Refinement and application of a new paleotemperature estimation technique

Abstract

A recently developed technique for determining past sea surface temperatures (SST), based on an analysis of the unsaturation ratio of long chain C37 methyl alkenones (Uk37) produced by Prymnesiophyceae phytoplankton, has been applied to late Quaternary sediment cores. Previous studies have shown that the Uk37 ratio of these alkenones is linearly proportional to the sea-water temperature in which the plankton grow, both in culture and water column samples. Furthermore, a reasonable correlation has been found between open ocean paleo-SST estimates based on Uk37 values and those derived from δ180, for the period spanning approximately the last 100,000 years (Brassell, 1986b). These results indicate this technique has potential for determining paleo-SST from analysis of alkenones extracted from marine sediments. In order to apply the Uk37 method quantitatively, it is necessary to calibrate the method for sediment samples, and to assess how well the alkenones maintain their temperature signal under some common conditions of sediment deposition and sample handling. It is also necessary to determine the method's usefulness downcore, that is, back in time, by comparing it to established methods. This study examined the effect on Uk37 of conditions that cause dissolution of carbonates in the sediment, and methods of storage and 'sample handling. These are two problems that must be resolved before the method can be applied rigorously and quantitatively to sediments for paleotemperature estimations. A comparison of duplicate samples collected and stored frozen versus those stored at room temperature for up to four years showed no resolvable differences in Uk37. Laboratory experiments of carbonate dissolution indicated there is no effect on Uk37 values under the acidic conditions that dissolve carbonates. Initial field results support this, but indicate more studies are necessary. The Uk37 "thermometer" was calibrated by analyzing Uk37 in coretops from widely varying open ocean sites. Sediment values of Uk37 reflected overlying SST for the appropriate season of the phytoplankton bloom, which for this study was assumed to be summer in high latitudes. These results fall on the same regression line for culture and water column samples derived by Prahl and Wakeham (1987), indicating that their equation (Uk37 = 0.033 T + 0.043) is suitable for use in converting Uk37 values in sediments to overlying SST for the season of coccolith bloom. Using this calibration for sediments, the Uk37 paleotemperature method can be quantitatively applied down core to open ocean sediments. In the Equatorial Atlantic, Uk37 temperature estimates were compared to those obtained from δ18O of the planktonic foraminifer Globigerinoides sacculifer, and planktonic foraminiferal assemblages for the last glacial cycle. The alkenone method showed ~1.56°C cooling at the last glacial maximum. This is about half the decrease shown by both the isotopic method ( ~3.40°C) and foraminiferal assemblages (~3.75°C), implying that, if Uk37 estimates are correct, SST in the equatorial Atlantic was only reduced slightly in the last glaciation. In the Northeast Atlantic, Uk37 temperature estimates show a profile downcore which is similar to the estimates from foram assemblages but with a constant offset toward warmer values throughout the core. Uk37 SST estimates are substantially warmer than foraminiferal estimates at all times, which may indicate inaccuracy in Uk37 temperatures at this site. Uk37 indicates a SST of 12°C for the late glacial and 18°C for the Recent, whereas assemblages give estimates of 9°C and 13°C, respectively. At 12,700 yrs BP, the Uk37 and foram assemblage methods indicate a 2°C warming. A temperature change of 2°C can account for only 0.44°/oo of the observed 1.2°/oo δ18O signal, indicating that the additional 0.8°/oo change in δ18O must result from changes in surface salinity most likely due to a meltwater lid. Uk37 estimates show the major temperature shift from glacial to interglacial temperatures occured at about 9,000 yrs BP disagreeing with assemblage data which shows the shift to Holocene values at about 12,700 yrs BP. If Uk37 temperature estimates are accurate, this disagreement may reflect differing habitats of flora and fauna under the unusual sea surface conditions in this area during the deglaciation.