High-resolution stalagmite stratigraphy supports the Late Holocene tephrochronology of southernmost Patagonia

Volcanic ash layers are important markers for the chronostratigraphy of paleoclimate and paleoenvironmental archives at the southern tip of South America. However, this requires that tephras are well-dated. We report geochemical data from stalagmite MA1 formed in a non-karst cave near Mt. Burney volcano in southernmost Patagonia (~53°S). High-resolution LA-ICP-MS analyses, SEM imagery, EPMA data, and NanoSIMS enable to identify volcanogenic signals during the last 4.5 kyrs from sub-annual trace element variations and tephra particles in distinct laminae. Our new 230Th/U-chronology of MA1 provides precise dating of tephra from Mt. Burney (MB) and, probably, Aguilera (A) at 4,216 +93/−193 yrs BP (MB2), 2,291 ± 33 yrs BP (MB3), 853 +41/−60 yrs BP (MB4) and 2,978 +91/−104 yrs BP (A1). This unique high-resolution record holds potential to date further eruptions from Southern Andean volcanoes for the tephrochronology in this critical region, and potentially also large-volume explosive volcanism off South America. South Patagonian speleothems yield chemical volcanogenic signals and tephra, which can be used to date regional volcanic eruptions and correlate them with the existing tephrochronology, according to geochemical analyses and micro-/nanoscale observations.

S evere and long-lasting impact on regional ecosystems from explosive volcanic eruptions and ash (tephra) fallout has been observed in fragile environments such as the superhumid South Patagonian Andes 1,2 . While geochemical fingerprinting links tephra glass shard compositions to their volcanic source and represents a valuable tool in tephrochronology 3 , the dating of distinct tephra layers enables a reconstruction of the regional history of volcanic activity and provides isochronous and stratigraphic marker horizons for the correlation of paleoenvironmental archives over large geographical areas 4 . Despite intensive research and rapid advances in the development of tephrochronological methods in the past few decades, the accurate identification of distinct tephra layers and the precise synchronization of single volcanic events recorded in geological archives from different depositional environments remains challenging 4,5 .
In southernmost Patagonia, the ages of widespread Holocene tephra layers from Plinian eruptions from volcanoes of the Patagonian Andes 6 in the southern segment of the Southern (41.5-46°S) and the Austral (49-55°S) Volcanic Zones were mainly deduced from 14 C-dated materials of terrestrial outcrops, peat and marine/lacustrine sediment cores 3,5,7,8 . This tephrochronological framework constitutes the principal reference for the stratigraphy of most geological records used in paleoclimate/ paleoenvironmental studies from this climatically unique and poorly explored region 9 .
But accurate age constraints for sedimentary climate and environmental archives are particularly difficult to obtain because age uncertainties and problems with the correlation of tephra layers in regional soils and sediment or peat core archives are due to variable composition, unclear stratigraphy, and diverging chronologies based on 14 C-dated carbonates, bulk organic matter (OM) or single plant remains 3,4 . In particular, 14 C-dated components of soils and sediments can have age offsets of hundreds of years due to reservoir effects caused by, e.g., the re-sedimentation and bioturbation of older OM or carbonate dissolution 10,11 , which have serious implications for age models. Large errors in age determination of volcanic ash layers also occur from the interpolation of bracketing 14 C dates 4 and/or age constraints based on estimated sedimentation/peat accumulation rates 12,13 .
Further challenges include the correct identification of the source volcano through the characteristic physical/textural properties as well as the mineralogical and geochemical composition of tephra pumice and glass shards 8 . Site-dependent, potentially intense weathering processes in acidic peatland ecosystems 14 can compromise geochemical fingerprinting methods for tephra correlation 3 . Chemical weathering over long periods strongly alters the composition of volcanic glass and embedded phenocrysts, leading to changes in the concentration of major and trace elements [15][16][17] .
Despite their increasingly important role as high-resolution terrestrial archives for the reconstruction of past climate and environmental changes, only a few speleothem records have so far contributed to regional tephrochronologies worldwide [18][19][20][21][22] . This is due to the difficulty of detecting unequivocal volcanic signals in speleothems. Furthermore, speleothem layers containing tephra glass shards or even small pumice particles haveto our knowledgenot yet been documented. Paine et al. 23 , for example, emphasized that the site-specific hydrological regime and soil/vegetation perturbation may attenuate the chemical signals from explosive volcanism that are preserved in stalagmites. Moreover, chemical signals of volcanic ash deposition can outlast the actual eruptive events by millennia 1 . These effects contribute to complicate the identification of volcanic eruptions by the sole stalagmite trace element profiling.
Nevertheless, speleothems can be very useful for such studies because 230 Th/U dating allows exceptionally accurate and precise dating and variations in chemical elements can be detected with up to (sub-) annual resolution. Therefore, speleothems have great potential for the high precision dating of volcanic events comparable to ice core tephra records 18,24 . Combined with stable isotope data (δ 18 O, δ 13 C) and trace element proxies, speleothem records enable both the dating of volcanic activity and the evaluation of possible environmental and climate impacts following the documented eruptions [18][19][20][21] .
In this work, we analyze a non-karst speleothem (stalagmite MA1 25 ) to better constrain the Late Holocene tephrostratigraphy in southernmost Patagonia and provide evidence for soil-environmental changes related to tephra deposition under extreme climate conditions. This speleothem formed in a small cave ca. 40 km S of Mt. Burney volcano and recorded sub-annual trace element variations detected by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). These high-resolution compositional profiles were combined with a refined 230 Th/U-chronology age model constructed from published 25 and additional new 230 Th/U dates. Based on this improved age model, our new LA-ICP-MS data, and principal component analysis (PCA), we (i) detect and analyze pyroclastic particles incorporated into the stalagmite and (ii) correlate these particles to pristine tephra of Mt. Burney volcano by electron microprobe (EPMA) glass composition measurements. This allows (iii) to precisely date the MB 2 , MB 3 , and MB 4 tephra using the improved age model for the speleothem archive based on 230 Th/U-ages. Furthermore, (iv) the MA1 record documents the environmental response to changes in the hydrological/pedological regime caused by the eruptions and their deposits. We also identify a chemical signal of the A 1 eruption of Aguilera and provide a new age estimate for this eruption as well. In addition, (v) EPMA measurements, scanning electron microscopy (SEM), and nanoscale secondary ion mass spectrometry (NanoSIMS) of detrital components preserved in distinct laminae confirm the identification of pristine and reworked tephra particles and their alteration products. Within the context of the data reported here, previously published stable isotope data (δ 18 O, δ 13 C) of MA1 25 provide additional constraints on short-term perturbations of the local surface hydrology and soil composition following the eruptions.
Environmental context and characteristics of stalagmite MA1. The MA1 stalagmite was obtained from the Marcelo Arévalo (MA) cave 25 , situated at 20 m a.s.l. in the Magellanic moorlands at 52°41.50′S/73°16.28′W (Fig. 1a). The cave formed in a nearly vertical tectonic fracture zone cross-cutting a plutonic sequence of coarse-crystalline granite with mylonitic orthogneiss and rare mafic dykes 26 . Granite and orthogneiss mainly comprise quartz, plagioclase, and mica. The most prominent accessory minerals are pyrite, apatite, and zircon. Rarely occurring mafic intrusives are mainly composed of hornblende, plagioclase, clinopyroxene, and a few mica. Accessory phases include, e.g., apatite and zircon. The hydrological system in this fracture zone directly connects the cave with a small catchment (~2500 m 2 ) at 80 m a.s.l. that is characterized by acidic (pH 3.9-5.7), waterlogged peatland soils, and surrounding bare rock outcrops 26 (Fig. 1b). The super-humid climate in this region is controlled by pronounced seasonal variations of the southern westerly wind belt [27][28][29] (SWW), present annual precipitation of~3800 mm yr −130 , and mean annual temperatures of 5.3°C 30 (recorded at the automatic weather station Arévalo; Fig. 1c). Reconstructed annual precipitation during the Late Holocene reached up to 6500 mm yr −125 . Monitoring data between 2005 and 2008 reveal that cave temperatures do not differ from atmospheric values and that average relative humidity is 94% 25 . The drip water has a pH of 7.5 and elevated average drip rates (~150-250 cm 3 d −1 ) measured at the MA1 sampling site vary within the seasonal trend 25 . An estimated maximum delay of only 4-6 weeks between major rainfall events and related drip rate increase 25 indicates a relatively high transmissivity of the fractured aquifer system due to good connectivity of the cave with the local surface hydrology in combination with the extraordinary precipitation. Stalagmite MA1 shows a rather constant mean growth rate of 67 μm yr −1 (minor variations of 25-100 μm yr −1 ) 25 and continuous growth over the last 4.5 kyrs BP (before present, year 1950 of the common era). A significant hiatus, interrupting the growth of MA1, has not been observed 25 . Along the growth axis, MA1 is dominated by compact calcite, whereas marginal areas and distinct sections rich in OM and detrital silicate components (volcanic glass, basement-derived mineral grains; Fig. S2) are more porous. However, microscopic analyses provided no evidence for CaCO 3 dissolution and/or recrystallization [31][32][33] in these laminae.
The cave site is located only 40 km S of the Mt. Burney volcano (52°19′S/73°22′W, Fig. 1a), the most active volcano of the Austral Volcanic Zone during the Late Quaternary 5 . Originating from this volcano, ten Holocene Plinian eruptions dispersed tephra layers across wide areas of southernmost South America 3 . After the large-volume MB 2 eruption, regional pristine ecosystems suffered millennium-scale damage caused by SO 2 -induced acidification associated with the deposition of >10 cm of volcanic ash 1,2 . By contrast, the tephra deposits from the notably smaller Mt. Burney eruptions with a volcanic explosivity index below 5 3 are only a few millimeters thick 12,13 and probably had minor impacts on the environment. All volcanic eruptions we address in this paper are identified and named according to the database published by Fontjin et al. 3 .

Results
Chronology. We used StalAge 34 to construct a new age-depth model for the MA1 speleothem compared to the previous study by Schimpf et al. 25 (Fig. 2). Our age-depth model (Table S1) is now based on 16 published and 2 new 230 Th/U-ages and a slightly modified detrital correction factor for Th 35 . The applied StalAge model for MA1 covers the entire 280 mm length along the growth axis of the speleothem, representing about 4.8 kyrs. For details of the 230 Th/U-age constraints, detritus correction, and the construction of the age model, see the "Methods" section.
Element concentrations determined by LA-ICP-MS. Concentrations of S, P, U, Si, Sr, and Zrselected by virtue of the well-known composition of Mt. Burney tephra 7,8,36 and the environmental impact after the deposition of MB 2 ash 1were measured on multiple overlapping thin sections prepared along the entire length of MA1 (4.5-0.6 kyrs BP) and are presented in Figs. 3-5. Average elemental concentrations were determined from a total of 59,200 LA-ICP-MS measurements for S (132 ppm), P (2212 ppm), U (2.4 ppm), Si (1400 ppm), Sr (197 ppm), and Zr (3.2 ppm). Against this background, we observe numerous high amplitude spikes identified as short (few months up to three years) and strong variations in the deposition of these elements. Such peak concentrations reached >10 ppm for U and Zr, >400 ppm for Sr, >1000 ppm for S and P, and easily exceeded 10,000 ppm in the case of Si (Figs. [3][4][5]. These high amplitude spikes are well correlated for S, P, U, Si, Sr, and Zr and appear highly irregular with respect to frequency and intensity along with the profiles. Particular conspicuous spikes of S, P, U, Sr, and partly of Si and Zr, occur at 4.216, 2.978, 2.291, and 0.853 kyrs BP (orange bars; Figs. 3 and 4). These conspicuous spikes are followed by (i) a series of mostly smaller, but equally discrete, short-term peaks for the same elements decreasing in height or by (ii) more extended periods characterized by consistently elevated concentrations from 4.20 to 4.14 kyrs BP, respectively (gray bars; Fig. 4). As would be expected for elements that are contributed from detrital particles sourced from tephra, Si should be closely related to Zr. Indeed, Fig. 5 shows that a large number of LA-ICP-MS measurements fall on a tight correlation trend that projects from the origin with a Si/Zr ratio of 3300 toward the composition of Mt. Burney glass with average values of 77 wt.% SiO 2 and 110 ppm Zr 7,8 . A 1 glass from Aguilera has a similar Si/Zr ratio of 3600 (75.7 wt.% SiO 2 and 97 ppm Zr on average 8 ), however, high Si and Zr peaks are absent in the expected section from 3.1 to 2.9 kyrs BP (Fig. 5d). By using these Si/Zr ratios, we calculated mixing lines between two endmembers: the Si-and Zr-rich Mt. Burney/Aguilera glass and the Si-and Zr-free CaCO 3 matrix of the speleothem (blue lines in binary plots in Fig. 5). Concentrations for Zr and Si that are higher than the concentrations in the tephra are an artifact of normalization to Ca in LA-ICP-MS measurements that include distinct but unknown amounts of a non-carbonate detrital component (i.e., basement-derived minerals, which are lower in Ca compared to CaCO 3 ). This artifact in the concentration data is unavoidable and does not affect the characteristic Si/Zr ratio of the volcanic glass mixing line. However, a large number of analyses fall off this correlation and can be attributed to silicate minerals from basement rocks. One additional component is particularly rich in Si (and low in Zr) and identifies quartz or silica gels, whereas a few data points tend towards high Zr and are likely caused by traces of detrital zircon. These minerals have also been identified on SEM micrographs.
In order to further characterize elemental peaks of Si, Zr, S, and other elements in the speleothem record and to provide for tests that these may represent volcanic signals, we analyzed the data set using a PCA 22,37 . The first three principal components (PC) explain 80.7% of the variability within the entire data set (Figs. 6 and S1 and Table S2). High scores for PC-1 (44.4%), PC-2 (19.4%), and PC-3 (16.9%) reflect the correlated trace element peaks as discussed above and displayed in Figs Composition of detrital silicate components. Major element compositions of detrital silicate particles embedded in MA1 were analyzed by EPMA ( Fig. S2 and Tables S3-S5). We detected pristine rhyolitic glass fragments in laminae at 4.21, 2.29, and 0.85 kyrs BP (78.12 wt.% SiO 2 , 1.77 wt.% K 2 O, 3.46 wt.% Na 2 O, and 0.29 wt.% MgO on average). These compositions are identical to matrix glasses from pumice originated from Mt. Burney, but differ significantly from known compositions of other documented wide-spread tephras from Hudson and Reclus volcanoes in the Southern and Austral Volcanic Zones 5,7,8 (Fig. 3b, c). A few oscillatory zoned plagioclase and euhedral clinopyroxene crystals are also typical for Mt. Burney tephra 36 (Fig. S2c, e).
Other mineral components detected in laminae at 2.29, 1.79, 0.85, and 0.73 kyrs BP ( Fig. S2 and Table S4) include ilmenite, apatite, pyroxene, biotite, plagioclase, quartz, and zircon. The occurrence of quartz at 0.76 kyrs BP is also confirmed by X-ray diffractometry, showing a peak at 3.344 Å (Fig. S2g). The composition, intergrowth textures, and shape of these mineral grains indicate that they originate from a granitic/gneissic bedrock. Additional EPMA measurements on Si-rich particles found in porous fabrics can neither be related to the composition of Mt. Burney glass nor to bedrock minerals (Table S5). These particles show a large range in composition with no relation to any mineral stoichiometry. They are, on average, significantly lower in SiO 2 (61.39 wt.%), TiO 2 (0.08 wt. % or below detection limit) Na 2 O (0.50 wt.%) and K 2 O (0.47 wt.%) and higher in Al 2 O 3 Fig. 2 New age-depth model of the MA1 stalagmite. a StalAge 34 age-depth model, including the corresponding 95% confidence limits (thin red lines). Uncorrected ages and ages after detritus correction 35 are shown with standard deviations as 2σ error bars (see Table S1 for details). b Photo of MA1 including the 18 sampling points for 230 Th/U dating. NanoSIMS measurements on inclusions at~3.50 and~3.11 kyrs BP did not detect such pristine siliciclastic particles (Fig. 8a,  b). By contrast, areas with higher 16 O − , 28 Si − , 27 Al 16 O − , and 56 Fe 16 O − counts appear as fine flaky structures. These structures occur scattered throughout the observed areas and correlate with high 12 C − , 12 C 14 N − , and 32 S − signals, here interpreted as indicative of OM. At~0.73 kyrs BP, particles with strong localized signals of 28 Si − and moderate 16 O − in the absence of other measured elements (Fig. 8c) are interpreted as phenocrysts or grains of bedrock-derived mineral detritus, such as quartz (cf. identified minerals; Fig. S2 and Table S4). However, in this case, thin 27 Al 16 O − rims have not been observed.

Discussion
Identifying volcanic input by compositional spikes of characteristic elements in the MA1 speleothem record provides means for precise dating and the correlation of volcanic ash inputs into Patagonian soils and environmental archives 9 . In order to identify such volcanic signals, we need to consider the direct primary elemental deposition from tephra, volcanic volatiles, and leached components. In addition, there may be a secondary response to tephra deposition on peatland soils that potentially affects nutrient cycling 38 , chemical leaching 15,16 , and soil C stocks 39,40 . Pristine peatland ecosystems such as those of southernmost Patagonia may be particularly sensitive to volcanic impacts from tephra-loading under extreme climate conditions 1,41 . Finally, the improved age constraints of tephra layers that we detected in MA1 allow addressing the impact that volcanic eruptions had on this super-humid environment.
There is direct evidence for tephra deposition and alteration during the growth of stalagmite MA1. The principal elemental input from tephra comes from fine-grained glassy pyroclastic particles and is documented by synchronous and sharp peaks of Zr and Si in the MA1 record. These peaks occur at and after 4.216, 2.978, 2.291, and 0.853 kyrs BP. Our geochemical fingerprinting based on EPMA data of preserved glass shards (Fig. 3b (Fig. 5). This extended input of Zr and Si is most likely related to the deposition of fine-grained ash particles (nanotephra) that cannot be resolved by optical microscopy observations in the CaCO 3 matrix. Higher input of such fine-grained glass following the eruption events over decades can be explained by repeated periods of increased tephra reworking and mobilization into the aquifer during distinct phases characterized by storm-related high rainfall 9,27 . These periods coincide with decreased δ 18 O and δ 13 C values, which is in accord with higher precipitation rates 25 at that time. Owing to differences in Late Holocene rainfall 29 shown by variable δ 18 O and δ 13 C levels 25 , a climate-induced influence on the hydrological soil-tostalagmite transport is suggested, explaining the slightly dissimilar response to tephra fallout recorded by stalagmite MA1 during 4.5-4.0, 3.1-2.9, 2.4-2.0 and 0.9-0.7 kyrs BP (Figs. 3 and 4).
Micrometer-sized volcanic glass fragments with surface alteration rims were also observed by NanoSIMS (Fig. 7d-f). These particles differ considerably from crystalline, bedrockderived silicate grains (Fig. 8c) with respect to their texture and 28 Si − and 16 O − secondary ion count levels at their location, which asserts them as volcanic glass. This is in accord with an expected low 27 Al 16 O − signal and 56 Fe 16 O − counts near to zero for vitreous material, which does not exclude that they contain greater amounts of Al or Fe 42,43 . The rims rich in 27 Al 16 O − around these particles suggest the formation of amorphous, secondary alumino-silicate phases with higher ionization rates in comparison with more crystalline structures [43][44][45] . We interpret the 12 C − and 12 C 14 N − signals detected on the particles as reflecting enrichments of OM 42 in abundant micro-vesicles. This is consistent with the results of recent studies that documented pronounced glass leaching and the subsequent formation of Al-(hydr)oxides related to tephra weathering in other humid and acidic environments elsewhere 16,46,47 . The close textural relation between 27 Al 16 O − enrichments and OM ( 12 C − , 12 C 14 N − , and 32 S −48,49 ) surrounding the observed particles ( Fig. 7d-f) suggests that leaching processes, e.g., microbially mediated weathering, also occur in-situ within the bio-chemical micro-environment existing at the stalagmite surface after the deposition of the tephra particles. Based on microscopic observations, we suggest that the tephra may have been fragmented mechanically while passing through 40 m of the fractured aquifer system. Subsequently, residual vitreous particles accumulated together with soil-derived OM in small pools formed at the stalagmite surface (Fig. S5e). In these accumulation pools, glass dissolution by microbes 50,51 is probably encouraged. A thin glaze of CaCO 3 that differs from the surrounding calcite fabric preserved the organic and inorganic components within these bio-geochemical micro-environments (Fig. S5e).
Our data constrain the chemical signals from tephra deposition: Apart from the direct observation of pristine glass shards and their in-situ alteration products, there is also evidence of secondary effects in soils as a response to volcanic fallout. Various    volcanogenic chemical compounds dissolve rapidly in surface environments after the direct wet deposition of acid rain from the eruption plume and/or after leaching by water from freshly deposited tephra 15,38 . Typical chemical compounds leached from pristine tephra are sulfates and a range of soluble salts containing other mobile and volatile metals 38 . Here we focus on S, P, Sr, and U to trace solutes potentially derived from Mt. Burney and Aguilera tephra for which the composition is well known 7,8,36 . Amongst elements released from volcanic ash after deposition, S and Sr but also P are known to reach high concentrations in surface waters 38 and U was also efficiently applied to document long-term leaching after the MB 2 eruption 1 . The low pH of 3.9-5.4 in soils of the MA catchment 26 promotes the mobilization of water-soluble elements from volcanic glass 15,17,38 . We, therefore, interpret the short-term release of Sr and U as a signal of tephra leaching in acidic waterlogged peat soils 1,14 at 4.216, 2.978, 2.291, and 0.853 kyrs BP (Figs. 3 and 4). We associate the periodically increasing Sr and U concentrations in periods following the eruptions with the storm phase-induced incorporation of (i) leached elements from tephra in the catchment or (ii) glass and its weathering residues. The high peaks of Sr could additionally be explained by a contribution from plagioclase phenocrysts 21,22 embedded in MA1 (e.g., glass-coated phenocrysts; Fig. 7b, c), which are-next to volcanic glass-a dominant component of tephra from Mt. Burney and Aguilera 7,36 .
These chemical signals also determine the high amplitude spikes of PCs 1-3 (Figs. 6 and S1). We use our PCA results (Figs. 6 and S1 and Table S2) to discriminate different compositional domains in stalagmite MA1. PC-1 mainly reflects high concentrations of presumably volcanogenic SO 4 2− and PO 4 3− 15,18,22 . We expect a simultaneous mobilization of S-and P-bearing species adsorbed to or contained in pedogenic OM 52,53 due to the perturbation of peat soil hydro-chemistry after ashfall 1,2,41 . Accordingly, high amounts of 32 S − , 12 C − , and 12 C 14 N − secondary ions counts are correlated (Figs. 7 and 8) and imply a direct linkage between soil OM and S in the stalagmite 48,54-56 even though OM is obviously not directly associated with the composition of fresh tephra 38 . In addition, PC-1 implies a relatively high contribution of siliciclastic components in the form of fine-grained detritus in combination with intense leaching, mobilization, and deposition of mobile lithophile elements from rock substrate or tephra in soils. PC-2 and PC-3 primarily mirror leaching processes and siliciclastic components without a concurrent deposition of volcanogenic SO 4 2− and PO 4 3− . According to Jamieson et al. 22 , volcanic signals in a speleothem do not automatically produce a clear corresponding chemical excursion in the stalagmite stratigraphy. However, in our particular case, all three PCs are indicative of volcanic impacts on the soil-to-stalagmite transfer and deposition system in the form of pristine glass, elements leached from the eruption plume, or products issuing from tephra alteration. As would be expected from loadings of PC-1, the main spikes of the PC-1 record coincide with peaks of the lithophile elements (Si, Sr, Zr, and U) as well as S and P concentrations (Figs. [3][4][5]. Such high amplitude compositional spikes, significantly exceeding average concentrations, are comparable to characteristic volcanogenic signals observed in other speleothem records 18,[20][21][22][23] . Therefore, it is reasonable to infer that these distinct compositional peaks mark times of trace element mobilization related to volcanic events that occurred at 4.216, 2.978, 2.291, and 0.853 kyrs BP (Figs. 3-6 and S1).
Chemical signals appear to indicate a prolonged chemical footprint of tephra deposition (i.e., elevated Si, Sr, and Zr; Figs. [3][4][5] in speleothem laminae from 3.8 to 3.0 kyrs BP, similar to that at 4.216, 2.978, 2.291, and 0.853 kyrs BP. However, there is no direct evidence for tephra deposition from nearby Mt. Burney, Reclus, or Aguilera volcanoes at that time 3,5,8 . Despite our intensive search by optical microscopy, wavelength dispersive spectroscopy (WDS) element mapping, and SEM, we could not detect any intact vitreous particles between 3.8 and 3.0 kyrs BP. The measured compositions of Si-rich particles from this section of MA1 cannot be associated with the composition of fresh volcanic glass or stoichiometric basement-derived minerals (Fig. 3d and Table S5). Alkali metal contents are exceedingly low and pronounced Al-/Fe-(hydr)oxide enrichment, as suggested by high Al 2 O 3 /TiO 2 and FeO/TiO 2 ratios, is observed. We, therefore, categorize these high Al-Fe-Si compounds to be the result of glass decomposition where newly formed alteration products were partly mixed with surrounding CaCO 3 (higher CaO values during EPMA analysis).
Based on our data, we argue that the environmental impact on soils from volcanic depositions is modulated by climate variations. Kilian et al. 1 documented a millennium-scale acidification phase in regional soils following the Plinian MB 2 eruption. This may have caused efficient chemical alteration of tephra and transport of leached elements from the catchment onto the speleothem. Accordingly, we observe correlated patterns of 12 C 14 N − , 32 S − , 28 Si − , 27 Al 16 O − , and 56 Fe 16 O − counts in NanoSIMS images of the speleothem during the period following the MB 2 eruption (Fig. 8a, b). This indicates the accumulation of silicate weathering products together with OM 44,48,49 . These weathering products are typical for the decomposition of vitreous tephra in acidic soils and are likely comprised of organo-mineral complexes enriched in Al-/Fe-(hydr)oxides and silica gels 46,47 . These compounds preferentially form in highly reactive aquatic redox environments 57 subjected to such extremely variable climate conditions 28,30 .
We also observe variations of δ 18 O and δ 13 C values 25 by 1-1.5‰ in the record following the events at 4.216, 2.978, 2.291, and 0.853 kyrs BP. First, the δ 18 O and δ 13 C values decrease for ca. 10 years followed by higher values lasting for~20 years (Fig. 4). We interpret these isotopic changes as a combined effect of varying rainfall intensity and short-term perturbations of the hydro-chemical soil system within the catchment-subsequent to and genetically related to events of tephra deposition 21 . Environmental changes such as variations in redox-pH conditions, water-table fluctuations, and nutrient release from ash deposition may strongly affect plant decomposition, and thus, the delivery of dissolved organic matter (DOM) and organo-mineral compounds from peat soils [57][58][59] . The DOM then can percolate into cave drip waters and be recorded by the δ 13 C signatures of speleothems 20,[54][55][56][57] . A close relationship between the abundance of OM that derives from DOM or mineral-associated OM in the drip water and 12 C − , 12 C 14 N − and 32 S − signals in detritus-rich laminae of MA1 is shown by the NanoSIMS measurements 48,49 (Figs. 7 and 8). Therefore, a link between detrital and volcanic input in MA1 and the fundamental changes in soil chemistry and hydrology in the catchment is clearly established.
Finally, an import of elements such as S and Sr in the MA catchment area can also occur by sea salt aerosols 25,60,61 . It is known that sea salt contribution from the nearby Pacific Ocean through surface deposition and transfer to cave drip waters is responsible for overall elevated concentrations compared to more continental speleothems 21,22,24,25,62 . Such speleothems contain concentrations of, i.e., Sr ranging between 10 and 20 ppm 22,62 . Consistent and strong SWW-induced wind velocities and frequent storm events in the region 9,27 cause a constant deposition of sea spray-derived elements. These elements are characterized by short residence times in waterlogged peatland soils 14 , and thus, we argue that the high annual precipitation 9,28,30 leads to a persistent input of excess sea salt S and Sr signatures in MA1 compared to other speleothems. Therefore, sea salt-derived S and Sr contribution should be of secondary importance for the distinct high amplitude spikes of these elements. Their deposition would not result in the documented compositional spikes and a close correlation that we observe with other mobile and immobile lithophile trace elements (e.g., Si, U, and Zr, Figs. 3-5) would not be expected.
Using the evidence from primary and secondary volcanic signals in the speleothem record, these data have implications for tephrochronology because our results prove that the MA1 stalagmite recorded distinct volcanic events for which we can determine precise 230 Th/U-ages from the refined age-depth model based on the StalAge algorithm 29 with errors represented by 95% confidence limits (Fig. 2a). Our speleothem record shows no distinct hiatus and because of the proximal location and high temporal resolution, it also comprises volcanogenic signals from so far insufficiently dated, notably smaller Mt. Burney eruptions with unknown ash dispersal and regional environmental impact (MB 3 and MB 4 ). It is important to note thatapart from stratigraphic correlationa differentiation between fresh ash depositions and rainfall-induced soil erosion/reworking of older tephra layers 1,63 is necessary. Evidence for a distinct major tephra reworking event is high peaks of Si and Zr without the recording of a compositional volcanic signal at the same time. In addition, phases of increased reworking should also correlate with paleoprecipitation data (δ 18 O and δ 13 C) and known past environmental conditions. For instance, a major reworking event is recorded during the termination of a well-documented cold phase 9 at~2.5 kyrs BP. Due to the environmental damage caused by the millennium-scale acidification 1 and the subsequent cold phase 9 , a series of storm events with extreme rainfall may have mobilized MB 2 ash from bare, sparsely vegetated surfaces at that time (Fig. S3). With respect to the rapid alteration of volcanic glass in acidic soils under humid climate 64 , the reworked material from older, surface-near tephra should not contain pristine glass and high S concentrations would not be expected. Indeed, the peaks of S at~2.5 kyrs BP are considerably reduced by a factor of ten in comparison with the distinct volcanic events that we detected (Fig. 3). Accordingly, the LA-ICP-MS data from a section older than the MB 2 eruption (4.8-4.6 kyrs BP) indicate similar depositional events comprising Mt. Burney ash (Fig. S4). Unlike the other tephra signals discussed above, no published 14 C-ages exist for this time, and thus, a correlation with a Mt. Burney eruption would be highly speculative. However, a possible eruption of Mt. Burney (MBK 3 3 ) was identified based on a 1 mm thick tephra layer found in only one local sediment core with a 14 C-age of >4.96 ± 0.09 kyrs BP 13 . Here, further interpretations with respect to primary ashfall or reworked inputs must await a more detailed analysis of this oldest section of MA1.
We consider the most important findings of this study that, based on 230 Th/U-ages, three Mt. Burney eruptions can now be dated more accurately to 4216 +93 / −193 yrs BP (MB 2 ), 2291 ± 33 yrs BP (MB 3 ), and 853 +41 / −60 yrs BP (MB 4 ). The 230 Th/U-age for the MB 2 eruption is in good agreement with the 14 C-age of 4200 ± 50 cal yrs BP published by Breuer et al. 63 , which is based on macro-plant remains included in the >10 cm thick MB 2 ash deposit. Our new MB 3 age falls into the lowermost range of bracketing 14 C-ages presented by Biester et al. 12 and Kilian et al. 13 . In both papers, the MB 3 eruption was roughly dated tõ 2020 ± 90 cal yrs BP (>1830 ± 40 and <2210 ± 90 cal yrs BP 12 , >1980 ± 40 and <2060 ± 90 cal yrs BP 13 ). This new age is, therefore, more precise than previous datings. A further Mt. Burney eruption (MB 4 ) that was only tentatively dated through stratigraphic correlation (~800 yrs BP 65 ) is now more precisely dated to 853 +41 / −60 yrs BP, based on the data of this study.
Following our arguments developed above, our trace element profiling and PCA analysis (Figs. 3-6 and S1), indicate that an additional volcanic event occurred at a 230 Th/U-age of 2978 +91 / −104 yrs BP. This event could not be linked to a specific eruption by chemical correlation of tephra because no glass was detected in MA1 at this time. However, the A 1 eruption from the remote Aguilera stratovolcano has a poorly defined 14 C-age bracketed between~3067 and 3339 cal yrs BP 7 and, therefore, it is the most likely candidate for this event. This is in accordance with the regional distribution of A 1 tephra 7 , which has also been discovered in lacustrine sediments nearby the MA cave site 13 .
With this successful identification of volcanic eruptions, we propose that our high-resolution speleothem archive enable precise 230 Th/U-age constraints for a refinement of existing tephrochronologies, closing a significant gap in the reconstruction of the volcanic activity in South Patagonia. Potentially, the variations in LA-ICP-MS trace element concentrations in this stalagmite imply environmental impacts induced by further ashfall events from more distal volcanic centers of the Southern and Austral Volcanic Zones but are much more difficult to detect. For example, there are peaks of elevated S concentrations in MA1 (Figs. 3 and 4) at times when no eruption is documented by the nearby volcanoes. These distinct S peaks likely record atmospheric signals of more distant volcanoes, such as Reclus and Cerro Hudson and, potentially, other high-magnitude volcanic events in the southern hemisphere off South America. This opens the possibility of correlation with Antarctic ice cores 66 to evaluate large-scale southern hemispheric patterns of Late Holocene volcanism.
We, therefore, conclude that the unique, sub-annually resolved speleothem LA-ICP-MS data presented here may constitute an important advancement in our knowledge about the Late Holocene tephrochronology of southernmost South America. This will allow a better understanding of the volcanic history in the region, and an improvement for the reconstruction of paleoclimate and paleoenvironmental conditions from regional geological archives. Besides well-preserved tephra layers, MA1 provides multiple climatically and environmentally important information due to its unique nature as a non-karst stalagmite with enclosed organic and inorganic detrital components. This information includes potential feedback signals to SWW dynamics and environmental impacts of regional volcanism back to~4.8 kyrs BP, documenting the evolution of one of the most remote and pristine ecosystems worldwide 9,27 .

Methods
Chronology. Stalagmite MA1 was dated at the Heidelberg Academy of Sciences. Sample preparation and analytical methods were described in detail by Schimpf et al. 25 . We included two additional samples (Lab. No. 3672 and 3436, Table S1) that were not considered in Schimpf et al. 25 due to their large content of detrital 232 Th. Stalagmite MA1 generally contains high amounts of detrital Th. This is obvious from the 230 Th/ 232 Th activity ratios that range from 4.257 to 93.70 (Lab. No. 3672 and 3653, Table S1). However, these values are well below the limit of >200 suggesting insignificant detrital contamination 67 . Detrital contamination of MA1 also results in age inversions (i.e., uncorrected ages that are not in stratigraphic order). For instance, the sample with Lab. No. 3435 at 268.5 mm distance from the top (dft) is significantly older (5.128 kyrs BP) than the sample below at 274 mm dft (4.803 kyrs BP).
Schimpf et al. 25 assumed a 232 Th/ 238 U weight ratio of 4.8 ± 50% for the detritus, which resulted in a smooth age-depth relationship. This value agrees with the conventionally used value for the upper continental crust (3.8 ± 1.9 68 ). Here we used a different approach to account for detrital contamination, based on an algorithm recently published by Budsky et al. 35 . This algorithm calculates corrected ages assuming a specific 232 Th/ 238 U weight ratio of the detritus and 230 Th, 234 U, and 238 U in secular equilibrium, and then determines the 232 Th/ 238 U weight ratio of the detritus, which results in a minimum sum of age inversions (in years). For stalagmite MA1, this approach results in a 232 Th/ 238 U weight ratio of the detritus of 3.69, which is close to the mean value of the upper continental crust 68 and in agreement within the error with the value used by Schimpf et al. 25 . As suggested by Budsky et al. 35 , we assumed a conservative uncertainty of ±50% for the determined 232 Th/ 238 U ratio of the detritus. Figure 2a shows a comparison of the uncorrected and the corrected ages.
The final age-depth model for MA1 (Fig. 2a) was calculated in R 69 using the Monte-Carlo-based StalAge algorithm 34 , which has been specifically designed for speleothems. In contrast to the previously used Akima interpolation, StalAge provides 95% confidence limits, which allows to assess the uncertainty of the age model.
Drip water pH. The pH of cave drip water collected at the MA1 sampling site was measured on-site with a digital pH-meter during expeditions between 2006 and 2008.
Mineralogical analyses. The mineral composition of a powdered specimen from 37 mm dft (Fig. S2g) was identified by X-ray diffraction with a Siemens D500 X-ray diffractometer at the Geology Department of Trier University. The diffractogram was evaluated using the DIFFRAC plus 5.0 software from Bruker. We accepted the calcite peak at 3.036 Å (Cu-Kα radiation) as the internal standard.
LA-ICP-MS data. The MA1 stalagmite was analyzed with laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) at GZG, University of Göttingen. LA-ICP-MS elemental analyses were performed as two parallel ablation tracks on 14 overlapping, 30 μm thick and 40-50 mm long polished thin sections covering the growth axis of the stalagmite from 24 to 280 mm dft (Fig. S5a). This approach accounts for lateral heterogeneities with respect to small-scale differences during stalagmite growth. We used the same thin sections as for EPMA presented in Schimpf et al. 25 . Figure S6 documents that the resin used for thin section manufacturing does not contain S, P, U, Sr, Si, or Zr in any significant amounts. Therefore, potential contamination caused by adhesive compounds affecting the LA-ICP-MS element concentrations presented in this study can be excluded. The ICP-MS used was a Perkin Elmer ELAN DRC II Q-ICP-MS coupled to a Lambda Physik COMPex 110 Argon Fluoride laser (193 nm) with a GeoLas optical bench by MikroLas and a small volume sample chamber. The ablation setup was operating with a beamwidth of 120 µm, a repetition rate of 10 Hz for the laser pulses (dwell time 40 ms per isotope, 1.078 s per sweep), and a scan speed of 20 μm s −1 . Background data were taken before and after the measurements for 2-3 min. The measurements were calibrated to the standards NIST610 and NIST612 by bracketing analyses and the internal standard 43 Ca was used to convert the measurements into absolute concentrations. Matrix effects between NIST610/NIST612 and CaCO 3 were considered to contribute less than 10% error. Measured isotopes were 27 Al, 29 Si, 31 P, 34 S, 43 Ca, 88 Sr, 90 Zr, and 238 U. Element concentrations have been normalized to the stoichiometric 400,000 ppm 43 Ca, assuming that the stalagmite is composed of 100% CaCO 3 . The background was subtracted and baseline drift corrected. For ablation spots on lamina with more than 10% detrital siliciclastic material (minerals/volcanic glass containing considerably lower Ca compared to speleothem CaCO 3 ), the internal normalization to 43 Ca will result in systematically higher values for lithophile elements. However, their elemental ratios (e.g., Si/Zr) will be unaffected. The analytical background of Si (<5,000 ppm) and Zr (<3 ppm) represents analytical scatter at low concentrations, and possibly, a weak signal caused by continuous tephra input from reworking. This background range was defined based on the limits of detection for Si and Zr 70 and the observed scatter in our LA-ICP-MS data set.
The final LA-ICP-MS record was compiled by aligning the overlapping ablation tracks using a wiggle-matching technique based on the Al concentrations measuredtaking small-scale variations of lamina thickness into account. It is characterized by sub-annual resolution of in total 59,200 LA-ICP-MS measurements covering a time span of 4212 years (4809-597 yrs BP). Finally, the consistency of the LA-ICP-MS data set with already published ICP-MS drill core data 25 was ensured by matching averaged concentrations of the LA-ICP-MS measurements with the contents previously determined by ICP-MS. For this purpose, we chose U as soluble, calcite-compatible element 25 and used the original drilling diameter of 1.5 mm in accordance with the exact distance from the top for every single coring point to calculate the average of the LA-ICP-MS concentrations for comparison (~290 LA-ICP-MS measurements per drill core). The LA-ICP-MS and ICP-MS U concentrations are in good agreement and correlate with r = 0.68 (Fig. S7).
The principal component analyses (PCA) for the assessment of speleothem high-resolution LA-ICP-MS data was done according to Orland et al. 37 and Jamieson et al. 22 . The PCA was performed with Origin (version Pro 2019b) to further constrain the occurrence of signals in MA1 derived from volcanic activity and was calculated for the trace elements P, S, Sr, Si, Zr, and U after z-score normalization.
EPMA analyses of silicate components in MA1. The major element compositions of silicate minerals and glass shards incorporated in laminae of MA1 that formed at 4.216, 2.291, 0.853 kyrs BP and shortly after were measured using a JEOL JXA-8900 RL EPMA equipped with five WDS spectrometers at GZG, University of Göttingen. On the polished thin sections, areas of interest (AOIs) with embedded silicates (Fig. S5d, e) in porous fabrics were selected by (i) optical microscopy, (ii) SEM imagery, and (iii) WDS elemental mappings of Si, Al, and Fe (15 kV acceleration voltage, a beam current of 60 nA, 6 μm beam diameter, a grid of 500 × 500 steps, step size of 6-8 µm, and 60 s dwell time/step). Quantitative individual silicate analyses were performed at an acceleration voltage of 15 kV and a beam current of 15 nA. Depending on the size of the particles, the beam diameter was adapted between 2 and 10 µm. The counting times of the characteristic X-ray signals were chosen between 8 and 10 s (Na, K), 15 s (Si, Al, Mg, Ca, Fe), and 30 s (Ti, Mn, P). The known effect of alkali-loss under electron bombardment has been checked and was found to be insignificant. For measurements on small grains, counting times of Na and K were reduced to 8 s. Natural (wollastonite: Si, Ca; anorthite: Al; hematite: Fe; olivine: Mg; rhodonite: Mn; albite: Na; sanidine: K) and synthetic (TiO 2 : Ti; ScPO 4 : P) reference materials were used for primary calibration. Routine analyses on secondary standards were performed on the two reference glasses VG-2 and NKT-1G (see Table S6 for details). Matrix correction was applied by using the phi-rho-z algorithm of the CITZAF program 71 . We discarded glass measurements that contain more than 80 wt.% SiO 2 after normalization to 100 wt.%.
SEM and NanoSIMS of incorporated detrital components. Stalagmite sections formed during and after the two Plinian eruption periods of the Mt. Burney volcano at 2.3-2.0 kyrs BP, 4.2-3.8 kyrs BP and during the millennium-scale acidification phase (~3.50 and~3.11 kyrs BP) after the MB 2 eruption 1 , and at~0.73 kyrs BP include laminae rich in detrital components. Representative samples have been investigated with SEM and NanoSIMS. These samples contain the same detritus-rich laminae as analyzed by LA-ICP-MS. We chose NanoSIMS because this technique allows to differentiate embedded inorganic (e.g., 28 Si − , 27 Al 16 O − ) and organic (e.g., 12 C 14 N − ) components from ubiquitous, very low electrically conductive CaCO 3 72,73 .
SEM imaging was conducted with a LEO 435VP and a JEOL JXA-8900 RL at the Geology Department of Trier University and the GZG, University of Göttingen. Replicate samples were analyzed (i) on thin sections and (ii) after dissolution in 5 M ultra-pure HCl and an acidic acetate buffer solution 74 to extract incorporated detritus from distinct laminae. An acceleration voltage of 15 kV was applied to produce high-resolution images in secondary electron and back-scattered electron mode.
Samples for NanoSIMS analyses were taken from the identical blocks the thin sections were prepared of (Fig. S5a). Prior to NanoSIMS measurements, the sample documentation and the selection of AOIs was guided by optical microscopy. We preferentially chose porous laminae, in which abundant organic and inorganic components are embedded in transition zones of different fabrics, so-called accumulation pools (Fig. S5d). NanoSIMS analyses were carried out at the Chair of Soil Science of the Technical University of Munich with a Cameca NanoSIMS 50 L using a Cs + primary ion beam with an impact energy of 16 keV. The selected stalagmite samples were polished and coated with a conductive Au/Pd layer (ca. 30 nm, Polaron Emitech SC7640 sputter coater) to account for charging during the measurements. Charging was additionally compensated using the electron flood gun of the NanoSIMS. Contaminants and the Au/Pd coating layer were locally sputtered away using a high primary beam current (pre-sputtering/implantation), while the reactive Cs + ions were implanted into the sample. The secondary ion yields were enhanced until reaching a steady state. The primary beam (ca. 2 pA) was focused at a lateral resolution of ca. 120 nm and was scanned over the sample, with 12 C − , 16 27 Al + , and 56 Fe + , were measured in combination with C − and O − ions, respectively 43,49 . To accurately separate mass isobars, a suitable mass resolution was achieved with appropriate slits and apertures (D1_3, ES_3, AS_2). The secondary ions were recorded using a dwell time of 1 ms pixel −1 , with 256 × 256 pixels for a 30 × 30 μm field of view with 30 planes per scan. Single planes were corrected for dead time, drift, and accumulated using the ImageJ software 75 in combination with the Open-MIMS plugin 76 .
The sensitivity of SIMS spans more than five orders of magnitude and the SIMS secondary ion yield notably depends on the ionization potential and the electron affinity of each species 73 . The secondary ion intensities do not reflect the relative element concentrations of the sample since different species have largely different ionization probabilities 42 . Furthermore, the ionization probability is very specific to the composition and chemistry of the surrounding matter, the so-called matrix effect 73 . Due to this effect, the ion yield of one element can be enhanced while the ion yield of another element can be suppressed under the same sputtering conditions 49,73 . Caution is therefore needed when interpreting NanoSIMS data on a specific mineral phase. As the count rate of an ion is not directly proportional to the elemental concentration in the mineral, a linear interpolation based on its elemental concentration is not expected 42,43 .
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data sets generated during the current study can be obtained from the Zenodo repository (https://doi.org/10.5281/zenodo.5856380). The LA-ICP-MS data may also be found in the online version of this article (Supplementary Data 1).