Spatial and temporal variability of Crenarchaeota in Lake Superior and implications for the application of the TEX86 temperature proxy

We present results of a study of Crenarchaeota in the water column of Lake Superior, in which their vertical distribution was examined through in situ filtration of suspended particulate matter (SPM) and temporal variability through a three-year time-series sediment trap deployment. We observed that Crenarchaeotal cells and membrane lipids are distributed throughout the water column during overturning conditions, but mainly reside in the hypolimnion when Lake Superior is thermally stratified. TEX86 derived temperatures in suspended particulate matter were in very good agreement with the actual water temperatures at times and depths that the SPM was sampled. Fluxes of isoprenoid crenarchaeotal membrane lipids towards the sediment mainly occur in winter and late spring/early summer and starts when the lake is overturning. We observed a strong covariance of all the fluxes analyzed in the sediment traps suggesting a role of resuspension and sediment focusing to the study site. A lack of seasonality in the trend of TEX86 derived temperatures in settling particles shows that the crenarchaeotal membrane lipid flux during the stratified period is likely dominated by lipids that have been produced in the hypolimnion. Both TEX86 derived temperatures from surface sediments and the flux weighted averaged temperatures from the sediment traps yield temperatures that are close the average hypolmnetic water temperature as well as the temperature observed during the start of overturning. These data suggest that the TEX86 from the sediments of Lake Superior largely reflects a subsurface water temperature and not an annual mean surface water temperature. The sensitivity of hypolimnetic water temperatures of Lake Superior to changes in atmospheric temperatures surrounding Lake Superior are small, even when large changes in atmospheric temperatures would occur over a century time scale. Therefore caution needs to be applied when interpreting trends of TEX86 temperatures from Lake Superior, as unless the niche of export production of crenarchaeotal membrane lipids moved towards the epilimnion, temperature changes are expected to be relatively small compared to the uncertainties surrounding the TEX86 method. Our study provides insight into where and when are the lipids that make up the TEX86 proxy produced in the water column that are subsequently incorporated into the sediment record, thus placing constraints on what kind of temperature is actually represented by the TEX86 proxy in sediment records of Lake Superior or similar lacustrine systems.