A variety of geological proxies has been developed to reconstruct the history of the low-latitude hydrological cycle, which plays a crucial role in the global moisture and heat transport. In the past decade, the strong similarity of δ18O variations between atmospheric O2 (δ18Oatm) and Chinese stalagmite, either on orbital or on millennial time scales, has raised the debate on the correlation between δ18Oatm and the low-latitude hydrology (1–5). As a chemical signal of the atmosphere, δ18Oatm is an integrated result of different processes that occurred over the globe, which is able to reflect biogeochemical and thus climatic changes at a planetary scale compared with regional proxies.
Interpreting the climate significance of δ18Oatm, however, turns out not to be straightforward. The δ18Oatm value is determined by the oxygen isotopic fractionation associated with photosynthesis and respiration of biosphere as well as the hydrological cycle, which is today more enriched by 23.5 ± 0.3‰ than oceanic water. This isotopic offset is traditionally referred to as the Dole effect. The respiration of both terrestrial and marine biosphere preferentially consumes 16O over 18O, causing a nearly equivalent isotope effect of ~19‰ [e.g., (6)]. This is the biggest source of the Dole effect. On land, plant transpiration can cause an isotopic enrichment of leaf water relative to soil water, resulting in a globally integrated photosynthetic Dole effect of 4 to 8‰ [e.g., (6, 7)]. In the marine realm, O2 produced by phytoplankton shows an 18O enrichment of less than 6‰ with respect to ambient seawater (8). The hydrological cycle, however, changes δ18Oatm in an opposite direction to the aforementioned two processes. This is because the ocean-to-land moisture transfer can always result in an isotopic depletion of terrestrial rainfall with respect to seawater. Rainwater is subsequently used by the terrestrial biosphere, which thus decreases the Dole effect (2, 6).
The fluctuation of the Dole effect (∆DE) over the past 800 thousand years (ka) has been reconstructed by removing the imprint of seawater δ18O from ice core–derived δ18Oatm records [e.g., (3, 6)]. During the mid-to-late Pleistocene, varying global ice volume is the first-order reason causing mean oceanic water δ18O changes. Accordingly, the reconstructed or the modeled ice-volume isotopic effect (δ18Osw) (9, 10) was subtracted from δ18Oatm to obtain the ∆DE [e.g., (3)]. Using this approach, the reconstructed ∆DE shows strong variance in the precession bands with an amplitude of ~1.2‰, believed related to low-latitude hydrological changes (3, 6). However, the ∆DE record contains additional and important variance at the ~100-ka periodicity, suggesting a substantial influence of ice-sheet volume on the ∆DE, possibly via changing terrestrial productivity at mid-to-high latitudes (3). This, therefore, impedes the use of the ∆DE as a proxy for the…