Applied Geochemistry 98 (2018) 265–277

Contents lists available at ScienceDirect

Applied Geochemistry
journal homepage: www.elsevier.com/locate/apgeochem

Influence of rainy season and land use on drinking water quality in a karst T
landscape, State of Yucatán, Mexico
David T. Longa,b,c,d,∗, Amber L. Pearsonc,e, Thomas C. Voicea,b,c,d, Angel G. Polanco-Rodríguezf,
E. Cuauhtemoc Sanchez-Rodríguezd,g, Irene Xagorarakib,d, Fanny G. Concha-Valdezf,
Miguel Puc-Francof, Rafael Lopez-Cetzf, Amanda T. Rzotkiewicze
a
Department of Earth and Environmental Sciences, Michigan State University, East Lansing, MI, USA
b
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI, USA
c
Environmental Science and Policy Program, Michigan State University, East Lansing, MI, USA
d
Institute for Global Health, Michigan State University, East Lansing, MI, USA
e
Department of Geography, Environment, and Spatial Sciences, Michigan State University, East Lansing, MI, USA
f
Department of Social Medicine and Public Health and Laboratory of Microbiology of Regional Research Center, Faculty of Chemistry, Autonomous University of Yucatán,
Mexico
g
Department of Hyperbaric Medicine, Hospital General Augustín O'Horan, Mérida, Yucatán, Mexico

A R T I C LE I N FO A B S T R A C T

Keywords: To understand the impacts of dry and wet seasons and land use on water quality in a karst system, twenty-one
Karst deep municipal wells were sampled synoptically across the State of Yucatán. Surrounding land uses comprised
Water quality urban, crop, and livestock as defined by the National Institute of Statistics and Geography of Mexico.
Seasonal effects Measurements included pH, temperature, specific conductance, Ca2+, Mg2+, Na+, K+, Cl−, SO42−, alkalinity,
Land use
NO3−, and fecal coliforms. Data were reduced using various geochemical, statistical, and spatial analytical
Flow paths
methods and compared to a similar study on cenotes (sinkhole). Results show that 1) groundwater is diluted
during the rainy season, but not for all parameters; 2) the geochemical structure of the water is obscured during
the wet season; 3) the change in values for most parameters differs from that in the cenotes, however for both,
fecal coliforms are higher in the wet season; 4) the relationship of land use to water quantity is unclear, however
some influence of agriculture and urban were observed, and 5) water quality is at risk in both dry (NO3−) and
wet (fecal coliforms) seasons. It is hypothesized that these results may be explained by changes in dominant flow
paths between dry (matrix, fracture flow) and wet (conduit flow) seasons. In addition, results suggest that rock-
water interactions, hydrology, and land cover maybe more important in understanding the influence on water
quality than land use. Clearly, understanding the impact of both land use and season on groundwater quality in
any karst environment is challenging. Further work should include using different methods for defining land use
need to be explored and a more diverse set of water quality measurements need to be made.

1. Introduction water quality influences in karst aquifers.

Karst terrains comprise limestone bedrock that has typically
weathered to a state allowing for channelized flow of groundwater and
a collapsing of the land surface, forming sinkholes (cenote in Spanish)
(Cvijić, 1893; Schmitter-Soto et al., 2002). Karst aquifers occupy
12–25% of the Earth's surface with an estimated 25% of the world's
population relying on water from such formations now (Brahana,
2007a; b), and the potential for significant increased use in the future
(Guo and Jiang, 2009). Regardless of uncertainty in these estimates,
there is a compelling need to bolster the current dearth of knowledge on
Most geologic and hydrogeologic research in karst environments
(e.g., Brahana, 2007a; b; Hollingsworth et al., 2008) has been aimed at
understanding processes and rates of limestone dissolution (Butscher
et al., 2011), especially those related to global warming (Moore et al.,
2010; Ford and Williams, 2007; Guo et al., 2005) and optimization of
non-Darcian flow models (Rehrl et al., 2008). While there is a sig-
nificant body of knowledge on karst hydrogeology, a process-level un-
derstanding of water quality influences that may be relevant to human
health, is severely lacking.
In karst settings, the presence of microbes from human waste has

∗
Corresponding author. Department of Earth and Environmental Sciences, Michigan State University, East Lansing, MI, USA.
E-mail address: long@msu.edu (D.T. Long).

https://doi.org/10.1016/j.apgeochem.2018.09.020
Received 28 March 2018; Received in revised form 25 September 2018; Accepted 27 September 2018
Available online 04 October 2018
0883-2927/ © 2018 Elsevier Ltd. All rights reserved.
D.T. Long et al.
been widely reported (e.g. Meusburger et al., 2006; Stambuk-
Giljanovic, 2006; Alcocer et al., 1998; Tranter et al., 1997). Chemical
contaminants from human activities have also been found, including
estrogen (Wicks et al., 2004), soap and bleach (Smith and Ley, 2009),
agricultural fertilizers (Guo and Jiang, 2009), mining (Hg-Deng et al.,
2011 and Cr-Yolcubal and Alyol, 2007), pesticides (Polanco-Rodríguez
et al., 2015) and petroleum products (Fels, 1999). Contaminant trans-
port in karst is fundamentally different from the traditional Darcian
processes that occur in porous media. For example, particle transport is
significant (Schwarz et al., 2011), filtration is limited (Domenico and
Schwartz, 1990), and rapid flushing occurs following precipitation
events, which has been believed to dilute contaminants. However, in-
creased microbial levels have been observed after heavy rainfall and
flood events (Dura et al., 2010; Boyer and Kuczynska, 2003). Chemical
contaminants could also be introduced into the aquifers from such
events since filtration is limited. Furthermore, flow pathways and the
timing of sampling have been shown to both increase and decrease
solutes during storm events (Huebsch et al., 2014). Due to the lack of
knowledge and the complexity of karst systems, in this study, we ex-
amined potential influences on water quality by season and land uses in
the karst aquifer of the State of Yucatán, Mexico.
2. Background and hypotheses
The study area comprised sites proximate to Mérida in the State of
Yucatán, Mexico (Fig. 1; area ∼40,000 km2), whose bedrock is mainly
Eocene and Pliocene with coastal Holocene sediments (Gondwe et al.,
2011). Dolomite and evaporites occur locally as well as gypsum and
anhydrite ejecta from the Chicxulub impact (Perry et al., 2009). The
climate is tropical with a wet season (typically June–October) which is
influenced by strong storms and hurricanes (Few et al., 2008; Farfán
et al., 2014). The State of Yucatán contains a vast groundwater reserve,
which is subject to limestone and dolomite dissolution and salt-water
intrusion (e.g., Back and Hanshaw, 1970; Back et al., 1986). Recent
studies have also identified human-related contaminants in the
groundwater, including NO3−, fecal coliforms and metals (Rocha et al.,
2015; Sanchez et al., 2015; Arcega-Cabrera et al., 2014, 2017; Derrien
et al., 2014; Arcega-Cabrera and Fargher, 2016).
The aquifer is an unconfined flat-lying karst landscape with abun-
dant cenotes, and little surface drainage (Escolero et al., 2000; Graniel
et al., 1999; Lesser and Weidie, 1988). The topographic high is in the
Fig. 1. Geologic map (Gondwe et al., 2011) of the state of Yucatan showing the sampling localities. See Table 1 for more information on the sampling sites.
266
Applied Geochemistry 98 (2018) 265–277
southwest (water Table 5-10 m above sea level) and slopes downward
toward the coast (Bauer-Gottwein et al., 2011). This dominant flow
path is locally influenced by conduits formed by dissolution, especially
along the Ring of Cenotes and the Ticul fault (Bauer-Gottwein et al.,
2011; Perry et al., 2002) (Fig. 1). As a result, groundwater chemistry
does not appear to change or evolve along a particular flow path, but
instead appears to be largely controlled by water-rock reactions (Back
and Hanshaw, 1970; Perry et al., 2002) and influenced by local factors.
The thin soils of the State of Yucatán are terra rossa-like consisting of
mainly Leptosols, Cambisols, Luvisols, and Vertisols (Bautista et al.,
2011, 2015; Krasilnikov et al., 2013). Thus, the aquifer can be easily
contaminated.
The State of Yucatán includes 106 municipalities, and the largest
urban area is the city of Mérida (greater metropolitan population ∼1
million). Landuse/landcover is characterized mainly the intermix of
agricultural activities and undeveloped areas (INFOFLR, 2016; Delgado
et al., 2010). Animal farming tends to be concentrated regionally, with
pig farming in the center of the state and cattle farming in the north
(INFOFLR, 2016; Delgado et al., 2010, 2011). A longitudinal study
(1970–1991) showed that wells in Mérida were highly contaminated by
human waste (Graniel et al., 1999). Other research identified other
water quality issues related to pesticides, fertilizers, NO3−, fecal coli-
forms, metals and untreated wastewater (Rocha et al., 2015; Sanchez
et al., 2015; Arcega-Cabrera et al., 2014, 2017; Arcega-Cabrera and
Fargher, 2016; Gonzalez-Herrera et al., 2014; Pacheco et al., 2000;
Escolero et al., 2002; Li et al., 2010). In Mérida, contamination from the
effects of urbanization such as untreated sewage, industrial effluent,
and landfill leachate, have led to the upper 20 m of the aquifer being
deemed unfit for human consumption (Graniel et al., 1999; Escolero
et al., 2002).
Several studies on the influence of the wet season on water quality
in the State of Yucatán help set the stage for this study. Pacheco et al.
(2001) focused on the role of agriculture on groundwater quality. They
measured NO3−, K+, and Cl− concentrations monthly in 12 municipal
wells over the period April 1992 to March 1993. Municipal wells are
pumped to supply water to local villages and are typically deeper (e.g.,
30 m) than cenotes and differ in that they are not directly open to the
atmosphere. These authors concluded that the source for both elevated
NO3− and K+ was from agricultural activities. Our study on municipal
wells includes five sampling sites of Pacheco et al. (2001). Castro et al.
(2018) studied the influence of wet season on karst water quality in the
D.T. Long et al. Applied Geochemistry 98 (2018) 265–277

Table 1
Details of sampling sites including well depths, location, land use, and cluster number (see Results and Discussion).
Site Village Depth (m) Latitude Longitude Land use Cluster Number

May October

1 Peto 40 20.125347° −88.921867° crop 1 4

3 Tzucacab 32 20.079563° −89.051687° crop 1 4
5 Tzucacab 32 20.067570° −89.051890° crop 1 4
6 Tekax 35 20.208224° −89.286216° crop 2 5
7 Tekax 70 20.217270° −89.305041° crop 3 6
8 Oxkutzcab 60 20.315802° −89.433785° crop 2 5
9 Ticul 38 20.396766° −89.556712° crop 2 5
10 Telchac puerto 5 21.338606° −89.262872° live stock 2 6
11 Dzilam gonzalez 13 21.277303° −88.920699° live stock 3 5
12 Dzilam gonzalez 30 21.280980° −88.921355° live stock 2 5
13 Dzilam gonzalez 21 21.281533° −88.921431° live stock 3 6
14 Buctzotz 24 21.201310° −88.787991° live stock 3 6
15 Tizimin 40 21.135554° −88.161862° live stock 2 5
16 Tizimin 40 21.137610° −88.132960° live stock 3 6
17 Sucila 20 21.164269 −88.312693 live stock 2 6
18 Montes de ame 18 21.032084° −89.620230° urban 2 6
19 Konchem 12 21.103809° −89.661704° urban 3 5
20 Caucel 44 20.999829° −89.709787° urban 2 6
21 Uman 40 20.877301° −89.757645° urban 2 6
22 Dzununcan 18 20.858959° −89.654078° urban 2 6
23 Kanasin 35 20.926631° −89.553737° urban 3 6

Table 2 Delgado et al. (2010) sampled 113 municipal wells covering the
Climate data for Cancún and Mérida for 2016. Data are from https://www. state of the Yucatán to investigate groundwater quality for irrigation.
wunderground.com/. Samples were collected during the wet season: July to October 2003.
Month Cancun (Aduana, Mexico) Merida (Lic Manuel Crecencio Spatial maps were created for the concentrations of Ca2+, Mg2+, Na+,
Rejon) K+, HCO3−, SO42−, NO3−, Cl−, and specific conductance across the
state of the Yucatán. Results from the Delgado et al. (2010) study are
Total Average Total Average
used to help interpret the results from this study.
Precipitation temperature Precipitation temperature
(mm.) o
C (mm.) o
C Building on the foundational work of Edmonds and others (as
summarized in Edmunds, 2009), we designed a study to determine the
January 59.7 23 2.55 24 influence of wet season on groundwater quality in the state of the Yu-
February 33.0 22 6.35 24 catán. We also consider that human activities on the landscape (e.g.,
March 35.6 26 16.8 28
April 40.6 27 3.3 30
agriculture, urbanization) can influence subsurface water quality;
May 10.1 28 33.3 31 therefore, a second objective of this study was to determine if there is a
(sam- relationship between the groundwater geochemistry and land use.
pling) These overarching hypotheses guide this research and are based on the
June 127.0 28 138.0 30
previous work discussed above:
July 78.7 28 42.4 30
August 138.0 28 90.2 30
September 72.4 28 110.0 29 1. groundwater will be diluted in the rainy season,
October 72.4 26 105.0 27 2. not all water quality parameters will decrease during the rainy
(sam- season due to the influence of flushing off the land surface (e.g.,
pling)
NO3− and fecal coliform concentrations are expected to increase),
3. the chemistry of the shallower cenote water will be more affected by
northwest region of the Yucatán. The area encompassed the city of surface runoff than the deeper municipal wells during the wet
Mérida. They identified the influence of seawater intrusion, water-rock season, and
reactions, and human contamination on water quality. They also ob- 4. different land uses will impart unique biogeochemical fingerprints
served that one of the impacts of wet season was the dilution of the on groundwater chemistry.
groundwater.
Perez-Ceballos et al. (2012) and Arcega-Cabrera et al. (2014) took Hypothesis 4 originates from our work that shows that different
dry and wet season samples from sites within the Ring of Cenotes. land uses can impart unique biogeochemical fingerprints (relative
Perez-Ceballos et al. (2012) found, for the most part, NO3− con- abundances of dissolved solutes) on river water quality (Fitzpatrick
centrations increased in the wet season, Cl− was diluted or did not et al., 2007; Wayland et al., 2003).
change, Mg2+ and K+ were highly variable, Ca2+ mostly decreased,
and Na+ and specific conductance mostly remained the same or de- 3. Methods
creased. They interpret the differences in water quality amongst the
cenotes to be due to hydrology and sourcing, e.g., intrusion and che- Water samples were collected from twenty municipal wells and one
mical variability associated with groundwater flow. Arcega-Cabrera individual household well (Fig. 1) representing three broad land use
et al. (2014) obtained samples during the 2011 wet season (October) types. These are urban (in and around Mérida), crop agriculture in the
and 2012 dry season (May). They observed the flushing of fecal sterols south, and livestock farming in the north (Table 1) that are defined
into the Ring of Cenotes during high rainfall events. Here we compare based on previous studies (Delgado et al., 2010; Perez-Ceballos et al.,
our results on water quality changes in municipal wells as a function of 2012; and Pacheco et al., 2001). Five of the municipalities (Tekax,
wet and dry seasons to their results on water quality changes in cenotes. Tzucacab, Peto, Oxcutzab, Ticul) were sampled in the nitrate study

267
D.T. Long et al. Applied Geochemistry 98 (2018) 265–277

Table 3
Percent change in water quality parameters between the dry (May) and wet (October) seasons.
Site pH SPC Temp Ca Mg Na K alkalinity Cl SO4 NO3 Fecals

Crop 1 10 −20 −2 −15 −26 −37 −18 −21 −19 13 −42 100
3 19 −13 −5 −34 2 −51 −2 −4 −1 320 −10 400
5 11 −14 −4 −37 -<1 −45 11 2 −8 420 −47 0
6 13 −19 1 −35 −8 −65 9 −4 −12 −94 −19 1400
7 11 −15 1 −28 −8 −53 0.00 −7 2 −66 −43 400
8 7 −24 <1 −26 −41 −9 0.00 −7 30 −37 −34 100
9 9 −24 −4 −23 −44 −32 50 2 0.00 −62 −13 100
Livestock 10 10 −13 3 −18 −20 −13 50 −7 −5 100 −29 600
11 7 30 −3 −23 5 2 180 −7 34 −62 −30 0
12 8 −12 −5 −28 12 −46 −11 −9 8 −89 −71 18000
13 5 −10 < -1 −23 86 −48 50 −4 1 −86 −32 300
14 12 −12 −4 −37 < -1 −33 −18 −9 14 −94 −24 0
15 10 −12 −2 −50 21 −44 50 2 −13 −91 −37 100
16 8 −11 −3 −21 −10 −36 200 −19 2 −85 −15 100
17 9 −14 −19 −30 −8 −16 −33 −23 −6 −66 −15 160000
Urban 18 8 −10 −3 −3 −53 −27 50 6 −3 −79 −13 250
19 8 −9 −4 −18 −26 −23 −14 −12 6 −38 11 400
20 9 −28 −9 −5 51 12 −25 −7 22 −93 −16 1600
21 8 −25 −4 3 −8 −23 −36 6 27 −89 −14 100
22 8 −10 −1 −3 −81 10 0.00 −9 −24 −58 −20 1000
23 6 −13 <1 −35 19 −39 −25 −7 −23 −20 −34 100
average 9 −13 −3 −23 −10 −29 22 −6 2 −17 −26 8900
median 9 −13 −3 −23 −8 −33 0.00 −7 0.00 −66 −24 250
decrease 20 15 20 16 18 7 17 9 17 20
no change 4 1 3
increase 21 1 6 1 7 3 10 4 11 4 1 19

Table 4
Mean values and one-way analysis of variance for selected solutes as a function of land use. Bold P values indicate significance at the 0.05 level.
Group May

Land Use Cl− Na+ K+ SO42- NO3− Cl/K Na/Cl

1 Crop 1 433 318 18.3 113 102 23.6 0.73

2 Crop 2 223 157 6.14 104 48.0 36.1 0.70
3 Livestock 234 125 3.02 37.9 42.0 77.0 0.53
4 Urban 183 102 3.84 37.9 35.1 47.8 0.56
P value Groups 1,2,3,4 < 0.001 0.619 < 0.001 0.004 0.017 0.366 0.004
P value Groups 2,3,4 0.35 0.445 0.26 < 0.001 0.636 0.354 0.099

Group October

Land Use Cl− Na+ K+ SO42- NO3− Cl/K Na/Cl

1 Crop 1 400 174 16.9 275 71.09 23.7 0.39

2 Crop 2 229 92.3 6.73 37.5 36.43 34.1 0.41
3 Livestock 241 87.8 3.71 11.9 27.9 64.9 0.38
4 Urban 185 84.9 3.32 13.5 30.6 55.7 0.46
P value Groups 1,2,3,4 0.01 0.01 < 0.001 < 0.001 0.04 0.21 0.43
P value Groups 2,3,4 0.43 0.954 0.17 0.06 0.69 0.39 0.32
seawater 49.9 0.65
halite 0.57
EPA 50

(Pacheco et al., 2001). Samples were collected 23–26 May and 18–21
October, 2016 representing dry and wet seasons respectively (Table 2).
Wells were pumped until a stable water temperature was obtained. Any
water treatment systems were bypassed. Temperature, specific con-
ductance, pH, and GPS coordinates were measured on-site. Cation
samples (Ca2+, Mg2+, Na+, K+) were filtered through 0.45 μm mem-
brane filters (Millipore) filters, acidified to pH < 2 with high-grade
nitric acid, and stored on ice for transport to the laboratory. Anion
samples (Cl−, SO42−, alkalinity, NO3−) were only filtered and also
stored on ice for transport. Bottles for the solutes were rinsed three time
prior to collection. In the laboratory, the solute samples were re-
frigerated until analyzed, which was typically within two days. Mi-
crobial samples for fecal coliforms determination were collected in
sterilized glass bottles, which were rinsed three times prior to
collection, stored on ice in the field. Analyses were started immediately
after samples were returned to the laboratory. All analyses were per-
formed using Standard Methods (http://www.standardmethods.org/
store/).
The solute and microbial data were reduced by solute-solute plots,
Piper plots, geochemical modelling, ratio analyses, and multivariate
statistical analysis (e.g., Sanchez et al., 2015), Pacheco et al., 2001;
Perez-Ceballos et al., 2012; Mohammadi, 2009; Arcega-Cabrera et al.,
2014; Castro et al., 2018). PHREEQC (developed by the U.S. Geological
Survey), using the WATEQ4F database was used for the geochemical
modelling (Parkhurst and Appelo, 1999; Ball and Nordstrom, 1991;
Long et al., 2012). Statistical analyses were performed using JMP5 (SAS
Institute) and StatMost (Dataxiom Software) (e.g., Long et al., 1992;
Wayland et al., 2003; McGuire et al., 2005; Fitzpatrick et al., 2007). For

268
D.T. Long et al. Applied Geochemistry 98 (2018) 265–277

Table 5 NO3− concentrations in groundwater in all land use types (Table 4). In
Comparison of the changes in water quality parameters between cenotes and Table 4, crop land use was split into two groups. The first group (Crop
municipal well due to the wet season. Majority means > 70% of the samples 1) comprises those sites characterized by geochemical cluster 1 and the
for a parameter either had either an increase or decrease in concentration be- second group (Crop 2) the remainder of the sites considered to be crop
tween the dry and wet season.
land use in Table 1. The mean concentrations are above the U.S. En-
Parameter Cenotes (Perez-Ceballos Municipal wells (this study) vironmental Protection Agency drinking water maximum contaminant
et al., 2012) N = 21 level of 10 mg/L (as N). The higher values for the Crop 1 group samples
N = 22
were dominated by high values from Peto and Tzuacab, similar to the
pH About equal mix of increase All increase observations by Pacheco et al. (2001). The decrease in NO3− con-
& decrease centrations is consistent with the results of Pacheco et al. (2001) who
Temp Variable increase, decrease, Majority decrease found concentrations to be lowest in the wet seasons. Their samples
same
were taken from the agricultural area in the south where we also col-
SPC Mostly same, some decrease Decrease, except 1
Ca2+ All decrease All decrease
lected samples (Peto, Tzucacab, Tekax, Oxkutzcab, and Ticul; Table 1).
Mg2+ Mix of increase & decrease Mix of increase & decrease, but They attributed the NO3− concentrations to be from fertilizers and the
more decrease lower concentrations in the wet season to result from dilution. We see
Na1+ Mostly same, some decrease Majority decrease no reason for any different interpretation. Pacheco et al. (2001) col-
K1+ Mostly same, some increase Variable increase, decrease,
lected their samples over a year (1992 and 1993) and thus the similarity
& decrease same
Alkalinity About equal mix of increase Majority decrease in results between this study and theirs can be interpreted to indicate
& decrease that the landscape processes (e.g., sources, wet season) influencing
Cl− Majority decrease About equal mix of increase & NO3− have remained much the same for a considerable period of time.
decrease
SO42- Mostly same or decrease Mostly decrease, few increase
NO3− Mostly increase, few Decrease, except 1
4.1.2. Chloride, potassium, sodium, and sulfate
decrease Potassium concentrations of 2–4 mg/l are typical for groundwater
Fecal coliforma Majority increase Majority increase (Long et al., 2017) and we have found that they are often elevated
a
across the State of Yucatán (Table 4). Concentrations are most elevated
Arcega-Cabrera et al., 2014. in the crop land use (Table 4), but all land uses have some sites with
elevated K+ concentrations. Sea salts and seawater can be sources,
the geochemical modelling, we focused on the state of equilibrium of however Escolero et al. (2000) and Perez-Ceballos et al. (2012) con-
the groundwater with respect to calcite, which is the major mineral of cluded from a study of Cl/K ratios that elevated K+ most likely results
the aquifer. The state of equilibrium was calculated as the log10 ratio of from fertilizers and possibly wastes. This is consistent with our findings
the activity product of cation and anion divided by the equilibrium for sites associated with crop land use (Table 4), however, we also find
constant for the dissolution of the minerals of interest. The resulting sites that had Cl/K ratios below that of seawater in all land use types.
number represents the saturation index (SIm) for the mineral where Elevated SO42− concentrations in karst ground water have been
positive values reflect supersaturated conditions and negative values interpreted to be at least partially due to human activities (e.g., agri-
undersaturated. culture, waste) (Long et al., 2012). The mean values in both crop groups
Spatial land use determination was obtained from data provided by are significantly elevated above those in the livestock and urban groups
the Instituto Nacional de Estadística y Geografía (INEGI, 2014) for for May (Table 4). In October, concentrations in Crop 1 are significantly
2013. Data were translated into English and processed in ArcGIS v10.5 higher than the others. Concentrations in Crop 2 are also higher than
(ESRI, Redlands, CA). We then generated Euclidean buffers around each Livestock and Urban, but not significantly. Except for Crop 1 and 2 in
site coordinate at 500 m intervals, in order to characterize the land uses May and Crop 1 in October, mean concentrations are not elevated.
at varying distances around each sampling location. For each buffer, we For both May and October, mean Cl− concentrations are sig-
then calculate the percentages of each buffer's area occupied by various nificantly higher in the Crop 1 group than the others are.
land uses using ArcMap v10.5. Concentrations in the Urban group are the lowest but not significantly
different from the Crop 2 and Livestock groups. Major sources for Cl−
4. Results and discussion (and Na+) in the State of Yucatán are halite (both rock and human-
sourced), sea salt spray, and seawater intrusion) (e.g., Escolero et al.,
4.1. What are the changes in water chemistry between dry and wet seasons? 2000). Sodium-chloride ratios are often used to understand the source
for elevated Cl− (Long et al., 2017). In May, there is a significant dif-
Except for one sample, the specific conductance decreased in the ference between the two Crop groups but both have ratios closer to
wet season and the responses of other water quality parameters were seawater than halite (Table 4). The Livestock and Urban ratios are
variable (Table 3). These results are consistent with the first two hy- consistent with halite sources. In October, all the land use ratios are
potheses. All sites had increases in pH, while calcium concentrations similar and fall below that for either halite or seawater.
decreased. Except for one sample, NO3− values decreased. The number Thus, sources for these solutes (e.g., halite, gypsum, salt-water in-
of sites in which Cl− and K+ concentrations increased was slightly trusion, fertilizers, wastes) and processes influence inputs (e.g., differ-
more than those where it decreased. Decreasing concentrations is ential recharge zones; hurricanes) would be quite variable in this
mainly observed for the other solutes (Mg2+, Na+, K+, Cl−, alkalinity, system. We consider that it is this variability that leads to the compli-
SO42−) and temperature. In general, total coliforms and fecal coliforms cated responses of these solutes to the change from dry season to wet
concentrations increased, albeit total coliforms were more variable. In season (e.g. Graniel et al., 1999; Pacheco et al., 2001; Escolero et al.,
the sections below, we consider four trends that are associated with 2000; Escolero et al., 2005; Delgado et al., 2010; Long et al., 2012;
NO3−; Na+, K+, Cl−, and SO42−; fecal coliform, and pH, Mg2+, Ca2+. Arcega-Cabrera et al., 2014; Zhou et al., 2011; Knierim et al., 2015; He
et al., 2016).
4.1.1. Nitrate
Nitrate is also normally elevated in the groundwater of the Yucatán 4.1.3. Fecal coliforms
and the high concentrations have been linked to human activities, Various studies have observed increases in microbial pollution
particularly those associated with agricultural practices, especially during wet seasons and storm events (Reed et al., 2011; Arcega-Cabrera
fertilizer use (e.g., Perez-Ceballos et al., 2012). We found elevated et al., 2014; Sinreich et al., 2014; Knierim et al., 2015; He et al., 2016).

269
D.T. Long et al.
Fig. 2. Solute plot showing the relationship between Ca2+ and Mg2+ concentrations. A. May samples and B. October samples.
The increase is attributed to the flushing of the microbial communities
from surfaces sources, typically human and animal wastes, as reported
in Arcega-Cabrera et al. (2014) for cenotes. We believe this same
phenomenon to be responsible for the elevated bacterial concentrations
(Table 3.) observed in wells in this study. As far as we know, this is the
first report of evidence for this process in the deeper municipal wells of
the state of the Yucatán. This contradicts a commonly held belief in the
region that these wells draw from a deeper aquifer that is largely un-
affected by pollution because it is isolated from a surface aquifer that
supplies the cenotes.
4.1.4. Alkalinity, calcium, magnesium, and pH
One difference between these parameters and the previous biogeo-
chemical parameters is that the geochemistry of Mg2+, Ca2+, and al-
kalinity will be more influenced by rock-water interactions than vari-
able anthropogenic inputs (Long et al., 2012). Typically, Mg2+ and
Ca2+ and concentrations are correlated in carbonate systems because of
common sourcing from calcite and dolomite minerals and this corre-
lation was observed for the samples taken in May (Fig. 2a). However, in
October, Mg2+ and Ca2+, were no longer correlated (Fig. 2b). Calcium
concentrations at all sites decreased, except for one increase, while at
sixteen sites, magnesium concentrations decreased and at seven, con-
centrations increased. Dilution might partially explain the decreases in
Mg2+ and Ca2+, but the percentage dilution for each is not the same
and Mg2+ concentrations increased at seven sites. Thus, we consider
three other mechanisms that could account for these observations.
One long-accepted mechanism for the decoupling of Mg2+ and
Ca 2+
concentrations in karst systems is dedolomitization (Bischoff
et al., 1994). The process is driven by gypsum dissolution in a system in
which calcite and dolomite are near equilibrium and gypsum (or an-
hydrite) is undersaturated. The dissolution of gypsum causes the dis-
solution of dolomite and subsequent precipitation of calcite through the
following reaction:
MgCa(CO3)2 + CaSO4·2H2O = 2CaCO3 + Mg2+ + SO42− + 2H2O
Magnesium is a relatively conservative solute and will only pre-
cipitate out of solution at very high pH levels or when a high degree of
evaporation has occurred. Dolomite would not be expected to pre-
cipitate under the conditions observed in these samples, even though
they are supersaturated (Land, 1985). Therefore, as this reaction pro-
ceeds, Mg2+ concentrations will increase and this could explain the
trend observed here. Calcite and dolomite were in near equilibrium in
the May samples but became supersaturated in October (Fig. 3a and b).
Gypsum was undersaturated in May and became more undersaturated
in October (Fig. 3c), thus the potential for the dissolution of gypsum
270
Applied Geochemistry 98 (2018) 265–277
increased. It should be noted that these SIm values are not unusual for
karst systems (Mohammadi, 2009). Sulfate concentrations did increase
at Peto, and both Tzucacab sites (Table 3). Gypsum beds are found in
this area of the Yucatán causing higher SO42− concentrations in the
aquifer (Delgado et al., 2010). Thus, it is possible that this reaction is
responsible for the observed Mg2+ increases.
Another mechanism decoupling Mg2+ and Ca2+ is the enhanced
dissolution of magnesium-rich limestone due to an increase ionic
strength of the groundwater (Long et al., 2012). Dissolution of this
mineral increases the dissolved Mg2+ concentrations, while solution-
mineral equilibrium controls the concentration of Ca2+. However, for
this mechanism to occur, the interacting solution needs to become more
concentrated to increase its ionic strength, for example, from the input
of anthropogenic chemicals or salty water. However, only one sample
(site 11. Dzilam Gonzalez) had a concentration increase, apparently
from increases in Cl− concentrations (Table 3). Magnesium con-
centrations did increase in this sample and it is possible that this me-
chanism caused the increase.
Finally, the decrease in Ca2+ concentration could be caused by
calcite precipitation and the Mg2+ increase caused by the flushing of
the deeper water mass. The increase in the degree of calcite super-
saturation in October (Fig. 3a) would work to facilitate the precipita-
tion of calcite and the removal of Ca2+. The increase in SIc is driven by
the increase in pH, which occurred at all sites. Decreased temperatures
might be one cause of the pH increases, however not all temperatures
decreased. This might suggest that interaction with another water mass.
Sanchez et al. (2015) reported dissolved Mg2+ increases in a karst
aquifer during high flow. They attributed the increase to input of
groundwater with greater water-rock (dolomite) interaction (e.g., dif-
fuse flow system). They argue that a long residence time would allow
more time for the dolomite to come into equilibrium with the
groundwater.
In sum, it is possible the dedolomitization and enhanced dissolution
might explain the decoupling of Mg2+ and Ca2+ concentrations in the
wet season, albeit on a local scale. On a larger scale, the precipitation of
calcite and flushing of a Mg2+ rich water mass should be considered
along with dilution. We submit however, that more studies are neces-
sary that includes a larger database (e.g., sample number, parameters
such as stable isotopes) are needed to fully understand the role of de-
coupling.
4.2. What are the changes in water chemistry between cenotes and
municipal wells between dry and wet seasons?
In comparing the changes in water chemistry between the cenotes
D.T. Long et al.
Fig. 3. Frequency histograms showing the season changes in the saturation indices of A. calcite, B. dolomite and C. gypsum. Red bars are for the May samples and
black bars are for the October samples. The shaded box is the error envelope for considering equilibrium. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
(Perez-Ceballos et al., 2012) and municipal wells, one needs to consider
the difference between the samples that could influence these changes.
The municipal well samples are collected from the aquifer at a specific
depth interval defined by the well screen. The cenotes on the other
hand are bodies of water much like lakes: water may originate from
multiple depths and both be mixed and stratified. The result is that
water chemistry can be quite variable as a function of depth. In the
study by Perez-Ceballos et al. (2012), vertical samples were taken and
the chemical results were integrated.
Regardless of these differences, for six of the twelve parameters
compared, the change between dry and wet seasons were the same for
both the municipal wells and the cenotes (Table 5). These parameters
include Ca2+, whose concentrations decreased at all sample sites, ex-
cept in one municipal well. Fecal coliforms concentrations increased at
a majority of sites; and Mg2+, K+, and SO42− whose concentration
changes varied (i.e., both increased and decreased). For the solutes, we
assume the explanation for the behaviors discussed above might also
apply for their behaviors in the cenotes. The changes in fecal coliforms
are discussed below.
The parameters that exhibited clear differences in responses be-
tween wells and cenotes were pH, temperature, specific conductance,
Na+, alkalinity, Cl− and NO3−. The pH values in all municipal well
samples increased from the dry to wet season. There was an approxi-
mately equal mix of increases and decreases in the cenote samples.
Except for one well, the specific conductances decreased, but were
mostly unchanged in the cenotes. The majority of the municipal well
water temperatures and concentrations of Na+, and alkalinity de-
creased. For the cenotes, the changes in temperature and alkalinity
were variable with little change in Na+. In the cenotes, the majority of
Cl− concentrations decreased while in the wells there was about an
equal mix of increases and decreases. Finally, in the cenotes, NO3−
concentrations mostly increased while in the wells, except for one,
decreased.
The differences in temperature could be attributed to differences
resulting from sampling, i.e. variation in the vertical temperature gra-
dient in the cenotes, versus the fixed sampling depth in the wells.
However, the differences in the other parameters may not relate to
sampling, but are more likely related to the importance of surface
runoff and pathways of transport (e.g., Huebsch et al., 2014).
For example, fecal coliforms were flushed into both the cenotes and
the municipal wells, but only NO3− was flushed into the cenotes. The
municipal wells are clearly diluted (i.e. decreased specific conductance)
while in the cenotes, the change in the specific conductance is variable.
Chloride in the wells was diluted but its concentrations both increased
and decreased amongst the cenotes.
Most of these observations might be explained if one were to assume
that this karst system is characterized by a simple dual-flow system;
water flow through conduits and through fractures and pores (e.g., Ford
and Williams, 2007). During the wet season, conduit flow would allow
rapid transfer of the water into the deeper municipal well system with
271
Applied Geochemistry 98 (2018) 265–277
less interaction with the near surface system, hence the presence of
colder more dilute water. Fracture flow, which is more diffuse than
conduit flow and surface runoff would have more influence on the
shallower cenote water chemistry during the wet season (e.g., variable
specific conductance, Cl− decrease, NO3− increase).
If the dual-flow system mechanism was influencing the transport of
the fecal coliforms, one might expect a behavior similar to NO3−, in
which the wells would show no increase in concentrations. Since this is
not observed, we suggest that sourcing for the concentration changes in
fecal coliforms is more important than transport pathway. The primary
sources for fecal coliforms are human and other animal wastes and this
indicator of contamination is frequently found to degrade water quality
in karst formations in populated settings (e.g., Grimmeisen et al., 2016).
By their nature, municipal wells in the study area are in villages, ty-
pically with no wastewater treatment plants (Escolero et al., 2000;
González and Rodríguez, 2006). Thus, the potential exists for localized
high concentrations of coliforms that can be flushed into the deeper
aquifer. This potential would also exist for cenotes located near villages.
The cenotes would also be susceptible to near surface flushing of other
animal wastes.
4.3. What is the biogeochemical structure of the water?
Multivariate statistical techniques (R-mode factor and cluster ana-
lyses) and a graphical technique (Piper diagrams) were used to de-
termine if there is a systematic pattern amongst the solutes or biogeo-
chemical finger print (Fitzpatrick et al., 2007) and if and how this
pattern differs from dry season to wet season. R-mode factor analysis
explores the relationship amongst water quality parameters. Calcula-
tions were done on a correlation matrix using JMP5 (SAS Institute) and
employed principal components analysis retaining those factors with
eigenvalues values greater that one for a varimax rotation (Wayland
et al., 2003). The microbial data did not correlate with any solute
parameter in either the May or October data sets, which is consistent
with the discussion on the changes in their concentrations between dry
and wet season. Therefore, we only studied the relationships among the
solutes.
Three factors were extracted (Table 6) based on eigenvalues greater
than one. Factor 1 defines most of the variability in the data and Factor
3, the least among the three factors. We consider parameter loadings
above ± 0.7 to be the most significant in defining the characteristic of
the factor (Wayland et al., 2003). In May, Factor 1 is characterized by
high loadings for Ca2+, Mg2+, Na+ and Cl−. Factor 2 is characterized
by K+, SO42− and NO3− and Factor 3 is characterized by pH and al-
kalinity that are off-loaded with one another. We interpret Factor 1 to
represent water-rock interactions (Ca2+, Mg2+) and influence of sea
salt and saltwater intrusion (Na+, Cl−) (Long et al., 2012) and Factor 2
to the influence of human activities (crop agriculture.) Factor 3 is re-
lated the relative behaviors of pH and alkalinity i.e., as pH decreases,
alkalinity can increase because of increased water-rock interaction.
D.T. Long et al. Applied Geochemistry 98 (2018) 265–277

Table 6 (Fig. 4) that give further insight into the nature of the geochemical
Results of R-Mode factor analysis for the May and October samples. Bold changes between the dry and wet seasons. Here we use cluster analysis
numbers signify a high loading, which is considered a value of ± 0.7 (Wayland to group water samples in terms of their chemical parameters, i.e.,
et al., 2003). Values in bold italics are the highest for the variable but not at which samples have a similar geochemical structure. Hierarchical cal-
the ± 0.7 level.
culations in JMP5 (SAS Institute) were done using Wards method on
May October standardized data.
In both May and October, three clusters are recognized in both data
Factor F1 F2 F3 F1 F2 F3
sets that are summarized for each sample in Table 1. Except for Clusters
pH −0.462 0.129 0.702 0.075 0.896 −0.015 1 and 4, the other two clusters from the May data set cannot be directly
Ca 0.972 −0.070 −0.008 0.190 −0.25 0.791 compared to the October data set because they do not comprise exactly
Mg 0.902 −0.229 −0.076 −0.077 0.403 0.742 the same sample sites. Therefore, each cluster is given a unique number.
Na 0.882 −0.376 −0.132 0.056 −0.152 0.890
Since cluster analysis does not readily reveal the chemical associations
K 0.395 −0.787 −0.334 0.657 −0.233 0.595
Alkalinity −0.113 −0.115 −0.875 −0.786 −0.419 0.182
that define each cluster, spider diagrams are used to help reveal the
Cl 0.930 −0.226 −0.106 0.1375 0.1717 0.897 associations (Fig. 5). The diagram is constructed by taking the ratio of
SO4 0.033 −0.943 0.033 0.376 −0.099 0.811 the average concentration for a parameter in a cluster to the average of
NO3 0.379 −0.766 −0.200 0.583 −0.054 0.465 the whole data set. The values of pH and temperature and alkalinity
% explained 57 17 12 50 16 11
(ALK) concentrations are similar for all clusters in May and October.
Relative to the other clusters, Clusters 1and 4 are higher in the con-
Perez-Ceballos et al. (2012) also performed principle component centrations of Ca2+, Mg2+, Na+, K+, Cl−, SO42−, and NO3−. The
analysis on their cenote water chemistry data. They extracted two fac- sample sites for these clusters are in the southern portion of the State of
tors, with the first factor characterized by conductivity, Na+, Cl− and YucatánYucatán, which as previously mentioned, are influenced by
SO42− and the second by pH, Ca2+, and HCO3−. However, the loadings evaporite deposits (e.g., halite and gypsum) and agricultural activities
of the parameters were not strong, with values defining the factors (Pacheco et al., 2000; Delgado et al., 2010).
ranging from 0.371 to 0.518. The low loadings may be due to their use In May, Ca2+, Mg2+, Na+, K+, alkalinity, and Cl− values are si-
of the combined seasonal averages for the calculations. Although they milar between Clusters 2 and 3 and SO42−, NO3−, and fecal coliform
did not make geochemical interpretations of the factors, two associa- (Fec) values differ between these clusters. We interpret the similarity
tions were similar to ours: Na+ and Cl− on Factor 1 and pH and HCO3− the chemical parameters of Clusters 2 and 3 to be the result of the si-
(alkalinity) were off-loaded on Factor 2. milarity in water-rock reactions (e.g., limestone) and influences of sea-
The factor loadings in our study changed in October (Table 6), with salts and seawater intrusion. The differences in SO42−, NO3−, and fecal
Ca2+, Mg2+, Na+ and Cl− defining Factor 3, the less important factor. coliform values can be attributed to the degree of anthropogenic
SO42− is also loaded on this factor. K+ and NO3− were equally loaded sourcing. In October, Mg2+, Na+, K+, and Cl− values differ between
on Factors 1 and 3. Alkalinity is the major parameter defining Factor 1 clusters 5 and 6 and SO42−, NO3−, and fecal coliform values between
and this was no longer correlated with pH, which was the sole para- these clusters are more similar both in values and in trend. Cluster
meter defining Factor 2. Providing a geochemical interpretation for analysis shows that the wet season specifically disrupts the water-rock
these factors is difficult. Considering the associations found in May, the signature, but homogenizes the anthropogenic parameters.
October associations are of mixed processes and weak (e.g., Factor 3) or A complimentary widely used tool to study changes in the geo-
not easily interpretable. For example, one might assume that the chemical structure of water groundwater as function of space and time
loading of K+ and NO3− on Factor 1 implies the influence of agri- (e.g., Herojeet et al., 2017) is the Piper Diagram (Piper, 1953). In ad-
culture. However, NO3− concentrations decreased and both are loaded dition to giving further insights in the biogeochemical signature of
on two factors. Therefore, we interpret the October results not in terms water (termed hydrochemical facies), patterns on Piper Diagrams can
of biogeochemical processes, but on the influence of the wet season. be used to identify water masses and processes such as the mixing of
The rainy season works to mix the water chemistry and lessen the water masses (Fig. 6a and b). For reference, rain chemistry from the
geochemical structure that developed during the dry season. area (Ceron et al., 2005), seawater chemistry (Langmuir, 1997), and
The influence of the wet season to lessen the geochemical structure karst water chemistry unaffected by sea salts (Long et al., 2012) are
of the system is also observed in the results of the cluster analysis shown.
As might be expected, the chemistry of rain in this area is dominated

Fig. 4. Cladograms for the water chemistry from the results of the cluster analyses. A. May, B. October.

272
D.T. Long et al. Applied Geochemistry 98 (2018) 265–277

Fig. 5. Spider diagrams showing the chemical associations amongst the clusters. A. May, B. October. Clusters 1 and 4, are the red dots; 2 and 5, green dots; and 3 and
6 the blue dots. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

by sea salts. Because of water-rock interactions on this karst landscape,
the cation and anion chemistries are changed from having a sea salt
signature to one in which Ca2+, Mg2+ and HCO3− become more
dominant. In karst (triangles in Fig. 6) unaffected by seawater or dis-
solution of halite, the hydrochemical facies of the waters is defined as
Ca-HCO3 because the water plots in the Ca and HCO3 dominate areas of
the cation and anion ternaries, respectively. Neither the May nor the
October data sets show a grouping of the water chemistry around the
triangles. In addition, there is no dominant hydrochemical facies, rather
the data plot across diagrams. Trends in the scatter of the data can often
be related to process such as mixing trends identified by scattering of
the data along a straight line.
The trends in the cation and anion ternaries in May (Fig. 6a) in-
dicate mixing between typical karst water and seawater. In the anion
ternary, there is more scatter due to some samples having higher SO42−
concentrations. As previously discussed, the source for higher SO42−
concentrations is most likely related to agricultural activities. The
trends for Clusters 2 and 3 overlap in the ternaries and central diamond.
Cluster 1 occurs more towards the seawater endmember in the ternary
diagrams and the central diamond shows its chemistry to be distinct
from the other two clusters. In October, the trends change. There is
more scatter of the data in the cation ternary and a mixing trend not
well defined. Clusters 5 and 6 become more dissimilar to each other,
but the geochemistry of Cluster 4 remains unique.
In sum, R-mode factor analysis showed that the geochemical
structure of the water is disrupted during the wet season and that fecal
coliform concentrations are not related to any geochemical parameter.
Cluster analysis also showed the disruption in the geochemical sig-
nature during the wet season and that this disruption was for the solutes
influences by water-rock interactions and sea salts. For the anthro-
pogenically sourced solutes, structure increased in the wet season. Piper
diagram corroborate these results but also show evidence for mixing of
water influenced by the dissolution of limestone and sea-salts/seawater.
If the results of this study are representative of the seasonal change
in the groundwater chemistry of the State of the YucatánYucatán, then
it is reasonable to assume that the geochemical structure observed for
May will be reestablished during the next dry season. If this assumption
is correct, then there must be a systematic processes causing these
changes. Here we posit that the transitions are the result of a two-phase
(conduit or direct; matrix or diffuse) flow mechanism (e.g., Martin and
Screaton, 2001; Huebsch et al., 2014). During the wet season, conduit
flow would be the dominant process causing the mixing and disruption
of the geochemical structure. During the dry season, matrix/fracture
flow dominates and work to reset the geochemical structure. More
detailed temporal sampling is necessary to better understand these
transitions and influence on drinking water quality.
4.4. What is the relationship of the water chemistry to land use?
Regardless of season, the geochemical clusters do not clearly map to

Fig. 6. Piper plots for a. May and b. October. Areas in the cation and anion ternaries in which a solute comprises greater than 50% of the total equivalent parts per
million for ion are depicted by the inverted triangle. Black circles and ovals are rain from the region, open squares are seawater, open triangles are typical karst
uninfluenced by sea salts (Long et al., 2015), red dots are cluster 1, green dots are cluster 2, and blue dots are cluster 3 for May and red dots are cluster 4, green dots
are cluster 5, and blue dots are cluster 6 for October. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of
this article.)

273
D.T. Long et al. Applied Geochemistry 98 (2018) 265–277

Table 7
Results of R-Mode factor analysis for the May and October samples including land-use type. Bold numbers signify a high loading, which is considered a value of ± 0.7
(Wayland et al., 2003). Values in bold italics are the highest for the variable but not at the ± 0.7 level.
Factor May October

F1 F2 F3 F4 F1 F2 F3

urban 0.312 0.795 0.425 −0.037 −0.222 0.777 0.378

crop −0.450 0.101 −0.796 −0.203 0.643 0.095 −0.636
livestock 0.147 −0.838 0.377 0.232 −0.418 −0.814 0.266
pH 0.513 −0.077 −0.002 −0.678 −0.112 −0.152 −0.703
Temp 0.030 0.737 −0.166 0.053 0.042 0.763 −0.542
Ca −0.969 0.017 −0.065 −0.017 0.853 0.093 0.224
Mg −0.887 −0.178 −0.300 0.034 0.513 −0.687 −0.124
Na −0.893 0.029 −0.345 0.146 0.861 −0.141 0.185
K −0.436 0.193 −0.645 0.467 0.820 0.042 −0.210
Alkalinity 0.105 −0.179 −0.088 0.830 −0.059 −0.148 0.742
Cl −0.932 −0.071 −0.170 0.128 0.801 −0.424 0.000
SO4 −0.041 0.093 −0.958 0.102 0.909 −0.015 −0.109
NO3 −0.408 0.069 −0.586 0.404 0.634 0.003 −0.162

the three land uses (crop, livestock, and urban) that were used to
characterize the area around each municipal wells (Table 1). R-mode
factor analysis (Table 7) that includes land use, also shows little cor-
relation between land use and geochemistry. For example, in May, none
of the land uses load on the most dominant factor, Factor 1. Of the
three, only crop (Factor 3) correlates with some solutes (K+, NO3−,
SO42−). Urban and livestock land uses (Factor 2) are off loaded with
respect to each other and correlate with temperature. Similar to May, in
October, urban and livestock land uses (Factor 2) are off loaded with
respect to each other and correlate with temperature and perhaps
magnesium. In October, crop loaded positively on Factor 1 and nega-
tively on Factor 3. The absolute values were similar but weaker than in
May. It has some degree of correlation to all the geochemical variables
except temperature.
The R-mode factor analysis involving land use (Table 7) assumes
that one particular land use is dominant (i.e., 100%) for each sampling
site. The interpretation of the results of the cluster analysis assumes
similarly. From our on-the-ground observations, we considered that this
classification scheme might be too broad and not accurately reflect
specific land uses that influence each municipal well. For example, in
regions where the land use is characterized as crop, the municipal wells
are often located in small villages that could be described as urban or
suburban. Therefore, we attempted to better define the types of land
use surrounding each municipal well through spatial modelling. Cal-
culations on the percentage of these three land uses were done for
buffer radii of 500, 2,000, and 5000 m. Table 8 is an example of the
results of this calculation for sites 6 and 7 that were initially char-
acterized as crop (Table 1). At the small radius (500 m), site 7 is
dominated by agriculture but for site 6, urban. As the radius increases,
the dominant land use changes with natural areas (e.g., not distinctly
urban, livestock, or crop) becoming important. Such changes were
observed at all sites. In fact, only one site we initially classified as urban
had urban as the dominant land use.
Considering this complexity in attempting to characterize land use
for a site, we picked three sites for which one of the land uses (crop
urban and livestock) was dominant at all radii. These sites were 8
(crop), 13 (livestock), and 18 (urban) (Table 9). A Spider diagram was
then constructed for these sites for the two seasons (Fig. 7). Although
there are geochemical changes between seasons, the general pattern for
each land use for is similar for both seasons (Fig. 7). Of the geochemical
parameters compared, pH, temperature, Ca2+, alkalinity do not appear
to be useful in distinguishing differences amongst the land use types.
Compared to the other two types of land use, crop is characterized by
intermediate conductivity, high Na+, K+, SO42−, SIc, values and low
fecal coliform values. The livestock land use has higher conductivity,
intermediate Na+, K+, and SIc values and low NO3− and fecal coliform
values. Urban has the lowest conductivity, low Mg2+, Na+, K+, Cl−,
and SIc values and high fecal coliform values. NO3− values for crop and
urban are similar. The spider diagram also shows that change in the
values (magnitude and direction) of the geochemical parameters be-
tween seasons differ. For example, for all three land uses, Mg2+ values
decrease from dry to wet season with livestock having the largest de-
crease and urban the least amongst the three. However, changes be-
tween seasons for Na+ (increase), K+ (decrease), SO42− (increase),
fecal coliform (decrease) and SIc (decrease) values are similar for both
crop and livestock and opposite to urban. These results suggest that
urban land use may have a unique influence and biogeochemical sig-
nature compared to crop and livestock. We recognize that this analysis
lacks robustness and needs more study.
The results from this study are inconclusive in terms of supporting
hypothesis 4, but may suggest that either 1) water-rock interactions
may be more of an influence on geochemical patterns than land use or
2) the land-use classification scheme did not adequately capture the
land use characteristics surrounding a well. It is most likely a combi-
nation of both of these possibilities. To help understand the importance
of processes during the wet season, other than water-rock, additional
biogeochemical chemical measurements (e.g., Br−, I−, B, DOC, NH4+,
PO43−, E. coli, organic contaminants, heavy metals) are needed.
Classifying the landscape as to a particular land use influencing water
chemistry is more challenging. For example, it is not clear at what
spatial scale this karst landscape influences municipal well chemistry or
the relative importance of landscape processes to that of subsurface

Table 8
Percent land use for sites 6 and 7 as a function of radius.
Land use Site 1 Radius (m) Site 6 Radius (m) Site 7 Radius (m)

500 2000 5000 500 2000 5000 500 2000 5000

% Crop 0.0 43.4 20.6 0 11.7 26.6 92.4 21.3 34.5

% Urban 98.7 22.1 3.5 94.3 36.3 6.0 7.6 10.8 6.0
% Livestock 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
% Natural 0.0 13.2 72.2 0.0 42.5 65.9 0.0 61.9 57.4

274
D.T. Long et al.
Table 9
Selected sites representing a dominant land use for all radii analyzed. Symbols are C (crop), L (Livestock), U (Urban), # (site number). Values in bold are highest
percentage landuse.
# 500 m 2000 m
C Urban Livestock Natural C Urban
C 8 66.3 0.0 0.0 33.7 51.3
L 13 0.0 14.4 46.0 0.0 0.0
U 18 0.0 96.2 0.0 0.0 0.0
lateral flow. Municipal wells, most of which are in villages are con-
sidered urban at smaller spatial scales but typically not a larger scales.
The importance of other land uses is similarly variable at all scales.
Considering these observations, it is possible that the intensities these
land uses in general are not sufficient to uniquely influence water
chemistry. Rather than land use, land cover may be the landscape
feature influencing water chemistry.
5. Conclusions
Water samples were collected from 21 municipal wells across the
State of Yucatán to study the influence of the rainy season on water
chemistry. Four hypotheses were tested:
1 groundwater will be diluted in the rainy season,
2 not all water quality parameters will decrease during the rainy
season due to the influence of flushing off the land surface (e.g.,
NO3− and fecal coliform concentrations are expected to increase),
3 the chemistry of the shallower cenote water will be more affected by
surface runoff than the deeper municipal wells during the wet
season, and
4 different land uses will impart unique biogeochemical fingerprints
on groundwater chemistry.
The results of this study support the first three hypotheses. Although
the results do not support hypothesis 4, there is some evidence that
agricultural practices and urban land use uniquely influence water
chemistry. The results also show that understanding the influence of the
rainy season on water quality is challenging. For example, the well
water was diluted during the rainy season and temperatures decreased,
which would be expected. Nitrate concentrations in the cenotes in-
creased during the rainy season and, decreased in the wells, which is
consistent with a more surface source influence on the water quality in
the cenotes than in the wells during this season.
However, microbial concentrations increased in both the cenotes
and wells during the wet season. All Ca2+ concentrations were diluted
in the wells in the wet season, but not all Mg2+ were, thereby decou-
pling the behaviors of calcium and magnesium during the wet season.
Na/Cl ratios decrease in the wet season to values less than that of a
seawater or halite source. Saturation indices increase during the wet
season. Although we suggest possible causes for most of these
Applied Geochemistry 98 (2018) 265–277
5000 m
Livestock Natural C Urban Livestock Natural
15.8 0.0 32.6 56.1 6.2 0.0 34.3
10.9 75.8 0.0 8.4 1.7 74.4 13.4
64.6 0.0 2.8 2.6 43.3 0.0 12.5
Fig. 7. Spider diagram for selected sites representing
Crop (red), Livestock (green), and Urban (blue) land
sues. Dashed line is for the May (dry) season and the
solid line is for the October (wet) season. Additional
variable include conductivity (cond) and calcite
disequilibrium index (SIc). (For interpretation of the
references to colour in this figure legend, the reader
is referred to the web version of this article.)
observations, other explanations cannot be eliminated as the system
and processes are complex.
It is clear from these results and those of others that changing sea-
sons influences water quality, and that the result cannot be explained
simply as just dilution by wet season precipitation producing improved
water quality. For example, microbial concentrations are higher in the
wet season but NO3− concentrations are higher in the dry season. It is
also evident that different karst landscapes or different systems on the
same karst landscape (e.g., shallow, deep) respond differently to sea-
sonal changes.
One influence on changes in water quality between the dry and wet
season, we feel arises from this work that deserves more study in karst,
is that of changes in flow path. More specifically, it is the change in the
relative importance of matrix flow and conduit flow between seasons.
We suggest that this change in flow path for the State of Yucatán system
is related to our observations of the seasonal concentrations changes of
NO3− and fecal coliforms, for example. We further speculate that it is
the change from the more dominant conduit flow during the wet season
to more matrix dominated flow during the dry season that works to
reset the biogeochemical structure of the groundwater that was re-
moved during the wet season. Since our study shows that water quality
can be degraded in both the dry and wet seasons, a better under-
standing of changes and mechanisms for change from the wet to dry
seasons needs more study on this and other karst systems.
We consider our results preliminary, but those that might guide
future studies. Needed are more spatial coverages, synoptic temporal
samplings, and more diverse chemical and microbial measurements.
These data need to be studied in light of detailed land use analyses to
help differentiate natural processes such as water-rock interactions and
human influences.
Acknowledgements
This work was funded by the Michigan State University Watercube
Program. We would like to sincerely acknowledge Centro de
Investigaciones Regionales Dr. Hideyo Noguchi, Autonomous
University of Yucatán for providing laboratory space and field vehicles.
References
Alcocer, J., Lugo, A., Marin, L.E., Escobar, E., 1998. Hydrochemistry of waters from five
275
D.T. Long et al.
cenotes and evaluation of their suitability for drinking-water supplies, northeastern
Yucatán, Mexico. Hydrogeol. J. 6, 293–301.
Arcega-Cabrera, F., Velazquez-Tavera, N., Fargher, L., Derrien, M., Norena-Barroso, E.,
2014. Fecal sterols, seasonal variability, and probable sources along the ring of
cenotes, Yucatán, Mexico. J. Contam. Hydrol. 168, 41–49.
Arcega-Cabrera, F., Fargher, L., 2016. Education, fish consumption, well water, chicken
coops, and cooking fires: using biogeochemistry and ethnography to study exposure
of children from Yucatán, Mexico to metals and arsenic. Sci. Total Environ. 568,
75–82.
Arcega-Cabrera, F., Fargher, L.F., Oceguera-Vargas, I., Noreña-Barroso, E., Yánez-Estrada,
L., Alvarado, J., González, L., Moo-Puc, R., Pérez-Herrera, N., Quesadas-Rojas, M.,
Pérez-Medina, S., 2017. Water consumption as source of arsenic, chromium, and
mercury in children living in rural Yucatán, Mexico: blood and urine levels. Bull.
Environ. Contam. Toxicol. 99, 452–459.
Back, W., Hanshaw, B.B., 1970. Comparison of chemical hydrogeology of the carbonate
peninsulas of Florida and Yucatán. J. Hydrol. 10, 330–368.
Back, W., Hanshaw, B.B., Herman, J.S., Vanddriel, J.N., 1986. Differential dissolution of a
Pleistocene reef in the groundwater-mixing zone of coastal Yucatán, Mexico. Geology
14, 137–140.
Ball, J.W., Nordstrom, D.K., 1991. User's manual for WATEQ4F with revised thermo-
dynamic data base and test cases for calculating speciation of major, trace, and redox
elements in natural waters. U.S. Geol. Surv. Open-File Rep. 91–183, 189.
Bauer-Gottwein, P., Gondwe, B.R.N., Charvet, G., Marín, L.E., Rebolledo-Vieyra, M.,
Merediz-Alonso, G., 2011. Review: the Yucatán peninsula karst aquifer. Mexico
Hydrogeol. J. 19, 507–524.
Bautista, F., Aguilar, D.Y., Batllori, S.E., 2011. Amenazas, vulnerabilidad y riesgos de
contaminación de las aguas subterráneas de la Península de Yucatán. Teoría y Praxis
9, 9–13.
Bautista, F., Frausto, O., Ihl, T., Aguilar, Y., 2015. An update soil map of the Yucatán
State, Mexico: geomorphopedological approach and WRB Ecosystems. Ecosistemas Y
Recursos Agropecuarios 2, 303–315.
Bischoff, J.L., Julia, R., Shanks, W.C., Rosenbauer, R.J., 1994. Karstification without
carbonic acid: bedrock dissolution by gypsum-driven dedolomitization. Geology 22,
995–998.
Boyer, D.G., Kuczynska, E., 2003. Storm and seasonal distributions of fecal coliforms and
Cryptosporidium in a spring. J. Am. Water Resour. Assoc. 39, 1449–1456.
Brahana, J.V., 2007a. Karst aquifers. In: Trimble, Stanley W. (Ed.), Encyclopedia of Water
Science, second ed. CRC Press2008, pp. 693–697 Print Version.
Brahana, J.V., 2007b. Karst aquifers water quality and water resource problems. In:
Trimble, Stanley W. (Ed.), Encyclopedia of Water Science, second ed. CRC Press
2008, pp. 698–699 Print Version.
Butscher, C., Auckenthaler, A., Scheidler, S., Huggenberger, P., 2011. Validation of a
numerical indicator of microbial contamination for karst springs. Groundwater 49,
66–76.
Castro, R.P., Avila, J.P., Ye, M., Sansores, A.C., 2018. Groundwater quality: analysis of its
temporal and spatial variability in a karst aquifer. Groundwater 56, 62–72.
Ceron, J.G., Cordova, A.V., Zavala, J., Muriel, M., 2005. Chemical composition of pre-
cipitation at coastal and marine sampling sites in Mexico. Global NEST J. 7, 212–221.
Cvijić, J., 1893. Das karstphanomen. Versuch einer morphologischen Monographie 3
Geographischen Abhandlung, Wien 218-329.
Derrien, M., Cabrera, F.A., Tavera, N.L.V., Manzano, C.A.K., Vizcaino, S.C., 2014. Sources
and distribution of organic matter along the Ring of Cenotes, Yucatán, Mexico: sterol
markers and statistical approaches. Sci. Total Environ. 511 233-229.
Delgado, C., Pacheco, J., Cabrera, A., Batllori, E., Orellana, R., Bautista, F., 2010. Quality
of groundwater for irrigation in tropical karst environment: the case of Yucatán,
Mexico. Agric. Water Manag. 97, 1423–1433.
Delgado, C., Carranza, C.D., Bautista, F., Lanza, R.O., Hernández, H.R., 2011.
Classification and Agroclimatic Zoning Using the Relationship between Precipitation
and Evapotranspiration in the State of Yucatán, Mexico. Investigaciones Geográficas
75. Boletín del Instituto de Geografía, UNAM, pp. 51–60.
Deng, C.B., Zhang, C.L., Li, L.H., Li, Z.Y., Li, N., 2011. Mercury contamination and its
potential health effects in a lead-zinc mining area in the karst region of Guangxi,
China. Appl. Geochem. 26, 154–159.
Domenico, P.A., Schwartz, F.W., 1990. Physical and Chemical Hydrogeology. John Wiley
and Sons. Inc., pp. 824.
Dura, G., Pandics, T., Kadar, M., Krisztalovics, K., Kiss, Z., Bodnar, J., Asztalos, A., Papp,
E., 2010. Environmental health aspects of drinking water-borne outbreak due to karst
flooding: case study. J. Water Health 8, 513–520.
Edmunds, W.M., 2009. Geochemistry's vital contribution to solving water resource pro-
blems. Appl. Geochem. 24, 1058–1073.
Escolero, O.A., Marin, L.E., Steinich, B., Pacheco, A.J., Cabrera, S.A., Alcocer, J., 2002.
Development of a protection strategy of karst limestone aquifers: the Mérida Yucatán,
Mexico case study. Water Resour. Manag. 16, 351–367.
Escolero, E., Marin, L.E., Steinich, B., Pacheco, J.A., Maldonado, A.M., Anzaldo, J.M.,
2005. Geochemistry of the hydrogeological reserve of Mérida, Yucatán, Mexico.
Geofisc. Int. 44, 301–314.
Escolero, O.A., Marin, L.E., Steinich, B., Pacheco, J., 2000. Delimitation of a hydro-
geological reserve for a city within a karstic aquifer: the Mérida, Yucatán example.
Landsc. Urban Plann. 51, 53–62.
Farfán, L.M., D'Sa, E.J., Liu, K., Rivera-Monroy, V.H., 2014. Tropical cyclone impacts on
coastal regions: the case of the Yucatán and the baja California peninsulas, Mexico.
Estuar. Coast 37, 1388–1402.
Fels, J.B., 1999. Source-identification investigations of petroleum contaminated
groundwater in the Missouri Ozarks. Eng. Geol. 52, 3–13.
Few, R., Mercado, A., Graizbord, B., Albornoz Mendoza, L., 2008. Climatic Hazards,
Health Risk and Response. Case Study 4: Yucatán, Mexico. Research Report.
276
Applied Geochemistry 98 (2018) 265–277
University of East Anglia, Norwich, UK, pp. 58.
Fitzpatrick, M.L., Long, D.T., Pijanowski, B.C., 2007. Exploring the effects of urban and
agricultural land use on surface water chemistry, across a regional watershed, using
multivariate statistics. Appl. Geochem. 22, 1825–1840.
Ford, D., Williams, P., 2007. Karst Hydrogeology and Geomorphology. Wiley, pp. 576.
Gondwe, B.R.N., Merediz-Alonso, G., Bauer-Gottwein, P., 2011. The influence of con-
ceptual model uncertainty on management decisions for a groundwater-dependent
ecosystem in karst. J. Hydrol. 400, 24–40.
González, R., Rodríguez, R., 2006. Contamination of the Yucatán aquifer by leachate from
municipal waste. In: In: Demuth, S., Gustard, A., Planos, E., Scatena, F., Serva, E.
(Eds.), Climate Variability and Change. Hydrological Impacts 308 IAHS Publication.
Gonzalez-Herrera, R., Martinez-Santibañez, E., Pacheco-Avila, J., Cabrera-Sansores, A.,
2014. Leaching and dilution of fertilizers in the Yucatán karstic aquifer. Environ.
Earth Sci. 72, 2879–2886.
Graniel, C.E., Morris, L.B., Carrillo-Rivera, J.J., 1999. Effects of urbanization on
groundwater resources of Mérida, Yucatán, Mexico. Environ. Geol. 37, 303–312.
Grimmeisen, F., Zemann, M., Goeppert, N., Goldscheider, N., 2016. Weekly variations of
discharge and groundwater quality caused by intermittent water supply in an urba-
nized karst catchment. J. Hydrol. 537, 157–170.
Guo, F., Jiang, G.H., 2009. Nitrogen budget of a typical subterranean river in peak cluster
karst area. Environ. Geol. 58, 1741–1748.
Guo, Q.H., Wang, Y.X., Ma, T., Li, X., 2005. Variation of karst spring discharge in the
recent five decades as an indicator of global climate change: a case study at Shanxi,
northern China. Sci. China Earth Sci. 48, 2001–2010.
He, Q., Qiu, S., Jiang, Y., Wu, Z., Liu, Z., 2016. Land-use change caused microbial pol-
lution in a karst underground river, Chongqing, China. Environ. Earth Sci. 75, 709.
https://doi.org/10.1007/s12665-016-5530-8.
Herojeet, R., Rishi, M.S., Lata, R., Dolma, K., 2017. Quality characterization and pollution
source identification of surface water using multivariate statistical techniques,
Nalagarh Valley, Himachal Pradesh, India. Appl. Water Sci. 7, 2137–2156.
Hollingsworth, E., Brahana, V., Inlander, E., Slay, M., 2008. Karst regions of the world
(KROW): global karst datasets and maps to advance the protection of karst species
and habitats worldwide. In: In: Kuniansky, Eve L. (Ed.), U.S. Geological Survey Karst
Interest Group Proceedings U.S 5023. Geological Survey Scientific Investigations
Report 2008-, pp. 142.
Huebsch, M., Fenton, O., Horan, B., Hennsessy, D., Richards, K.G., Jordan, P.,
Goldscheider, N., Butsche, r C., Plum, P., 2014. Mobilisation or dilution? Nitrate
response of karst springs to high rainfall events. Hydrol. Earth Syst. Sci. 18,
4423–4435.
INEGI, 2014. Guía para la interpretación de cartografía Uso del suelo y vegetación Escala
1:250 000 Serie V. Instituto Nacional de Estadística y Geografía (México) 195 pp.
INFOFLR, 2016. Mexico. https://infoflr.org/countries/mexico.
Knierim, K.J., Hays, P.D., Bowman, D., 2015. Quantifying the variability in Escherichia
coli (E. coli) throughout storm events at a karst spring in northwestern Arkansas,
United States. Environ. Earth Sci. 74, 4607–4623.
Krasilnikov, P., Gutiérrez-Castorena, C., Ahrens, R.J., Cruz-Gaistardo, C.O., Sedov, S.,
Solleiro-Rebolledo, E., 2013. The Soils of Mexico. Springer, New York 182 pages.
Land, L.S., 1985. Failure to precipitate dolomite at 25°C from dilute solution despite
1000-fold oversaturation after 32 years. Aquat. Geochem. 4, 361–368.
Lesser, J., Weidie, A., 1988. Region 25, Yucatán peninsula. In: Back, W., Rosenhein, J.S.,
Seaber, P.R. (Eds.), Geology of North America, Hydrogeology. Geol Soc Am, Boulder,
pp. 237–241.
Li, X.D., Liu, C.Q., Harue, M., Li, S.L., Liu, X.L., 2010. The use of environmental isotopic
(C, Sr, S) and hydrochemical tracers to characterize anthropogenic effects on karst
groundwater quality: a case study of the Shuicheng Basin, SW China. Appl. Geochem.
25, 1924–1936.
Long, D.T., Voice, T.C., Niagolova, N.D., McElmurry, S.P., 2012. Effects of human ac-
tivities on groundwater geochemistry in a rural karst area affected by Balkan
Endemic Nephropathy. Appl. Geochem. 27, 1920–1931.
Long, D.T., Fegan, N.E., Lyons, W.B., Hines, M.E., Macumber, P.E., 1992. Geochemistry of
acid-brines: lake Tyrrell, Victoria, Australia. Chem. Geol. 96, 19–32.
Long, D.T., Voice, T.C., Xagaroraki, I., Chen, A., Wu, H., Lee, E., Oun, A., Xing, F., 2017.
Patterns of c-q hysteresis loops and within an integrative pollutograph for selected
inorganic and organic solutes and E. coli in an urban salted watershed during winter-
early spring periods. Appl. Geochem. 83, 93–107.
Langmuir, D., 1997. Aqueous Environmental Geochemistry. Prentice Hall, N.Y., pp. 600.
Long, D.T., Voice, T.C., Chen, Ao., Xing, F., Li, S.-H., 2015. Temporal and spatial patterns
of Cl- and Na+ concentrations and Cl/Na ratios in salted urban watersheds. Elem. Sci.
Anthropocene. https://doi.org/10.12952/journal.elementa.000049.
Martin, J.B., Screaton, E.J., 2001. Exchange of matrix and conduit water with examples
from the Floridian Aquifer. In: Kuniansky, Eve L. (Ed.), 2001, U.S. Geological Survey
Karst Interest Group Proceedings. Water-Resources Investigations Report 01-4011,
pp. 38–44.
McGuire, J.T., Long, D.T., Hyndman, D.W., 2005. Analysis of recharge-induced geo-
chemical change in a contaminated aquifer. Ground Water 43, 518–530.
Meusburger, S., Reichart, S., Kapfer, S., Schableger, K., Fretz, R., Allerberger, F., 2006.
Outbreak of acute gastroenteritis of unknown etiology caused by contaminated
drinking water in a rural village in Austria, August 2006. Wien Klin. Wochenschr.
119, 717–721.
Mohammadi, Z., 2009. Assessing hydrochemical evolution of groundwater in limestone
terrain via principle component analysis. Environ. Earth Sci. 59, 429–439.
Moore, P.J., Martin, J.B., Screaton, E.J., Neuhoff, P.S., 2010. Conduit enlargement in an
eogenetic karst aquifer. J. Hydrol. 393, 143–155.
Pacheco, J., Cabrera, A., Marin, L.E., 2000. Bacteriological contamination in the karstic
aquifer of Yucatán, Mexico. Geofisc. Int. 39, 285–291.
Pacheco, J., Marín, L., Cabrera, A., Steinich, B., Escolero, O., 2001. Nitrate temporal and
D.T. Long et al.
spatial patterns in 12 water-supply wells, Yucatán, Mexico. Environ. Geol. 40,
708–714.
Parkhurst, D.L., Appelo, C.A.J., 1999. User's guide to PHREEQC (version 2)–A computer
program for speciation, batch-reaction, one-dimensional transport, and inverse geo-
chemical calculations. In: U.S. Geological Survey Water-resources Investigations
Report, vols. 99–4259 312 pp.
Perez-Ceballos, R., Pacheco-Avila, J., Euan-Avila, J.I., Hernandez-Arana, H., 2012.
Regionalization based on water chemistry and physicochemical traits in the Ring of
Cenotes, Yucatán, Mexico. J. Cave Karst Stud. 74, 90–102.
Perry, E., Velazquez-Oliman, G., Marin, L., 2002. The hydrogeochemistry of the karst
aquifer system of the northern Yucatán Peninsula, Mexico. Int. Geol. Rev. 44,
191–221.
Perry, E., Paytan, A., Pedersen, B., Velazquez-Oliman, G., 2009. Groundwater geochem-
istry of the Yucatan Peninsula, Mexico: constraints on stratigraphy and hydrogeology.
J. Hydrol. 367, 27–40.
Piper, A.M., 1953. A graphic procedure in the geochemical interpretation of water ana-
lysis. US Geol. Surv. Groundwater Note 12.
Polanco-Rodríguez, A.G., Navarro-Alberto, J.A., Solorio-Sánchez, J.S., Mena-Rejón, G.J.,
DelValls-Casillas, T.A., 2015. Contamination by organochlorine pesticides in the
aquifer of the ring of cenotes in Yucatán, Mexico. Water Environ. 29, 140–150.
Reed, T.M., Fryar, A.E., Brion, G.M., Ward, J.W., 2011. Differences in pathogen indicators
between proximal urban and rural karst springs, Central Kentucky, USA. Environ.
Earth Sci. 64, 47–55.
Rehrl, C., Birk, S., Klimchouk, A.B., 2008. Conduit evolution in deep-seated settings:
conceptual and numerical models based on field observations. Water Resour. Res. 44.
https://doi.org/10.1029/2008WR006905.
Rocha, H., Cardona, A., Graniel, E., Alfaro, C., Castro, J., Rude, T., Herrera, E., Heise, L.,
2015. Fresh/brackish water interfaces in the mérida-progresso region. Yucatán.
277
Applied Geochemistry 98 (2018) 265–277
Technology Y Ciencias Del Aqua 6, 89–112.
Sanchez, D., Barbera, J.A., Mudarra, M., Andreo, B., 2015. Hydrogeochemical tools ap-
plied to the study of carbonate aquifers: examples from the karst systems of Southern
Spain. Environ. Earth Sci. 74, 199–215.
Schmitter-Soto, J.J., Comin, F.A., Escobar-Briones, E., Herrera-Silveira, J., Alcocer, J.,
Suarez-Morales, E., Elias-Gutierrez, M., Diaz-Arce, V., Marın, E., Steinich, B., 2002.
Hydrogeochemical and biological characteristics of cenotes in the Yucatán Peninsula
(SE Mexico). Hydrobiologia 467, 215–228.
Schwarz, K., Gocht, T., Grathwohl, P., 2011. Transport of polycyclic aromatic hydro-
carbons in highly vulnerable karst systems. Environ. Pollut. 159, 133–139.
Sinreich, M., Pronk, M., Kozel, R., 2014. Microbiological monitoring and classification of
karst springs. Environ. Earth Sci. 71, 563–572.
Smith, B.G., Ley, D., 2009. Sustainable tourism and clean water project for two
Guatemalan communities: a case study. Desalination 248, 225–232.
Stambuk-Giljanovic, N., 2006. Evaluation of water resources in Dalmatia for human
health. Water Int. 31, 499–513.
Tranter, J., Gunn, J., Hunter, C., Perkins, J., 1997. Bacteria in the castleton karst, der-
byshire, England. Q. J. Eng. Geol. 30, 171–178.
Wayland, K.G., Long, D.T., Hyndman, D.W., Pijanowski, B.C., Woodhams, S.M., Haack,
S.K., 2003. Identifying relationships between baseflow geochemistry and land use
with synoptic sampling and R-mode factor analysis. J. Environ. Qual. 31, 180–190.
Wicks, C., Kelley, C., Peterson, E., 2004. Estrogen in a karstic aquifer. Ground Water 42,
384–389.
Yolcubal, I., Akyol, N.H., 2007. Retention and transport of hexavalent chromium in
calcareous karst soils. Turk. J. Earth Sci. 16, 363–379.
Zhou, L., Chen, X., Li, X., Yang, X., Huang, D., Xia, W., Cui, C., 2011. Nitrate vertical
transport and simulation in soils of rocky desertification in Karst regions, Southwest
China. Environ. Earth Sci. 63, 273–278.