Limnological dynamics of methane (CH₄) and carbon dioxide (CO₂) emissions from a tropical hypertrophic reservoir lake

Freshwater ecosystems are an important component in the global carbon cycle, because they store and release important amounts of carbon to the atmosphere. Although they cover barely 3.7% of the Earth's land surface (Verpoorter et al. 2014), continental freshwater ecosystems are responsible for about 16% of the total CH₄ emissions (Saunois et al. 2016), and 15% of the total CO₂ emissions to the atmosphere (Tranvik et al. 2009). In particular, in freshwater reservoirs, where there is a large stock of terrestrial organic matter due to both flooding and external loading, microbial decomposition may drive important production and emissions of CH₄ and CO₂ (Deemer et al. 2016).

The rate and the direction of CH₄ emission fluxes from freshwater ecosystems depend on the interlinkage of biological processes that affect the CH₄ budget with their limnological dynamics. Indeed, although the CH₄ concentration (⁠⁠) and CH₄ emission to the atmosphere are mainly a result of the balance between anaerobic CH₄ production (methanogenesis) and aerobic CH₄ oxidation (methanotrophy) (Miller et al. 2004), both of these processes may be differentially affected by the limnological structure and dynamics specific of the system, which can also affect the CH₄ transport and storage.

In freshwater ecosystems, methanogenesis takes place mainly in anaerobic sediments. It is the final step of the anaerobic degradation of organic material, which releases CH₄, and depends mainly of the availability of organic matter (Glissmann et al. 2004). In counterpart, methanotrophy also plays an important role in global methane cycling, mitigating the emissions of CH₄ through its oxidation to CO₂ and water. This process takes place mainly in oxic/anoxic interfaces, and it is carried out by methanotrophs, an important and specialized group of ubiquitous bacteria (Hrsak & Begonja 1998).

In terms of its emission pathways, CH₄ can reach the atmosphere by ebullition or by diffusion. Methane ebullition flux depends mainly on the net CH₄ production rate and on the hydrostatic pressure that the bubbles have to overcome to leave the sediments (Fendinger et al. 1992). In the case of CH₄ exported through diffusion from the sediments, CH₄ can be oxidized by methanotrophs as soon as it reaches oxic water (Bastviken et al. 2002). Additionally, the emission to the atmosphere depends on the concentration difference between the air and the water surface, but the exchange rate depends also on the turbulence driven by the wind speed (Podgrajsek et al. 2014). In stratified freshwater ecosystems, CH₄ can build up in anoxic layers, until eventually this CH₄ storage can be emitted rapidly because of the full overturn of the system at the onset of the circulation period in monomictic freshwater ecosystems (Encinas Fernández et al. 2014). Additionally, CH₄ emissions might also occur and be significant in systems with high water level fluctuations (WLF) (Bartlett & Harriss 1993; Bartosiewicz et al. 2015, 2016) due to the short-term boundary mixing events that can occur during the stratification period in systems such as VB (Merino-Ibarra et al. 2021).

In the case of CO₂, the emission rates and the flux direction depend mainly on the trophic condition of the lake – the ratio of primary production to heterotrophic respiration (Guimarais-Bermejo et al. 2018), which are affected both by light penetration and the availability of inorganic nutrients and of organic matter (Bartosiewicz et al. 2016), dependent on the external loads it receives (Ramírez-Zierold et al. 2010) and the degree of eutrophication of the ecosystem, which in the case of VB is very high (Barjau-Aguilar et al. 2022). Additionally, the CO₂ budget is also affected by the allochthonous inputs of dissolved inorganic carbon coming from the weathering of mineral and soil respiration in the watershed (Marcé et al. 2015).

The majority of reports on CH₄ and CO₂ cycling dynamics in freshwater ecosystems are from temperate or boreal water bodies, while reports addressing this information in tropical water bodies are very scarce and heterogeneous. For example, in the global synthesis by Saunois et al. (2016), 83% of the >900 ecosystems they report on are located >50°N. Because of that, the assessment of greenhouse gas (GHG) dynamics in tropical aquatic ecosystems is crucial to reach a full understanding about global GHG balances, particularly because, as recognized by different studies (Bartlett & Harriss 1993; Merino-Ibarra et al. 2008), CH₄ emissions from tropical aquatic ecosystems may represent more than a half of the global emissions. Holgerson & Raymond (2016) have specifically addressed and outlined the critical lack of data on the CH₄ and CO₂ cycling dynamics under the 30° latitudes.

For these reasons, measuring CH₄ and CO₂ dynamics and emissions in reservoirs where these processes and fluxes have not been assessed is important to achieve both tropical and global understandings of the budgets of these GHG. This assessment is particularly needed for the numerous reservoirs throughout the world that receive high loads of organic matter and nutrients, as is the case of VB (Ramírez-Zierold et al. 2010), where GHG emissions are likely enhanced by the fueling additional of decomposition (Deemer et al. 2016).

It is also important to address the processes that may drive significant variations in these emissions, as is the case of WLF, a condition that affects numerous reservoirs because of water management or hydroelectric withdrawals, but also as a result of sharp local variations in rainfall and droughts derived from climate change. Independently of their origin, these high WLF may in turn cause high variations in the mechanisms and magnitude of GHG emissions, through hydrostatic pressure effects, but also through the impact of boundary mixing events that intensify during low water levels (Merino-Ibarra et al. 2008, 2021).

Therefore, the objectives of the present work are to measure both the dissolved concentrations of CH₄ and CO₂ and the magnitude and direction of their atmospheric fluxes in order to assess the annual cycling dynamics of these two GHG and their variability, mainly on the seasonal scale, in the hypertrophic monomictic tropical reservoir of VB, a system that is important in itself and is also likely representative of many other reservoirs that are increasingly turning toward a higher trophic condition.

With these results, we expect to contribute to a better understanding of the processes that take place in tropical water bodies under eutrophication and water level pressures, and the variability that their CH₄ and CO₂ concentrations and emissions may undergo as a consequence of these processes. These results can in turn also be useful to improve the global estimates of GHG, and specifically global estimation of carbon fluxes, and the relative importance of tropical freshwater ecosystems for these fluxes and budgets.

The selection of the VB reservoir for this study on GHG exchange also relied on the availability of limnological and biogeochemical data derived from the long-term monitoring of this system (Guimarais-Bermejo et al. 2018; Merino-Ibarra et al. 2021; Barjau-Aguilar et al. 2022) that has been ongoing since 2001. In previous studies, the lake's features, bathymetry, and its spatial variability were initially studied with an extensive, high-density network of sampling stations (17) distributed throughout the lake (Merino-Ibarra et al. 2008) to figure out the possible intricacy of the lake. Similarly, the influence of the four main rivers and sewages that discharge to VB was addressed in detail in a study of its N and P mass budgets and external loading (Ramírez-Zierold et al. 2010).

In these studies, it was found the lake is daily homogenized by the strong daily winds that blow in VB and that totally mix and homogenize its surface layer (Merino-Ibarra et al. 2008; Ramírez-Zierold et al. 2010). Functional studies on the primary production and community metabolism of the lake (Valdespino-Castillo et al. 2014) showed that this homogenization encompasses the metabolic processes and exchange fluxes (i.e., Guimarais-Bermejo et al. 2018; Barjau-Aguilar et al. 2022). Therefore, since 2008 VB monitoring has been realized using a single central station (Figure 1) which was shown to be representative of the lake in these previous works.

The CH₄ and CO₂ seasonal dynamics of the VB reservoir were therefore assessed at this long-term monitoring station (S in Figure 1) during four monitoring campaigns separated by 3-month intervals in seasonally characteristic months: April 2019 (early stratification period), July 2019 (stratification period), October 2019 (just before mixing period), and January 2020 (mixing period). Taking into account that the marked diel wind pattern of VB (Merino-Ibarra et al. 2008), which during the strong wind daytime period might bring GHG-rich water from the hypolimnion to the epilimnion (Merino-Ibarra et al. 2021), and trigger bubble release from the sediment (Podgrajsek et al. 2014), in each monitoring campaign GHG concentrations and emissions were studied both before and after the onset of these strong diurnal winds (i.e., ∼11:00 and ∼17:00) to include the influence of this wind diel pattern on GHG cycling in VB. Additionally, as part of the ongoing long-term monitoring of the reservoir underway (Merino-Ibarra et al. 2021), physicochemical parameters were monitored monthly throughout 2019 and 2020.

The thermocline depth was inferred from the water column stability, following Coloso et al. (2011). The oxycline depth was identified, when present, by the sharpest DO gradient and/or the presence of an oxic/anoxic gradient.

CH₄ and CO₂ analyses were made using an ultraportable greenhouse gas analyzer (UGGA, model 915-0011, Los Gatos Research, Inc., USA) and manual injection of the samples to an open circuit with a continuous flow of CH₄- and CO₂-Free nitrogen (Infra, Mexico) as a carrier gas, regulated at 1.5 L min⁻¹ by a mass flow controller (GFC17, Alborg, Denmark). All the tubing used in the circuit was made of polyurethane with a 6 mm external diameter (Festo, Mexico). Additionally, gas samples were measured by gas chromatography using an Agilent 7890A GC system in order to test accuracy. The gas chromatography system was configured with a single channel using two detectors (FID and micro-ECD) for the analysis of CO₂ and CH₄. The total CH₄ and CO₂ emissions from the reservoir were calculated using the mean flux measurements before and after the onset of the strong diurnal wind and the surface area of the reservoir during each campaign.

The results obtained are presented as the average ± one standard deviation. Additionally, temporal contour maps of T, DO, and RWCS were plotted using Surfer 11.0 software (Golden Software, USA). The best method of interpolation based on the mean absolute error (MAE) and the mean bias error (MBE; (Willmott & Matsuura 2006)) was used. Since ⁠, ⁠, DO, and GHG fluxes were not normally distributed and did not meet the assumption of homoscedasticity, data were compared (e.g., among season and water depths) using the Mann–Whitney U test. Prior to correlation analysis, the data were normalized using logarithmic base 10 transformation, and linear regression analysis was used to quantify relationships. Before and after transformation, the data were tested for normality using the Shapiro–Wilk test. All statistics analyses were done with the NCSS200 Statistical Analysis System software (Number Crucher Statistical System, Utah, USA).

Table 1 summarizes the physicochemical data measured in VB during 2019–2020. Water levels changed seasonally in VB during this period due to seasonal rainfall variations and the water withdrawal from the reservoir for human use in Mexico City, Toluca, and other metropolitan areas. The water volume of the reservoir varied smoothly during this period, it was 85, 76, 81, and 95% (April, July, October, and January, respectively) of the maximum capacity of VB (391 × 10⁶ m³, Merino-Ibarra et al. 2008), and the reservoir area was 89, 85, 87, and 95% relative to its maximum area (18.55 km²Merino-Ibarra et al. 2008). The depth at the sampling site also varied slightly, it was 26 m, 24, 25, and 28 m, respectively, for each of the monitoring campaigns (April, July, October, and January).

Table 1 also summarizes other physicochemical parameters measured in VB during this period. pH ranged from 7.88 ± 0.49 (whole water column average) during January to 8.76 ± 0.17 during October, showing a chemocline at the same depth of the thermocline, and a vertical pH gradient when thermal stratification occurred (i.e., during July pH was 8.75 ± 0.16 in the epilimnion and 7.69 ± 0.19 in hypolimnion). The Secchi depth ranged from 0.60 to 3.05 m, reaching its minimum in July (at noon, after the onset of the diurnal wind) and its maximum in January (before the onset of the diurnal wind), values which are in agreement with the conditions found in eutrophic lakes (Vollenweider & Kerekes 1982).

Figure 3 demonstrates that the vertical and temporal variation of DO concentration in VB was tightly coupled to the stratification/circulation cycle of the reservoir, as shown by temperature. During April, July and October a strong oxycline was observed, below which totally anoxic conditions were registered, while during January, the full water column of the reservoir was partially oxygenated.

The measured in VB during 2019–2020 varied over more than four orders of magnitude: from 0.015 to 176.808 μmol L⁻¹, particularly between the surface and bottom layers when the reservoir was stratified (Figure 4(b)). The annual surface average was 0.132 ± 0.107 μmol L⁻¹, while the bottom averaged 62.585 ± 71.786 μmol L⁻¹ when the reservoir was stratified. The at the bottom was significantly higher than that observed at the surface during the stratification (p < 0.05). Nevertheless, the data obtained for VB are overall still within the range previously reported for tropical hypertrophic freshwater ecosystems (Holgerson & Raymond 2016).

exhibited a trend similar to that of DO, in which the highest differences between surface and bottom were observed during the stratification, with a concentration of methane up to 2,000 times higher in the bottom than in the surface during October. This vertical gradient, which has also been observed in other stratified freshwater ecosystems, is likely a result of (i) CH₄ production in anoxic sediments and its hypolimnetic accumulation, (ii) the existence of a strong and sharp thermal stratification, that prevents the whole lake water circulation, and (iii) methane oxidation that may have occurred with high intensity in the oxycline due to the absence of subtract limitation (Utsumi et al. 1998). Even though methane oxidation was not measured in this study, it was inferred by the existence of inversely proportional and DO profiles. Over time, a high accumulation of CH₄ was observed in the anoxic hypolimnion of VB during the stratification (as also reported elsewhere, e.g., Bartosiewicz et al. 2016; Gerardo-Nieto et al. 2017; Thalasso et al. 2020), which outlines the importance of limnological dynamics on the concentration distribution of GHG in freshwater water bodies.

The surface varied only from 0.015 to 0.2795 μmol L⁻¹ during the stratification period, values similar to the lower reported for other tropical ecosystems (Engle & Melack 2000), particularly those with a similar size range (5–50 km²), and that are deep enough to allow water column stratification (Kankaala et al. 2013).

Nevertheless, in all the sampled seasons, was higher than the equilibrium value with the current atmospheric CH₄ concentration of 1.82 ppmv (average concentration measured 1 m above the surface), indicating a permanent oversaturation of CH₄. These oversaturation surface concentrations provide a clear indication that VB behaved as a source of CH₄ even during the stratification, which was also confirmed by the CH₄ flux measurements, which are discussed further ahead. In spite of the occurrence of CH₄ emission during the stratification, CH₄ accumulated in the hypolimnion of the lake, and the bottom reached levels much higher than at the surface, ranging from 0.292 to 176.808 μmol L⁻¹ (with a higher inter-sample variation; CV = 115%), outlining an important hypolimnetic CH₄ storage in VB during the stratification period.

In fact, the increase of the bottom during stratification reached up to 153.097 μmol L⁻¹ in the October sampling, which was prior to the overturn of the reservoir. Assuming that the measured in the bottom was similar for the entire hypolimnion, based on the high stability and homogeneity previously reported for VB (Merino-Ibarra et al. 2008), a total CH₄ mass accumulation of 250.57 ton for the entire system was estimated. Considering the surface area of VB during the October sampling, the storage contribution was 0.971 mol CH₄ m⁻². This storage mass would then have been released during the overturn that took place between the October 2019 and January 2020 samplings, as previously reported for other ecosystems (Bartosiewicz et al. 2015).

A positive correlation between temperature (near the sediments) and was found (R² of 0.92), which is consistent with the temperature dependence of the methanogenic community. Methanogenesis is a mesophilic process, in which higher temperatures imply higher CH₄ production rates, with an optimum temperature between 35 and 40 °C (Schulz et al. 1997; Yvon-Durocher et al. 2014). This allows identifying a connection between water level decrease and CH₄ emissions in VB.

Since Merino-Ibarra et al. (2021) have shown that the rate of hypolimnetic warming in VB during stratification increases rapidly as the water level of the reservoir drops, the decrease of the water level – a scenario likely to become frequent due to climate change (Valdespino-Castillo et al. 2019) – would derive in an enhancement of CH₄ emissions in VB and water bodies that behave similarly.

The measured in the reservoir varied from 15.062 μmol L⁻¹ (in the surface on April 2019) to 219.505 μmol L⁻¹ (in the bottom, January 2020; Figure 5), with an annual average surface of 65.783 ± 60.525 μmol L⁻¹. As in the cases of and DO, when the reservoir was stratified, the in the bottom was significantly higher than in the surface (p < 0.05), and in the bottom of VB averaged overall 139.893 ± 57.207 μmol L⁻¹.

During the stratification, surface varied from 15.062 to 78.218 μmol L⁻¹, which are values within the order of magnitude reported previously for other tropical freshwater ecosystems (Marotta et al. 2010) and ecosystems with a similar size range (López Bellido et al. 2009; Kankaala et al. 2013). In contrast to the case of ⁠, during stratification, was below the value needed for equilibrium with the current atmospheric CO₂ concentration of 412 ppm (measured 1 m above the water surface). This surface sub-saturation of CO₂ provides an indication that VB behaved as a CO₂ sink during the stratification period, which was also confirmed by the negative CO₂ fluxes measured during that period, as will be discussed further ahead.

Bottom varied from 59.772 to 219.505 μmol L⁻¹ in VB during 2019–2020, values also within the range of values reported for ecosystems with similar size range (Kankaala et al. 2013). However, bottom concentrations were always higher than at the surface, and exhibited a similar trend to and DO, due to CO₂ accumulation in the anoxic hypolimnion. The hypolimnetic reached up to 165.188 ± 4.977 μmol L⁻¹ during October. Assuming bottom concentration was similar throughout the hypolimnion, based on the high stability and homogeneity reported by Merino-Ibarra et al. (2008), a total CO₂ mass accumulation of 390.11 ton for the whole lake was estimated, which is equivalent to 0.537 mol CO₂ m⁻² considering its surface area in October 2019.

No evidence of ebullition was detected in VB during any of the sampling campaigns, either by visual observation or during SC measurement. This could be in part due to the fact that our monitoring station was near the center of the reservoir and at a depth greater than 20 m, where ebullition fluxes are less commonly observed (Natchimuthu et al. 2016; West et al. 2016). However, the existence of ebullitive GHG emissions cannot be totally discarded in VB, mainly in the case of CH₄ due to its low solubility in water (mole fraction solubility of 2.81 × 10⁻⁵ at 20 °C), and because ebullition events occur on a time frame of seconds and are followed by long periods without ebullition events (Saunois et al. 2016; Gerardo-Nieto et al. 2019). So, our estimates on CH₄ emissions are a minimum, a lower bound, and could be higher if there were any ebullition events that went undetected.

CH₄ emissions from VB varied between 23.25 μmol m⁻² day⁻¹ in the October sampling (before the onset of the diurnal wind) and 1,220.80 μmol m⁻² day⁻¹ in April, after the onset of the diurnal wind. As we mentioned in the Methods section, GHG fluxes were measured before and after the onset of the strong diurnal wind. Figure 4 shows that the methane emissions from VB are higher after the onset of the diurnal wind in all seasons. These higher CH₄ fluxes are likely a result of the wind speed effect on the surface gas exchange, since the water was oversaturated with CH₄ throughout the samplings in 2019–2020. These fluxes could also be enhanced by the boundary mixing events the daily wind drives in VB (Merino-Ibarra et al. 2021), and less likely by any bottom currents – which are prone to CH₄ liberation (Joyce & Jewell 2003) – that may derive from internal waves breaking against the bottom of VB, a mechanism that still would need to be assessed there.

The mean seasonal CH₄ flux measured was 816.64 ± 571.56 μmol m⁻² day⁻¹ in April, 125.11 ± 40.08 μmol m⁻² day⁻¹ in July, 140.14 ± 165.31 μmol m⁻² day⁻¹ in October, and 444.36 ± 562.08 in January. Assuming that each of the four campaigns was equally representative of one quarter of the year, the annual CH₄ emissions from VB derived from direct measurements would be 0.139 mol m⁻² year⁻¹, which implies 36.76 ton CH₄ for the entire reservoir. This is to be added to the storage component of the annual flux, which is the main way CH₄ is released from reservoirs (Kemenes et al. 2007). As previously calculated, VB would release on an annual basis a total of 287.33 ton CH₄ to the atmosphere, 87.21% of which is released during the autumn overturn.

Figure 5 shows the CO₂ fluxes, which varied between −60.11 mmol m⁻² day⁻¹ (July, before the onset of the diurnal wind) and 254.99 mmol m⁻² day⁻¹ (January, after the onset of the diurnal wind). As in the case of CH₄ fluxes, in all the sampling campaigns, the CO₂ fluxes were higher after the onset of the diurnal wind, supporting the importance of physical processes and the limnological dynamics for the emission rates, particularly the epilimnetic wind mixing, and the boundary mixing events driven by the shoaling of the pycnocline with the bottom and shoreline (Merino-Ibarra et al. 2021). Considering the measurements before and after the wind onset, the means of seasonal CO₂ fluxes were −7.39 ± 7.53 mmol m⁻² day⁻¹ in April, −53.36.11 ± 9.54 mmol m⁻² day⁻¹ in July, −25.05 ± 46.07 mmol m⁻² day⁻¹ in October, and 143.26 ± 158.01 mmol m⁻² day⁻¹ in January.

The negative CO₂ fluxes observed in VB throughout the stratification are consistent with the high autotrophic condition of the epilimnion of the reservoir during this limnological period (Valdespino-Castillo et al. 2014; Guimarais-Bermejo et al. 2018), which drives a CO₂ assimilation rate high enough to maintain continuously subsaturated values at the surface, in spite of the enhancement of the atmospheric CO₂ flux that the alkalinity of VB would cause (Wanninkhof & Knox 1996). In spite of the high productivity of the epilimnion during the stratification period, when the reservoir operates as a CO₂ sink, the net annual heterotrophy of the reservoir lake (Valdespino-Castillo et al. 2014; Guimarais-Bermejo et al. 2018) implies that overall it will behave as a net source of CO₂, with a very strong pulse of CO₂ emission during the turnover period and the lake circulation, as the high flux (143.26 ± 158.01 mmol m⁻² day⁻¹) observed in January supports.

Assuming that each of the four campaigns was equally representative of one quarter of the year, the calculated average annual CO₂ emission from VB was 5.24 mol m⁻² year⁻¹, which implies an annual emission of 3,806.22 ton CO₂ for the entire reservoir lake per year. Additionally, the storage component must also be included, and once it is calculated as previously described and added, a total emission of 4,196.34 ton of CO₂ from VB on a yearly basis was calculated, of which in this case only 9.30% of it would be released during the circulation period. This proportion contrasts with the case of CH₄, where most (∼87%) of the emission occurs during the overturn.

To integrate the GHG emissions of CH₄ and CO₂ from VB here calculated, and to assess their relative contribution to global warming, both emissions were converted to CO₂ equivalents. Table 2 summarizes the emissions from VB. Overall, the GHG total emissions from VB expressed in CO₂ equivalents (CO_2eq; 25 g CO_2eq g⁻¹ CH₄; Shindell et al. 2009) were calculated at 11,379.619 ton CO_2eq on an annual basis, 63.12% of which corresponds to CH₄ (Table 2). This proportion is similar to that suggested by Tranvik et al. (2009), who considered equal CH₄ and CO₂ contributions. However, these results contrast with those reported by Holgerson & Raymond (2016) who found that the contribution of CH₄ (in CO₂ eq) in diffusive fluxes depends on the lake area, and in the case of lakes with areas from 10 to 100 km², as is the case of VB, the CH₄ contributes less than 5% of total emissions. In this sense, VB has a much higher relative contribution of CH₄ that is an order of magnitude higher.

Another consequence of the high relative contribution of CH₄ is that the overturn release of the hypolimnetic (deep) GHG accumulated during the stratification period represents 58.48% of the total GHG emissions of VB (Table 2).

These results show that hypertrophic reservoirs like VB can be emitting a much higher relative amount of CH₄ than would be expected for their relatively small surface area (Holgerson & Raymond 2016). They also reveal that a strong wing regime as found in VB can enhance the emissions as the fluxes measured before and after the onset of the wind in VB clearly show. Systems with a strong wind regime would be emitting faster, while systems with light winds may have higher storage that would be released during extreme weather events.

This case study also shows that water level changes, which are likely to increase in the near future due to the effects of climate change can also enhance the GHG emissions from the bottom of the systems, either due to reduced pressure that would allow ebullition, to increased vertical mixing and to the enhancement of methanogenesis due to the bottom temperature increase derived from vertical mixing or to overall warming of reservoirs.

These effects should be considered by water managers along with their water administration considerations, as the impact and the likely feedback on climate change they may have could be of greater importance than the water management priorities themselves.

The annual CH₄ and CO₂ dynamics monitored in VB during 2019–2020 are overall consistent with previously reported work on the carbon dynamics of tropical freshwater bodies of a similar size range, although in VB, the relative importance of CH₄ was much higher, accounting for half of the carbon emissions, one order of magnitude higher than previously expected. In spite of this and of the highly eutrophic condition of VB, the overall GHG emissions observed in 2019–2020 in this reservoir were close to the center of the range of values previously reported for lakes with a similar size. A GHG emission of 1.889 gCO_2eq m⁻² year⁻¹ was calculated, corresponding to a total emission from VB of 11,379.619 ton CO_2eq on an annual basis, ∼63% of which would correspond to CH₄ emissions.

In VB, the strong diurnal winds that characterize this reservoir lake enhanced the emission rates, pointing out an important effect of wind that should be accounted for in the scope of the climate changes to come. Seasonal monitoring in VB also revealed that the mixing regime and the limnological dynamics of water bodies have important effects on their greenhouse gas emissions. On the short scale, water levels that can alter both the intensity of boundary mixing events and the local warming rate are also important factors that can affect GHG emissions. In the VB reservoir lake, thermal stratification narrowed down the CH₄ emissions, and the CO₂ fluxes were negative. Therefore, VB functioned as a net sink of GHG during the stratification period. As a result, GHG accumulated in the hypolimnion until the winter overturn of the lake, and the emissions derived from the storage of GHG in the lake represented a large fraction (more than 58%) of total annual fluxes. Stratification storage and overturn emission were particularly important in the case of CH₄, for which most (∼87%) of its emission was due to this storage and its emission during the overturn period, outlining the importance of limnological dynamics on CH₄ emission from stratifying water bodies. Because of the importance of wind and mixing events, future work should be directed to understand the short-scale emissions that may occur associated with these two critical factors.

We acknowledge CONAHCYT (project CF-2023-G-155, Postdoctoral scholarships to O. G-N, P.M V-C and J.A. R-Z), and UNAM (PAPITT IN111321 and PASPA, DGAPA, UNAM to M.M-I.) for their funding. We also acknowledge ‘Patronato ProValle de Bravo A.C.’ and Club Naútico Avándaro A.C. for their logistical support, and many students and collaborators, including Jaqueline Hernández-Angeles, Ariadna Esther Gómez Montesinos, Adriana Hernández Cruz, Zubia Jocelyn Cisneros Ramos and many others for their collaboration in the field and laboratory work needed for this publication.

This work was funded by CONAHCYT through project CF-2023-G-155 and PAPITT IN111321 from DGAPA, UNAM. Authors also acknowledge individual funding received: O. G-N. and P.M.V-C. were supported by CONAHCYT (#747276), and M.M-I. by PASPA of DGAPA, UNAM. The funding from these sponsors made possible the writing, review, and publication of this paper, but they did not have any involvement in the study design, in the collection, analysis and interpretation of data, in the writing of the report, and in the decision to submit the article for publication.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.