Understanding the timing and magnitude of natural variability in the climate system is critical to contextualise current trends and predict future change. Cave deposits have the potential to provide these insights, however ensuring correct interpretations of their geochemical proxies is challenging.

Cave deposits, or speleothems (e.g., stalagmites, stalactites, flowstones), contain a plethora of detailed information about the past. Geochemical proxies from speleothems are used to decipher changes in past rainfall, groundwater recharge, vegetation dynamics, and fire occurrence. These terrestrial archives can be precisely dated back to 600,000 years using U-series disequilibrium (Cheng et al., 2016). Oxygen isotopes (δ¹⁸O) are the most common geochemical proxy investigated in tropical speleothems, and are mainly used to explore changes in hydroclimate. However, controls on speleothem δ¹⁸O are complex, especially across long timescales and dynamic climate regimes. Changes in moisture transport pathways, rainfall amount, convective activity, and monsoon strength are all potential drivers of δ¹⁸O variability.

To address uncertainties around the interpretation of δ¹⁸O as a rainfall indicator, we investigated stalagmites that grew over the same time interval (coeval), and across similar climatic events. Multiple proxy types were measured from each stalagmite, including oxygen-carbon isotopes, Mg/Ca, and Sr/Ca. Combined, these additional proxies can help to decouple controls on δ¹⁸O that may not be related to rainfall amount.

Identifying coeval stalagmites from within a cave is a time-intensive task. Formations located right next to each could easily have grown 10’s of thousands of years apart. Selecting samples for replication can take multiple field trips spanning many years. Our sampling protocol supports cave conservation and focuses mostly on already broken material while promoting limited and targeted collection of in-situ stalagmites (Scroxton et al., 2016). During the ‘initial’ field season, stalagmites thought to have high potential of producing a long paleoclimate record are subsampled by drilling small minicores (~1 cm wide). These mincores are taken near the top and base of the formation. The innermost material from the core is used for dating. Each subsampled stalagmite is tagged and photographed, and its location in the cave carefully noted. The subsamples are then brought back to the lab and sent for U-series dating, allowing us to estimate an upper and lower age of each stalagmite. These reconnaissance dates make it possible to identify stalagmites that grew over the same time interval and map periods of continuous coverage. In the following field season (generally 2 years later), we return to the cave and, with local permissions, remove targeted samples that span critical gaps in the record or provide necessary replication.

The four stalagmites used in this study from southwest Sulawesi, Indonesia were minicored for dating and eventually collected from Abadi cave across the 2009 and 2011 field seasons. A year later, these stalagmites were cut open and prepared for geochemical analysis as part of my PhD thesis.

The formations were retrieved from two different areas of the cave, with two fast-growing stalagmites collected from the ‘basement chamber’ and two slower-growing stalagmites collected from a small side chamber ~10 metres above the basement (see banner photo). The stalagmites form two separate record pairs, and what interesting pairs they make!

The pairs each span a glacial termination: TIV (342–332 ky BP) and TIIIa (220–210 ky BP). Each glacial-interglacial transition is recorded by one fast-growing stalagmite (~6 mm per 100 years) and one slow-growing stalagmite (~2 mm per 100 years).

The fast and slow-growers show excellent δ¹⁸O replication, indicating a well-preserved signal of rainfall δ¹⁸O. The δ¹³C records also have good replication, which is not often achieved in speleothems due to the multitude of drivers influencing carbon isotopes in a cave system (Fohlmeister et al., 2020). We found that the Sulawesi stalagmite δ¹³C signal is likely dominated by glacial-interglacial changes in vegetation productivity and soil CO₂, as it tracks well with global temperature and atmospheric CO₂ reconstructed from Antarctic ice cores.

Mg/Ca was measured on all four stalagmites to provide information about water infiltration through the cave system. In general, when there is less water moving through the cave (drier conditions), there is a higher likelihood of seepage waters degassing into void spaces and calcite precipitating out of solution, prior to reaching a stalagmite surface. This process is known as prior calcite precipitation (PCP). Drier conditions with increased occurrence of PCP result in an increase in stalagmite Mg/Ca. 

Interestingly, Mg/Ca of the slow growers track changes in δ¹⁸O, while Mg/Ca of the fast growers corresponded with changes in δ¹³C. After much cross-examination of the geochemical records and enquiry into karst processes, our explanation to this dilemma comes down to plumbing.

Cave waters feeding the fast-growing stalagmites, in the ‘basement’ chamber, are likely delivered via fractures within the bedrock that have a stronger connection to the soil zone. Mg/Ca variations in these stalagmites appear to be driven by changes in soil CO₂, which can influence bedrock dissolution and PCP processes, thus providing a link between Mg/Ca and δ¹³C in the fast-growing stalagmites. Cave waters feeding the slower-growing stalagmites in the ‘upper corridor’ are likely delivered by seepage flow. These waters also contain a δ¹³C signature driven by changes in soil CO₂, but increased opportunity for PCP along slower flow pathways dominate the Mg/Ca signal. During periods of reduced rainfall, seepage waters move more slowly through the karst, increasing water-rock interaction and exposure to void spaces, resulting in more PCP. Thus, slow growing stalagmites fed by seepage flow are most likely to provide information about rainfall amount in the Abadi Cave system.

After decades of careful sample selection and data analysis, we present evidence that speleothem oxygen isotopes are responding to rainfall amount across major glacial-interglacial climate transitions. We show that, at our site, Mg/Ca of slow-growing stalagmites and the δ¹⁸O of all four fast and slow-growing specimens indicate a shift in monsoon rainfall in the final stage of deglaciation. We also identify time periods where these proxies do not agree, thus δ¹⁸O is not always strictly tied to rainfall amount and, with further exploration, may provide insight into changes in rainfall source and seasonality.

Head to our paper https://doi.org/10.1038/s43247-023-00873-8 to learn more about this work.

References

Cheng, H., Edwards, R.L., Sinha, A., Spötl, C., Yi, L., Chen, S., Kelly, M., Kathayat, G., Wang, X., Li, X., Kong, X., Wang, Y., Ning, Y., Zhang, H. The Asian monsoon over the past 640,000 years and ice age terminations. Nature 534, 640–646 (2016). https://doi.org/10.1038/nature18591 

Scroxton, N., Gagan, M.K., Dunbar, G.B., Ayliffe, L.K., Hantoro, W.S., Shen, C.-C., Hellstrom, J.C., Zhao, J., Cheng, H., Edwards, R.L., Sun, H., Rifai, H. Natural attrition and growth frequency variations of stalagmites in southwest Sulawesi over the past 530,000 years. Palaeogeogr. Palaeoclimatol. Palaeoecol. 441, 823–833 (2016). http://dx.doi.org/10.1016/j.palaeo.2015.10.030 

Kimbrough, A.K., Gagan, M.K., Dunbar, G.B., Hantoro, W.S., Shen, C., Hu, H., Cheng, H., Edwards, R.L., Rifai, H., Suwargadi, B.W. Multi-proxy validation of glacial-interglacial rainfall variations in southwest Sulawesi. Commun. Earth Environ. 4, 1–13 (2023). https://doi.org/10.1038/s43247-023-00873-8