Introduction

Copper’s criticality stems from its broad range of industrial applications, particularly its key role in advancing green energy technologies and electrification infrastructure. The Earth’s crust typically contains copper (Cu) concentrations of less than 30 ppm1, making its economic extraction viable only in ore deposits. Magmatic-hydrothermal deposits such as porphyry Cu-Mo-Au and related epithermal Au-Ag systems are a primary source of global copper and gold production and supply considerable amounts of molybdenum and silver, among other critical elements, including rhenium2,3. Most commonly found in subduction zones, porphyry-epithermal systems are formed by the efficient transport and accumulation of ore metals that precipitate predominantly as sulfides from hydrothermal fluids released from evolving arc magmas4,5,6. Although it is increasingly recognized that oxidizing conditions and elevated concentrations of water, sulfur, and halogens in arc magmas are essential to the formation of porphyry-epithermal deposits over Earth’s history7,8,9, the mechanisms governing the occurrence of both fertile and barren magmatic-hydrothermal systems within restricted arc segments remain puzzling10.

An increasingly recognized hypothesis holds that the ore metal fertility of a productive crustal domain is given by a combination of processes operating during the evolution of magmas and fluids throughout the whole lithospheric column6,11,12,13. These include the production of oxidized magmas rich in water and ligands10,14,15,16, metal enrichment in the magma sources or their melting products17,18,19, the efficiency of metal and ligand transfers from magmas to hydrothermal fluids20,21, the duration of the magmatic-hydrothermal activity in the shallow crust22, and the efficiency of ore precipitation13. Another important factor relates to the intricate nature of transcrustal/deep-seated structural controls on magma emplacement and fluid circulation, which can significantly influence ore formation23,24. A recent view argues that the most favorable structural conditions for the formation of an intrusion-related hydrothermal ore system are provided by deep-seated faults oriented at a high angle with respect to the maximum principal the regional stress, i.e., fault systems that are misoriented for reactivation with respect to the prevailing stress field25,26. Such misoriented faults may promote magmatic differentiation at depth, leading to increasing residence times and multi-episodic magmatic recharge processes that ramp up the volatile content of magmas7,13,27. Accumulation of magma-derived volatiles favors host rock hydraulic fracturing at nearly lithostatic fluid pressures fostering positive feedback between fracturing and fluid flow28. Also, the elevated water and ligand content of magmas is expected to result in a more efficient transfer of ore metals such as Cu to hydrothermal fluids8,29,30. Although crustal thickening is considered a first-order control on metal transport and concentration in ore systems10,31,32, the structural controls on magmatic-hydrothermal plumbing systems and ore metal enrichment—and thus, their impact on the size and grade of the deposit—remain essentially unknown. This challenge is mainly because ore deposits represent fossil magmatic-hydrothermal systems that record a time-integrated sequence of fluid-flow pulses, hindering direct estimations of the enhancing effect that fault systems may have on mineralization events.

We tackle this gap in knowledge by determining the variations in the Cu concentration of fluids sampled from active hydrothermal systems in central-southern Chile as modern analogs of shallow mineralizing systems33,34,35. We do this by sampling subaerial fumaroles and associated hot springs that occur in contrasting volcano-tectonic domains. A key advantage of our case scenario is that the interaction between magmatism, hydrothermal activity, and seismicity in contrasting volcano-tectonic domains is well established27,34,36,37,38,39. By combining the chemical and helium isotopic composition of fumarole gases and related hot spring waters sampled along different structural domains, we show that the nature of fault systems actively influences the Cu budget of ascending hydrothermal fluids.

Results and discussion

Volcano-tectonic domains

The studied fumaroles and hot springs are associated with active Pleistocene-to-Holocene stratovolcanoes and are distributed along a ~450 km segment (36°–40°S) of the central and southern portions of the Chilean Southern Volcanic Zone (SVZ) (Fig. 1). The thickness of the continental crust in this segment (approximately ~40 km) varies little along-strike and is located above a slightly shallow dipping (<25°) Benioff zone40.

Fig. 1: Simplified structural map of the central portion of the Southern Volcanic Zone (SVZ) of Chile.
figure 1

Solid and dashed lines represent mapped and inferred faults, respectively. Circles, hexagons, and diamonds indicate the location of fumarole samples associated with different structural domains. A detailed version of this map is presented in Supplementary Fig. S1. ATF Andean Transverse Faults, LOFS Liquiñe-Ofqui Fault Systems, NC Nevados de Chillan, PCC Puyehue-Cordón Caulle, SN Sierra Nevada, TH Tolhuaca, SP Sollipulli, CP Copahue.

Full size image

Figure 1 shows the location of the studied hydrothermal manifestations and a simplified sketch of the main associated structural features. A detailed structural map is reported in Supplementary Material Fig. S1. Geophysical data, including magnetotelluric, ambient seismic noise Rayleigh-wave tomography, and InSAR inversion data suggest that these structural features exert a first-order control on the magmatic-hydrothermal plumbing system (Supplemental Material Table S1). As such, the volcanic and hydrothermal activity of this segment can be classified into three main volcano-tectonic domains, as follows.

  1. 1.

    A transtensional domain is associated with an NE striking network of extensional and extensional shear fractures that are broadly parallel to the current maximum horizontal stress in the SVZ (hereafter NE transtensional) (Fig. 1; Supplementary Material Fig. S1; Methods). These faults-fracture networks typically form under relatively low differential stress and correspond to secondary structures of the intra-arc NNE-striking dextral strike-slip Liquiñe Ofqui Fault Systems or to transverse ENE-striking basement faults36,41. NE faults can, for instance, form pull-apart structures in accommodation zones or act as transtentional faults under volcanic systems42. Under the NE-transtensional domain, the magmatic plumbing system is dominated by interconnected vertical faults and tension cracks favoring a rapid ascent of basaltic to basaltic-andesite magmas from the deep crust with little crustal inputs27,43.

  2. 2.

    A compressional to the transpressional domain is associated with a network of NW-striking faults and subhorizontal cracks (hereafter NW transpressional) (Fig. 1; Supplementary material Fig. S1; Methods). In this domain, the controlling NW faults are oriented at a high angle with respect to the current maximum horizontal stress in the SVZ (σ1 = N60E) and can only be activated by fault-valve mechanisms that require supra-lithostatic fluid pressures25,36. Plumbing systems associated with this compressional regime favor relatively shallow (<10 km) magma reservoirs, longer residence times, and episodic magma fractionation forming basaltic to rhyolitic volcanic products27 (Supplementary Material Table S1).

  3. 3.

    A third volcano-tectonic association occurs where the NW striking faults intersect with the transcurrent NNE striking faults of the intra-arc Liquiñe-Ofqui Fault Systems (hereafter NW-NNE interplay), forming a local combined compressional and strike-slip regime (Fig. 1; Supplementary Material Fig. S1; Methods). In this case, both systems are likely to accommodate deformation depending on local and transient changes in the stress field36,37. In this structural regime, the magmatic plumbing system is tapped by NW compressive faults and fed by deep crustal inputs that ascend through the deep-seated transcurrent NNE faults27.

Sampled hydrothermal fluids

A total of eight shallow fumarole gas and hot spring water samples were used in this study, spanning the three main tectonic-magmatic domains mentioned above. Sampling sites represent all currently known stratovolcano-related hydrothermal systems with active fumaroles within the 36°–40°S segment of the SVZ. Two samples are associated with a purely NW transpressional domain (NC1, NC2, Nevados de Chillán; Fig. 1), three samples are associated with purely NE transtensional domains (CP1, CP2, Copahue; SP, Sollipulli; Fig. 1), and three samples were collected in systems related with the NW-NNE interplay domain (TH, Tolhuaca; SN, Sierra Nevada; PCC, Puyehue-Cordón Caulle; Fig. 1).

To obtain representative values for sulfur, carbon dioxide content, helium isotope ratios, and trace metal concentration of the hydrothermal fluids, we sampled both the acidic hot spring waters and the associated fumarole gas at the surface. The acidic waters are produced by the natural condensation of water vapor in the fumarole gas due to the temperature drop close to the surface, capturing the acid species (SO2, HCl, HF) and metals44. The fumarole gases were also condensed through a cooling system to preserve the metal content remaining in the gas phase after the natural condensation process (See Methods). All samples were collected in sites where it was possible to obtain both the gas (for the analysis of Stot, CO2, and 3He/4He ratios) and the highly acidic waters (pH < 3; for Cu, Co, and Ni analyses), this strategy allowed to obtain representative data of the magmatic-hydrothermal system. Sulfur was measured both in the gas (as Stot = SO2 + H2S) and in the acidic waters (SL = S in SO42-).

Sources and composition of the hydrothermal fluids

The chemical and isotopic composition of gases and water are reported in Table 1. The concentration of CO2 and Stot in gases ranges from 72.99% to 95.22% and 0.65% to 15.5%, respectively. The SO2 species, along with HCl and HF, are mostly scrubbed (dissociated in water) when the water vapor condenses during ascent towards the surface45.

Table 1 Location, chemical, and rare gases isotope data of sampled hot spring and fumarole samples from the Southern Volcanic Zone of Chile
Full size table

Helium isotopes (3He/4He) are expressed as Rc/Ra, which is the air-corrected 3He/4He ratio of the sample (Rc; See Methods) over the present-day 3He/4He atmospheric ratio (Ra = 1.384 × 10−6)46. Because 4He is derived from the radiogenic decay of U-Th, the Earth’s crust is mainly dominated by 4He with low Rc/Ra ratios (0.05)47 while the depleted mantle shows elevated Rc/Ra ratios (8 ± 1)48, caused by the presence of the preserved primordial 3He isotope. These contrasted ratios between the atmosphere, the crust, and the mantle reservoirs are very useful to distinguish the sources of volatiles in a hydrothermal system49. The Rc/Ra of sampled fumaroles range between 3.54 and 7.80, with an analytical error <0.5 (Table 1; Methods). All the fumarole gas samples display much higher Rc/Ra ratios than purely crustal values, indicating that the whole studied crustal segment receives a strong contribution of mantle-derived helium. However, whereas samples associated with the NW transpressional domain (NC1, NC2) present the most radiogenic Rc/Ra values (3.54 and 3.68, respectively), samples associated with NW-NNE interplay (TH, SN, PCC) present moderately radiogenic values (4.61 and 5.70), and samples in the NE transtensional domain (CP1, CP2, SP) exhibit mantle Rc/Ra values between 7.52 and 7.72. The similarity between the Rc/Ra of hydrothermal fluids and fluid inclusions in associated volcanic rocks further supports a magmatic primary source of volatiles in the sampled fumaroles with limited to no modifications after fluid exsolution50,51,52. In addition, the correspondence of Rc/Ra ratios in hydrothermal fluids with literature Sr isotope data in the associated volcanic rocks suggest that Rc/Ra variations are related to radiogenic crustal inputs to the magmatic reservoir during differentiation prior to fluid exsolution37,50.

Copper was detected in all the acidic water samples. Its content, along with the SL, Cl, and other base metal concentrations, are presented in Table 1. Most of the metal concentrations in the gas condensates are negligible, indicating that the metal load of the ascending hydrothermal fluids partitions into the liquid phase. The narrow range of water pH, between 2.2 and 3, and the overall similar nature of the sampled systems exclude the possible change in Cu and S solubility in water due to shallow processes, ensuring that the metal concentrations reflect only variations in the fluid’s source44. Low-temperature hydrothermal fluids are the ultimate result of a complex evolution of more primary, high-temperature, magmatic-hydrothermal fluids that may have undergone cooling and expansion, brine condensation and/or precipitation of metal-bearing phases, and/or scrubbing processes53. Nonetheless, all sampled systems occur within a similar geodynamic setting, and fluids share physicochemical features (e.g., temperature and pH) that suggest analogous evolutionary pathways. The Cu composition of the hydrothermal fluids is expressed as Cu/SL ratios (and not absolute Cu concentrations) to avoid the metal content variability that may be due to the unquantifiable degree of mixing with surficial waters, this provides a more reliable measure of their primary metal composition and allows comparisons of samples from different systems. Notably, the Cu/SL ratios of our samples are comparable to those reported for high-temperature volcanic gases and ore-forming fluids53,54,55. In addition, the distinctive correlation of Cu/SL with independent geochemical proxies—such as Rc/Ra and Stot/CO2—strongly suggests that the Cu/SL ratio responds to primary fluid compositional variations (Fig. 2, see below). Therefore, and although the absolute Cu concentration of the acidic waters is lower than those typically found in high-temperature volcanic gasses and fluid inclusions53,55, we posit that the Cu/SL ratio of the sampled hot springs stands as a reliable proxy to compare Cu compositional variabilities in magmatic-hydrothermal fluids.

Fig. 2: Binary plots showing the helium isotope and metal composition of modern hydrothermal fluids in the SVZ.
figure 2

A Cu/SL*1000 vs. Rc/Ra. Linear trendline (R2 = 0.94) and 95% confidence interval is reported. The color gradient bar represents the Stot/CO2 ratio of each sample. B, C show the Rc/Ra vs. Ni/SL*1000 and Co/SL*1000, respectively. Error margins are lower than each mark size (see Methods). NC Nevados de Chillan, PCC Puyehue-Cordón Caulle, SN Sierra Nevada, TH Tolhuaca, SP Sollipulli, CP Copahue.

Full size image

Figure 2A shows that the Cu/SL ratio negatively correlates with the Rc/Ra ratio (R2 = 0.95). The color gradient in Fig. 2A illustrates the correspondence of the Cu/SL vs. Rc/Ra correlation with the Stot /CO2 ratio of the hydrothermal fluids. The Stot/CO2 ratio serves as a proxy of the fluid exsolution depth. Under higher pressure conditions, less soluble gas species in magmas like CO2 are exsolved (even at depths of over 25 km), whereas more soluble species such as H2S and SO2 are released at later stages at shallower depths56,57. Thus, increasing Stot/CO2 ratios reflect the decreasing depth of fluid exsolution. Nonetheless, we warrant caution in this interpretation, given the low-temperature and shallow nature of the studied systems. Figure 2A shows that samples containing deep magmatic volatiles—corresponding to higher Rc/Ra—have lower Cu/SL and were likely sourced at greater depths. By contrast, samples with more radiogenic He contributions (lower Rc/Ra) have higher Cu/SL and were most likely produced at shallower crustal depths. In addition, Ni/SL and Co/SL also show a negative correspondence with Rc/Ra (Fig. 2B, C and Table 1). Copper, Ni, and Co have similar geochemical behavior and partitioning into similar phases during the magma differentiation, including in mantle and crustal magmatic sulfides as well as in hydrothermal sulfides in both active and fossil systems18,20,35,58,59. The Ni/Cu and Co/Cu ratios of our samples (Table 1) are consistent with the same ratios in magmatic sulfides and volcanic fumaroles in arc settings (e.g., Merapi volcano, Indonesia), pointing to a potential source relationship20. However, unlike Cu, Co, and Ni also present a lithophile affinity during magmatic differentiation, which may explain some of the dispersion in Fig. 2B, C.

Structural controls on the Cu concentration of hydrothermal fluids

Tectonic and magmatic processes in the porphyry-epithermal environment result in the formation of high permeability conduits that efficiently focus and transport ore-forming fluids28. An increasing body of evidence suggests that giant porphyry Cu deposits (e.g., El Salvador, Spence, Centinela, and Escondida) and deposit clusters tend to form in association with compressional to transpressional faults that are oblique with respect to the regional stress field25,26. This structural configuration is thought to limit the ascent of magmas, which in turn promotes further differentiation and fluid accumulation, increasing the ore-forming potential of a magmatic-hydrothermal system25. We argue that this optimal structural configuration also impacts the Cu budget of fluids based on the contrasting Cu/SL vs. Rc/Ra signature of modern hydrothermal fluids associated with different structural arrays.

Our data reveal three main cases that relate the helium isotopes and metal composition of hydrothermal fluids to the structural domains recognized in the Chilean SVZ. On the one hand, samples associated with NE transtensional structural domains display the highest (most primitive) Rc/Ra signatures, the lowest Cu/SL ratios, and the lowest Stot/CO2 ratios (Fig. 2). On the other hand, samples associated with the NW transpressional structural domain show the lowest Rc/Ra ratios and the highest Cu/SL and Stot/CO2 ratios (Fig. 2). Finally, samples associated with the NW-NNE interplay domain have intermediate Rc/Ra, Cu/SL, and Stot/CO2 ratios (Fig. 2).

A conceptual model that explains the three observed scenarios is presented in Fig. 3. The crustal faults of the NE transtensional domain (Fig. 3A), optimally oriented for opening with respect to the regional stress field, promote the deep exsolution of hydrothermal fluids facilitated by the formation of open fractures36. Thus, fluid exsolution is produced from a deep and relatively primitive source, consistent with the high measured Rc/Ra and low Stot/CO2 values. These NE transtensional faults networks prevent fluid accumulation and magmatic differentiation and, thus, source magmas may have not yet evolved to a stage where the concentration of ligands such as Cl is high enough to allow efficient extraction of Cu and other metals from magma8,10. This is consistent with the more mafic character of erupted products in these volcanic systems27. The favorable scenario for Cu ore formation is shown in Fig. 3B, where the dominant structural control is exerted by faults of the NW transpressional domain. Because these structures are misoriented for reactivation with respect to the regional stress field, they prevent the early separation of fluids from magmas. Instead, they promote increasing residence times and magmatic differentiation and produce magmatic enrichment in volatiles and ligands, leading to a more efficient Cu extraction at the time of fluid exsolution29. Their lower Rc/Ra values are explained by either interaction with crustal materials or by the magmatic production of radiogenic He (magma aging) associated with their more protracted residence times50. Fluids separated at this stage would thus present relatively high contents of Cu and can migrate through the shallow local faults controlling the hydrothermal system. The sourced Cu may be in the silicate melt fraction or in the sulfide phases of a saturated magma that could be redissolved to transfer their metal load to the hydrothermal fluid20,60. The latter is consistent with the similarity of the Ni/Cu and Co/Cu ratios of hydrothermal fluids and magmatic sulfides reported in volcanic systems in arcs20. As such, in the NW transpressional domain, fluids are characterized by radiogenic He isotope compositions, higher Stot/CO2 ratios, and higher Cu/SL. Finally, in the NW-NNE interplay domain (Fig. 3B), the misoriented fault systems promote magmatic differentiation and fluid accumulation as in the previous scenario. However, the interplay with the deep-seated NNE-striking transcurrent faults also allows the involvement of deep magmatic or hydrothermal inputs from more primitive sources. As a result, the hydrothermal fluids represent the mixing of two contrasting endmembers.

Fig. 3: Conceptual model of the structural-geochemical links in variable volcano-tectonic domains.
figure 3

A Optimally oriented NE transtensional faults promote fluid to escape from deep and primitive magmatic systems where Cu remains mainly in the magma at the time of fluid exsolution (in silicate melt or sulfides). Panel B shows both the hydrothermal systems associated with the NW transpressional faults (red), where hydrothermal fluids form from a more evolved, ligand-rich magma, scavenging Cu more efficiently (form silicate melt or sulfides). Systems associated with the NW-NNE interplay (green) are also represented in panel B, where the mixture with deeply sourced fluids leads to intermediate compositions.

Full size image

It is relevant to note that the relative paucity of economic ore deposits in the studied area is likely the result of fundamental tectono-magmatic conditions, such as the effects of the crustal thickness on the low initial Cl and S composition of near-primary melts10. The hydrothermal systems studied here occur within a mostly barren crustal segment of relatively homogeneous thickness. Thus, the variable structural configurations allow us to better examine the extent to which fault systems that control the ascent of magmas can impact the geochemistry of fluids. Whereas active fault systems misoriented for reactivation are favorable for ore formation (fluids with higher Cu/SL and lower Rc/Ra), the presence of fault systems optimally oriented for reactivation is unfavorable for the genesis of coeval magmatic-hydrothermal ore deposits (fluids with lower Cu/SL and higher Rc/Ra). These results align well with helium isotope data extracted from fluid inclusions within various mineralized systems worldwide, including porphyry Cu, epithermal Au-Ag, and skarn deposits (see Supplementary Material Table S2). Such deposits are formed by hydrothermal fluids derived from variably evolved arc magmas with low Rc/Ra ratios, between ~3 and ~0.2 (Supplementary Material Table S2). Our study reveals variable Rc/Ra ratios in geothermal manifestations of the SVZ, with the lowest values (Rc/Ra ~3.5) in the Cu-richest samples NC1 and NC2. The lowest Rc/Ra values that we observe draw near the fertile systems’ threshold, in agreement with their association with the NW transpressional domain, their higher Cu/SL ratio, and the role of misoriented fault systems on the fertility of hydrothermal systems. Our results highlight that the structures associated with magmatic-hydrothermal systems can contribute to modulating the Cu budget of produced fluids, ultimately impacting the location and size of a magmatic-hydrothermal deposit. Similar examples may include Andean-type iron oxide-copper-gold (IOCG) and iron oxide-apatite (IOA) systems, which occur within the world-class metallogenic province of the Coastal Cordillera in central-northern Chile and southern Perú61,62,63. The formation of IOA and IOCG deposits was intimately associated with the activity of the sinistral strike-slip Atacama Fault System during the early Cretaceous64. As reported27,65, Fe-(±Ti-V)-rich and Cu-S-poor IOA deposits occur mostly along the NNE principal trace of the fault, while Cu-S-rich IOCG systems often occur along NW structures, pointing to structurally driven Fe vs. Cu partitioning. Finally, this study also supports the hypothesis that essential tectono-magmatic elements, such as increasing crustal thickness and elevated volatile contents, need to be given for an arc segment to be productive. Our conclusions contribute to understanding the local variations that produce barren and Cu-rich deposits within restricted arc segments of the same age and crustal thickness, having direct implications for conceptual mineral exploration.

Methods

Sampling and analysis of fumarole gases, fumarolic condensates, and acidic waters

Fumarole gases were sampled using a titanium tube, inserted into the fumarole vent, and connected to a condenser to force the separation of a dry gas and a fumarolic condensate. The dry gas aliquot for chemical components and helium isotope analysis was collected in pre-evacuated alkaline glass containers with vacuum valves at both ends. This method for volcanic gas sampling is exhaustively described by ref. 37. Both fumarolic condensates and hot spring water samples were collected in pre-cleaned high-density polyethylene bottles and were analyzed for their chemical composition.

Six out of eight helium, CO2, and Stot data have been documented in prior works37,66. The chemical composition (Stot and CO2) of sampled gases was measured using a quadrupole mass spectrometer. Experimental errors are ±10%. The gas mixture in the copper tubes was diluted manually in a constant volume until reaching a pressure lower than 10 mbar. The reactive gases (e.g., H2O) were removed using two Ti-getter, operated at 600 °C for 15 min, followed by 10 min at ambient temperature. One SAES ST-707 getters were also used at 100 °C for 15 min followed by 10 min at ambient temperature. Gases were then adsorbed onto an Advanced Research System cryogenic trap containing activated charcoal at 10 K and released at 40 K for He and 90 K for Ne. He and Ne isotopes were measured at the Montreal Noble Gas Laboratory (GRAM) of GEOTOP on a Thermo® HELIX-MC Plus using an axial Faraday detector by peak, except for 3He, which was measured by ion counting on an axial Compact Discrete DynodeTM (CDD) detector. Obtained signals were calibrated against standard air. Blanks are typically on the order of 0.01% for He and Ne. Typical standard reproducibility for 4He and 20Ne are 1.5–2%. Errors on the 3He/4He ratios are 2% at 1σ for high 3He (magmatic) samples.

The measured 3He/4He ratios, normalized to the atmosphere Ra = 1.384 × 10−6, have been corrected for the presence of atmospheric helium (Rc/Ra) using the 4He/20Ne ratio of the sample67:

$$\frac{{R}_{c}}{{R}_{a}}=\left(\frac{R}{{R}_{a}}-r\right)/\left(1-r\right)$$
(1)

where “r” is:

$$r = {\left(\frac{{4\atop}\!He}{{20\atop}\!Ne}\right)}_{ATM} \Bigg/ {\left(\frac{{4\atop}\!He}{{20\atop}\!Ne}\right)}_{obs}$$
(2)

where (4He/20Ne)ATM and (4He/20Ne)obs are the atmospheric and observed 4He/20Ne ratios, respectively.

Anions (Cl, S) of hot spring waters were analyzed in ACTLABS laboratories in Canada using an Ion Chromatography (IC, Dionex ICS 2100) with the reported errors of Cl = ± 0.15 mg/L and S = ± 0.1 mg/L, and detection limits of Cl = 0.3 mg/L and S = 0.3 mg/L. Concentrations of trace elements (Cu, Co, Ni) were analyzed in ACTLABS using an Inductively Coupled Plasma–Mass Spectrometry (ICP-MS; iCap). Reported errors are the following, Cu = ±0.2 µg/L, Co = ±0.03 µg/L; Ni = ±0.4 µg/L. Detection limits are the following, Cu = 2 µg/L, Co = 0.05 µg/L; Ni = 3 µg/L.

Structural map

The regional and local structural domains documented in this work were defined using a combination of photointerpretation on Digital Elevation Models and quick bird images (from Google Earth), and a compilation of previous works described in Supplementary Fig. S1.

Online content

Any methods, additional references, Nature Research reporting summaries, source data, statements of code and data availability, and associated accession codes are available in the supplementary information.