An array of morphologies is present in the CCC samples from WWC. The smallest size fractions are spherulitic, with large aggregates reaching diameters of 50 µm, and individual spherules from 10 to 20 µm in diameter (Fig. 2c). This observation is corroborated by results from laser scattering, which reveal mean grain sizes from 24 to 42 µm in the <75-µm fraction. Rounded grains are typically clumped in botryoidal aggregates with smooth surfaces, and dumbbell structures are common. Occasionally more sharp-edged morphologies are present, with needle-like points that rarely exceed 5 μm in length. Larger size fractions (250-75 μm) contain rafts of aggregated grains in excess of 200 μm long. The TF samples, which were collected from the water surface, are characterized by arrow-shaped blades 10 to 20 µm long, grouped into thin rafts (Fig. 2d). All of these observations match CCC morphologies reported in the literature. For instance, previous work has established that CCC often occurs as raft-like aggregates of crystals, and spherical forms21,22,26. Rafts are generally flat and composed of interlocking crystals in a manner similar to floating carbonate minerals reported from non-cryogenic cave environments27. Crystal splitting has been invoked as a mechanism for the growth of spherical forms28, which are sometimes superimposed on a branching, sheaf-life skeleton29. Finally, spherules are reported to display smooth surfaces, and often connect during growth to form dumbbell shapes21 or chains22.
On the other hand, the CCC in WWC is considerably finer than precipitates with similarly depleted δ18O values reported from European caves. Although there is a wide range in sizes, CCC is typically described as having crystals from <1 mm to ~40 mm21,22,26 compared with the 10 to 20-µm diameter spherules common in WWC. The significance of this difference is unclear; it may simply indicate that CCC in WWC formed from smaller-volume pools, or from water with a lower total dissolved load.
Types of CCC Present in WWC. Although, as their names suggest, early studies differentiated CCCfine and CCCcoarse by grain size, δ13C and δ18O data reflecting different C and O isotope fractionation mechanisms are now recognized as the only valid criteria for distinguishing between the two groups18. With this in mind, values of δ13C and δ18O indicate that both types of CCC are present in WWC (Fig. 3a). Samples YS-3 and YS-4 have δ13C averaging 6‰ and an average δ18O of -7.3‰ (Table S2), suggesting that both are CCCfine. In contrast, samples YS-1, YS-2, YS-5, YS-6, CP, and YP have lower values of δ13C, averaging 3.5‰, and notably more depleted δ18O, averaging -16‰. These values identify these samples as CCCcoarse (Table S2). Sample TF also plots in the CCCcoarse field (Fig. 3a), although it was collected from a pool that had just begun to freeze, which may explain why its δ18O value is less depleted than the other CCCcoarse samples. For comparison, samples of the Madison Limestone collected in WWC have an average δ13C of 2.2‰ and δ18O of -5.9‰ (Fig. 3a). Consistency of stable isotope values within individual size fractions (250-165 µm, 165-75 µm, and <75 µm) emphasizes that YS-5, YS-6, CP, and YP are pure CCCcoarse (Fig. 3b). In YS-1 and YS-2, CCCcoarse is concentrated in the <75-µm fraction. In contrast, YS-4 is consistently CCCfine in all size fractions.
The results of additional analyses permit further interpretation of these samples. XRD reveals that YS-3 and YS-4 (CCCfine) are a mixture of calcite and quartz (Fig. S1), and rounded quartz grains were visible in the <250-µm size fraction. The presence of quartz is confirmed in the XRF results, where sample YS-4 contains 52% SiO2 (Table S3). YS-3 and YS-4 also have much higher chondrite-normalized REE values (Table S4, Fig. S2). These observations, along with the isotope data for size fractions presented above, are evidence for a detrital component in these CCCfine samples, which were collected from bedrock shelves above the current ice surface (Fig. 1). In contrast, XRD analysis indicates that the CCCcoarse samples are composed solely of calcite, as are the bedrock samples (Fig. S1). Major element analysis (Table S3) reinforces the abundance of calcite in YS-6 (70% as CaO), and Mg is considerably more abundant in YS-6 (12.2% as MgO) than in YS-4 (1.8%). Collectively the mineralogical and geochemical analyses support the conclusion from the isotope results that both CCCcoarse and CCCfine are present in WWC.
CCC Ages. Previous studies have successfully applied 230Th/234U disequilibrium dating26 to CCC15. Accordingly, an attempt was made to date CCCcoarse from WWC (Table S5). The resulting ages are imprecise due to low 230Th/232Th ratios, a situation that has complicated other work22,26, meaning that the correction for detrital 230Th, and the uncertainty on that correction, are large. On the other hand, one of the samples analyzed (TF) is clearly modern because it was collected in 2018 from a pool of water that was not present in 2016. The initial 230Th/232Th for this sample was, therefore, applied to the others (with a 25% 2-σ uncertainty) to refine the age calculations. Results indicate that the CCC in WWC likely formed during the Holocene, and nearly all of the ages have error estimates that overlap with modern (Table S5, Fig. S3). Thus these ages, imprecise as they are, are consistent with CCC formation as a recent or current process in WWC. This result is significant because the majority of published CCC ages are from the Late Pleistocenee.g. 21,30, with only a few reports of Holocene ages22,26,31.
Two additional lines of evidence support the interpretation that the CCC in WWC is young. First, rafts of calcite from sample YS-6 yielded a radiocarbon result of fModern 1.111 ± 0.005, consistent with formation in the late 20th Century when calibrated with the NH1 bomb curve in Oxcal 4.432 (Fig. S4). Second, samples of rodent fecal pellets from the ice beneath the surface lag of CCC yielded radiocarbon ages that calibrate to between AD 1600 and 185033 using the IntCal20 calibration curve34. Together this evidence strongly supports the conclusion that CCC formation in WWC occurred in the late Holocene, and is still occurring today.
Winter Wonderland Cave as a Unique Natural Laboratory for the study of CCC.
Because of its utility as a paleoclimate proxy, numerous studies have investigated CCCe.g. 21,30,35,36. However, there are only a few reports of CCC from locations outside Eurasia19,37, and all of these are categorized isotopically as CCCfine. Winter Wonderland Cave is, therefore, the first location in the Western Hemisphere where the unique paleoclimate indicator CCCcoarse has been identified.
Furthermore, nearly all previous observations of CCC describe these minerals from locations where they are present as loose concentrations of mineral grains on an ice-free cave floor21,26. Only twice has CCCcoarse been reported in association with modern, perennial ice31,36, and a recent comprehensive review noted that “Despite increasing evidence for Holocene CCCcoarse, actively forming sites have not yet been observed”13. In WWC, CCCcoarse with late Holocene to modern ages is present in association with perennial ice, making this cave an exceptional natural laboratory in which to study these precipitates and their genesis.
Furthermore, firsthand observations indicate that the situation within WWC is consistent with the model for the formation of CCCcoarse. During summer visits in 2014, 2015, and 2016, the cave was dry and the ice surface exhibited a complex pattern of elongated ridges and troughs with local relief on the order of 30 cm (Fig. 4) formed as a result of sublimation38. Short-term studies in an ice cave in Alberta, Canada suggest ice sublimation rates on the order of 3 mm per year39, although rates up to 10× higher have also been reported13. Either way, extrapolation from these benchmarks suggests that the relief observed on the ice surface in 2016 reflects 101 to 102 years of sublimation without the addition of new water. In contrast, in summer 2018, and again in 2019, liquid water entered the cave, filling many of the furrows, and freezing to create a new ice surface (Fig. 4). Deeper pools of water had lids of ice from 1 to 5 cm thick (Fig. 2b), precisely the mechanism proposed for the formation of CCCcoarse15,21,23. This water ranged from clear to a deep yellow color, and CCC was observed on the floor of each pool (Table S1). From the available data we cannot determine if this CCC precipitated from the water, or whether it was present as a lag on the ice surface before the pools formed. However, the concentration of solutes in this pool water was very high, with the highest values (K, Mg, and Ca >200 mg/L) corresponding to the yellow color (Fig. S5), consistent with conditions necessary for mineral precipitation through freezing-induced saturation14.
The change from 2016 to 2018 suggests that the thermal state of the epikarst has recently shifted, at least locally, to allow liquid water to penetrate to the level of the cave (~100 m below the ground surface), or to produce meltwater from ice farther back in inaccessible parts of the cave system. Under extensive permafrost conditions, caves are unlikely to contain ice sourced from dripwater because the permafrost inhibits the downward movement of liquid water40. Only when permafrost is degrading is it possible for liquid water to reach a cave where temperatures remain consistently subzero. The available evidence supports the interpretation that WWC and its surrounding host rock comprise a sporadic permafrost body that is currently undergoing this transition. Numerical modeling suggests that the time window between thawing and ultimate loss of permafrost is relatively short41, and studies have noted that cave ice is rapidly ablating in various locations around the world42. Future investigations in WWC should take advantage of this singular opportunity to observe the formation of different types of CCC in (near) real time, with the goal of improving our ability to use the presence of these features in currently ice-free caves as a dateable indicator of past permafrost thaw21.