EVALUATION OF AQUEOUS GASTROPOD SHELLS AS GROUNDWATER RADIOCARBON PROXIES ACROSS SPECIES AND SITES

Abstract

We evaluate carbonate gastropod shells as ¹⁴C proxies for groundwater discharge at springs. Groundwater ¹⁴C is commonly used to estimate groundwater transit times, and a carbonate shell proxy would present a different way of collecting groundwater ¹⁴C data. Specifically, we test the hypothesis that in exclusively groundwater-fed spring systems, water ¹⁴C is preserved in carbonate shells at multiple sites, species, and water ¹⁴C. We first present isotopic and water temperature variability over several years at three spring sites in Utah. We then compare the ¹⁴C of contemporaneously collected water, sediment, and shells of benthic gastropods (Melanoides tuberculata, Pyrgulopsis pilsbryana, and Physella gyrina). We show that water and shell ¹⁴C activities at each site are correlated (slope = 1.00, R² = 0.999, n = 22). These results support the hypothesis that ¹⁴C from groundwater is preserved in carbonate shells, and that aqueous gastropods a viable groundwater ¹⁴C proxy. Finally, we describe the utility and limitations of using gastropod shells as a groundwater ¹⁴C proxy.

INTRODUCTION

Radiocarbon (¹⁴C) can be used to estimate apparent transit times in groundwater aquifers (also referred to as residence times and groundwater “ages” in the literature) (Ingerson and Pearson 1964; Mook et al. 1974; Tamers 1975; Fontes and Garnier 1979; Fontes 1992; Kalin 2000; Geyh 2000; Plummer and Glynn 2013; Han and Plummer 2016; Cartwright et al. 2020; and others). Modeling ¹⁴C depletion in groundwater must account for the variation of atmospheric or initial ¹⁴C activities and for soil-gas exchange and C dissolution in both open and closed systems. These processes are commonly accounted for using stable carbon isotopes (δ¹³C), dissolved inorganic carbon (DIC) concentration, and knowledge of the aquifer system mineralogy and flow geometry (with a variety of methods reviewed in Clark and Fritz 2013; Han and Plummer 2016; Cartwright et al. 2020 and references therein). Because of these considerations, ¹⁴C is useful to estimate groundwater mean transit times of 100–30,000 years and are particularly useful (if imprecise) in arid and semi-arid regional aquifers (Cartwright et al. 2020). Calculation of these averaged transit times are critical for understanding groundwater flow rates, which in turn inform groundwater recharge rates and management strategies for sustainable resource use (Cook and Böhlke 2000).

Gastropods and other mollusks build carbonate shells, the chemistry of which have been used as a proxy for a variety of paleoenvironmental conditions. Gastropods are excellent proxies for aquatic geochemical environments because they are globally distributed and live in a wide variety of environments from hypersaline and marine, to freshwater environments (Fortunado 2016). Many studies show that aqueous δ¹³C signatures are retained in carbonate shells are preserved in the sedimentary record (Keith et al. 1964; Mook and Vogel 1968; Fritz and Poplawski 1974; von Grafenstein et al. 1992; Tevesz et al. 1997; Balakrishnan et al. 2005; Shanahan et al. 2005; Salvador 2018; and others). This paper addresses the idea that gastropod shells may provide material which records ¹⁴C of groundwater, a proxy for groundwater transit times. The collection of spring-dwelling gastropod shells requires little equipment (a shovel and a container) and the ¹⁴C analysis is a simple and cost-effective laboratory procedure.

Previous studies have suggested this direct relationship between the ¹⁴C activity of springwaters and shells. Riggs (1984) described the springwater at two sites with very depleted water ¹⁴C activities (11.6 percent modern carbon, or pMC, at Crystal Pool and 3.9 pMC Big Springs in Ash Meadows, Nevada) measured in 1973 (Pearson and Bodden 1975; Winograd and Pearson 1976) and two shells collected the same sites at an unspecified later date (10.6 pMC and 3.3 pMC for Big Spring and Crystal Pool, respectively). Riggs reports a third shell and water at King Spring with a 4.3 pMC difference (where water is 1.6 pMC and shell is 5.9 pMC), which they attribute to partial water ¹⁴C reequilibration with the atmosphere due to the shallow flow from the discharge are to the shell collection pool. Notably, Brennan and Quade (1997) use Riggs’ (1984) correlation to apply this ¹⁴C depletion in sedimentary records as a for past groundwater transit times. However, they suggest but do not model the potential effect of springwater reequilibration with air as it flows away from the discharge points, nor do they investigate a numerical relationship between modern waters and modern shells at the study sites (although these sites are in the same general region of southern Nevada, U.S.A. as the Riggs 1984 study). Lastly, Copeland et al. (2012) suggest that the ¹⁴C depletion due to groundwater chemistry described by Riggs (1984) and Brennan and Quade (1997) may explain their observed ¹⁴C depletion of aquatic snails relative to stratigraphic age markers in Late Pleistocene archaeological deposits, but do not argue that the shells are directly representing groundwater transit times.

To add to these previous studies, this study includes (1) a seasonal evaluation of water isotopic variation through time, (2) contemporaneously collected water and shell samples, (3) shells collected directly at spring discharge sites, (4) with a range of groundwater transit times (modern and Holocene) to derive a robust correlation at different spring types, (5) shells of several genera (including endemic genera) to constrain inter-species variation of Δ¹⁴C_s-gw, (6) a comparison of ¹⁴C of organic C in detritus to shell and water ¹⁴C, and (7) a transect of ¹⁴C in shells with increasing distance from the subsurface discharge point at one site.

We measure the relationship between modern water and modern shells, testing the viability of gastropod shells as a proxy for groundwater ¹⁴C. Specifically, we test the hypothesis that groundwater and shell ¹⁴C activities (C _gw and C _s , respectively) are equal across gastropod genera, sites, and at different ¹⁴C activities. We propose the notation of Δ ¹⁴ C _s-gw to represent this comparison calculated by the following equation:

(1) $${\Delta ^{14}}{{{C}}_{{{s - gw}}}}{ = ^{14}}{{{C}}_{{s}}}{ - ^{14}}{{{C}}_{{{gw}}}}$$

where Δ ¹⁴ C _s-gw is 0 if the hypothesis is supported.

In comparison, stable carbon isotopes (δ¹³C) have been shown to be generally retained in carbonate shells but also exhibit measurable variations with shell mineralogy, species, growth season, temperature, salinity, or other environmental parameters often described as “vital effects” (Balakrishnan et al. 2005; Shanahan et al. 2005; Salvador 2018). We consider the following three C sources that contribute to carbonate shell formation: (1) Carbon dioxide in the atmosphere equilibrates with dissolved gasses in the water at the air-water interface, where ¹⁴C_air (or “modern carbon”) is approximately 100 pMC. (2) Inorganic carbon dissolved in groundwater seeps into the spring through the bottom sediments or fractured bedrock. This course of DIC is likely depleted in ¹⁴C due to the transit time in the aquifer saturated zone, with contributions from other DIC sources such as dissolution of carbonate aquifer material, among others (for more information, please see a more thorough examination of these considerations in the discussion). (3) Organic carbon in detritus from plant, algal, or microbial material either falls into the spring or grows in the water column. The detritus is consumed by gastropods and used to form shell carbonate. We measure the organic ¹⁴C in sediment which we treat as a relative proxy for metabolic ¹⁴C. These C fluxes are presented in Figure 1A and described further by McConnaughey and Gillikin (2008).

Figure 1 Site information. (A) A schematic cross-section (not to scale) of spring with potential carbon sources that contribute to the formation of carbonate shells. (B) Sample sites are shown as dots on a map of Utah. The purple cross represents a weather station. (C) Sampling sites within the Blue Lake (BL) spring system. (D) Sampling site from Horseshoe Spring (HRS). (E) Field photo from Red Butte Spring.

Overall, this study examines the relationship between groundwater and aquatic gastropod shells as an alternative sampling material for researchers investigating groundwater transit times.

Site Description

We sampled three spring sites in northwestern Utah, on traditional and ancestral lands of the Newe/Western Shoshone, Goshute, and Ute peoples (Figure 1B). Blue Lake (BL) and Horseshoe (HRS) springs are large wetland spring systems in Utah’s West Desert. The third site is Red Butte Spring (RBS): an urban spring in Salt Lake City that runs down a hillside into Red Butte Creek. All three spring sites support perennial wetland ecosystems and were selected as they represent a range of groundwater transit times, spanning the Holocene (BL) to modern (RBS) (Fitzgerald 2019; Lerback et al. 2019).

BL is in the Salt Spring Waterfowl Management Area, near the Utah-Nevada border and HRS is just south of the Great Salt Lake (Figure 1C). Neither BL nor HRS are fed by surface water runoff as there is no visual evidence of streams directed to the springs, and both springs occur near fractured limestone outcrops and flow along a slight topographic slope towards the basinal playa deposits, remnants of glacial Lake Bonneville (Louderback and Rhode 2009; Lerback et al. 2019). BL and HRS are fed by groundwater discharge from the bottom of the springs. Prior investigation at BL-spring modeled ¹⁴C apparent groundwater transit times of Holocene ages (Lerback et al. 2019). BL is the largest of the spring systems sampled and was sampled from three sites: (1) a shallow (0.5 m depth) spring (called “BL-spring”), which seeps from fractured limestone a meter above the other pools, (2) a pond (“BL-pond”) 2–3 m in depth just below the BL-Spring, and (3) the main lake bottom, about 17 m in depth (“BL-lake”).

HRS has two primary ponds with perennial active discharge that converge as the water runs west towards the playa sediments (Figure 1D).

RBS discharge had a tritium activity over 5 TU, representing modern groundwater recharge, and a modeled apparent transit time between 0 and 5 years using ³H-³He dating methods (Solomon et al. 1993; Fitzgerald 2019). This site is on a slope such that water pools only to about 5 cm depth. The springwater discharges from a piped spring box, which is where aquatic gastropods were found (Figure 1E).

Gastropod Physiology

Gastropods build their shells in a series of conical spirals, or whorls. Calcium and CO₂-bicarbonate ions are primarily supplied by direct uptake from the surrounding environment, although in some species metabolic carbonate can supply up to 10% of the carbon (Wilbur and Saleuddin 1983; McConnaughey 2003; Shanahan et al. 2005). Shell carbonate is precipitated in the extrapallial fluid or extracellular calcifying fluid between the mantle (soft inner tissues) and the solid shell. Calcareous material (either calcite or aragonite) is fixed in layers with an organic matrix to extend or expand the shell whorls (Wilbur 1964). We note that lung-breathing gastropods may use more metabolic C in their shells, making them less ideal representatives of water chemistry (Goodfriend 1992; Goodfriend and Ellis 2002; Stott 2002; Copeland et al. 2012; Hill et al. 2017; Padgett et al. 2019).

MATERIALS AND METHODS

Water and sediments (containing gastropod shells) were collected at BL, HRS, and RBS spring sites in Utah’s West Desert to understand spatial and temporal variation in spring chemistry. Exact locations are provided in the data hosted under the doi:10.5281/zenodo.5587202. Gastropods were collected under a research agreement with Utah Division of Wildlife Resources (4COLL10642).

Spring Environmental Characteristics

Spring environmental characteristics were sampled through multiple seasons from 2018–2020. Sample types and locations are summarized in Table 1. A Hydrolab® MS5 multiparameter Mini Sonde (OTT Hydromet, Loveland, Colorado) or a YSI® multiparameter probe (YSI Inc, Yellow Springs, OH) was used for in-situ measurements of water temperature and specific conductance with a measurement error of less than 0.5°C and 10 μs/cm, respectively. Water was collected in high-density polyethylene (HDPE) bottles that were washed with 5% HCl stable hydrogen and oxygen isotope composition analyses (δ²H and δ¹⁸O, respectively). Measurements of δ²H and δ¹⁸O were performed with a Laser Water Isotope Analyzer Picarro L2130i at the SIRFER laboratory at the University of Utah, and the reported measurement error for these is less than 0.1‰. Weather station data from station DPG25 from 2016 to 2021 (located at latitude 40.52 and longitude –113.75) was accessed through MesoWest. This data provides air temperatures measured from January 2016 to April 2021. Measured and modeled isotopes of precipitation from the University of Utah were collected from 2016 to 2022 from the Online Isotopes in Precipitation Calculator (OIPC; Bowen et al. 2005; West et al. 2009; Bowen et al. 2021).

Table 1 Summary of samples and analyses in study.

^a National Ocean Sciences Accelerator Mass Spectrometer is abbreviated as NOSAMS.

Carbon Isotope Measurements

Water samples to be analyzed for C isotopes were collected from several points where temperature and flow rates indicated spring discharge. Samples were collected in 250-mL glass bottles with solid glass stoppers and cleaned according to Woods Hole Oceanographic Institution specifications. Sample water collected was pumped into the bottle, which was held underwater and overflowed to prevent air equilibration and bubbles within the container. The stopper was then coated with Apiezon-M high vacuum grease and twisted into the bottleneck to ensure the seal.

The top 5 cm of sediment at the spring bottom was collected in clean plastic containers. Sediment for organic carbon analysis was transferred to a clean glass vial and freeze-dried within 48 hours of sample collection. Stored, dried sediments were disaggregated using Calgon and deionized (DI) water following Forester (1999).

Bulk sediment was processed at the University of Utah to disaggregate and sort gastropod shells. Shells collected for isotopic analysis were culled and stored in 90% ethanol. Gastropods analyzed for ¹⁴C in this study include Melanoides tuberculata and Pyrgulopsis pilsbryana, and Physella gyrina (Hovingh 2018). All are aquatic, benthic, bottom-feeding genera, which we chose to study aquatic, gill-breathing taxa because it is likely the C used to build shells is from the aqueous environment (Fritz and Poplawski 1974; Rhoads and Lutz 1980; Wilbur and Saleuddin 1983). Melanoides are an invasive species to Utah and distributed globally (Facon et al. 2003). This species on average lives from 1 to 5 years (Dudgeon et al. 1985; Pointier et al. 1992). This species lives both in brackish and freshwater, as well as at a wide range of water temperatures (Raw et al. 2016). Pyrgulopsis species is found in the Idaho, Utah, and Nevada (Hershler 1994; Liu et al. 2017). Pyrgulopsis are of the subclass Hydrobiidae and are mostly found in water temperatures from 22–35°C (Hershler 1998). Finally, Physella is a widely distributed aquatic species found throughout the United States (Newman et al. 1996). The Physella subspecies sampled in this study (utahensis) is endemic to Utah, Nevada, Colorado, and Wyoming. They are found in vegetated shallow spring-fed pools. This species is categorized as a generalist with a diet that consists of detritus, diatoms, fungi, and microscopic living organisms.

Sieves were used to separate size fractions of sediments, from which gastropods were collected and sorted by genus and photographed. Carbonate shells were then cleaned for chemical analysis using the following practices, also detailed in Caporaletti (2011) and described here. To remove organic material, we soaked samples for 10 minutes with 3% hydrogen peroxide (Xia and Engstrom 1997; Janz and Vennemann 2005). Shells were then rinsed with deionized water and stored in glass vials. Vials with shells and deionized water were sonicated in 10-min increments and rinsed until water remained clear after sonication (Lawrence 2008). Shells were then rinsed with ethanol and dried in a dust-free environment (Mischke 2007). Bulk shells were homogenized using a clean mortar and pestle, and smaller species were combined to acquire an adequate sample size.

Dissolved inorganic carbon in water, inorganic carbonate, and organic carbon in sediment were analyzed for ¹⁴C and δ¹³C at the National Ocean Sciences Accelerator Mass Spectrometer facility at Woods Hole Oceanographic Institution (NOSAMS). ¹⁴C to ¹²C in an unknown sample relative to the known ratio in appropriate measured standards. ¹⁴C was reported as fraction modern C, which we discuss in the text as percent modern C (pMC), as specified by Stuiver and Polach (1977). Modern ¹⁴C is defined at NOSAMS as 95% of the ¹⁴C concentration of NBS Oxalic Acid I, normalized to δ¹³C_VPDB = –19‰ (Olsson 1970). The average reported error is 0.12 pMC. δ¹³C was reported as per mil (‰) relative to Vienna Pee Dee Belemnite (subscript VPDB), where error was not reported but NOSAMS lab accession number is provided. Data processing and statistical analyses were performed using R v3.6 in RStudio.

RESULTS

Spring Environmental Characteristics

BL, HRS, and RBS exhibit relatively constant characteristics through seasons as sampled from 2016 to 2021 (Table 2 and Figure 2; Lerback et al. 2019). BL and HRS are mesothermal spring systems, whose average temperatures are 10 above the mean annual air temperature of 12°C. BL-Lake, BL-Spring, and HRS have average water temperatures of 23°C (max = 24°C, min = 22°C, n = 6), 28°C (max = 31°C, min = 26°C, n = 9), and 21°C (max = 28°C, min = 17°C, n = 10): respectively. RBS has an average temperature of 11°C (max = 12°C, min = 10°C, n = 5), which is lower than the mean annual air temperature because it is approximately 100 m higher in elevation (1372 m MSL) than the weather station (1286 m MSL). The air temperature monthly averages range from 28°C in July to –3°C in December (see boxplots in Figure 2A).

Table 2 Spring environmental data. N.D. means no data.

Figure 2 Seasonal variation in spring characteristics. (A) Water temperature from spring sites and air temperature from weather station DPG25 are shown by month. Daily average air temperature data are represented by monthly boxplots representing four quartiles, mean, and outliers. Temperature Monthly averages of precipitation from 2016-2021 are shown as pink bars, with the axis labeled on the left of the figure. (B) The δ²H isotope composition of springwaters and regional precipitation is shown by month. The isotopic composition of precipitation is represented by measured and modeled data from OIPC. The dashed horizontal lines represent the annual average of precipitation weighted by precipitation amount. (C) The δ¹⁸O isotope composition of springwaters and regional precipitation is shown by month. The isotopic composition of precipitation is represented by measured and modeled data from OIPC. The dashed horizontal lines represent the annual average of precipitation weighted by precipitation amount.

As shown in Figure 2B and Table 2, δ²H isotope composition of springwaters varied by a maximum of 6.4‰ (at RBS) and less than 2‰ at BL and HRS. BL-Lake and BL-Spring yielded an average δ²H isotope composition of –123.8‰ (max = –123.7‰, min = –124.0‰, n = 6), and –124.7‰ (max = –123.9‰, min = –125.0‰, n = 9). The δ²H isotope compositions of HRS and RBS had averages of –122.1‰ (max = –121.5‰, min = 123.2‰, n = 11), and –118.1‰ (max = –114.4‰, min = –120.8‰, n = 4). The measured δ²H isotope composition of precipitation varied from –17.0‰ in May to –193.3‰ in January whereas the modeled δ²H isotope composition of precipitation varied from –63.0‰ in August to –193.0‰ in January. The annual weighted average (by average monthly precipitation) is –107.3‰ and –142.5‰ for measured and modeled δ²H isotope composition of precipitation, respectively.

As shown in Figure 2C and Table 2, δ¹⁸O isotope composition of springwaters varied by a maximum of 2.2‰ at RBS and less than 0.5‰ at BL and HRS. BL-Lake and BL-Spring yielded an average δ¹⁸O isotope composition of –15.8‰ (max = –15.7‰, min = –15.9‰, n = 6), and –16.0‰ (max = –15.9‰, min = –16.1‰, n = 9). The δ¹⁸O isotope compositions of HRS and RBS had averages of –16.0‰ (max = –15.9‰, min = –16.2‰, n = 11), and –15.2‰ (max = –13.8‰, min = –16.0‰, n = 4). The measured δ¹⁸O isotope composition of precipitation varied from –1.0‰ in May to –25.0‰ in December. The modeled δ¹⁸O isotope composition of precipitation varied from –9.2‰ in July to –25.2 in January. The annual weighted average (by average monthly precipitation) is –14.5‰ and –19.3‰ for measured and modeled δ²H isotope composition of precipitation, respectively.

In addition, the specific conductance (SpC) of BL-Lake is 9070 μS/cm (max = 9620 μS/cm, min = 8700μS/cm, n = 6), BL-Spring is 873 μS/cm, (max = 9310 μS/cm, min = 7560 μS/cm, n = 9), and HRS is 10310 μS/cm (max = 12120 μS/cm, min = 9640 μS/cm, n = 11). RBS had an average of 1540 μS/cm (max = 2110 μS/cm, min = 1080 μS/cm, n = 5). We did not observe a seasonal variation in SpC at these sites despite rainfall events, likely because there is less than 30 mm/year on average, and we did not observe evidence of surface water runoff entering the spring systems. Runoff from snowmelt and rain events is not channeled directly to these spring systems.

Carbon Isotope Measurements

The measured radiocarbon activity of water, shell, and sediment are reported in Table 3 and discussed in the following subsections.

Table 3 Measured carbon isotope data. N.D. means no data.

¹⁴C values from shells and water, range from 7.9 pMC at BL-Spring to 81.8 pMC at RBS, and the ¹⁴C of bulk organic C in sediment was 29.5 pMC at BL-Pond and 42.0 pMC at HRS. The average Δ ¹⁴ C _s-gw across gastropod genera and site is 0.27 pMC (SD = 0.77, n = 22). A paired t-test for all shell and water samples resulted in a small, statistically insignificant difference where p > 0.05 (mean difference = 0.27 pMC, t = 1.65, df = 21, p = 0.11). This close relationship does not appear to vary by ¹⁴ C _gw ; a linear model between ¹⁴ C _s and ¹⁴ C _gw yielded a slope of 1.002 with an R² of 0.999 and an RMSE of 0.751 (Figure 3A). The difference between ¹⁴ C _s and ¹⁴ C _metabolic is more than 20 pMC at BL-Pond and 5 pMC at HRS.

Figure 3 C isotope analyses. (A) Water ¹⁴C activity is plotted against the ¹⁴C activity of solid (shell and sediment) samples collected at the same place and same date. The red solid line represents a linear regression between water and shell samples, and regression statistics are annotated on the plot. Analytical error is smaller than the points shown. (B) Water δ¹³C isotope composition is plotted against the δ¹³C isotope composition of solid (shell and sediment) samples collected at the same place. The red solid line represents a linear regression between water and shell samples, and regression statistics are annotated on the plot. Analytical error is smaller than the points shown. (C) A schematic diagram of the bathymetry at HRS. Shells and water were collected at the four marked HRS subsites with increasing distance from the primary HRS discharge point. Δ ¹⁴ C _s-gw and ϵ are annotated for each sample.

Table 4 provides summarize this studies’ sample Δ ¹⁴ C _s-gw by mean, sd, and n, and divided by species and site. An analysis of variance test was run to understand how Δ ¹⁴ C _s-gw varies as a function of species and as a function of site. We found slight differences in average Δ ¹⁴ C _s-gw by both genera (f(2) = 25.35, p < 0.001) and by site (f(4) = 18.5, p < 0.001). Separated by genera, the mean Δ ¹⁴ C _s-gw for Melanoides and Physella shells is below 0.25 pMC, whereas the Pyrgulopsis shells had a Δ ¹⁴ C _s-gw greater than 1. Excluding the data from Pyrgulopsis shells, a paired t-test for shell Δ ¹⁴ C _s-gw (Melanoides and Physella) resulted in a small difference at p < 0.05 (mean difference = −0.21 pMC, t = −2.53, df = 13, p = 0.02). Separated by site, HRS had Δ ¹⁴ C _s-gw greater than 1, whereas the other sites yielded Δ ¹⁴ C _s-gw less than 0.3 pMC.

Table 4 Summary of ¹⁴C isotopic differences measured between water and shells (Δ¹⁴C_s-gw). Analyses do not include data from HRS subsites. N.A. indicates that the statistic is not applicable.

Sediment, shells, and water were analyzed for δ¹³C to understand potential effects of environment-related and/or biomediated carbon isotope fractionation. δ¹³C_VPDB values range between –11.3‰ to –1.8‰ for water and shell samples whereas the sediment δ¹³C_VPDB is –26.6‰ and 20.6‰ at BL-pond and HRS, respectively, which are within the range of values expected of plant material in the region (Hart et al. 2010). A linear model between δ¹³C_shells and δ¹³C_water yielded a slope of 0.897 and an R² of 0.926 (Figure 3B).

To directly compare shell δ¹³C and water δ¹³C, we calculated the difference between shell and water, and plotted the average and SD by species represented by ϵ (Table 4). The ϵ of all samples was –1.88‰ (n = 21, sd = 0.77‰). A paired t-test resulted in a low p-value, where we see there is a small difference in means (t = 11.14, df = 20, p < 0.01). An analysis of variance test was run to understand how ϵ varies as a function of species and as a function of site. We found a statistically-significant difference in ϵ by both genera (f(2) = 7.7, p < 0.001) and by site (f(4) = 18.2, p < 0.001).

While the ¹⁴C activity of modern shells is similar to the ¹⁴C activity of water, it is apparent that the ¹⁴C of sediment is higher in both locations; 29.50 pMC at BL-Pond and 42.04 pMC at HRS. Due to the small sample size, we do not share statistical observation.s. The observed difference could be due to a fraction of the organic carbon in the sediment deriving from vascular plant material, which would reflect an atmospheric ¹⁴C signal (Pigati 2002; Shanahan et al. 2005).

We note that half (4) of the Pyrgulopsis shells were found at the HRS site. At BL-Spring, where Pyrgulopsis and Melanoides shells were collected from the same site, we see a difference in average Δ¹⁴C_s-gw by genera (f(1)=11.96, p = 0.014). At this site, Pyrgulopsis shells had an Δ¹⁴C_s-gw of 0.67 pMC (SD = 0.15, n = 4) and the Melanoides shells had an Δ¹⁴C_s-gw of –0.10 pMC (SD = 0.42, n = 4). This Δ¹⁴C_s-gw difference of 0.76 pMC between species indicates that there may be genera-dependent biomediated fractionation or differences from gastropod shell deposition (if the shell grew in a more shallow environment than where it was deposited).

Potential explanations for the observed Δ ¹⁴ C _s-gw include incorporation of metabolic carbon, live gastropod mobility, or movement after initial shell deposition by the pond bathymetry, circulation, or other sediment disturbance. Previous studies show that it is possible for shells to incorporate up to 10% of their ¹⁴C from metabolic C from consumed detritus (Wilbur and Saleuddin 1983; Raw et al. 2016). The δ¹³C of a shell sample may be used to estimate how much of the ¹⁴C is derived from atmospheric and metabolic C sources. However, this may not be appropriate in all cases, as we show that the δ¹³C composition of shells is less negative than both measured water and sediment sources and that other spring-dwelling gastropod shells been reported to demonstrate some vital effect offsets (Shanahan et al. 2005). It is also possible the shells formed at a shallower area of the pond, where modern ¹⁴C from the air (around 100 pMC) was equilibrating with the water and raising the average ¹⁴C activity in the water. The shells would have grown in an aqueous environment with a slightly elevated ¹⁴C activity, and moved, fell, or were moved by fish or circulation to the deeper sampling point.

To address potential groundwater reequilibration with air as springwater moves away from the discharge point, we provide ¹⁴C data from four subsites at HRS and a contemporaneously collected water sample from the discharge point. These samples have increasing the distance from the bottom discharge site and decreasing depth, which are shown shown schematically in Figure 3C. The shell and water data are provided in Table 3 and annotated Figure 3C, where the water at HRS-subsite0 had 33.7 pMC, and Pyrgulopsis shells ranges from 36.5 pMC at HRS-subsite4 (within a meter of the discharge) to a high 41.7 pMC at HRS-subsite2 (approximately 10 meters from the discharge area). Although there is only one sample at each subsite, these data indicate the distance from the discharge source, pond circulation patterns, and potential for gastropod movement after shell formation should be considered when using this proxy, as has been suggested by Riggs (1984) and Copeland et al. (2012).

DISCUSSION

The data presented in this study support the hypothesis that the ¹⁴ C _s represents ¹⁴ C _gw across three species and three spring wetlands (five sites). The data presented in this study demonstrate that modern shells do not reflect modern air ¹⁴C activities nor bulk sediment organic ¹⁴C (as a proxy for metabolic ¹⁴C). We note that only one site had ¹⁴C data for more than one species.

The ¹⁴ C _s proxy described in this study can be used to model groundwater transit times rather than direct measurement of ¹⁴ C _gw . Measurements of ¹⁴ C _gw are commonly used to estimate aquifer transit times, and many methods have been developed to do this by not only considering the radioactive decay of ¹⁴C along the flowpath, but also corrections including the distiubution of flowpath lengths sampled (e.g., Jurgens et al. 2016; Broers et al. 2021) and the input of ¹⁴C in pre-recharge atmosphere, DIC from the dissolution of aquifer materials, or soil CO₂ (Fontes and Garnier 1979; Kalin 2000; Clark and Fritz 2013; Clark 2015; Han and Plummer 2016; Cartwright et al. 2020; Wang et al. 2020 and references therein). Models of groundwater transit time demonstrate relatively small sensitivity to errors in ¹⁴C input values compared with flowpath geometries, model type, and other associated uncertainties (e.g., initial δ¹³C, heterogeneity in the stable isotope ratios of matrix calcite, open-system carbonate dissolution, methanogenesis; Cartwright et al. 2020). For example, the apparent ¹⁴C transit time at BL-Spring was modelled by Lerback et al. 2019) to be on average Holocene recharge. They used the measured ¹⁴C of 8.5 pMC, δ¹³C of –4.5‰, and alkalinity of 226 mg/L as CaCO3 and pH of 7.3, with an initial ¹⁴C of 100 pMC and δ¹³C of –22‰. DIC from limestone was modelled to have a ¹⁴C activity of 0 pMC and δ¹³C of 0‰, following Gardner and Heilweil (2014) transit times models of waters in the same greater aquifer system (Kennedy and Genereaux 2007). The model developed by Han and Plummer (2013) as an update to the Fontes and Garnier (1979) model uses these inputs to estimate the contribution of carbonate rock dissolution, leading Lerback et al. 2019) to calculate an apparent transit time of 5600 years. In comparison, the International Atomic Energy Agency model (Gonfiantini 1972) focuses on the potential mixing of different flowpaths, which Lerback et al. 2019) used to calculate an apparent transit time of 10,800 years, which aligns with their modelled transit times of 11,000 in a particle simulation using the regional steady-state numerical groundwater model (Brooks et al. 2014). The difference between models is consistent with the observed model difference reviewed by Cartwright et al. (2020). Using these BL-Spring parameters and the Han and Plummer (2016) model through NETPATH, a 0.27 pMC (the average Δ ¹⁴ C _s-gw ) change to measured ¹⁴C the modeled transit time by approximately 250 years, whereas the difference between models was several thousand years (Plummer et al. 1994; Parkhurst et al. 2008; El-Kadi et al. 2010; Lerback et al. 2019).

This study builds upon three studies which touch upon this idea of shells as groundwater transit time proxies: Riggs’ (1984), Brennan and Quade (1997), and Copeland et al. (2012). To directly compare our data with that of Riggs (1984), we calculate that their three sites reported Δ ¹⁴ C _s-gw of 0.6, 1.0, and 4.2 pMC, where the latter (higher) Δ ¹⁴ C _s-gw is associated with some potential water reequilibration with atmosphere. Building upon these initial measurements, this study reports shell and water samples collected at the same time, adds an understanding of seasonal springwater variability, and investigate the change in Δ ¹⁴ C _s-gw as distance increases from the discharge source.

In addition, Brennan and Quade (1997) apply the Δ ¹⁴ C _s-gw proxy to the fossil record in Nevada, USA. This paper reports the ¹⁴C-ages in years B.P. rather than ¹⁴C activities as pMC or fraction modern C, and as such, they report aquatic shell ¹⁴C-age deficiencies (Δt) of 0–10,000 years (recognize that these ages would need other corrections for groundwater transit times as is described previously). Visual interpretation from their Figure 3 shows that this is a ¹⁴C activity depletion of 0–4 pMC. While Brennan and Quade (1997) discuss the impact of springwater reequilibration with air as a function of springwater circulation or morphology, they do not assess this impact or validate this Δ ¹⁴ C _s-gw proxy in modern settings. They note that to apply this proxy to the fossil record it is important to select shells without evidence of overgrowth and diagenesis to account for potential post-depositional C-exchange, and they also emphasize the importance of reliable depositional age controls.

Copeland et al. (2012) suggest the Δ ¹⁴C _s-gw proxy to explain the apparent ¹⁴C depletion of shells in the archeological record at a cultural site in Sonora, Mexico. They identify an approximately 2000 ¹⁴C-year lag for the aquatic shells (Planorbella sp.). Their data provided for Stratigraphic Unit B3 has an average charcoal ¹⁴C activity of 67.7 pMC (n = 3) and two aquatic shells to have 51.2 pMC and 55.8 pMC. Stratigraphic Unit B5 has an average charcoal ¹⁴C activity of 79.1 pMC (n = 8) and the co-occurring aquatic shell to have 58.1 pMC. Copeland et al. (2012) explain that the depletion is likely due to surface water chemistry and suggest that it may in part be due to groundwater depletion (citing Brennan and Quade 1997), but do not describe the depositional environment as directly groundwater-fed, recognizing that it might be an anthropogenic, managed wetland environment.

In the application of this proxy, researchers should consider the following site characteristics: (1) Morphology: the spring should have a pool directly fed by groundwater discharge. The pool shape and circulation should contain carbonate shells that live in springs whose flow rate is such that the advective transport of C is greater than the downward diffusion of atmospheric C, or flowing water such that the water that has not re-equilibrated with air. (2) Temporality: these springs should be persistent over several months (hillside springs that go dry seasonally may not harbor aquatic snails). (3) C inputs: the pond recognizably spring-fed without significant water inputs from rain, ephemeral streams, rivers or the ocean. This is most likely in arid regions without storm runoff or river flooding that might carry modern carbon into the pool.

In sum, we test the hypothesis that in exclusively groundwater-fed spring systems, water ¹⁴C is preserved in carbonate shells. Water, sediment, and shells of benthic gastropods (Melanoides, Pyrgulopsis, and Physella) from three sites in Utah were analyzed for ¹⁴C. We show that water (n = 6) and shell (n = 22) ¹⁴C activities at each site are correlated (R² = 0.999). These results support the hypothesis that ¹⁴C from groundwater is preserved in bicarbonate shells, and future studies should expand upon potential species- or site- specific vital effects.