Abstract
Investigating controls on past variability of South American hydroclimate is critical to assessing its response to future warming scenarios. δ18O records from South America offer insight into past variability of the South American Monsoon System (SAMS). The controls, however, on precipitation δ18O values can be decoupled from precipitation amount at a given site and, thereby, limit local moisture condition reconstructions. Here we use a principal components analysis to assess the coherence of speleothem and lake core Holocene δ18O records in tropical and subtropical South America to evaluate the extent to which δ18O variability reflects changes in SAMS intensity at different sites across the region. The main mode of variability across Holocene δ18O records (PC1) closely tracks austral summertime insolation, consistent with existing work. Sites towards the periphery of the continent are heavily weighted on PC1, whereas interior sites as not. Further δ18O variability at interior sites bear little similarity to each other and implicate controls, beyond regional monsoon intensity, on these δ18O records. Further, we develop speleothem 87Sr/86Sr records spanning the Holocene from Tamboril Cave (Brazilian Highlands), Paraíso Cave (eastern Amazon Basin), Jaraguá Cave (Mato Grosso do Sul Plateau), and Botuverá Cave (Atlantic coastal plain) to investigate coupling between reconstructed monsoon variability (reflected by PC1) and local moisture conditions (interpreted from 87Sr/86Sr records). We interpret speleothem 87Sr/86Sr variability as a proxy of local moisture conditions, reflecting the degree of water-rock interaction with the cave host rock as driven by variations in water residence time. Speleothem 87Sr/86Sr records from all the sites, except Botuverá cave, do not co-vary with PC1, suggesting that local moisture conditions do not necessarily follow variations in regional monsoon intensity at these interior sites. These speleothem 87Sr/86Sr records generally suggest dry mid-Holocene conditions relative to the early- and late-Holocene, consistent with interpretations of other paleo-moisture records in the region, but timing of wet-dry transitions varies between sites. These results highlight that controls, in addition to SAMS variability, might influence δ18O variability. Further, our results suggest spatially variable local moisture conditions at interior sites that do not uniformly respond to regional monsoon intensity, and stress the need for δ18O-independent reconstructions of moisture conditions.