Comprehensive molecular‐isotopic characterization of archaeal lipids in the Black Sea water column and underlying sediments

The Black Sea is a permanently anoxic, marine basin serving as model system for the deposition of organic‐rich sediments in a highly stratified ocean. In such systems, archaeal lipids are widely used as paleoceanographic and biogeochemical proxies; however, the diverse planktonic and benthic sources as well as their potentially distinct diagenetic fate may complicate their application. To track the flux of archaeal lipids and to constrain their sources and turnover, we quantitatively examined the distributions and stable carbon isotopic compositions (δ13C) of intact polar lipids (IPLs) and core lipids (CLs) from the upper oxic water column into the underlying sediments, reaching deposits from the last glacial. The distribution of IPLs responded more sensitively to the geochemical zonation than the CLs, with the latter being governed by the deposition from the chemocline. The isotopic composition of archaeal lipids indicates CLs and IPLs in the deep anoxic water column have negligible influence on the sedimentary pool. Archaeol substitutes tetraether lipids as the most abundant IPL in the deep anoxic water column and the lacustrine methanic zone. Its elevated IPL/CL ratios and negative δ13C values indicate active methane metabolism. Sedimentary CL‐ and IPL‐crenarchaeol were exclusively derived from the water column, as indicated by non‐variable δ13C values that are identical to those in the chemocline and by the low BIT (branched isoprenoid tetraether index). By contrast, in situ production accounts on average for 22% of the sedimentary IPL‐GDGT‐0 (glycerol dibiphytanyl glycerol tetraether) based on isotopic mass balance using the fermentation product lactate as an endmember for the dissolved substrate pool. Despite the structural similarity, glycosidic crenarchaeol appears to be more recalcitrant in comparison to its non‐cycloalkylated counterpart GDGT‐0, as indicated by its consistently higher IPL/CL ratio in sediments. The higher TEX86, CCaT, and GDGT‐2/‐3 values in glacial sediments could plausibly result from selective turnover of archaeal lipids and/or an archaeal ecology shift during the transition from the glacial lacustrine to the Holocene marine setting. Our in‐depth molecular‐isotopic examination of archaeal core and intact polar lipids provided new constraints on the sources and fate of archaeal lipids and their applicability in paleoceanographic and biogeochemical studies.

CL ratios and negative δ 13 C values indicate active methane metabolism.Sedimentary CL-and IPL-crenarchaeol were exclusively derived from the water column, as indicated by non-variable δ 13 C values that are identical to those in the chemocline and by the low BIT (branched isoprenoid tetraether index).By contrast, in situ production accounts on average for 22% of the sedimentary IPL-GDGT-0 (glycerol dibiphytanyl glycerol tetraether) based on isotopic mass balance using the fermentation product lactate as an endmember for the dissolved substrate pool.Despite the structural similarity, glycosidic crenarchaeol appears to be more recalcitrant in comparison to its non-cycloalkylated counterpart GDGT-0, as indicated by its consistently higher IPL/ CL ratio in sediments.The higher TEX 86 , CCaT, and GDGT-2/-3 values in glacial sediments could plausibly result from selective turnover of archaeal lipids and/or an archaeal ecology shift during the transition from the glacial lacustrine to the Holocene marine setting.Our in-depth molecular-isotopic examination of archaeal core and

| INTRODUC TI ON
The Black Sea is a permanently stratified, euxinic basin characterized by anoxic conditions in deep waters below ~100 m below sea level (mbsl).
This basin is traditionally referred to as a modern model system for the deposition of organic-rich sediments under anoxic conditions (Tissot & Welte, 1984).An extensive body of work exists regarding the presence and transformation of a wide array of lipid biomarkers in this basin (Boon et al., 1979;Freeman et al., 1994;Freeman & Wakeham, 1992;King et al., 1998;Kuypers et al., 2003;Repeta et al., 1989;Schubotz et al., 2009;Wakeham, 1989;Wakeham et al., 2003).High concentrations of archaeal lipids were previously observed in the water column and sediments of the Black Sea (Coolen et al., 2007;Schubotz et al., 2009;Sollai et al., 2019;Wakeham et al., 2003Wakeham et al., , 2004) ) and in putative geological analogues to the Black Sea depositional setting such as Cretaceous black shales (Arthur & Sageman, 1994;Kuypers et al., 2001), making it a useful system for studying the sources and fate of archaeal lipids in oxygen-deficient oceanographic settings.
Moreover, the well-characterized Late Quaternary sedimentation history and stratigraphy (Degens & Ross, 1972;Kwiecien et al., 2008;Schmidt et al., 2017) allow an in-depth study of archaeal lipids in relation to past oceanographic conditions.
Archaeal lipids are present either as the comparatively labile intact polar lipids (IPLs; e.g., Biddle et al., 2006) or as their more recalcitrant degradation products, the core lipids (CLs; e.g., Schouten et al., 2002), with the latter being frequently used for paleoenvironmental reconstruction, such as TEX 86 (Equation 1, Schouten et al., 2002), CCaT (Equation 2, Wörmer et al., 2014) and MI (Equation 3, Zhang et al., 2011).TEX 86 is used as a paleothermometer based on the relationship between archaeal lipid cyclization and sea surface temperature (Schouten et al., 2002), while the CCaT was introduced as a surrogate of the TEX 86 for reconstructing sub-annual sea surface temperature records in applications of matrix-assisted laser desorption/ionization coupled to Fourier transform ion cyclotron resonance mass spectrometry to sedimentary archives (Alfken et al., 2020;Wörmer et al., 2014).The MI was proposed to evaluate the relative contribution of GDGTs from anaerobic methaneoxidizing archaea (ANMEs) in settings characterized by anaerobic oxidation of methane (AOM; Zhang et al., 2011).
Both archaeols and GDGTs can be attached to different polar head groups, forming IPLs.IPLs are relatively rapidly degraded after cell lysis even though monoglycosyl (1G) and diglycosyl (2G) derivatives of archaeol and GDGTs appear to be resistant to degradation in sediments (Lengger et al., 2014;Schouten et al., 2010;Xie et al., 2013;Zhu et al., 2021), at least when compared to fatty acid-based phospholipids (Xie et al., 2013).Thus, archaeal IPLs in sediments can be regarded as mixed signals of both fossil and active archaea.
Archaeal lipids, in particular in their core lipid form, have been studied in the context of anaerobic oxidation of methane and aerobic ammonia oxidation in the Black Sea's water column and surficial sediment (Wakeham et al., 2003(Wakeham et al., , 2007)).Based on the comparison of core lipid carbon isotopic composition in suspended particulate matter (SPM) and surface sediments, Wakeham et al. (2003) found that the majority of archaeal core GDGTs reaching the surface sediment of the Black Sea were derived from the chemocline.Meanwhile, the 13 C-depleted lipids produced by anaerobic oxidation of methane in deep anoxic waters were not incorporated into surface sediments (Schouten et al., 2001;Schubert et al., 2006;Wakeham et al., 2003).Zhu et al. (2021) examined the compound distribution and carbon isotopic composition of the archaeal CLs and IPLs in a diverse set of surface and subsurface sediment samples from the Mediterranean Sea, Marmara Sea, and Black Sea.They found that IPL-and CL-GDGTs were predominantly (1) intact polar lipids provided new constraints on the sources and fate of archaeal lipids and their applicability in paleoceanographic and biogeochemical studies.

K E Y W O R D S
archaea, core lipids, intact polar lipids, lipid sources, lipid turnover, lipidomics, stable carbon isotope derived from planktonic sources, demonstrating the resistance of the polar derivatives on millennial time scales to degradation.By contrast, archaeol was predominantly derived from benthic archaea.
Nonetheless, the majority of previous studies involving isotopic analysis of archaeal lipids in the Black Sea did not distinguish between CLs and IPLs (Blumenberg et al., 2004;Schouten et al., 2001;Wakeham et al., 2003).Discriminative analysis of these two compound pools might add new insights regarding their sources and turnover.Moreover, little is known about how archaeal lipid signals from the water column are incorporated into the sedimentary record, preserved on geological timescales, and overprinted by benthic archaeal activity (cf.Lipp et al., 2008).A combination of molecular and stable isotopic analysis could provide useful constraints on lipid producers, fate, and consequently their contribution to carbon and nutrient cycles as well as the imprint of IPLs on paleoenvironmental proxies (e.g., Biddle et al., 2006;Schubotz et al., 2009Schubotz et al., , 2011;;Wakeham et al., 2003Wakeham et al., , 2007;;Zhu et al., 2021).To constrain the sources and turnover of archaeal lipids and their influence on various environmental proxies, we established a comprehensive inventory of molecular-isotopic and distributional data of archaeal lipids (CLs and IPLs) in the stratified water column (N = 9) and a directly underlying 8-m-long sediment core (N = 28) in the southwestern Black Sea.The water column plus sedimentary high-resolution molecular record is complemented by detailed geochemical and stratigraphic information that puts this data into a paleoenvironmental and biogeochemical process-related context.The resulting dataset adds valuable constraints on the taxonomic sources and the turnover of archaeal lipids in the depositional model setting of the Black Sea, including the relationship to frequently applied lipid proxies.

| Sampling and geochemical information
The sampling site (43.538°N and 30.885°E; GeoB15105) is located in the southwestern Black Sea (Figure 1).Samples were retrieved in February 2011 during cruise No. 84, Leg 1 of R/V Meteor (Zabel et al., 2011).The water column chemistry of the study site has been previously reported (Becker et al., 2018;Chuang et al., 2021;Schröder, 2015).In brief, geochemical data reveal a strong vertical stratification of the Black Sea water column (Figure S1).The oxic, suboxic chemocline, and anoxic zones are defined from 0 to 70 mbsl, 70 to 150 mbsl, and 150 mbsl down to the seafloor at a water depth of 1266 mbsl, respectively (Becker et al., 2018).The sediment core shows two distinct lithologies: an upper part (Unit I, 0-415 cmbsf, 0-8 ka) that is gray to olive brown characterized by very-fine-scale-lamination of coccolith ooze (less than 1 mm to 1 cm); a lower part (Unit 2, 415-827 cmbsf, >8 ka) that is light gray to black and homogenous, corresponding to lacustrine deposits (Zabel et al., 2011).The top of Unit II is marked by an organic-rich sapropel deposited coeval to the onset of water column anoxia in the Black Sea (Eckert et al., 2013).Total organic carbon (TOC) increases with depth until 400 cm below seafloor (cmbsf) the sapropel horizon with a TOC content of 3.67% (Figure S2).From 400 cmbsf to the core bottom, TOC decreases gradually from 1.3% to 0.39%.Sulfate (SO 4 2− ) concentration is the highest (16 mmol L −1 ) at the surface sediments and decreases rapidly with increasing depth but remains detectable until around 400 cmbsf (Figure S2).Based on the shape of the sulfide profile and the steep decrease of methane concentrations, we place the sulfate-methane transition zone (SMTZ) between 50 and 160 cmbsf; the sulfate-bearing zone (SZ) is found above.The methanic zone (MZ) underlies the SMTZ and is characterized by low to undetectable SO 4 2− and an accumulation of methane (CH 4 ).Acetate recovery and split into subsamples for onshore and offshore analysis.Sediment samples (N = 28) were collected by using a multicorer and a gravity corer and stored at −20°C in Teflon containers.
All samples are now stored in the MARUM GeoB Core Repository with supporting data archived in PANGAEA Data Publisher for Earth & Environmental Science or related publications (Becker et al., 2018;Chuang et al., 2021;Coffinet et al., 2020;Schmidt et al., 2017).

| Analysis of bulk sediment and pore water
For SPM analysis in the water column, an aliquot of each filter was decalcified under HCl atmosphere for 48 h and transferred into tin capsules.For TOC analysis, an aliquot of sediment was homogenized and freeze-dried before being decalcified with 10% HCl.
Afterward, the sediment was washed with water and freeze-dried.
Eventually, between 10 and 30 mg of decalcified sediment was weighed into tin capsules.δ 13 C values of SPM (δ 13 C SPM ) and TOC (δ 13 C TOC ) were analyzed on a Thermo Scientific Flash 2000 elemental analyzer connected to a Thermo Delta V Plus isotope ratio mass spectrometry (IRMS, Thermo Fisher Scientific, Bremen, Germany).
The carbon isotopic composition of dissolved inorganic carbon (DIC) in pore water was analyzed using a gas bench coupled to a Finnigan MAT 252 mass spectrometer (Thermo Fisher Scientific).
Samples were prepared according to Heuer et al. (2009), such that 100 μL of phosphoric acid (H 3 PO 4 ) was transferred to glass tubes, which were subsequently sealed with butyl septa and plastic caps and purged with helium.The liquid sample was injected into the purged tubes by using a syringe.Samples were allowed to degas CO 2 from the acidified aqueous matrix for 5 h before the carbon isotopic composition of CO 2 was analyzed in subsamples of the gas phase.The concentration and carbon isotopic composition of acetate and lactate were analyzed by liquid chromatography (LC)isolink-IRMS (Thermo Fisher Scientific) as previously described by Heuer et al. (2009).

| Lipid extraction and analysis
We extracted the freeze-dried samples by using a modified Bligh and Dyer method (Sturt et al., 2004).In brief, samples were extracted four times by ultra-sonication in a mixture of dichloromethane (DCM): methanol (MeOH): buffer (1:2:0.8;v/v/v) for 10 min.A phosphate buffer (K 2 HPO 4 , 50 mmol L −1 at pH 7.4) was used the first two times, and a trichloroacetic acid buffer (TCA, 50 g L −1 , pH 2) for another two times.We washed the total lipid extracts (TLEs) with distilled water to remove salts and evaporated the TLEs to dryness under a stream of nitrogen.Archaeal lipids were quantified by injecting an aliquot of the TLEs on a Dionex Ultimate 3000 ultra-high performance liquid chromatography (UPLC) system.Archaeal lipids were separated using reversed-phase UPLC on an ACE3 C18 column (2.1 by 150 mm; 3 μm particle size; Advanced Chromatography Technologies, Aberdeen, Scotland) maintained at 45°C as described previously (Zhu et al., 2013).The UPLC was connected to a Bruker maXis Ultra-High Resolution quadrupole time-of-flight tandem mass spectrometer (qToF-MS) with an electrospray ionization (ESI) ion source operating in positive mode (Bruker Daltonik, Bremen, Germany).The MS was set to a resolving power of 27,000 at m/z 1222.Each analysis was mass-calibrated by loop injections of a calibration standard and correction by lock mass, leading to a mass accuracy of typically less than 1 ppm (Wörmer et al., 2013;Zhu et al., 2013).Ion source and other MS parameters were optimized by infusion of standards (GDGT-0, 1G-GDGT-0, 2G-GDGT-0) into the eluent flow from the UPLC system using a T-piece.Lipid quantification was achieved by injecting an internal standard glycerol trialkyl glycerol tetraether (GTGT, Huguet et al., 2006) along with the samples.Lipid abundances were corrected for response factors of commercially available as well as purified standards as described by Elling et al. (2014).
GDDs were present both in the water column and sediments but most abundant at the chemocline (120 mbsl, 7% of the CL pool).The Nitrososphaerota biomarker MeO-AR was most abundant in oxic and suboxic waters (on average 6%) but was also present in deep anoxic waters and all analyzed sediment samples.BDGT-0 was detected in the anoxic water column (up to 3.8% of the CL pool) and in smaller relative proportions in the sediments (0% to 0.3% of the CL pool).PDGT-0 was only detected in some sediments with extremely low abundance (less than 0.1% of the CL pool).CL-AR was detected in all water column samples but was generally more abundant in the anoxic zone, where it ranged from 9% to 17% of the CL pool; it was present in all sediments with exceptionally high proportions in two of the deepest samples.OH-AR showed a similar distribution as CL-AR except that it was not detected in surface waters and that its proportion was generally lower than that of CL-AR. the IPL-AR pool of the water column, while in sediments mono-and diglycosidic AR (1G-AR and 2G-AR) were dominant (Figure S7).IPL-ARs had low concentrations relative to GDGTs in sediments deposited during the marine phase and were more abundant in the lacustrine (Figure 2b).1G-AR was only detected below 272.5 cmbsf, i.e., within the methanic zone, where it was even the dominant IPL-AR in some horizons.This compound was relatively abundant in the anoxic zone of the water column.IPL-BDGTs ranged from 0% to 2.8% with the highest proportion in the anoxic water column but were undetectable in the upper oxic water column.In sediments, 1G-BDGT-0 was the only representative of the IPL-BDGT pool, ranging from 0% to 2% of the total IPL pool, with an average value of 0.8%.(Figure 2b).In the sediments, 1G-GDGTs were dominant in most of the samples (on average 58%) except for two samples that were dominated by IPL-ARs (Figure 2b).

| Quantitative relationships between IPLs and their corresponding CLs
Total lipids in the water column ranged from 40 to 252 ng L −1 with maximum concentrations in the suboxic zone and at 300 mbsl in the anoxic zone (Figure 2d).Lipid concentrations in sediments ranged from 0.3 to 19.7 μg g −1 dry weight (dw), with maximum levels occurring in and around the sapropel layer (Figure 2d).The pronounced peak around the sapropel layer is consistent with a strong influence of preservation on lipid concentration.
We obtained a more detailed view of the turnover status of IPLs from compound-specific IPL proportions (Figure 3) based on IPL/ (CL + IPL)*100%.Conceptually, high IPL proportions represent zones where biosynthesis of a particular IPL is relatively more important than its degradation and/or, in the case of the water column, its loss due to sinking, while low values indicate zones where its degradation predominates overproduction.The four selected lipids exhibit strikingly different patterns across the combined water column and sediment profile (Figure 3).In the oxic and suboxic zone, the highest proportions of >60% are observed for OH-GDGT-0, while minimum values of just a few percent are found in the sediment's methanic zone.Archaeol is most abundant in the anoxic zone of the water column and the methanic zone of the sediment (particularly in the zone deposited under lacustrine conditions) while the lowest abundances are found in the oxic and suboxic zones.For crenarchaeol, the highest IPL proportions are observed in the suboxic zone and the upper portion of the anoxic zone.The pattern of GDGT-0 resembles that of crenarchaeol with elevated values in the suboxic and upper anoxic zone but its IPL proportions tend to be distinctly lower in the majority of sediment samples.
To explore the impact of varying lipid sources and turnover on various lipid-based proxies, TEX 86 , CCaT, MI, and GDGT-2/-3 were calculated for the pools of CLs, 1G-GDGTs, and 2G-GDGTs

| Bulk sediment geochemistry and volatile fatty acids: carbon isotopic compositions
In the water column, δ 13 C SPM ranged from −24.5 to −28.3‰, with the most positive and negative values at 120 and 900 mbsl, respectively (Figure 5a).The δ 13 C TOC value in sediments stays relatively stable in the marine sediment but shifts to slightly more negative values in the lacustrine sediment.δ 13 C DIC decreased from −11.0‰ in surface sediments to a minimum value of −23.8‰ in the upper SMTZ and below increased gradually to +7.2‰ at 810 cmbsf (Figure 5a).The isotopic compositions of acetate and lactate were measurable from 65 to 810 cmbsf, ranging from −7.8 to −33.8‰ and from −29.0 to −34.2‰, respectively (Figure 5a).

| Sources of archaeal lipids in surface sediments
The characterization of the archaeal lipidome in the Black Sea water column is key to evaluating how planktonic microbial signals are exported to, and preserved within sediments (Wakeham et al., 2003).
Given the fact that the majority of archaeal lipid types detected in this study could have terrestrial, planktonic, and benthic sources, the source assignment of a lipid found in the sediment is not straightforward and requires multiple lines of information.Based on compound distributions from the water column and the underlying sediments in combination with compound-specific isotopic compositions we can, however, derive useful source constraints.
The relative abundance of CLs and IPLs for individual lipids could provide useful information on the activity of the respective archaeal source since typically more than 90% of lipids produced by growing archaeal cultures are IPLs (Becker et al., 2016;Elling et al., 2014;Meador et al., 2015).By contrast, in deep subsurface sediments, for GDGTs, this proportion is around 10% or lower (Liu et al., 2011), while in shallow coastal sediments, this ratio might go up to around 50% (Meador et al., 2015).We observed higher IPL proportions in the water column (38%) than in sediments (22%).This is expected given the comparatively short residence time of lipidbearing particles in the water column in comparison to the millennial timescales in the sediments.Nevertheless, values in the water column are substantially lower than in cultures (Becker et al., 2016;Elling et al., 2014;Meador et al., 2015) and thus suggest substantial turnover of IPLs in SPM.
Based on IPL/CL ratios, we can tentatively identify the major zones of production for certain lipids within the water column (Figure 3): OH-GDGTs are preferentially produced in the oxic and suboxic zone, with a likely major source being Nitrososphaerota (Elling et al., 2014(Elling et al., , 2017)).Crenarchaeol and GDGT-0, again both major lipids in Nitrososphaerota, appear to have their major zone of production in the suboxic and uppermost anoxic zone.The major zone of archaeol production in the water column is in the anoxic zone where its IPL/CL ratios are highest; possible sources include methane-metabolizing archaea.
δ 13 C DIC decreased with water depth (Fry et al., 1991) while δ 13 C BP0 increased from the oxic to the lower suboxic zone (120 mbsl).We suggest that different AOA communities (Rattanasriampaipong et al., 2022) or photoheterotrophic marine group II Euryarchaeota (Lincoln et al., 2014;Ma et al., 2020;Merkel et al., 2015;Sollai et al., 2019;Zhu et al., 2016) are among the possible sources of 13 Cdepleted BP0, noting that the contribution of the latter group to the pool of GDGTs is still under debate (Besseling et al., 2020).We attribute the strong increase of the GDGT-2/-3 ratio in the deep anoxic water column to the presence of anaerobic methane-oxidizing archaea, whose GDGT distribution tends to maximize at GDGT-2 (e.g., Liu et al., 2011).
Archaeol-based lipids are present throughout the water column.
We observed a greater 13 C depletion (~10‰) of Phy than BPcren in the oxic and upper suboxic zones, which indicates distinct sources of archaeol and crenarchaeol.Marine group II Euryarchaeota would be the likeliest candidate for archaeol production in these zones.Zhu et al. (2016) showed that archaeols with 0-4 double bonds are the dominant IPLs in the upper water column and suggested that these compounds are derived from marine group II Euryarchaeota.
IPL-archaeol is present predominantly in the anoxic zone, which is consistent with the previously demonstrated activity of ANMEs (Schubert et al., 2006;Wakeham et al., 2003), i.e., major producers of archaeol-based lipids in the deep anoxic zone.This contribution of methane-oxidizing archaea to the archaeal lipid pool is also evident in the notable 13 C-depletion of both Phy and BP0 in the anoxic zone (Figure 5) and in the MI of CL-GDGTs, which increases steadily with water depth in the anoxic zone (Figure 4).δ 13 C values of both Phy and BP0 in surface sediments suggest that the contribution of archaeol and GDGT-0 from the deep anoxic water column is minimal.
This likely reflects the combination of relatively low lipid production rates compared to shallower waters and the absence of grazers that package archaeal cells into rapidly sinking fecal pellets (Schouten et al., 2001).Nearly identical isotopic compositions of BPcren in the lower suboxic zone and surface sediments (0.5 cmbsf), combined with the rather invariable isotopic compositions downcore, suggest that the majority of crenarchaeol in sediments is derived from the lower suboxic zone.This is in line with previous observations from Mediterranean and Black Sea sediments that suggested the absence of any sedimentary sources for crenarchaeol (Zhu et al., 2021).
Accordingly, smaller deviations of δ 13 C of crenarchaeol from the water column and sediment values, especially in the lacustrine portion of the sediment column, likely reflect past changes in the isotopic composition and concentration of DIC (Hurley et al., 2019;Schouten et al., 2013).
For sedimentary archaeol, the assignment of sources is more complicated.IPL distributions in the sediment substantially differ from those in the water column, with a strong predominance of glycosidic derivatives in the sediment as opposed to phosphatidic derivatives in the water column (Figure S7).This compositional decoupling between sediment and overlying water column suggests that the export of archaeol into the sediments is quantitatively less important than for the major crenarchaeol and GDGT-0 (Wakeham et al., 2003(Wakeham et al., , 2004(Wakeham et al., , 2007)).Interestingly, major contributions of 1Garchaeol to total archaeol-based IPLs coincide with elevated concentrations of methane in both the water column's anoxic zone (Kessler et al., 2006) and the sediment (Figure S7; cf. Figure S2).IPL/CL ratios suggest that deep anoxic waters, i.e., the habitat of ANME archaea, are the major zone of production of archaeol within the water column (Figure 3).As noted earlier, this signal does not appear to be efficiently transmitted into the sediments where the isotopic composition of Phy is around −30‰ (Figure 5).The isotopic composition of Phy in both the water column and sediments in relation to the diagnostic crenarchaeol (Nitrososphaerota) suggests that other archaeal groups are the major source of the archaeol pool (heterotrophic and/or methanogenic contribution).
The isotope values of archaeol in the marine portion of the sediment are more negative than TOC but the profile's shape is inconsistent with that of DIC (Figure 5).This suggests that the contribution of methane-metabolizing autotrophic archaea (e.g., CO 2 reducing methanogens or CO 2 assimilating ANME) to the archaeol pool in this section is minor.Plausible sources of sedimentary archaeol are thus heterotrophic and/or methylotrophic methanogens, yet with comparatively low rates of production as suggested by its low IPL/CL ratios (Figure 3).
Further evidence for the low contribution of terrestrial organic matter also comes from the low BIT values in the marine phase (0.06 ± 0.02, Table S1) and lacustrine phase (0.16 ± 0.06, Table S1).
No discernable differences were observed between δ 13 C CL-Cren and δ 13 C IPL-Cren in both the water column and the 8-m long sediment core.Moreover, δ 13 C IPL-Cren did not relate to the large range of pore water δ 13 C DIC , suggesting that the production of crenarchaeol by autotrophic Nitrososphaerota within the sediment is negligible; with the consequence that crenarchaeol must predominantly be derived from water column sources.The simultaneous shift of 1.5‰ in both δ 13 C CL-Cren and δ 13 C IPL-Cren below 647 cmbsf (Figure 5) presumably represents a shift in the isotopic composition of the DIC pool from marine to lacustrine Black Sea.
This further validates the carbon isotopic compositions of crenarchaeol serving as a proxy for paleo δ 13 C DIC in the ocean (Elling et al., 2019;Schoon et al., 2013).
The concentration profiles of the IPL-and CL-derivatives of both GDGT-0 and crenarchaeol did not provide any clues regarding preferential production or degradation of either GDGT-0 or crenarchaeol (Figure S8).However, crenarchaeol shows consistently higher IPL/CL ratios than GDGT-0 in the sediment (Figure 3), despite relatively similar values of both compounds in the water column.Higher IPL/CL ratios of ether-based glycerol lipids can be explained by either higher production or lower degradation of the IPL relative to the corresponding CL; e.g., IPL-BDGTs, which are presumed to be primarily produced by archaeal activity in sediments, are generally more abundant relative to their corresponding CLs than the accompanying GDGTs, which have substantial sources in the water column (Coffinet et al., 2020).As discussed above, we rule out sedimentary sources for crenarchaeol and consequently, we attribute the higher IPL/CL ratios of crenarchaeol to its retarded degradation relative to GDGT-0.Additional support comes from the 1G-and 2G-CCaT values, which increase steadily in the sediments and thus are consistent with the preferential degradation of 1G-and 2G-GDGT-0 (Figure 4).In fact, the strong correlation between the IPL/CL ratio of GDGT0 and 1G-CCaT (r 2 = 0.71, p < .0001; Figure 6a) suggests that the latter ratio is strongly influenced by more rapid degradation of 1G-GDGT-0 compared to 1G-crenarchaeol.A similarly strong relationship is observed if we limit the IPL/CL ratio to only 1G-GDGT-0 (r 2 = .58,p < .0001; Figure 6b).
If the glycosidic GDGT-0 derivatives would primarily react to CL-GDGT-0, we would expect that the preferential degradation indicated by the downcore increase of 1G-CCaT (Figure 4) would in turn result in a downcore decrease of CL-CCaT.This is, however, not observed (Figure 4), and the CL-CCaT is in contrast to the 1G-CCaT not significantly correlated with IPL/CL ratios (Figure 6c,d).Based on these relationships, we conclude that glycosidic GDGT-0 is more reactive than the corresponding crenarchaeol but hydrolytic cleavage of the glycosidic headgroup to yield the core lipid is not the major degradation pathway.Alternative sinks include the initial formation of glycosidic dibiphytanyl diethers (Meador et al., 2014) and additional downstream products formed after hydrolysis of one or more glycerol ether bonds (Liu et al., 2016(Liu et al., , 2018)), possibly in association with lipid recycling, as previously demonstrated for benthic archaea (Takano et al., 2010).
Lipid recycling could be one explanation for the seemingly higher reactivity of glycosidic GDGT-0 relative to glycosidic crenarchaeol because the former compound is almost universally present among the archaea (Becker et al., 2016;Elling et al., 2017;Koga & Nakano, 2008;Schouten et al., 2013;Zhu et al., 2022) while the latter has so far only been found in Nitrososphaerota (Elling et al., 2017;Pitcher et al., 2011;Sinninghe Damsté et al., 2002, 2012) and may not be as widely recycled by benthic archaeal communities.

| In situ production of archaeal lipids in sediments
In marine surface sediments, GDGT-0 is commonly depleted in 13 C by 0.6‰ relative to crenarchaeol (Pearson et al., 2016).This depletion in GDGT-0 was attributed to additional heterogeneous sources (Pearson et al., 2016) such as terrigenous input (Zhu et al., 2021) or sedimentary in situ production (Biddle et al., 2006;Zhu et al., 2021).The pool of IPL-GDGTs contains a fossil portion (Schouten et al., 2010;Xie et al., 2013;Zhu et al., 2021) but due to its higher reactivity and smaller pool size compared to the CL pool, it is presumably more sensitive to the influence of sedimentary archaea.
In our study, the difference between δ 13 C CL-BP0 and δ 13 C CL-Cren in the top 50 cm of sediment is on average 0.3‰, i.e., within instrumental precision.This suggests that CL-caldarchaeol and CL-crenarchaeol in Black Sea surface sediments are predominantly derived from the same planktonic sources.In contrast to BPcren, BP0 in the sedimentary IPL pool is depleted in 13 C relative to the CL pool by 3 ± 1.8‰.This isotopic offset is strong evidence for the contributions of benthic archaea, which use either 13 depleted organic or inorganic carbon for lipid biosynthesis to the BP0 pool, which is derived from GDGT-0, −1, −2, −3 as well as BDGT-0.We have previously shown on a small set of samples from the Mediterranean and Black Seas that GDGT-0 and BDGT-0 contribute equally to this isotopic offset (Coffinet et al., 2020;Zhu et al., 2021).In contrast to BP0, and in analogy to BPcren, we did not observe significant differences in the isotopic composition of IPL-vs.CL-derived BP1 and BP2 (Figure 5c).Consequently, there is no isotopic evidence for substantial sedimentary production of mono/and bicyclic GDGTs.
To further explore the carbon sources used by benthic archaea, we measured the carbon isotopic compositions of TOC, DIC, acetate, and lactate as potential carbon substrates for lipid biosynthesis (Figure 5a).
We observed that δ 13 C IPL-BP0 did not covary with δ 13 C DIC and δ 13 C acetate but instead had roughly similar trends with δ 13 C TOC .Moreover, δ 13 C IPL-BP0 did not change in the SMTZ, suggesting that the contribution of methanotrophic archaea to the BP0 signal was at most minimal.
This cumulative evidence thus suggests that GDGT-0-producing benthic archaea in Black Sea sediments assimilate mainly organic carbon but little inorganic or methane-derived carbon (δ 13 C of methane: −50 CCaT values as a function of IPL/CL ratios, suggesting the preferential turnover of glycosidic GDGT-0 relative to glycosidic crenarchaeol. to −90‰; Becker et al., 2018).These results are consistent with previous studies that suggested the predominance of heterotrophic archaea in organic-rich sediments (Biddle et al., 2006;Lloyd et al., 2013;Zhu et al., 2021).Based on the knowledge of the isotopic compositions of the various carbon pools, we estimated the heterotrophic contribution to the BP0 pool by using an isotopic mass balance (Equation 4).
Here, δ 13 C IPL-BP0 is the measured value of IPL-BP0, which presumably reflects contributions from isotopically distinct sources, either water column-derived fossil input (δ 13 C Em1 in Equation 4, where Em means endmember) or sedimentary in situ production (mainly heterotrophic; δ 13 C Em2 in Equation 4).The 6‰-value in Equation 4 represents the presumed isotopic fractionation during lipid biosynthesis and is derived from simultaneous isotopic analysis of archaeal cells and BP0 from Peru Margin sediments (Biddle et al., 2006; average value from ten and nine samples, respectively).
f is the fraction of sedimentary in situ production.We note that this approach will not account for potential contributions by autotrophic or mixotrophic benthic archaea, which we cannot fully exclude, in particular for the fraction of BP0 derived from relatively 13 C-depleted BDGTs (Coffinet et al., 2020;Meador et al., 2015;Zhu et al., 2021).Assuming identical isotopic compositions of Nitrososphaerota-derived crenarchaeol and GDGT-0 and given the evidence for the diagenetic stability of δ 13 C CL-BPcren (e.g., Figure 5), we, therefore, use δ 13 C CL-BPcren as an approximation of historical δ 13 C Em1 .Sedimentary heterotrophic microbes mainly utilize small organic molecules such as acetate, lactate, and amino acids, which were enzymatically released from particulate organic matter.While the isotopic composition of pore water acetate is strongly influenced by acetogenesis and acetoclastic methanogenesis, lactate has been shown to resemble isotopic compositions of bulk dissolved organic carbon (Heuer et al., 2009) and can thus serve as the isotopic endmember representation of the carbon pool utilized by heterotrophic archaea (δ 13 C lactate = δ 13 C Em2 ).Through mass balance, the contribution of heterotrophic sedimentary production (f) ranges from 4 to 54%, with an average value of 22% (Table 1).Accordingly, roughly three-quarter of BP0-containing IPLs are derived from the water column and represent a fossil signal.This range is consistent with previous estimates based on the degradation kinetics of glycosidic ether lipids for the subsurface depth range of our Black Sea samples (Xie et al., 2013).We did not find a relatively high sedimentary production in the sapropel layer where DOC and acetate accumulate (Chuang et al., 2021) and sustain microbial activities.This might be related to the relatively large fossil pool of IPL-GDGT-0 from the water column during sapropel deposition.

| Isotopic relationships according to sedimentary diagenetic zones
Isotopic relationships between the three major lipid types (BP0, BPcren, phytane) in both the sedimentary CL and IPL pool of the Black Sea have been extended by previously published data from the Mediterranean Sea (Zhu et al., 2021) to illustrate general trends within the three diagenetic zones (Figure 7).Independent of the site, δ 13 C CL-BP0 values are equal or lower than δ 13 C CL-BPcren , consistent with a general contribution of benthic archaea to the BP0 precursors IPL-GDGT-0, −1, −2, −3 and IPL-BDGT-0.The fact that it is reflected in the CL pool requires IPL turnover to the corresponding core lipids, and possibly other diagenetic products (Liu et al., 2016(Liu et al., , 2018) ) that release BPs upon ether cleavage in the laboratory.Likewise, the isotopic offset between δ 13 C IPL-BP0 and δ 13 C CL-BP0 is observed in different sedimentary environments and reflects the relatively higher impact of benthic archaeal BP0 production on the IPL pool (Figure 7).
For the isotopic difference (CL-BP0 and CL-BPcren, IPL-BP0 and CL-BP0) depicted in Figure 7, we observe somewhat lower isotopic offsets in the sulfate-bearing zone compared to the SMTZ and methanic zone.This is consistent with a relative increase in BP0 production by methane-metabolizing archaea in the latter two zones.
IPL-and CL-BPcren are isotopically highly similar (Figure 7), suggesting IPL-crenarchaeol shared the same sources as its CL counterpart and experienced little post-sedimentary influence.Compared with GDGT-0 and crenarchaeol, the isotopic offset between IPL-and CL-Phy showed a substantially wider range with both positive and negative values (Figure 7).We suggest that the lack of a clear trend may be caused by the relatively high turnover of archaeol across sediment depth.

| δ 13 C Phy values indicate in situ methanogenic activities in sediments
Archaeol concentrations in the sediments were substantially lower than the concentrations of GDGT-0 or crenarchaeol (Table S2), both of which have a relatively large contribution from fossil planktonic archaea; accordingly, the isotopic signal of benthic archaeal activity will be more strongly buffered by the large pool of allochthonous GDGTs compared to the small archaeol pool.δ 13 C Phy is generally more negative than δ 13 C BPcren and δ 13 C BP0 (Figure 5).This indicates that the sources of archaeol are distinct from those of GDGT-0 and crenarchaeol.Its relatively high isotopic variability in the methanic zone (Figure 5) compared to the overlying sediments, together with its elevated IPL/CL ratios (Figure 3) is consistent with this zone being inhabited by active archaeol-producing archaea.Within the methanic zone, acetate and DIC carbon isotopic compositions became progressively 13 C-enriched with depth (Figure 5), reflecting the utilization of both carbon sources by methanogenic archaea (Heuer et al., 2009;Summons et al., 1998;Zhuang et al., 2018), i.e., major producers of archaeol (Koga et al., 1993a).This source assignment is consistent with recent radiotracer experiments that suggested the sedimentary production of core archaeol by methanogens (Evans et al., 2019).The apparent increase in the isotopic isotopic fractionation between substrate and lipid biomarkers of hydrogenotrophic methanogenesis.For other methanogenic pathways, the relationship between energy availability and isotopic fractionation remains to be studied.The decreasing substrate availability with depth in Black Sea sediments would likewise result in larger fractionation as observed in our samples.From 700 cmbsf downwards, δ 13 C IPL-Phy values became more negative than δ 13 C CL-Phy values, averaging at −48.5 and −37‰, respectively.For the isotopic fractionation between archaeol and DIC, this results in values as large as 55‰, which are suggestive of methanogenesis at low substrate concentrations (Nguyen et al., 2020).

| The influence of sources and turnover of archaeal lipids on lipid-based proxies
The Black Sea provides a useful modern analogue to past oceanographic conditions during which sediments have been deposited under euxinic or oxygen-deficient conditions.Our study setting with detailed information on archaeal lipid distributions within the water column and the underlying sediments as well as contextual geochemical data enables an informed examination of the commonly applied proxies TEX 86 , CCaT, and MI in the pools of CLs, 1G-GDGTs, and 2G-GDGTs.In previous sections, we have presented evidence for both IPL production and degradation in Black Sea sediments.We found that the CL-based proxy ratios in the surface sediment resemble those in the suboxic zone (Figure 4).TEX 86 values of CLs vary between 0.3 and 0.6, with a tendency to higher values in the deep anoxic zone of the water column and the methanic sediments deposited during the lacustrine phase of the Black Sea during the last glacial.The seemingly warmer signal in sediments deposited under colder waters suggests that sedimentary sources of archaeal lipids and/or the ecology of Nitrososphaerota differed during the lacustrine phase since shifts in the dominant ecotypes can have a substantial impact on the TEX 86 -temperature relationship (Elling et al., 2015;Pitcher et al., 2011;Qin et al., 2015).We  a Represents the average value of δ 13 C lactate of the whole sample set due to the absence of data at these depths.
observed higher CL-GDGT-2/−3 ratios in sediments deposited in the lacustrine phase compared to the marine phase (Figure S9) even though both are within the range of shallow AOA types (Rattanasriampaipong et al., 2022).The CL-MI has been proposed as a proxy to evaluate the potential impact of GDGTs produced by AOM on the TEX 86 signal (Zhang et al., 2011) since methaneoxidizing archaea have been shown to be strong sources of GDGTs with one to three cycloalkyl moieties (Blumenberg et al., 2004;Liu et al., 2011).Consistent with AOM in the deep anoxic zone of the water column, we observe a corresponding signal in the CL-MI.
There is no positive MI signal in the SMTZ, but values above 0.5 in the methanic zone of the lacustrine portion.We cannot distinguish whether this signal originates from methane-producing archaea that are currently active in this zone or if this is a fossil signal from methane-oxidizing archaea caused during the inflow of brackish, sulfate-bearing waters into methane-laden lacustrine sediments during the early Holocene.In any case, the potential influence of methane-related sedimentary processes on the TEX 86 is evident in our dataset.
The profile of the CL-CCaT in sediments resembles that of CL-TEX 86 , with a trend to the highest values in the lacustrine portion of the methanic zone (Figure 4).Since this trend, caused by increasing amounts of crenarchaeol relative to GDGT-0, can neither be explained by increased activity of methane-metabolizing archaea in the methanic sediments nor by the history of SST in the Black Sea, we cannot rule out that all three proxy ratios have also been influenced by the ecology of planktonic archaeal GDGT sources.

| CON CLUS IONS
In this study, we have dissected the archaeal lipidome covering all major redox zones in the water column and the underlying sediments • Sedimentary IPL-and CL-crenarchaeol originate exclusively from planktonic AOA, with the major source being the chemocline.
• Sedimentary GDGT-0 originates predominantly from the water column but contributions from benthic archaea (22 ± 12%) are reflected by isotopic deviations from its chemocline values and its analogue crenarchaeol produced by Nitrososphaerota.
• Compound-specific carbon isotopic compositions of archaeol in the Black Sea respond sensitively to contributions of methanemetabolizing archaea.
• IPL/CL ratios of individual archaeal ether lipids highlight differences in the turnover of the archaeal lipids crenarchaeol, GDGT-0, OH-GDGT, and archaeol and identify zones of their elevated biosynthesis.

ACK N OWLED G M ENTS
We are grateful to the crew and the scientific shipboard party of the FS Meteor cruise M84-1 (DARCSEAS highly variable and range from 2 to 250 μmol L −1 with the highest concentration observed in the sapropel layer.Lactate (C 3 H 5 O 3 − ) concentrations are less variable than acetate, ranging from 29 to 54 μmol L −1 .Filter samples (N = 9) from the different depths of the water column at the same station were collected by in situ pumps through two stacked 0.7 μm glass fiber filters.Recovered filters were wrapped immediately in combusted aluminum foil and stored at −20°C.Pore water was extracted from sediment cores with Rhizon micro-suction samplers (0.1 μm filter width; Rhizosphere Research Products, Wageningen, the Netherlands) immediately after core F I G U R E 1 Study location in the southwestern Black Sea (red circle) at a water depth of 1266 mbsl.The map was created with Ocean Data View (ODV 5.5.1;Schlitzer, 2015).
Stable carbon isotopic compositions of Phy and BPs were determined by gas chromatography (GC)-IRMS.Briefly, the samples were injected into the Trace GC Ultra (Thermo Finnigan, Germany) equipped with a Restek Rxi-5 ms column (30 m × 250 μm × 0.25 μm, Restek, Bad Homburg, Germany) and coupled to a Delta V Plus IRMS via GC IsoLink connected to a ConFlow IV interface (Thermo Fisher Scientific GmbH).The initial oven temperature was held at 60°C for 1 min, increased to 150°C at a rate of 10°C min −1 , then raised to 310°C at a rate of 4°C min −1 and held at 310°C for 40 min.The carrier gas was helium with a constant flow rate of 1.0 mL min −1 .The injector temperature was set at 310°C.The accuracy was monitored by routine C 20 -C 40 nalkane standard measurements, which gives a long-term precision of ±0.5‰.The precision of the replicate analysis of samples (n = 2) was ≤0.7‰.All isotopic values are reported in the delta notation as δ 13 C relative to the Vienna PeeDee Belemnite (VPDB) standard.3| RE SULTS 3.1 | Compositional distributions of both core and intact archaeal lipids in the water column and sediments 3.1.1| Core lipids (CLs) To examine the distribution of diverse archaeal lipids across the Black Sea water column and sediments, both CLs and IPLs have been measured (Figure 2, Figures S3-S8).For a broad overview of distributional changes among CLs, they were classified into eight groups (Figure 2a): GDGTs, hydroxylated (OH)-GDGTs, glycerol dialkanol diethers (GDDs), OH-GDDs, BDGTs and PDGTs (B/PDGTs), archaeol (AR), OH-AR, and methoxy archaeol (MeO-AR).

3. 1
.2 | Intact polar lipids (IPLs) IPLs were broadly classified into six groups (Figure 2b): monoglycosyl (1G)-GDGTs, diglycosyl (2G)-GDGTs, hexose-phosphohexose (HPH)-GDGTs, IPL-OH-GDGTs, IPL-BDGTs (PDGTs were not detected in their IPL form), and IPL-ARs.There are pronounced changes in the overall lipid distribution at the transition to the anoxic zone of the water column (Figure 2).The IPL distribution in the marine portion of the sediment (Figure 2b) resembles that of the water column samples at 90 and 120 mbsl.HPH-GDGTs are most prominent in samples at 90, 120, and 150 mbsl; they remain detectable in the majority of the water column and sediment samples but at substantially lower relative proportions.Interestingly, their GDGT distributions in sediments generally differ from those in the water column (Figure S6).1G-GDGTs and IPL-OH-GDGTs were dominant in the oxic and suboxic zones of the water column while IPL-ARs became dominant in the deep anoxic zone, reaching up to 95% of the total IPL pool.ARs with phosphatebased headgroups, i.e., phosphatidylethanolamine (PE), phosphatidylglycerol (PG), and phosphatidylinositol (PI) contributed substantially to F I G U R E 2 The relative contribution of various ARs and GDGTs in the CL pool (a), IPL pool (b), and the relative abundance of CLs and IPLs in the total lipid pool (c) and the absolute concentration of total archaeal lipids in the Black Sea and its underlying sediments (d).SZ: sulfatebearing zone; SMTZ: sulfate-methane transition zone; MZ, methanic zone.The red dashed line represents the sapropel layer sample, which marks the transition between marine (shallower) and lacustrine (deeper) sediments.

(
Figure 4, Figure S9).CL-TEX 86 and 2G-TEX 86 in surface sediment (0.5 cmbsf) had similar values as in samples from the oxic or suboxic water column while 1G-TEX 86 values in surface sediments were generally higher than those of the water column.TEX 86 in all three lipid pools displayed a notable pattern change in lacustrine sediments compared to the overlying marine sediments.Likewise, MI and CCaT values differed between these two sedimentary zones.CL-MI increased gradually with water depth from 0.07 at 40 mbsl to 0.59 at 1200 mbsl.Moreover, CL-MI values in surface sediments resembled those from the suboxic zone of the water column while 1G-and 2G-MI in surface sediments did not match any zones in the water column.CL-CCaT values slightly decreased with water depth and the surface sediment value is equal to the value at 120 mbsl in the lower suboxic water.1G-CCaT in surface sediments was not consistent with any horizons in the water column.1G-CCaT increased slightly with sediment depth from 0.64 to 0.71 during the marine sedimentation.In summary, all CL-based proxies in surface sediment were consistent with signals derived from the suboxic zone while IPL-based ratios did not show this pattern.The proxies exhibited dif-ferent patterns for the lacustrine and marine sediments but there was no obvious trend related to the prevailing sedimentary redox regime, for example, no change of three MI ratios in the SMTZ compared to the overlying SZ.CL-GDGT-2/-3 showed a similar trend with CL-MI and increased with water depth from 1.8 at 40 mbsl to 26 at 1200 mbsl (FigureS9).CL-GDGT-2/-3 values in surface sediments resembled those from the suboxic zone of the water column with a mean value of 3.8.1G-and 2G-GDGT-2/-3 data was incomplete in the water column, especially in the deep water but their values in the suboxic and shallow anoxic zone were less than 7.8.Their values in sediments were less than 5.6.

TA B L E 1
Estimate of the in situ production (based on Equation4) of IPL-GDGT-0 in Black Sea sediment.
in the southwestern Black Sea, including the lacustrine phase deposited during the last glacial.The combined distributional and isotopic depth profiles enabled us to constrain sources and turnover of sedimentary lipids and to assess the impact of benthic processes on the sea surface temperature proxies TEX 86 and CCaT.Specifically, we demonstrated that:• Water-column-derived archaeal IPLs are well preserved in the anoxic depositional setting of the Black Sea as indicated by relatively high IPL/CL ratios.•The isotopic composition of archaeal lipids indicates CLs and IPLs in the deep anoxic water column have negligible influence on the sedimentary pool.
(Zhu et al., 2021)g the isotopic differences of various archaeal lipid types from the sediments of the Black Sea (this study) and other sites(Zhu et al., 2021).The symbol shapes and colors in the figure indicate study sites and geochemical zonation, respectively.MZ, methanic zone; SMTZ, sulfatemethane transition zone; SZ, sulfatebearing zone.
). Jenny Wendt, Xavier Prieto, and Heidi Tauber are thanked for supporting sampling and instrumental analyses.This study was funded by the European Research Council under the European Union's Seventh FrameworkF I G U R E 7