Radiolarian-based transfer functions for estimating mean surface ocean temperatures and seasonal range

New radiolarian-based transfer functions to estimate sea surface temperature (SST) and seasonal range are presented. The transfer fonctions are based on the approach originated by Imbrie and Kipp [1971]. The transfer functions differ from previous studies in the following three important ways: (1) extensions to Q-mode factor analysis provide an objective method to cull species in the very diverse radiolarian population; (2) a log transform of the relative abundance data is used to normalize the species percent data; and, (3) rather than writing equations for specific seasons, which are not independent data sets, statistically independent equations are developed to predict mean annual sea surface temperatures as well as seasonal temperature range. One hundred and seventy surface sediment samples from the Pacific Ocean are used to generate the SST and season temperature range transfer functions. All samples were counted using a standardized radiolarian taxonomy. Forty one radiolarian species were used in the final regression equation. Q-mode factor analysis of this data set identified seven assemblages. These assemblages, the tropical, transitional, Antarctic, Bering Sea, western Pacific, central gyre, and eastern boundary current, are named for the oceanographic regions where these assemblages are important. The seven assemblages are used in a regression analysis to predict SST and seasonal temperature range. The standard error of estimates for both mean SST and seasonal temperature range is 1.6°C. Comparison between radiolarian-based SSTs and SST estimates from alkenone U_k³⁷ in a 20,000 year long record from the northeast Pacific shows excellent agreement. Comparison of mean SST estimates for the last glacial maximum (LGM) based on radiolarian and foraminifera in 10 eastern equatorial Pacific also show excellent concordance. These new LGM estimates suggest that the original Climate: Long-Range Investigation Mapping, and Prediction (CLIMAP) reconstruction for this region underestimated surface ocean cooling.