We accept our hypothesis that tropical karstic marshes locally called sabanas are phreatotrophic wetlands, the hydrochemical data demonstrated that groundwater is the dominant water source in this marsh. Stein et al. (2004) reported that faults may be the conduit responsible for water delivery into wetlands; thus, supporting the assumption of GW provenance for this tropical karstic marsh. The water of the marsh of La Esperanza is calcium bicarbonate type regardless the origin of the sample. We consider that the homogeneous chemical conditions in low flood conditions reflect stable hydraulic conditions, which changed in response to seasonal events (such as frontal rains), the presence of vegetation (Yang et al. 2008) and hydrodynamics (Karstens et al. 2015). It is important to mention that all GW eventually was in the ground surface short after it precipitated, so the SW and IW plotting in the recharge section of the Chadah diagram is attributed to the transit of water from the surface to the vadose zone and the into the shallow aquifer. The water moving through the vadose zone into the saturated zone, interacts with sedimentary rock, dissolving carbonates and as HCO3− and Ca2+ (Ravikumar et al. 2017) as showed by the saturation indexes. IW has a more defined chemical composition, the carbonate minerals are around equilibrium, but it also acquired characteristics of infiltrating water that has passed through the soil, such as acidic pH, reductive conditions and organic matter decomposition (Craft et al. 1991). Similar electrical conductivity in GW and IW is assumed as comparable amount of dissolved solids was found in those compartments; the former due to the water-rock interaction (Smith et al. 2020), the latter due to the solubilisation of several substances around the rhizosphere (Raghavendra et al. 2016). In summary, this marsh is a good site to evaluate in detail the water infiltration process and the chemical changes that occur in the water during this process.
Changes in EC in SW are interpreted as entrance of precipitation with the consequent dilution effect in the previously existent surface water (McDonough et al. 2020). Slightly more acidic pH in IW might be attributed to elevated organic acids and secreted substances by organisms in the rhizosphere (Lynch et al. 2012). The water oxidation-reduction potential, together with the behaviour of redox-sensitive species and ferrous iron in the soil, give additional hints of the most probable processes occurring inside the marsh. For instance, moderately reduced condition in GW (≈ 250 mV) enable nitrates and sulfates presence in GW; whereas reduced conditions in SW (≈ 90 mV) and IW (≈ -16 mV) indicate that most nitrate has been reduced, supported by very low N-NO3−/L concentrations measured in SW and IW (Close et al. 2016). In addition, ammonium concentrations increase in SW and IW because of organic matter decay, mineralization, retaining or transformation of NH4+ (Gribsholt et al. 2006). Ammonia could also be a product of nitrogen fixation, since high nitrogen fixation (expressed as nitrogenase activity) has been described in karstic marshes (Rejmánková and Komárková 2000). DIN in low flood conditions (October 2019) was higher in SW, but changed to higher DIN in IW with greater flood conditions (February 2020). We are not able to estimate N transformation rates; however, we consider that high DIN in SW is the result of water evaporation from the marsh, yielding greater concentration of N in a reduced water volume. The water saturated, anoxic soil supported by the ferrous iron tests in soil, provides evidence of reductive conditions in the rhizosphere, which allowed sulphate reduction; thus, promoting high alkalinity in IW (Norton et al. 1988, Pester et al. 2012). To the best of our knowledge, high alkalinity (≥ 10 mM HCO3−) can inhibit growth of some plants (Cartmill et al. 2008); the alkalinity in this marsh was around 6.2 (± 2.4 mM HCO3−) and did not seem to affect the growth of grasses and sedges. High alkalinity in wetlands has been identified as a sink of atmospheric CO2 (Saderne et al. 2021) accounting for an overlooked component of the C cycle and storage in wetlands.
Low phosphates concentrations in karstic ecosystems is attributed to Ca- and Fe/Al-bound P, as they vary spatially and seasonally (Gao et al. 2019). With the measured pH-Eh, the expected P species is H2PO4− (di-hydrogen phosphate ion, Takeno 2005). With oxygen deficiency and stagnant conditions, phosphorus can be released back from sediments to the water column and with oscillating flood conditions; sediments rich in Fe and Mn are capable of becoming sinks again (Karstens et al. 2015).
The similar COD between SW and IW reveals that with stable hydraulic conditions, the water in the rhizosphere and above the ground had similar amount of organic matter. This situation changed in February 2019 (higher flood level), when higher COD in IW indicated increase in dissolved organic matter in the rhizosphere. High COD has also been associated to elevated sulfate reduction and inhibition of some trophic groups by H2S (Barber and Stuckey 2000). This situation was likely occurring in the karst marsh as low sulfate concentrations were measured with high flood. High COD/N ratio (600 to 1800 in La Esperanza marsh) relates to very active denitrifiers, yielding the low nitrate conditions observed, due to high availability of organic matter as electron donor (Hou et al. 2018).
In case of silica, its variable concentration can be related to the hydrology of the marsh and the dominant plant species. Both Grasses (Poaceae) and sedges (Cyperaceae) are Si-accumulating species (Struyf and Conley, 2009) which can release it back to the environment after biomass decay. Marshes are recognized as silica recyclers and re-suppliers (Struyf et al. 2005). We do not know microalgae Si demand, but it is reported that some microalgae can regulate fluxes and DSi concentrations in aquatic environments for as long as six months (Sigmon and Cahoon, 1997). We cannot provide plausible explanations because we ignore the microalgae composition. The knowledge of Si cycling in wetlands is incomplete (Struyf et al. 2005); this topic requires further research, especially due to the relevance of Si as groundwater tracer.
The under-saturation of the evaporitic minerals (anhydrite, gypsum and halite) suggest that those rocks are not dominant underlying this marsh; however, the trend might be opposite in southern areas of the Holbox Fracture system, where other geological formations are dominated by gypsum (Perry et al. 2002). Aragonite, calcite and dolomite had variable trends, there might exists precipitation, deposition and resuspension of biogenic carbonates (Morse 1986) as observed for the variability in saturation indexes between sampling events. The Chadah diagram clearly points towards recharge processes. Additionally, the Gibbs diagram associates some water samples with rock dissolution. Rainwater is the main entrance of water into the aquifer, large part of the precipitation rapidly infiltrates until reaching the saturated zone of the aquifer (Lases-Hernandez et al. 2019). Then, as the phreatic level rises because of increases in the hydraulic head, the GW already in contact with sedimentary rock, increases its interaction with calcite and dolomite, enhancing water erosion (Dai et al. 2017), increasing the dissolved solids in IW and even in SW as the water table elevates above the ground level. Finally, as indicated by the Gibbs diagram, evaporation is another dominant process in the marsh. This evaporation is supported by the increase in dissolved solid in SW during low flood conditions. It is important to mention that, what we consider SW, is a combination of groundwater discharged into the marsh (Gondwe et al. 2010) in addition to intercepted precipitation inside the marsh; thus, the evaporation process affects SW after the combination of the two main processes above-mentioned, yielding water subject to evaporation and rock dissolution.
Regarding our second hypothesis, we have less evidence that the highway had an influence on some parameters in the marsh. For instance, ferrous iron is soil was present in both portions, with higher concentration in the south portion, suggesting prolonged or predominant anoxic conditions. Nonetheless, X-ray spectroscopy suggested that iron (and chloride) were present only in the north portion. This last piece of evidence could be inferred as the effect of surface runoff form the highway. Road traffic has been related to the production and deposition of dust with metals and elements related to engine combustion and tires wear (Adachi and Tainosho 2004, Aguilera et al. 2018). The production of these residues, together with the preferential flow northward of the water (Perry et al. 2002, McKay et al. 2020), help explaining its presence only in one side of the road, partially supporting our hypothesis that the highway caused changes in the biogeochemistry of the marsh. We were not able to find effects of the highway in the hydrology of the marsh, the similar flood level in both sides of the road suggest that there is not any apparent hydraulic cut so far. We cannot discuss the fact that the largest polygon located at the north of the highway has something to do with changes in the groundwater flow, discharge or changes in the microtopography.
The soil of this tropical karstic marsh has very high water retention capacity, holding as much as seven grams of water per gram of soil. This condition is not exclusive of marshes; it is a common feature across all types of wetlands (Campos et al. 2011). The soil in these marshes do not develop deep profiles (Fragoso-Servón et al. 2020). Assuming that the mean soil depth in the marsh is 50 cm, with an average apparent density of 0.25 g m− 3 (250 kg m− 3), there are approximately 97,173,142.5 kg of soil in the marsh with an average gravimetric water capacity of 5.8 kg of water per kg of soil. This represents an estimated water retention capacity in the soil of this marsh of 563,604 m3 of water (563.6 million liters). When the water column is above the ground such as in November 2020 (1 m flood lever after Hurricanes Delta and Zeta), the micro-basin of this karstic wetland would store at least 700,000 m3 of surface water, in addition to the water stored in the soil. This goes to show the magnitude and relevance of the provisioning and supporting services that tropical karstic marshes provide just by existing, and the importance of maintaining in good conditions for keeping these environmental services, commonly acknowledged but barely understood and quantified (Shepard et al. 2011).
The elemental composition of the soil clearly reflect its sedimentary origin, with calcium as the element comprising 90% of the mineral fraction. The presence of Si and Mg are also result of rock weathering (Cejudo et al. 2020). In case of Al, kaolinite is naturally present in young and intermediate karstic landscapes (Bautista et al. 2011) and it is commonly in aluminosilicates and oxide minerals. Its availability do not depend of the redox potential of the environment and is one of the most prevalent contaminants directly attributed to human activities and potentially toxic to aquatic biota (Gensemer and Playle, 1999). Its deposition in the soil is considered result of weathering and mobilization of clays (Cabadas et al. 2010). This element and Si deserves much more attention in wetlands biogeochemistry.