Dipole Response of Millennial Variability in Tropical South American Precipitation and δ18Op during the Last Deglaciation. Part II: δ18Op Response.

Understanding the hydroclimate representations of precipitation δ¹⁸O (δ¹⁸O_p) in tropical South America (TSA) is crucial for climate reconstruction from available speleothem caves. Our preceding study (Part I) highlights a heterogeneous response in millennial hydroclimate over the TSA during the last deglaciation (20–11 ka before present), characterized by a northwest–southeast (NW–SE) dipole in both rainfall and δ¹⁸O_p, with opposite signs between central-western Amazon and eastern Brazil. Mechanisms of such δ¹⁸O_p dipole response are further investigated in this study with the aid of moisture tagging simulations. In response to increased meltwater discharge, the intertropical convergence zone (ITCZ) migrates southward, causing a moisture source location shift and depleting the isotopic value of the vapor transported into eastern Brazil, which almost entirely contributes to the δ¹⁸O_p depletion in eastern Brazil (SE pole). In contrast, the moisture source location change and local condensation change (due to the lowering convergence level and increased rain reevaporation in unsaturated subcloud layers) contribute nearly equally to the δ¹⁸O_p enrichment in the central-western Amazon (NW pole). Therefore, although a clear inverse relationship between δ¹⁸O_p and rainfall in both dipole regions seems to support the “amount effect,” we argue that the local rainfall amount only partially interprets the millennial δ¹⁸O_p change in the central-western Amazon, while δ¹⁸O_p does not document local rainfall change in eastern Brazil. Thus, the paleoclimate community should be cautious when using δ¹⁸O_p as a proxy for past local precipitation in the TSA region. Finally, we discuss the discrepancy between the model and speleothem proxies on capturing the millennial δ¹⁸O_p dipole response and pose a challenge in reconciling the discrepancy. Significance Statement We want to comprehensively understand the hydroclimate footprints of δ¹⁸O_p and the mechanisms of the millennial variability of δ¹⁸O_p over tropical South America with the help of water tagging experiments performed by the isotope-enabled Community Earth System Model (iCESM). We argue that the millennial δ¹⁸O_p change in eastern Brazil mainly documents the moisture source location change associated with ITCZ migration and the change of the isotopic value of the incoming water vapor, instead of the local rainfall amount. In contrast, the central-western Amazon partially documents the moisture source location shift and local precipitation change. Our study cautions that one should not simply resort to the isotopic “amount effect” to reconstruct past precipitation in tropical regions without studying the mechanisms behind it. [ABSTRACT FROM AUTHOR]