EXPLORING THE CONSISTENCY OF ALKENONE AND FAUNAL-BASED SEA SURFACE TEMPERATURE RECONSTRUCTIONS FROM THE SOUTHWEST PACIFIC DURING THE PLEISTOCENE

Geochemical and biotic proxies are useful tools for the paleoclimate community in reconstructing past climate conditions in order to better contextualize future changes in climate. Because not every proxy is viable at all geographic locations, sea surface temperature (SST) estimates from multiple proxies are often compiled into global or regional climate reconstructions with the implicit assumption that estimates derived from different proxies can be used interchangeably. However, limited evidence currently exists to support the validity of this assumption. Using paleotemperature data from sediment collected from ODP Site 1125 (42˚33’S, 178˚10’W, 1365 meters water depth) in the southwest Pacific, we provide a ~1Myr, orbital-scale SST proxy comparison of novel alkenone-derived SST data to previously published faunal assemblage SST data from the Pleistocene. These U^K’₃₇ and faunal assemblage SST records show strong structural similarity and yield remarkably similar estimates for basic climate metrics across each of the time series, including mean (14.9°C versus 14.4°C, respectively), standard deviation (both 1.6°C), and range (7.8°C versus 7.5°C, respectively). Spectral analysis reveals that the alkenone and faunal records are spectrally similar, both containing a dominant 100k beat, with additional spectral power in the ~41k and 23k bands. Regression analysis yields a fairly strong (r=0.63) and statistically significant correlation between the two SST records. Our preliminary results indicate that these two proxies would yield very similar estimates for the paleoclimate metrics most commonly used in empirical paleoclimate reconstructions that seek to document the evolution of climate over time. However, significant disparities between SST estimates exist for some time intervals, particularly during glacial times. Thus, treating these proxies interchangeably when employing the time slab or time slice approaches that are typically employed in modelling studies could be problematic.