Validating lipid biomarker-based sea surface temperature proxies Uk'37 and TEX86 in the Mediterranean Sea

Abstract

Sea surface temperature (SST) reconstructions based on lipid biomarker proxies Uk’37 and TEX86 show offsets from expected values in the Mediterranean Sea compared to other parts of the global ocean. In surface sediments, SSTs based on Uk’37 are often 2–4°C colder than mean annual values, whereas SSTs based on TEX86 are generally 2–6°C warmer. Uk’37 is known to be prone to biases towards the production season of alkenones, a non-linear temperature response to temperatures >24°C and resuspension of alkenones when exported to the sea floor. TEX86, on the other hand, can be biased by the input of isoGDGTs produced by non-Thaumarchaeota or Thaumarchaeota living in the deep sea. This study assesses sources of bias in these proxies in the Mediterranean Sea by using sediment traps moored in the Bannock Basin at 500 m, 1500 m and 2500 m depth, collecting alkenones and isoGDGTs in a near-continuous time serie between 2008 and 2011. The data show that SST estimates based on Uk’37 underestimate mean annual values by 4°C in the upper trap and 6°C in the middle and lower trap. This colder bias is caused by the bias towards the production season of the alkenones, proved by the high alkenone fluxes in winter. SST estimates based on TEX86 overestimate mean annual values by 8-9°C at each sediment trap. This warmer bias is a result of isoGDGT contributions by Group I.1b Thaumarchaeota and deep-water dwelling Thaumarchaeota, proved by [Cren’]/([Cren] + [Cren’]) ratio values of >0.04 and the [GDGT-2]/[GDGT-3] ratio values of >5, respectively.