Seven volcanoes have been recognized so far in the SMVF, which covers an area of ∼400 km². Three of them have known ages, and four are of unknown age. The former are 1) San Diego volcano (Figure 2A), a maar formed 20 ka ago and a lava dome to the NE of the maar, which records the last stage of the eruption (Borrero et al., 2017). The maar has an elongated crater (2.1 × 1.5 km in diameter) and hosts a ∼50-m-deep lake; its deposits are recognized as dilute pyroclastic density currents formed by phreatic and phreatomagmatic activity, distributed mainly toward the NE side of the volcano (Borrero et al., 2017). 2) El Escondido volcano (Figure 2B), a 38-ka tuff cone (Sánchez-Torres et al., 2019); it presents a semi-circular crater (∼1.7 km diameter) and flanks with slopes between 10 and 15°. El Escondido deposits are distributed toward the N and E sides of the volcano and records a spectrum of concentrated and dilute pyroclastic density currents and secondary deposits formed by both magmatic and phreatomagmatic activities (Sánchez-Torres et al., 2019). 3) Pela Huevos volcano (Figure 2C), a 154-ka lava dome, 250 m high with an elongated conical shape (Rueda-Gutiérrez, 2019); it is located on the SE limit of El Escondido volcano and was disrupted by the eruption that formed the El Escondido volcano (Sánchez-Torres et al., 2019). The volcanoes of unknown age are 4) Piamonte volcano (Figure 2D), a ∼220-m-high lava dome with a conical shape, flat at the top, 5) Morrón volcano (Figure 2E), a 350-m-high lava dome with a conical morphology, elongated toward E, 6) Guadalupe volcano (Figure 2F), a lava dome with a conical morphology and volcaniclastic deposits (block and ash pyroclastic flow deposits) around it, and 7) Norcasia volcano (Figure 2G) defined by volcaniclastic products without a clear morphology.

Nine samples were collected from the seven studied volcanoes (Table 1). Thin sections were made in Teclab laboratories (Colombia). The petrographic analysis was carried out by point counting in each section, using a Nikon Eclipse E200 petrographic microscope. The size of phenocrysts was defined as >0.5 mm, that of microphenocrysts between 0.5 and 0.05 mm, and that of microlites (groundmass) <0.05 mm (González, 2008). Mineral chemistry analyses (591) were performed using a JXA-8530F field emission electron probe microanalyzer (FE-EPMA) equipped with five wavelength-dispersive spectrometers, at the Facility for Analysis, Characterization, Testing, and Simulation (FACTS) in Nanyang Technological University (NTU), Singapore. Point analyses were acquired using a focused beam at a probe current of 20 nA and an accelerating voltage of 15 kV for all minerals. Current was reduced to 10 nA and defocused beam diameters of 3 and 10 μm for analysis of plagioclase and glass, respectively. The results were quantified using well-characterized natural and synthetic external calibration standards and a ZAF matrix correction procedure. The standards used were T1-G for Si and Al, P&H block Geo MkII for Ca (wollastonite), Ti (rutile), Mg (periclase), K (orthoclase), Fe (specularite), Mn (rhodonite), and P (apatite), and Astimex block MINM25-53 for Na (albite). The oxygen content was assumed from cation abundance, with all iron present as Fe²⁺. Error on repeat analysis of standard reference materials was <1% of measured values. Kα X-ray lines were monitored for 20–60 s for each element, depending on expected concentrations with the exception of Na Kα, which was monitored for only 10 s. Background measurements were performed on either side of each peak position for combined counting times equaling the corresponding peak counting times. Measured peak and background positions were found to be free of interferences within the sample and standard matrices. The analyses were carried out on plagioclase, olivine, pyroxene, amphibole, biotite, Fe–Ti oxides, and glassy groundmass. The analyses obtained in the microprobe correspond to the total sum of oxides >98 wt.% for anhydrous phases and >95 wt.% for hydrous phases. In the cases of the analyses in biotite and Fe–Ti oxides, the analyses with a total >93 wt.% were used since the majority were shown with low values by the microprobe. The cations per formula unit were calculated using the weight percentage obtained by the microprobe and molecular weight of each oxide, taking into account the amount of oxygen present in the chemical formula of each mineral. For Fe–Ti oxides, the total iron content was recalculated by the Carmichael (1967) method to obtain de FeO and Fe₂O₃ contents.

The nine samples were analyzed for whole-rock chemistry at Actlabs laboratories (Colombia and Canada). Major elements were analyzed using the ICP-OES (Inductively Coupled Plasma Optical Emission Spectrometry) technique, while trace elements were analyzed using the ICP-MS (Inductively Coupled Plasma Mass Spectrometry) technique. The samples were run for major and selected trace elements on a combination simultaneous/sequential Thermo Jarrell-Ash ENVIRO II ICP for a Varian Vista 735 ICP. Calibration was performed using seven prepared USGS- and CANMET-certified reference materials. For the lithium metaborate/tetraborate fusion—ICP/MS portion of the analysis, the samples were fused, diluted, and analyzed by Perkin Elmer Sciex ELAN 6000, 6100, or 9000 ICP/MS. Loss-on-Ignition (LoI) was calculated from the weighed samples, and iron was reported as total Fe₂O₃. >100x detection limit ±5% for major oxides and >100x detection limit ±100% for minor and trace elements. The sample IIES-V-007 from the Guadalupe volcano yielded a high LoI value (∼7 wt.%), and therefore it was not used for major whole-rock and geothermobarometric analysis.

Three samples from Guadalupe, Piamonte, and Morrón volcanoes were collected for K/Ar geochronology. The samples were crushed and then sieved to separate the 0.5-mm fraction. From these, 20 g of groundmass was extracted by hand picking in order to date the cooling time of the magma. This avoids overestimating ages by mixing the crystals that would introduce older crystallization ages. Freshness of the groundmass was defined not only by looking at the glass using the microscope but also by using the rock samples with low LoI values as indicated by the whole-rock chemistry results. The dating analyses were carried out in ActLabs (Canada) after a further separation of 2 g of fresh groundmass, where the procedure used is described as follows: aliquots of the samples were weighed into an Al container, loaded into the sample system of the extraction unit, and degassed at 100 °C for 2 days to remove the surface gases. Argon was extracted from the sample in a double vacuum furnace at 1,700 °C. The determination of radiogenic argon content was carried out twice on a MI-1201 IG mass spectrometer by the isotope dilution method using ³⁸Ar as a spike, which was introduced to the sample system prior to each extraction. The extracted gases were cleaned up in a two-step purification system. Then, pure Ar was introduced into a custom-built magnetic sector mass spectrometer (Reynolds type). Two globally accepted standards (P-207 muscovite and 1/65 “Asia” rhyolite matrices) were measured for ³⁸Ar spike calibration. For age calculations, the international values of constants were used as follows: λK = 0.581 × 10⁻¹⁰ y⁻¹, λβ- = 4.962 × 10⁻¹⁰ y⁻¹, and ⁴⁰K = 0.01167 (аt.%). Calculated errors were 2σ.

A paleosol located under volcaniclastic products from Norcasia volcano was also selected for ¹⁴C dating. The analysis was carried out at the Center Radiochronology Laboratories, Université Laval (Canada) by the AMS (Accelerator Mass Spectrometry) method. The sample was chemically cleaned, burned, and transformed into CO₂, followed by oxidation and reduction to graphite. The graphite produced was pressed into a target for AMS measurement.

The studied rocks (Table 1) are characterized by their porphyritic texture (30–45 vol.% crystals). The groundmass is glassy (holohyaline) in El Escondido and Morrón volcanoes; glassy with microlites (hypocrystalline) in Pela Huevos, Piamonte, and Norcasia volcanoes; and microcrystalline and cryptocrystalline in Guadalupe and San Diego volcanoes (Table 1). Plagioclase is the most abundant mineral phase in five volcanoes (San Diego, El Escondido, Piamonte, Pela Huevos, and Norcasia; Table 1), with at least three populations 1) “clean” crystals, 2) coarse sieve texture crystals (Figure 3A), and 3) fine or dusty sieve texture crystals (Figure 3B); all populations of plagioclase present twins, normal, reverse, and oscillatory zonation. Of the studied volcanoes, San Diego does not display sieve textures of any kind, and Guadalupe only presents fine sieve textures. Amphibole is the most abundant mineral phase in the other two volcanoes (Morrón and Guadalupe), and it is not present in the San Diego volcano (Table 1). Amphibole in El Escondido and Piamonte volcanoes is green and strongly pleochroic (Figure 3C), whereas in the other volcanoes, it is dark brown and highly oxidized on the whole crystal and/or on the rims (Figure 3D). The two types of amphibole do not coexist together. Amphiboles with disequilibrium textures such as oxidation rims (Figures 3C–E) and resorption (Figure 3D) are observed in Pela Huevos, Morrón, Guadalupe, Norcasia, and Piamonte volcanoes; the latter only exhibit oxidation rims. Biotite was observed only in El Escondido, Morrón, San Diego, and Guadalupe volcanoes. In San Diego (Figure 3F), the biotite is reddish brown and highly oxidized on the whole crystals, while in Guadalupe, some of the biotite crystals show oxidation rims (Figure 3E). Quartz (Figure 3F) is only present in El Escondido and San Diego volcanoes (Table 1). Pyroxene (clinopyroxene and orthopyroxene) appear in Pela Huevos and Norcasia volcanoes (Figure 3G; Table 1), while olivine is present in Pela Huevos and Guadalupe volcanoes (Table 1), commonly surrounded by brown amphibole crystals (Figure 3H). Accessory minerals such as Fe–Ti oxides (<1 vol.%) appear in all rock samples. Glomerocrysts (Figure 3G) of different mineral associations are also observed (Figure 3G; Table 1).

Plagioclase composition in the studied rocks varies from oligoclase to bytownite. San Diego volcano hosts crystals with a low and narrow An range (An_26–32), compared (Figure 4A) with the other volcanoes —El Escondido (An_27–57), Pela Huevos (An_33–73), Guadalupe (An_30–58), Norcasia (An_36–68), and Piamonte (An_42–74). Some crystals in El Escondido, Pela Huevos, Norcasia, Piamonte, and Guadalupe volcanoes show oscillatory compositional zonation (Figure 5A), and some others in El Escondido and Pela Huevos volcanoes show reverse zonation (Figure 5B). Amphiboles are calcic —tschermakite and magnesio-hastingsite— with only magnesio-hornblende crystals in Pela Huevos and Guadalupe volcanoes (Figure 4B), without significant differences between both types of amphibole crystals. Normal (i.e., decreasing Mg# toward the rim; Figure 5C) and reverse (i.e., increasing Mg# toward the rim; Figure 5D) compositional zonation are common in both amphibole crystals. Biotite is magnesiobiotite in both San Diego and Guadalupe (Figure 4C). Unfortunately, we did not get accurate measurements of biotite in El Escondido and Morrón volcanoes. Pyroxene from Norcasia volcano corresponds mainly to enstatite (Wo_2–3, En_70–76, Fs_21–28), diopside, and augite (Wo_44–45, En_41–42, Fs_13–15, and Wo_42–44, En_46–47, Fs_10–11, respectively) (Figure 4D). Olivine corresponds to Fo_82–88 in Pela Huevos volcano and Fo_81–83 in Guadalupe volcano (Figure 4E). Fe–Ti oxides (Figure 4F) are magnetite (Morrón: Usp_6–89 Mag_94–11 and Guadalupe: Usp_4–85 Mag_96–15) and ilmenite (Morrón: Ilm_78–92 Hem_22–8, and Guadalupe: Ilm_61–91 Hem_39–9). Table 2 shows representative analyses of all mineral phases; all mineral chemistry data can be found in Supplementary Material S1.

Whole-rock chemistry analysis shows that the rocks from the SMVF range between andesite and dacite (Figure 6A), with those from San Diego (SiO₂ ∼69 wt.%) and El Escondido (SiO₂ ∼66 wt.%) being the most evolved of the group and those from Norcasia, Morrón, and Guadalupe the least evolved (SiO₂ ∼60 wt.%). All samples are of a calc-alkaline affinity, with medium potassium content (Figure 6B). The studied rocks show similar behavior of the incompatible trace elements, with a slight enrichment in LILE with respect to HFSE (Figure 6C). Positive anomalies of Ba, U, K, Pb, Sr, and Nd and negative anomalies of Th, Nb, Ti, and P are observed, with the exception of the San Diego volcano, which lacks a Ti anomaly (Figure 6C). Rare earth elements (REE), normalized to chondrite, show a strong LREE enrichment with respect to HREE. Of all the volcanoes, the samples from Morrón and San Diego are the most and least enriched, respectively (Figure 6D). Incompatible element ratio diagrams (Nb/Th vs. Nb/Zr and Zr/Y vs. Nb/Th) show similar ratios for all the samples, with only San Diego as an outlier (Figure 6E). Binary diagrams of major and trace elements vs. SiO₂ show similar compositions between San Diego and El Escondido volcanoes, between Piamonte and Pela Huevos volcanoes, and between Morrón, Guadalupe, and Norcasia volcanoes (Figure 7). A negative correlation between TiO₂, Al₂O₃, MgO, CaO, Sr, V, and Sc with respect to SiO₂ can be recognized and a positive correlation with K₂O (Figure 7). Whole-rock chemistry data from all samples are reported in Table 3.

The K/Ar analysis yields eruptive ages of 0.05 ± 0.04 Ma for the Guadalupe volcano, 0.46 ± 0.04 Ma for the Piamonte volcano, and 1.32 ± 0.06 Ma for the Morrón volcano, while the ¹⁴C analysis yielded an age of 16,919 ± 220 years Cal BP for the Norcasia volcano. These ages together with the ages already known for San Diego (20,056 ± 96 years Cal BP, ¹⁴C; Borrero et al., 2017), El Escondido (38,553 ± 596 years Cal BP, ¹⁴C; Sánchez-Torres et al., 2019), and Pela Huevos (153,700 ± 38,500 years, ⁴⁰Ar-³⁹Ar on amphibole; Rueda-Gutierrez, 2019) volcanoes (Table 4) indicate that the Samaná field is a long-lived (∼1.3 Ma—17 ka) monogenetic volcanic field. In addition, ages indicate that the area is volcanogenic and therefore, the limit previously considered non-volcanogenic due to flat subduction angles begins at ∼6°N and not at 5° (c.f. Londoño et al., 2020).

The wide textural variation identified in the volcanoes forming the SMVF allows us to infer physicochemical processes that the distinct magma batches were subject to, from the source to surface. The existence of both monomineralic and polymineralic glomeroporphyritic texture is indicative of convective movements in the magma (Hogan, 1993; Jeffery et al., 2013), suggesting stagnation zones linked to crystallization, as the magma ascended. This occurred in all magma batches as both glomeroporphyritic texture types were observed in all analyzed rocks (Table 1). Changes in the intrinsic parameters of magma (i.e., pressure, temperature and/or water content in the melt) produced disequilibrium between the crystals and the liquid, and therefore features such as zonation, resorption, and oxidation rims were found (c.f. Best, 2003). Reverse zonation in plagioclase crystals from El Escondido and Pela Huevos volcanoes may be associated with crystals that grew in slowly cooled melts or in melts with high concentrations of volatile components (Loomis, 1982). The oscillatory zonation, which is very common in El Escondido, Pela Huevos, Guadalupe, and Norcasia, is associated with kinetic effects and/or convective movements of the magma within the magmatic reservoir (Pearce & Kolisnik, 1990; Berlo et al., 2007; Shcherbakov et al., 2010; Viccaro et al., 2010). Resorption textures (Guadalupe, Pela Huevos, and Morrón volcanoes) and oxidation rims in amphibole (Piamonte, Pela Huevos, Morrón, Guadalupe, and Norcasia volcanoes) and biotite crystals (Guadalupe and Morrón volcanoes) probably occurred due to the loss of water in the melt caused by decompression during magma ascent (Gill, 1981; Rutherford & Hill, 1993; Ridolfi et al., 2008) and/or the dehydrogenation of the melt (Feeley & Sharp, 1996). Coarse sieve texture in plagioclase crystals from El Escondido, Piamonte, Pela Huevos, Morrón, and Norcasia volcanoes is associated with disequilibrium due to decompression and fast ascent rates that produced dissolution in the crystals (Nelson & Montana, 1992; Monfaredi et al., 2009; Viccaro et al., 2010; Viccaro et al., 2012), while fine sieve texture in all volcanoes (but San Diego) is formed by partial melting of the crystal due to the reaction linked to the influx of magma (i.e., magma recharge or mixing); therefore, the clean rims that surround the dusty zone are interpreted as being formed by overgrowth at a post-mixing stage (Tsuchiyama, 1985; Viccaro et al., 2010; Viccaro et al., 2012).

Taking the aforementioned characteristics, we propose that, at the SMVF, the magma that fed the volcanoes was affected by sudden changes in pressure, temperature, melt water content, and convective movements. However, the San Diego volcano is the most stable system in relation to the other volcanoes due to little evidence of disequilibrium of its mineral phases. In summary, the textural analysis of SMVF volcanoes suggests that the magmas had periods of stagnation before eruption. During these periods, convective movement of the magma, followed by decrease of pressure, gave rise to different degrees of resorption and disequilibrium processes in the different mineral phases. Later, the ascent of magma with degassing produced destabilization and oxidation of the previously formed crystals; this process was continuous to shallow levels.

Two types of geothermobarometers were applied to obtain crystallization conditions of the main mineral phases identified in the volcanic products from the SMVF 1) based on mineral–liquid chemical equilibrium, which involved olivine, clinopyroxene, orthopyroxene, and plagioclase mineral phases; 2) based on mineral composition, which involved amphibole and Fe–Ti oxides. For the mineral–liquid method, the composition of the liquid was assumed to be the composition of the whole rock for the olivine and pyroxene phases, while for the plagioclase, the composition of the liquid was assumed to be the glass composition based on the partition coefficient (K_D).

Olivine crystals were not found in equilibrium (K_{D Fe-Mg}: 0.27 ± 0.03), which in turn suggests that these crystals correspond to antecrysts or xenocrysts (c.f. Jerram and Martin, 2008; Zellmer, 2021). For clinopyroxene (K_{D Fe-Mg}: 0.28 ± 0.08) and orthopyroxene (K_{D Fe-Mg}: 0.29 ± 0.06) in equilibrium, the geothermobarometers applied were equations 32d and 32c (clinopyroxene) and 28a and 29b (orthopyroxene) of Putirka (2008). Three out of four clinopyroxene crystal analyses and one out of two orthopyroxene crystal analyses met the equilibrium test. For plagioclase, equations 24a, 25a, and 25b of Putirka (2008) were applied to 212 out of 232 crystal analyses, which were the ones that met the equilibrium test, based on a proxy of ∼70 wt.% of SiO₂ liquid and a K_D (An-Ab) of 0.10 ± 0.05 for T <1,050°C. This liquid composition coincides with the glass composition of the Norcasia volcano and therefore was used as the proxy for all volcanoes. In addition, the plagioclase geothermobarometer requires input parameters such as pressure and H₂O; the former was estimated from the clinopyroxene crystallization (0.7 GPa) (c.f. Putirka, 2008) and the latter defined as 3.2 wt.% from the glass composition results. All the plagioclase analyses were applied exclusively to “clean” crystals, with the exception of the Guadalupe volcano, where valid microprobe data (i.e., the total sum of major oxides >98 wt.%) were obtained for crystals with sieve texture only. The results after applying these mineral–liquid chemical geothermobarometers indicate that clinopyroxene crystallized at 1,194–1,165°C and 0.9–0.7 GPa, orthopyroxene crystallized at 1,148°C and 0.6 GPa, and plagioclase crystallized at 943–891°C, 0.8–0.1 GPa, and 4.9–2.9 wt.% H₂O (Table 5; Supplementary Material S2). The range of the calculated temperatures is narrow and similar between volcanoes, whereas the pressure shows relatively wider ranges but still relative similar among the volcanoes (Figure 8). Perhaps the only exception is the pressure range for the plagioclase phase of the San Diego volcano, which is noticeably higher (>0.5 GPa) than the other volcanoes. All data obtained including K_D, T (°C), P (GPa), H₂O (wt.%), and log fO ₂ are included in Supplementary Material S2 and summarized in Table 5.

A geothermobarometer based solely on the composition of amphibole was evaluated following Ridolfi et al. (2010) for fO ₂ and Ridolfi & Renzulli (2012) for T, P, and H₂O melt. Thus, 51 analyses were assessed yielding crystallization conditions of 987–810°C and 0.8–0.2 GPa (Figures 8, 9A). H₂O and fO ₂ were also obtained, which yielded values of 12.2–4.3 wt.% H₂O (Figure 9B) and −8.6 to −11.5 log fO ₂ , respectively,(Figure 9C). Data for each volcano are discriminated in Table 5. The composition of Fe–Ti oxides also allowed us to determine temperature and fO ₂ , following the geothermobarometer of Andersen & Lindsley (1985) through the ILMAT Excel worksheet (Lepage, 2003) based on the equilibrium between the coexistence of ilmenite and magnetite phases proposed by Bacon & Hirschman (1988). Using the pair of phases that met the equilibrium test, the results indicated crystallization conditions of 871–687°C and −11.9 to −15.8 log fO ₂ (Figure 9C) (Table 5; Supplementary Material S2).

The depths at which the different mineral phases crystallized (Table 5) were estimated by using the results of the pressure calculations, following White (2013): density × gravity = pressure/depth, and assuming a density of 2.7 g/cm³ (c.f. Lucassen et al., 2001). Based on this, we estimate that diopside and augite pyroxene crystallized between 33 and 28 km and enstatite pyroxene at 21 km. Plagioclase crystallization occurred between 31 and 5 km. For amphiboles, magnesio-hastingsite crystallized between 31 and 9 km, tschermakite crystallized between 31 and 7 km, and magnesio-hornblende crystallized at 9 km. The Supplementary Material S2 shows discriminated data for each geothermobarometer.

The compositional characteristics of the SMVF (i.e., calc-alkaline affinity, behavior of trace and REE elements) are typical of magmas subduction-related. The enrichment of K, U, and Ba and the negative anomalies of Nb, Ta, and Ti (Figure 6C) are typical of volcanic arc rocks and represent processes of fractionation of Fe–Ti oxides and crustal contamination and influence of subduction fluids in the partial fusion of the mantle wedge (Pearce, 1983; Rollinson, 1993; Best, 2003; White, 2013). Incompatible element ratio diagrams (Figure 6E) indicate that the volcanoes of the SMVF are likely fed by the same magmatic source, with the exception of the San Diego volcano, whose source chemical characteristics are slightly different.

At least seven monogenetic volcanic edifices are part of the northernmost volcanism in the Andean chain. This volcanism is not only long-lived but also potentially active. Recognizing this volcanism is important because the area has been commonly considered non-volcanogenic linked to a flat subduction. This volcanism also sheds light onto magmatic evolution associated with evolved monogenetic volcanism, which is common, although poorly known in the literature, in other subduction zones. In addition, this study gives insights into how the magma evolution of monogenetic fields can be more complex than that given by individual batches of magma reaching the surface uninterrupted, as is normally described for monogenetic volcanic fields of more mafic compositions.

As discussed earlier, the magmatic evolution of the SMVF reveals that fractional crystallization is the major differentiation process that magmas underwent during their ascent to the surface; this fractionation took place during multiple stagnation zones as evidenced by the geothermobarometric calculations. In addition to the process of fractional crystallization, (e.g., disequilibrium textures linked to increases of temperature), magma recharge was also evidenced not only at the source but also at different stagnation levels. The interplay of these two major processes is responsible for the range of compositions displayed by the volcanoes of the SMVF. Taking this into account, we propose the following model of magma evolution: The magma that feeds the SMVF originates by the fusion of the mantle wedge, caused by the subduction of the Nazca plate under the South American plate. The magma then rises until it stagnates at crustal levels around 20–35 km depth, according to the magmatic accumulation zone proposed by Londono (2016), using geophysical methods such as regional 3D tomography of seismic velocity. This zone is also assumed to be the main magmatic supply for the SMVF volcanoes (Murcia et al., 2019). Our geothermobarometric calculations indicate that the main mineral phases in these volcanoes form at depths that coincide with this magmatic accumulation zone (Figure 11). Diopside and augite crystallize between 33 and 28 km, and enstatite crystallizes at 21 km, if the values are extrapolated from those obtained from Norcasia crystals. These crystals form monomineralic glomeroporphyritic texture in both Norcasia and Pela Huevos volcanoes. Plagioclase crystallization occurs between 31 and 5 km, indicating that it starts to form within the accumulation zone up to shallow levels of the crust. These crystals exhibit disequilibrium textures such as zonation and sieve crystals, which suggest convective movement and physicochemical variations within the magma during ascent. Likewise, the crystallization of amphibole starts at the magma accumulation zone (31 km) and continues up to 7 km, with magnesio-hastingsite crystallizing between 31 and 9 km, tschermakite between 31 and 7 km, and magnesio-hornblende at 9 km. The last phases to crystallize are biotite at <800°C (c.f. Castro & Dingwell, 2009), identified in the San Diego, El Escondido, Morrón, and Guadalupe volcanoes, quartz observed in El Escondido and San Diego volcanoes, and microphenocrysts of Fe–Ti oxides in all volcanoes (Figure 11). In addition to the described sequence of crystallization, evidence of magma recharge in the accumulation zone is also observed. For example, the incorporation of olivine antecrysts in the magmas of Pela Huevos and Guadalupe volcanoes and reverse zonation were observed in amphiboles of El Escondido, Pela Huevos, Guadalupe Norcasia, and Morrón volcanoes (Figure 11). Similarly, oxidation rims were observed in amphibole crystals in rocks from Piamonte, Pela Huevos, Guadalupe, and Norcasia volcanoes (Figure 11). After eruption, the rapid cooling produced crystallization of microlites and solidification of the melt. As evidenced in the volcanic edifices, the magma effusively reached the surface to form the Pela Huevos, Piamonte, Morrón, and Guadalupe volcanoes and explosively to form El Escondido and San Diego volcanoes, the latter associated with the presence of water on the surface (Figure 11).