Glycerol dialkyl glycerol tetraethers (GDGTs) in high latitude Siberian permafrost: Diversity, environmental controls, and implications for proxy applications.

• GDGT sources, distribution and environmental controls examined in 129 permafrost samples. • Isoprenoid GDGTs not controlled by physicochemical parameters. • Branched GDGTs show (anti-)correlations with %TOC, %TN, and soil pH. • Branched GDGT distributions yield a permafrost-specific soil pH calibration. • Deposit types show distinct GDGT fingerprints, which may bias sedimentary proxy records. Archaeal and bacterial glycerol dialkyl glycerol tetraethers (GDGTs) are globally abundant in soils under various climatic conditions, but little is known about their sources, relative distribution, and environmental controls on their diversity in high latitude permafrost deposits. Thus, constraints on GDGT-based proxies, such as methylation of branched GDGTs (MBT) or cyclization of branched GDGTs (CBT) used to infer mean annual temperature or soil pH, are also sparse. We investigated the GDGT diversity in typical North Siberian permafrost deposits including the active layer of polygonal tundra soils (seasonally frozen ground), fluvial terrace/floodplain sediments, Holocene and Pleistocene thermokarst sediments, and late Pleistocene Ice Complex (Yedoma). Our data show that isoprenoid GDGTs are produced by both methanotrophic and methanogenic Euryarchaeota, as well as Thaumarchaeota, but their abundance does not seem to be controlled by the investigated physicochemical parameters including %TOC, %TN, and soil pH. Branched GDGTs (brGDGTs) show similar distributional changes to those observed in other high latitude soil samples, i.e., a dominance of pentamethylated and hexamethylated brGDGTs, likely reflecting the adaptation to low temperatures and a positive correlation of the degree of cyclization with soil pH. Specifically, brGDGT-IIIa correlates positively with %TOC and %TN and negatively with soil pH, while brGDGT-Ib and brGDGT-IIb correlate negatively with %TOC and %TN and positively with pH. Moreover, we observe a negative correlation between 5-methyl and 6-methyl brGDGTs without cyclopentane moieties (except brGDGT-IIIa), but this anticorrelation does not seem to be related to the investigated physicochemical parameters. The observed brGDGT distribution yields a permafrost-specific soil pH calibration, p H ' PF = 1.8451 × C B T ' PF + 8.5396 (r2 = 0.60, RMSE = 0.074; n = 109). The different investigated deposit types show some distinct GDGT distributional changes and appear to be distinguishable based on the relative abundance of crenarchaeol, GDGT-0/(crenarchaeol + GDGT-0) ratios, and CBT' PF values, although we also observe strong heterogeneity for each deposit type. In particular, Yedoma and the active layer of polygonal tundra soils represent distinct endmembers, which differ from each other, as well as from fluvial terrace/floodplain sediments and thermokarst sediments, while the latter two deposit types have similar GDGT fingerprints that are not easily distinguishable. Yet, the observed GDGT distributional differences have implications for GDGT proxies analyzed in aquatic suspended matter and sediments. Quantitative estimates of permafrost erosion, as well as soil pH inferred using BIT indices or CBT' PF , respectively, may be biased by changing relative contributions of different deposit types (carrying their respective GDGT signals) to the exported permafrost OC, particularly from Yedoma and the active layer of polygonal tundra soils. [ABSTRACT FROM AUTHOR]