Storage and distribution of organic carbon in cave sediments: examples from two caves in the northern karst region of Puerto Rico

The clastic sediments that accumulate in cave settings can be an important storage reservoir for organic carbon. This study reports on grain size, total organic carbon (TOC) concentrations, and total organic carbon: total organic nitrogen (TOC:TON) ratios measured in sediments from two caves in Puerto Rico. El Tallonal Cave (TAL) is a small cave with a flowing stream; the sediments in TAL were collected from a deposit that is being actively eroded. Clara Cave (CAM) is an upper level of the Río Camuy Cave System; the sediments from CAM were newly deposited by an internal river that rose in response to Hurricane Maria. Sediments collected from both caves were poorly sorted and contained no apparent stratigraphic correlation. CAM sediments contained a larger range in TOC concentrations but were overall lower than TOC in the TAL sediments. In TAL, the TOC concentrations were higher in sediments collected from below the erosional terrace. TOC:TON ratios from sediments at both caves were highly variable, highlighting the heterogeneous deposition and storage of organic matter. Despite the observed variation, TOC concentrations in both cave systems could cause retardation of organic contaminants by up to two orders-of-magnitude, implying that deposited sediments influence the fate of organic contaminants in the groundwater; therefore, cave sediments could facilitate long term storage of organic carbon and associated contaminants.


Introduction
The information stored in clastic cave sediments has been poorly explored relative to the more commonly studied chemical precipitates. Clastic cave sediments have the potential to provide evidence relevant to paleoclimate (Panno et al. 2004;White 2007;van Hengstum et al. 2010), elemental cycling (Vesper and White 2003;Hartland et al. 2012), subterranean ecosystems (Birdwell and Engel 2010;Pipan and Culver 2013;Husic et al. 2017;de Paula et al. 2020), paleohydrology (Granger et al. 2001;McCloskey and Keller 2009;Chess et al. 2010;Polk et al. 2013), and contaminant storage (Vesper and White 2004;Mahler et al. 2007;Torres et al. 2018). Organic carbon is a controlling factor in many of these processes; however, organic carbon concentrations in clastic cave deposits are generally not reported. While spatiotemporal heterogeneity in flow regime, geomorphology, and water chemistry adds complexity, deposited cave sediments experience consistent environmental conditions compared to subaerial fluvial systems (Bull 1981;de Paula et al. 2020). This property makes cave sediments valuable environments to explore organic matter dynamics between systems with differing hydrogeology.
Clastic cave sediments include all materials not precipitated in-situ. These may include detrital matter derived from the weathering of the host rock or surface-derived material (White 1988;Bosch and White 2007). The latter is generally considered to be the dominant source of clastic sediments and its introduction may be associated with collapsed structures, such as sinkholes, in response to major storms, or associated with more global climate changes. Surface-derived sediments are transported into a system via gravity or associated with incoming recharge flow (Hart and Schurger 2005). These surface-derived sediments are injected and subsequently deposited in the subsurface, 1 3 36 Page 2 of 14 where they are relatively protected from the erosive processes occurring at the surface (e.g., aeolian processes and freeze thaw cycles). During speleogenesis, a cave system may undergo cycles of sediment accumulation and erosion in response to external forcing due to climatic events or, on a longer time scale, glacial and inter-glacial cycles (Anthony and Granger 2007;Farrant and Smart 2011;Asanidze et al. 2017). Geomorphological and land use changes may also alter the types and volumes of input from the surface.
Sedimentary deposits within fluvial systems are a known carbon sink, storing organic carbon introduced from upland areas (Martínez-Mena et al. 2019). Cave systems are the subterranean equivalent of river systems, where materials from upland areas are transported and deposited within a natural sink (Ford and Williams 2007;Nichols 2009). While the sediment carrying capacity of cave systems is significantly less (with some exceptions) than that of fluvial systems, sediment transport is sustained for longer periods of time after a hydrologic event (Husic et al. 2017). Recent works have shown that organic carbon concentrations associated with suspended sediment are a potentially significant pool of carbon in cave systems (McCarthy and McKay 2004;Cruz et al. 2005;Simon et al. 2007;Ban et al. 2008;Hartland et al. 2012;Husic et al. 2017). Deposited sediments are found in nearly all cave systems, and while suspended sediment as a carbon source has been explored, organic carbon associated with deposited cave sediments remains largely unreported.
The sediments in this study were collected from the northern karst region of Puerto Rico, where abundant groundwater resources have promoted industrial and urban development, resulting in 25 Superfund sites on the EPA National Priority List (Padilla et al. 2011). Groundwater contamination in this region of Puerto Rico includes chlorinated volatile organic compounds (CVOCs), pesticides, and heavy metals. Although concentrations have decreased over time, contamination has persisted for more than 40 years (Yu et al. 2015;Torres et al. 2019). The organic matter concentration in a sediment is one of the primary controls on how organic contaminants are partitioned between the sorbed and dissolved phase (Schwarzenbach et al. 2016). Storage of sediment organic carbon within a karst system has important implications for contaminant transport and storage. Extreme weather events, such as hurricanes, could mobilize previously deposited subsurface sediments and stored contamination, resulting in potential human exposure to harmful compounds. The organic carbon data from cave systems in Puerto Rico reported in this study will help predict the long-term contaminant fate and support potential future remediation efforts.
In this study, organic carbon concentrations were obtained for clastic sediments from two caves in Puerto Rico with different surface connections and hydrological properties. The goal of this study was to determine the range and distribution of organic carbon in cave sediments along with potential correlations between carbon and sedimentological or hydrological properties. Paired nitrogen data were obtained to further characterize the sediment organic carbon using total organic carbon (TOC) and total organic nitrogen (TON) ratios. In addition, sediment organic carbon data collected in this study were then compared to previously published relevant data sets including cave sediments, cave water, tropical soils, and tropical fluvial sediments. Finally, the TOC data collected were used to estimate the potential increase in retardation of organic contaminants found in Puerto Rico's groundwater.

Sample locations and collection
Sediments were collected from two caves located in the northern karst region of Puerto Rico: Tallonal (TAL) Cave and Clara (CAM) Cave ( Fig. 1a) in February 2019. This karst region is eogenetic, having undergone little to no postdepositional deformation (Giusti 1978). A total of 13 cores from TAL and 9 cores from CAM were collected. The sediment cores were subsampled for analysis, resulting in 94 and 59 individual sediment samples from TAL and CAM caves, respectively.
TAL Cave is comprised of one main sinusoid passage, approximately 60 m in length, with no branching side passages, no known upper or lower levels, and a continuously flowing cave stream (Fig. 1b). Water enters the cave from an upward-flowing sump spring in the back of the cave and flows downstream to the cave entrance (Fig. 2a). A datalogger installed at TAL shows that turbidity coincides with most rain events (F. Pantoja-Agred, pers. comm., January 5, 2016). The baseflow level in the cave is highest during the tropical wet season (April-November). A dam at the cave entrance supports a private water supply and regulates the water level and flow in the cave. The dam is periodically opened to flush the system, manage storm surges, or to allow access to the cave. There is a significant increase in water level at the dammed portal in response to heavy rains; however, the exact water level fluctuation is unknown. The onset of turbidity coupled with the presence of sediment deposits within ceiling crevasses indicates TAL cave is currently undergoing active erosion.
Samples were collected from sediment banks located approximately 25 m from the entrance of TAL Cave (Fig. 2a). The sediment bank in this room contains an erosional terrace which has been interpreted as the baseflow water level when the dam is closed (Fig. 2a). To compare between samples, the ceiling height in the room was defined as an arbitrary datum and all sediments were given a vertical location relative to that datum (Fig. 2b). Based on the height of the terrace, the dammed water level in the cave is 137 cm below datum. Samples collected below the terrace are considered saturated (typically underwater); those collected above the terrace level are considered unsaturated (typically above water, Fig. 2b). The categorization of sediments between saturated and unsaturated conditions is based on the dam height and the inferred erosional terrace and is, therefore, not an exact distinction. Furthermore, it is likely to vary during storms and over longer time periods. Small (< 2 cm thick) speleothem deposits were observed on top of the sediment banks in TAL suggesting that the sediment deposit is not new. Because the angle of the core samples relative to the bank could not be consistently maintained, the sediment cores are not stratigraphically consistent and an accurate correlation between the cores could not be determined; thus is not included in the assessment (Fig. 2c).
The second sampling location, CAM, is operated by a National Parks Company and is part of a natural protected area designated by the Commonwealth of Puerto Rico (Fig. 1c). CAM sample location is a dry cave due to the vertical abandonment of the Río Camuy River, a subterranean river within the northern karst region of Puerto Rico (Fig. 1c). CAM is a toured portion of the Camuy Cave system. The Río Camuy Cave system contains multiple drycave levels that were left behind as the Río Camuy River baseflow lowered and diverted deeper underground (Miller 2009). The Río Camuy River is currently at the base of a large sinkhole approximately 20 m below the CAM sample locations (Fig. 1c). Cave sediments from the uppermost levels of the Río Camuy Cave system, dated using cosmogenic Fig. 1 a Puerto Rico's major carbonate-karst areas with the locations of the two caves included in this study. b Aerial imagery and regional topography surrounding El Tallonal Cave (TAL) with the plan view of TAL included. c Aerial imagery and topography surrounding Clara Cave (CAM) with the plan view of CAM included. The Río Camuy River is also included. The point in which the river enters the subsurface is represented by an "X". Dashed line represents subsurface flow of the Río Camuy River. The location in which the Río Camuy River flooded CAM is labeled. Sampling locations for CAM are also included Observed water level (darker blue) and dammed water level (lighter blue) are at 218 cm and 137 cm, respectively. c The spatial distribution of sample locations 16 and 17. Sediment cores were collected in a diagonal transect along both sediment banks. Both observed and dammed water level is shown in (c) isotopes ( 10 Be and 26 Al), have been in place for up to 4.5 million years (Miller et al. 2017).
On September 20, 2017, Hurricane Maria made landfall on Puerto Rico. The storm surge combined with a tidal swell produced a maximum surge of 1.5 m along the northern coast. Over 2.3 m of rainfall fell on the island in just over 32 h (Pasch et al. 2017). The Río Camuy River level rose at least 20 m as a result of Hurricane Maria flooding the generally dry room (Miller 2018), where the sediment samples were collected. Once the river returned to baseflow after the storm, approximately 0.4 m of sediment were deposited throughout the cave. The new sediment was clearly identified, because it accumulated on the paved walkways used for the public tour after Hurricane Maria. The 20 m rise was determined because the river, located approximately 20 m below CAM, is believed to be the source of the new storm sediments. There is no perennial flowing water source coming into or within the cave that could account for the newly deposited sediment. The samples were collected from 10 locations in the main chamber of CAM (Fig. 1c).
While the relationship between sediment input, deposition, and erosion in response to a rain event remains unknown in TAL Cave, there is still a clear hydrological difference between these two cave systems. TAL Cave contains a continuously flowing stream which experiences variable discharge accompanied by the onset of turbidity, whereas CAM Cave is consistently dry and presumably experiences sediment deposition only under exceptionally high flow conditions as it did during Hurricane Maria.

Sample preparation and laboratory methods
Core samples were collected by pushing a 5 × 30.5 cm polyethylene core sleeve into the sediment. After collection, the cores were capped, wrapped, refrigerated, and kept from direct light until analysis. In the laboratory, the core sleeves were split in half with a straight edge blade exposing two clean core faces for subsample collection. Subsamples were collected for analysis from 1 ± 0.5 cm sections of the core at depths in which visible change had occurred (e.g., color, grain size, or density changes). Subsamples for particle size analysis were prepared using a 1:1.5 sediment to 5% Calgon ® solution mass ratio to minimize clumping, then placed on a rotary shaker overnight at 70 rpm. Sediment slurries were analyzed using a Beckman Coulter single wavelength LS13-320 particle size analyzer that measures sizes between 0.4 µm and 2000 µm. Particle size is reported as a volume percent. Raw data files from the instrument were organized via R and processed through the GRADISTAT program (Blott and Pye 2001).
Subsamples collected for carbon and nitrogen were homogenized, air dried for 24 h, and subsequently oven dried at 60 °C for another 24 h. Total carbon (TC) and total nitrogen (TN) were determined using a Carlo Erba NA1500 CNHS elemental analyzer at the University of Florida's Stable Isotope Mass Spectroscopy Laboratory. Total inorganic carbon (TIC) was determined by acidifying the sediment under an N 2 blanket. Evolved CO 2 resulting from the acidification was then quantified coulometrically using a UIC 5017 CO 2 coulometer. Total organic carbon (TOC) was determined as the difference between TC and TIC.

Physical properties
Apparent layering observed in TAL could not be accounted for by changes in grain size (Fig. 3). Following the Folk and Ward (1957) devised sorting scale (unitless), sediments collected from TAL were categorized as poorly to extremely poorly sorted (poorly sorted: 1.0-2.0; very poorly sorted: 2.0-4.0; extremely poorly sorted: > 4.0). TAL sediments were, on average, categorized as extremely poorly sorted (average 4.3), whereas CAM sediments were very poorly sorted (average 3.2). Due to the poorly sorted nature of the sediments, proportions of sand, silt, and clay are referenced rather than mean grain size. Sediments from both cave locations are dominantly fine sand to coarse silt sized grains. Proportion of sand from TAL sediments (average 50.2%) was significantly higher than CAM sediments (average 38.9%; p value < 0.05). No significant differences (p value < 0.05) in grain size were observed between saturated and unsaturated sediments from TAL location 16 (Fig. 4).
Discontinuous thin banding of a black material was present throughout core TAL 16-04 (Fig. 3b). This core was in contact with the water table and experienced saturation in response to fluctuations in stream water level (Fig. 2). Preliminary XRF data were collected on a select number of grab samples for semi quantitative characterization using the mudrock calibration created by Rowe et al. (2012). This preliminary analysis showed elevated levels of Mn and Fe (molar units, > 2σ of the mean), common redox sensitive oxide forming elements, in sediments from core 16-04 (Downey 2020). Precipitation of Fe or Mn oxides could be a possible explanation for the black banding present in core 16-04 (Fig. 3c).
The remaining cores collected from location 16 were fully saturated (Fig. 2b). Variable color zonation was present in all saturated cores (Fig. 3c). These zones ranged in size and the contacts between zones were variable in shape and clarity. Color within these zones ranged from a dark brown or dark green hue to lighter tan (Fig. 3c). All samples collected from location 17 were saturated and showed similar color variation to the saturated cores collected from location 16 (see supplementary Fig. S1). Sediment cores were collected throughout CAM (Fig. 1c). In general, CAM sediments contained more distinct sedimentary layering than those collected from TAL (see supplementary Fig. S2). The CAM cores varied significantly between sample locations with no observable stratigraphic correlation. Large desiccation cracks covered the sediment surface throughout CAM, suggesting that the sediments were not saturated after their initial deposition via Hurricane Maria.

Chemical properties
The concentrations of TOC (wt%) ranged from 0.13 to 0.73% (average 0.33, n = 23) in unsaturated TAL samples and from 0.11 to 2.36% (average 0.96%, n = 71) in saturated TAL samples (Fig. 5a). TOC concentrations ranged from below detection to 3.43% (average 0.42%, n = 59) in CAM sediment samples. For the purpose of this study, we included non-detectable concentrations as having values of 0 in the calculation of averages. This may result in concentrations that are biased low but has been consistently applied for all averages. Of the 153 samples collected in total, only 2 of those contained no measurable TOC so this simplification does not have a major impact on the outcome. Combined TAL saturated and unsaturated sediment TOC concentrations were, on average (average = 0.8%), twice as high as CAM sediments (Fig. 5a). TOC comprised most of the total carbon in samples from TAL (average 83%) with a strong linear relationship (R 2 = 0.97; p value < 0.05) between TOC and TC (Fig. 5b). There was no relationship between TOC concentration and proportion of sub-silt sized grains at either location 16 or 17 in TAL sediments or CAM sediments (R 2 = 0.02; p value < 0.05). TOC made up a lesser proportion of the total carbon in CAM (average 70%). Of the 59 samples analyzed from CAM, 19 contained less than 40% of the carbon in the organic form (Fig. 5b). Whereas, of the 94 samples collected from TAL, only one sample contained less than 40% of carbon in the organic form (Fig. 5b).
Concentrations of TOC were significantly higher in saturated sediments from TAL (p value < 0.01) relative to either unsaturated TAL or CAM samples (Fig. 5a). TOC concentrations from TAL sediments, when plotted against depth relative to the datum, reveal a sharp increase in TOC concentration at the inferred stream level (Fig. 6). The sharp increase in TOC concentration between unsaturated and saturated sediments present in TAL sediments could not be evaluated in the CAM sediments, because those sediments were not saturated after their initial deposition. The TOC variation between unsaturated and saturated TAL sediments suggests that the presence or absence of water in the sediments plays a role in carbon storage and/or carbon generation after initial input into the system.
Complementary TN data collected for all sediments provides a framework for initial sediment organic carbon characterization. A strong linear relationship exists between TOC and TN in both saturated (R 2 = 0.82; p value < 0.05) and unsaturated (R 2 = 0.65; p value < 0.05) TAL sediments. Following the method of Goni et al. (1998), total inorganic nitrogen (TIN) can be estimated from TOC by applying a linear regression to a TN vs. TOC plot. The concentration of TIN (C TIN ) is mathematically represented by the y-intercept. C TIN was estimated to be 0.036 (C TIN = 0.08C TOC + 0.036) and 0.042 (C TIN = 0.10C TOC + 0.042) for saturated and unsaturated sediments from TAL, respectively. Like TAL sediments, TOC and TN were strongly correlated (R 2 = 0.77; p value < 0.05) in sediments from CAM. However, the y-intercept was lower (C TIN = 0.10C TOC + 0.015). This reveals that CAM sediments contained less TIN than TAL sediments.
By subtracting the estimated TIN from TN, total organic nitrogen (TON) can also be estimated (C TON = C TN -C TIN ). From this, TOC:TON molar ratios were calculated. Average TOC:TON ratios were 17.7 (standard deviation, SD = 13.4) and 13.7 (SD = 10.1) for saturated and unsaturated TAL sediments, respectively (Fig. 7). Average TOC:TON from sediments collected at CAM was 15.9 (SD = 12.4). The high SD calculated from the TOC:TON ratios from each cave location hints at the heterogeneous nature of natural organic matter. Despite this, average TOC:TON values are still informative for initial organic matter characterization and comparison to standard types of natural organic matter. Humic and fulvic acids, a major component of natural organic matter, average C:N ratios are 18.0 and 20.4, respectively (Rice and MacCarthy 1991;Essington 2015). Amino acids, the main component of non-humic-like natural organic matter, has a lower average C:N ratio at 3.15 (Jover et al. 2014). TOC:TON ratios collected from sediments at both locations show ratios within the range of humic and fulvic acids, independent of TOC concentration (Fig. 7). Highest TOC:TON values were found in cave sediments with lower TOC concentrations at both TAL and CAM locations. TOC:TON ratio values consistent with amino acids were observed only in CAM sediments.

Variability between cave systems
Sediment mobility within a conduit is dependent on cave morphology, groundwater flow, the magnitude of rain events, and flow velocity; most karst systems are likely to contain both gradual and threshold-based changes. For example, Herman et al. (2008) reported on temporal trends in sediment transport to a spring in response to precipitation events and found that a hurricane event triggered a large, punctuated increase in sediment transport to same spring. The process was interpreted as a threshold event in which flow conditions were significantly altered. The hydrogeology of the two cave systems included in this study are very different. TAL is a wet cave with a continuously flowing cave stream; in contrast, CAM does not have flowing or perennial water in the upper (older) level of the Río Camuy system, where samples were collected (Miller et al. 2017). The cave stream flowing through TAL becomes turbid during most rain events, whereas sediment input at CAM occurs only when the Río Camuy River rises to reach the dry cave, as it did during Hurricane Maria (Fig. 1c). While the cave stream in TAL frequently carries suspended sediment, the relationship between turbidity, deposition, and remobilization remains under investigation. Stalagmites (< 2 cm thick) coating portions of the sampled sediment banks was observed, indicating the deposited sediments have been in place at TAL for an extended time. The sediments collected from CAM were known to be deposited by flooding via the Río Camuy River resulting from Hurricane Maria. These sediments have not been reworked since deposition and were in place approximately 18 months at the time of sampling. TOC variation between the two caves reflects the differences in sediment sources, input dynamics, and post depositional conditions.
The heterogeneity of a sedimentary environment facilitates the presence of microenvironments leading to variations in limiting nutrients and electron acceptors. Work by Schlüter et al. (2018) showed that conditions change from anoxic to oxic surrounding a sediment particle in just over 0.4 mm. Redox zonation results from variation in oxygen availability and the thermodynamic availability of electron acceptors in an environment. The semi quantitative elemental XRF analysis on TAL sediments suggests that the black banding present in TAL 16-04 is possibly microcrystalline Mn oxide banding (Downey 2020). This is likely from fluctuation in water level as Mn oxides rapidly precipitate when environments transition from suboxic to oxic (Dixon et al. 1990) (Fig. 3b). The electromotive potential of a soil has been correlated with the characteristics of the pore water solution within a sedimentary environment (Wanzek et al. 2018). Chess et al. (2010) described sediment facies from Butler Cave, VA as "low and wet" and "high and dry" based on color variation attributed to varying degrees of hydration of iron oxides. Sediment saturation in TAL Cave could enhance the redox conditions within the sediments resulting in precipitation reactions causing the colored zonation observed in nearly all saturated sediment cores (Fig. 3c).
Organic carbon was significantly higher in saturated sediments compared to unsaturated sediments in TAL (Fig. 6). Work by Cruz et al. (2005) found that organic carbon concentrations were highest in more reducing conditions, linking organic carbon to redox conditions in the subsurface. In addition to facilitating enhanced electromotive potential, pore water is the principal mode of transport for materials into and out of these sediments. The exchange and transport of surface derived organic matter, including DOC, is likely occurring between the water column and the sediment pore solution. The groundwater transport of freshly derived organic matter into sediment pore water could facilitate long term storage and subsequent transformation and/or mineralization. The data reported in this study suggest that sediment saturation plays an important role in carbon storage within the subsurface likely through a combination of pore water DOC, flux in and out of the sediments, organic carbon adsorption to mineral surfaces, and limited mineralization due to less oxidizing conditions (when compared to unsaturated sediments).
TOC:TON ratios differed between cave systems. TOC:TON ratios consistent with amino acids were found only in CAM sediments. Amino acids typically represent protein-like organic matter when evaluating spectrographic data and has historically been interpreted as a microbial signature (Coble 1996). Saturated TAL sediments contained slightly higher TOC:TON values (average, 17.7), closer to a humic acid-like signature. This is, however, not evidence that TAL lacks a microbial signature. The exact structure and composition of natural organic matter is complex and not well-defined. Therefore, categorizing the organic matter within these sediments has humic acid-like, fulvic acidlike, or amino acid-like based on TOC:TON ratio values are shown in Fig. 7 as an exploratory comparison to accepted values for various known types of organic carbon. These data are provided as a framework for comparison. Moreover, the calculated TOC:TON ratios from both caves had relatively large standard deviations (Fig. 7). The high variability in TOC:TON ratios in sediments collected from the same cave, some only centimeters apart is likely due to the complex and heterogeneous nature of natural organic carbon.

Comparison of organic carbon concentrations to similar sites
Despite the complexity of sedimentary organic matter, TOC:TON is a widely reported parameter and the analyzed cave sediments in this study enables a useful preliminary comparison to other soil and sediment data sets (Table 1) (Table 1). TOC:TON ratios from cave sediments reported by Fichez (1990) were lower than TAL and CAM sediments. This is, to our knowledge, the only study on cave sediments that report TOC:TON values for comparison.
Overall, there is a paucity of data for organic carbon in cave sediments. To provide a framework for better understanding the data in this study, the TAL and CAM carbon concentrations were compared with available data from other cave sediments from various locations plus data from tropical surface sediments (Table 1; Fig. 8). Of the five studies that report cave sediment organic carbon compiled for comparison, three were collected from deposited cave sediments and the remaining two report sediment organic carbon associated with suspended sediment from a cave stream (Table 1). Studies reporting on suspended cave sediments reveal higher overall TOC concentrations than those reporting on deposited cave sediments. Husic et al. (2017) demonstrated that organic carbon associated with suspended sediments decreased from input to output through a cave stream suggesting some form of retention within the system (Fig. 8; Table 1). De Paula et al. (2020) reported deposited cave sediment organic carbon and found that higher concentrations occurred during the tropical wet seasons from a cave in Brazil. CAM sediments had the largest range and lowest concentration of TOC. This is also the only cave in the compiled data set known to be dominantly dry year-round. This, coupled with higher TOC values associated with suspended cave sediments (Table 1; Fig. 8) corroborates the idea that sediment saturation is an important control on organic carbon deposition and subsequent storage within cave systems.
DOC concentrations within cave streams from a variety of locations have also been reported and are included for comparison (Table 1; Fig. 8). DOC concentrations within these cave streams are comparable to the compiled suspended sediment organic carbon data. Values of DOC reported from cave water are only slightly higher than TOC sediment concentrations collected in this study and reported in the literature (Table 1: lines 1-4). The cave sediment data compiled here lead to the conclusion that TOC associated with sediments are a significant additional pool of organic carbon within the subsurface and should be considered when quantifying carbon fate and transport in the subsurface.
Surface soils are an important sink for carbon storage (Essington 2015). Tropical climates promote abundant vegetation and rapid soil carbon turnover (Sayer et al. 2019). Work by Moyer et al. (2013) analyzed small mountain streams throughout Puerto Rico and found that organic matter is rapidly mobilized and subsequently stored within the coastal floodplains due to the abundant rainfall occurring on the island. Due to the efficient surface to groundwater transfer occurring within the Puerto Rico karst region, it is assumed that some of this sediment and associated organic carbon will make it into the groundwater system. The TOC values reported in this study were compared to various tropical sediments and soils, including from Puerto Rico ( Fig. 8; Table 1). River POC and soil pore water DOC collected from central Puerto Rico were significantly lower than any of the reported cave sediment concentrations (number 14 and 15 in Fig. 8). While only a few studies from surface sediments and soils are reported here for comparison, cave sediment TOC concentrations are within range of these values, further illustrating the importance of organic carbon in the subsurface.

Implications for contaminant fate and transport
The organic carbon associated with deposited sediments has important implications for contaminant fate and transport in a karst aquifer. The organic carbon-water partition coefficient (K iocw ) is used to quantify the affinity of a given chemical, i, for natural organic matter (Schwarzenbach et al. 2016). Due to the complex, apolar structure of most natural organic matter, it often exhibits a high sorption affinity for most aromatic organic contaminants. Consequently, the solid-water partition coefficient (K id ) of a chemical, i, is proportional to the concentration of organic carbon within the sedimentary environment ( Table 2). The presence of fine-grained sediments and natural organic matter in a groundwater flow path with known organic contamination increases the possibility of retention or storage through sorption. The retention of an organic contaminant is often quantified by calculating the ratio of the average linear velocity of groundwater to the velocity of the contaminant, referred to as a retardation factor. Higher retardation factors suggest that the transport of a chemical will be slower than groundwater flow.
The high solid-water partition coefficients, controlled by the concentration of natural organic matter, lead to higher retardation factors ( Table 2). Concentrations of TCE, CCl 4 , and PCE are among the highest detected contaminants in Puerto Rico's groundwater (Torres et al. 2019). The fraction of organic carbon (f oc ) within the cave sediments collected in this study ranged from 0.0002 to 0.0236. The presence of f oc = 0.0236 (maximum found in this study) would increase the retardation of TCE, CCl 4 , and PCE by 582%, 608%, and 545%, respectively, when compared to the minimum f oc found in this study (Table 2). Natural organic matter where C is is the concentration of compound i sorbed to a solid and C iw is the concentration of Compound i in the dissolved phase; data compiled from EPA (1996)  associated with sediments within an aquifer may account for significant sorption and storage of organic compounds in groundwater; this sorption has been cited as a possible reason for the persistence of contamination within a karst aquifer (Padilla and Vesper 2018;Torres et al. 2019). Work by Lin et al. (2020) found that concentrations of CVOC contaminants in Puerto Rico's drinking water increased significantly following Hurricane Maria. While sorption removes contaminants from the water column, large storm events such as Hurricane Maria will result in severe flooding potentially mobilizing previously deposited sediments and any associated sorbed contaminants. These mobilized sediments act as a vector for contaminant transport, potentially entering public or private water supplies (Mahler et al. 2007).

Conclusions
Organic carbon associated with suspended sediment within a cave system as a potentially significant pool of organic carbon has been studied; however, deposited cave sediments as an organic carbon pool has remained largely unexplored. The results from this study show that concentrations of organic carbon within deposited cave sediments are within the range of organic carbon associated with suspended cave sediments or DOC within the groundwater. Sediments and associated organic carbon from CAM were deposited recently via Hurricane Maria and have not been reworked since, whereas the TAL cave stream experiences turbidity during most rain events. Organic carbon concentrations varied significantly (p value < 0.05) between cave systems. The range in organic carbon concentrations within TAL sediment was lower than CAM sediments but contained on average higher concentrations of organic carbon. These differences are likely controlled by many factors, including but not limited to, post-depositional metabolic activity, redox, ambient environmental conditions, and time. The TOC:TON ratios of the collected cave sediments are consistent with terrestrial materials and comparable to TOC:TON ratios of sediments found in the central highlands of Puerto Rico.
Significant differences between unsaturated and saturated sediments from TAL were observed. It is highly likely that the colored zonation present within saturated TAL sediments is a result of changing redox conditions facilitating precipitation of some redox sensitive minerals. In addition to the colored zonation, organic carbon was significantly higher in TAL saturated sediments. Sediment pore solution is the most dominant mode of transport for freshly input organic matter to enter a subsurface sedimentary environment. However, exact mechanism resulting in the higher concentrations of organic carbon in the saturated sediments from TAL remains under investigation.
It is important to remember that organic carbon concentrations from only two caves are reported here. Given the heterogeneity in karst systems, this is unlikely to provide a full range of conditions and carbon concentrations. However, this study does add to our knowledge about the distribution and range of TOC concentrations in these systems which provides important information to direct future work. The bulk organic carbon concentrations from the cave sediments reported in this study show that organic carbon associated with long term subsurface sedimentary environments is significant enough to warrant the inclusion of such deposits when attempting organic carbon flux calculations.
The presence of natural organic carbon in the subsurface is exceptionally important for contaminant transport. Retardation factors calculated using the minimum and maximum organic carbon concentrations collected from the sediments in this study reveals drastic increases in retardation of common contaminants found in Puerto Rico's groundwater. The sorption of organic contaminants onto the karst aquifer organic matter also contributes to the long-term storage of contamination in this system. This may play an important role for remedial design and for public health concerns in the region.