Lithologic controls on hydrologic and geochemical processes in constructed Everglades tree islands

Chemical Geology xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Chemical Geology journal homepage: www.elsevier.com/locate/chemgeo

Lithologic controls on hydrologic and geochemical processes in constructed everglades tree islands ⁎

Andres E. Prieto Estradaa, René M. Pricea,b, , Leonard J. Scintoa,b, Florentin J-M.R. Maurrassea, Thomas W. Dreschelc, Fred H. Sklarc, Eric A. Clinec a

Department of Earth and Environment, Florida International University, 11200 SW 8th St., Miami, FL 33199, USA Southeast Environmental Research Center, Florida International University, 11200 SW 8th St., Miami, FL 33199, USA c Everglades Systems Assessment Section, South Florida Water Management District, West Palm Beach, FL 33406, USA b

A R T I C LE I N FO

A B S T R A C T

Keywords: Tree islands Evapotranspiration Peat Limestone Calcite supersaturation

Tree islands have been hypothesized to induce calcite precipitation by concentrating ions in the groundwater through evapotranspiration processes. This research investigates how lithology regulates the hydrologic and geochemical conditions within two types of constructed Everglades tree islands: peat-based and limestone-core surrounded by peat. Three years of hydrochemical and hydrologic data (2013–2015) obtained from the constructed tree islands suggest that under current climatic and managed hydrologic conditions, calcite dissolution is prevalent within the top meter of soil in both types of tree islands. Mass-balance calculations along groundwater flow paths in one peat-based tree island that contained clay at depth, indicated that calcite precipitation is likely 1 m below the soil surface. The lithological characteristics of that peat-based island supported a persistently depressed groundwater table, while tree islands with elevated ratios of peat-to-sand content did not. Limestone-core tree islands also supported a depressed water table, particularly during the dry season. This study determined that the lithology of a set of man-made Everglades tree islands played a primary role in regulating the seasonal fluctuation of the water table and hydrogeochemical processes. Understanding the mechanisms of tree-island formation and maintenance is important for preserving the overall ecosystem function of the freshwater-Everglades.

1. Introduction The influence of tree vegetation on soil mineralization processes and groundwater flow patterns have been documented in wetland environments. For instance, in the Mkuze wetland in South Africa and the Okavango Delta in Botswana, chemical precipitation of calcite, silica, and other minerals was found to occur in soils of tree islands in response to elevated evapotranspiration (ET) rates (Humphries et al., 2011; McCarthy et al., 2012). Elevated tree-island ET rates in the Okavango Delta also results in a density-driven accumulation of saline water into deep sand layers beneath clayey soils across the wetland, contributing to the salt balance of the ecosystem (Bauer-Gottwein et al., 2007; Ramberg and Wolski, 2008). The unique climatic and geologic conditions of each wetland add to the complexity of understanding wetlands ecohydrological functioning. However, the similar ET-driven patterns of groundwater flow and hydrochemical concentrations below

tree islands around the world frame the scope of this investigation, which benefits from long-term geochemical and hydrologic monitoring in a managed Everglades setting. Some of the vital ecosystem services provided by the freshwaterEverglades wetland, such as water quality, carbon storage, and nutrient cycling, result from the complex biogeochemical interplay between its geomorphological features (Maltby and Dugan, 1994). These distinct features, namely, sloughs, sawgrass ridges, and tree islands are positioned on a large carbonate platform (Wetzel et al., 2011). Tree islands represent only 2% of the landscape, but play a critical role in maintaining oligotrophy in the Everglades freshwater ecosystem by functioning as biogeochemical hot spots (Gleason and Stone, 1994; Wetzel et al., 2005, 2009, 2011). Seasonal variations in evapotranspiration (ET) rates have been reported to induce drawing water into the tree islands, thereby accumulating ions in the groundwater, precipitating calcium carbonate minerals, and storing phosphorus (Ross et al., 2006;

Abbreviations: WCAs, Water Conservation Areas; ET, evapotranspiration; LNWR, Loxahatchee National Wildlife Refuge; LILA, Loxahatchee Impoundment Landscape Assessment; SRS, Shark River Slough; M1E, Eastern tree island in macrocosm 1; M1W, Western tree island in macrocosm 1; M2E, Eastern tree island in macrocosm 2; M2W, Western tree island in macrocosm 2; M3E, Eastern tree island in macrocosm 3; M3W, Western tree island in macrocosm 3; W4E, Eastern tree island in macrocosm 4; M4W, Western tree island in macrocosm 4 ⁎ Corresponding author at: Department of Earth and Environment, Florida International University, 11200 SW 8th St., Miami, FL 33199, USA. E-mail addresses: aprie050@fiu.edu (A.E. Prieto Estrada), pricer@fiu.edu (R.M. Price). https://doi.org/10.1016/j.chemgeo.2018.04.001 Received 15 September 2017; Received in revised form 14 March 2018; Accepted 2 April 2018 0009-2541/ © 2018 Published by Elsevier B.V.

Please cite this article as: Prieto Estrada, A.E., Chemical Geology (2018), https://doi.org/10.1016/j.chemgeo.2018.04.001

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 1. Map showing the location and description of the Loxahatchee Impound Landscape Assessment (LILA) facility in relation to the Water Conservation Areas (WCAs) and Shark River Slough (SRS), in south Florida, USA. The LILA map illustrates the location of the 8 tree islands, their lithologic composition and groundwater well locations.

Most of the remaining Everglades tree islands are located at the western edge of WCA3, in the Loxahatchee National Wildlife Refuge (LNWR, aka WCA1) in the northern Everglades, and in Shark River Slough (SRS) in the southern Everglades (Wetzel et al., 2005; Fig. 1). General explanations concerning tree-island formation suggest that small and rounded tree islands in the northern Everglades (WCA1 and northern WCA2), also known as “battery” islands, formed about 1300 years ago as woody plants colonized floating masses of mixed water lily and sawgrass peat (Loxahatchee peat) (Gleason et al., 1974; Gleason and Stone, 1994). Conversely, large and elongated “fixed” tree islands in the southern Everglades formed along the direction of water flow on relict topography of the substrate, with limestone or sand mounds underlying the peat soils (Craighead, 1974; Gleason et al., 1974). A series of archaeological excavations within the peat soils of 20 large tree islands in central SRS revealed the presence of calcium

Wetzel et al., 2011; Sullivan et al., 2014a; Sullivan et al., 2016). However, in the present Everglades, the biogeochemical functioning of tree islands has been compromised by water management practices meant to “reclaim” the wetlands for expanding development (Light and Dineen, 1994). For instance, drainage by canals and increased water storage in Water Conservation Areas (WCAs) 1, 2 and 3 (Fig. 1) modified the natural hydrologic flow in the Florida Everglades (Light and Dineen, 1994). In the past century, this alteration has led to the degradation of tree islands either because of increased fire frequency in dried areas or increased drowning of trees in increasingly flooded areas (Sklar and van der Valk, 2002). Hence, there has been a great reduction in the number of tree islands, i.e. approximately 90% and 60% disappearance in WCA2 and WCA3, respectively, in the Everglades conceding valuable ecosystem services (Davis et al., 1994; Patterson and Finck, 1999; Larsen and Harvey, 2010). 2

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

contrast. The limestone-core tree islands contain limestone rubble (excavated elsewhere) in the center of the tree islands, surrounded on the top and sides with the organic soils excavated from the adjacent slough area (Van Der Valk et al., 2008). The grass ridges (Fig. 2) were the only undisturbed topographic features during the construction process of LILA (Sklar et al., 2004). Over 700 tree saplings from ten species native to the Everglades were planted on each tree island (Sullivan et al., 2011; Stofella et al., 2010). The tree islands in macrocosms 1 and 4 (M1 and M4) were planted in March 2006, while the tree islands in macrocosms 2 and 3 (M2 and M3) were planted in March 2007 (Sullivan et al., 2011; Fig. 1). The climatic conditions at LILA are like those across south Florida, with distinct wet and dry seasons between June and November and December and May, respectively. The area has an annual average of 133 cm of precipitation (Ali et al., 2000). The surface-water levels in LILA are controlled to simulate the seasonal variations of the surfacewater levels in the Everglades. They are maintained according to an established hydrograph with the highest levels typically occurring from September to January, flooding the tree islands, and the lowest levels occurring from April to June (Sullivan et al., 2011, 2014a, 2016). The water within LILA is managed by a large electric pump (1.84 m3 s−1), a series of water control structures, and recording stage gauges, which allows the management of water levels and flows within each macrocosm (Aich et al., 2011).

carbonate layers with thicknesses of 50 to 70 cm situated at depths of 25 to 50 cm below the soil surface (Schwadron, 2006; Coultas et al., 2008). However, the original environmental conditions, the depositional and cementation processes under which these calcitic horizons formed are poorly understood (Graf et al., 2008). Calcium carbonate layers, aka calcrete or caliche, have been reported in seasonal wetlands where vadose and shallow phreatic groundwaters are saturated with calcium carbonate (Wright, 2007). The main factors implicated in controlling groundwater saturation with respect to CaCO3, including Ca2+ concentration, partial pressure of carbon dioxide (pCO2), and temperature, are largely regulated by the activity of vegetation through evapotranspiration (ET) (Langmuir, 1997; Meyer et al., 2014). Essentially, phreatophytic plants would be able to induce the chemical precipitation of calcrete from groundwater around the capillary fringe above the water-table and around the rooting zone under appropriate thermodynamic conditions (AlonsoZarza and Wright, 2010; Meyer et al., 2014). Groundwater supersaturation with respect to carbonate minerals, caused by ion exclusion during root water uptake, suggests that calcite could accumulate in the soils of Everglades tree islands (Sullivan et al., 2011, 2014a, 2014b). Consequently this mechanism of calcite precipitation may explain the origin of the calcitic horizons found in peat soils of SRS tree islands (Graf et al., 2008). Additionally, ET-driven calcite precipitation could influence P cycling and contribute to soil accretion rates on time scales of centuries to millennia (Sullivan et al., 2016). This study attempts to determine if calcium carbonate is precipitating in Everglades tree islands. The study was conducted on manmade tree islands constructed of two types of substrates: 1) peat; and 2) limestone and peat. Building on previous studies conducted on these man-made islands and on natural Everglades tree islands (Sullivan et al., 2011, 2014a, 2014b, 2016), two main hypotheses were formulated and tested in this study: (1) a water-table depression caused by seasonal ET promotes calcite supersaturation in the groundwater below constructed tree islands during low surface-water stages, and (2), calcite precipitates in the capillary fringe above the water table during the dry season, potentially forming a “calcrete” layer in peat soils of tree islands. This study focused on analyzing and modeling multi-year hydrogeological and hydrochemical data from the two types of constructed tree islands (defined above). Additionally, the chemical and mineralogical compositions of the upper 1 m of soil within the tree islands were characterized. The results provide new insights into the hydrogeochemical functioning of tree islands, as well as supporting ongoing restoration efforts of the Everglades freshwater ecosystem.

2.2. Field sampling and monitoring The LILA facility has an extensive network of PVC-encased wells with an average depth of 1.34 m across the tree islands, ridges, and sloughs (Figs. 1 and 2). All wells have a 50–60 cm screened interval at the bottom (Sullivan, 2011). This well-construction design gives a sturdy foundation for the wells, while allowing the collection of groundwater samples from the piled peat soils and the deeper mineral substrates. The wells located at the edges of the islands are referred to as edge wells, and those at the center of the islands are referred to as center wells. The center wells of the peat-based tree islands are referred to as peat-center wells, while the center wells of the limestone-core tree islands are referred to as limestone-center wells. Each peat-based tree island has an additional well with a depth of 2 m referred to as peatdeep wells. The peat-based island M1W has one additional center well with a depth of 0.6 m referred to as peat-shallow well. Groundwater sampling in the shallow (0.6 m), center (1.34 m), and deep (2.0 m) wells in M1W provided for the investigation of varying hydrogeochemical conditions with soil depth.

2. Materials and methods 2.2.1. Water sampling A total of 27 groundwater samples (Fig. 1) and eight surface water samples were collected biannually in October (wet season) and May (dry season) from October 2013 until October 2015. All samples were collected with a peristaltic pump, and each well was purged of three well volumes before sampling. During sampling, temperature, salinity, and conductivity were measured using a YSI™ meter, and pH was measured using a Thermo-Scientific Orion™ 3-Star pH meter. Two filtered (0.45 μm membrane filter) and two unfiltered samples were collected at each sampling location. At each site, one filtered and one unfiltered sample were preserved with 10% hydrochloric acid (HCl) for cation and total phosphorus analysis, respectively. All samples were stored at 4 °C and transported to Florida International University (FIU).

2.1. Study area This project was conducted at the Loxahatchee Impoundment Landscape Assessment (LILA) facility (Aich et al., 2011), located at the Arthur R. Marshall Loxahatchee National Wildlife Refuge (LNWR), Boynton Beach, FL, USA (Fig. 1). LILA was constructed between 2002 and 2003 as a large-scale physical model of the freshwater-Everglades (Figs. 1–2), and includes tree islands, sawgrass ridges, and slough habitats (Stofella et al., 2010). These physiographic features were “created” by sculpting an existing LNWR impoundment into four 8.1-ha macrocosms, each of which contains two types of tree islands: a solely peat-based island and a limestone-core island capped by approximately 0.3 m of peat (Fig. 1). The peat-based tree islands were constructed to simulate the “battery” islands in the LNWR (northern Everglades), whereas the limestone-core tree islands were built to mimic the “fixed” tree islands in SRS (southern Everglades). Each island measures approximately 43 × 71 m. The top of the tree islands are approximately 0.9 m higher than the bottom of their surrounding sloughs. The peatbased tree islands were constructed from the adjacent organic soils excavated to create the sloughs, and piled to generate the topographic

2.2.2. Water levels Pressure transducers (In-Situ Level TROLL 500®) were deployed in two edge wells, the center, slough, and ridge wells in the eastern half of each macrocosm (Fig. 1), allowing for cross-sectional monitoring of the water table across the created landscape (Fig. 2). In addition, three pressure transducers were installed in PVC-case structures, which acted as stilling wells, in the sloughs of M2E, M3E, and M4E to measure 3

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 2. Schematic cartoon showing the distribution of piezometers monitoring the groundwater levels across the eastern ends of the macrocosms at LILA (courtesy of PK Sullivan).

the groundwater samples were used to evaluate nutrient seasonal dynamics in the LILA groundwater. All filtered water samples were analyzed for total alkalinity and major ions: chloride (Cl−), sulfate (SO42−), calcium (Ca2+), magnesium (Mg2+), sodium (Na+), and potassium (K+) at the FIU Hydrogeology Laboratory using a Brinkman Titrino™ 751 Titrator and a Dionex-120™ Ion Chromatograph, respectively. Total alkalinity was calculated as the concentration of HCO3– by titrating 30 mL of samples with 0.1 N HCl to an inflection point closest to pH = 2. The charge balance error of all water samples was < 5%. Ion concentrations were used to analyze hydrochemical trends, and as inputs in thermodynamic and mass-balance modeling. The filtered samples were also analyzed for the stable isotopic composition of hydrogen (δD) and oxygen (δ18O) using a LGR DTL-100™ Liquid Water Isotope Analyzer at the FIU Hydrogeology Laboratory. The δ18O values and Cl− concentrations were used to trace groundwater movement. Statistical analyses were conducted using IBM® SPSS® Statistics version 23. Two-tailed Pearson correlations were computed to determine significant linear correlations between hydrochemical constituents and geochemical parameters in the water samples, grouped in categories according to sampling location and tree island substrate (peat-based vs limestone-core islands). Additionally, box-and-whisker plots were generated to depict the statistical distribution of mineral saturation states, as a function of seasonality and sampling location.

surface water levels. The pressure transducers were vented to the atmosphere and set to record groundwater levels every 30 min, with an accuracy of ± 3.5 mm. The elevation of the top of each groundwater well and the position of the pressure transducers were determined using standard surveying techniques, relative to the National Geodetic Vertical Datum of 1929 (NGDV29), with a ± 3 mm level of accuracy. Surface water levels at M1E were determined from the South Florida Water Management District (SFWMD) network of recording gauges located at the eastern and western boundaries of each macrocosm at LILA (http://my.sfwmd.gov/dbhydroplsql/show_dbkey_info.main_ menu). A linear interpolation between the stages was used to determine surface water levels along the boundaries of the eastern tree island in M1 (M1E), and to fill in data gaps in the other macrocosms when necessary. 2.2.3. Soil sampling Soil samples were collected from three peat-based tree islands using a soil auger. These islands were chosen because their soils allowed hand augering to 1 m soil depth, which corresponded approximately to the lowest groundwater levels observed in the tree islands. Eight soil cores were obtained between April and May 2015 to assess possible CaCO3 accumulation in the soils of the islands. Two preliminary cores were collected at the center of island M2E from the ground surface in section increments of 8 cm down to 96 cm each, for a total of 24 soil samples. Three additional cores were obtained from the center of island M1W, and three more from the center of island M3E. The samples from M1W and M3E were obtained in increments of 8 cm down to 40 cm, in increments of 4 cm down to 72 cm, and in increments of 8 cm down to 96 cm. The tree island M1W was chosen for being one of the first islands where trees were planted (March 2006), and M3E for showing the lowest groundwater levels at LILA.

2.3.2. Groundwater-flow directions Groundwater-flow directions across the eastern macrocosms were simulated for two seasons using a finite-difference solution to the twodimensional Laplace's equation, which states that the sum of the second order partial derivative of h (hydraulic head) with respect to x and y (spatial variables), is zero (Wang and Anderson, 1982).

2.3. Laboratory analyses and hydrogeochemical modeling

∂ 2h ∂ 2h + 2 =0 ∂x 2 ∂y

2.3.1. Water chemistry and nutrients Unfiltered water samples preserved with HCl were analyzed for total phosphorus (TP) and total organic carbon (TOC) at the Southeastern Environmental Research Center Nutrient Analysis Laboratory (SERCNAL) using an Alpkem Rapid Flow Analyzer with a 2-Channel ER Detector and a Shimadzu TOC-V, respectively. Filtered (0.45 μm) water samples were analyzed for soluble reactive phosphorus (PO43−) and ammonium (NH4+) at the FIU Soil/Sediment Biogeochemistry Laboratory (SBL) using an Alpkem 300 Series 4 Channel Rapid Flow Analyzer. Total and dissolved phosphorus and carbon concentrations in

Assumptions associated with this equation include groundwater flow through an isotropic, homogeneous aquifer under steady-state conditions (Wang and Anderson, 1982). The domain of the model used in this study consisted of a 75-by-5 matrix, with 75 cells representing the distance in meters between the northern edge well and the slough well in the eastern ends of the macrocosms. No-flow conditions were assigned to the left, right, and bottom boundaries of the mathematical domain. The top boundary consisted in a spline interpolation of monthly-averaged groundwater levels, simulating the water table in October 2014 and May 2015 across the eastern ends of the macrocosms. 4

(1)

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

These four boundary conditions were implemented to represent treeisland ET as the groundwater flow driving mechanism.

Table 1 Parameters including calcite saturation index (SIcalcite) used to constrain possible geochemical reactions in mass-balance models for groundwater in the center, edge, and deep wells of tree islands and in the adjacent slough during sampling events in October 2014 and May 2015.

2.3.3. Speciation-solubility and mass-balance modeling The speciation and activities of dissolved ions, including HCO3−, Cl−, SO42−, PO43−, Ca2+, Mg2+, Na+, and K+ were calculated for all water samples using their corresponding pH and temperature values and using the AquaChem software version 2014.1, with the thermodynamic database of the geochemical modeling code PHREEQC version 3.3.5 (Parkhurst and Appelo, 2013). This code was also used to calculate the pCO2 and the Saturation Indices (SI), defined as the logarithm of the ratio of the ion-activity product (IAP) to the solubility product constant (Ksp) (Nordstrom and Campbell, 2014), to determine the mineral saturation state of all water samples.

IAP ⎤ SI = log ⎡ ⎢ K sp ⎥ ⎣ ⎦

Date/Tree island Oct 2014 M1E M2E M3E

M4E May 2015 M1E

(2)

M3E

Samples with SI > 0 were considered supersaturated and mineral precipitation was expected, while samples with SI < 0 were considered undersaturated and mineral dissolution was predicted (Plummer et al., 1983; Langmuir, 1997). Water samples with SI values within ± 0.05 were considered at equilibrium, with both mineral dissolution and precipitation equally expected. The graphical user interface PhreeqcI version 3.3.5 for the code PHREEQC version 3.3.5 (Parkhurst and Appelo, 2013) was used to construct mass-balance models to test hypothesized geochemical reactions. The resultant mass-transfers were selected to account for the observed hydrochemical composition in samples collected in October 2014 and May 2015 along simulated flow paths. Because this geochemical modeling technique requires the definition of all the potential reactants and products in the groundwater system (Plummer et al., 1983; Parkhurst et al., 1993; Parkhurtst, 1997), the following phases were considered to be present in the soils of the LILA macrocosms in both seasons: calcite, CO2(g), H2O(g), and cation exchange (CaX2, MgX2, and NaX). Cation exchange can be represented by the following equation:

2Na+ + CaX2 = 2NaX + Ca+2

M4E

Wellsite

pCO2

SIcalcite

Na+:Cl−

Edge Center Edge Center Edge Center Deep Edge Center

0.16 0.21 0.07 0.14 0.15 0.18 0.08 0.09 0.10

0.00 0.54 −0.32 0.03 0.03 0.98 1.13 0.08 0.63

meq/L 1.5/1.3 2.1/2.9 0.5/0.4 2.8/3.6 1.2/1.2 7.5/9.1 5.4/7.1 1.0/1.0 2.0/2.2

Edge Center Edge Center Deep Slough Edge Center

0.19 0.05 0.14 0.01 0.02 0.01 0.15 0.05

−0.02 0.53 −0.8 1.26 1.44 1.22 −0.75 0.73

1.7/1.9 1.8/4.1 1.0/1.2 3.2/4.2 3.4/4.0 0.9/0.3 1.3/1.4 1.6/1.9

Organic matter was determined by loss-on-ignition (LOI) at 550 °C with an accuracy of ± 2% (Nelson and Sommers, 1996). These results were used to quantify vertical proportions of organic and inorganic carbon concentrations and mineral content in the soils of three peat-based tree islands. Additionally, six samples and their corresponding mineral residues were described for every soil core using a binocular microscope. These observations allowed the determination of the minerals present in the soils. Based on of those determinations, a few smear slides were prepared and analyzed using a polarizing microscope to assess CaCO3 cementation processes. 3. Results

(3)

3.1. Groundwater movement and hydrochemical conditions

where X indicates the exchanger. Sodium is taken up by the exchanger and Ca2+ is released (Langmuir, 1997). Saturation indices, pCO2 values, and Ca2+/Na+ exchange processes (inferred from Na+:Cl− ratios) were used to constrain possible geochemical reactions (Table 1). In addition to the abovementioned phases, the following were assumed to be reacting during the wet season: organic matter (CH2O), methane (CH4), and the hydrated calcium oxalate mineral whewellite (CaC2O4·H2O). These additional carbon-bearing phases were necessarily included to generate models satisfying thermodynamic constraints (Table 1) in a reducing environment. Models that resulted in CaC2O4·H2O dissolution were discarded because of its low solubility at pH values around 6.5 (Ruiz-Agudo et al., 2013; Gadd et al., 2014), such as those observed at LILA. The option for minimal model calculations in PhreeqcI was used to test the most essential geochemical reactions and further reduce the number of output models. Uncertainty limits of 5% were specified for initial and final solutions, as well as for ion concentrations in each solution. As a model simplification, and because the mineralogy of the deeper soils in the LILA macrocosms has not been investigated, the concentrations of dissolved Cl−, K+, PO43−, and SO4− were balanced in all simulations. Thus, mineral sources and sinks for these ions were omitted.

The surface water levels in the LILA macrocosms varied according to the operational hydrograph, simulating the seasonality of the Everglades (Fig. 3). However, two exceptions occurred in macrocosm M1 during hydrologic reversals characterized by a rapid re-flooding and subsequent dry-down, or vice versa. These non-seasonal events took place in April 2014 and in April 2015 over four-week periods (Fig. 3), and were implemented for other scientific projects. Except for the short reversals in M1, the surface water levels in all macrocosms varied similarly throughout the study period, with highest and lowest elevations in October 2014 and July 2015, respectively (Fig. 3). 3.1.1. Groundwater-flow directions Groundwater-flow directions in the peat-based tree islands M2E and M3E differed markedly from each other (Fig. 4). In October 2014, when M2E was flooded (Fig. 4A), groundwater flowed from the center of the island towards the slough and the ridge. In May 2015, groundwater flowed across M2E from the slough towards the southern edge of the island, and from the center of the island towards the northern edge (Fig. 4B). The water table in M2E fluctuated horizontally across the landscape, while a water-table depression persisted in the center of M3E throughout the study period. Groundwater flowed from the edge towards the center of M3E tree island in both seasons (Fig. 4C and D), but in May 2015, groundwater also flowed from the slough towards the center of the island in response to a greater drawdown of the water table. By comparison, in the limestone-core tree islands M1E and M4E

2.3.4. Chemical and microscopic analyses of soil Concentrations of total carbon (TC) and total inorganic carbon (TIC) were measured on oven-dried (80 °C) soil samples by the high-temperature method of dry combustion (Nelson and Sommers, 1996) using a Thermo Finnigan FlashEA® 1112 Nitrogen and Carbon Analyzer. 5

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 3. Hydrograph showing the surface water stages in the LILA macrocosms from October 2013 until October 2015. The timing of the water sampling events and the ground surface elevation of the tree islands, ridges, and sloughs are included. Elevation is given in meters relative to NGVD-29.

undersaturation and equilibrium. The surface water was undersaturated with respect to calcite during the wet seasons (Fig. 9A) and slightly supersaturated during the dry seasons (Fig. 9B). In mass-balance calculations, negative values indicate a phase leaving the aqueous solution (precipitation or gaseous loss from groundwater), while positive values refer to a phase entering the aqueous solution (dissolution or gaseous uptake by groundwater; Tables 3 and 4). Calcite dissolution and CH2O uptake in the October 2014 wet season were the main mass transfers influencing groundwater geochemistry. These geochemical reactions were essential to satisfy SI and pCO2 constraints (Table 1). Only groundwater flow from the deep well to the center well of the M3E island included calcite precipitation and groundwater uptake of CO2. The October-2014 models (Table 3) also included water loss (through ET), except during groundwater discharge from the center of M2E towards the edge of the island. This flow path required CH4 to leave the groundwater system, as no other model solution reactions were found to satisfy the conceptualized water-rock interactions. Furthermore, the precipitation of hydrated calcium oxalate (CaC2O4·H2O) in wet-season models (Table 3) did not regulate calcite dissolution in tree islands M1E, M3E, and M4E, as tested by sensitivity analysis. In both October 2014 and May 2015, Ca2+/Na+ exchange influenced the groundwater chemistry in M1E and M4E (limestone-core), while Ca2+/Mg2+ exchange influenced the groundwater chemistry in M3E (peat-based). Processes of Ca2+/Na+ exchange along groundwater flow paths were also important in all tree islands in May 2015, but the essential geochemical processes were CO2 degassing and calcite precipitation (Table 4). However, the amounts of calcite precipitation in M1E (limestone-core) and M3E (peat-based) in the dry-season models, were an order of magnitude smaller than the amounts of dissolved calcite in the wet-season models. Additionally, calcite dissolution was prevalent in M4E (limestone-core) in both seasons (Tables 3 and 4).

groundwater-flow directions were similar during the dry season in May, but differed from each other during the wet season in October (Fig. 5). In M1E, groundwater flowed towards the center of the island in both seasons (Fig. 5A and B), responding to a continuously depressed water table. Conversely, groundwater flowed across the island M4E from a hydraulic divide formed at its southern edge in October 2014 (Fig. 5C). In May, however, groundwater flowed from both edges and from the slough towards the center of M4E (Fig. 5D). 3.1.2. Ions and nutrients Most ion and nutrient concentrations in the water samples, particularly in those collected from the peat-shallow, peat-deep, peat-center, and limestone-center wells, varied seasonally with higher and lower values in October and May, respectively (Fig. 6B through E). Conversely, the nutrient and ion concentrations in the slough, ridge, and occasionally the edge groundwater, were relatively stable throughout the study period (Fig. 6B through E). Ion and nutrient concentrations were highest in peat-center groundwater compared to the limestonecenter groundwater (Fig. 6B through E). The concentrations of TP in all groundwater samples followed a similar trend since October 2014, with higher and lower values in October and May, respectively, while the opposite trend was observed in surface water samples (Fig. 6E). In the peat-center groundwater, both TP (Fig. 7A) and Cl− concentrations (Fig. 7B) were significantly and linearly correlated with TOC concentrations. Furthermore, an excess in dissolved Ca2+ (Fig. 8A) and HCO3– (Fig. 8B) with respect to Cl− concentrations was observed in all samples. The groundwater samples showed a depletion in dissolved Na+ with respect to Cl− concentrations, when the latter exceeded 2 meq/L (Fig. 8C). 3.2. Hydrogeochemical modeling Saturation indices with respect to calcite varied by season and by sampling location (Fig. 9). All the peat-shallow and peat-deep groundwater samples were supersaturated with respect to calcite and dolomite at all sampling events. The calcite saturation indices in the edge, ridge, and slough groundwater oscillated between

3.3. Soil mineralogy and chemistry Results from the LOI method, shown in Fig. 10, revealed different vertical distributions of mineral content and carbon concentrations 6

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 4. Monthly-averaged directions of groundwater flow (arrowed lines) and water-table elevations (continuous blue lines) across two macrocosms containing peatbased tree islands: a M2E (October 2014); b M2E (May 2015); c M3E (October 2014); d M3E (May 2015). The dashed lines represent the ground-surface elevation and the vertical lines indicate monitoring wells. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

within the first meter of soil of the peat-based tree islands. Peaks of elevated TC/TIC concentrations occurred at different depths and varied systematically alongside with mineral content (shown as mineral residue in Fig. 10) in every soil core. Specifically, elevated carbon concentrations were generally associated with low mineral content, and vice versa. Elevated mineral content in all the soil profiles was observed in the first 40 cm below the ground surface (Fig. 10). The soils of M1W and M2E showed an increased mineral content at depths approximately

70 cm below the ground surface (Fig. 10A through D), whereas the soil of M3E exhibited a decreasing mineral content with depth below the ground surface (Fig. 10E). The structure of the investigated peat soils, as well as the components accounting for their TIC concentrations and mineral content, were analyzed using a binocular microscope (Fig. 11). In sampling intervals with high TC/TIC concentrations and low mineral content (e.g., 50 to 70 cm below the ground surface in Fig.10C), the soil consisted in a 7

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 5. Monthly-averaged directions of groundwater flow (arrowed lines) and water-table elevations (continuous blue lines) across two macrocosms containing limestone-core tree islands: a M1E (October 2014); b M1E (May 2015); c M4E (October 2014); d M4E (May 2015). The dashed lines represent the ground-surface elevation and the vertical lines indicate monitoring wells. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

spongy, fine-grained, dark-brown-to-black peat matrix mixed with sawgrass fragments and fine rootlets (Fig. 11A). Additionally, in the upper 40 cm below the ground surface, samples with both elevated mineral content and TC concentrations (e.g., 0 to 30 cm below the

ground in Fig. 10A) were associated with hardwood-vegetation roots and quartz grains (Fig. 11B). Quartz grains with a medium-to-fine size range (~500–125 μm) and moderately well sorted were found as the main mineral component in 8

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 6. Time-series graphs of: a δ18O enrichment; b Cl−; c Ca2+; d TOC; and e TP concentrations averaged by sampling location. The vertical bars represent one standard error.

calcination, supporting these observations (Fig.11E). Additionally, a few aggregates of quartz grains cemented with CaCO3, were found randomly distributed within the original (non-calcinated) samples obtained from the first 24 cm below the ground surface in M1W and M3E (Fig. 11F and G). The calcareous nature of that cement and of the gastropod shells, was confirmed by their effervescence upon addition of

the soil samples (Fig. 11C). Namely, samples with elevated mineral content showed elevated amounts of quartz, and vice versa. Similarly, freshwater-gastropod shells made of CaCO3 accounted for the TIC concentrations in the soil samples; i.e., samples with elevated TIC concentrations showed larger amounts of gastropod shells (Fig. 11D). Remains of gastropod shells were found in the mineral residues after 9

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 7. Significant linear relationships between concentrations of: a TP and b Cl− concentrations with TOC concentrations in the peat-center groundwater.

(Wetzel et al., 2011). Sullivan et al. (2014b) showed groundwater drawdown in the tree islands at LILA was greatest in the dry season when soil moisture and surface water levels were low. The drop in the groundwater table was hypothesized to draw surface water into the tree island from the surrounding wetland (Ross et al., 2006), and confirmed by previous hydrochemical tracing (Cl− and δ18O) studies at LILA (Sullivan et al., 2016). In this investigation, whether or not a seasonal drop in the groundwater table was observed seemed to be controlled by lithology. Seasonal groundwater movement transporting solutes into the peatbased tree islands was regulated by mixtures of peat with sand in varying proportions along the first meter of soil (Fig. 10). The different

HCl. Furthermore, the morphology of the CaCO3 cementing the quartz aggregates was characterized with a petrographic microscope in crosspolarized light. The cement was identified as isopachous micrite (Fig. 11H) uniformly distributed around the quartz grains, some of which showed organic-matter coatings. 4. Discussion 4.1. Influence of lithology on groundwater flow and solute transport The ability of trees to draw down the groundwater table via transpiration on a daily, monthly, and seasonal basis has been reported

Fig. 8. Comparisons of a Ca2+, b HCO3−, and c Na+ concentrations with Cl− concentrations in all water samples. 10

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

limestone-core tree islands M1E and M4E, particularly during the dry season (Fig. 5). Additionally, the drawdown of the water table resulted in increasing Cl− concentrations in the limestone-center groundwater observed throughout the study period (Fig. 6B), suggesting a long-term potential for those islands to concentrate solutes. In both types of tree islands at LILA, the presence of low K intervals towards the bottom of the first meter of soil (i.e., the mixture of limestone rubble with peat in M1E and M4E and the clay sediments present in M3E) may have facilitated the depression of the water table by ET, resulting in the subsequent solute accumulation in the groundwater beneath the center of those islands. Similar effects of impermeable layers on groundwater flow patterns were observed by McNeill and Cunningham (2003) in tree islands in WCA3, where horizons of limestone-rich mud situated below the surficial peat layers limited vertical surface water-groundwater exchange, contributing to a hydraulic gradient towards the center of those islands. These observations contrast the surface-water seepage observed across M1W and M2E at LILA, which is associated with elevated sand content in their shallow peat soils. 4.2. Water-rock interactions and calcite saturation Increasingly elevated Ca2+ concentrations in the groundwater beneath the center of most tree islands supported supersaturation with respect to calcite (Fig. 9). However, mass-balance results (Tables 3 and 4) suggested M3E as the only tree island where calcite could accumulate under current climatic and managed hydrologic conditions, below 1 m soil depth. The loss of groundwater through ET and CO2 outgassing to the atmosphere, coupled with increasing dissolved Ca2+ concentrations, were likely the cause of the supersaturated conditions with respect to calcite in M3E. The gaseous exchange in M3E may have been facilitated by an elevated K in the upper portion of the soil, associated with an elevated sand content (Fig. 10E and F). Similarly, the contribution of CO2 outgassing from groundwater to CaCO3 supersaturation (Price and Herman, 1991) was also observed in M1W, as the groundwater pCO2 values were consistently lower in the center well compared to those in the shallow well (Table 2). Furthermore, the elevated proportion of sand-to-peat within the soil of M2E (Fig. 10C and D) may have accounted for the calcite dissolution tendency (Table 3), by facilitating groundwater seepage across the island during the dry season (Fig. 4B) and groundwater discharge towards the slough during the wet season (Fig. 4A). Additionally, in the limestone-core islands M1E and M4E, surface water-groundwater interactions would prevent calcite accumulation under current climatic conditions, as suggested by mass-balance results in both seasons (Tables 3 and 4). Mass-balance results also indicated that whewellite (CaC2O4·H2O) could precipitate in the soils of M1E, M3E, and M4E during the wet season (Table 3), if the groundwater beneath those islands were saturated with respect to calcium oxalate (CaC2O4) because of fungal decomposition of leaf litter (Manning, 2000; Verrecchia et al., 2006; Gadd et al., 2014). The precipitation of whewellite, the most insoluble form of CaC2O4 (Ruiz-Agudo et al., 2013), did not affect calcite solubility in the mass-balance calculations. Inverse models without whewellite were also tested (data not shown) using the same thermodynamic constraints, and calcite dissolution also occurred. Furthermore, the dihydrate form of calcium oxalate, weddellite (CaC2O4·2H2O), has been found in the northern Everglades as a primary constituent in the inorganic fraction of peat soils (Griffin et al., 1984), and as an accessory mineral in clays and quartzose sands underlying peat deposits (Sawyer and Wieland, 1988; Harris, 2011). The accumulation of calcium oxalate minerals in the soils of the tree islands at LILA could be favored by the mildly acidic pH conditions in the groundwater (Ruiz-Agudo et al., 2013), which remained typically around 6.5. Because Cl− is a conservative anion, i.e., it tends to not be affected by adsorption and reaction processes (Bibby, 1981), the excess concentrations of Ca2+ and HCO3– with respect to Cl− concentrations (Fig. 8A and B) in all water samples indicated a tendency for carbonate

Fig. 9. Seasonal and locational variability of saturation indices with respect to calcite. a October (wet seasons) and b May (dry seasons). The circle and asterisk marks represent mild and extreme outliers, respectively. The dashed horizontal lines indicate equilibrium with respect to calcite.

peat-to-sand proportions are expected to control the hydraulic properties (e.g. hydraulic conductivity K and specific yield Sy) of the treeisland soils thus affecting hydrochemical transport and gaseous exchange with the atmosphere (Boelter, 1968, 1969; Rezanezhad et al., 2016). In M2E for example, the water table remained almost horizontal across the tree island and the macrocosm (Fig. 4A and B) suggesting enhanced hydraulic connectivity with the adjacent slough, most likely because of the elevated sand content within the soil of that peat-based tree island (Fig. 10C and D). Similarly, the elevated sand content within the peat soil of M1W (Fig. 10A and B) may have facilitated surface water seepage into the island, accounting for the increasing δ18O enrichment and decreasing Cl− concentrations in the center groundwater which approached those in the edge groundwater by the end of the study (Table 2). In contrast, the decreasing sand content towards the bottom of the first meter of soil in M3E (Fig. 10E and F), parallel with an increase in clay content (Sullivan et al., 2011), may have contributed to the persistent water table depression in that peat-based tree island (Fig. 4C and D) because of suspected lower K values versus ET rates (Sullivan et al., 2014a). Elevated seasonal ET rates coupled with the lowest K values amongst the LILA tree islands (Sullivan et al., 2014a) may also explain the observed drawdown of the water table in the center of the 11

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 10. Vertical profiles showing concentrations of total carbon (TC), total inorganic carbon (TIC), and mineral residue in the soils of three peat-based tree islands. a M1WA and b M1WB are two soil cores collected from the M1W tree island. c M2EA and d M2 EB are two soil cores collected from the M2E tree island. e M3EA and f M3 EB are two soil cores collected from the M3E tree island. Concentrations of TC/TIC are expressed in mg/g and mineral residue is expressed in percent of dry weight (dw%) soil. The blue horizontal lines represent the water table below the center of the islands in May 2015 (refer to Fig. 7B and D), when the samples were collected. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

samples, suggested that cation exchange potentially increased dissolved Ca2+ concentrations in the groundwater beneath the tree islands at LILA. Mass-balance results (Tables 3 and 4) indicated that Ca2+/Na+ and 2+ Ca /Mg2+ exchange influenced the geochemical evolution of groundwater flowing towards the center of the limestone-core tree islands, and towards the center of the peat-based tree islands in both seasons, respectively. However, Mg2+ depletion can be explained by adsorption (MgX2 precipitation) as well as Mg2+ uptake by plants (Landmeyer, 2012). Furthermore, Ca2+/Na+ exchange in M2E during

mineral, most likely calcite, dissolution within the LILA groundwater system. However, the dissolution of carbonate minerals may not have been the only source of dissolved Ca2+ and Mg2+ concentrations in the groundwater at LILA. Because the precipitation of sodium minerals is unlikely in this freshwater environment (Reading, 2009), and Na+ uptake by trees is insignificant (Landmeyer, 2012), cation exchange processes were inferred by examining the ratios of Na+ and Mg2+ to Cl− concentrations in the groundwater. The observed depletion of Na+ (Fig. 8C) and Mg2+ with respect to Cl− concentrations in the shallowpeat, peat-center, peat-deep, and limestone-center groundwater 12

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Fig. 11. Photomicrographs of soil samples: a Loxahatchee peat; b hardwood-vegetation roots; c residual quartz grains after calcination; d freshwater-gastropod shells; e remnants of gastropod shells after calcination; f and g aggregates of quartz grains cemented with CaCO3; h quartz grains embedded in isopachous-micritic cement with organic-matter coatings (cross polarized light). The metal bar used for scale in a through f is 1 mm wide, and the black lines used for scale in g are 1 mm apart.

4.3. Shallow soil composition

the wet season (Table 3) and in M3E during the dry season (Table 4), suggests that the cation exchange capacity (CEC) of the organic soils (Reddy and DeLaune, 2008) at LILA promoted cation exchange processes. Additionally, the occurrence of mixed-layer smectite along the flanks and at the southern boundary of the Loxahatchee basin (Sawyer and Wieland, 1988), may have also contributed to cation exchange processes in the soils at LILA, as clayey sediments were observed by Sullivan et al. (2011) in the soils of macrocosm M3.

The examination of the first meter of soil from the ground surface in three peat-based tree islands suggests that little to no calcite has accumulated in the capillary fringe above the water table in response to seasonal ET. Microscopic observations (Fig. 11) indicate that fine-sized quartz grains account for the mineral content of the soils (Fig. 10). Thus, peaks of elevated TC/TIC concentrations associated with lower residual mineral content, correspond to peat sections with lower quartz content. Moreover, freshwater-gastropod shells (Fig. 11D and E), which are characteristic in calcitic muds and peat deposits in the northern 13

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

1974; Sawyer and Wieland, 1988). Thus, the construction process of the peat-based tree islands at LILA (Sklar et al., 2004) may have resulted in a partially reverse hydrostratigraphy in their shallow substrate, as compared to natural tree islands in the Everglades (McNeill and Cunningham, 2003).

Table 2 Hydrochemical parameters including partial pressure of carbon dioxide (pCO2) in the center, edge, and shallow wells of tree island M1W obatined from all groundwater sampling events. Date/Wellsite

Cl− (mg/L)

δ18O (‰ VSMOW)

pCO2

Oct 2013 Edge Center Shallow

116.2 193.6 135.1

0.2 1.2 −5.3

0.22 0.28 0.24

May 2014 Edge Center

83.9 176.2

5.3 1.2

0.15 0.18

Oct 2014 Edge Center Shallow

144.0 170.6 249.4

0.2 2.6 −2.6

0.18 0.16 0.14

May 2015 Edge Center Shallow

81.8 150.0 173.8

6.6 3.4 1.0

0.08 0.19 0.03

Oct 2015 Edge Center Shallow

118.3 121.5 400.9

3.6 4.0 −5.9

0.16 0.18 0.17

4.4. Implications of mineral solubility on soil biogeochemical processes Although groundwater chemistry analyses did not include testing for specific organic compounds, such as concentrations of methane, calcium oxalates, and other dissolved organic constituents, these were considered in mass-balance calculations corresponding to the wet season in October 2014 (Table 3). Dissolved organic matter (CH2O) and CH4 were necessarily considered as reacting phases to attain masstransfer results consistent with CaCO3 SIs and pCO2 values (Table 1). For example, groundwater discharge from the center of M2E towards the slough in October 2014 included CH4 emissions, according to geochemical simulations. This result was consistent with elevated CH4 production within the first 40 cm below the peat soil surface of M1W and M4W, detected by Schonhoff (2015), when prolonged flooding of the soils during the wet season in 2014 induced reducing conditions. Furthermore, reducing conditions associated with the decomposition of organic matter and CH4 production, could also inhibit calcite precipitation in the soils through the introduction of humic substances (Hoch et al., 2000). Such substances could block potential sites for crystal nucleation and enhance calcite solubility (Amrhein and Suarez, 1987; Meyer et al., 2014), resulting in the CaCO3 supersaturation observed in the groundwater beneath some of the tree islands. Groundwater uptake of dissolved organic matter during high surface-water stages in the eastern tree islands at LILA was estimated through geochemical modeling (Table 3), but future calculations of mineral saturation states should include measurements of specific organic compounds in the groundwater at LILA. Also, measuring the δ13C and 14C isotopic composition of dissolved inorganic carbon (DIC), carbonate minerals in the soils, and CO2 gas from the unsaturated zone could provide a more robust geochemical/isotopic dataset to better understand the geochemical evolution of the groundwater system at LILA (Han et al., 2014; Meredith et al., 2016). The anaerobic decomposition of organic matter under flooded conditions also promotes soil P release from restoration wetland soils (Aldous et al., 2005). In the Everglades tree islands, the decomposition of leaf litter results in higher soil P concentrations (Troxler Gann and Childers, 2006; Rodriguez et al., 2014). Thus, the significant and positive correlations between TOC and TP concentrations observed in the peat-center groundwater (Fig. 7A), which were higher during the wet seasons (Fig. 7D), may have resulted from the breakdown of organic matter. Moreover, the availability of P for plant uptake during highsurface water stages at LILA may be favored by a preferential capture of dissolved Ca2+ into calcium oxalate minerals (Ruiz-Agudo et al., 2013); which would then prevent the precipitation of calcium phosphate minerals within the tree-island soils. Flooding cycles of the tree-island

Everglades (Gleason et al., 1974; Sawyer and Wieland, 1988), accounted for the majority of the TIC concentrations in the soil samples. Although, aggregates of quartz grains cemented with calcium carbonate were observed (Fig. 11G and H), their presence were few and randomly distributed within the first 24 cm of soil in islands M1W and M3E, contributing only a small fraction of the TIC concentrations. The petrographic identification under crossed polarized light of the CaCO3 cementing the quartz aggregates indicated isopachous micrite, uniformly distributed around a few of the quartz grains (Fig. 11H), typical of a meteoric cement that originated in the phreatic zone. The morphology of this micritic cement (Fig. 11G and H) indicated a complete saturation of the porous media with water during meteoric diagenesis; while an ET-driven mechanism would have generated a meniscus cement around the quartz grain contacts, as a result of capillary rise (Scholle and Ulmer-Scholle, 2003). These findings suggest that the cement on the quartz aggregates was most likely not formed in the upper 24 cm of the tree island soils, but may be debris derived from the construction process of LILA, which involved sculpting the topography from existing impoundments (Sklar et al., 2004). This construction process would also account for the higher sand content observed in the upper sections of the tree-island soils, which was introduced in the peat soils from an underlying sediment layer which consists of quartzose sand, shell fragments, and clay (Sullivan et al., 2011). A sedimentary unit with those characteristics, the Pamlico formation, borders the northern Everglades and underlies peat deposits, including tree-island soils in the Loxahatchee basin (Gleason et al.,

Table 3 Mass-balance results along groundwater flow paths in October 2014. Positive (mass entering water) and negative (mass leaving water) phase mole transfers indicate dissolution and precipitation, respectively. Phase transfers (mol/Kg) along groundwater flow paths in different tree islands Mineral phases Calcite CaX2 CH2O CH4(g) CO2(g) H2O(g) MgX2 NaX CaC2O4·H2O

M1E edge → center 4.721e − 02 3.616e − 03 3.179e − 02

M2E center → edge 1.014e − 03 1.456e − 04 4.634e − 03 −2.317e − 03

M3E edge → center 6.196e − 02 9.390e − 04 3.983e − 02

−2.220e + 02 −9.196e − 04 −5.392e − 03 −6.358e − 02

3.877e + 01

−2.979e + 02 −9.390e − 04

−2.913e − 04 −7.966e − 02

14

M3E deep → center −2.146e − 03 5.080e − 05

2.924e − 03 −2.119e + 01 −5.080e − 05

M4E edge → center 1.412e − 02 4.348e − 04 9.267e − 03

−1.077e + 02 −8.696e − 04 −1.853e − 02

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al.

Table 4 Mass-balance results along groundwater flow paths in May 2015. Positive (mass entering water) and negative (mass leaving water) phase mole transfers indicate dissolution and precipitation, respectively. Phase transfers (mol/Kg) along groundwater flow paths in different tree islands Mineral phases Calcite CaX2 CO2(g) H2O(g) MgX2 NaX

M1E edge → center −1.336e − 03 1.905e − 04 −8.512e − 03 −1.517e + 01

M3E edge → center −2.756e − 03 3.774e − 04 −1.715e − 02 −1.299e + 02 −3.774e − 04

M3E deep → center −2.543e − 03 2.241e − 04 −2.206e − 03 −1.063e − 04 −2.355e − 04

−3.811e − 04

M3E slough → center −4.195e − 02 6.323e − 03 −4.525e − 02 −5.136e + 02 −3.512e − 03 −5.621e − 03

M4E edge → center 1.239e − 03 3.586e − 04 −5.068e − 03 −4.215e + 01 −7.171e − 04

LILA groundwater system. Hydrochemical analyses indicated that the breakdown of organic matter regulated phosphorus concentrations in the groundwater below the peat-based islands. Biogeochemical P sequestration in the subsurface of the LILA tree islands is not currently governed by carbonate geochemistry in the groundwater, as suggested by hydrochemical trends and geochemical modeling. Calcite dissolution estimated by mass-balance calculations, occurred regardless of the possibility for calcium oxalate precipitation during the wet season, and P concentrations were too low for hydroxyapatite precipitation. Thus, P mineralization and adsorption onto calcite would take longer to occur and could be inhibited by calcium oxalate formation. An additional step for predicting the evolution of biogeochemical cycling and water-rock interactions in the LILA groundwater system could include measuring the carbon isotopic composition in DIC, vadose CO2 gas, and soil CaCO3. Analyzing the interactions of these carbon-bearing sources would support further geochemical modeling efforts fitting with hydrochemical data. Predicting the evolution of the LILA groundwater geochemistry remains an essential task to restore healthy tree islands in the Everglades under varying hydrologic regimes. The results of this investigation can be applied to wetland tree islands around the world such as in the Pantanal of South America, the Great Vasyugan Bog of Siberia, and Okavango Delta in South Africa, and the Sian Ka'an karstic wetland in the Yucatan, Mexico.

soils promoting reducing conditions (Schonhoff, 2015) would also favor the preservation of CaC2O4 minerals over CaCO3 minerals (Griffin et al., 1984; Manning, 2000). Furthermore, SIs with respect to CaCO3 did not correlate with TP nor with PO43− concentrations, suggesting that the geochemistry of CaCO3 did not regulate P dynamics in the LILA hydrogeochemical system. 5. Conclusions Identifying petrologic characteristics of the soil in three peat-based LILA tree islands provided insights fundamental to understanding the causes for different groundwater flow patterns in response to ET, and the water-rock interactions affecting potential CaCO3 precipitation in tree islands. Peat-based tree islands with the highest sand content along soil vertical profiles promoted surface water seepage and calcite dissolution, particularly during the wet season. Islands with a lithology expected to have a decreasing K (e.g. limestone rubble mixed with peat, or peat with lower sand content and some clay) with depth allowed for a drop in the groundwater table with ET and an increase in Cl− concentrations. Specifically, surficial, sandy peats allowed for continuous ET-driven groundwater flow and solute transport towards the center of the islands. A clay layer underlying the sandy peat reduced surface water seepage across one of the islands, supporting greater solute concentrations in the groundwater driven by ET processes. Although natural tree islands exist on different geologic substrates across the Everglades and globally, our results show that the lithologic characteristics of the first meter of soil is critical in the regulation of geochemical and hydrologic processes driven by seasonal ET rates. These controlling factors have been identified in other freshwater wetlands around the world, but the geological characteristics of the deep treeisland substrates in the Everglades have yet to be studied in relation to groundwater geochemistry. Deeper soil coring at LILA is recommended as a next step to achieve a holistic understanding of tree-island ecohydrological functioning across the Everglades. The elevated sand content in the peat soils of the LILA tree islands supported a simplification of anisotropic conditions along the simulated groundwater flow paths. In turn, these average flow lines provided a seasonal hydrologic frame to quantify mass transfers influencing CaCO3 mineralization. Geochemical modeling using speciation-solubility parameters (CaCO3 SIs and pCO2) to constrain mass-balance calculations proved a useful tool for testing reactions hypotheses to explain the observed groundwater chemistry. Elevated Ca2+ concentrations in the groundwater beneath most tree islands supported calcite supersaturation, but mass-balance results along groundwater flow paths indicated that long-term calcite accumulation was only possible in one tree island with a high clay content. Excess concentrations of Ca2+ and HCO3– with respect to Cl− concentrations in all the water samples indicated an overall tendency for the dissolution of carbonate minerals. Despite the insights on calcite dynamics in the groundwater at LILA provided by mass-balance and thermodynamic calculations, supported by analysis of hydrochemical trends, the complex effects of dissolved organic compounds on calcite kinetics need to also be considered in future studies to achieve a better geochemical conceptualization of the

Acknowledgments This project was funded by the South Florida Water Management District Contract No. 4600002848. The Arthur R. Marshall Loxahatchee National Wildlife Refuge and the U.S. Fish and Wildlife Service contributed to the creation of LILA. Field work was completed with the help of: Edward Linden, Joshua Allen, Dillon Reio, Kalli Unthank, Shimelis Dessu, and Paul Kuhn. This research was developed in collaboration with the Florida Coastal Everglades Long-Term Ecological Research (LTER) program under National Science Foundation Grant No. DEB-1237517 and the Centers for Research Excellence in Science and Technology (CREST) Program under Grant No. HRD-1547798 awarded to FIU. Additional support was provided by the Southeast Environmental Research Center (SERC) at FIU, and this manuscript is SERC contribution number 861. References Aich, S., Dreschel, T., Cline, E., Sklar, F., 2011. The development of a Geographic Information System (GIS) to document research in an everglades physical model. J. Environ. Sci. Eng. 5, 289–302. Aldous, A., McCormick, P., Ferguson, C., Graham, S., Chris, Craft, 2005. Hydrologic regime controls soil phosphorus fluxes in restoration and undisturbed wetlands. Restor. Ecol. 13 (2), 341–347. Ali, A., Abtew, W., Van Horn, S., Khanal, N., 2000. Temporal and spatial characterization of rainfall over Central and South Florida 1. J. Am. Water Resour. Assoc. 36 (4), 833–848. Alonso-Zarza, A., Wright, V., Alonso-Zarza, A., Tanner, L., 2010. Calcretes. In: Developments in Sedimentology. Vol. 61. pp. 225–267. Amrhein, C., Suarez, D., 1987. Calcite supersaturation in soils as a result of organic matter mineralization. Soil Sci. Soc. Am. J. 51 (4), 932–937. Bauer-Gottwein, P., Langer, T., Prommer, H., Wolski, P., Kinzelbach, W., 2007. Okavango

15

Chemical Geology xxx (xxxx) xxx–xxx

A.E. Prieto Estrada et al. Delta Islands: interaction between density-driven flow and geochemical reactions under evapo-concentration. J. Hydrol. 335 (3), 389–405. Bibby, R., 1981. Mass transport of solutes in dual-porosity media. Water Resour. Res. 17 (4), 1075–1081. Boelter, D., 1968. Important physical properties of peat materials. In: Proceedings, Third International Peat Congress, pp. 150–154. Boelter, D., 1969. Physical properties of peats as related to degree of decomposition. Soil Sci. Soc. Am. J. 33 (4), 606–609. Coultas, C.L., Schwadron, M., Galbraith, J.M., 2008. Petrocalcic horizon formation and prehistoric people's effect on everglades tree island soils, Florida. Soil Horizons 49 (1), 16–21. Craighead, F., 1974. Hammocks of South Florida. In: Gleason, P.J. (Ed.), Environments of South Florida: Present and Past. Miami Geological Society, Naples, pp. 53–60. Davis, S.M., Gunderson, L.H., Park, W.A., Richardson, J.R., Mattson, J.E., 1994. Landscape Dimension, Composition, and Function in a Changing Everglades Ecosystem. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosystem and its Restoration. St. Lucie Press, Boca Raton, pp. 419–444. Gadd, G.M., Bahri-Esfahani, J., Li, Q., Rhee, Y.J., Wei, Z., Fomina, M., Liang, X., 2014. Oxalate production by fungi: significance in geomycology, biodeterioration and bioremediation. Fungal Biol. Rev. 28, 36–55. Gleason, P., Stone, P., 1994. Age, origin, and landscape evolution of the everglades peatland. In: Ogden, J.C., Davis, S.M. (Eds.), Everglades: The Ecosystem And Its Restoration. St. Lucie Press, Boca Raton, Florida, USA, pp. 149–198. Gleason, P.J., Cohen, A.D., Smith, W.G., Brooks, K.H., Stone, P.A., Goodrick, R.L., Spackman Jr., W., 1974. The environmental significance of Holocene sediments from the Everglades and saline tidal plain. In: Gleason, P.J. (Ed.), Environments of South Florida: Present and Past. Miami Geological Society, West Palm Beach, pp. 287–341. Graf, M., Schwadron, M., Stone, P., Ross, M., Chmura, G., 2008. An enigmatic carbonate layer in everglades tree island peats. EOS Trans. Am. Geophys. Union 89 (12), 117–118. Griffin, G.M., Sawyer, R.K., Melkote, S.R., 1984. Weddellite occurrence in peats and other organic-rich sediments of Florida. J. Sediment. Res. 54 (3), 861–868. Han, L.-F., Plummer, L.N., Aggarwal, P., 2014. The curved 14C vs. δ13C relationship in dissolved inorganic carbon: a useful tool for groundwater age- and geochemical interpretations. Chem. Geol. 387, 111–125. Harris, W., 2011. Mineral distribution and weathering in the greater Everglades: implications for restoration. Crit. Rev. Environ. Sci. Technol. 41 (S1), 4–27. Hoch, A.R., Reddy, M.M., Aiken, G.R., 2000. Calcite crystal growth inhibition by humic substances with emphasis on hydrophobic acids from the Florida Everglades. Geochim. Cosmochim. Acta 64 (1), 61–72. Humphries, M., Kindness, A., Ellery, W., Hughes, J., 2011. Water chemistry and effect of evapotranspiration on chemical sedimentation on the Mkuze River floodplain, South Africa. J. Arid Environ. 75, 555–565. Landmeyer, J.E., 2012. Fundamentals of plant anatomy and physiology related to water use. In: Landmeyer, J.E. (Ed.), Introduction to Phytoremediation of Contaminated Groundwater. Historical Foundation, Hydrologic Control, and Contaminant Remediation. Springer, Netherlands, pp. 43–93. Langmuir, D., 1997. Aqueous Environmental Geochemistry. Prentice Hall, Upper Saddle River. Larsen, L.G., Harvey, J.W., 2010. How vegetation and sediment transport feedbacks drive landscape change in the everglades and wetlands worldwide. Am. Nat. 173 (3), E66–E79. Light, S.S., Dineen, J.W., 1994. Water control in the everglades: a historical perspective. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosystem and its Restoration. St. Lucie Press, Boca Raton, pp. 47–84. Manning, D.A., 2000. Carbonates and oxalates in sediments and landfill: monitors of death and decay in natural and artificial systems. J. Geol. Soc. 157 (1), 229–238. McCarthy, T., Humphries, M., Mahomed, I., Le Roux, P., Verhagen, B., 2012. Island forming processes in the Okavango Delta, Botswana. Geomorphology 179, 249–257. McNeill, D.F., Cunningham, K.J., 2003. Hydrostratigraphy of tree island cores from Water Conservation Area 3. In: U.S Geological Survey. U.S. Department of the Interior, Miami Retrieved from. http://sofia.usgs.gov/publications/ofr/03-68/. Meredith, K., Han, L., Hollins, S., Cendón, D., Jacobsen, G., Baker, A., 2016. Evolution of chemical and isotopic composition of inorganic carbon in a complex semi-arid zone environment: consequences for groundwater dating using radiocarbon. Geochim. Cosmochim. Acta 188, 352–367. Meyer, N.A., Breecker, D.O., Young, M.H., Litvak, M.E., 2014. Simulating the effect of vegetation in formation of pedogenic carbonate. Soil Sci. Soc. Am. J. 78 (3), 914–924. Maltby, E., Dugan, P., 1994. Wetland ecosystem protection, management, and restoration: an international perspective. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: The Ecosystem and its Restoration. St. Lucie Press, Boca Raton, Florida, USA, pp. 29–46. Nordstrom, D.K., Campbell, K., 2014. Modeling low-temperature geochemical processes. In: Holland, H., Turekian, K. (Eds.), Treatise on Geochemistry, Second Edition. vol. 7. Elsevier Ltd, Boulder, Colorado, pp. 27–68. Nelson, D.W., Sommers, L.E., 1996. Total carbon, organic carbon, and organic matter. In: Sparks, D. (Ed.), SSSA Book Series: 5. Methods of soil analysis. Part 3 - Chemical Methods. Soil Society of America, Inc, Madison, WI, pp. 961–1010. Parkhurst, D.L., Appelo, C., 2013. Description of input and examples for PHREEQC version 3—a computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. In: Geological Survey, U.S. (Ed.), Techniques and Methods, Book 6. US Geological Survey, Denver, Colorado, pp. 43–497. Parkhurst, D.L., Scott, C., Breit, G., 1993. Ground-water-quality assessment of the central Oklahoma Aquifer, Oklahoma; geochemical and geohydrologic investigations. In: U.S. Geological Survey Water-supply Paper 2357. U.S. Geological Survey, Denver, CO. Parkhurtst, D.L., 1997. Geochemical mole-balance modeling with uncertain data. Water Resour. Res. 33 (8), 1957–1970.

Patterson, K., Finck, R., 1999. Tree Islands of the WCA 3A Aerial photointerpretation and trend analysis project summary report. In: South Florida Water Management District. Geonex Corporation, West palm Beach. Plummer, L., Parkhurst, D.L., Thorstenson, D.C., 1983. Development of reaction models for ground-water systems. Geochim. Cosmochim. Acta 47, 665–686. Price, R.M., Herman, J.S., 1991. Geochemical investigation of salt-water intrusion into a coastal carbonate aquifer: Mallorca, Spain. Geol. Soc. Am. Bull. 103 (10), 1270–1279. http://dx.doi.org/10.1130/0016-7606(1991)103<1270:GIOSWI>2.3.CO;2. Ramberg, L., Wolski, P., 2008. Growing islands and sinking solutes: processes maintaining the endorheic Okavango Delta as a freshwater system. Plant Ecol. 196 (2), 215–231. Reading, H.G., 2009. Sedimentary Environments: Processes, Facies and Stratigraphy. Blackwell Science Ltd. Reddy, K., DeLaune, R., 2008. Biogeochemistry of Wetlands: Science and Applications. CRC Press. Rezanezhad, F., Price, J.S., Quinton, W.L., Lennartz, B., Molojevic, T., Van Cappellen, P., 2016. Structure of peat soils and implications for water storage, flow and solute transport: a review update for geochemists. Chem. Geol. 429, 75–84. Rodriguez, A.F., Serna, A., Scinto, L.J., 2014. Soil accretion influenced by elevation, tree density, and substrate on reconstructed tree islands. Soil Sci. Soc. Am. J. 78 (6), 2090–2099. Ross, M.S., Mitchell-Bruker, S., Sah, J.P., Stothoff, S., Ruiz, P.L., Reed, D.L., Jayachandran, K., Coultas, C.L., 2006. Interaction of hydrology and nutrient limitation in the Ridge and Slough landscape of the southern Everglades. Hydrobiologia 569 (1), 37–59. Ruiz-Agudo, E., Álvarez-Lloret, P., Putnis, C., Rodriguez-Navarro, A., Putnis, A., 2013. Influence of chemical and stuctural factors on the calcite-calcium oxalate transformation. CrystEngComm 9968–9979. Sawyer, R.K., Wieland, C.C., 1988. The clay mineralogy of calcitic seat earth in the northern Everglades of Florida. J. Sediment. Petrol. 58 (1), 81–88. Scholle, P.A., Ulmer-Scholle, D.S., 2003. A Color Guide to the Petrography of Carbonate Rocks: Grains, Textures, Porosity, Diagenesis. vol. 77 The American Association of Petroleum Geologists, Tulsa, Oklahoma. Schonhoff, B.R., 2015. Gaseous carbon emissions (methane and carbon dioxide) from wetland soils in a re-created everglades landscape. In: FIU Electronic Theses and Dissertations. Florida International University, Earth & Environment, Miami. Schwadron, M., 2006. Everglades tree islands prehistory: archaeological evidence for regional holocene variability and early human settlement. Antiquity 80 (310), 1–6. Sklar, F., van der Valk, A., 2002. Tree Islands of the Everglades: an overview. In: Sklar, F., van der Valk, A. (Eds.), Tree Islands of the Everglades. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 1–18. Sklar, F., Coronado-Molina, C., Gras, A., Rutchey, K., Gawlik, D., Grozier, G., ... McVoy, C., 2004. The Loxahatchee Impoundment Landscape Assessment (LILA). In: Chapter 6: Ecological Effects of Hydrology. Everglades Consolidated Report, South Florida Water Management District, West Palm Beach, FL, pp. 2004. Stofella, S.L., Ross, M.S., Sah, J.P., Price, R.M., Sullivan, P.L., Cline, E.A., Scinto, L.J., 2010. Survival and growth responses of eight Everglades tree species along an experimental hydrological gradient on two tree island types. Appl. Veg. Sci. 13 (4), 439–449. Sullivan, P.L., 2011. Groundwater-surface water interactions on tree islands in the Everglades, South Florida. In: FIU Electronic Theses and Dissertations. Florida International University, Earth & Environment, Miami. Sullivan, P.L., Price, R.M., Ross, M.S., Scinto, L.J., Stofella, S.L., Cline, E.A., Sklar, F.H., 2011. Hydrologic processes on tree islands in the Everglades (Florida, USA): tracking the effects of tree establishment and gorwth. Hydrogeol. J. 19 (2), 367–378. Sullivan, P.L., Price, R.M., Miralles-Wilhelm, F., Ross, M.S., Scinto, L.J., Dreschel, T.W., Cline, E., 2014a. The role of recharge and evapotranspiration as hydraulic drivers of ion concentrations in shallow groundwater on Everglades tree islands, Florida (USA). Hydrol. Process. 28 (2), 293–304. Sullivan, P.L., Engel, V., Ross, M.S., Price, R.M., 2014b. The influence of vegetation on the hydrodynamics and geomorphology of a tree island in Everglades National Park (Florida, United States). Ecohydrology 7 (2), 727–744. Sullivan, P.L., Price, R.M., Ross, M.S., Stofella, S.L., Sah, J.P., Scinto, L.J., Sklar, F.H., 2016. Trees: a powerful geomorphic agent governing the landscape evolution of a subtropical wetland. Biogeochemistry 1–16. Troxler Gann, T., Childers, D.L., 2006. Relationships between hydrology and soils describe vegetation patterns in seasonally flooded tree islands of the southern Everglades, Florida. Plant Soil 279, 271–286. Van Der Valk, A.G., Wetzel, P., Cline, E., Sklar, F.H., 2008. Restoring tree islands in the Everglades: experimental studies of tree seedling survival and growth. Restor. Ecol. 16 (2), 281–289. http://dx.doi.org/10.1111/j.1526-100X.2007.00311.x. Verrecchia, E.P., Braissant, O., Cailleau, G., 2006. The oxalate-carbonate pathway in soil carbon storage: the role of fungi and oxalotrophic bacteria. In: Fungi in Biogeochemical Cycles, pp. 289–310. Wang, H.F., Anderson, M.P., 1982. Introduction to Groundwater Modeling: Finite Difference and Finite Element Methods. Academic Press. Wetzel, P.R., van der Valk, A.G., Newman, S., Gawlik, D.E., Troxler, T., Coronado-Molina, C.A., Sklar, F.H., 2005. Maintaining tree islands in the Florida Everglades: nutrient redistribution is key. Front. Ecol. Environ. 3 (7), 370–376. Wetzel, P.R., van der Valk, A.G., Newman, S., Coronado, C.A., Troxler-Gann, T.G., Childers, D.L., Sklar, F.H., 2009. Heterogeneity of phosphorus distribution in a patterned landscape, the Florida Everglades. Plant Ecol. 200 (1), 83–90. Wetzel, P.R., Sklar, F.H., Coronado, C.A., Troxler, T.G., Krupa, S.L., Sullivan, P.L., ... Orem, W.H., 2011. Biogeochemical processes on tree islands in the Greater Everglades: initiating a new paradigm. Crit. Rev. Environ. Sci. Technol. 41 (S1), 670–701. Wright, V.P., 2007. Calcrete. In: Nash, D.J., McLaren, S.J. (Eds.), Geochemical Sediments & Landscapes. Blackwell Publishing Ltd., Malden, pp. 10–45.

16