OPTiMAL: a new machine learning approach for GDGT-based palaeothermometry

Abstract

Abstract. In the modern oceans, the relative abundances of glycerol dialkyl glycerol tetraether (GDGT) compounds produced by marine archaeal communities show a significant dependence on the local sea surface temperature at the site of deposition. When preserved in ancient marine sediments, the measured abundances of these fossil lipid biomarkers thus have the potential to provide a geological record of long-term variability in planetary surface temperatures. Several empirical calibrations have been made between observed GDGT relative abundances in late Holocene core-top sediments and modern upper ocean temperatures. These calibrations form the basis of the widely used TEX86 palaeothermometer. There are, however, two outstanding problems with this approach: first the appropriate assignment of uncertainty to estimates of ancient sea surface temperatures based on the relationship of the ancient GDGT assemblage to the modern calibration dataset, and second, the problem of making temperature estimates beyond the range of the modern empirical calibrations (> 30 ∘C). Here we apply modern machine learning tools, including Gaussian process emulators and forward modelling, to develop a new mathematical approach we call OPTiMAL (Optimised Palaeothermometry from Tetraethers via MAchine Learning) to improve temperature estimation and the representation of uncertainty based on the relationship between ancient GDGT assemblage data and the structure of the modern calibration dataset. We reduce the root mean square uncertainty on temperature predictions (validated using the modern dataset) from ∼ ±6 ∘C using TEX86-based estimators to ±3.6 ∘C using Gaussian process estimators for temperatures below 30 ∘C. We also provide a new quantitative measure of the distance between an ancient GDGT assemblage and the nearest neighbour within the modern calibration dataset, as a test for significant non-analogue behaviour.

Archaeal lipids trace ecology and evolution of marine ammonia-oxidizing archaea

“…We interpret GDGTs with unusually high GDGT-2/GDGT-3 ratios as a diagnostic fingerprint of today’s deep AOA communities and consequently lead to nonthermal distribution patterns of GDGT-2/GDGT-3 signals. This view is consistent with arguments that Eocene and Cretaceous GDGT distributions generally are not analogous to modern GDGT assemblages ( 38 ). Interestingly, some Mesozoic and early Cenozoic marine sediments do record the nonthermal pattern (arrows in Fig.…”

“…Intrinsic distributions of archaeal lipid assemblages have rarely been used to study the ecology and evolution of archaea in past oceans. Building upon Taylor et al ( 52 ) and the whole-assemblage machine learning approach of Dunkley Jones et al ( 38 ), we used GDGT-2/GDGT-3 ratios to infer changes in marine archaeal ecology over the geologic past. The clear distinction of GDGT-2/GDGT-3 signatures between shallow and deep marine samples support taxonomic arguments for two distinct groups of AOA in the ocean water column (e.g., refs.…”

“…Despite extensive use of TEX 86 and other GDGT-based metrics to reconstruct ocean temperature history during the Mesozoic and Cenozoic (e.g., refs. 33 – 38 ), GDGT data are rarely used to study archaea themselves (cf. ref.…”

Development of global temperature and pH calibrations based on bacterial 3-hydroxy fatty acids in soils

“…This limitation, common to all linear models, can be overcome using non-parametric methods such as some of the machine learning algorithms (e.g. nearest neighbours or random forest; Dunkley Jones et al, 2020). The reliability of the latter models lies in the fact that they are non-linear, which helps capture the intrinsic complexity of the environmental setting, and that they avoid the regression dilution phenomenon observed in most linear models.…”

“…the alkenone unsaturation index (U k 37 ; Brassell et al, 1986) and inorganic (Mg/Ca ratio and 18 O/ 16 O ratio of foraminifera; Emiliani, 1955;Erez and Luz, 1983) fossil remains were notably developed for the reconstruction of sea surface temperatures. Some of the existing proxies are based on membrane lipids synthesized by certain microorganisms (Eglinton and Eglinton, 2008;Schouten et al, 2013). These microorganisms are able to adjust the composition of their membrane lipids in response to the prevailing environmental conditions in order to maintain an appropriate fluidity and to ensure the optimal state of the cellular membrane (Singer and Nicolson, 1972;Sinensky, 1974;Hazel and Williams, 1990;Denich et al, 2003).…”