The chromitite bodies consist of massive (>80 vol.% chromite), semi-massive (50-80 vol.% chromite), and to a lesser extent, disseminated (30-50 vol.% chromite) textures. These chromitites were intensely modified by metamorphism, which has produced the three textural varieties of chromites defined byColáset al. (2016) (Table 1;Figure 3). Type I chromite, without exsolved phases and optically homogeneous (Figure 3a), shows a trend of increase of Cr# [Cr/(Cr+Al) atomic ratio] coeval with a decrease of Mg# [Mg/(Mg+Fe²⁺) atomic ratio] displaced at lower Mg# from massive (Mg# = 0.52-0.68 and Cr# = 0.47-0.72), semi-massive (Mg# = 0.48-0.55 and Cr# = 0.60-0.73) to disseminated (Mg# = 0.42-0.52 and Cr# = 0.55-0.63) chromitites (Table 1;Figure 4a to4c). The Fe³⁺/(Fe³⁺+Fe²⁺) ratio varies from 0.43 to 0.54 in Type I chromite from massive and disseminated chromitites, but is lower in Type I chromite from semi-massive chromitites (Fe³⁺/(Fe³⁺+Fe²⁺) = 0.24-0.27) (Table 1;Figure 4a to4c). Type II chromite shows a magnetite-rich phase (Mg# = 0.21-0.23, Cr# = 0.85-0.87, and Fe³⁺/(Fe³⁺+Fe²⁺) = 0.62-0.63) that forms irregular blebs randomly distributed along the rims or linear strings, and fine lamellae with apparent crystallographic control hosted in a spinel-rich phase (Mg# = 0.56-0.63, Cr# = 0.50-0.54, and Fe³⁺/(Fe³⁺+Fe²⁺) = 0.43-0.52) (Table 1;Figures 3c and4d to4f). Type III chromite, with symplectitic texture, is composed of variable proportions of magnetite-rich (Mg# = 0.29-0.33, Cr# = 0.76-0.79, and Fe³⁺/(Fe³⁺+Fe²⁺) = 0.62-0.64) and spinel-rich (Mg# = 0.64-0.73, Cr# = 0.41-0.43, and Fe³⁺/(Fe³⁺+Fe²⁺) = 0.40-0.54) phases (Table 1;Figures 3d and4d to4f). This chromite prior to exsolution have similar Mg# (0.44-0.50) and Cr# (0.54-0.59), but higher Fe³⁺/(Fe³⁺+ Fe²⁺) (0.60-0.61) than Type I chromite from disseminated chromitites (Mg# = 0.42-0.52, Cr# = 0.55-0.63, and Fe3+/(Fe³⁺+Fe²⁺) = 0.43-0.48) (Table 1;Figures 4a to4c).

Figure 3
Backscattered-electron (BSE) images of the different textural varieties of chromites identified in Los Congos and Los Guanacos chromitite bodies. Type I chromite in massive (a) and semi-massive chromitites (b) with relics of clinopyroxene; Type II chromite from Los Congos (c) and Type III chromite from Los Guanacos (d). Abbreviations: Atg: antigorite, Cb: carbonates, Chr: chromite, Cpx: clinopyroxene, Mag: magnetite, Spl: spinel (Whitney and Evans, 2010).

Figure 4
Compositional variations of chromite grains in the studied chromitites from Los Congos and Los Guanacos in terms of Mg# [Mg/(Mg+Fe²⁺) atomic ratio] versus Cr# [Cr/(Cr+Al) atomic ratio] (a and d) and Fe³⁺/(Fe²⁺+Fe³⁺) ratio (b and e), and Al-Cr-Fe³⁺ compositions (c and f). Legend provided as inset in the figure. Dashed lines in (a and d) denote the composition of olivine in equilibrium with chromite at a nominal temperature of 1,200 ºC (Dick and Bullen, 1984). Shaded fields in plots are the composition of chromite from Los Congos and Los Guanacos chromitites reported byEscayolaet al. (2004) andProenzaet al. (2008).

Table 1
Characterization of the chromitite samples investigated in this study.

Note: Mineral abbreviations afterWhitney and Evans (2010).^{* a)} Composition calculated using the proportion of each phase in exsolved chromites according toColás et al. (2016).

Type I chromite is found in clinopyroxene-bearing massive, semi-massive and disseminated samples from Los Congos and Los Guanacos massifs, whereas Type II chromite is preserved exclusively in clinopyroxene-bearing disseminated samples from Los Congos, and Type III chromite occurs in clinopyroxene-free massive chromitites from Los Guanacos (Table 1).

The three varieties of chromite are made up by subhedral and less frequently anhedral crystals (from 0.2 to 1 cm across) with rounded shapes, and have a fracture network of variable density (Figure 3); these fractures are often filled with secondary silicates (i.e., antigorite and chlorite), and less frequently with carbonates (Figure 3c). No primary olivine was observed in the chromitites, whereas antigorite and chlorite forms the interstitial matrix between chromite grains (Figure 3). Clinopyroxene are also found intergranular to Type I and Type II chromites (Figures 3a and3b). Small inclusions of these silicate minerals are scattered within chromite grains.

In situ LA-ICP-MS analyses of Type I chromite in massive chromitites from Los Congos and Los Guanacos show variable albeit similar contents of Ga (10.32-32.70 ppm), Ti (615-1354 ppm), Zn (1282-2853 ppm), and Mn (2890-8215 ppm), higher contents of Sc (1.5-10.07 ppm) and Ni (1012-2386 ppm), and lower contents of Co (287-370 ppm) and V (657-1007 ppm) than those in semi-massive chromitites from Los Guanacos (Ga = 18.96-27.46 ppm, Ti = 746-1449 ppm, Zn = 1974-2508 ppm, Mn = 2540-3172 ppm, Sc = 0.33-1.23 ppm, Ni = 634-851 ppm, Co = 372-397 ppm, V = 1118-1255 ppm) (Figure 5a to5b). Spinel-rich phase of Type III chromite in massive chromitites from Los Guanacos are enriched in Ga (26.95-30.13 ppm), Co (616-699 ppm) and Zn (5118-6024 ppm), but depleted in Sc (1.03-2.02 ppm), V (598-836 ppm), Ti (302-526 ppm), Mn (3244-4197 ppm) and Ni (1781-2877 ppm) than the magnetite-rich phase (Ga = 9.42-14.88 ppm, Co = 280-407 ppm, Zn = 846-1082 ppm, Sc = 2.19-4.67 ppm, V = 830-1282 ppm, Ti = 876-1378 ppm, Mn = 3097-5083 ppm, Ni = 3700-5812 ppm) (Figure 5c and5d).In situ analyses of Type II chromite from Los Congos were excluded because the beam size of LA-ICP-MS (~30-50 μm) is larger than the size of the exsolved phases (seeFigure 3c).

Antigorite is present in Type I, Type II and Type III chromites (Table 1), but its composition varies in the different ultramafic massifs. Antigorite crystals in the chromitite bodies from Los Congos (n= 9) vary within the following ranges: 42.42-44.66 wt.% SiO₂, 36.04-39.33 wt.% MgO, 3.01-3.99 wt.% Al₂O₃, 1.48-3.19 wt.% FeO, and 0.68-1.39 wt.% Cr₂O₃. Antigorite from Los Guanacos (n= 12) has relatively higher MgO (37.24-41.12 wt.%) and FeO (1.77-3.30 wt.%), but relatively lower SiO₂ (39.71-43.16 wt.%), Al₂O₃ (up to 0.71 wt.%), and Cr₂O₃ (0.03-0.26 wt.%) than those from Los Congos (Appendix 2).

Chlorite (n= 8) associated with Type I chromite in chromitites from Los Guanacos (Table 1) have narrowly variable SiO₂ (29.01-31.87 wt.%), Al₂O₃ (17.45-18.36 wt.%), MgO (32.78-34.27 wt.%), and FeO (1.83-2.63 wt.%) (Appendix 2). All the analysed grains contain significant amounts of Cr₂O₃ (1.94-3.95 wt.%), but their low Fe/(Fe+Mg) ratios (up to 0.04) overlap the compositional field of clinochlore defined byHey (1954) (Appendix 2).

Clinopyroxene (n= 26) identified in Type I and Type II chromites (Table 1) has a Mg-rich augite composition (Wo_28-29 En_67-69 Fs_3-5) (Appendix 2), but its composition varies in different types of chromite. Clinopyroxene grains which are associated with Type I chromite in chromitites from Los Congos and Los Guanacos (Table 1) show similar Mg# (0.94-0.96), SiO₂ (55.29-57.19 wt.%), Al₂O₃ (2.19-4.06 wt.%) and CaO (12.96-13.89 wt.%), with relatively higher Na₂O (0.16-0.36 wt.%) and Cr₂O₃ (0.55-1.09 wt.%), and lower TiO₂ (0.10-0.25 wt.%) and MnO (0.02-0.13 wt.%) than those from Type II chromites in Los Congos (Mg# = 0.93-0.94, SiO₂ = 54.53-56.55 wt.%, Al₂O₃ = 3.50-4.96 wt.%, CaO = 13.22-13.38 wt.%, Na₂O = 0.13-0.26 wt.%, Cr₂O₃ = 0.43-0.72 wt.%, TiO₂ = 0.22-0.33 wt.%, and MnO = 0.13-0.15 wt.%) (Appendix 2).

Type III chromite prior to exsolution (Mg# = 0.44-0.50) and Type I chromites in disseminated chromitites have lower Mg# (0.42-0.52) with respect to those in massive chromitites (Mg# = 0.61-0.68), expanding between and parallel to the Fo₈₅ and Fo₉₃ contours (Table 1;Figure 4). This trend is similar to that described in non-metamorphosed chromitites from ophiolites elsewhere, where accessory chromites hosted in olivine are enriched in FeO and the olivine interstitial to chromite in chromitite bodies is enriched in MgO (Leblanc and Nicolas, 1992). It could be attributed to the Mg²⁺↔Fe²⁺ exchange between chromite and olivine at high temperatures (i.e., 1000-900 °C) during postmagmatic cooling (Irvine, 1967).Thus, a decrease of Mg# at relatively constant Cr# should be expected in the chromite grains from chromitites with lower chromite/silicate ratios. However, this hypothesis is not valid for Type I and Type III chromites prior to exsolution from Los Congos and Los Guanacos because the decrease of Mg# is associated with an increase of Cr# (from 0.47 to 0.73 in Type I chromite and from 0.54 to 0.59 in Type III chromite prior to exsolution) (Table 1;Figure 4a and4c).

The trend of Cr₂O₃ and FeO gain coupled with the MgO and Al₂O₃ loss showed in Type I and Type III chromites prior to exsolution (Table 1;Figure 4a and4d) is similar to those reported in chromitites from metamorphosed ophiolites elsewhere (e.g.,Gervillaet al., 2012;Prabhakar and Bhattacharya 2013;Ahmed and Surour, 2016;González-Jiménezet al., 2016;Kapsiotiset al., 2018;Yuet al., 2019). These chemical variations are usually characteristic of Fe²⁺-rich porous chromite rims or grains formed by the reaction of magmatic chromite with olivine and/or pyroxene during the retrograde hydrous metamorphism, in presence of SiO₂-rich and MgO-poor fluids derived from the host ultramafic rocks (e.g.,Bachet al., 2006;Frostet al., 2013;Majumdaret al., 2016;Coláset al., 2017). This hydration process produced the mass loss of primary chromite and the preferential partitioning of MgO and Al₂O₃ towards chlorite that crystallized inside the pores of Fe²⁺-rich porous chromite, and also the formation of antigorite in the matrix (e.g.,Gervillaet al., 2012;Barraet al., 2014;González-Jiménezet al., 2016;Coláset al., 2019). Interestingly, Type I chromite in disseminated chromitites is significantly more enriched in Co and V and depleted in Ni than chromite in massive chromitites (Figure 5a and5b). This suggests that a greater element mobility could take place as a result of highly effectiveness in chromite grains with lowest chromite/silicate and high fluid/rock ratios (i.e. semi-massive chromites) (Loferski, 1986;Candia and Gaspar, 1997;Proenzaet al. 1999;Merliniet al. 2009;Barraet al. 2014;Coláset al. 2017).

Figure 5
Box chart diagrams illustrating the content of divalent (a and c), trivalent and tetravalent (b and d) minor and trace elements in the different textural varieties of chromite identified in Los Congos and Los Guanacos chromitites. Elements arranged according toColáset al. (2019).

In contrast, the high contents in Fe₂O₃ in the Type I and Type III chromites prior to exsolution from Los Congos and Los Guanacos [Fe³⁺/(Fe³⁺+Fe²⁺) from 0.24 to 0.54 in Type I, and from 0.60 to 0.61 in Type III chromite prior to exsolution] (Table 1;Figure 4) are usually associated with the modification of Fe²⁺-rich porous chromite to Fe³⁺-rich non-porous chromite during the retrograde hydrous metamorphism. This process is a consequence of the exhaustion of olivine and pyroxene from the chromitites and host ultramafic rocks, and the progressive increase of ʄO₂ and SiO₂ in the fluid (i.e., more oxidizing conditions), which promoted the oxidation of Fe²⁺ and the subsequent increase of Fe³⁺ in the system (e.g.,Alt and Shanks, 1998;Bachet al., 2004,2006;Altet al., 2007). These fluids would have reacted with Fe²⁺-rich porous chromite through the pre-existing pore network, thus dissolving chlorite and precipitating magnetite in the remaining porosity (e.g.,Satsukawaet al., 2015;Coláset al., 2019;Gervillaet al., 2019). The diffusion of Fe³⁺ and Fe²⁺ through pores resulted in the development of Fe³⁺-rich non-porous chromite, obliterating the igneous chemical signature of the pre-existing chromite (e.g.,Gervillaet al., 2012;Barraet al., 2014;Coláset al., 2019;Yu et al, 2019). In this context, the lowest Fe³⁺/(Fe³⁺+ Fe²⁺) ratios (0.24-0.27) showed by Type I chromite in semi-massive chromitites and the highest Fe³⁺/(Fe³⁺+ Fe²⁺) ratios (0.60-0.61) displayed by Type III chromite prior to exsolution (Table 1;Figure 4) suggest an uneven efficiency of the alteration process, which could be significantly affected by the presence of secondary permeability in the chromitites and/or by variable fluid/rock ratios (e.g.,Loferski, 1986;Candia and Gaspar, 1997;Coláset al., 2017).

The absence of Fe²⁺-rich porous chromite and/or Fe³⁺-rich non-porous chromite rims in Type I chromites and the presence of exsolved textures in Type III chromites (Figure 3) suggest a change in the metamorphic conditions. Type I and Type III chromites prior to exsolution have albeit similar Mg# (from 0.42 to 0.68 in Type I, and from 0.44 to 0.50 in Type III chromite prior to exsolution) and Cr# (from 0.47 to 0.73 in Type I, and from 0.54 to 0.59 in Type III chromite prior to exsolution), but higher (Fe³⁺/(Fe³⁺+ Fe²⁺) ratios (from 0.24 to 0.54 in Type I, and from 0.60 to 0.61 in Type III chromite prior to exsolution) than primary chromite reported in Los Guanacos byProenzaet al. (2008) (Cr# = 0.48-0.56, Mg# = 0.61-0.75, and Fe³⁺/(Fe³⁺+ Fe²⁺) = 0.30-0.36;Figure 4). According toGonzález-Jiménezet al. (2015), the dehydration of previously hydrated chromitite bodies during the thermal prograde metamorphism may produce the reaction of Fe²⁺-rich porous and/or Fe³⁺-rich non-porous chromite with antigorite and chlorite from the pores and/or matrix. This process formed an optically homogeneous chromite with a composition close to igneous chromite -i.e., a chromite with lower Cr₂O₃ but higher MgO, Al₂O₃ and Fe₂O₃ than that formed by the retrograde hydrous metamorphism- in equilibrium with Mg-rich olivine and/or pyroxene (e.g.,Padrón-Navartaet al. 2010,2013).

It is worth noting that Type I chromites from Los Congos and Los Guanacos shows patterns of minor and trace elements with lower Ga and higher Co, Zn and Mn than those chromites from non-metamorphosed ophiolitic chromitites (i.e., chromites form back-arc and oceanic suprasubduction zone (SSZ) mantle;Figure 6a to6c). This suggests that the distribution of minor and trace elements in Type I chromite was disturbed significantly by the formation of Fe³⁺-rich non-porous chromite during retrograde hydrous metamorphism, and the subsequent re-equilibrium between those hydrated chromites and silicates during the prograde metamorphism. This is consistent with the fact that the enrichment of Co, Zn and Mn coupled with the depletion of Ga in chromite was interpreted as a result of the formation of chlorite and antigorite by the hydration of chromite during the retrograde metamorphism (Coláset al., 2014). This hypothesis is also supported by (i) the absence of Fe²⁺-rich porous and/or Fe³⁺-rich non-porous chromite grains and/or rims in Type I chromites, and (ii) with the fact that these chromite grains have albeit similar distribution of minor and trace elements than Fe³⁺-rich non-porous chromite in metamorphosed ophiolitic chromitites from the Eastern Rhodopes in Bulgaria (Figure 6d;Coláset al., 2014;González-Jiménezet al., 2015,2018).

Figure 6
MORB-normalized minor and trace element distribution in Type I chromite grains from Los Congos and Los Guanacos chromitites (a and b) compared with high-Al non-metamorphosed chromitites (c) and with metamorphosed chromitites (d). MORB normalizing values are fromPagé and Barnes (2009). Data sources for high-Al chromites from suprasubduction zone (SSZ) mantle-hosted podiform chromitites: Sartohay, China (Zhouet al., 2014); Islahiya, Doğanşehir (Akmazet al., 2014) and AntalyaIsparta (Uysalet al., 2016), Turkey; Mercedita (Coláset al., 2014) and Mayarí (Zhouet al., 2014), Cuba. Data source for back arc chromitites is from Coto, Philippines (Yao, 1999), and from Sagua de Tánamo, Cuba (González-Jiménezet al., 2015); and that of from metamorphic chromitites is from Golyamo Kamenyane (Coláset al, 2014) and Chernichevo (González-Jiménezet al. 2018), Bulgaria.

Additionally, Type I chromite in massive chromitites exhibits similar Ga, Ti, Zn, and Mn, higher Sc and Ni, and lower Co and V than that chromite in semi-massive chromitites (Table 1;Figure 5a and5b). This is consistent with the fact (i) the dehydration of previously hydrated chromitites forms Mg-rich olivine and/or pyroxene with lower Co and higher Ni than Type I chromite (e.g.,Zhanget al., 2018;Suet al., 2019), and (ii) the lowest contents of Sc and Fe₂O₃ [Fe³⁺/(Fe³⁺+ Fe²⁺) = 0.24-0.27] (Table 1;Figure 4b and4c) showed by Type I chromite in semi-massive chromitites is due to the preference of Sc for Fe³⁺-rich chromites (Wijbranset al., 2015). Hence, these observations allow us to suggest that the major, minor and trace element composition of Type I and Type III chromites prior to exsolution from Los Congos and Los Guanacos (Figure 4a to4f) was likely modified by the hydration of primary chromite during the retrograde hydrous metamorphism, and their subsequent dehydration during the prograde metamorphism.

Recent studies conducted byGonzález-Jiménezet al. (2015,2016) andColáset al. (2017) reproduced the phase relations and temperature conditions of the alteration of chromite during the retrograde hydrous metamorphism in the system Cr₂O₃-MgO-FeO-Al₂O₃-SiO₂-H₂O (CrMFASH). However, this thermodynamic approach has limitations to model Fe³⁺-rich non-porous chromite as Type I, Type II and Type III chromites prior to exsolution from Los Congos and Los Guanacos because the solution model used for Cr-spinel does not include a calibration for the magnetite (Fe₃O₄) end-member with an inverse spinel structure. Recent refinements of this thermodynamic modelling (González-Jiménezet al., 2018;Coláset al., 2019) now consider that the Fe₂O₃ content in Fe³⁺-rich non-porous chromite can be modelled as a magnetite component in the FeO-Fe₂O₃-MgO-Al₂O₃-SiO₂-H₂O (FMASH) system, thus allowing a better constrain the temperature of formation of Fe³⁺-rich non-porous chromite in chromitites from Los Congos and Los Guanacos.

Such an approach allowed us to construct a pseudosection for Type I chromite, contoured for Cr# and Mg#, in the fluid-saturated FeO-Fe₂O₃-Cr₂O₃-MgO-Al₂O₃-CaO-SiO₂-H₂O (FCrMACaSH) system (Figure 7). In our computations, we have used an initial assemblage of chromitite bulk composition of semi-massive (57:35:04:04) molar mix of chromite, antigorite, clinopyroxene and chlorite (Table 1). The composition of chromite is the Type I chromite with the lowest Fe³⁺/(Fe²⁺+Fe³⁺) ratio [(Mg_0.48Fe_0.52)(Cr_0.72Al_0.28Fe_0.08)O₄], and that of antigorite [(Mg_0.48Fe_0.52)(Cr_0.73Al_0.27Fe_0.08)O₄], clinopyroxene [(Ca_0.5Mg_1.44Fe_0.06)Si₂O₆] and chlorite [(Mg_4.68Fe_0.16)Al₂Si₃(OH)₈] is the average composition of those silicates in semi-massive chromitites including Type I chromite (Figure 4a;Table 1;Appendices 1 and2).

Our modelling shows that the maximum slope of Cr# isopleths in chromite is located between ca. 800 ºC at 20 kbar and 500 ºC at 1 kbar (Figure 7a and7b), predicting that under water-saturated conditions, primary chromite in semi-massive chromitites is not stable with olivine and pyroxene below 800 ºC. At these conditions, the stable assemblage is made up of chromite with higher Cr# and lower Mg# in equilibrium with chlorite, antigorite, brucite and magnetite (Figure 7a to7c). The model also predicts that at the P-T conditions for ocean floor metamorphism (4 kbar and 300 °C;Escayolaet al., 1996,2004) estimated for the host ultramafic rocks at Los Congos and Los Guanacos, chromite with Cr# = 1.00 and Mg# = 0.25 in semi-massive chromitite should form (green circle-1 inFigure 7b and7c). Such modelled values are in accordance with the composition of Fe³⁺-rich non-porous chromite in metamorphosed chromitites at Tehuitzingo, Loma Baya in southern Mexico (Proenzaet al., 2004;González-Jiménezet al., 2015,2017;Coláset al., 2019;Farré-de-Pabloet al., 2019), Tapo in Peru (Coláset al., 2017), and La Cabaña ultramafic massifs in Chile (Barraet al., 2014;González-Jiménezet al., 2016;Plissartet al., 2019).

Figure 7
P-T pseudosection (a) and isopleths of Cr# (b) and Mg# (c) in chromite and Mg# (d) in olivine calculated for a semi-massive chromitite (57:35:04:04 chromite:antigorite:clinopyroxene:chlorite molar proportions) in the fluid-saturated FCrMACaSH system. Green circles show the metamorphic conditions obtained byEscayolaet al. (1996,2004) in the ultramafic host rocks at Los Congos and Los Guanacos massifs. Orange rectangle show the temperature conditions estimated byColáset al. (2016) for the exsolution of Type II and Type III chromites and the pressure conditions obtained byRapelaet al. (1998) andMartinoet al. (2010) in the associated granulites. Abbreviations: Atg: antigorite, Ath: antophillyte, Chl: chlorite, Chr: chromite, Cpx: clinopyroxene, Hem: hematite, Mag: magnetite, Ol: olivine, Opx: orthopyroxene (Whitney and Evans, 2010).

The occurrence of brucite along with chlorite and antigorite below 500 ºC and 20 kbar further support that ocean floor metamorphism took place under SiO₂-poor water-saturated conditions (Coláset al., 2017). Furthermore, the predicted amount of magnetite is <2.58 vol.%, which can be explained as a result of olivine runoff and the oxidation of Fe²⁺ to Fe³⁺ in the solutions (e.g.,Bachet al. 2004;Altet al. 2007). This oxidizing fluids circulated through the network of pores in the Fe²⁺-rich porous chromite, thus dissolving chlorite and diffusing Fe²⁺ and Fe³⁺ into the chromite lattice and converts it to Fe³⁺-rich non-porous chromite (e.g.,Gervillaet al., 2012;Barraet al., 2014;Coláset al., 2014,2017,2019;González-Jiménezet al., 2016,2018).

However, Type I chromite in semi-massive chromitites have lower Cr# (0.60-0.73) and higher Mg# (0.48-0.55) and Fe³⁺/(Fe³⁺+Fe³⁺) (0.24-0.27) than modelled Fe³⁺-rich non-porous chromite (Figure 4a to4c), which suggests a change in the metamorphic conditions (Figure 7a). According toFigure 7a to7c, an increase in Al₂O₃, MgO and Fe₂O₃ could be related to the breakdown of chlorite and antigorite if a prograde metamorphism affects the chromitite bodies. Hence, at the P-T conditions of the metamorphic peak (10 kbar and 800 °C as ofEscayolaet al., 1996,2004) (green circle-2 inFigure 7) Fe³⁺-rich non-porous chromite reacts with chlorite+antigorite+brucite in the matrix. Such reaction of dehydration leads to the formation of homogeneous Fe³⁺-rich chromite with Cr# = 0.65, Mg# = 0.57 (i.e., Type I chromite), olivine, orthopyroxene and clinopyroxene (Figure 7a to7c). Clinopyroxene is very rare in non-metamorphosed ophiolitic chromitites, although diopside is found in Al-rich ophiolitic chromitites from the Mayarí-Baracoa Ophiolitic Belt in Cuba (Proenzaet al., 1998,1999;Pujol-Solàet al., 2018), which are spatially associated with gabbro dikes and sills. However, our thermodynamic model also predicts that olivine with Fo₉₃ (Figure 7d) and Mg-rich clinopyroxene (the predicted Mg# is 0.97;Figure 7a) were in equilibrium with a chromite with the composition of Type I chromite in semi-massive chromitites (Figures 4a to4c and7).

Significantly, the presence of exsolved textures in Type II and Type III chromites indicate a late cooling from granulite facies metamorphism to lower P-T conditions (Figures 3c,3d and7a). Contrary to the petrographic observations, our thermodynamic model cannot predict the exsolution of homogeneous Fe³⁺-rich non-porous chromite to spinel-rich and magnetite-rich phases in Type II and Type III chromites. However, at the minimum temperature estimated for the exsolution of such chromites (ca. 600 ºC;Coláset al., 2016) and the lowest pressures recorded in the associated granulites (from 4 ± 0.5 to 6.2 ± 1 kbar;Otamendiet al., 1999,2004,2005;Rapelaet al., 1998;Martinoet al., 2010) (orange rectangle-3 inFigure 7), chromite with higher Cr# (0.94-0.98) and lower Mg# (0.36-0.38) than Type I chromite (Figure 7b and7c) is in equilibrium with olivine, clinopyroxene, chlorite and antophillyte (Figure 7a). The presence of antigorite in the studied chromitites (Figure 3) vs. the predicted occurrence of olivine, clinopyroxene and chlorite by means of thermodynamic modelling (Figure 7), suggests that a final hydration of silicates did not fully progress at amphibolite facies conditions. Further, anthophyllite is not present in the studied chromitite samples (Figure 3), but it was described in ultramafic host rocks to the chromitite bodies at Los Guanacos (Escayolaet al., 1996,2004). Such occurrence can be explained as a consequence of higher rates of reaction between the infiltrating fluids and silicates in chromitites than between these fluids and the host ultramafic rocks. Therefore, these observations suggest (i) Type I, Type II and Type III chromites experienced high-T metamorphism at granulite facies condition after the formation of Fe³⁺-rich non-porous chromite by the hydrous metamorphic reaction that generated Fe²⁺-rich porous chromite and magnetite above 650-800 ºC (Figure 7a to7d;Coláset al., 2019); and (ii) the ultimate mineral assemblage preserved in Type I chromite, as well as the exsolution of Type II and Type III chromites started once the ultramafic body already entered within the P-T field of amphibolite facies (Figure 7).

Coláset al. (2016) suggested that the variety of textures in Los Congos and Los Guanacos chromitites is mainly controlled by the Cr₂O₃ and Fe₂O₃ content in chromite (Table 1;Figure 4a to4f). Under the assumption that Type II and Type III chromites had a similar composition prior to exsolution, such a process could be represented on an ideal phase diagram of chromite (Figure 8). In this diagram, the composition in major elements of Type I chromite and spinel-rich and magnetite-rich phases in Type II and Type III chromites may be represented by an hypothetical solvus on the (Mg_0.5Fe²⁺_0.5)Cr₂O₄-MgAl₂O₄-(Mg_0.5Fe²⁺_0.5)Fe³⁺₂O₄ plane, included in the Cr-spinel prism (Figure 8a and8b) (Tamura and Arai, 2004). As shown inFigure 8b to8d, the extension of the solvus is more restricted in chromites with high Cr₂O₃, than in those with low Cr₂O₃. Therefore, Type I chromite with higher Cr₂O₃ but lower Fe₂O₃ contents (Cr# = 0.47-0.73 and Fe³⁺/(Fe^{2 +}+Fe³⁺) = 0.24-0.54) than Type II and Type III chromites (Cr# = 0.54-0.59 and Fe³⁺/(Fe²⁺+Fe³⁺) = 0.60-0.61) (Table 1;Figure 4a to4f), should be expected to plot out of the solvus at >600 ºC, which is the blocking temperature calculated by Coláset al. (2016) -i.e., the minimum possible temperature for exsolution (Figure 8b-c). However, Type II and Type III chromites prior to exsolution (i.e., 800 ºC, the metamorphic peak conditions estimated byEscayolaet al., 1996,2004) would be found in the miscibility gap during the cooling from granulite to amphibolite facies, as recorded in the host granulites (Escayolaet al., 1996,2004;Rapelaet al., 1998;Otamendiet al., 1999,2004,2005;Martinoet al., 2010).

InFigure 8d is also shown that Type III chromite prior to exsolution has, in a binary system, an intermediate composition of Fe₂O₃ (Fe³⁺/(Fe²⁺+Fe³⁺) = 0.60-0.61; (Table 1 andFigure 4d to4f), which lies inside the spinoid at the blocking temperature. Thus, the symplectitic texture in Type III chromite (Figure 3c) could be effectively formed by phase separation of spinel-rich and magnetite-rich phases into the spinoid, as a consequence of a continuous and slow exsolution process through time (e.g.,Kingeryet al., 1976;Tamura and Arai, 2004). This exsolution process also produced the enrichment in Co, Zn, Sc, and Ga in the spinel-rich phase and the enrichment in Mn, Ni, V and Ti in the magnetite-rich phase in Type III chromite (Figures 5a and5b). We have assumed that the original composition of Type II chromite was analogue to Type III chromite prior to exsolution (Mg# = 0.44-0.50, Cr# = 0.54-0.59, and Fe³⁺/(Fe³⁺+Fe²⁺ = 0.60-0.61;Figure 4a to4f). This implies that Type II chromite should also plot inside the spinoid at the blocking temperature (Figure 8d). However, the presence of a magnetite-rich phase in Type II chromite as irregular blebs or forming very fine lamellae within the spinel-rich phase (Figure 3d) indicates a discontinuous and relatively fast exsolution process.

Figure 8
Schematic illustration for interpretation of the different textural varieties of chromite identified in Los Congos and Los Guanacos chromitites according to their composition. (a) Cr-spinel prism including the approximate plane for Type I, Type III and Type III chromite compositions. (b) Hypothetical solvus on the Cr-Al-Fe³⁺ plane calculated bySack and Ghiorso (1991a). Al³⁺-Fe³⁺ planes that cut through the solvus of (b) at different high and low Cr³⁺ contents equivalent to Type I (c), Type II and Type III (d) chromite compositions (modified fromTamura and Arai, 2004).

The random distribution of magnetite-rich phase blebs along the rims or forming linear strings in Type II chromite grains (Figure 3d) suggests that such blebs are due to a slow cooling, probably at high temperatures (i.e., the metamorphic peak conditions at 800 °C;Escayolaet al., 1996,2004) in which a relatively effective and fast intracrystalline diffusion of cations is favoured. However, the crystallographic control of the magnetite-rich phase lamellae implies that they could form at lower temperatures (i.e., the blocking temperature at ca. 600 ºC;Coláset al., 2016) because the intracrystalline diffusion of cations would be less efficient and slower, which would mean that exsolved phase could only concentrate along crystallographic planes. These two exsolution processes are analogue to those reported in chromites from the Chilas (Pakistan;Janet al., 1992) and the Eastern Desert (Egypt;Ahmedet al., 2008) magmatic arcs, and from the Iwanai-dake peridotite complex (Japan;Tamura and Arai, 2004,2005). Therefore, the different textures in Type I, Type II and Type III chromites in chromitite bodies from Los Congos and Los Guanacos are the consequence of: i) gradual variations in the Cr₂O₃ and Fe₂O₃ contents during the dehydration of previously hydrated chromitites by prograde metamorphism within the granulite facies, and ii) the cooling rates during the retrograde overprint within the amphibolite facies.

*Corresponding author: (V. Colás)vcolas86@gmail.com