Shallow groundwater chemical evolution, isotopic hyperfiltration, and salt pan formation in a hypersaline endorheic basin: Pilot Valley, Great Basin, USA

Endorheic basin brines are of economic significance as sources of boron, iodine, magnesium, potassium, sodium sulfate, sodium carbonate, and tungsten, and they are a major source of the critical metal lithium. Although evaporation is the primary hypersalinization driver for evaporative water bodies, recent investigations have proposed more novel mechanisms for some subsurface brine. This investigation explores shallow groundwater hypersalinization. The chemical evolution and isotopic fractionation of shallow hypersaline groundwater in the clay-rich arid endorheic basin sediments of Pilot Valley, Great Basin (USA), were investigated. Groundwater evolves from fresh in the mountain bedrock and alluvial fans, to brackish and saline at the alluvial fan–playa interface, and to hypersaline in the upper 12 m of basin sediments. Alluvial fan systems are isolated from each other and have varying groundwater 3H and 14C travel times. Nonevaporative in-situ isotopic fractionation of up to −8‰ in δ18O is attributed to clay sequence hyperfiltration. Groundwater flow-path sulfate and chloride mineral dissolution is the primary driving mechanism for both interface and basin groundwater evolution. Evaporation only impacts the groundwater quality in a small portion of the basin where the groundwater is within ~1 m of the ground surface. Here capillary action carries dissolved soluble salts to the land surface. Episodic flooding redissolves and carries the precipitated salt to the annually flooded salt pan where it accumulates as a salt crust during the dry season. The Pilot Valley model may help explain the buildup accumulative layers of soluble salt that when remobilized becomes subsurface brine.


Introduction
Approximately 20% of the Earth's land surface is covered by endorheic basins, but the basins only account for 2.3% of the total worldwide annual river runoff (UNESCO 2014). The basins typically occur in arid and semiarid regions where the worldwide average precipitation is 54.1 mm year −1 compared to the average for exoreic regions of 321.5 mm year −1 (Meybeck et al. 2001). Although endorheic basins are ubiquitous, the evolution of their groundwater geochemistry, particularly high saline waters, has not received the same level of attention as has other aspects of hydrogeology. Coastal and hydrogeological processes that involve anthropogenic and paleo-groundwater influences have been found to be a factor in the evolution of some high salinity groundwaters (Lowenstein and Risacher 2009;Post et al. 2014;Sabarathinam et al. 2018;Van Engelen et al. 2017). Salinization may also be the result of diffuse recharge of residual water associated with transpiration and evaporation (Cartwright et al. 2007), the dissolution and mobilization of evaporite soil-zone and aquifer minerals (Massuel et al. 2006;Mayo et al. 2010;Scanlon et al. 2009), biogenic processes (Mayo et al. 2007), and discharge zone evaporation (Shuya et al. 2018;Tweed et al. 2011). In recent years portions of the Great Australian Basin have been reevaluated by geochemical means Priestley et al. 2017: Shand et al. 2013 as has the geochemical evolution of a few endorheic basins in China, Central and South America, and Turkey (Bayari et al. 2008;Edmonds et al. 2006;He et al. 2015;Herrera et al. 2016;Johnson et al. 2010;Ortega 2003;Xiao et al. 2017Xiao et al. , 2018. Endorheic basin brines are of great economic significance as sources of boron, iodine, lithium, magnesium, potassium, sodium sulfate, sodium carbonate, and tungsten (Hardie 1991;Eugster 1980;Riascher and Fritz 2009;Rosen 1994)-for example, one-third of the world's lithium is from a single source, the Salar de Atacama in Chile, and the Salar de Uyuni in Boliva may contain the world largest lithium reserves. Traditionally mineral leaching coupled with evaporation has been considered to be the primary driver of basin hypersalinization; however, recent investigations have cast a wider net. Duffy and Al-Hassen (1998) and Hamann et al. (2015) have modeled coupled evaporation-convective groundwater flow as a salinization mechanism. Marazuela et al. (2020) found that thermohaline flow associated with fault damage zones promote the remobilization of lithium in the Salar de Atacama. Rissman et al. (2015) proposed that brine leakage from active salars (salt flats) coupled with groundwater flow in the high Andes of the Atacama Desert concentrates brines in the Salar de Atacama. Of interest here is the development of a better understanding of the hypersalinization processes in shallow, active zone endorheic basin groundwater. Pilot Valley, located in the Western United States Great Basin (Fig. 1), was selected for study because it is a readily accessible, has not been significantly impacted by anthropogenic processes, is partially covered by a salt pan, is of a manageable size for in-depth research, and has shallow groundwater chemistry that evolves from fresh to hypersaline along sort flow paths. Pilot Valley was a western embayment of the vast Pleistocene age pluvial Lake Bonneville.

Previous Pilot Valley investigations
Early Pilot Valley investigations focused on Lake Bonneville history and the origin of the salt crust (Gilbert 1890;Nolan 1928;Stansbury 1852). The near surface geology of the valley has been well documented (Anderson 1957(Anderson , 1960Miller Fig. 1 Google Earth image of Pilot Valley and the bounding Pilot and Silver Island Ranges. In this article, the Playa outline is used in figures without location attribution 1984, 1985, 1990a, b, 1993Miller and Lush 1994;Miller and Phelps 2016;Miller et al. 1990Miller et al. , 1993O'Neil 1968;Schaeffer and Anderson 1960;South et al. 2016). Harrill (1971) and Stephens and Hood (1973) conducted hydrologic reconnaissance investigations. Lines (1979) described the zonation of evaporate minerals and provided an overview of the aquifers. Peterson (1993) undertook the first examination of the shallow groundwater geochemistry. Mason et al. (1995) and Liutkus and Wright (2008) reported some chemical and isotopic data. Carling et al. (2012) investigated Pilot Valley mountain front groundwater recharge and Rey et al. (2016) characterize the shallow basin stratigraphy. The basin margin fresh/salinewater interface has been investigated by Duffy and Al-Hassan (1988), Oliver (1992), Fan et al. (1997), Winkel (2007), and Mayo et al. (2020).

Methods of investigation
Field and laboratory investigations included: (1) construction of shallow monitoring wells, (2) water level measurements, (3) mineralogical analysis of lake bed cores, (4) collection and laboratory analysis of stream, spring, and well waters for solutes and isotopes, and (5) geochemical modeling of the fresh to hypersaline groundwater evolution. Solute and isotopic samples were collected from 247 shallow monitoring wells, 16 boreholes, 15 springs, 10 ephemeral streams, and snow. As part of this study, 110 wells and the boreholes were constructed, and the remaining 137 wells were previously constructed (Carling et al. 2012;Mason et al. 1995;Oliver 1992;Peterson 1993). Well depths range from <1 to 30 m below ground surface (bgs). Most wells were completed within 6 m of the ground surface in hand augured boreholes using 2.54 cm inside diameter (ID) polyvinyl chloride (pvc) pipe fitted with 0.5 m of factory slotted screen. Water level measurements were density corrected.
Major sampling campaigns included 2000, 2001, and 2006 for major ions, 2004 for density, and 2002 and 2018 for stable isotopes and tritium. The wells were purged, often to dryness, prior to sampling the following day. Solute and isotopic samples were analyzed at Brigham Young University (USA) facilities for major ions, total dissolved solids (TDS), density, δ 2 H, δ 18 O, δ 13 C, 3 H, and 14 C. Laboratory instruments included a PerkinElmer 760 AA for cations, a Dionex ICS-90 ion chromatograph for anions, automatic titration for bicarbonate, a Los Gatos cavity-ringdown stable isotope analyzer for δ 2 H and δ 18 O, a PerkinElmer Quantalus low activity liquid scintillation counter for tritium, and atomic mass spectrometry (AMS) for carbon-14. Selected wells sampled in 2002 for stable isotopes were resampled in 2018 for δ 2 H and δ 18 O to validate the laboratory results. Wells constructed after the 2002 stable isotopic sampling event were sampled in 2018 for stable isotopes. TDS (mg/L) was calculated by summing the major ion concentrations, whereas temperature-corrected density (g/cm 3 ) was determined for 231 well samples collected in 2004 by mass-volume calculations. δ 2 H and δ 18 O samples were preprocessed by vacuum distillation and tritium samples were electrolytically enriched prior to analysis. Stable isotopes were analyzed relative to the standard VSMOW. Sample locations, well information, and average major ion chemistry and isotopic values are included in the electronic supplemental material (ESM). The mineralogy of sediment samples was determined by x-ray diffraction (XRD) and the computer program RockJock (Eberl 2003). The computer code NETPATH-WIN (El-Kadi et al. 2011) was used to model the geochemical evolution.

Geology
The Pilot Valley basin is bounded by Lemay Island to the north, the Pilot Range to the west, the Silver Island Range to the east and southeast, and the Leppy Hills to the south (Fig.  1). The basin is the result of crustal extension block faulting that began about 11 million years ago. The west and east bounding ranges rise to as much as 3,266 m above mean sea level (amsl). The valley floor is~1,295 m amsl and slopes gently to the north where it is closed at the low elevation divide between the Lemay and Crater islands. Pleistocene Lake Bonneville communicated with the Pilot Valley embayment thru this divide. The mountain ranges consist of Precambrian metamorphic core complex and Paleozoic carbonate and siliceous bedrock. The valley, a half graben, is filled with as much as 1,600 m of alluvial fan, shoreline, lake bed, and volcanic sediments (Cook et al. 1964;Lines 1979).
The mountain front alluvial fan and pediment surfaces extend 3-5 km from the mountain bedrock to the edge of the playa. The Pilot Range and Silver Island Range fan surfaces slope values are 0.085 and 0.051, and encompass about 151 and 87 km 2 , respectively ( Table 1). The 379-km 2 playa basin has been divided into the north, center, and south subbasins (Mayo et al. 2020). The playa slopes gently north and west toward the topographic low in the north subbasin (Fig. 2). A few cm thick, a 26-km 2 salt pan covers part of the north subbasin surface ( Fig. 1; Table 1). Most of the north and center subbasin surfaces are covered by mud flats containing no vegetation. The north subbasin is nearly flat with elevations ranging from 1,293.5 to 1,295 m amsl (Fig. 2). The south subbasin is a mud flat that occupies about 90 km 2 and is covered by sage bush in the south.
The upper~4 m of the basin sediments consists of a post-Lake Bonneville (Unit I) and Lake Bonneville (units II-IV) stratigraphic units ( Fig. 3; Rey et al. 2016). The Lake Bonneville sediments are massive marls with occasional thin  sandy interbeds. A pre-Lake Bonneville oolitic sand (unit V) that is up to 1 m thick on the basin margins, underlies the Lake Bonneville sequence. Away from the basin margins the unit V oolitic sand thins. Eight or more meters of pre-Lake Bonneville blue-colored marl underlies unit V. Except as noted in the following, references to basin sediments refer to stratigraphic units I-V and the pre-Lake Bonneville sequence. The boundary between pre-Lake Bonneville and Lake Bonneville sediments is nominally 3.8 m below ground surface (bgs). Except for a thin oolitic sand layer in unit III, the post-Lake Bonneville and Lake Bonneville silts, clays, and marls only slowly transmit measurable water. The pre-Lake Bonneville oolitic sand in unit V generally transmits water freely near the basin margins. RockJock mineralogical results for shallow basin sediments have been organized into five groups: water transported clay, detrital, carbonate, sulfate, and chloride (Fig. 4). Samples include 0-0.2 and 0.2-3.5-m bgs depth intervals. The clay and

Climate and precipitation
Pilot Valley is an arid region with hot summers and cold winters. Temperature is near freezing in the winter and peaks at about 30°C in the summer. Based on data from Wendover, Utah, located 10 miles to the south, the average monthly precipitation is only about 12.1 cm and precipitation peaks in the spring and fall. Precipitation rates in the high elevations of the Pilot and Silver Island Ranges are appreciably greater than at Wendover. During the spring, summer, and early fall, monthly pan evaporation rates are greater than 200 mm and peak at 400 mm in July and August (Lines 1979). The Pilot Valley basin average annual precipitation is 159.67 × 10 6 m 3 ( Table 1). The Pilot Range bedrock-alluvial fan and the playa surface each receive about one-third of the total precipitation, and the remaining one-third falls on the combined Silver Island Range and Leppy Hills bedrockalluvial fans. The fate of precipitation falling on the Pilot and Silver Island Ranges is asymmetrical in two aspects. First, some Pilot Range alluvial fan groundwater discharges from springs located at the distal end of the alluvial fan, whereas the Silver Island Range alluvial fan does not support springs. Second, surface flows and playa flooding are annual events from Silver Island Range storms and Pilot Range alluvial fan ephemeral stream flows do not reach the playa. Although about one-third of the total basin precipitation falls on the playa surface, only the north subbasin is partially flooded for a few weeks in most years. This flooding is due to Silver Island Range surface flow, although some Silver Island Range flooding events do not reach the salt pan. Every few years the center subbasin is partially flooded.

Groundwater
Pilot Valley contains three interconnected groundwater systems: mountain block bedrock, mountain front alluvial fan, and playa basin sediment (Mayo et al. 2020). Only the Pilot Range supports bedrock mountain springs and their discharges are quite small (Peterson 1993). The alluvial fans are recharged by underflow from the bedrock highlands, infiltration of ephemeral surface flows that originate in the bedrock highlands, and by the concentration of local overland alluvial fan flow (Table 2). Total annual bedrock and alluvial fan groundwater recharge is~5.7 × 10 6 m 3 of which about 85% occurs in the Pilot Range. Pilot Range alluvial fan springs discharge at or near the alluvial fan-playa interface and are referred to herein as interface springs. Measured spring flow rates range from <1 to~30 L/s and many spring areas do not have surface discharge, but are covered by meadow grass and other hydrophytes. About 31% (1.74 10 6 m 3 ) of Pilot Range bedrock and alluvial fan groundwater discharges at the 50 plus interface springs ( Fig. 5) and~64% (~2.5 × 10 6 m 3 ) of the groundwater underflows to shallow basin sediments. By contrast only about 0.82 × 10 6 m 3 of Silver Island Range bedrock and alluvial fan groundwater underflows to the basin sediments.
Although there is considerable annual water contribution to the playa surface from direct precipitation (55.6 × 10 6 m 3 ; Table 1) and surface runoff from the Silver Island Range (0.66 × 10 6 m 3 ), this water is mostly consumed by evaporation as the basin is closed to surface and groundwater outflow, and vegetation is limited in the south subbasin and absent elsewhere. The potential for recirculation of playa surface water to the shallow basin sediments and then back to the surface for evaporation is discussed in the following.
Two shallow groundwater systems have been identified in the basin sediments (Mayo et al. 2020). The systems are the Lake Bonneville sediments and the pre-Lake Bonneville sediments, which correspond to approximately 0-3.8 and 3.8-12 m bgs, respectively. Except for some basin margin areas the groundwater systems are leaky confined. The general groundwater flow direction is from the alluvial fan-playa interfaces to the subbasins, from the south subbasin to the center subbasin, and then to the north subbasin (Fig. 6). In addition to the northerly flow the dominant groundwater flow direction in the center subbasin is westward from the Silver Island Range alluvial fan toward the Pilot Range alluvial fan (Fig. 7). This westerly flow is the result of the gentle westward tilt of the basin due Basin and Range faulting. Along the A-A′ profile, as elsewhere in the basin, there are upward vertical groundwater gradients from the pre-Lake Bonneville sediments to the Lake Bonneville sediments. Vertical gradients have been calculated for nested well pairs using up to 15 years of data for center and north subbasin wells. Excluding wells located at the alluvial fan-playa interface, the average upward gradients in the north and center subbasins are 0.0135 and 0.0139, respectively. Upward gradient persist throughout the year.
What is particularly interesting about the westward groundwater flow in the center subbasin is the fact that at the Pilot Range alluvial fan-playa interface there is an inverted salinity gradient (Fig. 8) and a north-south trending, linear pressure ridge of freshwater beneath the overlying brine. Mayo et al. (2020) found that the east flowing freshwater in the deeper coarse grained Pilot Range alluvial fan sediments has sufficient hydraulic head, velocity, and flux to keep the west flowing basin sediment brine at bay, while the velocity and flux of the freshwater in the overlying clays are not sufficient to prevent the west movement of shallow brine at the transition zone. They further found that the pressure ridge was the result of an osmotic pressure condition between the deeper less saline water and overlying brine in the clay-rich sediments.

Results
Average major ion concentrations, δ 13 C, 3 H and 14 C, results for streams, springs, mountain block, alluvial fan-playa interfaces, and basin sediments are listed in Table 3. Average stable isotopic data (δ 2 H and δ 18 O) data values for each sampling location are included in the ESM.
The Pilot Range aluminosilicate mountain bedrock springs discharge mixed anion-HCO 3 type water with an average TDS of 53 mg/L, and the carbonate bedrock springs discharge

Leppy Hills -Leppy Hills bedrock and alluvial fan underflow to the playa
Ca-HCO 3 type water with an average TDS of 212 mg/L (Table 3; Fig. 10). The waters are undersaturated with respect to all mineral species of interest ( Fig. 11). Because direct measurements of alluvial fan groundwater chemistry are not possible, bedrock aluminosilicate and carbonate spring water chemistries are assigned as surrogates for their respective alluvial fan groundwaters. The chemistry of Pilot Range carbonate bedrock groundwater is also a reasonable surrogate for the Silver Island Range alluvial fan groundwater chemistry. Ephemeral Pilot Range alluvial fan streams contain Ca-HCO 3 type water with an average TDS of 71 mg/L. The waters are undersaturated with respect to carbonate, chloride, and sulfate minerals. The interface springs become progressively more saline from north to south (Fig. 12). North subbasin springs discharge fresh (TDS 261-469 mg/L) mixed cation-anion type water (Table 3). Center subbasin springs discharge brackish (TDS~1,200-1,400 mg/L) Na-Cl type water containing lessor amounts of Ca 2+ , Na + , and HCO 3 − , and south subbasin springs discharge saline (TDS~9,000-21,000 mg/L) Na-Cl type waters. The potassium in both the center and south subbasin spring waters are elevated and the reacting equivalents of K + in the south subbasin springs exceed the reacting equivalents of SO 4 2− . The north and center subbasin interface spring waters are undersaturated with respect to all mineral species of interest, whereas the south subbasin springs are supersaturated with respect to carbonate minerals (Fig. 11).
Regardless of density, all playa basin sediment groundwaters are Na-Cl type with Na + and Cl − concentrations ranging from~45 to 4,035 and~43 to 4,720 meq/L, respectively (Table 3). Density increases generally parallel to the groundwater flow direction with freshwater near the basin margins evolving to hypersaline in the north and center subbasins (Fig. 13). In Pilot Valley, density is a better indicator of groundwater flow direction than single-event water level measurements, because in the closed basin environment salinity Fig. 5 Locations of springs and seeps that discharge the distal at the alluvial fan-playa interface. Springs are shown as filled circle with a tail reflects the accumulated contact with soluble salts along the groundwater flow path. The density contours show an evolution from less saline water in the south subbasin and basin margins to more saline water in the western side of the center subbasin and at the topographic low in the north subbasin.
Based on average major ion compositions, all basin sediment groundwater density groups are saturated or supersaturated with respect to carbonate minerals (Fig. 11). Groundwaters with density < 1.15 g/cm 3 are undersaturated with respect to sulfate and chloride minerals. In the north subbasin the Lake Bonneville sediment, groundwaters become both halite and gypsum saturated with the area of gypsum saturation appreciably larger than the area of halite saturation (Fig. 14). Although there are not sufficient data to draw saturation index (SI) contours for pre-Lake Bonneville sediment groundwater, the available data suggests that the waters are gypsum supersaturated and halite undersaturated in the central area of the north subbasin.
Stable isotopic data for 127 precipitation samples collected from seven Silver Island Range continuous collectors, three Pilot Range ephemeral streams, seven Pilot Range alluvial fan springs, and 126 basin wells and boreholes are plotted relative to the Global Meteoric Water Line (GMWL; Fig. 15). The precipitation, spring, and stream data plot parallel to the GMWL (Fig. 15a). Most precipitation δ 18 O data fall in the broad range of about of −17 to −9‰, and the spring and stream data fall near the more negative portion of the GMWL with average δ 18 O and δ 2 H values of −14.8 and -117.3‰, respectively. Much of the groundwater in both Lake Bonneville and pre-Lake Bonneville sediments contain  Mayo et al. 2020) apparent evaporated water. Waters with δ 18 O < −11 to −10‰ are considered to be apparently evaporated. Apparent evaporated water includes groundwater in pre-Lake Bonneville sediments and a sample from a basin well screened at~30 m bgs (Fig. 15b).
Nonevaporative groundwater evolves to apparent strongly evaporative groundwater from the basin margins to the center of the north and center subbasins in Lake Bonneville sediment groundwaters (Fig. 16a). A similar trend may exist in the pre-Lake Bonneville sediment groundwater (Fig. 16b), but the data are too limited to confirm this. To better visualize the evolution to apparent evaporated groundwater the δ 18 O data for both Lake Bonneville and pre-Lake Bonneville sediments that lie along a linear trend from the distal end of the Pilot Range alluvial fan to the Silver Island Range fan-playa interface in the center subbasin is plotted in Fig. 17. Groundwater in the Pilot Range alluvial fan-playa interface is nonevaporated, water beneath the playa is apparently evaporated, and groundwater at the Silver Island Range fan-playa interface is transitioning from nonevaporated to apparent evaporated. At the boundary between the Pilot Range fan-playa interface and the playa there is an abrupt change from nonevaporated to apparent evaporated groundwater (Fig. 18). Over a horizontal distance of only about 100 m there is a remarkable~8‰ change in δ 18 O from approximately −15 to −7‰. About 400 m east of the playa margin the δ 18 O is only about −3‰. The shift from nonevaporated to apparent evaporated groundwater near the boundary of alluvial fan-playa interface and the playa involves both Lake Bonneville and pre-Lake Bonneville sediments groundwater, including wells that are screened at 14.5-30 m bgs. All of the wells at the interface boundary are screened more than 1.5 m bgs and most are screened more than 2 m bgs.  Fig. 7. Modified after Mayo et al. (2020) Tritium samples were analyzed for two streams, two alluvial fan wells, seven interface springs, and basin groundwaters in 27 Lake Bonneville and 18 pre-Lake Bonneville sediment wells (Table 3; Fig. 19). Carbon-14 was analyzed for two interface springs and one alluvial fan well. Regardless of playa location almost all groundwater in both Lake Bonneville and pre-Lake Bonneville sediments contain measurable tritium. Pre-Lake Bonneville sediment groundwater has an average of 1.52 ± 1.57 TU with two of the samples having less than the minimum detection limit (mdl) of 0.3 TU. Lake Bonneville sediment groundwater has an average of 3.58 ± 2.5 TU with two samples containing less than the mdl. The interface springs contain between <0.3 and 4.1 TU. Two north and center subbasin interface springs with 0.3 TU or less (i.e., AS-6 Tessa, and AS-1 Halls Meadow) have calculated 14 C ages of 3,100 and 2,100 years BP, respectively (Table 3). Groundwater in a 30-m deep irrigation well in the Pilot Range alluvial fan contains 58% modern carbon (pmc) and has a calculated 14 C age of modern.
Tritium data are not available for local precipitation; however, Lindon, Utah, precipitation data are a reasonable surrogate for Pilot Valley in that Lindon has a similar physiographic position and receives precipitation from the same storm tracks as Pilot Valley. The median Lindon 3 H in 79 samples collected between 2002 and 2012 is 7.6 TU (average 8.5 TU ± 4.8; Mayo and Tingey 2021).

Basin margin groundwater
The 3 H in the interface springs suggests two different alluvial fan groundwater flow-path travel times. One group of springs  differences. Some center subbasin springs, which likely have contact with saline lakebed sediments, also have dissimilar 3 H contents. Both AS 1 and AS 4 have apparent 3 H travels times >50 years, whereas AS 10 has a 3 H travel time of~12 years. The very low 3 H content and the calculated 14 C ages of 2,000-3,000 years BP of many interface springs are unexpected. The small 3 H contents in some interface springs suggests that the extremely slow shallow alluvial fan groundwater flow is isolated from modern groundwater recharge along much of the relatively short (3-5 km) and steep alluvial fans, and that inflows from ephemeral alluvial fan streams are not a major recharge source for this group of older alluvial fan groundwater. The fact that~84% of the calculated Pilot Range groundwater recharge occurs in the bedrock highlands before   (3)  Fig. 14 Contour maps of Lake Bonneville sediment groundwater a Halite saturation of groundwater in Lake Bonneville sediments, and b Gypsum saturation of groundwater in Lake Bonneville sediments. Saturation is Log SI =0.0 ± 0.1. Filled circles are sampling locations bgs, contain~7 TU and 58 pmc. Some are seasonally pumped for intervals of 2-4 days and have drawdowns as great as 22 m. The large drawdowns are consistent with a low hydraulic conductivity clayey matrix in the fan gravels. What is particularly significant is the fact that one of the well bores was dry to the 24.4-m bgs completion depth. The dry well bore and the clustering of the interface springs at the distal end of individual fans (Fig. 5) support the idea that the alluvial fan groundwater flow occurs in isolated systems. Isolated systems are consistent with the depositional history of the alluvial fans along the 23-km-long mountain front in that the mountain front alluvial fan system is a bajada (i.e., coalescing alluvial fans). Erosional fan debris carried by each of the major bedrock stream systems has its own internal depositional structure and distribution of coarse and fine-grained sediments. These internal sedimentary differences make it possible for some fans to more easily acquire modern recharge water than others.
South subbasin interface springs discharge from lake bed sediments and have elevated TDS (Table 3; Fig. 12). The upgradient bedrock and alluvial fan areas for both AS 9 and AS 12 are significantly undersized relative to the size of the groundwater recharge areas necessary to support the springs. During recent drought years these springs have only existed as ponds and do not discharge sufficient water to maintain their previously large wetland areas. The cause of the undersized upgradient recharge areas, particularly for AS 9, has historically been problematic. The idea of south-moving horizontal groundwater flow from the Pilot Range bedrock massive along the damage zone of a Basin and Range frontal fault, followed by upwelling at the springs, is attractive; however, the 3 H contents of 1.9-4.0 TU of the spring waters preclude this idea. Such short apparent 3 H travel times of~12 to 24 years and the absence of major ephemeral stream channels upgradient of the springs mean that the 3 H travel times are not consistent with the calculated 14 C ages and the long 3 H travel times associated with many of the north and center subbasin springs. It is more likely that the south subbasin springs AS 9 and AS 12 are supported by alluvial fan groundwater underflow from Pilot Creek valley located on the west side of the Pilot Range (Fig. 12). Here only a low elevation ridge of carbonate bedrock separates the Pilot Creek basin from the southwestern most Pilot Range alluvial fan. In the vicinity of AS 12, a distal reach of a Pilot Creek Valley anastomosing alluvial fan overlies the carbonate ridge. Likely alluvial fan groundwater flow paths are shown as arrows in Fig. 12.

Basin groundwater
Basin wide 3 H data indicate that the deeper pre-Lake Bonneville sediments contain older groundwater than the overlying Lake Bonneville sediments (Fig. 19a). The mean and median calculated 3 H ages of the deeper sediment groundwaters are~12 and 24 3 H years older than the overlying groundwater, respectively. The 3 H values for the two groundwater systems are statically different at the 95% confidence level with a p value of 0.0085. Near the alluvial fanplaya interface, Lake Bonneville sediment groundwater contains an average of 3.75 TU, whereas the deeper pre-Lake Bonneville sediment groundwater only contains an  (Fig. 19b). This difference suggests that groundwater recharging the pre-Lake Bonneville sediments has about a 15 year longer 3 H travel time than the shallower groundwater. The 30 m deep well located near the interface contains <0.2 TU indicating more than 50 years of travel time for deepest groundwater flow. The longer 3 H travel times of the deeper groundwaters suggests that most shallow basin groundwater recharge occurs in the Lake Bonneville sediments and that upward contributions from deeper pre-Lake Bonneville sediment groundwater to the overlying sediments is limited.
The idea of shallow basin groundwater recharge from alluvial fan underflow is well established (Carling et al. 2012;Mayo et al. 2020). What is less well understood is the contribution of basin recharge from direct playa precipitation and playa flooding. Upward gradients from pre-Lake Bonneville to the Lake Bonneville sediments preclude pre-Lake Bonneville sediment groundwater recharge from the playa surface. Although the areal distribution of apparently evaporated groundwater in center and north subbasin Lake Bonneville sediments is similar to areas that are occasionally or seasonally flooded (Fig. 16a), temporal water level data from nested wells in the topographic low of the north subbasin indicate that playa surface infiltration to the deeper portion of the Lake Bonneville sediments is not a factor (Fig. 20a). The data also show that there are continuous upward gradients throughout the stratigraphic column in the topographic low and that the pre-Lake Bonneville and Lake Bonneville sediment groundwater systems behave differently. In the north subbasin the shallowest Lake Bonneville sediment pressure surfaces are relatively stable throughout the water year, varying only about 0.2 m (Fig. 20b). In the deeper Lake Bonneville and pre-Lake Bonneville sediments water levels are seasonally dependent and vary more than 0.5 m in the pre-Lake Bonneville sediments. In the topographic low of the north subbasin the annual recharge pulse in the pre-Lake Bonneville sediments first arrives in May and continues to build until late fall. What this means is that the flux of the annual recharge slug in the deep oolitic sand is greater than the rate of the permeability inhibited upward groundwater movement through the overlying clay sequences. The several month delay in the annual recharge pulse reaching the deeper Lake Bonneville sediments is consistent with slow vertical movement through the thick Lake Bonneville sediment clay package.
Evaporation is an attractive idea to be the cause of the apparent evaporative groundwater in the center and north subbasin Lake Bonneville sediments (Fig. 16a) as all the groundwater is ultimately lost to evaporation. However, less than 50% of the wells in the north subbasin are screened within 1 m of the ground surface (Fig. 21) and only five wells have top of well screens within 0.5 m of the ground surface. Basin wide only 17.5% of the wells have a top-of-well screen within 1 m of the ground surface. Additionally, the abrupt change from nonevaporated to apparent evaporated groundwater at the Pilot Range alluvial fan-playa interface occurs over a horizontal distance of about 100 m and at a depth of more than 2 m bgs (Fig. 18). What this means is that an alternative mechanism is responsible for most of the isotopic fractionation.
The pre-Lake Bonneville and Lake Bonneville sediment packages are fundamentally carbonate-clay membranes with thin stingers of water transmitting carbonate sand (Figs. 3 and  4). The clays, predominantly halloysite, Fe-smectite, illite, Fechlorite, and montmorillonite, surface charges and a cation exchange capacity (CEC) which allows them to interact with a range of solutes. It is well established that clay membranes can restrict the migration of solutes in native environments and in engineered barriers (Coplen and Hanshaw 1973;Fritz 1986;Malusis et al. 2012;Neuzil and Person 2017;Neuzil and Provost 2009;Whitworth and Fritz 1994). Laboratory experiments and field observations have demonstrated that hyperfiltration in clay environments, due to advective fractionation, can affect hydrogen, oxygen, carbon, chloride, and lithium isotopes (Benzel and Graf 1984;Campbell 1985;Coplen and Hanshaw 1973;Fritz et al. 1987;Hitchon and Friedman 1969;Oerter et al. 2014Oerter et al. , 2018Philips and Bentley 1987;Whitworth 1993;Wang et al., unpublished data). Although reported δ 18 O fractionation due to hyperfiltration has been only 1.5‰ or less, isotopic fractionation due to advective hyperfiltration is the likely process in Pilot Valley. The idea of hyperfiltration isotopic fractionation is supported by several factors: (1) evaporation is not a factor for most Pilot Valley fractionation, (2) sluggish flow through clay membranes induce hydrogen and oxygen isotopic fractionation, (3) the Pilot Valley sedimentary packages are thick clay membranes with strong upward gradients originating in thin water-bearing oolitic sand layers, (4) the very low permeability of the clay packages inhibits the upward groundwater flux, thus promoting the membrane effect, and (5) the evolution of δ 18 O from approximately −16 to < −10‰ occurs over extremely short horizontal flow distances and at depths of 2 m or more bgs.

Groundwater chemical evolution
The shallow-groundwater chemical evolution can be visualized as four geochemical flow paths: (1) infiltration of precipitation to the bedrock and alluvial fans, (2) alluvial fan flow to the alluvial fan-playa interface and interface springs, (3) interface flow to the basin sediments, and (4) basin margin and south to north subbasin flow (Fig. 22). Mountain bedrock and alluvial fan groundwater chemically evolves as it flows down gradient in response to soil zone CO 2(g) and interactions with bedrock and alluvial fan minerals. At the alluvial fan-playa interface, water quality is impacted by contact with varying amounts of lake bed sediments that contain soluble sulfate and chloride minerals. Groundwater in the basin continues to evolve from fresh and brackish water at the alluvial fanplaya interface to saline, brine, and hypersaline Na-Cl type water as it flows in basin sediments to the terminal sink in the salt pan area.
The first three flow paths are straight forward and behave as anticipated. The bedrock and alluvial fan groundwaters evolve from precipitation to low TDS Ca-HCO 3 and Ca-Mg-HCO 3 type groundwater in the aluminosilicate and carbonate terrains, respectively. The very low TDS of groundwater in the aluminosilicate terrain (~50 mg/L) is due to the low solubility of the quarts, orthoclase, anorthite, and amphibole. The chemistry of carbonate terrain groundwater has a slightly higher TDS (~200 mg/L) due to interactions with calcite and dolomite.
Although the north subbasin interface springs are down gradient of alumninosilicate alluvial fans and there are no surface outcrops of lake bed sediments, the increase in both Na + and Cl − , from 0.2 to 2.86 and 0.1 to 2.63 meq/L, respectively, means that the groundwater flow paths encounter some lake bed sediments near the springs. Elsewhere, the progressive north to south increases in Na + , Cl − and SO 4 2− in the interface spring waters (Fig. 12) are consistent with contact with increasing amounts of playa-lake bed carbonate mud that contains sulfate and chloride minerals. The chemical evolution of pre-Lake Bonneville and Lake Bonneville sediment groundwaters are similar despite the fact that the compositions of their initial waters vary and are location specific. The general pattern is to evolve from low to moderate TDS water at the alluvial fan-playa interface to saline and hypersaline Na-Cl type water with elevated SO 4 2as exemplified in groundwater density contour maps (Fig. 13).
Evaluating the chemical evolution of basin sediment groundwaters is more complex. The easy solution would be to attribute the evolution to contact with sulfate and chloride bearing minerals in the basin sediments and to significant evapoconcentration of Lake Bonneville sediment groundwater. Taken at face value the evaporative δ 18 O signatures of north and center subbasin pre-Lake Bonneville sediments (Figs. 15b and 16a) support this idea. As discussed above, the problem is that there are upward gradients between the two groundwater systems and only a small portion of the north subbasin well screens are within 1 m of the ground surface (Fig. 21). Evapoconcentration deeper than 1 m bgs is unlikely. Additionally, there is essentially no vertical chemical evolution in the conservative species Cl − and SO 4 2− in nested wells in the north and center subbasin wells, although there is a substantial increase in Cl-from the center to north subbasin wells (Fig. 23). The decline in SO 4 2− concentrations in the north subbasin nested wells relative to the south subbasin wells is attributed to gypsum precipitation in the supersaturated water (Fig. 14).
There could be an influx of evaporated groundwater from the Pilot Range alluvial fan due to the re-infiltration of evaporated spring discharge water at the alluvial fan-playa interface; however, such a mechanism is unlikely. The stable isotopic data at the alluvial fan-playa interface along the western edge of the center subbasin clearly demonstrates that the apparent evaporative signatures evolve in-situ and at depth (Fig. 18). The data also suggest that the apparent evaporative signatures in the center of the basin in both pre-Lake Bonneville and Lake Bonneville sediments also evolve in-situ (Fig. 17). Although the most likely cause of the in-situ isotopic evolution is isotopic fractionation associated with hyperfiltration in the low hydraulic conductivity clays, hyperfiltration does not appear to impact the solute chemistry. The concentrations of the conservative species Cl − and SO 4 2− are relatively stable or slightly increase in the vertical profiles of the nested wells (Fig. 23). The absence of a solute membrane effect is consistent with in-situ dissolution of the abundant sulfate and chloride minerals in the basin sediments (Fig. 4). The concentrations of sulfate and chloride minerals, which approach 10% by weight of the sediment mineralogy, means that there are ample soluble salts in the sediment package to account for the hypersalinity of the basin groundwaters.

Geochemical modeling
Inverse modeling was used because initial and final water compositions and flow-path mineralogy are known. Wellcharacterized mineral assemblages for both reactant and product phases are critical for inverse reaction modeling (Bowser and Jones 2002). Modeling each groundwater flow path was based on the average compositions of end member waters (Table 3) and mineral phases identified by RockJock analysis. Rayleigh distillation calculations for δ 13 C were used to better constrain the carbon evolution for the interface systems. δ 13 C was not used for basin sediment models as the carbon composition is substantially altered by interaction with carbonate mud. Because NETPATH-WIN does not validate results against SI, only modeling results which were consistent with the calculated SI of modeled mineral phases were considered plausible.
Representative groundwater flow paths selected for modeling were: Alluvial fan-playa interface: Flow The chemistry at the interface springs was used to represent the interface chemistries. In NETPATH-WIN modeling, reactants and product phases are described in terms of constraints and phases. Constraints included the major elements Ca, Mg, Na, K, C, Cl, and S, and δ 13 C. Modeled phases were selected to be representative of the known mineralogy in the alluvial fans and basin sediments: (1) augite, K-feldspar, quartz, plagioclase (An35), and pyroxene for the dissolution of low solubility aluminosilicate minerals, and (2) calcite, dolomite, gypsum, halite, sylvite, and Na-Ca ion exchange for alluvial fan-playa interface and basin minerals. Alluvial fan-playa interface models included δ 13 C, and some basin models included evaporation. Alluvial fan to interface spring modeling results are consistent with the southerly increase in lake bed sediment content at the interface and indicate that evaporation could play a major role in the elevated chloride content (Fig. 24a). For example, in the south subbasin springs the modeled halite dissolution is 138. 6 mmole L −1 without evaporation and is only 8.7 mmole L −1 with~96% evaporation. A large evaporation contribution is unlikely because the sampled springs all flow freely. Except for the north subbasin springs, ion exchange was not a modeled result even though there is an abundance of clay minerals at the center and south subbasin fan-playa interfaces. Exchange was not a model outcome because the reacting equivalents of Na + and Cl − are similar in most waters (Table 3). Although basin sediments are not observed in the near surface geology associated with the north subbasin springs, the modeling indicates about~2 mmole L −1 of halite dissolution is required to account for the average 93 mg/L Cl − concentration.
Basin sediment models included both no evaporation and evaporation (Fig. 24b,c). Evaporation is only plausible in the topographic low of the north subbasin. Calcite precipitation is common in most model runs, which helps explain the carbonate mineral content of the basin sediments (Fig. 4). Not surprisingly, the modeling demonstrates that halite is the primary driver of elevated TDS with or without evaporation. Evaporation is likely a significant factor in the north subbasin areas where groundwater is near the land surface, but not elsewhere.

Salt transfer
The modeling demonstrates that pervasive sulfate and chloride mineral dissolution plus evaporation in the near surface sediments in the topographic low of the north subbasin are the driving mechanisms in basin groundwater evolution. However, the modeling does not answer the key questions: (1) why does the TDS and mineral saturation progressively increase along the flow path (Fig. 13), and (2) what is the origin of the salt pan? Two possible mechanisms could account for the progressive increase in TDS and mineral saturation. The sulfate and chloride mineral content of the sediments may progressively increase from the south to the north and from the basin margins. Alternatively, the groundwater flow may progressively wash out soluble minerals up gradient and transfer the salts to the topographic and structural low of the basin. The limited data suggest that halite and gypsum concentrations in Lake Bonneville sediments are relatively constant in the south and center subbasins at about 0.01-0.04 and 0.004-0.009 g/cm 3 for gypsum and halite, respectively. Gypsum increases to 0.028 g/ cm 3 in the most northern portion of the north subbasin and halite locally increases to 0.18 g/cm 3 in the center of the north subbasin. The fact that areas in both the north and center subbasins with elevated gypsum and halite saturation do not  correspond with the regions of elevated sediment mineral concentrations suggests that the distribution of gypsum and halite is not the primary driver of the progressive increase in TDS. The concentrations of gypsum and halite throughout the basin sediments suggests that continuous mineral dissolution occurs along the groundwater flow paths until mineral saturation is reached (Fig. 25). The most likely source of these minerals was co-precipitation with fine-grained detrital deposition associated with Lake Bonneville and earlier lakes in the nearly closed Pilot Valley embayment. Evaporation from the playa surface is a contributing factor in the increased gypsum and halite saturation in Lake Bonneville sediments where the water surface is within 1 m of ground surface. As the water moves to the surface via capillary action it carries the dissolved soluble salts to the land surface where they precipitate. Episodic flooding then redissolves the salt and carries it to the annually flooded salt pan where it accumulates and forms a salt crust during the dry season.

Conclusions
Pilot Valley shallow groundwater chemical evolution can be visualized along four geochemical flow paths: (1) infiltration of precipitation to the bedrock and alluvial fans, (2) alluvial fan flow to the alluvial fan-playa interface, (3) interface flow to the endorheic basin sediments, and (4) basin margin and south to north subbasin flow. Of interest here is the upper most 12 m of the basin sediment consisting of low hydraulic conductivity Pleistocene age Lake Bonneville and pre-Lake Bonneville lacustrine calcareous mud with a few oolitic sand units. The oolitic sands transmit most shallow-basin active groundwater flow and the clay sequences are effectively thick clay membranes. The oolitic sands are not near the top of the sedimentary sequence and the flux of vertical flow is inhibited by the thick lacustrine clay.
Shallow groundwater in much of the central and northern portion of the basin has apparent evaporative δ 2 H and δ 18 O signatures. Isotopic fractionation of up to −8‰ in δ 18 O is attributed to hyperfiltration in the clay sequences. Groundwater in the basin evolves from fresh and brackish water at the alluvial fan-playa interface to saline, brine, and hypersaline Na-Cl type water as it flows to the terminal sink in the salt pan area. Although hyperfiltration impacts the stable isotopes, hyperfiltration has not significantly affected most of the basin groundwater chemistry. The absence of a significant solute membrane effect is the result of the abundance of sulfate and chloride minerals in the basin sediments.
Based on an analysis of sediment mineralogy, water levels, and solute and isotopic compositions a model of shallow endorheic basin groundwater chemical evolution and hypersalinization that does not involve basin wide evaporation has been developed. The model envisions the remobilization of authigenic soluble salts that were deposited over time with fine-grained detrital sediments in the endorheic basin. Groundwater flow path sulfate and chloride mineral dissolution is the primary driving mechanism for hypersalinization. Evaporation only impacts the groundwater quality where the groundwater is within~1 m of the ground surface. Here capillary action carries dissolved soluble salts to the land surface. Episodic flooding redissolves and carries the precipitated salt to the annually flooded salt pan where it accumulates as a salt crust during the dry season. The Pilot Valley model may help explain the long-term accumulation of buried layers containing local concentrations of highly soluble ions in some endorheic basins. When these concentrated zones include high porosity layers, brines may be commercially extracted. who assisted us in both the field and laboratory. We also thank the three anonymous reviewers whose insights help us clarify and broaden the appeal of the manuscript.

Declarations
Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.