Local and regional climatic controls on high-resolution rainfall and cave dripwater oxygen isotopes in northern Borneo

The relationship between climate variability and rainfall isotopic variability is poorly constrained, especially in the tropics, where many key high-resolution paleoclimate records rely on past rainfall isotopes as proxies for hydroclimate. Multi-year, high-resolution monitoring studies of modern rainfall isotopes are needed in order to inform interpretations of isotope-based paleoclimate reconstructions, yet such studies are rarely undertaken. Here we present a daily-resolved, 5-yr-long timeseries of rainfall oxygen isotopes (δ18O) and 3-yr-long timeseries of bi-weekly dripwater δ18O from Gunung Mulu National Park, located in northern Borneo (4° N, 114° E), which we compare to instrumental records of local precipitation amount as well as globally-gridded climate variables. Daily variations in rainfall δ18O, ranging from +0.5 to -18.5‰, exhibit an inverse relationship with daily local precipitation amount (R = -0.20, p < 0.05), evidence of the 'amount effect'. We observe a stronger amount effect relationship when we compare daily rainfall δ18O values to the average of local precipitation amount over the preceding week (R = -0.46, p < 0.01). Similarly strong correlations for monthly averaged Mulu rainfall δ18O and precipitation amount (R = -0.57, p < 0.01) highlight the time-integrative nature of rainfall δ18O. A relatively weak bi-model seasonal cycle in rainfall δ18O of 2-3‰ is overshadowed by large, negative δ18O anomalies of up to 16‰ that persist for several days every 30-90 days, closely associated with the Madden-Julian Oscillation. Interannual rainfall δ18O anomalies of 6-8‰ are highly correlated with indices of the El Niño Southern Oscillation (ENSO), such that relatively dry El Nino conditions are associated with higher rainfall δ18O values, and vice versa during La Nina events. Mulu rainfall δ18O is highly correlated to basin-scale interannual climate variability, highlighting the advantages of using water isotope-based proxies, such as stalagmite δ18O, for hydroclimate reconstruction. Mulu cave dripwater δ18O samples taken from three distinct sites during the study period of the rainfall δ18O timeseries strongly reflect the interannual signal of amount-weighted Mulu rainfall δ18O (R = 0.90; 0.88; & 0.64). The dripwater timeseries most closely tracks a 2-3-month average of the rainfall δ18O timeseries, with a 1-2 month lag evident in the dripwaters. There is no difference in the absolute value of rainfall δ18O and dripwater δ18O over the study period, suggesting that evaporation is not an important source of dripwater δ18O variability in the Mulu karst system. We conclude that the three Mulu drips studied here are characterized by residence times of 2-3 months, over which time appreciable mixing occurs in the epikarst. While the 6‰ interannual variations of Mulu rainfall δ18O are largely preserved in both the fast [35 drips per minute (dpm)] and slow (7 dpm) dripping sites in Wind Cave, signal amplitude is reduced to approximately 3‰ at the moderate (14 dpm) dripping site in Lang's Cave. Overall, our study illustrates the great potential to reconstruct past ENSO variability from sufficiently fast-growing northern Borneo stalagmites, and more generally, strongly supports the interpretation of northern Borneo stalagmite δ18O records as proxies for regional hydroclimate variability.