1. Introduction

In this paper we present chemical data from groundwater, especially its sulfate ion concentration, as tracers, and we assert that these data provide strong indirect evidence for the existence of a previously unrecognized deep groundwater flow path in the southeastern part of the Mexican Yucatán Peninsula (henceforth MY) adjacent to, and probably extending into, northern Belize. The region is an elevated zone of recharge encompassing much of the southeastern part of Campeche state and adjacent parts of southern Quintana Roo. There, the Paleocene-Eocene Icaiche Formation is present at the surface over a wide area; and, because that formation contains massive and extensive beds of highly soluble gypsum (CaSO₄•2H₂O), it is a major regional source of groundwater sulfate ions.

Although the Icaiche Formation is widely exposed, relatively little is known about the formation because a thick cover of tropical vegetation combined with a low population density in the region and a correspondingly sparce road network make detailed geologic observation difficult (^{Perry, et al., 2019}). Nevertheless, geologic research in this largely undeveloped southern part of MY is important and warrants broad interest because, as pointed out by ^{Perry et al., (2019)}, the Icaiche Formation is present in the subsurface beneath much of MY far beyond its outcrop area. It contains both thick beds of highly soluble, massive gypsum as well as argillite layers of low permeability, and the formation strongly influences geomorphology and groundwater geochemistry in populous parts of the northern Peninsula. Hence, detailed study of the Icaiche Formation and its geologic associations in this southern area where its outcrops are exposed and can be examined directly is an effective way of developing an increased understanding of important aspects of the hydrogeology and geomorphology of the entire MY.

In a companion to this paper, ^{Perry et al., (2019)} discussed important hyrogeologic and geomorphic effects that massive gypsum beds of the Icaiche Formation have produced both in and adjacent to its area of outcrop. In this paper we extend the discussion of how the geology and geochemistry of the Icaiche Formation has affected more populous regions by examining a specific case in which we examine the chemistry of high-sulfate groundwater in the elevated recharge area where gypsum of the formation is present, and we then compare that groundwater to ground and surface water chemistry in a lowland area more than 50 km away -- where groundwater of similarly high sulfate content is discharging into the Rio Hondo Fault Zone (henceforth RHFZ), (Figure 1). We also establish that no high-sulfate surface water or shallow groundwater is present in the wide intervening zone between the area of Icaiche gypsum outcrops and the RHFZ. Finally, we conclude that the remarkably similar high-sulfate water chemistry in the Icaiche recharge area is the probable direct source of high-sulfate water in the RHFZ. From what is known about regional geology we then suggest that the two areas are connected by a deep aquifer consisting of a weathered upper Cretaceous carbonate surface overlain by weakly consolidated air-fall debris of the K-Pg Chicxulub impact event as we discuss further in Section 5.1.

Figure 1 Geologic map of the southeastern Mexican Yucatán Peninsula (MY) showing the outcrop area of the Icaiche Formation and the Chichen Itza Formation in relation to outcrops of Miocene and younger rocks (from ^{SGM, 2007}) as well as relevant outcrops of the Albion and Barton Creek Formations (^{Pope et al., 2005}) discussed in the text. The eastern boundary of the Elevated Interior Region (EIR) is shown. The region extends into Guatemala and Belize to the south. Its entire outcrop area within Mexico is shown in ^{Perry et al. (2019)}, a companion paper that discusses additional powerful influences on the geomorphology and hydrogeology of MY by the Icaiche Formation. 

Figure 2 Generalized Stratigraphic column for SE Mexican Yucatan Peninsula. 

Figure 3 Schematic cross section of proposed Icaiche Formation aquifer. 

2. Essential aspects of regional stratigraphy

Discussion of the hydrogeology of the study area requires basic knowledge about three geologic formations: the late Cretaceous Barton Creek Formation, the K-Pg Albion Formation, and the Palecoene-Eocene Icaiche Formation. The Barton Creek and the Icaiche Formations have had a major influence on groundwater movement in the region but have received little attention in the context of hydrogeology.

In evaluating water analyses from the northeastern part of the study area we shall have occasion to discuss briefly the influence of a local argillaceous facies of the Miocene/Pliocene Carrillo Puerto Formation. That formation is a major carbonate aquifer farther north, but, as discussed in more detail by ^{Perry et al., 2019}, in east central Quintana Roo, south of the town of Felipe Carrillo Puerto, the Carrillo Puerto Formation has acted as an aquitard, partially isolating surface water flow from the underlying groundwater system.

In addition to the relevant geologic formations, part of an ancient fault system that extends along the entire eastern coastal area of MY acts as an important regional groundwater conduit. (^{Tulaczyk et al., 1993}; ^{Bauer-Gotwein et al., 2011}).

The Paleocene-Eocene Icaiche Formation, the dominant rock type exposed at the surface in much of the Mexican EIR (Figure 1), contains thoroughly recrystallized carbonate rock together with marl and abundant bedded gypsum. The formation name was assigned (^{Butterlin and Bonet, 1960}) during a regional field study of the area undertaken by Bonet and Butterlin (B and B) between the years 1958-1960, a time when geologic exploration of the area was particularly challenging because much of the EIR was heavily wooded and was crossed by few roads (a situation that is not completely different today). B and B were successful in mapping the area in large part because their project took place just at the time a road crew was cutting the east-west Mexican Highway 186 across the center of the area of Icaiche outcrops and through heavily forested terrain. Highway excavations produced fresh exposures of highly soluble gypsum that B and B examined before those outcrops could dissolve and disappear in the wet climate of southern Campeche [where average annual rainfall, although variable, is normally about 1100-1600 mm/year (^{Dunning et al.,2012}; ^{Dunning et al., 2016})].

In fact, gypsum in the area rarely crops out except where it has been exposed by relatively recent excavation in road cuts and quarries. Those limitations may have caused B and B and subsequent investigators (^{Lopez Ramos, 1973}; ^{Weidie et al., 1985}) to significantly underestimate the abundance of gypsum in the Icaiche Formation and hence in the EIR. Where newly exposed (Figure 4), the gypsum unit in the Icaiche Formation is up to 25-35 m thick (^{Perry et al., 2019}). Using known gypsum outcrops and the gypsum quarries marked on the 2007 edition of the Servicio Geológico Mexicano geologic map of the Peninsula (henceforth ^{SGM 2007}) it is possible to outline an area, probably exceeding 10,000 Km², over which gypsum crops out in the EIR. This is the area indicated by shading in Figure 1.

Figure 4 Outcrop of massively bedded gypsum of the Iaiche Formation along Mexican highway 186 (at 18.530^oN, 89.576^oW) near Xpujil, Campeche. 

From a high point in south-central Campeche, the Icaiche Formation dips gently northward beneath younger rocks; and, whereas it is exposed extensively in Campeche state, it does not appear at the surface in northern Quintana Roo. In Yucatán state it crops out only in the southernmost corner (^{Perry et al., 2018} and ^{Perry et al., 2019}). Weathering of Icaiche gypsum releases sulfate that contaminates groundwater within the EIR; as demonstrated quantitatively by ^{Wagner, 2009} and ^{Perry et al., 2011}.

Although no pre-Paleogene formations are shown in ^{SGM 2007}, a narrow band of Upper Cretaceous rocks, which is the stratigraphic equivalent of the Barton Creek Formation (BCF) of neighboring Belize, is present in southeastern Quintana Roo within a block uplifted in the RHFZ (^{Pope et al., 2005}). These rocks are exposed in the Johnson limestone/dolomite quarry (18.517°N, 88.484°W) near the town of Juan Sarabia, Quintana Roo (Figure 1). The BCF is part of the Upper Cretaceous Yucatán Group (^{Viniegra-O., 1981}; ^{Peterson, 1983}). Paleogeographic analysis by ^{Viniegra-O. (1981)} shows that the whole of the Cretaceous System of MY and adjacent Guatemala and Belize is comprised of a thick sequence of carbonates and evaporites that formed on a large contiguous shallow platform.

At Ramonal (18.42° N, 88.53°W) on the western edge of the RHFZ (Figure 1) there is a small outcrop of deeply weathered BCF dolomite with iron oxide staining, calcite veins, and abundant solution cavities (^{Pope et al., 2005}). The BCF is also present at several localities in Belize on the eastern side of the Rio Hondo (which here marks the Mexico-Belize border and flows within the RHFZ). In particular, the BCF is well exposed in Belize at the Albion Island quarry (18.13°N, 88.70°W, labeled "Barton Creek" in Figure 1), about 48 km south of the Johnson quarry (^{Ocampo et al., 1996}; ^{King and Petruny, 2014}).

The upper BCF surfaces in both the Johnson and Barton Creek quarries exhibit karst weathering features, and, at the well exposed Ramonal BCF outcrop, there is "extensive karst weathering with local relief of 3-10 m in the surface of a heavily recrystallized limestone with iron-oxide staining, caliche deposits, abundant vugs and travertine deposits" (^{Pope et al., 2005}).

The Albion Formation disconformably overlies the karst erosion surface of the BCF at each of the three BCF outcrops mentioned. (^{Pope et al., 2005}) It is a basal Paleogene ejecta blanket composed of fragments expelled from the 65 my Chicxulub bolide impact crater (^{Schulte et al., 2010}). In addition to the outcrop at Ramonal where the Albion Formation overlies the BCF, ^{Pope et al.,(2005)} have described numerous other outcrops of the Albion Formation along Ucum-La Unión highway (which parallels the western edge of the RHFZ).

The formation is weakly consolidated, and the mineralogy, texture, and size of its components are remarkably diverse. ^{Pope et al., 2005} give field and laboratory descriptions of several of the accessible and easily identifiable outcrops of the formation. Their description of the outcrops at Ramonal does not include the entire range of the lithology of the Albion Formation, but it of does demonstrate its heterogeneity.

At Ramonal, they report micritic limestone and dolomite cobbles and boulders up to 3 m in diameter that are supported by a matrix of calcite and dolomite silt. In thin section they report devitrified glass converted to clay with vesicles filled with carbonate matrix, limestone clasts containing foraminifera, and accretionary lapilli, all in a matrix of carbonate grains.

It is reasonable to infer that the BCF is present at depths of 4-20 m beneath the Albion Formation outcrops along the Ucum- La Unión highway. It is especially relevant to the present study and worth emphasis here that, where it is exposed in the Johnson quarry and at Ramonal, the BCF is a unit of relatively pure limestone/dolomite rock whose upper surface is thoroughly karstified. And, as noted above, at the beginning of the Paleogene that weathered surface was covered only by a thin layer of impact ejecta (the Albion Formation). Both the upper, weathered part of the BCF and perhaps parts of the weakly consolidated Albion Formation are likely to be quite permeable; hence, if extensive they would be expected to form a highly conductive aquifer.

3. Regional groundwater and surface water analyses and calculation of SI_gypsum

A study by ^{Sánchez-Sánchez et al., (2016)} contains chemical analyses and geographic coordinates of groundwater and surface water samples of southern Quintana Roo that greatly increase the regional hydrologic database. The focus of the ^{Sánchez-Sánchez et al., (2016)} paper is the use of factor analysis to distinguish families of water. We have extended their study by using their chemical analyses to determine the saturation indices (SI) of the sampled waters with respect to gypsum (^{Parkhurst, 1995}). [SI = log(Q/K)], where Q is the ion activity product of sulfate and calcium ions in a given solution, and K is the solubility product constant for gypsum.

Thus, an SI of zero corresponds to a water whose activity product (Q) indicates precise saturation with gypsum. Note that calculation of SI requires activities (not concentrations) of calcium and sulfate ions. These activities can be determined from the ^{Sánchez-Sánchez et al.,(2016)} dataset because those authors have reported complete major element analyses for their samples, thus permitting us to determine activity coefficients of calcium and sulfate ions specific to each sample.

We have combined the data from ^{Sánchez-Sánchez et al., (2016)} with earlier results by ^{Perry et al.,2002} and ^{Perry et al., 2011} to produce both Table 1 and Figure 5 (the spatial representation of the geochemical data). As noted in section 2.1, gypsum of the Icaiche Formation crops out on the surface in roadcuts and quarries of the EIR over much of the shaded area of Figure 1. Not surprisingly, as can be seen from the figure, those water samples that are from wells or from springs in direct contact with Icaiche Formation gypsum have saturation indices that indicate nearly complete saturation, i.e. values near zero (ranging from +0.02 to -0.15, with antilog values in Figure 5 between 1.05 and 0.71 respectively) and correspondingly high sulfate ion concentrations ranging from 31 to 36.5 meq/l. These high values are plotted as relatively large circles in the left map (for SI_gypsum) and center map (for sulfate concentration).

Figure 5 Analyses of groundwater and surface waters of southeastern MY listed in Table 1. Chemical data are superimposed on a simplified version of the map of Figure 1. For sulfate and chloride, circle size is proportional to concentration. On the leftmost map, circle size is proportional to the antilog of the gypsum Saturation Index (SI_gypsum); thus, large circles indicate near-saturation with respect to gypsum. Comparing the three figures, there is an obvious positive correlation of sulfate concentration with SI_gypsum and a low or negative correlation with chloride (right hand figure). (It is important to note that the circles refer specifically to the chemistry of the water at the point marked by their center. Higher or lower values may, and probably do, exist within the map areas of the circles.) Samples labelled C, B L, and P are identified in the text. 

Samples from the area of gypsum outcrops in the EIR of southern Campeche (shaded area in figures 1 and 5) have relatively low but variable chloride ion concentrations that, in waters with high SI_gypsum , could indicate release of pockets of halite and other highly soluble evaporite minerals trapped within armoring layers of weathering gypsum. In the case of three samples from the community of La Guadalupe, part of the chloride content may come from pollution by humans.

La Guadalupe is a remarkably poor community established adjacent to an outcrop of gypsum that contains a spring (which could have been a reason to site the village in this inappropriate spot). Water in the two hand-dug community wells is so contaminated by sulfate (a laxative) that CONAGUA (the national water commission) delivers potable water here by tank truck. Two of the La Guadalupe water samples in Table 1 come from the hand-dug wells within the village.

The third sample is from the spring adjacent to the town. Each of these samples has an SI_gypsum indicating saturation with respect to gypsum (+.02 to -0.07). Both wells have nitrate concentrations indicating significant anthropogenic pollution by organic matter, and all have relatively high chloride contents, which could come from dissolution of evaporite minerals or from anthropogenic sources. Despite this contamination, we confidently include these as valid samples here for two reasons: 1) the spring sample is in direct contact with gypsum, and 2) the SI_gypsum values of each sample indicate that the sulfate concentration of each sample is controlled by equilibrium with gypsum.

A few samples in the area of gypsum outcrops have quite low sulfate ion concentrations and negative gypsum saturation indices, indicating strong undersaturation. These are from known shallow sources fed by rain; and, typically, they are isolated from dissolving gypsum by layers of impermeable surficial clay. As we noted in section 1.1, the Maya of the EIR used natural and anthropogenically modified shallow water bodies like these as primary water sources during their peak period of occupation of the EIR a few centuries before and after 100 CE.

4. Regional patterns in water chemistry

East of the high-sulfate waters of the EIR there are two distinct zones outlined by their qualitatively different groundwater chemistry (Figure 5). Nearest to the EIR is a zone, or gap, approximately 50 km wide, trending in the direction of the RHFZ, in which groundwater has relatively low concentrations of chloride and sulfate ions and corresponding gypsum saturation indices ranging as low as -3.1.

Within the gap between the high-sulfate water of the EIR and the RHFZ, there are surface channels up to several meters deep that were established and are maintained by recurring episodic overland flow from storms. We can assert from a reconnaissance study by ^{Leal-Bautista, et al., (2017)} that these are not responsible for transporting high sulfate to the RHFZ because ephemeral lakes and ponds that are present in their channels contain fresh water of low ion conductivity. The most remarkable feature clearly demarcated in Figure 5 is the relatively narrow band containing high-sulfate water within and adjacent to the RHFZ and lying of the zone of relatively ion-free water. This band is comprised of wells, flowing springs, and cenotes with gypsum saturation indices ranging from -0.22 for water from Cenote Azul to -0.5 for a well (Rancho del Toro) 200 m from the Rio Hondo (Figure 5). Water samples from the river itself and from Laguna Bacalar also have near-saturation indices of -0.33 and -0.29 respectively. Allowing for some mixing with surface water and shallow groundwater, these results strongly suggest a direct connection between high-sulfate groundwater generated by gypsum dissolution in the EIR and the water that appears in the RHFZ, presumably from a higher elevation recharge zone to the west.

A cautionary note is that gypsum probably is also present in the EIR south of the Mexican border, and water with high sulfate concentration may be present in water of the Rio Hondo flowing north into Mexico from Belize. However, such river water entering from outside the study area would represent only a complimentary sulfate source (and is quite probably an extension into Belize and Guatemala of the system we describe). Evidence for a major sulfate source in the Mexican EIR is that the high-sulfate springs in the Rio Hondo valley are under a strong positive hydrostatic head. For example, Acapulquito Spring Table 1) has an elevation at the park pool of 15 masl, but the source spring can be traced to a small arroyo that heads up at about 30 m (Google Earth, accessed 9 September, 2020). (We note that "E" in the acronym "EIR" stands for "Elevated" and implies that this is a likely recharge area.

5. A proposed deep aquifer that transports high-sulfate groundwater to the coast from the EIR

We have demonstrated that, in a sparsely populated part of MY, namely, in southern Campeche state and southwestern Quintana Roo, there exists a highland region of distinctive high-sulfate ground-water recharge produced by dissolution of gypsum (the Icaiche recharge zone). Although the detailed geologic structure and subsurface stratigraphy of the area are incompletely known, the near-surface rocks are nearly flat-lying. We have further established that 50 km away, an adjacent lowland region exists that is a zone of high-sulfate groundwater discharge (the RHFZ). The composition of ground-water in the two regions is strikingly similar as can be seen by comparison of water in the two regions documented in Table 1 and illustrated in Figure 5. Rather than asking whether the two groundwater systems are directly connected, we think it more reasonable to ask: "How can they not be connected?" Rocks in the right stratigraphic position and with reasonable permeability to form a deep aquifer connecting the Icaiche recharge zone and the RHFZ discharge zone do exist. These are the karstified upper Cretaceous rocks of the BCF and the overlying weakly consolidated K-Pg deposits of the Albion Formation exposed at both the Johnson and Albion quarries and along the Ucum-La Unión highway. There is no direct evidence at present to indicate that either or both of these formations continue westward and thus directly underly the Icaiche Formation in the zone of recharge. But, overlying near-horizontal Paleogene rocks do persist westward into the EIR, so there is no stratigraphic evidence against such a projection.

Accordingly, we propose here, as a hypothesis to be tested, that a composite aquifer consisting of the Albion Formation and the upper, permeable part of the BCF continue west and that together they form a deep aquifer carrying high-sulfate water, generated by dissolution of Icaiche Formation gypsum, eastward and discharge this water into the RHFZ. The suggested stratigraphy is shown in Figure 2, and a schematic of the proposed aquifer system is shown in Figure 3.

Our model, is consistent with the following observations and inferences: 1) Gypsum-saturated, sulfate-rich groundwater is present throughout that part of the EIR covered by Icaiche Formation rocks containing abundant gypsum, specifically in the shaded area of Figures 1 and 5. Given that the estimated annual rainfall in the region is about 1100-1600 mm/year it follows that the amount of recharge must be large. Yet, referring specifically to the eastern edge of the EIR, for which we have data, there is no evident continuous path by which this groundwater is discharged to the ocean. Surface water and shallow groundwater in the 50 km "gap area" west of the RHFZ is characterized by low sulfate content. 2) Water nearly saturated with respect to gypsum and containing exceptionally high concentrations of sulfate enters the RHFZ from an un-recognized source and is discharged by the Rio Hondo and by flowing springs in a narrow zone whose axis is the NNE-SSW Fault Zone, which is separated by 50 km from the nearest known high-sulfate groundwater. 3) An uplifted fault block of the BCF limestone/dolomite is brought to the surface in the RHFZ along the Mexico-Belize border between the Johnson quarry in Mexico and the Albion quarry in Belize (Figure 1). 4) In the late Cretaceous, even before emplacement of the K/Pg Albion Formation and any subsequent rocks, the uppermost part of the BCF had been altered by strong weathering to produce a karst surface. 5) It is reasonable to project the BCF westward beneath the younger formations to a position beneath the Icaiche Formation, where its weathered upper surface could serve as an effective aquifer for transport of high-sulfate water to the RHFZ on the east coast. One apparent anomaly in Figure 5 is the high SI of a near-coastal water from a well in the Chetumal suburb of Calderitas (labelled "C" in Figure 5). That water can be readily explained by our model if, as is likely, the linear coast north of Chetumal is actually a fault parallel to the main RHFZ and the other faults that define much of the east coastline as suggested in Figure 1). It is reasonable to expect that this fault also taps water from the presumptive Barton Creek aquifer.

6. Complications and Caveats

Factors complicating the deep aquifer hypothesis are: 1) Possible faulting in the BCF whose geology in Belize and Guatemala is known to be complex (^{King et al., 2003}; ^{King et al., 2004}; ^{King and Petruny, 2014}). 2) Probable facies changes and permeability variation in the Miocene-Pliocene Carrillo Puerto Formation and underlying eastern coastal Miocene formations as shown schematically in the stratigraphic column of ^{SGM 2007} and discussed in ^{Perry et al., (2019)}. 3) Variable quantity and chemistry of regional surface water and shallow groundwater input sources that could mix with and generally dilute the putative deep groundwater from the EIR. 4) Penetration of a saline intrusion into the northernmost part of the RHFZ (shown by high chloride concentration and correspondingly lower SI in water of the three northernmost wells along the RHFZ projection. Of these interesting uncertainties, the data set presented here in Table 1 is only useful in addressing the fourth one: what happens to groundwater flowing through the RHFZ as it moves north into the extensive, complex and poorly understood area south and east of Lake Chichancanab (19.88°N, 88.77°W) in the eastern part of Quintana Roo. That is the area where the Ticul Fault and the RHFZ intersect (^{Perry et al., 2019}).

Three samples presented in Table 1 have chloride values high enough to confirm that they have definitely mixed with water of marine origin. These are: Buenavista (18.889°N, 88.237°W), Pedro Antonio Santos (18.958°N, 18.958°W), and Los Limones (19.022°N, 88.110°W). They are marked B, P, and L respectively in Figure 5. The most noteworthy of these samples is from Buenavista, with a high sulfate concentration and SIgypsum (29.15meq/l and -0.51 respectively) and a chloride concentration of 44.51 meq/l. Thus, along with sulfate and SI values typical of the RHFZ, the Buenavista water has a chloride ion concentration that is 8% of the chloride concentration of seawater. It also has comparably high concentrations of sodium and magnesium (^{Sánchez-Sánchez et al., 2016}). This amount of chloride in water from a municipal well is remarkable. Among the large number of wells (including those in coastal areas) in the MY used to obtain potable water that were sampled by ^{Perry et al., 2002}, no similar value is reported.

Below a thick freshwater lens, deep east coast cenotes do actually have even much higher chloride concentrations than water from Buenavista. For example, in a cenote within the Calica quarry, 7 km from the Caribbean coast in northern Quintana Roo, Bianca ^{Pedersen (2007}, unpublished PhD thesis) reported that below a sharp interface separating fresh and saline water and at 14 m depth, chloride concentration increases rapidly, reaching an approximate seawater value of 400 meq/l at about 18 m depth. That profile makes it seem likely that a marine saline intrusion is present beneath the Buenavista well and that the well is drawing water from near the depth of the saline interface. That interpretation is supported by unpublished well logs provided by Joan Sánchez-Sánchez (Written Communication, 25 September, 2020). Those logs report 22.3 meq/l chloride in 2000 and 36.8 mq/l in 2009. This regular increase in chloride concentration with time suggests that a saline interface has progressively invaded the well over time, either by over-pumping of the well and consequent upconing or by a regional upward migration of the interface. We conclude that at Buenavista, significant mixing occurs between a marine saline intrusion and water flowing northward through the RHFZ.

Los Limones is the farthest north of the three chloride-rich cenotes in the northern RHFZ extension. It has the lowest sulfate and chloride concentrations of the three samples. Its 2.5 meq/l sulfate concentration is unremarkable, but its 17.15 meq/l chloride concentration is five times the average chloride concentration of samples in Table 1. Pedro Antonio Solis is located between Los Limones and Buenavista, and, among the three samples it is intermediate in sulfate, SI, and chloride values. The influence of sulfate derived from gypsum is obviously attenuated moving north from Buenavista. Chloride attenuation is also occurring perhaps as a result of increased fresh water input from the Ticul Fault and the northern fresh water lens.

Because groundwater from Buenavista is so unusual, we insert here for reference some observations whose significance we are unable to assess. Buenavista, which lies on a sharp bend near the northern end of Lake Bacalar, is 24 km south of Lake Noh Bec, the lake through which the extension of the Ticul Fault is projected on the ^{SGM 2007} map. The bent arms of both Lake Noh Bec and Lake Bacalar are approximately 5 km long. After this 5 km offset, the southeast arm of each lake bends back to parallel the direction of its western arm. Thus, the two lakes zigzag in parallel. The ^{SGM 2007} map shows the bend in Lake Noh Bec as an offset on the Ticul Fault. If, as implied, the bend in Lake Noh Bec marks a fault, it looks as though Lake Bacalar may have been displaced in parallel, suggesting faulting contemporaneous with or subsequent to deposition of the Miocene-Pliocene Carrillo Puerto Formation.

7. Independent Support

Two quite different and independent recent observations are consistent with the existence of the deep aquifer proposed here. One is the recent catastrophic draining of a lake precisely where we have suggested that recharge of the proposed aquifer takes place. The second recent observation of possible relevance is the discovery of evidence for a large submarine discharge off the Caribbean coast of Quintana Roo.

8. Significance

What is the regional significance of a possible deep aquifer containing sulfate-contaminated groundwater? Several factors may be relevant: 1) A human population wave, stimulated by tourism, is moving southward from Cancun and may soon stress the known exploitable supply of potable water on the Caribbean coast and in the adjacent EIR. Preparing for this influx will require better hydrologic information about the area than currently exists. 2) In contrast to Cancun and other populous parts of northeast Quintana Roo, it may be that the presence of an aquitard will isolate ground-water in the proposed deep, high-sulfate aquifer supplying Lake Bacalar and Cenote Azul and protect water in these recreation spots from contamination (or, conversely, it could produce greater contamination if the deep aquifer were selected as a prime waste disposal site). 3) A major objective of the Caribbean hydrographic study by ^{Carrillo et al.,(2016)} was to better understand marine conditions stressing corals in the largest coral reef system in the Atlantic Ocean. If a substantial part of the submarine discharge they discovered comes from the deep aquifer proposed here, then protecting that aquifer from pollution by sewage injection could be important to preservation of corals. 4) The Maya spent millennia learning how to thrive with the complex hydrologic systems of the EIR and the rest of the Peninsula. They persisted by developing sophisticated rainwater harvesting techniques ^{Perry et al.,(2019)}. Archeologists and others are working to understand how the Maya civilization interacted with climate dynamics and how this might help a modern burgeoning population to effectively respond to climate change. Ultimately, the Maya abandoned the EIR, a region in which they had coped with extreme hydrologic challenges for many generations. Learning more about regional hydrology may help understand why they left.

9. Conclusions

We present and evaluate data suggesting that a previously unrecognized deep aquifer exists in a sparsely settled but developing part of southern Quintana Roo and adjacent southern Campeche. Evidence for this aquifer is indirect, derived from a study of groundwater geochemical data. Confirmation will require obtaining much more detailed and focused geochemical information than we have been able to present. However, real understanding of the hydrogeology of the region is possible only as part of an integrated geoscience program that begins with a coordinated mapping of regional stratigraphy and structure that extends into Belize and Guatemala. Such a project would also require comprehensive geophysical study, followed by drilling. Strong current and accelerating political and economic pressure to develop the region justifies such a study.