Land-use change and managed aquifer recharge effects on ...online.sfsu.edu/jgurdak/Publications/Hejazian_etal_2017.pdfFreshwater resources on low-lying atoll islands are highly vulnerable

lable at ScienceDirect

Applied Geochemistry 80 (2017) 58e71

Contents lists avai

Applied Geochemistry

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

Land-use change and managed aquifer recharge effects on thehydrogeochemistry of two contrasting atoll island aquifers, Roi-NamurIsland, Republic of the Marshall Islands

Mehrdad Hejazian a, Jason J. Gurdak a, *, Peter Swarzenski b, Kingsley O. Odigie c, d,Curt D. Storlazzi d

a Department of Earth & Climate Sciences, San Francisco State University, San Francisco, CA, USAb International Atomic Energy Agency, Monaco, Principality of Monaco, Monacoc University of California, Santa Cruz, Santa Cruz, CA, USAd United States Geological Survey, Pacific Coastal and Marine Science Center, Santa Cruz, Santa Cruz, CA, USA

a r t i c l e i n f o

Article history:Received 16 July 2016Received in revised form14 February 2017Accepted 9 March 2017Available online 14 March 2017

Editorial handling by Prof. M. Kersten.

* Corresponding author.E-mail address: [email protected] (J.J. Gurdak).

http://dx.doi.org/10.1016/j.apgeochem.2017.03.0060883-2927/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

Freshwater resources on low-lying atoll islands are highly vulnerable to climate change and sea-levelrise. In addition to rainwater catchment, groundwater in the freshwater lens is a critically importantwater resource on many atoll islands, especially during drought. Although many atolls have high annualrainfall rates, dense natural vegetation and high evapotranspiration rates can limit recharge to thefreshwater lens. Here we evaluate the effects of land-use/land-cover change and managed aquiferrecharge on the hydrogeochemistry and supply of groundwater on Roi-Namur Island, Republic of theMarshall Islands. Roi-Namur is an artificially conjoined island that has similar hydrogeology on the Roiand Namur lobes, but has contrasting land-use/land-cover and managed aquifer recharge only on Roi.Vegetation removal and managed aquifer recharge operations have resulted in an estimated 8.6 � 105 m3of potable groundwater in the freshwater lens on Roi, compared to only 1.6 � 104 m3 on Namur. We usegroundwater samples from a suite of 33 vertically nested monitoring wells, statistical testing, andgeochemical modeling using PHREEQC to show that the differences in land-use/land-cover and managedaquifer recharge on Roi and Namur have a statistically significant effect on several groundwater-qualityparameters and the controlling geochemical processes. Results also indicate a six-fold reduction in thedissolution of carbonate rock in the freshwater lens and overlying vadose zone of Roi compared toNamur. Mixing of seawater and the freshwater lens is a more dominant hydrogeochemical process on Roibecause of the greater recharge and flushing of the aquifer with freshwater as compared to Namur. Incontrast, equilibrium processes and dissolution-precipitation non-equilibrium reactions are moredominant on Namur because of the longer residence times relative to the rate of geochemical reactions.Findings from Roi-Namur Island support selective land-use/land-cover change and managed aquiferrecharge as a promising management approach for communities on other low-lying atoll islands toincrease the resilience of their groundwater supplies and help them adapt to future climate changerelated stresses.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Low-lying atolls are generally 400 atolls worldwide may be uninhabitable by the end of the 21stcentury (Bailey et al., 2013). In addition to the SLR-induced inun-dation, climatic and population pressures are threatening the sus-tainability of the limited freshwater resources on most atoll islands

mailto:[email protected]://crossmark.crossref.org/dialog/?doi=10.1016/j.apgeochem.2017.03.006&domain=pdfwww.sciencedirect.com/science/journal/08832927http://www.elsevier.com/locate/apgeochemhttp://dx.doi.org/10.1016/j.apgeochem.2017.03.006http://dx.doi.org/10.1016/j.apgeochem.2017.03.006http://dx.doi.org/10.1016/j.apgeochem.2017.03.006

M. Hejazian et al. / Applied Geochemistry 80 (2017) 58e71 59

(Holding et al., 2016; Karnauskas et al., 2016). In some cases, thelack of freshwater, including groundwater resources, may be alimiting factor for habitability of atolls even before loss of land dueto SLR-induced inundation (Gurdak et al., 2015; Treidel et al., 2012).In addition to SLR, incremental seawater encroachment, morefrequent storm over-wash events, and droughts related to El Ni~no/Southern Oscillation (ENSO) variability will reduce land cover andthreaten groundwater supplies, forcing inhabitants to adapt orrelocate (White et al., 2007). Because groundwater is an importantsource of freshwater supply on low lying atolls, especially duringdroughts, any adaptation measures to climate stressors and SLRmust include an appropriate management plan that accounts forboth groundwater quantity and quality.

Freshwater in atoll aquifers floats on top of saline water, isroughly lenticular in shape, and is referred to as the ‘freshwaterlens’ (FWL). Here we define the FWL as

Fig 1. Roi-Namur Island is located on the northern most point of Kwajalein Atoll,Republic of the Marshall Islands. Location of the well clusters, transects, and horizontallens well onRoi-Namur Island are shown.

M. Hejazian et al. / Applied Geochemistry 80 (2017) 58e7160

are located on the northern and central part of this lobe, but thedifference in LULC between the Roi and Namur lobes is substantial.The groundwater on Namur is naturally recharged, but the volumeof the FWL is limited by higher ET rates compared to Roi. The FWLdynamics on Namur are more indicative of natural atoll islandsettings where human impact has been minimal, yet a relativelythin, viable FWL exists. A horizontal lens well is located on Namur,but it is not being actively pumped for water supply (not shownhere). In this study we identify Roi as the human-modified lobe andNamur as the natural, vegetated lobe.

2.2. Roi-Namur hydrogeochemistry

Observations of atoll island geology have identified two distinctstratigraphic layers that influence the depth of the FWL and tidalpropagation (Ayers and Vacher, 1986; Hunt and Peterson, 1980)(Fig. 2). The bottom layer is a highly permeable Pleistocene lime-stone layer that overlies basaltic basement rock and can be as thick

as 3000 m on some islands (Raitt and Perkins, 1954). The highpermeability is due to subaerial diagenesis and the karstification oflimestone during low sea-level stands. The upper layer is signifi-cantly less permeable and consists of unconsolidated fine-grainedHolocene carbonate sediments that are typically

Fig 2. Generalized hydrogeology and tidal mixing conceptual model of a typical atoll aquifer. The hydraulic conductivity (K) values and approximate depth of the ThurberDiscontinuity have been reported previously for Roi-Namur Island (Oberdorfer et al., 1990; Gingerich, 1992; Underwood et al.,1992).


layer is consistent with observations from nearby atolls of Enewe-tak, Bikini, and Majuro (Anthony et al., 1989; Buddemeier andOberdorfer, 1997; Emery et al., 1954). Using the groundwatermodel SUTRA (Voss,1984), the estimated hydraulic conductivity (K)of the Holocene sediments is 10e50 m day�1 (Underwood et al.,1992) and the Pleistocene limestone is 1000 m day�1 (Oberdorferet al., 1990) (Fig. 2).

In general, the FWL on atoll islands is a function of averageannual rainfall, recharge, K of the unconsolidated Holocene de-posits, island width, depth to the Thurber Discontinuity, and insome cases the reef flat plate, which is a relatively thin and shallow,lagoon-ward sloping limestone layer on the ocean side (Bailey et al.,2010). Other factors such as the position of the island relative towind direction, cross-island grain size differences, and tidal lag alsoaffect the FWL dynamics (Hunt and Peterson, 1980). Larger islandstend to form on the leeward side of atolls and have thicker Holo-cene deposits and finer sediments because they are not exposed tothe large swells and higher energy environment of windwardislands (Spennemann, 2006). Numerical model simulations haveindicated that the relatively less permeable Holocene deposits onthe leeward islands results in K values of about 50 m day�1, whilethe Holocene sediments on windward islands result in K values ashigh as 400 m day�1 (Bailey et al., 2009). The relatively lower K ofthe leeward islands results in less tidal propagation and mixing ofseawater with the FWL. Thus, the lower-energy environment onleeward islands tends to promote a thicker FWL than windwardislands. This is evident on Kwajalein Atoll and the thinner FWL onthe smaller windward island of Roi-Namur (Gingerich, 1996) ascompared to the thicker FWL on the larger, southern leewardKwajalein Island (Hunt and Peterson, 1980).

In addition to difference in hydrogeology between leeward andwindward islands, the cross-island lithology of an atoll island cancontribute to differences in FWL thickness. The lagoon side of atypical atoll island is a relatively low-energy depositional envi-ronment protected from big ocean waves where fine Holocenesands are deposited (Fig. 2). In contrast, the ocean facing section ofan atoll island is a higher-energy depositional environment wherecoral fragments and pebble to cobble size sediments have been

deposited in the Holocene. In general, sediment size in the un-consolidated Holocene deposits decrease in size from ocean tolagoon, as observed by Anthony et al. (1989). It has been demon-strated on Kwajalein Island that variations in cross-island lithologyaffect the rate and extent of tidal propagation, thus affecting theFWL geometry (Hunt and Peterson, 1980). The higher permeabilityand hydraulic conductivity found on the ocean side promotes fastertidal propagation and increased mixing, and explains the asym-metrical shape of the FWL on atoll islands (Fig. 2). A relatively thicksaline transition zone (sometimes thicker than the FWL) can formon the ocean side because of dispersive tidal mixing (Fig. 2). Forthese reasons, atoll islands generally have a thicker FWL on thelagoon side compared to the ocean side (Fig. 2).

2.3. Field methods and groundwater sampling

Between April 30 and May 7, 2015, groundwater samples werecollected from a total of 33 existing U.S. Geological Survey (USGS)monitoring wells from nine separate well clusters located on Roi-Namur (Fig. 1 and Supplementary Material, Table S1). Not all ofthe original USGS monitoring wells on Roi-Namur (Gingerich, 1996,1992) were sampled because of time and resource constraints.Therefore, well selection was prioritized so that lagoon and oceanside wells on both lobes of the island are represented and severalcross-island transects (Fig. 1) could be compared.

The nine well clusters include six on Roi and three on Namur(Fig. 1); each cluster represents a single location where a group ofwells were drilled to discrete depths so that groundwater samplescould be obtained along the depth profile of the vertical salinitygradient in the FWL and transition zone to the deeper salinegroundwater. A total of 21 discrete well depths on Roi and 12discrete well depths on Namur were sampled. All wells are iden-tified with the letter ‘R’ followed by a number that denotes the wellcluster and approximate depth below mean sea level of the screenmid-point for each well within each cluster (e.g. R1-1, R1-7, etc.)(Table S1), which is consistent with the original USGS well identi-fication (Gingerich, 1996, 1992). The monitoring wells are con-structed of 5.1-cm-diameter poly-vinyl-chloride (PVC) flush-


threaded pipe. A 0.61 m section of each pipe is screened 0.15 mfrom the bottom to allow groundwater to flow into the well onlyfrom the desired depths.

All groundwater wells were sampled using protocols of theUSGS National Field Manual for the collection of water-quality data(U.S. Geological Survey, variously dated). Using an electrical sub-mersible pump, three wet-casing volumes were purged prior tosample collection. A peristaltic pump and nylon tubing was used tolift groundwater to the surface, where it was run through a 0.45 mmfilter to remove particulates before being gravity drained intosample bottles. Prior to sample collection at each well, the inside ofthe nylon tubing was rinsed with well water equivalent to at leasttwo sample volumes to ensure only representative samples fromthat location and depth were being collected. Chemical pre-servatives were not used for any collected samples. Sampling pro-cedures were identical for all sample constituents, but samplevolume varied from 20 ml vials to 1.0 L bottles depending on theconstituent of concern. Samples were placed in a field cooler andtransported to a refrigerator for preservation and storage.

A calibrated YSI 556 multi-probe handheld multi-parameterfield meter was used to measure field water-quality parameters,including pH, temperature, oxidation-reduction potential (ORP),dissolved oxygen (DO), electrical conductivity, and salinity. Totalalkalinity was determined for each sample at the end of each fieldday using the inflection point method by performing multipointtitration with 1.6 M H2SO4.

2.4. Laboratory methods

Groundwater samples were analyzed at the following labora-tories. Major ions and trace elements (except for chloride, bromide,fluoride, and sulfate) were analyzed at the University of SouthernMississippi Center for Trace Analysis using Inductively CoupledPlasma Mass Spectrometry (ICP-MS) (Shim et al., 2012; Swarzenskiet al., 2007, 2006). The chloride, bromide, fluoride, and sulfate wereanalyzed using an ion chromatograph (IC), and dissolved inorganiccarbon (DIC) and d13C-DIC were analyzed via continuous flowcavity ring-down spectroscopy following wet chemical oxidation atthe USGS inMenlo Park, CA. The nutrient analysis was performed atWoods Hole Oceanographic Institution (WHOI), in Woods Hole, MAusing a SEAL AA3 four-channel segmented flow analyzer (Ganguliet al., 2014; Gonneea et al., 2014; Swarzenski and Izbicki, 2009).The dissolved organic carbon and d13C-DOC were analyzed usinghigh temperature catalytic con-version DOC analyzer, a GD-100 CO2trap and a continuous flow IRMS at WHOI's National Ocean Sci-ences Accelerator Mass Spectrometry Facility (NOSAMS) (Gonneeaand Charette, 2014).

2.5. Statistical analysis

To evaluate if water-quality parameters and geochemical pro-cesses on Roi and Namur are statistically different at the 95% con-fidence level, median concentrations of selected constituents arecompared using a Wilcoxon rank-sum test. All statistical analyseswere performed using the software package JMP version 12.1.0 (SASInstitute, Cary, NC). We used a Wilcoxon rank-sum test on thedistribution of residuals to evaluate whether the median concen-trations on Roi and Namur differ:

Ho : Cr ¼ Cn; Ha : CrsCn; a ¼ 0:05 (1)

where Cr is the median concentration on Roi and Cn is the medianconcentration on Namur, a is the significance level required toreject the null hypothesis (Ho), and Ha is the alternative hypothesis.The constituents are generally selected based on the mechanistic

hypothesis that the decay of vegetation in the soil and root zonewill lead to more microbial oxidation and increased carbonatedissolution through the following reactions:

CH2Oþ O2/CO2 þ H2O (2)

CaðMgÞCO3 þ CO2 þ H2O42HCO3� þ Ca2þ�Mg2þ

�(3)

The differences in vegetation cover between Roi and Namur mayalter the geochemical reactions in equations (2) and (3), whichwould be observed in the geochemical signal of each lobe of Roi-Namur. The specific mechanisms and subsequent geochemicalprocesses tested in the statistical analyses are described next.

A geochemical study on Kwajalein Island by Tribble (1997)showed that microbial oxidation in the root zone plays an impor-tant role in carbonate dissolution on atoll islands and that the mostactive dissolution occurred in the vadose zone and near the top ofthe water table. Increasing vegetative cover would likely result ingreater organic matter input to the soil, and thereby producegreater inorganic carbon through microbial oxidation (equation(2)).We also use results of theWilcoxon rank-sum test to evaluate ifthe removal of natural vegetation on Roi has significantly reducedCO2 flux in the groundwater and overlying vadose zone, thusreducing dissolution of the carbonate rocks (equation (3)).

The inorganic carbon formed by these reactions can either befrom microbial oxidation of organic matter (equation (2)) or car-bonate mineral dissolution (equation (3)). To distinguish whethermicrobial oxidation or carbonate dissolution is a greater source ofdissolved inorganic carbon (DIC) on each lobe, 13C/12C for DIC (d13C-DIC) was evaluated. Plants preferentially take up the lighter stableisotope of carbon-12 for photosynthesis and are more depleted inthe heavier carbon-13 isotope. Conversely, carbonate minerals areless depleted in carbon-13 and dissolution reactions produceinorganic carbonwith an isotopic signature reflective of the ViennaPee Dee Belemnite (VPDB) reference standard. Dissolved organiccarbon (DOC) in groundwater results from microbial breakdown ofleaf litter and other decaying organic matter or from hydrocarboncontamination. Therefore, we expect to find that groundwater onNamur has higher concentrations of DOC due to the larger input oforganic litter in the soil zone as compared to Roi.

Carbonate dissolution-precipitation reactions can affect calcium(Ca2þ), magnesium (Mg2þ), and bicarbonate (HCO3�) ion concen-trations in groundwater of carbonate aquifers (Plummer et al.,1976). Therefore, we also tested for statistically significant differ-ence in Ca2þ, Mg2þ, and HCO3� concentration in groundwater on Roias compared to Namur. If the mechanism for dissolution is relatedto DIC input, we expect to observe higher Ca2þ, Mg2þ, and HCO3�

concentrations in groundwater related to the dissolution of car-bonate minerals (equation (3)).

Nutrient cycling also plays an important role in geochemicalprocesses. Decaying organic matter in the soil is a source of mac-ronutrients, such as nitrogen (N) and sulfur (S). Biological decom-position of organicmatter in the soil produces themineralized formof these nutrients, which are subsequently taken up by plants and/or leached into the groundwater. Under anaerobic conditions,denitrification and sulfate reduction may reduce N and S to theirgaseous form (i.e. NO3� to N2 and SO42� to H2S) where they becomeunavailable for plant uptake and are released to the atmosphere(Korom, 1992). More vegetative input and microbial oxidation maylead to additional leaching of nutrients in the groundwater, thus wetested for statistically significant difference in nitrate (NO3�),ammonium (NH4þ), total dissolved nitrogen (TDN), and sulfur (S)concentrations in groundwater on Roi as compared to Namur.Similarly, decomposition of organic matter and oxidation reactions


influences the pH and oxidation-reduction potential (ORP) of theaqueous system, and thus we also tested for statistically significantdifferences in pH and ORP on Roi and Namur.

Prior to geochemical reactions with carbonate rocks, ground-water chemistry is a function of the percent seawater that is mixedwith recharge from precipitation or other freshwater sources.Therefore, the groundwater sample concentrations must benormalized by seawater to account for mixing before they can becompared. The mixing of recharge water with seawater in thetransition zone results in groundwater chemistry that is a result ofconservative and non-conservative reactions. Chloride (Cl�) doesnot react readily compared to other major ions, and variations inconcentrations of Cl� in natural waters is primarily due to physicalprocesses (Hem, 1985). Thus, the percent seawater that is mixedwith recharge in each groundwater sample can be determinedusing Cl� concentration and the assumption that all the Cl� in thegroundwater sample originates from seawater. Normalizing thegroundwater samples is done by plotting constituent concentra-tions with respect to percent seawater for both lobes and calcu-lating a least squares regression line from the combined data set.The residuals for each lobe are then calculated from the regressionline and the differences in residuals become statisticallycomparable.

2.6. Geochemical modeling

Solution concentrations from each groundwater sample wereinput into the geochemical modeling program PHREEQC (Parkhurstand Appelo, 2013) to obtain charge balance, saturation indices (SI),and pCO2 values. The WATEQ4 reference thermodynamic database(Ball and Nordstrom, 1991) was used for all PHREEQC simulations.The end-member chemistry used in the geochemical modeling wasderived from proportions of seawater and rainwater based on theconservative Cl� concentrations in each sample. The difference inchemistry between rain and artificial recharge from the MAR sys-tem is assumed to be negligible. The geochemical modeling wasalso used to identify reactions such as dissolution of the carbonateminerals in the groundwater that deviate from conservativefreshwater-seawater mixing processes. In the vadose zone andaquifer of carbonate islands, groundwater ion concentrations areaffected by the dissolution of calcite and aragonite minerals, whichis controlled predominantly by the pCO2, pH, and alkalinity. Dif-ferences between observed groundwater sample concentrationsand theoretical speciation values may indicate a system that hasadditional chemical inputs from diagenesis or non-equilibriumdissolution-precipitation reactions.

Geochemical modeling was also used to estimate dissolutionrates of carbonate minerals in the FWL of Roi and Namur, andevaluate the effects of recharge, inorganic carbon input, and resi-dence times on the rates of the carbonate-diagenetic reactions.Aragonite and magnesian-calcite are precipitated on atoll reefs andhave been shown to be more soluble than pure calcite (Morse andMackenzie, 1990). Based on core samples and thin sections onMajuro Atoll, the predominant dissolution of aragonite skeletalmaterial followed byminor amounts of calcite and low-magnesian-calcite, indicates a system favoring aragonite dissolution (Anthonyet al., 1989). Water flux through the subsurface dissolves sourcerock, increases secondary porosity, and ultimately results in a netflux of sediment out of the system. Studies have measured disso-lution for small carbonate islands and found that water in the FWLand transition zone increases permeability because of diageneticreactions (Anthony et al., 1989; Plummer et al., 1976). The impor-tant driver in this reaction is CO2 input, which in solution, dissolvescarbonate rock (equation (3)).

To estimate the dissolution rates of carbonate minerals in the

FWL, source rock contributions from carbonate dissolution weredetermined by measuring excess Ca2þ, Mg2þ, and HCO3� in theaquifer system. Given that the common naturally occurring sourceof these ions is from dissolution of carbonate minerals or seawater,the concentrations in solution that are outside what is expectedfrom conservative rainwater-seawater mixing represent the disso-lution of carbonate rock. Localized groundwater contaminationfrom septic systems could also result in reactions that involveorganic matter to produce HCO3�. Therefore, the molar ratios ofCa2þ:Cl�, Mg2þ:Cl�, and HCO3�:Cl� in local seawater were used todetermine the theoretical concentration in each groundwatersample under conservative reactions, and the sum of the excessconcentrations of Ca2þ, Mg2þ, and HCO3� represent the masscontribution from dissolution of source rock. Only one sample onNamur (R6-1, Table S1) was within the FWL (


CO2 þ H2O ¼ H2CO3ðaqÞ (5)

H2CO3 ðaqÞ4Hþ þ HCO3� (6)

HCO3�4Hþ þ CO32� (7)

Therefore, the pCO2 in the infiltration and recharge is an importantfactor in the subsequent geochemical processes of the aqueous andsolid carbonate system. Rainwater in equilibrium with the atmo-sphere will have a pCO2 of approximately 10�3.4 or 0.04% Vol CO2(400 ppm). In groundwater systems, pCO2 values of 10�2.5 to 10�1.0

have been observed (Hem, 1985), which are one to three orders ofmagnitude greater than rain. Plummer and others (1976) foundsimilarly elevated pCO2 in soil pore water samples. The elevatedpCO2 in soil pore water is driven by plant decay and microbialrespiration in the top soil, which is especially prevalent in tropicalclimates with dense vegetation. Studies have shown there is amoderate trend of increased dissolution of carbonateminerals withincreasing pCO2 (Anthony et al., 1989; Plummer et al., 1976).Respiration and aerobic decay has been shown to increase pCO2(equation (2)), which in turn reduces the pH (equations (5e7)) andreduces saturation and carbonate concentrations of water (Drever,1988). On Kwajalein Island, Tribble (1997) showed that the pre-dominant reaction in the system was the dissolution of calciumcarbonate driven by CO2 from microbial respiration. If residencetime is sufficient during water flux in the vadose zone, carbonatedissolution proceeds and Ca2þ, Mg2þ, and HCO3� is transported tothe groundwater where thermodynamically favored reactionscontinue.

3. Results and discussion

3.1. Freshwater lens asymmetry

Results of the groundwater sampling were used to estimate theshape of the FWL on Roi and Namur (Fig. 3). Most atoll islandsgenerally have an asymmetrically shaped FWL, with a relativethicker FWL on the lagoon side and thinner FWL on the ocean side(Ayers and Vacher, 1986; Hunt and Peterson, 1980). The asymmetryof the FWL on atoll islands is generally attributed to greater hy-draulic conductivity on the ocean side that increases tidal efficiencyand the dispersion of saline waters through the upper aquifer. As aresult, the transition zone is thicker on the ocean side, and the FWLis truncated at shallower depths compared to the lagoon side.However, the shape of the FWL on Roi is not consistent with mostatoll islands because a substantially thicker FWL and transitionzone is observed on the ocean side (Fig. 3a,b). Namur exhibits themore classic asymmetric shape, with a slightly thicker FWL on thelagoon side (Fig. 3c). The observation of a relatively thicker FWL onthe ocean side of Roi compared to natural atoll aquifers, such asNamur, is likely the result of the influx of additional diffuserecharge from the reduced ET and MAR on Roi, as discussed below.

In general, the FWL thickness and shape is function of recharge,K of the unconsolidated Holocene sediments, and depth to theThurber Discontinuity (Bailey et al., 2010). For example, Bailey et al.(2008) used numerical models to show that although the FWLthickness increases with increasing recharge rates and the FWLcould be truncated if the bottom extends to the depth of theThurber Discontinuity, the geometry and shape of the FWL did notsignificantly change. Drilling records on Roi-Namur indicate thereis a hard consolidated layer located 7e14 m bls (Gingerich, 1992),which likely represents the Thurber Discontinuity. However, theFWL on Roi is only 3 m and 5 m thick on the lagoon and ocean side,

respectively (Fig. 3a,b), and thus is not likely truncated by theThurber Discontinuity. Although the depth of the Thurber Discon-tinuity may vary from ocean to lagoon side, it is unlikely to result ina thicker FWL on the ocean side because the geology of most atollshave thicker Holocene deposits on the lower-energy lagoon side, asobserved on Enewetak Atoll (Buddemeier, 1981) and Majuro Atoll(Anthony et al., 1989).

Focused recharge from theMAR is a more likely cause of the FWLshape on Roi. If recharge was equally distributed across Roi, thepreviously discussed hydrogeologic controls on the lagoon sidewould likely limit seawater mixing, forming a thicker FWL.Groundwater level contours developed by Gingerich (1996) showsthat the highest hydraulic head on Roi is located in the eastern partof the lobe approximately half-way between the ocean and lagoonside under the concrete-lined catchment basin (Fig.1). The concrete-lined catchment covers a large area of the eastern section and lagoonside of Roi, and effectively eliminates direct recharge beneath thecatchment. Although recharge is limited, the hydraulic head(Gingerich, 1996) is highest and the FWL is relatively thick beneaththe concrete-lined catchment (Fig. 3a). This is likely due to therelatively low hydraulic conductivity on the lagoon side that limitstidal mixing and freshwater drainage. The general groundwater flowdirection in the FWL on Roi radiates from the highest head near theconcrete-lined catchment outward in all directions (Gingerich,1996). The MAR system on Roi reroutes rain that would normallyinfiltrate beneath the concrete-catchment and artificially rechargesthat water on the grassy area adjacent to the runway and near wellclusters R1 and R2 (Fig. 1). This focused artificial recharge hascreated the thickest part of the FWL under well cluster R1, eventhough it is on the ocean side (Fig. 3a). The groundwater levelcontours (Gingerich, 1996) indicates that the artificial rechargeadjacent to the runway will flow downgradient towards the oceanside of the island, with the strongest gradients favoring flow towardswell clusters R4 and R10 and producing a thick FWL (Fig. 3b). Theresult of the concrete-lined catchment and MAR is that the rainfallthat would have been recharged on the lagoon side has beendiverted and recharged on the ocean side and in an area that favorsgroundwater flow towards the ocean, which helps explain the un-expected asymmetry of the FWL. It is likely that without the LULCmodifications and MAR on Roi, the FWL geometry would reflect amore typical atoll aquifer, similar to what is observed on Namur.

3.2. Statistical differences in groundwater quality on Roi andNamur

Results of the Wilcoxon rank-sum test indicate that pH, ORP,Ca2þ, NH4þ, TDN, and d13C-DIC in groundwater of Roi and Namur arestatistically different at the significance level of 5% (a ¼ 0.05)(Fig. 4). Alkalinity, DO, S, HCO3�, Mg2þ, NO3�, DIC, and DOC ingroundwater of Roi and Namur are not statistically different ata ¼ 0.05. Statistically lower pH values and higher Ca2þ concentra-tions are observed on Namur (Fig. 4a,b), indicating relatively moreacidic groundwater and greater dissolution of carbonate mineralsthan on Roi. Groundwater on Namur is depleted in d13C-DIC relativeto groundwater on Roi (Fig. 4c). Although the DIC concentrationsare not statistically different on Roi and Namur, the statistical dif-ferences in d13C-DIC between Roi and Namur indicates that mi-crobial oxidation of organic matter is a more important source ofDIC than dissolution reactions on Namur compared to Roi.Considering that Roi and Namur have statistically similar DICconcentrations and that Namur has a higher calculated carbonatedissolution rate in the FWL than Roi, as discussed below, there arelikely important processes other than decaying vegetation on Roithat contribute to DIC. Comparison of groundwater quality from thewell clusters on Roi (Table S1), reveals that well clusters R3 and R4

Fig 3. Salinity profiles for cross-sections (a) AeA’ and (b) BeB’ on Roi and (c) CeC’ on Namur.


have elevated concentrations of DIC (Fig. 5a) and correspondingelevated pCO2 and nutrients (Fig. 5bee), as compared to otherclusters. As discussed next, the elevated nutrients are likely anindication of fecal contamination, which could represent animportant process on Roi that contributes to DIC.

Given the relatively greater vegetation density on Namur, weexpected the organic matter and microbial respiration and con-sumption of oxygen in the soil would result in more reduced (lowerORP) groundwater on Namur than Roi. However, the results of thestatistical tests indicate significantly more reduced groundwater onRoi than Namur (Fig. 4d), which may be related to the elevatedconcentrations of NH4þ and other nutrients in groundwater on Roi(Fig. 5cee). Because the landscape on Roi is not fertilized, point-source contamination associated with human activity is the likelysource of elevated nutrients in the groundwater on Roi. NH4þ canderive from human and animal waste and is often found at elevatedconcentrations in groundwater near septic systems and leakysewage pipes. There are three actively used septic tanks and leachfields on Roi-Namur; one is located on the dredge-filled, conjoinedsection of the island that is adjacent to well cluster R3, and two arelocated on Namur (Thomas Hutchinson, Roi Operations Manager,

2015; personal communication).The elevated concentrations of TDN, NO3�, and NH4þ (Table S1),

generally occur in the well clusters R3 on the lagoon side and R4 onthe ocean side of Roi (Fig. 5). Of the R3 cluster, NO3� concentrationsare highest in well R3-1, but are not detected at deeper depthswhere NH4þ concentrations are elevated (Fig. 5d,e). Similarly at theR4 cluster, the shallow wells at (R4-1, R4-4, R4-7) have the highestNO3� concentrations and NH4þ increases with depth, but to a lesserextent than the deep well at R3 (Fig. 5e). The observation ofincreasing NH4þ and decreasing NO3�with depth likely indicates thatdissimilatory nitrate reduction to ammonium (DNRA) may beoccurring in the more anoxic conditions in the deeper parts of theaquifer (Korom,1992). Atwell cluster R4, the rate of increase in NH4þ

with depth is less what is observed at R3 (Fig. 5e). These differencesR3 and R4 can be attributed to the location of R4 further from theleach field and closer to the ocean side of Roi. The closer proximityof R4 to the ocean side will likely result in relatively greater waterflux, mixing of freshwater and seawater, and lower residence timeof freshwater in the FWL compared to R3, which would have thecumulative effects of increasing NO3� transport with depth andlimiting DNRA at R4.

Fig 4. Distribution of residuals of water quality constituents (aef: pH, Ca2þ, d13C-DIC,ORP, TDN, NH4þ) in groundwater of Roi and Namur that are significantly differentat the significance level of 5% (p-values from the Wilcoxon-rank sum test are shown).Residuals are calculated from a least squares regression line of the combined data fromRoi and Namur.


3.3. Carbonate dissolution

The calculated average source rock contributions (SRC) from theFWL and overlying vadose zone (7.27 and 10.32 mmol L�1 for Roiand Namur, respectively) indicate that relatively more dissolutionis occurring on Namur than Roi (Table 1a). However, with a largerFWL and overlying vadose zone area (0.46 and 0.14 km2 for Roi andNamur) and greater flushing (4.08 � 108 and 6.74 � 107 L yr�1 forRoi and Namur), the rate of sediment volume dissolved is

Fig 5. Vertical profiles of (a) DIC, (b) pCO2, (c) TDN, (d) NO3- , and (e) NH4þ fromwell clusters RThe R3 wells include R3-1, R3-4, R3-7, R3-11, and R3-16; and the R4 wells include R4-1, R4

139 m3 yr�1 on Roi compared to 32 m3 yr�1 on Namur (Table 1b).Using the estimated recharge for each lobe (886 and 482 mm yr�1

for Roi and Namur), the area underlain by a FWL, the volume of theFWL, a rock density of 2900 kg/m3 for aragonite, and assuming aneffective porosity of 0.3, the estimated annual increase in porosity is0.004% on Roi and 0.024% on Namur (Table 1b). Therefore, theannual increase in porosity in the FWL and overlying vadose zoneon Namur is approximately six times greater than Roi, which in-dicates that the dissolution of carbonate rock (increase in annualporosity) has been reduced on Roi relative to Namur as a result ofLULC change (loss of vegetative cover) and use of MAR. Futureadditional sampling of the FWL, particularly on Namur, will helpconstrain the spatial variability of these dissolution rates.

Using a similar method as the FWL calculations, we calculatedthe average SRC calculations at deeper depths in the aquifer toestimate changes in excess ion concentrations with depth andincreasing salinity. The SRC for three different vertical zones andassociated seawater percentages were calculated: FWL, �1.3%;transition zone, 1.3e50%; and deep aquifer, >50% (Table 1a). Aspreviously discussed, the greatest average SRC and dissolution ratesoccur on Namur in the FWL (�1.3% seawater) and overlying vadosezone sediments. The relatively lower dissolution rates in the FWLand overlying vadose zone on Roi are likely a result of relativelygreater flushing due to the relative increase in recharge from thelower ET rates and MAR. On Namur, the transition zone from 1.3 to50% has a smaller average SRC (8.65 mmol L�1) than in the FWL.However, the opposite pattern is observed on Roi where the tran-sition zone (9.35 mmol L�1) has greater average SRC than in theFWL (7.27 mmol L�1) (Table 1a). In both lobes, the lowest averageSRC occurs in deep aquifer (>50% seawater), either as a result oftidal flushing and removal of dissolved sediment and/or non-equilibrium reactions as a result of mixing.

Estimating the changes in dissolution rate and porosity indeeper, more saline parts of the aquifer is challenging becausedissolved minerals can be transported from shallower depths andprecipitate at deeper depths if the saturation conditions change.However, we find that the SI values on Roi and Namur (Fig. 6) areconsistent with the SRC calculations (Table 1). The groundwater atlow (�1.3% seawater) and intermediate (1.3e70% seawater) salin-ities in Roi are generally undersaturated with respect to aragoniteand become supersaturated at the most saline depths (>70%seawater), with only one exceptions at 40% seawater (Fig. 6). This isconsistent with the interpretation that undersaturated water is

3 (white circle) and R4 (black circle) on Roi plotted as function of percent seawater (%).-4, R4-7, R4-11, and R4-16 (Table 1).

Table 1(a) Calculated source rock contribution (SRC) on Roi and Namur for the freshwater lens (FWL) (�1.3% seawater), transition zone (1.3e50% seawater), and deep aquifer (>50%seawater); and (b) the calculated dissolution rates for the FWL (�1.3% seawater) and overlying sediments on Roi and Namur.

(a) Calculated Source Rock Contribution (SRC)

Ca2þ (mmol L�1) Mg2þ (mmol L�1) HCO3� (mmol L�1) Average SRC (mmol L�1)

�1.3% seawaterRoi 1.63 0.80 4.60 7.27Namur 2.28 1.16 6.37 10.321.3e50% seawaterRoi 1.91 1.21 4.89 9.35Namur 2.29 0.59 5.18 8.65>50% seawaterRoi 1.05 0.99 2.16 6.13Namur 1.13 0.80 1.89 5.79

(b) Calculated Dissolution RatesParameters Roi Namur

rainfall (mm yr�1) 1,927 1,927recharge (mm yr�1) 886 482area underlain by FWL (km2) 0.46 0.14flushing (L yr�1) 4.08Eþ08 6.74Eþ07source rock contribution (kg L�1) 6.90E-04 9.78E-04source rock flushed (kg yr�1) 2.81Eþ05 6.59Eþ04volume dissolved (m3 yr�1) 139 32annual increase in porosity (%) 0.004 0.024


being transported from the FWL on Roi to deeper transition zonewhere dissolution persists until the most saline depths (>70%seawater) where the groundwater is supersaturated with respect toaragonite. Most groundwater samples on Namur stay close tosaturation (SI ¼ 0 þ/� 0.2) throughout the profile (Fig. 6).Groundwater reaches equilibrium conditions in the FWL beforebeing transported deeper in the aquifer, and the vertical gradientsin the geochemical processes are explored using the results fromthe geochemical modeling, as discussed in the next section.

3.4. Equilibrium/non-equilibrium reactions

The equilibrium/non-equilibrium dissolution-precipitation re-actions in carbonate aquifers are described in this study using the SIwith respect to aragonite and pCO2 of groundwater. Changes in theSI of groundwater in carbonate aquifers can be generally attributedto five possible factors: variability in soil pCO2, evasion of CO2through the soil zone, CO2 flux in the phreatic zone, dissolution-precipitation non-equilibrium reactions of carbonate minerals,and seawater mixing (Plummer, 1975). To investigate how

Fig 6. Saturation indices with respect to aragonite plotted as a function of percentseawater for groundwater samples from Roi and Namur. Linear regressions and cor-responding r-square values are shown for Roi (solid line) and Namur (dashed line). Theseawater sample is shown for comparison.

geochemical processes have been altered on Roi due to the LULCchanges and MAR operation, we analyze these five factors andcompare to processes on Namur using geochemical modeling.

3.4.1. Soil pCO2 and CO2 evasionFor this study, we assume a closed system exists once water

enters the phreatic zone, and CO2 flux is only caused by dissolutionor mixing process that transport groundwater to sea. There is suf-ficient soil cover to limit CO2 evasion and there are no directsources of additional CO2 input in the groundwater, such as frommarshes or other natural sources. These are reasonable assump-tions because it was observed on Kwajalein Island that most of theproduction of CO2 by microbial oxidation occurs in the vadose zoneor near the top of the water table (Tribble, 1997).

3.4.2. Groundwater CO2 fluxThe only FWL sample collected on Namur is fromwell R6-1, and

it has the greatest log pCO2 (�1.51) of either lobe (Fig. 7). The logpCO2 of the FWL samples on Roi range from �2.52 to �1.53 (Fig. 7and Table S1). The linear regressions (Fig. 7) indicate that there isrelativelymore CO2 in the FWL and transition zone of the aquifer onNamur compared to Roi. The pCO2 values in the most salinegroundwater on Namur tend to be higher than Roi (Fig. 7). On Roi,the lowest pCO2 concentrations occur in the most saline ground-water (>80% seawater) (Fig. 7) and at the shallowest well at site R3-1 (Fig. 5b). Additionally, the position of Roi on the corner of the atollversus Namur's position along themore straight-line shorelinemayalso contribute to relatively enhanced tidal flushing on Roi, whichmay also contribute to a shorter residence times and moreseawater-like geochemistry at depth as compared to Namur.

Although the overall DIC concentrations in groundwater on Roiand Namur are not statistically different, there are important dif-ferences in the vertical DIC profiles of the two lobes (Fig. 8). OnNamur, the FWL and shallowest sections of the aquifer have thegreatest DIC and concentrations generally decrease with depth(Fig. 8c). In contrast on Roi, the FWL and shallowest sections of theaquifer generally have some of the lowest DIC concentrations(6.00 mmol C/L) in many parts of the transition zone and deepaquifer (Fig. 8a,b). These results indicate an apparent shift to

Fig 7. pCO2 plotted as a function of percent seawater for groundwater samples fromRoi and Namur. Linear regressions and corresponding r-square values are shown forRoi (solid line) and Namur (dashed line). The seawater sample is shown forcomparison.


greater DIC concentrations deeper in the profile on Roi compared toNamur. The apparent differences in vertical DIC profiles (Fig. 8) maybe attributed to a lower residence time in the overlying vadose zoneand near the top of the water table on Roi due to higher recharge

Fig 8. Dissolved inorganic carbon (DIC) profiles interpolated from groundwater sam

rates, which limits dissolution compared to Namur. Increases in DICdeeper in the profile on Roi are consistent with the relatively lowerobserved SI values (Table S1) in the transition zone that promotesgreater dissolution (Fig. 6).

The well cluster specific anomalies in pCO2 and DIC may beattributed to the previously discussed point-source contamination.The low pCO2 at well R3-1 may be related to the elevated NO3�

concentrations from the nearby leach field. It is likely that NO3�

predominates in the shallowest well where sufficient oxygen ispresent (DO ¼ 3.40 mg L�1, Table S1). The next deepest well (R3-4)has elevated NH4þ, a spike in TDN, and an increase in DOC, DIC, andpCO2 relative to R3-1 (Fig. 5), which is attributed to the lower DOconcentration and greater mass of nutrients at this lower depth inthe aquifer. Wells R4-1, R4-4, and R4-7 also have elevated NO3�,TDN, DOC, DIC, and pCO2, but NH4þ concentrations are low or notdetected (Fig. 5). The source of nutrients at this location is unknownsince well R4 is not located near any septic systems, but the lack ofNH4þmay indicate a non-organic source of N, leaky sewage pipes, orpossibly a lack of mineralization due to low residence times as aresult of a large recharge flux.

3.4.3. Non-equilibrium reactionsThe FWL on Roi is generally undersaturated with respect to

ples along cross-sections (a) AeA’ and (b) BeB’ on Roi and (c) CeC’ on Namur.

Fig 9. Saturation indices with respect to aragonite plotted as a function of percentseawater for groundwater samples from (a) Roi and (b) Namur compared to thetheoretical mixing model (solid line) for groundwater saturated with carbonate andseawater. Polynomial regression (dotted line) and corresponding r-square values areshown.


aragonite (Fig. 6), which indicates that the residence time of thegroundwater in the FWL is shorter than the rate of equilibriumreactions with the carbonate aquifer. These non-equilibrium con-ditions are likely the result of more recharge and greater water fluxthrough the groundwater system, which allows for discharge offreshwater out of the system at a relatively faster rate. By contrast,the FWL and transition zone on Namur is saturated or supersatu-rated with respect to aragonite (Fig. 6), indicating a longergroundwater residence times on Namur that contributes to moredissolution and the relatively greater average SRC as compared toRoi.

Fig 10. Saturation indices with respect to aragonite plotted as a function of pCO2 forgroundwater samples from Roi and Namur. Linear regressions and corresponding r-square values are shown for Roi (solid line) and Namur (dashed line). The seawatersample is shown for comparison.

3.4.4. Seawater mixingThe resulting SI of seawater mixes with carbonate groundwater

is a function of the pCO2 and SI of the original FWL groundwater(Matthews, 1971). For example, observations of mixing supersatu-rated and/or saturated waters can produce undersaturated water(Thrailkill, 1968; Wigley and Plummer, 1976), and the mixture oftwo undersaturated waters can produce supersaturated water(Runnells, 1969). Here, we used PHREEQC to calculate a theoreticalmixing line for seawater and carbonate groundwater for Roi andNamur to predict and compare the groundwater chemistryassuming mixing only to the calculated SI values from the fieldsamples (Fig. 9). Note that the difference in the theoretical mixinglines on Roi and Namur is attributed to the difference in end-member chemistry that was calculated using the FWL samples onRoi and Namur. The SI values of field samples from Roi follow thenon-linear theoretical mixing line reasonablywell (polynomial bestfit R2 value ¼ 0.43) (Fig. 9a). Similar to the theoretical mixing line,groundwater on Roi skews towards undersaturated waters at lowerseawater percentage and supersaturated water at mixtures con-taining predominantly seawater (>75%) (Fig. 9a) These resultsindicate that seawater-groundwater mixing is a dominantgeochemical process on Roi. Conversely, SI values of field samplesfromNamur do not correspondwith the theoretical mixing line, butgenerally remain close to equilibrium (SI ¼ 0 þ/� 0.2) (Fig. 9b).Although seawater-groundwater mixing occurs on Namur, it isapparently less of an important process than on Roi.

The groundwater samples from Roi have a strong inverse rela-tion (R2 ¼ 0.714) between SI and pCO2 while the samples fromNamur have a weaker relation (R2 ¼ 0.106) of decreasing SI withincreasing pCO2 (Fig. 10). The supersaturated conditions and lowpCO2 from deep parts of the aquifer on Roi resemble seawater(SI ¼ 0.68, log pCO2 ¼ �3.45) (Fig. 10). Conversely, most of thegroundwater samples from Namur are relatively close to saturationwith respect to aragonite (SI ¼ 0 þ/� 0.2), and deep wells haverelatively higher pCO2. These trends in SI and pCO2 are consistentwith a larger flux of water through the aquifer and observations ofundersaturated water in the FWL, which indicate shorter residencetimes of groundwater on Roi. Mixing of seawater and carbonategroundwater is greater on Roi compared to Namur as a result of theincreased recharge and flux through the system, and consequently,residence time in the aquifer on Roi is reduced. Relatively lowerrecharge rates on Namur corresponds to less flushing and slowermixing processes that increases residence time in the FWL, asindicated by waters that are closer to saturation with respect toaragonite (Fig. 10). The higher pCO2 and saturated or slightly under-saturated conditions in deep wells on Namur indicates processesother than mixing are dominant. The apparent CO2 transport out ofthe system on Namur is occurring at a significantly slower rate andthe kinetics of equilibrium and non-equilibrium dissolution-pre-cipitation reactions is a more dominant geochemical process thanon Roi.

4. Summary and conclusions

LULC change and MAR on Roi have altered natural rechargepatterns and hydrogeochemical processes in the FWL. Greater tidalefficiency on the ocean side of atoll islands typically limits the FWLvolume relative to the lagoon side. However, the influx of waterfrom the relatively lower ET and MAR operations on Roi hasincreased recharge and the volume and thickness of the FWL on theocean side. These findings indicate that the ocean side of an atollaquifer has the capacity for additional storage of fresh groundwater.LULC change and MAR have also resulted in a thicker FWL on thelagoon side, which is naturally more efficient in storing freshwaterbecause it has less tidal mixing and relatively slower drainage


compared to the ocean side. On Roi-Namur, the area directlybeneath the concrete-lined catchment basin may have the greatestcapacity to store fresh groundwater due to its central location,reduced ET, and suitable hydrogeology.

LULC change and MAR have also contributed to statistical dif-ferences in pH, ORP, Ca2þ, NH4þ, TDN, and d13C-DIC and relatedhydrogeochemical processes in groundwater of Roi and Namur.Depleted d13C-DIC, lower pH, and higher Ca2þ indicate decay ofvegetation and increased microbial respiration is responsible forthe higher dissolution rate on Namur compared to Roi. The annualincrease in porosity in the FWL and overlying vadose zone is 0.004%on Roi and 0.024% on Namur. The greater dissolution of aragoniteand magnesian-calcite should theoretically produce more Mg2þ

and inorganic carbon species, and consequently differences inalkalinity and HCO3�. However, statistically significant differenceswere not observed for these parameters. These unexpected findingsare likely related to contamination from human activities on theisland, chemically altered recharge water from the MAR system,and increased groundwater flux and reduced residence time. Ob-servations of individual well concentrations provide more insightinto the responsible geochemical processes. High DIC concentra-tions at wells R3 and R4 correlate with elevated pCO2 and nutrientconcentrations. Therefore, changes in groundwater-quality pa-rameters due to LULC change can be masked by geochemical pro-cesses related to contamination from human activity.

There is considerable evidence that residence time is reduced onRoi due to the large influx of water and strong flow gradient towardthe more permeable ocean side. Dissolution rates are higher in theFWL on Namur and thewater is saturatedwith respect to aragonite.In contrast, most shallow wells on Roi have groundwater that isundersaturated with respect to aragonite, which indicates that theresidence time is short relative to the reaction rate for carbonatedissolution. Undersaturated waters are transported and persistdeeper in Roi compared to Namur, and SRC calculations indicatesome dissolution has been shifted to the intermediate transitionzone. In the deepest wells on Roi, the SI values indicate supersat-uration with respect to aragonite and have other geochemicalsimilarities to seawater. These observations and low pCO2 at deeperdepths support a mixing-dominated signal in Roi groundwater. Incontrast, longer residence times on Namur allow equilibrium re-actions to take place over a longer period time, while less mixingreduces CO2 transport out of the system. This interpretation isfurther supported by higher pCO2 in general and groundwater onNamur that is close to saturation with respect to aragonite at mostdepths.

The reduction in natural vegetation and the implementation ofanMAR system on Roi has increased potable groundwater supply inthe FWL. The LULC change has also altered natural geochemicalreactions and reduced dissolution rates in the FWL without detri-mental effects to the overall groundwater quality. Aside fromseawater intrusion of the FWL, contamination from human activ-ities poses the greatest threat to groundwater quality, but this canbe mitigated by a groundwater management plan that emphasizesoversight of this limited and vulnerable resource. Findings fromRoi-Namur Island support selective LULC change and MAR as apromising management approach for communities on other low-lying atoll islands to increase the resilience of their groundwatersupplies and help them adapt to future climate change relatedstresses. The substantially larger FWL on Roi compared to Namurillustrates the potential for targeted LULC change and MAR to in-crease the sustainability of freshwater resources and the resilienceof atoll communities during drought. Such management strategiesmay also have the potential to help prolong the habitability of manyatolls globally, particularly those that are relatively less vulnerableto SLR-induced inundation, but relativelymore threatened bywater

scarcity.

Acknowledgments

Funding for this research was provided by the U.S. GeologicalSurvey; U.S. Department of Defense's Strategic EnvironmentalResearch and Development Program (SERDP) under Project RC-2334; Research Institute for Humanity and Nature (RIHN) R-08-Init project; and a student travel grant from the California StateUniversity (CSU) Council on Ocean Affairs, Science, and Technology(COAST). The IAEA is grateful for the support provided to its Envi-ronment Laboratories by the Government of the Principality ofMonaco. We thank Dr. Mary Leech (SFSU) for constructive com-ments on previous versions of this paper. Carol Golby-Saunders andEric Nystrom (Roi-Namur water treatment plant) and Stanley Jaz-winski (Liquid SystemsManager) assisted with field data collection.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.apgeochem.2017.03.006.

References

Anthony, S.S., Peterson, F.L., MacKenzie, F.T., Hamlin, S.N., 1989. Geohydrology of theLaura fresh-water lens, Majuro atoll: a hydrogeochemical approach. Geol. Soc.Am. Bull. 101, 1066e1075.

Ayers, J.F., Vacher, H.L., 1986. Hydrogeology of an atoll island: a conceptual modelfrom detailed study of a micronesian example. Groundwater 24, 185e198.

Bailey, R.T., Jenson, J.W., Olsen, A.E., 2010. Estimating the ground water resources ofatoll islands. Water 2, 1e27. http://dx.doi.org/10.3390/w2010001.

Bailey, R.T., Jenson, J.W., Olsen, A.E., 2009. Numerical modeling of atoll island hy-drogeology. Ground Water 47, 184e196. http://dx.doi.org/10.1111/j.1745-6584.2008.00520.x.

Bailey, R.T., Jenson, J.W., Rubinstein, D., Olsen, A.E., 2008. Groundwater Resources ofAtoll Islands: Observations, Modeling, and Management. Water and Environ-mental Research Institute of the Western Pacific, University of Guam.

Bailey, R.T., Jenson, J.W., Taboro�si, D., 2013. Estimating the freshwater-lens thicknessof atoll islands in the Federated States of micronesia. Hydrogeol. J. 21, 441e457.http://dx.doi.org/10.1007/s10040-012-0923-6.

Ball, J.W., Nordstrom, D.K., 1991. User's Manual for WATEQ4F, with Revised Ther-modynamic Data Base and Test Cases for Calculating Speciation of Major, Trace,and Redox Elements in Natural Waters.

Buddemeier, R.W., 1981. Geohydrology of Enewetak Atoll Islands and Reefs. Law-rence Livermore National Lab, CA (USA).

Buddemeier, R.W., Oberdorfer, J.A., 1997. Hydrogeology of Enewetak atoll. In: Ge-ology and Hydrogeology of Carbonate Islands. Elsevier, pp. 667e692.

Buddemeier, R.W., Oberdorfer, J.A., 1986. Internal hydrology and geochemistry ofcoral reefs and atoll islands: key to diagenetic variations. In: Schroeder, J.H.,Purser, B.H. (Eds.), Reef Diagenesis. Springer Berlin Heidelberg, Berlin, Heidel-berg, pp. 91e111.

Dickinson, W.R., 2009. Pacific atoll living: how long already and until when? GSAToday 19, 4e10, 0.1130/GSATG35A.1.

Drever, J.I., 1988. The Geochemistry of Natural Waters, second ed. Prentice Hall,Englewood Cliffs, N.J.

Emery, K.O., Tracey Jr., J.I., Ladd, H.S., 1954. Geology of Bikini and Nearby Atolls(Marshall Islands).

Ganguli, P.M., Swarzenski, P.W., Dulaiova, H., Glenn, C.R., Flegal, A.R., 2014. Mercurydynamics in a coastal aquifer: Maunalua Bay, Oahu. Hawaii. Estuar. Coast. ShelfSci. 140, 52e65. http://dx.doi.org/10.1016/j.ecss.2014.01.012.

Gingerich, S.B., 1996. Ground-water Resources and Contamination at Roi-NamurIsland, Kwajalein Atoll, Republic of the Marshall Islands, 199091 (Water-Re-sources Investigations Report No. 95e4275). U.S. Geological Survey, Reston, VA.

Gingerich, S.B., 1992. Numerical Simulation of the Freshwater Lens on Roi-NamurIsland, Kwajalein Atoll Repubic of the Marshall Islands. University of Hawaii,Honolulu, Hawaii.

Golby-Saunders, C.J., 2015. Personal Communications, Roi-Namur Water TreatmentPlant Operations Manager.

Gonneea, M.E., Charette, M.A., 2014. Hydrologic controls on nutrient cycling in anunconfined coastal aquifer. Environ. Sci. Technol. 48, 14178e14185.

Gonneea, M.E., Charette, M.A., Liu, Q., Herrera-Silveira, J., Morales-Ojeda, S., 2014.Trace element geochemistry of groundwater in a karst subterranean estuary(Yucatan Peninsula, Mexico). Geochim. Cosmochim. Acta 132, 31e49.

Gurdak, J.J., Allen, D.M., Holding, S., Carvalho Resende, T., Leblanc, M., Aureli, A.,Hejazian, M., Swarzenski, P.W., Antoniou, A., Dumont, A., 2015. GraphicGroundwater and Climate Change. Small Island Developing States (SIDS).United Nations Educational, Scientific, and Cultural Organization (UNESCO)-

http://dx.doi.org/10.1016/j.apgeochem.2017.03.006http://dx.doi.org/10.1016/j.apgeochem.2017.03.006http://refhub.elsevier.com/S0883-2927(16)30139-1/sref1http://refhub.elsevier.com/S0883-2927(16)30139-1/sref1http://refhub.elsevier.com/S0883-2927(16)30139-1/sref1http://refhub.elsevier.com/S0883-2927(16)30139-1/sref1http://refhub.elsevier.com/S0883-2927(16)30139-1/sref2http://refhub.elsevier.com/S0883-2927(16)30139-1/sref2http://refhub.elsevier.com/S0883-2927(16)30139-1/sref2http://dx.doi.org/10.3390/w2010001http://dx.doi.org/10.1111/j.1745-6584.2008.00520.xhttp://dx.doi.org/10.1111/j.1745-6584.2008.00520.xhttp://refhub.elsevier.com/S0883-2927(16)30139-1/sref5http://refhub.elsevier.com/S0883-2927(16)30139-1/sref5http://refhub.elsevier.com/S0883-2927(16)30139-1/sref5http://dx.doi.org/10.1007/s10040-012-0923-6http://refhub.elsevier.com/S0883-2927(16)30139-1/sref7http://refhub.elsevier.com/S0883-2927(16)30139-1/sref7http://refhub.elsevier.com/S0883-2927(16)30139-1/sref7http://refhub.elsevier.com/S0883-2927(16)30139-1/sref8http://refhub.elsevier.com/S0883-2927(16)30139-1/sref8http://refhub.elsevier.com/S0883-2927(16)30139-1/sref9http://refhub.elsevier.com/S0883-2927(16)30139-1/sref9http://refhub.elsevier.com/S0883-2927(16)30139-1/sref9http://refhub.elsevier.com/S0883-2927(16)30139-1/sref10http://refhub.elsevier.com/S0883-2927(16)30139-1/sref10http://refhub.elsevier.com/S0883-2927(16)30139-1/sref10http://refhub.elsevier.com/S0883-2927(16)30139-1/sref10http://refhub.elsevier.com/S0883-2927(16)30139-1/sref10http://refhub.elsevier.com/S0883-2927(16)30139-1/sref11http://refhub.elsevier.com/S0883-2927(16)30139-1/sref11http://refhub.elsevier.com/S0883-2927(16)30139-1/sref11http://refhub.elsevier.com/S0883-2927(16)30139-1/sref12http://refhub.elsevier.com/S0883-2927(16)30139-1/sref12http://refhub.elsevier.com/S0883-2927(16)30139-1/sref13http://refhub.elsevier.com/S0883-2927(16)30139-1/sref13http://dx.doi.org/10.1016/j.ecss.2014.01.012http://refhub.elsevier.com/S0883-2927(16)30139-1/sref15http://refhub.elsevier.com/S0883-2927(16)30139-1/sref15http://refhub.elsevier.com/S0883-2927(16)30139-1/sref15http://refhub.elsevier.com/S0883-2927(16)30139-1/sref15http://refhub.elsevier.com/S0883-2927(16)30139-1/sref16http://refhub.elsevier.com/S0883-2927(16)30139-1/sref16http://refhub.elsevier.com/S0883-2927(16)30139-1/sref16http://refhub.elsevier.com/S0883-2927(16)30139-1/sref17http://refhub.elsevier.com/S0883-2927(16)30139-1/sref17http://refhub.elsevier.com/S0883-2927(16)30139-1/sref17http://refhub.elsevier.com/S0883-2927(16)30139-1/sref18http://refhub.elsevier.com/S0883-2927(16)30139-1/sref18http://refhub.elsevier.com/S0883-2927(16)30139-1/sref18http://refhub.elsevier.com/S0883-2927(16)30139-1/sref18http://refhub.elsevier.com/S0883-2927(16)30139-1/sref19http://refhub.elsevier.com/S0883-2927(16)30139-1/sref19http://refhub.elsevier.com/S0883-2927(16)30139-1/sref19http://refhub.elsevier.com/S0883-2927(16)30139-1/sref19


International Hydrologic Programme (IHP), Paris, France.Hem, J.D., 1985. Study and Interpretation of the Chemical Characteristics of Natural

Water. Department of the Interior, US Geological Survey.Holding, S., Allen, D.M., Foster, S., Hsieh, A., Larocque, I., Klassen, J., Van Pelt, S.C.,

2016. Groundwater vulnerability on small islands. Nat. Clim. Change 6,1100e1103.

Hunt, C.D., Peterson, F.L., 1980. WRRCTR No. 126 Groundwater Resources of Kwa-jalein Island (Marshall Islands).

Jevrejeva, S., Moore, J.C., Grinsted, A., 2010. Howwill sea level respond to changes innatural and anthropogenic forcings by 2100? Geophys. Res. Lett. 37 http://dx.doi.org/10.1029/2010GL042947.

Karnauskas, K.B., Donnelly, J.P., Anchukaitis, K.J., 2016. Future freshwater stress forisland populations. Nat. Clim. Change. http://dx.doi.org/10.1038/nclimate2987.

Korom, S.F., 1992. Natural denitrification in the saturated zone: a review. WaterResour. Res. 28, 1657e1668. http://dx.doi.org/10.1029/92WR00252.

Matthews, R.K., 1971. Diagenetic environments of possible importance to theexplanation of cementation fabric in subaerially exposed carbonate sediments.Carbonate Cem. Johns Hopkins Univ. Stud. Geol. vol. 19, 127e132.

Morse, J.W., Mackenzie, F.T., 1990. Geochemistry of Sedimentary Carbonates.Elsevier.

Nath, S., Roberts, J.L., Madhoo, Y.N. (Eds.), 2010. Saving Small Island DevelopingStates: Environmental and Natural Resource Challenges. CommonwealthSecretariat, London, UK.

Oberdorfer, J.A., Hogan, P.J., Buddemeier, R.W., 1990. Atoll island hydrogeology: flowand freshwater occurrence in a tidally dominated system. J. Hydrol. 120,327e340. http://dx.doi.org/10.1016/0022-1694(90)90157-S.

Pachauri, R.K., Allen, M.R., Barros, V.R., Broome, J., Cramer, W., Christ, R., Church, J.A.,Clarke, L., Dahe, Q., Dasgupta, P., others, 2014. Climate Change 2014: SynthesisReport. Contribution of Working Groups I, II and III to the Fifth AssessmentReport of the Intergovernmental Panel on Climate Change.

Parkhurst, D.L., Appelo, C.A.J., 2013. Description of Input and Examples for PHREEQCVersion 3: a Computer Program for Speciation, Batch-reaction, One-dimen-sional Transport, and Inverse Geochemical Calculations (Report No. 6eA43).Techniques and Methods, Reston, VA.

Plummer, L.N., 1975. Mixing of sea water with calcium carbonate ground water.Geol. Soc. Am. Mem. 142, 219e236.

Plummer, L.N., Vacher, H.L., Mackenzie, F.T., Bricker, O.P., Land, L.S., 1976. Hydro-geochemistry of Bermuda: a case history of ground-water diagenesis of bio-calcarenites. Geol. Soc. Am. Bull. 87, 1301. http://dx.doi.org/10.1130/0016-7606(1976)872.0.CO;2.

Racz, A.J., Fisher, A.T., Schmidt, C.M., Lockwood, B.S., Huertos, M.L., July/August 2012.Spatial and temporal infiltration dynamics during managed aquifer recharge.Ground Water 50 (4), 562e570. http://dx.doi.org/10.1111/j.1745-6584.2011.00875.x.

Raitt, R.W., Perkins, B., 1954. Seismic-refraction Studies of Bikini and KwajaleinAtolls. US Government Printing Office.

Runnells, D.D., 1969. Diagenesis, chemical sediments, and the mixing of naturalwaters. SEPM J. Sediment. Res. 39 http://dx.doi.org/10.1306/74D71DCA-2B21-11D7-8648000102C1865D.

Russo, T.A., Fisher, A.T., Lockwood, B.S., 2015. Assessment of managed aquiferrecharge site suitability using a GIS and modeling. Groundwater 53, 389e400.http://dx.doi.org/10.1111/gwat.12213.

Schmidt, C.M., Fisher, A.T., Racz, A.J., Lockwood, B.S., Huertos, M.L., 2011. Linking

denitrification and infiltration rates during managed groundwater recharge.Environ. Sci. Technol. 45, 9634e9640. http://dx.doi.org/10.1021/es2023626.

Shim, M.-J., Swarzenski, P.W., Shiller, A.M., 2012. Dissolved and colloidal trace ele-ments in the Mississippi river delta Outflow after hurricanes katrina and Rita.Cont. Shelf Res. 42, 1e9.

Spennemann, D., 2006. Freshwater lens, settlement patterns, resource use andconnectivity in the Marshall Islands. Transform. Cult. EJournal 1.

Storlazzi, C.D., Elias, E.P.L., Berkowitz, P., 2015. Many atolls may be uninhabitablewithin decades due to climate change. Sci. Rep. 5, 14546. http://dx.doi.org/10.1038/srep14546.

Swarzenski, P.W., Izbicki, J.A., 2009. Examining coastal exchange processes within asandy beach using geochemical tracers, seepage meters and electrical re-sistivity. Estuar. Coast. Shelf Sci. 83, 77e89. http://dx.doi.org/10.1016/j.ecss.2009.03.027.

Swarzenski, P.W., Orem, W.G., McPherson, B.F., Baskaran, M., Wan, Y., 2006.Biogeochemical transport in the Loxahatchee river estuary: the role of sub-marine groundwater discharge. Mar. Chem. 101, 248e265.

Swarzenski, P.W., Simonds, F.W., Paulson, T., Kruse, S., Reich, C.D., 2007.A geochemical and geophysical examination of submarine groundwaterdischarge and associatednutrient loading estimates into Lynch Cove, HoodCanal, WA. Environ. Sci. Technol. 41, 7022e7029.

Thrailkill, J., 1968. Chemical and hydrologic factors in the excavation of limestonecaves. Geol. Soc. Am. Bull. 79, 19. http://dx.doi.org/10.1130/0016-7606(1968)79[19:CAHFIT]2.0.CO;2.

Thurber, D.L., Broecker, W.S., Blanchard, R.L., Potratz, H.A., 1965. Uranium-seriesages of Pacific atoll coral. Science 149, 55e58.

Treidel, H., Martin-Bordes, J.L., Gurdak, J.J. (Eds.), 2012. Climate Change Effects onGroundwater Resources: a Global Synthesis of Findings and Recommendations,first ed. CRC Press/Balkema, Leiden, Netherlands ; Boca Raton.

Tribble, G.W., 1997. Ground-water Geochemistry of Kwajalein Island, Republic of theMarshall Islands, 1991 (Water-Resources Investigations Report No. 97e4184).Geological Survey, Honolulu, HI (United States) (Water Resources Div., Hono-lulu, Hawaii).

Underwood, M.R., Peterson, F.L., Voss, C.I., 1992. Groundwatr lens dynamics of atollislands. Water Resour. Res. 28, 2889e29902.

U.S. Geological Survey, variously dated. National field manual for the collection ofwater-quality data (U.S. Geological Survey Techniques of Water-Resources In-vestigations), book 9, chaps. A1eA9. U.S. Geological Survey, Reston, VA.

USEPA, 2016. Drinking Water Contaminants. U.S. Environmental Protection Agency.Vandenbohede, A., Van Houtte, E., Lebbe, L., 2009. Sustainable groundwater

extraction in coastal areas: a Belgian example. Environ. Geol. 57, 735e747.Voss, C.I., 1984. SUTRA (Saturated-unsaturated Transport). A Finite-element Simu-

lation Model for Saturated-unsaturated, Fluid-Density-Dependent Ground-Water Flow with Energy Transport or Chemically-reactive Single-species So-lute Transport (DTIC Document).

Wardlaw, B.R., Quinn, T.M., 1991. The record of Pliocene sea-level change at Ene-wetak Atoll. Quat. Sci. Rev. 10, 247e258. http://dx.doi.org/10.1016/0277-3791(91)90023-N.

White, I., Falkland, T., Metutera, T., Metai, E., Overmars, M., Perez, P., Dray, A., 2007.Climatic and human influences on groundwater in low atolls. Vadose Zone J. 6,581. http://dx.doi.org/10.2136/vzj2006.0092.

Wigley, T.M.L., Plummer, L.N., 1976. Mixing of carbonate waters. Geochim. Cosmo-chim. Acta 40, 989e995. http://dx.doi.org/10.1016/0016-7037(76)90041-7.

http://refhub.elsevier.com/S0883-2927(16)30139-1/sref19http://refhub.elsevier.com/S0883-2927(16)30139-1/sref20http://refhub.elsevier.com/S0883-2927(16)30139-1/sref20http://refhub.elsevier.com/S0883-2927(16)30139-1/sref21http://refhub.elsevier.com/S0883-2927(16)30139-1/sref21http://refhub.elsevier.com/S0883-2927(16)30139-1/sref21http://refhub.elsevier.com/S0883-2927(16)30139-1/sref21http://refhub.elsevier.com/S0883-2927(16)30139-1/sref22http://refhub.elsevier.com/S0883-2927(16)30139-1/sref22http://dx.doi.org/10.1029/2010GL042947http://dx.doi.org/10.1029/2010GL042947http://dx.doi.org/10.1038/nclimate2987http://dx.doi.org/10.1029/92WR00252http://refhub.elsevier.com/S0883-2927(16)30139-1/sref26http://refhub.elsevier.com/S0883-2927(16)30139-1/sref26http://refhub.elsevier.com/S0883-2927(16)30139-1/sref26http://refhub.elsevier.com/S0883-2927(16)30139-1/sref26http://refhub.elsevier.com/S0883-2927(16)30139-1/sref27http://refhub.elsevier.com/S0883-2927(16)30139-1/sref27http://refhub.elsevier.com/S0883-2927(16)30139-1/sref28http://refhub.elsevier.com/S0883-2927(16)30139-1/sref28http://refhub.elsevier.com/S0883-2927(16)30139-1/sref28http://dx.doi.org/10.1016/0022-1694(90)90157-Shttp://refhub.elsevier.com/S0883-2927(16)30139-1/sref30http://refhub.elsevier.com/S0883-2927(16)30139-1/sref30http://refhub.elsevier.com/S0883-2927(16)30139-1/sref30http://refhub.elsevier.com/S0883-2927(16)30139-1/sref30http://refhub.elsevier.com/S0883-2927(16)30139-1/sref31http://refhub.elsevier.com/S0883-2927(16)30139-1/sref31http://refhub.elsevier.com/S0883-2927(16)30139-1/sref31http://refhub.elsevier.com/S0883-2927(16)30139-1/sref31http://refhub.elsevier.com/S0883-2927(16)30139-1/sref31http://refhub.elsevier.com/S0883-2927(16)30139-1/sref32http://refhub.elsevier.com/S0883-2927(16)30139-1/sref32http://refhub.elsevier.com/S0883-2927(16)30139-1/sref32http://dx.doi.org/10.1130/0016-7606(1976)872.0.CO;2http://dx.doi.org/10.1130/0016-7606(1976)872.0.CO;2http://dx.doi.org/10.1111/j.1745-6584.2011.00875.xhttp://dx.doi.org/10.1111/j.1745-6584.2011.00875.xhttp://refhub.elsevier.com/S0883-2927(16)30139-1/sref35http://refhub.elsevier.com/S0883-2927(16)30139-1/sref35http://dx.doi.org/10.1306/74D71DCA-2B21-11D7-8648000102C1865Dhttp://dx.doi.org/10.1306/74D71DCA-2B21-11D7-8648000102C1865Dhttp://dx.doi.org/10.1111/gwat.12213http://dx.doi.org/10.1021/es2023626http://refhub.elsevier.com/S0883-2927(16)30139-1/sref39http://refhub.elsevier.com/S0883-2927(16)30139-1/sref39http://refhub.elsevier.com/S0883-2927(16)30139-1/sref39http://refhub.elsevier.com/S0883-2927(16)30139-1/sref39http://refhub.elsevier.com/S0883-2927(16)30139-1/sref40http://refhub.elsevier.com/S0883-2927(16)30139-1/sref40http://dx.doi.org/10.1038/srep14546http://dx.doi.org/10.1038/srep14546http://dx.doi.org/10.1016/j.ecss.2009.03.027http://dx.doi.org/10.1016/j.ecss.2009.03.027http://refhub.elsevier.com/S0883-2927(16)30139-1/sref43http://refhub.elsevier.com/S0883-2927(16)30139-1/sref43http://refhub.elsevier.com/S0883-2927(16)30139-1/sref43http://refhub.elsevier.com/S0883-2927(16)30139-1/sref43http://refhub.elsevier.com/S0883-2927(16)30139-1/sref44http://refhub.elsevier.com/S0883-2927(16)30139-1/sref44http://refhub.elsevier.com/S0883-2927(16)30139-1/sref44http://refhub.elsevier.com/S0883-2927(16)30139-1/sref44http://refhub.elsevier.com/S0883-2927(16)30139-1/sref44http://dx.doi.org/10.1130/0016-7606(1968)79[19:CAHFIT]2.0.CO;2http://dx.doi.org/10.1130/0016-7606(1968)79[19:CAHFIT]2.0.CO;2http://refhub.elsevier.com/S0883-2927(16)30139-1/sref46http://refhub.elsevier.com/S0883-2927(16)30139-1/sref46http://refhub.elsevier.com/S0883-2927(16)30139-1/sref46http://refhub.elsevier.com/S0883-2927(16)30139-1/sref47http://refhub.elsevier.com/S0883-2927(16)30139-1/sref47http://refhub.elsevier.com/S0883-2927(16)30139-1/sref47http://refhub.elsevier.com/S0883-2927(16)30139-1/sref48http://refhub.elsevier.com/S0883-2927(16)30139-1/sref48http://refhub.elsevier.com/S0883-2927(16)30139-1/sref48http://refhub.elsevier.com/S0883-2927(16)30139-1/sref48http://refhub.elsevier.com/S0883-2927(16)30139-1/sref48http://refhub.elsevier.com/S0883-2927(16)30139-1/sref49http://refhub.elsevier.com/S0883-2927(16)30139-1/sref49http://refhub.elsevier.com/S0883-2927(16)30139-1/sref49http://refhub.elsevier.com/S0883-2927(16)30139-1/sref51http://refhub.elsevier.com/S0883-2927(16)30139-1/sref52http://refhub.elsevier.com/S0883-2927(16)30139-1/sref52http://refhub.elsevier.com/S0883-2927(16)30139-1/sref52http://refhub.elsevier.com/S0883-2927(16)30139-1/sref53http://refhub.elsevier.com/S0883-2927(16)30139-1/sref53http://refhub.elsevier.com/S0883-2927(16)30139-1/sref53http://refhub.elsevier.com/S0883-2927(16)30139-1/sref53http://dx.doi.org/10.1016/0277-3791(91)90023-Nhttp://dx.doi.org/10.1016/0277-3791(91)90023-Nhttp://dx.doi.org/10.2136/vzj2006.0092http://dx.doi.org/10.1016/0016-7037(76)90041-7

Land-use change and managed aquifer recharge effects on the hydrogeochemistry of two contrasting atoll island aquifers, Roi ...1. Introduction2. Materials and methods2.1. Site description2.2. Roi-Namur hydrogeochemistry2.3. Field methods and groundwater sampling2.4. Laboratory methods2.5. Statistical analysis2.6. Geochemical modeling

3. Results and discussion3.1. Freshwater lens asymmetry3.2. Statistical differences in groundwater quality on Roi and Namur3.3. Carbonate dissolution3.4. Equilibrium/non-equilibrium reactions3.4.1. Soil pCO2 and CO2 evasion3.4.2. Groundwater CO2 flux3.4.3. Non-equilibrium reactions3.4.4. Seawater mixing

4. Summary and conclusionsAcknowledgmentsAppendix A. Supplementary dataReferences

Documents

Land-use change and managed aquifer recharge effects on ...online.sfsu.edu/jgurdak/Publications/Hejazian_etal_2017.pdfFreshwater resources on low-lying atoll islands are highly vulnerable