A comparison of western Great Basin paleoclimate records for the last 3000 yr: Evidence for multidecadal- to millennial-scale drought

This paper summarizes the hydrological variability in eastern California (cen-tral Sierra Nevada) for the past 3000 yr based on three distinct paleoclimate proxies, δ18O, total inorganic carbon (TIC), and magnetic susceptibility (chi). These proxies, which are recorded in lake sediments of Pyramid Lake and Walker Lake, Nevada, and Mono Lake and Owens Lake, California, indicate lake-level changes that are mostly due to variations in Sierra Nevada snowpack and rainfall. We evaluated lake-level changes in the four Great Basin lake systems with regard to sediment-core locations and lake-basin morphologies, to the extent that these two factors influence the paleoclimate proxy records. We documented the strengths and weaknesses of each proxy and argue that a systematic study of all three proxies together significantly enhances our ability to characterize the regional pattern, chronology, and resolution of hydrological variability. We used paleomagnetic secular variation (PSV) to develop paleomagnetic chronostratigraphies for all four lakes. We previously published PSV records for three of the lakes (Mono, Owens, Pyramid) and developed a new PSV record herein for Walker Lake. We show that our PSV chronostratigraphies are almost identical to previously established radiocarbon-based chronologies, but that there are differences of 20–200 yr in individual age records. In addition, we used eight of the PSV inclination features to provide isochrons that permit exacting correlations between lake records. We also evaluated the temporal resolution of our proxies. Most can document decadal-scale variability over the past 1000 yr, multidecadal-scale variability for the past 2000 yr, and centennial-scale variability between 2000 and 3000 yr ago. Comparisons among our proxies show a strong coherence in the pattern of lake-level variability for all four lakes. Pyramid Lake and Walker Lake have the longest and highest-resolution records. The δ18O and TIC records yield the same pattern of lake-level variability; however, TIC may allow a somewhat higher-frequency resolution. It is not clear, however, which proxy best estimates the absolute amplitude of lake-level variability. Chi is the only available proxy that records lake-level variability in all four lakes prior to 2000 yr ago, and it shows consistent evidence of a large multicentennial period of drought. TIC, chi, and δ18O are integrative proxies in that they display the cumulative record of hydrologic variability in each lake basin. Tree-ring estimations of hydrological variability, by contrast, are incremental proxies that estimate annual variability. We compared our integrated proxies with tree-ring incremental proxies and found a strong correspondence among the two groups of proxies if the tree-ring proxies are smoothed to decadal or multidecadal averages. Together, these results indicate a common pattern of wet/dry variability in California (Sierra Nevada snowpack/rainfall) extending from a few years (notable only in the tree-ring data) to perhaps 1000 yr. Notable hydrologic variability has occurred at all time scales and should continue into the future.