite-quartz veins represent the earliest mineralization event at Morelos and constrain the onset of hydrothermal activity to ~67 to 66 Ma at the northern Limon and Guajes deposits and ~66 to 65 Ma in the southern Naranjo and Media Luna deposits. These ages are consistent with zircon ages and crosscutting relationships of granodiorite and feldspar porphyry sills as well as younger dikes. Neglecting two outliers, garnet U-Pb LA-ICP-MS ages of both analyzed skarn deposits cluster in a relatively narrow age interval of ~64.6 to 63.6 Ma (Fig. 13), interpreted as the onset (prograde skarn stage) of the main hydrothermal stage. Garnet ages overlap with the age range of reported feldspar-biotite-hornblende dikes, which supports field observations that skarn and these dikes formed coevally. The garnet ages suggest that skarn formation occurred relatively late in the magmatic-hydrothermal evolution of the system and postdates the main solidification of the El Limon intrusive stock and early-stage hydrothermal veining. This relatively late occurrence of skarns in magmatic-hydrothermal evolution also was documented for other systems, and hence seems to be characteristic (Meinert et al., 2005; Wafforn et al., 2018).

In contrast with Re-Os molybdenite ages, there is no difference in garnet ages for the Limon and Media Luna deposits. Comparing garnet ages from different positions of the skarn (proximal, intermediate, and distal) reveals that there are no statistically robust age differences, either (Table 2; Fig. 13). The relative errors of ~0.5 to 1.5% of reported U-Pb garnet ages are evidently still too high to reveal the time scales of skarn formation, i.e., age differences between proximal and most distal skarnoid garnet, for systems of this age (Late Cretaceous-Paleocene). The Re-Os molybdenite ages with relative errors of ~0.35% are sufficiently precise to capture temporal differences in the timing of the early-stage hydrothermal event and thus provide an estimate of the minimum precision that would be required to resolve the time scales of skarn evolution for the given Late Cretaceous-Paleocene age. Although the precisions of garnet U-Pb ages are not yet sufficient for the studied system, reported relative errors of about 0.5% could be sufficiently precise to capture time scales in very young skarn systems (Chiaradia et al., 2014; Wafforn et al., 2018).

Titanite and phlogopite produced ages (65.2–65.4 Ma) that overlap, within their errors, with the age estimates of both the early-stage molybdenite-quartz veins and the prograde skarn stage (Fig. 13). Due to the large analytical errors and the age overlap, phlogopite and titanite ages do not provide any further constraints on the magmatic-hydrothermal evolution of the system but generally confirm the validity of the garnet U-Pb data.

The two major skarn deposits in the Morelos district, Limon-Guajes and Media Luna, share many similarities but also exhibit distinct differences, both relative to each other as well as to other gold skarns in the Guerrero gold belt. The calcsilicate skarn minerals, consisting dominantly of garnet and pyroxene, are zoned systematically relative to the causative intrusions, and the variations in garnet/pyroxene ratio and garnet and pyroxene color and composition provide an overall exploration vector (bulls eye) that can be used for target selection (Meinert et al., 2005). The common thread through all of the deposits in the Guerrero gold belt is the presence of abundant magnetite and/or pyrrhotite accompanying skarn minerals. Although magnetite and pyrrhotite are useful for geophysical exploration targeting, they are not of economic importance as a separate iron ore. Furthermore, high gold grades often spatially coincide with magnetite-rich zones of the skarn systems at Limon-Guajes, Media Luna, Los Filos, and Nukay (this study; Belanger, 2012), although magnetite and gold are paragenetically not related. The replacement of garnet and pyroxene by magnetite may have, however, resulted in local enhancement of rock permeability, providing preferential fluid migration pathways for the late-stage ore fluid, which precipitated gold and sulfides.

Most skarn deposits of the Guerrero gold belt, except for Media Luna, are gold-only deposits without significant copper (Belanger, 2012) and thus are appropriately categorized as gold skarns (Meinert et al., 2005). The biggest difference between the gold-only skarns such as Limon-Guajes, Nukay, Bermejal, Los Filos, and Media Luna is the composition of the carbonate protolith, specifically the dolomite component. The carbonate MgO content ranges up to 11.5 wt % at Media Luna, and this dolomitic part of the stratigraphic section is also intersected in deep drilling at Limon-Guajes (Drill hole SST-153), yet most of the protolith at Limon-Guajes is fairly pure limestone. Although detailed characterization of the sedimentary host rock is not available for most gold deposits of the Guerrero gold belt, their distinct calcic skarn mineralogy suggests rather magnesium-poor limestone as the protolith (this study; Belanger, 2012). In contrast, Media Luna is characterized by a distinct magnesium-rich skarn mineralogy, which is related to local differences in the composition of the sedimentary protolith. This difference in MgO content has several important implications. First, the magnesian protolith enables the formation of magnesium-rich minerals and may have forced the iron that would have gone into garnet and pyroxene in a calcic skarn system to precipitate as magnetite. In a calcic skarn during the sulfide stage, the excess iron commonly precipitates as pyrrhotite in a reduced system like Limon-Guajes, or pyrite in a more oxidized system. Thus, the magnetic signature of Limon-Guajes is dominated by pyrrhotite, whereas at Media Luna, it is dominated by magnetite. Second, precipitation of magnetite at relatively high temperature (or early in the paragenetic sequence) means that the available sulfur that would have gone into pyrrhotite formation is available to bind copper as chalcopyrite from the ore fluid. This explains why, although the gold contents of Limon-Guajes and Media Luna are similar, Media Luna has significant copper as chalcopyrite, whereas Limon-Guajes and most other deposits of the Guerrero gold belt do not. Dolomite-rich units seem to preferentially occur in the upper part of the Morelos formation (Aguilera-Franco et al., 2001; Aguilera-Franco and Hernández Romano, 2004); hence, the potential to discover new gold ± copper skarns is particularly high where Upper Cretaceous-Paleocene granitoids intruded near the pretectonic pristine sedimentary contact of the Morelos, Cuautla, and Mezcala Formations, as is the case in the Morelos district (Figs. 1, 2).

The Morelos project area, located in the center of the Guerrero gold belt, is the site of several recent discoveries of significant gold ± copper skarn deposits. On the north side of the project area are the Limon and Guajes deposits, where mineralization is mainly hosted in retrograde altered garnet-pyroxene skarn that developed along the contact of the El Limon granodiorite and calcareous sedimentary host rocks. The Limon and Guajes deposits show very similar mineralogical features as well as a rather similar metal gold-(silver-iron) tenor as skarns in other districts of the Guerrero gold belt (de la Garza et al., 1996; González-Partida et al., 2003b; Belanger, 2012) and share many mineralogical similarities with the skarns at Fortitude, Nevada (Myers and Meinert, 1991), and Hedley, British Columbia (Ettlinger et al., 1992). The Media Luna deposit at the southern margin of the El Limon granodiorite is, however, unique for the Guerrero gold belt, as it contains an average copper grade of 1.08 wt % in addition to gold and silver. In contrast to Limon-Guajes and other skarn deposits of the Guerrero gold belt, the distinct Mg-silicate mineralogy of the Media Luna deposit suggests an exceptionally Mg-rich sedimentary protolith. The high Mg concentrations favored early precipitation of magnetite, which potentially resulted in higher sulfur activities during the late skarn stage, in turn promoting precipitation of chalcopyrite. Hence, skarns that replaced magnesium-rich sedimentary protolith are economically favorable exploration targets on the Guerrero gold belt, as they may host significant copper resources in addition to gold and silver.

Rhenium-osmium ages for molybdenites from four localities in the Morelos project area show spatially distinct timings for early-stage mineralization. For two of the four localities, a second molybdenite stage from a different core sample confirms the accuracy of the age and again affirms the high precision and temporal dissection possible with Re-Os dating of molybdenite in relatively young systems.

The U-Pb zircon data of the magmatic units at Morelos suggests that the main emplacement of the El Limon intrusive system occurred at ~67.5 to 66 Ma, which is consistent with other magmatic intrusions related to the Guerrero gold belt. Although U-Pb data support the polyphase evolution of the El Limon system and are consistent with crosscutting relationships, the errors are too large to constrain the absolute timing of consecutive intrusive stages. The main emplacement of the El Limon granodiorite was followed by early-stage hydrothermal molybdenite-quartz veins, which formed slightly earlier in the northern part of the intrusion (~67–66 Ma) relative to the southern part (~66–65 Ma). Innovative U-Pb LA-ICP-MS garnet geochronology was applied successfully to the Limon and Media Luna skarn deposits and indicates skarn formation occurred between ~64 and 65.5 Ma, coeval with the late-stage feldspar-biotite-hornblende dikes.

The validity of garnet ages was confirmed by other radiometric isotope systems (Re-Os on molybdenite, U-Pb on titanite, and Ar-Ar on phlogopite). The reported analytical errors of garnet U-Pb LA-ICP-MS U-Pb are comparable to U-Pb CA-TIMS of titanite and are significantly smaller than errors of the Ar-Ar phlogopite ages reported here.

The ubiquitous presence of garnet in skarn deposits and the time- and cost-efficiency of in situ U-Pb garnet geochronology are striking advantages of this method to constrain the timing of skarn alteration. This method has been recently successfully applied to date skarn mineralization generally (Seman et al., 2017) and to constrain the timing of magmatic-hydrothermal activity on the district and province scale (Burisch et al., 2019a; Reinhardt et al., 2022). However, its limitation for scientific questions currently lies within its precision, as the errors associated are too large to constrain the duration and rates of individual skarn-forming events on the deposit scale (for the given age of the El Limon magmatic-hydrothermal system). Nevertheless, the example of the Morelos district demonstrated that U-Pb LA-ICP-MS dating of garnet is the most suitable method to date skarn deposits, especially in the absence of molybdenite. Hence, we expect that this method will also be of increasing interest to the mineral exploration industry.

We are grateful to Peter Megaw and Panagiotis Voudouris for their constructive reviews that improved a former version of this manuscript. We thank Torex Gold Resources for permission to publish the results of geochronological studies done at the Morelos project over the last four years, but note that views expressed are those of the authors and not necessarily endorsed by Torex. Appreciation is extended to all the geologists who have worked at the Morelos site over the years, including Teck Resources geologists who did early mapping of the entire project. In particular, we thank Torex geologists Olaf Scholtysek, Cristian Puentes, and Juanita Sierra for many lively discussions about the different intrusive rocks at Morelos, geologist Luis Parra for assistance in preparing molybdenite quartz vein samples, and Skapto Consulting geologists for assistance in the field mapping several topographically challenging areas of the project. Shelly M. Faizy is thanked for providing some of the petrographic images. FIERCE is financially supported by the Wilhelm and Else Heraeus Foundation and by the Deutsche Forschungsgemeinschaft. This is FIERCE contribution No. 103.

Mathias Burisch-Hassel is head of the Mineral Systems Analysis Group, a research associate professor at the Colorado School of Mines, and a visiting scientist at the Helmholtz-Institute Freiberg for Resource Technology. From 2016 to 2022, he was an assistant professor at the TU Freiberg after he received his Ph.D. from the University of Tübingen in 2016.

Mathias’ research aims to understand geologic processes that result in the formation, transformation, and preservation of mineral resources associated with hydrothermal systems. Together with his team, he constructs metallogenic models that provide a scientific framework for the mineral exploration industry.

Steven Bussey has been a consulting geologist with Western Mining Services for nearly twenty years, focused on exploration targeting for epithermal, porphyry, and skarn deposits at both regional and prospect scale. Prior to that, he was a geoscientist with Western Mining Corporation for 18 years, working on exploration projects throughout the Americas. His education includes Western Colorado University (B.A.), Southern Methodist University (M.Sc.), and Colorado School of Mines (Ph.D.).