Heterogeneity of speleothem records of North American monsoon rainfall: cave or climate?

Understanding the full range of hydroclimate variability in the southwestern United States during the late Holocene requires information from diverse paleoclimate archives. As a complement to regional tree-ring records, which have been used to reconstruct precipitation over the last few centuries, we present new high-resolution oxygen isotope records from two southern Arizona caves: Cave of the Bells (COB; 31.72°N, 110.77°W) and Ft. Huachuca Cave (FHC; 31.53°N, 110.42°W). Both records benefit from sub-decadal resolution (~4.5yrs/sample) and span at least the last 4000 years. Extensive monitoring over the last decade suggests that the COB record reflects a balance between winter and summer rainfall; dripwaters largely come from winter rainfall, but strong monsoons can affect the average dripwater oxygen isotopic value. Monitoring at FHC has been less frequent, but also highlights the dominance of winter precipitation on the overall moisture balance of the cave. We therefore interpret more positive values in the speleothem records as a strengthening of the North American monsoon relative to winter cyclonic activity. Both speleothem records are high-resolution and well constrained chronologically, which should provide robust information about local decadal-scale precipitation fluctuations at each cave site. In fact, an additional speleothem record from COB, which spans 6500-3500 years before present (Cole et al. in prep), is highly and significantly correlated within age model error with the COB record discussed here, further suggesting the caves record local climate. On timescales shorter than a few centuries, however, the COB and FHC records are neither significantly correlated nor coherent with one another. Extensive Monte Carlo age-modelling rules out the possibility that inter-site differences reflect errors in the chronologies of the two records, and instead suggests that COB and FHC are fundamentally discordant on shorter timescales (10 to 300 years). We argue that this disagreement is at least partly due to spatial heterogeneity of rainfall in the arid Southwest during summer. Cave processes, such as the kinetics of calcite precipitation at the speleothem site, may also play a role in either or both caves. On longer timescales (300 to 1000 years), the COB and FHC records are in better agreement with one another and they show that the North American monsoon has weakened over the past 4000 years.