The CO₂ efflux values measured at Cuicocha volcanic lake ranged between below the detection limit of the instrument (<0.5 g⋅m⁻²·d⁻¹) to 695 g·m⁻²·d⁻¹, with an average value of 54 g⋅m⁻²·d⁻¹ for the four surveys (Supplementary Table S1 in Supplementary Material). The surface water temperature presented a range between 14.5 and 17.3°C (average of 15.9°C). The pH value of the water was slightly basic, ranging between 7.17 and 9.02 (average of 8.2) and EC ∼700 μS cm⁻¹. No significant spatial variation of EC was observed at 30 cm depth. The water temperature values are ∼2°C higher in the bubbling zone, while the pH is ∼0.4 units lower with respect to the rest of the lake.

To check for the presence of overlapping log-normal populations of the efflux data, we applied the probability-plot technique (Sinclair, 1974) to the entire CO₂ efflux data (cumulative percentile frequencies versus class intervals). The inflection point of the curve allows the threshold value between different populations to be distinguished. The descriptive statistics summary of CO₂ efflux values partitioned populations at each survey are shown in Supplementary Table S1 (Supplementary Material). The result of the statistical-graphic analysis of CO₂ efflux data (Figures 2A–D) showed different populations for each survey: population I (background) with values from 6.1 to 58.0 g⋅m⁻²·d⁻¹ (average 15.9 g⋅m⁻²·d⁻¹) and represented between 29.6 and 53.8% of the total data (average 54%). Population III (anomalous or peak), which presented a range from 4.8 to 9.1% (average 6.9%) of the total data, showed values between 27.4 and 326 g⋅m⁻²·d⁻¹ (average 59 g⋅m⁻²·d⁻¹). The rest of the cumulative probability (population II) corresponding to the mixing of two log-normal populations (background and peak) is not considered as a product of a different source or mechanism, but rather the mixture of the previous ones.

Figure 3 shows the spatial distribution maps of CO₂ efflux at the Cuicocha volcanic lake for the period 2012–2018. Supplementary Figure 1S in Supplementary Material shows the omnidirectional experimental variogram of CO₂ efflux normal scores from the survey conducted at the Cuicocha volcanic lake, as well as the parameters that refer to the variogram models. An inspection of CO₂ efflux distribution maps shows that background CO₂ efflux values (∼16 g⋅m⁻²·d⁻¹) are identified across most of the studied area, except for the 2012 survey that shows values of ∼100 g⋅m⁻²·d⁻¹. For all the surveys, relatively high CO₂ efflux values were observed in the bubbling gas zone located in the northern corner of Yerovi island (>40 g⋅m⁻²·d⁻¹), close to CO₂ bubbling areas (temperature of 16-17°C and pH of 7–8). During the 2012 survey, the highest CO₂ degassing rate showed four areas with particularly high values (>200 g⋅m⁻²·d⁻¹): 1) in the NW shores of the lake; 2) along the eastern shores of Wolf Island; 3) in the southern zone of the lake; and 4) along the SW shores (Figure 3A). Relatively high CO₂ efflux values (>35 g⋅m⁻²·d⁻¹) were measured in 2014 at the western zone of the lake and at the eastern shores (Figure 3B). The spatial distribution map of the 2017 survey shows constant CO₂ efflux values (>60 g⋅m⁻²·d⁻¹) over the entire surface of the lake (Figure 3C). Finally, the 2018 CO₂ efflux map shows a significant decrease in the magnitude of the values, with relatively high values of CO₂ efflux measured in the small area located to the W and NW of Wolf Island (Figure 3D).

Guided by the variogram model, sGs of diffuse CO₂ efflux data were conducted covering an area of 3.95 km² for each survey in the period 2012–2018. The CO₂ output estimated shows a range from 76 ± 3 to 652 ± 25 t⋅d⁻¹ in the 2012–2018 period, with 2018 and 2012 surveys showing the lowest and maximum values, respectively (Table 1 and Figure 4A). The normalized CO₂ emission value by area (3.95 km²) ranged between 19 ± 1 t⋅d⁻¹·km⁻² (in 2018) and 165 ± 6 t⋅d⁻¹·km⁻² (in 2012; Supplementary Table S1 in Supplementary Material).

CO₂ efflux values ranged from 5.2 to 542 g⋅m⁻²·d⁻¹, with an average value of 83 g⋅m⁻²·d⁻¹. The surface water temperature ranges between 10.7 and 15.5°C (average 13.6°C), with pH ranging between 6.60 and 8.10 (average 7.3). Similar to Cuicocha lake, no significant spatial variations of EC were observed on the water at 30 cm depth. In the bubbling gas zone water temperature is ∼1–2°C higher and pH is 0.5 units lower than the rest of the lake.

The probability-plot technique applied to the diffuse CO₂ efflux values confirms the existence of two log-normal populations (Figures 2E–G; Supplementary Table S1 in Supplementary Material). Population I showed values from 11 to 74 g⋅m⁻²·d⁻¹ (average 39 g⋅m⁻²·d⁻¹), which represented between 21.9 and 73.2% of the total data (average 48.4%). Population III, which represented a range from 9.7 to 33.9% (average 18.3%) of the total data, showed values between 107 and 354 g⋅m⁻²·d⁻¹ (average 206 g⋅m⁻²·d⁻¹).

The CO₂ efflux maps (Figure 5) show the highest values are located mainly in the SW shores (>450 g⋅m⁻²·d⁻¹), where gas bubbles and warm water springs appear, being characterized also by a relatively high water temperature and pH (temperature of ∼22°C and pH of ∼6.8) (6). In 2014, other high values of the CO₂ efflux were observed in the SW shores (>450 g⋅m⁻²·d⁻¹) and in the north-eastern zone of the lake (>200 g⋅m⁻²·d⁻¹; Figure 5A). During the 2017 survey, relatively high values of CO₂ efflux were observed in the eastern zone of the lake, with values > 250 g⋅m⁻²·d⁻¹ (Figure 5B). Regarding the 2018 survey, a general decrease on the extension and magnitude of the values were registered, with the principal anomalies focused in the south-western zone (Figure 5C).

The diffuse CO₂ emission at Quilotoa was estimated between 141 ± 6 and 536 ± 35 t⋅d⁻¹ (average 330 t⋅d⁻¹; Table 1 and Figure 4B), corresponding to an area of 3.50 km². The normalized emission value ranged between 40 ± 2 t⋅d⁻¹·km⁻² (in 2018) and 153 ± 10 t⋅d⁻¹·km⁻² (in 2014; see Supplementary Table S1 in Supplementary Material).

The observed bimodal distributions reflected the existence of more than one population of CO₂ efflux (Figure 2), which suggests the occurrence of different sources for the CO₂, as well as the existence of different mechanisms of gas transport (Padrón et al., 2008; Cardellini et al., 2017). In this respect, during the period 2012–2018, Cuicocha volcanic lake presents background CO₂ emission values represented predominantly by the diffusion of CO₂ through the water-air interface. The mean value of the background with a confidence level of one standard deviation was used to estimate the contribution of CO₂ from population I to the CO₂ emission of Cuicocha lake, as previously described (Melián et al., 2014). Assuming an area of 3.95 km² for Cuicocha volcanic lake, the cutoff background emission was estimated as 114 t⋅d⁻¹ and the standard deviation of the background emission was computed as 65 t⋅d⁻¹ for the 16th percentile (−1σ) and 141 t⋅d⁻¹ for the 84th percentile (+1σ). These values are similar to those considered by Sierra et al. (2021), who estimated values of cutoff background emission approximately 79–119 t⋅d⁻¹ (20–30 g⋅m⁻²·d⁻¹) based on the Graphical Statistical Approach method calculations and the model proposed by Mazot et al. (2014).

The advective mechanism represents an important contribution in population III for Cuicocha volcanic lake, not only by the direct transport to the surface, but also because bubbling contributes to the dissolved CO₂, and as such increases the CO₂ gradient between the bottom and the surface of the lake. The origin of population III is likely the CO₂ released from a magma chamber that escapes to the surface with a crustal CO₂ contribution from carbonate decomposition. Thus, the isotopic composition of dissolved gases (see Dissolved and Bubbling Gases at Cuicocha Volcanic Lake section) and the isotopic composition of bubbling gases evidence the existence of deep-seating magmatic degassing that in turn affects the lake.

The temporal evolution of diffuse CO₂ emission measured at the water surface is depicted in Figure 4A for Cuicocha volcanic lake (red dots for the present work data, red pentagon for data from Padrón et al. (2008), and red squares for the data from Sierra et al. (2021)), plotted together with the ³He/⁴He isotopic ratio measured in the bubbling gases collected in the lake in the period 2006–2018 (blue dots for the present work data and blue diamonds for the data from Inguaggiato et al. (2010)). The seismic event data of Cuicocha are from the IG-EPN seismic network and reported by Sierra et al. (2021). The diffuse CO₂ emission values for Cuicocha volcanic lake were similar to those reported by Padrón et al. (2008) and Sierra et al. (2021; Figure 4). Cuicocha is a monomictic lake, i.e., it has an overturn period every year from June to August, when circulation reduces the CO₂ accumulated in the deepest water. Under these conditions, a maximum CO₂ emission to the atmosphere occurs (Padrón et al., 2008). This could be the mechanism that explains the maximum emission rate measured in July 2012, values higher than those estimated as background emission. However, since a much lower value was reported in a survey conducted in 2006 (Padrón et al., 2008), a significant increase in the input of magmatic CO₂ cannot be excluded. The magmatic component increased in 2009 as indicated in Supplementary Table S2 (Supplementary Material), and the CO₂ released from the fresh magmatic melts persisted during 2009–2012. However, the magmatic He was released more quickly as the magmatic component remained at similar values in 2006 and 2012 and experienced an increase in 2009. Such observations are coherent with the expected geochemical behaviors of He and CO₂. It is also worth noting that the CO₂ emission peak detected in 2012 occurred after episodes of high LP seismicity at the end of 2011 and beginning of 2012 (Sierra et al., 2021). LP events are generally low-amplitude signals linked to alterations in the shallow hydrothermal system. Such alterations might be due to injection of magmatic fluids that were observed at the surface several months later. ³He/⁴He ratio measured in bubbling gases in the lake showed an increase from 2006 to 2009 (Figure 4A), which means an increase in the magmatic fraction of helium (Supplementary Table S2 in Supplementary Material). Unfortunately, no data are available between 2009 and 2012 to confirm if the increase persisted after 2009. The increase in magmatic helium emission suggests a magmatic intrusion, which likely occurred in 2009 or before, and injected magmatic gases and perturbed the hydrothermal system, stimulating pressure fluctuations and causing fluid-driven cracks in the volcano-hydrothermal system of Cuicocha.

Vertical profiles of water temperature, pH, and EC (Figures 6A–C) showed that Cuicocha volcanic lake is comprised of discrete water masses. The water temperature shows variations at 30 cm depth, which are attributed to environmental influences (Figure 6A). The temperature differences depicted in Figure 6A were observed between the surface waters and 20 m depth in both 2017 and 2018 surveys, and hence the thermal stratification. This observation is consistent with the report of a stratification period (September to May) in Cuicocha volcanic lake by Padrón et al. (2008).

The EC of Cuicocha lake (<700 μS⋅cm⁻¹) is relatively elevated compared to values of the nonactive Ecuadorian Mojanda caldera (35 μS⋅cm⁻¹), reported by Gunkel et al. (2009). In general, anions and cations content are present with higher concentration in the hypolimnetic waters than in the epilimnion (Figures 6D–G). These results are comparable with the data reported by Gunkel et al. (2009). According to the Langelier-Ludwig classification diagram represented in Figure 10A, the samples along the water column display a Ca²⁺(Mg²⁺)-HCO₃ ⁻ composition typical of worldwide superficial waters and shallow aquifers (Tassi et al., 2009; Inguaggiato et al., 2010). Inguaggiato et al. (2010) report water composition enriched in Cl⁻ + SO₄ ²⁻, probably because the sample was collected in the bubbling zone richest in Cl⁻ and SO₄ ² located on the northern shores of Yerovi Island. A similar behavior is observed at a depth of 60 m in one of the profiles reported by Gunkel et al. (2009), where an increase in the concentration of CO₂ was also registered (Gunkel et al., 2009). The samples of Cuicocha present low values of total dissolved solids (TDS = 269 mg⋅L⁻¹) which could be explained by a weak water-rock interaction due to short residence time and/or low aquifer temperature.

The relative HCO₃ ⁻, SO₄ ²⁻, and Cl⁻ contents in the vertical profiles samples are presented in Figure 10B (Giggenbach, 1988) of Cuicocha volcanic lake for the 2017 and 2018 surveys. The samples from the Cuicocha plot close to the HCO₃ ⁻ vertex, in the peripheral waters zone, probably due to gas-water interaction processes, CO₂ addition/removal processes in the aquifer (Inguaggiato et al., 2010), and/or addition of organic CO₂ from decomposition of plants and animals (Aguilera et al., 2000). The variation of SO₄ ²⁻ and Cl⁻ in water samples relative to the volcanic hydrothermal fluids (VHFs), meteoric water (MW), and seawater (SW) end-members (Hernández et al., 2017) is shown in Figure 11. The Cuicocha samples show a low SO₄ ²⁻ and Cl⁻ content and plot close to MW, explained by rainfall and input of surface water from the catchment area feed the lake, along with hydrothermal water inflow (Gunkel et al., 2009).

Values for δ²H_VSMOW-H₂O and δ¹⁸O_VSMOW-H₂O for Cuicocha profiles are presented in Figure 7, Local Meteoric Water Line (LMWL; δ²H = 6.3×δ¹⁸O+8.1; Inguaggiato et al., 2010), and Global Meteoric Water Line (GMWL; δ²H = 8×δ¹⁸O+10; Craig, 1961). To evaluate the evaporation process, the isotopic composition of San Vicente, Salinas, and Guapán water from Inguaggiato et al. (2010) is also represented for comparative purposes (Figure 1A). The samples of Cuicocha present values of δ¹⁸O_VSMOW-H₂O and δ²H_VSMOW-H₂O very close to the LMWL and/or GMWL, suggesting that the volcanic lake water is mainly recharged by local precipitation that is characterized by a short residence time (Giggenbach, 1991). This result is concordant with the low SO₄ ²⁻ and Cl⁻ contents shown in Figure 11. There are differences in the δ²H_VSMOW-H₂O of Cuicocha between waters collected in 2017 and 2018 (∼−6‰). δ²H_VSMOW-H₂O changes are related to H₂S exchange from volcanic gases (Karolyte et al., 2017). During 2017, Cuicocha volcanic lake presents values of CO₂ emission and ³He/⁴He greater than in 2018 (Figure 4), which is consistent with higher values of δ²H_VSMOW-H₂O.

Dissolved gases in volcanic lakes are excellent tracers of gas-water interaction, due to their high mobility and different solubilities (Capasso and Inguaggiato, 1998; Capasso et al., 2000). The concentrations of dissolved gases (Table 3) in the Cuicocha samples collected during the 2017 and 2018 surveys are higher than expected values for air-saturated water (ASW; Capasso and Inguaggiato, 1998) at the sampling temperature, which suggests an important gas–water interaction. The content of CO₂ in all the samples (∼26 cm³·STP·L⁻¹) is much higher than the ASW values (0.32 cm³·STP·L⁻¹). For the 2018 survey, high concentrations of He (∼8·10^–3 cm³·STP·L⁻¹) were also measured in the water column (ASW = 4.8·10^–5 cm³·STP·L⁻¹). The high concentrations of CO₂ and He measured in dissolved gases suggest a significant contribution of such gases from volatile-rich fluids. No significant variation was observed in the chemical and isotopic composition of water in 2018 compared to 2017.

The thermal stratification observed in Cuicocha volcanic lake (Figure 6A) is confirmed by a corresponding stratification of dissolved gases (Figures 8A–E). CO₂ concentration in dissolved gases showing a decreasing trend toward the 40 m depth sample (epilimnion zone) could be due to the loss to the atmosphere and photosynthetic utilization, which then increases again to the bottom of the lake (Figure 8A). The CH₄ concentrations are lower in the hypolimion with respect to the epilimnion (Figure 8C), while high concentrations of dissolved CO₂, H₂, and CH₄ in water have been observed around 20–40 m (Figures 8A,C,D), in the metalimnion zone. The carbon isotopic signature (Figure 8E) indicates a clear endogenous origin for CO₂, with a greater contribution in the metalimnion zone. These results are congruent with the data observed by Gunkel et al. (2009) who observed emission of gases using sonar from the bottom of the crater lake (see Figure 1B).

The relative compositions of both dissolved and bubbling gases are plotted on the CO₂-O₂-N₂ ternary diagram (Figure 8K). All samples of dissolved gases of Cuicocha volcanic lake show an alignment with an O₂/N₂ ratio at around ASW (∼0.489; Figures 8A,K) and a relative increase of CO₂ with depth was also observed (Figure 8A). This result suggests different degrees of interaction between the bubbling gases and the water of Cuicocha volcanic lake. Bubbling gases present an O₂/N₂ ratio lower than that in air (Figure 8K), highlighting an excess of non-atmospheric N₂ and/or a consumption of O₂. The O₂ consumption due to reducing redox conditions could be the principal process causing the relative N₂-enrichment. The relative content of He, N₂, and CO₂ in the samples of both dissolved and bubbling gases is represented (Figure 8L). All the dissolved gases samples of Cuicocha volcanic lake show very similar He, N₂, and CO₂ content and He/N₂ ratios higher than the atmospheric ratio (∼1.4·10^–3). The He/N₂ ratios in dissolved gases were very different from those measured in the bubbling gases, except for Cuicocha in the 2017 survey. All bubbling gases samples plot close to the N₂ vertex, due principally to reducing redox conditions.

As in the Cuicocha volcanic lake, two distinct modes were found at Quilotoa volcanic lake (Figure 2) suggesting a deep perturbation of the volcanic system for the CO₂ and/or different mechanisms of gas transport (Padrón et al., 2008; Cardellini et al., 2017). Background CO₂ emission values (population I) are predominantly represented by the diffusion of CO₂ through water-air interface during the period 2014–2018. The contribution of CO₂ from population I to the CO₂ emission of Quilotoa lake (3.50 km²) is estimated as 137 t⋅d⁻¹ with a standard deviation as 89 t⋅d⁻¹ for −1σ and 212 t⋅d⁻¹ for the +1σ. Νo previous data are described in the literature. Advection might be the responsible transport mechanism to explain the relatively high observed CO₂ efflux values (population III). The isotopic composition of dissolved and bubbling gases (see Dissolved and Bubbling Gases at Quilotoa volcanic Lake section) and the observation of acoustic degassing plumes in Quilotoa (see Acoustic Degassing Plumes and Crater-Lake Bottom Morphology at Quilotoa volcanic Lake section) evidence the existence of deep-seating degassing from the bottom of the crater lake.

Figure 4B shows the temporal evolution of diffuse CO₂ emission measured at the water surface from Quilotoa volcanic lake (red dots for present work data), plotted together with the ³He/⁴He isotopic ratio measured in the bubbling gases (blue dots for present work data and blue diamonds for data from Inguaggiato et al. (2010)). The temporal evolution of diffuse CO₂ emission measured from the water surface of Quilotoa volcanic lake during 2014–2018 period and the ³He/⁴He values at the bubbling gases in the period 2009–2018 present maximum values in 2014 and a decreasing trend until the 2018 survey. Unfortunately, geophysical records are not available nor diffuse CO₂ studies prior to 2014. The lack of more records does not allow for the evaluation of the possible origin of this increase; however, it is worth noting that an increase in the magmatic fraction of helium was observed in 2014 (Supplementary Table S2 in Supplementary Material), so an additional contribution of deep-seated fluids to the volcanic system cannot be ruled out.

The differences observed in the ³He/⁴He ratio measured in bubbling gases reported in this work and by other authors between Cuicocha and Quilotoa could be related to the age of the recent volcanism in both systems: Cuicocha has experienced 4 confirmed holocenic volcanic eruptions and Quilotoa only one. In fact, bubbling gases in Quilotoa show a higher percentage of radiogenic (crustal) He (∼79%) than that in Cuicocha (∼48%), where the magmatic component is higher. It is worth highlighting that the largest magmatic component of Cuicocha bubbling gases was observed in 2009 (Inguaggiato et al., 2010), 2–3 years before the seismic unrest that occurred in the surroundings of Cuicocha. As stated before, a magmatic intrusion, likely occurred in 2009 or before injected magmatic gases, disturbed the hydrothermal system that later originated pressure fluctuations, and caused fluid-driven cracks in the volcano-hydrothermal system of Cuicocha.

Quilotoa presents a relatively low normalized CO₂ emission values by area that was approximately ∼94 t⋅d⁻¹·km⁻², which is a typical value for lakes filled with neutral pH waters and comparable to other volcanic lakes as Masaya and Apoyeque (Nicaragua; Pérez et al., 2011) and significantly higher than that of Cuicocha. With a value of normalized CO₂ emission by area of ∼27 t⋅d⁻¹·km⁻², excluding the maximum of the 2012 survey where a contribution of magmatic CO₂ cannot be excluded. These values are similar to those estimated in Nejapa (Nicaragua) or Monoun (Cameroon) volcanic lakes (Pérez et al., 2011).

The water temperatures (Figure 6A), pH (Figure 6B), and EC (Figure 6C) measured during this study in the vertical profiles indicated that Quilotoa volcanic lake is made up of different water masses. The temperature differences depicted in Figure 6A were observed between the surface waters and 30 m depth in both the 2017 and 2018 surveys, and thermal stratification. This observation is consistent with the report by Aguilera et al. (2000). The EC of Quilotoa lake (∼12,000 μS⋅cm⁻¹) was high, compared with Cuicocha lake (<700 μS⋅cm⁻¹) and Mojanda caldera (35 μS⋅cm⁻¹; Gunkel et al., 2009). In general, anions and cations content present higher concentration in the hypolimnetic waters than in the epilimnion ones. The Langelier-Ludwig classification diagram (Figure 10A) shows that the water samples of the Quilotoa volcanic lake plot in the Na⁺(K⁺)-Cl^-(SO₄ ²⁻) field, typical of crater lakes in active volcanic systems (Delmelle and Bernard 1994; Tassi et al., 2009; Hernández et al., 2017). These results are in accordance with those reported by other authors who observed similar findings about the chemical composition of Quilotoa volcanic lake (Figure 10A; Aguilera et al., 2000; Inguaggiato et al., 2010).

Figure 10B shows the relative HCO₃ ⁻, SO₄ ²⁻, and Cl⁻ contents in the vertical profiles (Giggenbach, 1988) for Quilotoa volcanic lake for the 2017 and 2018 surveys. Data from Aguilera et al. (2000), Gunkel et al. (2009), and Inguaggiato et al. (2010) are also shown. The water from Quilotoa shows an intermediate composition with a trend observed in the Na⁺ and SO₄ ²⁻ contents (Figure 10C) that suggests the addition of S-rich gases to the lake water. The relatively higher Cl⁻/SO₄ ²⁻ ratios (Figure 11) of Quilotoa samples are probably due to 1) the removal of S (precipitation of SO₄ ²⁻ minerals and elementary S and/or bacterial reduction of SO₄ ²⁻ to S²⁻ and precipitation of sulfide minerals); 2) the addition of Cl⁻ related to inflow of neutral Na-Cl^- geothermal waters or brines; or 3) both (Aguilera et al., 2000). Quilotoa water samples present a typical composition of VHF, a fact that is also endorsed by the δ²H_VSMOW-H₂O and δ¹⁸O_VSMOW-H₂O values (see Figure 7).

The Quilotoa water samples are all enriched in both δ²H_VSMOW-H₂O and δ¹⁸O_VSMOW-H₂O relative to the GMWL indicating a different origin and/or that waters have modified their isotopic composition (Figure 7). A very moderate isotopic shift of oxygen can be observed relative to GMWL values by up to +3.5‰ (Figure 7), suggesting the occurrence of enhanced water–rock interactions, mixing with VHF or andesitic water, or/and evaporation processes. There are marked differences in the isotopic composition between the Quilotoa volcanic lake waters collected in 2009 by Inguaggiato et al. (2010) and the samples collected in the present study. Inguaggiato et al. (2010) reported for Quilotoa water samples collected in bubbling gases values of -2.8‰ for δ¹⁸O_VSMOW-H₂O and -39‰ for δ²H_VSMOW-H₂O. However, in this study, we obtained values for δ¹⁸O_VSMOW-H₂O of ∼ −1.5‰ and δ²H_VSMOW-H₂O of ∼ −29‰. Different hypotheses could explain the observed changes of δ²H_VSMOW-H₂O values between the samples collected during 2009 and 2017–2018: 1) changes in evaporation because of an increase in water temperature; 2) increased rainfall (displacement of the data toward the LMWL); 3) incorporation of andesitic water due to an increase in hydrothermal volcanic fluids; and 4) hydrogen exchange between H₂S and H₂O (Hernández et al., 2017). The concentration of Cl⁻, SO₄ ²⁻ and most cations were similar in Quilotoa for 2009 and this study (2017 and 2018 surveys); therefore, isotopic changes cannot be explained as a consequence of variations in concentration. The water temperature of Quilotoa lake in 2009 (22°C; Inguaggiato et al., 2010) was significantly higher than in 2017 and 2018 (∼13.5°C), so the evaporation process could explain the changes observed in the isotopic composition of water for the different surveys. For Quilotoa volcanic lake, the values reported by Inguaggiato et al. (2010) for the 2009 survey present lighter values with respect to the 2017 and 2018 surveys, indicating an increase in the isotopic exchange of endogenous CO₂ with water. A similar behavior was observed for δ²H_VSMOW-H₂O, showing a shift toward lighter δ²H_VSMOW-H₂O values, which can be explained in terms of more abundant H₂S and associated isotopic exchange (Aguilera et al., 2000). These variations were agreed with the emission CO₂ and ³He/⁴He data observed in 2009 and 2017–2018 from Quilotoa volcanic lake.

The concentrations of dissolved gases (Table 3 and Figure 7) in the Quilotoa samples of vertical profiles collected during the 2017 and 2018 surveys are higher than the expected values for ASW (Capasso and Inguaggiato, 1998) at the sampling temperature, which suggests an important gas–water interaction. The content of CO₂ in all the samples (∼115 cm³·STP·L⁻¹) is much higher than the ASW values (0.32 cm³·STP·L⁻¹; Capasso and Inguaggiato, 1998). For the 2018 survey, high concentrations of He (∼7.0·10^–2 cm³·STP·L⁻¹) were also measured in the water column (ASW = 4.8·10^–5 cm³·STP·L⁻¹; Capasso and Inguaggiato, 1998). The high concentrations of CO₂ and He measured in dissolved gases suggest a significant contribution of such gases from volatile-rich fluids.

The relative composition of both dissolved and bubbling gases is plotted on a CO₂-O₂-N₂ ternary diagram (Figure 7K). All samples of dissolved gases at Quilotoa show an alignment with an O₂/N₂ ratio at around ASW (∼0.529) and an increase of CO₂ with depth was also observed (Figure 7K). This result suggests different degrees of interaction between the bubbling gases and the water of Quilotoa volcanic lakes. Bubbling gases present an O₂/N₂ ratio lower than that in air (Figure 7J) with CO₂ content and O₂/N₂ ratio higher than Cuicocha, suggesting a reducing redox condition in Quilotoa volcanic lake.

The relative content of He, N₂, and CO₂ in the samples of both dissolved and bubbling gases are represented (Figure 7). All the dissolved gases samples of Quilotoa volcanic lake show very similar He, N₂, and CO₂ content with He/N₂ ratios (∼1.35·10^–2) higher than the atmospheric and very different from those measured in the bubbling gases. The physicochemical conditions of the Quilotoa volcanic lake (pH, temperature, and salinity) control the behavior of CO₂ in water due to the different solubility and reactivity between He and CO₂ (Capasso et al., 2000).

The distribution of dissolved H₂ and CH₄ and CO₂, together with the O₂/N₂ ratios along the Quilotoa vertical profile, are shown in Figure 7L. For Quilotoa volcanic lake, a stratification of the dissolved gases was observed at ∼30 m depth. CO₂ and CH₄ concentrations in dissolved gases show an increasing trend toward 120–140 m depth (Figures 7F,H). High concentrations of H₂ at depth stand out in the 2017 survey (Figure 7G), suggesting a reducing redox condition in Quilotoa volcanic lake. The isotopic signature of carbon (Figure 7J) also indicates a clear endogenous origin for CO₂ consistent with the occurrence of significant CO₂ sources rising from the bottom of the lake through fumarolic discharges.

Based on both frequencies studied (200 and 83 kHz), it can be interpreted that the acoustic data are composed of true and false bathymetric echoes (Figures 6A–C). False echoes forming acoustic positive anomalies are interpreted as zones of intense degassing generating false acoustic domes that mask the real lake floor as reported in other active acid crater lakes and verified in the Taal crater lake 2 years before the explosion of the crater lake, where intense degassing partially masked the lake floor (Hernández et al., 2017). Thus, two main positive anomalies (labeled as A and B in Figures 6A,B) are located in the center of ∼ 330 m crater floor. Aguilera et al. (2000) described that the deepest lake floor was located at 200 m water depths within an elongated deep basin surrounded by steep flanks, with a morphology and depth very different from those obtained in this work. The northern anomaly A is identified on both 200 and 83 kHz, whereas the southern anomaly B is only identified with the lowest frequency of 83 kHz (Figure 6B). Both anomalies are NW-SE oriented, suggesting they are related to deep vents sourced from the deepest crater lake. Otherwise, minor positive anomalies identified within the flanks are interpreted as minor degassing vents sourced from the flanks of crater (Figure 6C). These minor degassing vents do not mask the water column fully, which allows the possibility to image the degassing sourced from the lake’s bottom. Thus, two main vertical acoustic degassing plumes have been imaged by the echosounder in both the western and eastern sectors of the crater lake (Figure 12). The acoustic plume identified at a depth of 35 m along the western flank of the crater lake (Figure 12A) corresponds to a flare-like type interpreted as being related to a focused active continuous degassing vent according to the classification by Melián et al. (2017). The fluid of the vent is sourced from a system of normal faults (brown colors in Figure 12A) that constitute the basement of the western flank of the crater lake and migrates upward through a layer of bottom-lake deposits ∼20 m in thickness of (orange colors in Figure 12A). We interpret this layer beneath the lake bottom as hydrothermal-altered deposits embedded by fluids, since their impedance is very similar to the vertical acoustic plume rising throughout the water column. Therefore, the implication of hot water involved into this type of plume might be related to the occurrence of a groundwater reservoir below the bottom of the crater lake. This degassing vent seems to be related to an active zone of bubbling detected along the western shore of the crater lake (Figure 12D). Vents on the southwestern sector of the crater lake were highly active in 2014 when peaks of diffuse CO₂ efflux were measured on the lake surface at this sector (Figure 4A).

Otherwise, on the eastern side of the crater lake, another degasification plume was identified sourcing from the bottom at 120 m water depths (Figure 12B). The shape of this acoustic plume corresponds to a puffing-like flare caused by intermittent burst of a focused degassing according to the classification mentioned above. In this type of acoustic plumes, frequency of degassing may vary between 1 and 30 s (Melián et al., 2017). Contrary to the observations in the western sector, the degassing is sourced directly from the substrate of the lake bottom. This degassing vent at the bottom of the lake is related to a maximum in the measured CO₂ efflux measured on the lake surface in 2017 (Figure 4B, Figure 12C). Above the vertical plume, a highly reflective stratified layer is located at 30 m below the surface of the lake (Figure 12B). We suggest that this stratified acoustic boundary might represent the abrupt increase in the CO₂ pressure identified at ∼ 25 m below the surface of the lake by Aguilera et al. (2000). Another explanation for occurrence of this stratified plume is that they might be created by the formation of particles of calcite precipitated as a result of CO₂ loss, remaining in suspension in epilimnetic waters for long periods of time, perhaps several months or more, as proposed by Aguilera et al. (2000).

Surface water temperature has also been obtained from the echosounder sensor (Figure 6D). The intense anomaly detected along the western side (up to 21°C) is related to a bubbling zone identified in the shallow water (Figure 6D). The distribution of the surface temperatures suggests a clockwise convective gyre around the crater lake. Aguilera et al. (2000) calculated the expected changes in density of the Quilotoa crater lake waters at different depths, according to the variations in the temperature. Thus, at 8 m below the surface and a temperature of 20°C, the density is 1.0075 g⋅cm⁻³, increasing exponentially to 1.0090 g⋅cm⁻³ as the temperatures drop to values of 10–12°C. Thus, the rapid cooling of these warm water gyre might lead to the sinking of water as it is cooling. This process could trigger the renewal of stratified waters of the crater lake by opening “holes” within the sub-stratified layer of CO₂.