Introduction
Colombia has the most extended history of gold production and exploitation in South America, which dates back to pre-Columbian goldsmithing to the extraction of porphyry and epithermal gold and silver deposits in the present day (Martinón-Torres & Uribe-Villegas, 2017; Scott & Seeley, 1983). Even though Colombia ranks as the 20th largest gold producer globally, with 46.3 tons of gold mined in 2019 (GoldHub, 2020), proven reserves are above 50 Au Moz (Celada et al., 2016). A considerable proportion of such reserves are found in the unexploited and recently discovered La Colosa and Nuevo Chaquiro porphyry deposits, with 881 and 176 Au tons, respectively (Bartos et al., 2017; Celada et al., 2016; Naranjo et al., 2018). These deposits are part of twelve Late Miocene mineral districts located in the Cauca Metallogenic Belt (CMB), a strip comprising numerous Au-Ag-Cu-Mo satellite deposits between the Central and Western Andean Cordillera of Colombia (Fig. 1). The CMB deposits display the typical characteristics of Andean metallogenetic systems, including calc-alkaline magmatism of intermediate composition and high Sr/Y ratios. These attributes make the CMB a natural laboratory for modeling gold endowments in porphyry and epithermal systems associated with arc environments.
Recent studies have used Monte Carlo simulations to model and predict sensitive parameters in the gold endowments associated with Au- Cu porphyry deposits (Chiaradia, 2020; Chiaradia & Caricchi, 2017). The simulation works on four main parameters: (1) melt productivity; (2) the H2O concentrations in solution, in excess and exsolvable; (3) the amounts of gold in the exsolvable water, and (4) the precipitation efficiency (Au efficiency = (Xinicial - Xfinal) / Xinicial). A database of the world’s largest gold and copper deposits and their modeling has shown that Au endowments are predominantly controlled by magma volumes, duration of the magmatic-hydrothermal activity, and precipitation efficiency. Nevertheless, Monte Carlo simulations can provide new petrogenetic information for a genetically related chain of Au deposits, compared to a large group of deposits scattered around the world, even if they have similar Au endowments and share similar geochemical characteristics. Therefore, to reduce uncertainties and have better statistical approximations, it is possible to refine some parameters such as melt productivity, H2O concentrations, and precipitation efficiency tied to spatially and temporally consistent input data. Thus, the CMB provides an unprecedented opportunity to model and study the Au endowment mechanism of a group of spatially and temporally related hydrothermal gold deposits. This research work aims to analyze the trigger mechanism in the formation of magmatic-hydrothermal gold deposits in the CMB of Colombia by using Monte Carlo modeling.
The rationale of the model
Data collection
Geochronological, chemical (Sr/Y ratios and alkalinity), and gold endowment information were obtained mainly by the metallogenetic map of Colombia (Celada et al., 2016), previous studies, and repositories of master and doctoral theses on seven primary gold deposits and provinces in the CMB (Supplementary Information A). The information corresponding to the ore duration (Ma) was made based on geochronological data of granitogenesis and syn- to post-ore porphyry and epithermal U-Pb, Ar-Ar and Re-Os dating (Bissig et al., 2017; Celada et al., 2016; Lesage et al., 2013; Naranjo et al., 2018). There is an information gap for the northern and southernmost zones of the CMB, which needs to be studied in detail in the future. However, the central zone is well detailed to model the Au endowments for the primary deposits in the CMB (Supplementary Information A).
Monte Carlo simulation
The Monte Carlo simulation was carried out in the RStudio program (https://rstudio.com), with more than 2 million iterations, to ensure reproducibility and replicability. The algorithm was based on Chiaradia (2020) and Chiaradia & Caricchi (2017), according to the model and mass balance of Annen et al. (2006) and Annen (2009). The simulation works with four main parameters, using concatenated equations (detailed in the attached Supplementary Information B), as follows: (1) melt productivity, defined as the hybrid melt produced in the crust; (2) the H2O concentrations in solution, in excess and exsolvable at pressure of saturation; (3) the amounts of gold in the exsolvable water, at the pressure of saturation within the crust, and (4) the precipitation efficiency, defined as the ratio of gold precipitated regarding the initial gold content in the fluid (Gold efficiency [%] = (m Au initial - m Au final) / m Au initial), where m is the molality in the initial and final fluid (Sanchez-Alfaro et al., 2016). The values and intervals used for each parameter to model hypothetical reservoirs of the lower, middle, and the upper crust of the CMB are explained in Table 1.
For reservoirs of the upper crust in archetypal gold deposits of the world, injection rates of10-2 m/year with radii greater than 7.5 km have been proposed to explain the volume of super-eruptions (> 450 km3) (Chang et al., 2007; Mason et al., 2004; Sparks et al., 2005). However, the present work models smaller-scale plutons and plugs (<40km2), for which a rate of 10-2 m/year with a smaller radius is proposed, plus injection time intervals between 40,000 -200,000 years and random pressures at which magma accumulation occurs between 0.3 - 3 kbar, corresponding to crustal depths of ~1 to ~10 km (Sanchez-Alfaro et al., 2016).
Parameters analyzed for reservoirs in the lower and middle crust of the CMB | Parameters analyzed for reservoirs in the upper crust of the CMB |
---|---|
|
|
Results
State of the art in the CMB
The CMB, located throughout the vast Cauca-Romeral fault systems (Fig. 1), between the Western and Central Andean Cordillera of Colombia (Cediel et al., 2003; Pulido, 2003; Restrepo & Toussaint, 2020; Velásquez et al., 2019), is an inter-Andean area of ~620 km long by 50 km wide hosting several Late Miocene porphyries, epithermal and intrusion-related deposits of gold, silver, and copper plus molybdenum, which has south continuity to Miocene porphyry and epithermal deposits in Ecuador (Schütte et al., 2012). The geology and tectonics of the CMB were studied at first by Grosse, 1926, in which he drew the cartography and mapped the artisanal mines of the region. Based on this classic work, researchers conceptualized the economic geology associated with the Cauca-Romeral fault system. The CMB comprises seven large deposits (> 1 Au Moz) in the central area, defined by (Sillitoe, 2008) as the Middle Cauca Belt and also five poorly explored zones to the south with numerous associated satellite deposits (Fig. 1), for which no geochemical information was available until now. The CMB also continues north to the Andean foothills, with a tradition of ancestral alluvial gold mining (OCDE, 2016), but also with no basic geochemical or geochronological information available. The main deposits include La Colosa (28.3 Au Moz), Quinchia (2.4 Au Moz), Marmato (8.7 Au Moz), Nuevo Chaquiro (5.6 Au Moz), La Mina (1.62 Au Moz), Titiribí-Cerro Vetas (4.6 Au Moz) and Buriticá (3 Au Moz) (Bartos et al., 2017; Bissig et al., 2017a; Lesage et al., 2013; Naranjo et al., 2018; Tassinari et al., 2008). These deposits and their associated plutons have been dated by the Colombian National Mining Agency giving ages between 5.6 and 10 Ma (Celada et al., 2016).
The magmatic rocks associated with deposits of the CMB display calc-alkaline andesitic to dacitic compositions, porphyritic textures and adakitic-like signatures, with high Sr/Y ratios mainly between 40-60 (Fig. 2b), except for the La Colosa deposit, which shows an average value of 26.8 (Gil-Rodríguez, 2010). The mineralization style of the Middle Cauca Belt varies in a south-north sense, where La Colosa is gold-dominant without copper and displays veinlets and bands of low grade (0.92 g/t) (Naranjo et al., 2018), whereas the Titiribí, La Mina and Quinchía deposits are low-grade gold-copper (<0.8 g/t) (Bissig et al., 2017b). Nuevo Chaquiro along with El Roble (~13.8 Ma) are copper-dominant porphyry deposits (Bartos et al., 2017). The Buriticá epithermal deposit shows thick (>30 cm) mineralized high-grade veins of gold-dominant (10.4 g/t) and silver, distributed in two main E-W vein systems (www.continentalgold.com), called Veta Sur and Yaraguá. The Marmato Au deposit is prominent in the area, due to its mining traditions for more than 100 years of gold exploitation, with veins and bands of 0.9 g/t (Tassinari et al., 2008).
In general, the basement of the mineral deposits associated with the Western Cordillera consist mainly of andesitic flows, tuffs, agglomerates and sandstones of the Late Miocene Combia formation (6 - 10 Ma; Jaramillo et al., 2019; Weber et al., 2020) and marine volcano-sedimentary sequences of the Late Cretaceous Barroso formation (Rodríguez & Zapata, 2013). Otherwise, the basement of the deposits towards the Central Cordillera consist mainly of black schists and micaceous metapelites of Triassic-Jurassic age from the Cajamarca Complex (240-230 Ma & 157-146 Ma; Blanco-Quintero et al., 2014; Vinasco et al., 2006) and the Early Cretaceous volcano-sedimentary rocks of marine origin from the Quebradagrande Complex (Nivia et al., 2006).
Monte Carlo simulationforAu endowments from the CMB
Low magma flux
Monte Carlo simulations with 3 million iterations, returned 1,919,000 plotted results. A group of endowments gave negative results, but they were not taken into account because a negative Au endowment has no geological significance. Results for a low magma flux between 0.0007 and 0.0011 km3/ year, yielded a wide spectrum of Au endowments, varying between 0 - 48,000 Au tons, with a mean of 8,277 tons and a median of 5,719 tons (Fig. 3a). These endowments are well above our case study (the CMB endowment interval is marked between the 2 red lines in Figs. 3a-d). Nonetheless, for injection time between 0 and 2 Ma, there is a volume of hydrous melt of up to 500 km3 with an Au endowment spectrum similar to those of the CMB (Fig 3b). For time intervals greater than 2 Ma there is no correlation with the CMB. The pressure at which magma accumulation occurs is sensitive up to 5 kbars (Fig. 3c), after this value and up to 9 kbars there is no good correlation as is shown for the injection time. Finally, the results for the efficiency parameter are very random between 0 -0.7%, and associates values of Au tons and volume of hydrous melt of0 - 20,000 and 0 - 1,500 km3 respectively. For an efficiency between 0.7 to 1%, the results are no longer random and a group of separate fields are observed (Fig. 3d).
High magma flux
Monte Carlo simulation for a high magma flux was only possible for a maximum number of 2 million iterations. For a larger number of iterations, the computer showed problems to concatenate the equations in Supplementary Information B. The results for a high magma flux exhibit a match with Au endowments of the CMB (also marked between the interval of 2 red lines in Figs. 4a-d). The iterations yielded a mean of 954 Au tons and a median of 760 Au tons, which are within the range of Au endowments in the CMB (Fig. 4a). Parameters such as pressure (at which magma accumulation occurs) and efficiency present random values in almost the whole graphic, although at low pressures and low efficiency, a trend is shown for 0 -1 kbars and 0 - 1.5% for pressure and efficiency respectively (Figs. 4c-d). The injection time on the other hand, shows a clear well-marked response to model the Au endowments, in which it is observed that for 0 - 160 ka, endowments between 0 - 1,000 Au tons and 0 - 600 km3 of volume of hydrous melt were found (Fig. 4b).
Discussion
Magmatic reservoirs in the Andean lower crust of Colombia in the early Miocene
The Monte Carlo simulations in the present work employing 3 million iterations, aimed to determine how the gold endowments were reached from the mantle to the shallow crust. The Late Miocene magmatic rocks located along the Cauca river valley include the volcanic rocks of the Combia Formation and the Cauca shallow volcanic intrusions (Weber et al., 2020). These rocks display calc-alkaline affinity (Jaramillo, 1976) and there is a group of rocks that chemically plots in the adakitic series (Borrero & Toro-Toro, 2016). In the CMB, there are also rocks of tholeiitic and shoshonitic series (Rodríguez & Zapata, 2013), but the analysis of the association of calc-alkaline, tholeiitic and shoshonitic magmas is beyond the scope of this work. However, the literature shows that a strong hypothesis for the association of these three magmatic series could be explained by the Caldas Tear (Vargas & Mann, 2013), which is described as a fracture of the Nazca plate under the South American plate, causing a flat slab, therefore there is an incipient post-collisional magmatism, placing rocks of the calc-alkaline, tholeiitic and shoshonitic series in the same regional context, however future geochronological studies need to be carried out to clarify this association. The calc-alkaline affinity rocks present garnets with rings of plagioclase, hornblende and plagioclase with zoned textures in the Chinchiná, Palestina, Jericó, Tesorito and El Poma areas (Alonso-Perez et al., 2009; Harangi et al., 2001), all of them related to the areas with polymetallic deposits within the CMB. Chemical outcomes showed that the garnet-bearing magmas undergone at least two stages of formation through the evolution of phenocrysts, where garnets were formed from 17-19 km and 950 °C and the rims and sieve textures in plagioclase indicate degasification and chemical/ physical disequilibrium (Weber et al., 2020). The previous results show that in the Late Miocene, the lithosphere associated with the CMB underwent enormous magmatic fractionation at the base of the earth's crust in the northern Andes of Colombia that later generates reservoirs in the upper crust.
The results obtained by the Monte Carlo simulation reflect the complex geodynamic context of the CMB, which is attributed to a transtensional basin setting (Weber et al., 2020). Monte Carlo simulations demonstrated that there is a low probability that Au deposits within the CMB are formed directly from the reservoirs generated in the lower and middle crust (Figs 3a-d), mainly because the Monte Carlo simulations show Au endowments generally outside the range of the CMB (Figs. 3 and 5), however, because earlier deposits reveal high early fractionation of garnet and hornblende and total rock chemistry (with high Sr/Y ratios Fig. 2) showing a development of rocks of the adakitic series (Borrero & Toro-Toro, 2016), it is probable that the reservoirs of the lower and middle crust, have fed more felsic reservoirs to the upper crust, forming a more evolved dacitic intermediate reservoirs. Monte Carlo simulations also shows a low probability of triggering hydrothermal gold deposits from basalt magmas formed directly from lower tectonostratigraphic levels (Figs. 3a-d), but at timescales of millions of years (Fig. 5). However, this works for porphyry-type Au deposits, while epithermal deposits are more associated with mineralization stages in windows of thousands years (Sanchez-Alfaro et al., 2016). Finally, Monte Carlo simulations show that the mean values of gold endowments associated with the CMB are associated with reservoirs in the lower and middle crust with dimensions between 500 and 2,000 km3 (Fig. 5), values that are above an order of magnitude, compared to reservoirs modeled for the upper crust.
Reservoirs of the upper crust and generation of porphyry and epithermal deposits
The CMB is characterized by having several porphyry and epithermal Au deposits, some of them located less than 30 km apart, along a 100 km long strip, indicating consecutive small-scale reservoirs in the upper crust, due to a previous process of magmatic differentiation in the middle and lower crust (Weber et al., 2020). Considering this, we propose an injection rate of 50 mm/y, through a disk of 6,500 m radius, corresponding with a high magma flux of 0.007 km3/ year, coupled with injection time intervals between 40,000 and 200,000 years for the shallow levels. An injection rate greater than 50 mm/y results in a greater volume of hydrous melt, and would model scenarios outside the CMB range (Fig. 5), which vary between 0 to 1000 Au tons. It is also observed that between 0 to 1,000 tons of Au, mineralization intervals <120 ka can develop in hydrous melt volumes of less than 300 km3, while longer intervals of mineralization reach a maximum of 800 km3 (Fig. 5). On the contrary, injection rates of less than 50 mm/y are expected for the lower crust but with larger time intervals (in Ma; Fig. 5) and would mean a transport of hydrous basaltic melt to the crust, with mean values well outside the range of the CMB. Therefore, by means of several interactions using a high magma flux of 0.007 km3/year, it can be observed that Au endowments match the amount found in the CMB (Fig. 4).
The mineralization style of the epithermal deposits in the CMB are of high-grade vein characteristics, with values of up to 10 g/t, which suggests that the efficiency would have values of tens of %, however the model shows that the obtained efficiencies for epithermal deposits varies between ~1 - 2%, slightly higher than the endowments for porphyry deposits (~0.7%). Thus, it was quantitatively observed that the gold efficiency from epithermal deposits is lower than the expected and slightly higher than porphyry type deposits or that the present method has no way of modeling resetting and flashing processes.
The emplacement of mineralizing hydrothermal fluids can be closely linked to the Cauca-Romeral fault system. While the ore ages vary between 10 -5.6 Ma, the reactivation of the Cauca-Romeral fault system is ~6.3 Ma (Weber et al., 2020). Thus, the mineralizing processes in the CMB can be associated with pre, sin and post deformation activity underwent in the Cauca Valley in Late Miocene. In the case of the La Colosa porphyry-type gold deposit, it is located along the Palestinian fault, while the deposits of Marmato, Nuevo Chaquiro, La Mina, and Quinchía are associated with the central and western Cauca fault system. Finally, the Buriticá epithermal deposit is closely associated with the Tonuzco fault, of which it is located in the main entrance to the Higabra Mine. Thus, the complex Cauca-Romeral fault system was permeated in the Late Miocene by the water expelled by the hydrated residual magma from the more superficial reservoirs. This complex fault system could have been responsible for generating the seismic activity to generate all the deposits associated with the CMB, as has been modeled in (Sanchez-Alfaro et al., 2016; Sanematsu et al., 2006).
Conclusions
In the present research project, Monte Carlo simulations were used to generate a unique formation model for the huge CMB in the Andes cordillera of Colombia. The results of this work show that for a high dacitic magma flux of 0.007 km3/year and pressures below 3 kbar, the gold endowments obtained were representative for epithermal and porphyry Au deposits of the region, whose hypothetical models fit an equivalent range up to 1,000 Au tons. For time intervals of magma injection <200 ka, the values are associated with epithermal deposits where there is a close interaction of exsolved fluids with the Cauca-Romeral fault system, because of a strong decompression in a transtensional basin setting. Otherwise, the modeling for a low basaltic magma flux between 0.0007 - 0.0011 km3/year and efficiencies of ~0.7% showed that only for pressures above 5 kbar that correspond to the middle crust, the modeled reservoirs showed sufficient available content of exsolvable H2O and gold to feed the reservoirs in the upper crust or generate gold deposits from them along the Cauca-Romeral regional fault system, but in longer time intervals.