For all three Lilburn cave sites, air pCO₂, RH, and temperature remain constant throughout the year (Summary: 678 +/− 43 ppm, 100% humidity, 7.4°C; Supplementary Table S1). A typical period between when automated sampling devices were deployed and retrieved is shown in Supplementary Figure S2. Drip rates at BYHT were not recorded when sampling as the drip-water flowed continuously from the ceiling of the cave to the top of a flowstone. At the nearest site to the entrance, GC (Figure 1), the drip rate ranges from the highest recorded value (six drips/second) during March to May and decreases to the lowest values (<1 drips/second) throughout the summer and into the early winter months (June through December) (Figure 2). At CP (Figure 1), drip rates range from one to three drips/second, with the highest drip rates occurring between March to May (Figure 2) and the lowest during August through December. All stable isotope and trace element results are tabulated in Supplementary Table S2. Seasonally, the drip-water δ¹⁸O compositions are similar at GC and CP, with the highest values occurring during the autumn and early winter months (September–December). In contrast, BYHT drip-water δ¹⁸O compositions reach their lowest values in autumn and early winter (September–December) and rise to their highest values in the spring (March–May) (Figure 3). The single δ¹⁸O measurement of “farmed” calcite grown in situ at GC is -8.7‰ and the δ¹⁸O of calcite at CP ranges between −8.6 and −9.4‰ (±0.06%).

Drip-water δ¹⁸O and δ²H values for Lilburn Cave define a slope and intercept (δ²H = 6.5 *δ¹⁸O–3.4) that lie between those defined by precipitation compositions from SEKI (δ²H = 7.6* δ¹⁸O + 8.5), and of surface water from Redwood Canyon (δ²H = 5.4* δ¹⁸O–17.8) (Figure 4). We compare the slopes of drip-water δ¹⁸O and δ²H compositions to that of precipitation from SEKI for 2001 to 2011 (McCabe-Glynn et al., 2016) as well as to snow and surface water compositions from Redwood Canyon sampled during the period of cave monitoring (Figure 4). The slope for the drip water at Lilburn Cave falls between that of the SEKI precipitation and the Redwood Creek stream water. Drip-water compositions do appear to not represent precipitation or surface water alone.

Seasonally, the TE/Ca ratios for CP and BYHT decrease in the autumn, contemporaneously with the increase in autumn precipitation in SEKI (Figure 4). The decrease of drip-water TE/Ca ratios at BYHT in the autumn is contemporaneous with a decrease in the δ¹⁸O compositions. Alternatively, the decrease in TE/Ca ratios at CP occurs when δ¹⁸O compositions are increasing. The signatures at GC, however, reveal that drip-water TE/Ca ratios at this site increase in the autumn and mirror an increase in δ¹⁸O compositions (Figure 4). The TE/Ca ratios at the three sites each respond differently to the variability in SEKI precipitation. To further examine the source of this variability, we analyzed the relationship between Ba, Sr, and Mg using the approach of Sinclair (2011), Sinclair et al. (2012), and Casteel and Banner (2015). A cross-plot of ln (Mg/Ca), ln (Sr/Ca), and ln (Ba/Ca) ratios in the drip-water of each cave site exhibits calculated slopes that range from 0.8 to 1.1 (Figure 5). The Black Chasm drip water slope for Mg/Ca and Sr/Ca is 0.99 (Figure 5). Values of Ba/Ca were not measured for Black Chasm.

The TE concentrations from the calcite grown in situ at Lilburn Cave are comparable (taking into account K_Ds) to the drip-water concentrations at GC and CP. Sr, Mg, and Ba concentrations in the in situ precipitated calcite at the shallowest and furthest sites from the Meyer Entrance (GC and CP, respectively) are comparable to those of the drip-waters from these sites (Supplementary Table S2). The concentrations of Mg at these Lilburn Cave sites range from 4,230 to 4,456 ppb at CP and 23989 to 2,790 ppb at GC as compared to 1,092 ppb and 701 ppb in the calcite from CP and GC, respectively. The Sr concentrations in the drip-water at these sites range between 190 and 196 ppb at CP and 110–128 ppb at GC as compared to 191 and 183 in the farmed calcite from CP and GC, respectively. The concentrations of Ba at these drip sites range between 89 and 102 ppb at GC and between 74 and 83 ppb at CP as compared to 68 and 154 ppb in the farmed calcite from GC and CP, respectively.

The concentrations of less commonly applied trace metals (Cu²⁺, Mn²⁺, Fe³⁺) in the Lilburn drip waters were examined to determine their viability as a new geochemical signal of seasonal recharge in this cave. To evaluate these elements, we compared the seasonal trend in their concentrations in the drip water to the precipitation amount in SEKI. Mn²⁺ concentrations in the shallowest and intermediate depth sites (GC and BYHT) vary seasonally with minimum values (0.0–1.5 ppb) in the spring and summer that increase 2 to 6-fold in autumn (0–3.0 ppb) contemporaneously with an increase in precipitation amount (Figure 6). In contrast, the Mn²⁺ concentration at CP is more variable, although there is a rise in values contemporaneous with an increase in precipitation amount, comparable to the other two sites (Supplementary Table S2). Cu²⁺ concentrations at all three drip sites exhibit a similar seasonal trend to Mn²⁺ concentrations and increase in concentration in the drip-water synchronous with an increase in precipitation amount. One notable exception is that Cu²⁺ concentrations at all three drip sites double in August of 2019 relative to August 2018 concentrations (Figure 6). Fe³⁺ concentrations at all three drip sites are more variable with overall low values through 2018 (∼5–10 ppb for the bulk of the dataset) but increase overall (range of ∼10.0–60.0 ppb) in 2019. Concentrations of all three metals in the “farmed” calcite at CP (Mn = 4 ppb, Fe = 223 ppb, and Cu = 2.4 ppb) and at GC (Mn = 2 ppb, Fe = 66 ppb, Cu = 22 ppb) are 10-fold higher overall than drip-water concentrations (ranges listed above; Supplementary Table S2) accountable by their K_D values.

To further interrogate the Lilburn Cave monitoring results, we compare the data to a five-year cave monitoring dataset from Black Chasm, a cave from the foothills (elevation of 300 m) of the central Sierra Nevada (Oster et al., 2012). To summarize that study, 5 years of drip water monitoring at five sites in Black Chasm, as well as surface monitoring, revealed that Black Chasm has an average temperature of 12.3°C, with 5°C annual variability (Oster et al., 2012). The pCO₂ at Black Chasm ventilates seasonally when cold dense air displaces summer air in the autumn. The relative humidity at Black Chasm varies seasonally but remains >80% through most of the year. Seasonally, the δ¹⁸O of Black Chasm drip water varies in response to changes in precipitation δ¹⁸O, with a lag of ∼3 weeks, and the precipitation δ¹⁸O varies depending on precipitation source and temperature. The δ¹³C and TE/Ca results from Black Chasm reveal that seasonal variability in rainfall amount drives prior calcite precipitation (Oster et al., 2012).

The geochemical variability and drip rates at each drip site in Lilburn Cave reveal seasonal variability related to precipitation amount above the cave, and for GC and CP, airflow regimes in the cave (Figure 3). Drip rates at GC and CP increases in response to increased precipitation in the autumn and decreases in response to precipitation loss in the summer, albeit with a 3 months time lag. This relationship indicates that drip rates at these two sites respond to precipitation amount above the cave. Variability in Mg/Ca, Ba/Ca, and Sr/Ca ratios in cave drip waters has been attributed to the type and composition of bedrock, soil composition above the cave, the fluid flow-path through the epikarst (i.e., conduit vs. diffuse), and the change in degree of water-rock interaction driven by water residence time in the epikarst. Additionally, the extent of prior calcite precipitation controlled by both water amount and rate of CO₂ degassing impacts water TE concentrations, which may or may not be governed by the same processes controlling water availability, and the incongruent dissolution of dolomite and/or calcite (Fairchild et al., 2000; Fairchild and Treble, 2009; Sinclair, 2011; Wong et al., 2011; Oster et al., 2012; Sinclair et al., 2012; Casteel and Banner, 2015). Based on the aforementioned studies, when the TE/Ca ratios are controlled by water-rock interaction and precipitation amount, it is expected that TE/Ca ratios will increase during periods of lower precipitation amount and higher water-rock interaction and/or prior calcite precipitation. We interpret the seasonal trends in TE/Ca ratios at drip sites in Lilburn Cave as responding directly to prior calcite precipitation and/or water-rock interaction driven by local precipitation amount. At CP and GC, the trends in TE/Ca (see Sr/Ca in Figure 3) decrease as drip-rate increases as would be expected if the trend is driven by precipitation amount. The same relationship is also observed at BYHT, although no drip-rate information is available for this site. It is notable, however, that the change in TE/Ca ratios at the GC site lags several months behind the change in precipitation amount in the autumn; this lag is not observed at the CP and BYHT sites (Figures 2, 4) that both exhibit changes in TE/Ca ratios directly after the changes in precipitation amount in September and/or October.

Although it is useful to understand the seasonal relationship between precipitation amount, drip rate, and TE/Ca, it is not possible to use these trends to tease apart the separate processes of water-rock interaction (WRI) and prior calcite precipitation (PCP). This is because both can cause seasonal variability in TE/Ca values. To address this, we used a mathematical approach that isolates WRI from PCP and assesses the influence on drip-water Mg/Ca, Sr/Ca, and Ba/Ca values (Figure 5) (Sinclair, 2011; Sinclair et al., 2012; Casteel and Banner, 2015). Based on this method, the formula K_D-Sr–1/K_D-Mg -1 or K_D-Ba–1/K_D-Mg–1 is used to derive slopes for ln (Mg/Ca) vs ln (Sr/Ca) or ln (Mg/Ca) vs. ln (Ba/Ca) while replacing the K_D values for the appropriate TE/Ca ratio in question. These slopes and the TE data are subsequently used to produce plots that distinguish between PCP or incongruent calcite dissolution (ICD) and WRI in drip-water values (Figure 5) (Sinclair, 2011; Sinclair et al., 2012; Casteel and Banner, 2015). Based on these calculations, the TE/Ca ratios for Lilburn Cave and Black Chasm are driven by PCP or incongruent dolomite or incongruent calcite dissolution. Lilburn Cave is mostly developed in calcite marble, although dolomitic marble and limy dolomite marble have been observed (Tinsley et al., 1981). We cannot rule out the influence of incongruent dissolution of calcite or dolomite, by which trace elements (Mg, Sr, or Ba) are dissolved preferentially and result in increased TE/Ca ratios in the drip water. The observation that Mg, Sr, and Ba concentrations vary in the drip water in the same direction at each site (Figure 5) cannot be explained by variability in bedrock composition. We propose that the TE/Ca ratios in Lilburn Cave drip-waters primarily record PCP in the epikarst. Furthermore, the “farmed” calcite TE/Ca values indicate that the calcite directly records drip-water TE/Ca ratios (Supplementary Table S2), suggesting the high potential for Lilburn stalagmites to record temporal variability in precipitation amount in this cave.

Variability in the drip-water δ¹⁸O values at the three Lilburn sites further document the influence of seasonal variability in temperature and precipitation, and/or continuous CO₂ ventilation. Notably, the drip-water δ¹⁸O and δ²H values at all three sites are within the range of the precipitation and snow δ¹⁸O and δ²H in the region (Figure 4). The variability in precipitation δ¹⁸O and δ²H from 2001 to 2011 reflects the large temperature and source variability in storms that arrive at SEKI. Although the precipitation monitoring period does not overlap with the drip water monitoring period, we assume here that the McCabe-Glynn et al. (2016) dataset is representative of potential precipitation variability through the monitoring 2018 to 2021 in the same area. Alternatively, snow and drip-water δ¹⁸O and δ²H were not as variable during 2018–2021 in comparison to the period 2001–2011 (Figure 4). This is likely due to mixing of precipitation from different storm events in the soil, surface water, or epikarst zone. In comparison, drip-water δ¹⁸O and δ²H is much less variable than precipitation δ¹⁸O and δ²H at Black Chasm (Supplementary Figure S2; Oster et al., 2012), which is interpreted to be due to mixing in the epikarst.

To better understand the seasonal variability in drip-water δ¹⁸O (Figure 3), we modeled annually resolved time series for drip-water δ¹⁸O at Lilburn Cave using Karstolution (Treble et al., 2019) to determine the primary influence(s) on measured δ¹⁸O variability. Karstolution is a forward proxy model that includes measured precipitation δ¹⁸O, precipitation amount, evapotranspiration amount, and temperature in its parameterization. Furthermore, Karstolution considers evaporative enrichment of water from the soil zone, however, all other processes in Karstolution assume equilibrium fractionation when the RH is set to 95% humidity, a setting that is relevant for Lilburn Cave. Consequently, although calculated evaporative enrichment of the δ¹⁸O composition at the “modeled” surface of the cave, from the soil zone, may be impacting the drip-water compositions from Karstolution, in-cave kinetic effects are not incorporated into the modeled drip-water compositions in this analysis. The model applies a range of processes of mixing and storage and outputs a time series of drip-water δ¹⁸O compositions and five different time-series of calcite δ¹⁸O. To parameterize Karstolution, we used the 10-years record of precipitation δ¹⁸O composition and amount data from SEKI (McCabe-Glynn et al., 2016) and corresponding temperature and pan evaporation measurements from a nearby weather station (National Oceanic and Atmospheric Administration (NOAA), 2021). The modeled drip-water δ¹⁸O values from Karstolution are similar in composition to the measured δ¹⁸O in drip water from BYHT year-round and are similar to the measured δ¹⁸O of drip water from GC and CP in the spring months (Figure 7A). Accordingly, the comparison between the measured drip-water δ¹⁸O and the modeled drip-water δ¹⁸O indicate that the δ¹⁸O compositions at BYHT are representative of the variability in precipitation δ¹⁸O from 2001 to 2011 in SEKI, precipitation amount, temperature, and evaporation (Figure 7A). This is a significant result as this behavior suggests that out of the three drip sites monitored at Lilburn Cave, BYHT may be the most ideal to develop a speleothem paleoclimate record that captures precipitation variability in SEKI. In contrast, the δ¹⁸O compositions at GC and CP are experiencing seasonal enrichment (autumn and early winter) as demonstrated by the increased δ¹⁸O in drip water at these sites in comparison to the modeled δ¹⁸O compositions (Figure 7A). It is important to note that precipitation monitoring did not occur over Lilburn Cave during the monitoring period and that the increases at GC and CP could reflect variability in precipitation δ¹⁸O in 2018 and 2019 that was greater than the precipitation δ¹⁸O variability captured in the precipitation monitoring effort from 2001 to 2011. However, given that BYHT matches the modeled drip-water δ¹⁸O results closely, we posit that GC and CP are instead influenced by in-cave kinetic effects or dynamics such as increased airflow as they are near the two entrances of the cave (Figure 1). Others have found that increased ventilation near entrances potentially leads to variability in δ¹⁸O of calcite (Tremaine et al., 2011; Oster et al., 2012). The enrichment is due to decreased pCO₂ in well ventilated cave passageways that can lead to increase CO₂ degassing from drip water that rapidly changes the δ¹⁸O values of the dissolved inorganic carbonate (DIC) forcing the DIC pool out of equilibrium (Mickler et al., 2004; Mickler et al., 2006; Dreybrodt and Deininger, 2014; Carlson et al., 2020). Additionally, a decrease in cave air RH can lead to increased evaporation from the drip water, forcing the δ¹⁸O values of the water and the DIC pool out of equilibrium (Mickler et al., 2004). These two cave air parameters are the driving cause of is termed in-cave kinetic effects. Cave air pCO₂ and RH remained constant throughout the monitoring period, however, the pCO₂ in Lilburn Cave remained quite low (average of 678 ppm), potentially due to enhanced ventilation and may have led to the increased δ¹⁸O values observed at GC and CP. Although in-cave kinetic effects have been extensively studied (Mickler et al., 2004; Mickler et al., 2006; Riechelmann et al., 2013) they are mainly relegated as a cause for concern or as a reason to disregard a cave, region, or stalagmite sample. The seasonal signal at these two sites, however, can be reframed as a significant finding as the δ¹⁸O enrichment happens in the autumn season of both monitoring years. Future stalagmites taken from Lilburn Cave, given extensive monitoring to determine the relationship between δ¹⁸O, airflow, cave air parameters, and precipitation δ¹⁸O, could preserve a signal such as that observed at GC and CP that denotes seasonal variability in ventilation. Furthermore, the “farmed” calcite grown in situ at Lilburn Cave reveal calcite δ¹⁸O that are similar to the modeled calcite δ¹⁸O from Karstolution, but with an enrichment of ∼1% (Figure 6B), which would be expected if the δ¹⁸O are impacted by in-cave kinetic effects. Additionally, the calcite δ¹⁸O compositions from CP reveal an enrichment from the center of the watch glass to the edge of the calcite on the watch glass (Figure 7B) a finding consistent with the hypothesis proposed by many others that in-cave kinetic effects lead to variable δ¹⁸O of calcite away from the center of stalagmite growth (Hendy, 1971; Mickler et al., 2006; Tremaine et al., 2011; Riechelmann et al., 2013). Consequently, we propose that stalagmite paleoclimate records developed from these sites in the future be used as records of seasonal ventilation or airflow in the cave system but should be treated with caution as records of precipitation δ¹⁸O and temperature.

We further modeled drip-water δ¹⁸O for the central Sierran foothill cave, Black Chasm, using a five-year record of precipitation δ¹⁸O (Oster et al., 2012) and temperature and evaporation data from a corresponding weather station (National Oceanic and Atmospheric Administration (NOAA), 2021). The modeled drip-water δ¹⁸O for Black Chasm agrees with the measured drip-water δ¹⁸O from Black Chasm (Supplementary Figure S2) indicating that Black Chasm drip-water δ¹⁸O is not influenced by in-cave dynamics such as low RH or airflow. This is a significant result as it suggests that Black Chasm drip sites are ideal locations for developing stalagmite paleoclimate records of precipitation δ¹⁸O and temperature, as well as site BYHT from Lilburn Cave. Our interpretation is consistent with the original interpretation from Oster et al. (2012) that the cave hydrology, including drip rates, responds immediately (within hours to days) to large precipitation events (Oster et al., 2012). Dampened variability in the modeled drip-water δ¹⁸O compositions for Black Chasm (−10.5 ± 0.6%), relative to the measured values, is likely due to overrepresentation of mixing in the Karstolution model. The comparison of the modeled and measured drip-water δ¹⁸O compositions from Black Chasm and Lilburn Cave reveal that the seasonal enrichment experienced at two sites (GC and CP) and Lilburn Cave is not a regional phenomenon but is instead constrained to these two sites. It is important to note that Black Chasm and Lilburn Cave are separated by 2° of latitude and 4,000 ft of elevation. These differences, however, do not lead to observable difference in the overall variability in precipitation δ¹⁸O (Supplementary Figure S2). The elevation and latitude could explain the −1.5% difference in dripwater δ¹⁸O as this is consistent with a lapse rate of −0.35%/km. The explanation, however, could equally be a difference in time between the Black Chasm study and the study presented here. The cross-cave comparisons highlight the need to monitor unique cave environments.

The Lilburn Cave drip water compositions indicate that trace metal concentrations (Cu²⁺, Mn²⁺, and Fe³⁺) may be sensitive to hydrologic recharge in the cave system (Figure 7). The increase of Cu²⁺, Mn²⁺, and Fe³⁺ concentrations in the autumn suggests that Mn²⁺ is responding to the first available recharge to the system either through increase soil mobilization or through a wash through event by increased recharge in the cave or bedrock system. Additionally, we hypothesize that the change in concentrations of Cu²⁺ and Fe from 2018 to 2019 (Figure 7) is due to the region of SEKI experiencing a moderate drought in 2018 followed by a period of greater effective moisture in 2019 (PDSI, 2021). Other studies have demonstrated that these metals are transported and measured in drip waters under peak flow conditions due to the observations that humic-derived organic matter binds metals such as Cu²⁺, Mn²⁺, and Fe³⁺ and aides in their transport to the drip site (Hartland et al., 2011; Hartland et al., 2012). We propose that Mn²⁺, Cu²⁺, and Fe³⁺ concentrations in drip waters may be a sensitive proxy of recharge intensity and of the onset of the wet season after dry summer months or following a prolonged drought.

These metals can also be concentrated in soils during fire events (Treble et al., 2016) and/or be transported into the cave at a higher rate directly after fire events due to increased recharge (Treble et al., 2016). Although we know of no proximal fires in the SEKI region during the monitoring period, a controlled burn in Redwood Canyon (2016) may have contributed increased Cu, Mn, and Fe concentrations in the soil above Lilburn Cave. Additionally, the Ferguson Fire (96,901 acres) occurred in Yosemite National Park (∼160 km away) in summer 2018, potentially leading to increased aerosols that were transported to SEKI, such as found in other studies (Tadros et al., 2019). Although further work is needed to better understand the transport of trace metals in California cave systems, the observed seasonal trends of these trace metals in Lilburn drip waters and the strong correlation to variability in precipitation amount indicate that increases in the concentrations of these trace metals in drip waters may capture the onset of the rainy season in CA caves. If this hypothesis is shown by further study to be correct, then concentrations of these trace metals in stalagmites would be a critical proxy for reconstructing variability in the timing of the onset of the rainy season in CA, which has been strongly linked to the duration of the fire season (Lukovic et al., 2021; Swain 2021). Moreover, the concentrations of Mn, Cu, and Fe are recorded in the “farmed” calcite from Lilburn Cave at the same levels as they are measured in drip water. The results of this study, despite spanning only 2 years, document that many monitored variables respond to seasonal variability in precipitation amount and temperature. The monitoring results further demonstrate the potential for Lilburn Cave to archive a longer-term record of seasonal variability in stalagmites. Future studies of stalagmites from southern Sierra Nevada caves will benefit from the monitoring study presented in this paper.