Glycerol dialkyl glycerol tetraether variations in the northern Chukchi Sea , 1 Arctic Ocean , during the Holocene 2 3

19 Glycerol dialkyl glycerol tetraethers (GDGTs) have become a useful tool in 20 paleoclimate research in ocean environments, but their applications in the Arctic are yet to be 21 developed. GDGTs were analyzed in three sediment cores from the northern/northeastern 22 Biogeosciences Discuss., doi:10.5194/bg-2016-529, 2016 Manuscript under review for journal Biogeosciences Published: 19 December 2016 c © Author(s) 2016. CC-BY 3.0 License.


Introduction
The Arctic Ocean currently experiences fast environmental changes due to its high sensitivity to global warming on various time scales (Screen and Simmonds, 2010;Miller et al., 2010).In particular, the Chukchi Sea (Fig. 1) is a region of dramatic changes in sea-ice and ocean current conditions due to its proximity to the North Pacific (e.g., Shimada et al., 2006).As the observational period of these changes covers only the last 2-3 decades Biogeosciences Discuss., doi:10.5194/bg-2016Discuss., doi:10.5194/bg- -529, 2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.(Woodgate et al., 2005 a,b;Shimada et al., 2006), reconstruction of sea-ice cover and ocean circulation on time scales beyond this period is important to comprehend the ongoing processes and future climate.A number of paleo-proxy records has been obtained recently on sediment cores recovering Holocene deposits from the Chukchi region (e.g., Belicka et al., 2002;de Vernal et al., 2005de Vernal et al., , 2013;;McKay et al., 2008;Darby et al., 2009Darby et al., , 2012;;Ortiz et al., 2009;Faux et al., 2011;Polyak et al., 2009Polyak et al., , 2016)).This data suggests that the Holocene paleoceanography in the Chukchi Sea was considerably different from other Arctic margins.
For example, some of the proxies indicate that while the Arctic was overall warmer in the early Holocene owing to higher summer insolation (CAPE Project Members, 2001), the Chukchi region may have had expanded sea ice, possibly related to a diminutive inflow of warm Pacific water via the Bering Strait (e.g., de Vernal et al., 2005de Vernal et al., , 2013;;Polyak et al., 2016).There is also evidence that the Pacific inflow intensified and peaked in the middle Holocene (Ortiz et al., 2009;Polyak et al., 2016;Yamamoto et al., 2016).In addition to these long-term changes, higher frequency variabilities have also been identified in records with enhanced resolution (Polyak et al., 2016;Yamamoto et al., 2016).
These results pose numerous questions to the responses of the Chukchi Sea current system, sea ice, and biota to climatic changes, which can be addressed by more detailed and multifaceted proxy studies.Biomarker research, including glycerol dialkyl glycerol tetraethers (GDGTs), has a potential to augment paleoclimatic data as an independent proxy approach to paleoproductivity and hydrographic environments.We have analyzed GDGTs in three sediment cores from the northern/northeastern Chukchi Sea to evaluate changes in their abundance and composition during the Holocene (last ~10 ka) in the context of regional hydrographic and sea-ice environments.Some of the basic GDGT data on two of the studied Biogeosciences Discuss., doi:10.5194/bg-2016Discuss., doi:10.5194/bg- -529, 2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.cores have been used by Polyak et al. (2016) to corroborate biomarker-based sea-ice reconstructions, while regional distribution of GDGTs in surficial sediments has been reported in Park et al. (2014).In this paper we provide a comprehensive investigation of the GDGT distribution in the Holocene sediments under study.We also evaluate the applicability of GDGTs for paleoclimatic reconstructions in the Arctic seas by comparing them to other proxies in cores under study and data from other marine sites in the Chukchi-Alaskan region.

GDGT proxies
Glycerol dialkyl glycerol tetraethers (GDGTs) are increasingly used as proxies to trace environmental changes such as contribution of soil organic matter, sea surface temperature (SST), air temperature, and soil pH in the source areas (e.g., Sinninghe Damsté et al., 2000;Schouten et al., 2002;Hopmans et al., 2004;Weijers et al., 2007), but their application to Arctic marine sediments has been limited.Park et al. (2014) discussed the production, advection, and preservation of both isoprenoid and branched GDGTs as key processes determining the distribution of GDGTs and derived indices in surface sediments of the Chukchi Sea and adjacent areas of the Arctic Ocean (Fig. 2).In particular, this study demonstrates that GDGT composition in the Arctic Ocean north of the Chukchi margin (approximately 75°N) is strongly affected by allochtonous soil bacteria, while GDGTs on the Chukchi shelf have a higher marine component, and thus, a better potential for characterizing their sources and related paleoceanographic environments (Park et al., 2014).This knowledge can be now used for interpreting GDGT records in marine sediment cores from the Chukchi margin.Discuss., doi:10.5194/bg-2016Discuss., doi:10.5194/bg- -529, 2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.

Biogeosciences
In this study, in addition to measured isoprenoid and branched GDGT, we empolyed TEX 86 (TetraEther indeX of tetraethers consisting of 86 carbon atoms), MI (methane index), MBT' and CBT indices (methylation index and cyclization ratio of branched tetraethers), and BIT (Branched and isoprenoid tetraether).TEX 86 used to reconstruct sea surface temperature (Schouten et al., 2002;Kim et al., 2010) is based on an initial global core-top dataset (Schouten et al., 2002), whereas, TEX 86 L is based on an expanded dataset including polar waters (Kim et al., 2010).MI indicate a degree of anaerobic oxidation by Euryarchaeota (Zhang et al., 2011).High MI (> 0.4), representing high contribution of Euryarchaeota, is associated with bias of TEX 86 values.MBT' and CBT have been proposed as proxies of soil pH and mean annual air temperature (Peterse et al., 2012;Weijers et al., 2007), and BIT as a proxy of soil organic matter contribution (Hopmans et al., 2004;Kim et al., 2006).These indices have been shown useful for identifying the sources of branched GDGT in the study region (Park et al., 2014).

Samples and age constraints
Sediment cores ARA02B 01A-GC (gravity core) and HLY0501-05TC/JPC, -06JPC and -08TC/JPC (trigger/ jumbo piston cores), hereafter referred to as 01A-GC, 05JPC, 06JPC and 08JPC, were collected from the northern to northeastern margin of the Chukchi shelf during the 2011 cruise of the RV Araon and 2005 cruise of USCG Healy (Table 1; Fig. 1).Trigger (e.g., Keigwin et al., 2006).In contrast, cores 05JPC and 06JPC were raised from the continental slope at 462 m and 673 m depth, where sediment deposition was not interrupted by sea-level changes.Only a few samples from the bottom part of stratigraphically most extensive 06JPC are used in this study to augment the 05JPC record.
In total 110, 47, 2 and 34 samples were collected for the GDGT analysis from cores 01A-GC, 05JPC, 06JPC, and 08JPC, respectively.Samples were mostly taken from the Holocene marine sediments at intervals providing a multidecadal-to multicentury-scale resolution.In addition, several samples from the lower part of cores 05JPC and 06JPC span the pre-Holocene sedimentary sequence.Samples were stored in a refrigerator since collection, subsampled and freeze-dried for further processing.
Local reservoir corrections (ΔR) were taken as 500 years for 01A-GC and 08JPC washed by surface waters and 0 years for 05JPC washed by subsurface Atlantic waters (McNeely et al., 2006;Darby et al., 2012).
The age model was constructed by linear interpolation between the dating points, which fall within the interval of ca.1.5-8.6 ka (Fig. 3), as well as the assumed modern age of the core tops.Ages below the dated interval were extrapolated to the bottom of cores 01A-GC and 08JPC and to the bottom of marine unit in stratigraphically longer core 05JPC.Core 05JPC was further expanded by the addition of two samples from the nearby core 06JPC.
Rough age constraints for older sediments in cores 05JPC and 06JPC were estimated by correlation with cores from the adjacent western Arctic Ocean (Polyak et al., 2009), where samples from core 05JPC span the last deglaciation, and the two samples from core 06JPC possibly represent pre-LGM (> ca. 25 ka) environments.Due to inevitable inaccuracies in the age estimation beyond the dated interval, the bottom of marine sediments in the 01A-GC, 05JPC and 08JPC came out with slightly different ages between ca.8.5−9.5 ka, so we assume that the actual age of this stratigraphic boundary is close to 9 ka.The distribution of linear sedimentation rates shows maximal values in all studied cores around 5−6 ka, with especially high rates in core 08JPC (Fig. 3).The synchroneity of this peak corroborates the validity of the difference in ΔR used for cores from different water masses (01A-GC and 08JPC vs.We note that the difference applies only to the uppermost part of the stratigraphy and does not have a considerable effect on the conclusions of this study.

GDGT analysis
Freeze-dried and homogenized sediments were extracted using accelerated solvent
Crenarchaeol and GDGT-0 are the most abundant isoprenoid GDGTs in the studied samples (Fig. 4).In the averaged fractional abundance of isoprenoid GDGTs, crenarchaeol in cores 01A-GC and 08JPC is most abundant during middle and late Holocene, comparable to GDGT values in surface sediment from the shelf edge of the Chukchi Sea (Fig. 4).Fractional abundances of isoprenoid GDGTs show a considerable variability at the transition from early to middle Holocene, especially in core 05JPC.
Total concentrations of isoprenoid GDGT have highest values in core 01A-GC, from 2.6 to 31.6 μg/g, and do not exceed 18.4 μg/g in cores 05JPC and 08JPC (Fig. 5).
Concentrations vary between the cores but show similar, stratigraphically consistent downcore patterns, especially for cores 01A-GC and 05JPC.In the late deglacial interval to early Holocene (until ca. 9 ka) concentrations were low in all three cores, then increased markedly to ca. 7−8 ka, and reached a maximum around ca. 5−6 ka in cores 01A-GC and 05JPC.In the late Holocene, isoprenoid GDGTs had overall high but variable concentrations, with a distinct maximum around 3 ka in 08JPC.Near the core top in 01A-GC, 05JPC and 08JPC concentrations show a decrease.In the deglacial unit studied in core 05JPC, isoprenoid GDGT get relatively more abundant towards the bottom.In the yet older (possibly pre-LGM) sediment recovered in core 06JPC, the concentrations of isoprenoid GDGT were during the entire Holocene, GDGT-IIb is also abundant in core 01A-GC, especially in the middle and late Holocene, with values comparable to surface sediment from the shelf edge of the Chukchi Sea (Fig. 4).Branched GDGTs show a considerable difference in fractional abundances between middle and early Holocene, especially in core 05JPC, similar to isoprenoid GDGTs (Fig. 4).The total concentrations of branched GDGTs reach 1.3 μg/g, 1.1 μg/g, and 1.9 μg/g in cores 01A-GC, 05JPC, and 08JPC, respectively (Fig. 5).Like isoprenoid GDGTs, the branched GDGTs concentrations show a similar downcore distribution pattern.In all studied cores, concentrations were low until ca. 9 ka, and then increased to around 8 ka.The peak at ca. 5−6 ka is well expressed in cores 01A-GC and 05JPC, followed by variable concentrations with another, somewhat lower maximum at 1−2 ka.In core 08JPC maximal values were reached around 7 ka and 3 ka.An increase in branched GDGTs is also evident at the bottom of the deglacial unit in core 05JPC, whereas, low concentrations characterize the two older samples from core 06JPC (Fig. 5).
The BIT index shows highest levels of >0.5 in the deglacial unit, decreases to very low levels by ca.7−8 ka, and stays consistently low since then in cores 01A-GC and 05JPC (Fig. 5).In core 08JPC, BIT decreases with some fluctuations from >0.3 at 9−10 ka to 0.1 by 3 ka, and then slightly increases towards the core top.In the early deglacial and older sediments in cores 05JPC/06JPC, BIT values are lower than later in deglaciation to early Holocene, but somewhat higher than in the late Holocene (Fig. 5).The ratio of GDGT-0 to crenarchaeol (GDGT-0/cren) and the MI are overall low except for somewhat elevated values in the early Holocene until ca.8 ka in cores 05JPC and 08JPC (Fig. 5).
The distribution of the CBT index is similar to that of the BIT (Fig. 5).Maximal CBT values characterize the early Holocene and decrease to low levels by 7−8 ka in cores 01A-GC and 05JPC.In 08JPC the Holocene CBT is somewhat higher and shows an overall decrease towards ca. 3 ka and a slight increase thereafter.In the early deglacial and older section in cores 05JPC/06JPC, CBT values are relatively elevated, but not as much as in later deglacial to early Holocene sediments.
The MBT' index also shows relatively high values in the early Holocene with a peak around 10 ka in core 05JPC and another peak around 8 ka (Fig. 5).In core 08JPC MBT' is variable with peaks at ca. 6, 7 and 8 ka.In core 01A-GC the MBT' shows less variability.
TEX 86 -derived temperatures strongly fluctuate, ranging mostly between 5 and 15 °C with a slight general increase throughout the Holocene in cores 01A-GC and 05JPC (Fig. 5).
No such trend occurs in core 08JPC.TEX 86 L -derived temperatures are overall lower by up to 10 °C than TEX 86 -derived temperatures and show no trend in their distribution.In core 01A-GC, the amplitude of TEX 86 L fluctuations increases noticeably after ca. 6 ka.

Changes in production and sources of GDGTs during the Holocene
GDGT distribution in surface sediments from the Chukchi Sea and the adjacent western Arctic Ocean and northern Bering Sea shows abundant isoprenoid GDGTs on the outer shelf and slope of the Chukchi Sea and the upper slope of the Bering Sea (Fig. 2; Park et al., 2014).Biogeosciences Discuss., doi:10.5194/bg-2016-529, 2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.
The higher abundances are attributed to a combination of higher production of marine Archaea (Thaumarchaeota) at the shelf edge, redeposition of GDGT-carrying fine sediment particles from the shelf, and better preservation of GDGTs at sites with higher sedimentation rates.
The fast increase in isoprenoid GDGTs is observed in all three studied cores at the bottom of marine sedimentary unit, from ca. 9 to 8 ka (Fig. 6).This change is consistent with an increase in total organic carbon and some other biogenic proxies, such as silica, in these and nearby cores (Darby et al., 2001;Lundeen, 2005;McKay et al., 2008;Currie, 2009, and unpublished data for 01A-GC).This correspondence suggests that the increase in isoprenoid GDGT concentrations was driven by increasing bioproduction with the establishment of marine environments in the Chukchi Sea after the end of deglaciation and opening of sufficient inflow through the Bering Strait.
Further changes in isoprenoid GDGT concentrations are more complex.The peak values around 5-6 ka in cores 01A-GC and 05JPC correspond to maximal sedimentation rates (Fig. 5) suggesting that preservation may have been a factor.However, no isoprenoid GDGT peak occurs in core 08JPC at this time despite a well expressed sedimentation rate maximum.
This discrepancy indicates that isoprenoid GDGT concentrations were primarily controlled by factors other than preservation, like primary production, which could have varied spatially due to differing sea-ice conditions.A comparison with the distribution of sea-ice related biomarker IP 25 shows that concentrations of isoprenoid GDGTs in both cores increased after the decline of IP 25 values peaking at ca. 5-6 ka in 5JPC and ca. 3 ka in 8JPC (Fig. 5; Polyak et al., 2016).This offset of isoprenoid GDGT peaks relative to IP 25 is consistent with the inferred negative effect of sea ice on local GDGT production (Park et al., 2014).Another possibility is that the GDGT peak in cores 01A-GC and 05JPC was related to an increase in the Bering Strait inflow in the middle Holocene (Ortiz et al., 2009;Polyak et al., 2016), which may have had different effects in the northern and eastern parts of the Chukchi Sea washed by different branches of the Bering Strait inflow (Fig. 1).In the pre-Holocene section, a relative increase in isoprenoid GDGTs near the bottom of deglacial sediments in core 05JPC (Fig. 5) might represent a post-LGM warming and resultant sea ice retreat and enhanced primary production, such as during the Bølling/Allerød period, but the age control is insufficient to constrain this interval.
Branched GDGTs in surface sediments are abundant on the Chukchi shelf and in the Yukon and Mackenzie River estuaries (Fig. 2; Park et al., 2014).A concerted abundance of both branched and isoprenoid GDGTs at the Chukchi shelf edge indicates common concentration processes, such as sediment redeposition and enhanced preservation at sites with high sedimentation rates.High cyclization ratios of branched tetraethers (CBT) characterisze sediments from the Arctic Ocean north of the Chukchi margin (~75 °N), as well as the Yukon and Mackenzie River estuaries, in contrast to lower CBT in sediments from the Chukchi and Bering seas.This difference indicates two principal sources of branched GDGTs tentatively interpreted as soil bacteria and in situ marine bacteria, respectively.
High BIT and CBT vs. MBT' values peaking in the deglacial sediments and extending into lower Holocene until ca.8 ka, as expressed especially clearly in cores 01A-GC and 05JPC (Fig. 6), are similar to these indices in surface sediments of the study region north of 75 °N (Park et al., 2014).75 °N were interpreted as a result of very low marine production and/or severe degradation under multi-year ice, and thus relatively high content of imported terrestrial GDGTs.We infer that similar conditions prevailed in the study area in the early Holocene before ca.8 ka.This conclusion is consistent with the dinocyst-based proxy record from core 05JPC indicative of high sea-ice concentration in the early Holocene (McKay et al., 2008).Other dinocyst studies from this region also show a generally similar pattern (de Vernal et al., 2005;2008;2013;Farmer et al., 2011), but lack resolution or stratigraphic recovery for a comprehensive characterization of the early Holocene.In addition to the effect of high sea-ice concentration in the early Holocene, low Bering Strait inflow and elevated freshwater inputs due to incomplete sea-level rise and deglacial processes (Darby et al., 2001;Lundeen, 2005;Yamamoto et al., in review) could inhibit marine production and enhance import of terrestrial material in the northern Chukchi Sea.
BIT and CBT vs. MBT' values in the early deglacial and older sediments in cores 05JPC/06JPC show intermediate levels between the deglacial to early Holocene and middlelate Holocene data .This suggests the possibility of either relatively high organic production (low ice concentrations) during those times or redeposition of organic material from stratigraphically older deposits.The latter may be especially applicable to 06JPC samples that show very low isoprenoid GDGT concentrations.(Park et al., 2014).However, in full-marine sediments deposited in cores under study after ca.8-9 ka, these indices are consistently low (Fig. 5), suggesting that variation in TEX 86 and TEX 86 L here is not related to terrestrial contribution or appreciable methanotrophic euryarchaea, which can produce GDGT-1, GDGT-2 and GDGT-3.

Variations in TEX
A millennial-to multicentury-scale variability in TEX 86 -and TEX 86 L -derived SST is well expressed in the investigated Holocene records, especially in core 01A-GC studied with the highest resolution (Fig. 5).This variability appears to increase in amplitude, along with a slight increase in TEX 86 -and TEX 86 L -derived SST values, during the Holocene towards the core top.This variation does not resemble the record of chlorite abundance reflecting the strength of Bering Strait inflow (Ortiz et al., 2009) nor the proxy records of sea-ice cover (de Vernal et al., 2013;Polyak et al., 2016).This differing pattern suggests that neither Pacific water advection nor sea-ice cover had a major control on TEX 86 and TEX 86 L variability in the northern Chukchi Sea.More studies are needed to understand the controls on these indices and their applicability for SST reconstructions in the Arctic.

Conclusions
The analysis of GDGTs in three sediment cores from the northern/northeastern Chukchi

Figure captions
cores are used for a better representation of the uppermost soft sediments that may have been missed by respective piston cores due to overpenetration.Cores 01A-GC and 08JPC are sited on the outer shelf in water depths of 111 and 90 m, which were above the sea level at the time of the last glaciation and inundated during the postglacial transgression since about 15 ka Biogeosciences Discuss., doi:10.5194/bg-2016-529,2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.
from 210 Pb measurements in the upper 15 cm in 05TC suggest somewhat younger ages than those derived from available 14 C datings (McKay et al., Biogeosciences Discuss., doi:10.5194/bg-2016-529,2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.2008).While comparing these two approaches require more precise chronostratigraphic constraints, in this study we used the 14 C-based age model because of the uncertainty with extrapolating 210 Pb-based age estimates related to potential variability in sedimentation rates.
Fig. 4 also shows similarity in fractional distribution of branched GDGTs between the early Holocene interval in the studied cores and offshore areas of Chukchi Sea.These high BIT and CBT values along with low GDGT concentrations north of Biogeosciences Discuss., doi:10.5194/bg-2016-529,2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.
Biogeosciences Discuss., doi:10.5194/bg-2016-529,2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.Sea margin provides insights into GDGTs production and sources in this region of the Arctic during the Holocene.Concentrations of isoprenoid GDGTs reached high values by ca. 8 ka with the establishment of marine conditions after deglaciation and sea-level rise.Low GDGTs concentrations combined with high BIT and CBT values prior to ca. 8 ka may suggest high concentrations of sea-ice in the northern Chukchi Sea, with overall milder seaice conditions later in the Holocene.Higher inputs of terrestrial material were also likely during the deglaciation extending into the early Holocene.After ca. 8 ka, GDGTs distribution was variable and probably controlled by a combination of sea-ice conditions and Bering Strait inflow that affected primary production and sediment transport and deposition.Different patterns in GDGTs distribution between cores from the northern and northeastern sites may indicate spatial differences in the pathways of Pacific waters and sea-ice extent.TEX 86 and TEX 86 L indices potentially useful for SST reconstruction show millennial-scale variability, but the controls are not well understood.More investigations using multiple proxies are needed to comprehend sea-ice, temperature, and circulation history in the Chukchi Sea, a critical region for the Arctic climate change.

Fig. 1 .
Fig. 1.Index map with location of sediment cores under study and a generalized bathymetry.Arrows indicate major currents in the Chukchi Sea: Siberian Costal Current (brown color) and three branches of Pacific water flowing through the Bering Strait (orange color).The blue-sky and red dashed lines indicate summer sea-ice margin (15% concentration) for the late 20 th century average and the all-time observational 2012 minimum, respectively (data from the National Snow and Ice Data Center).

Fig. 2 .
Fig. 2. GDGT distribution in surface sediments of the Chukchi Sea and adjacent western Arctic Ocean and northern Bering Sea: concentrations of (A) isoprenoid GDGTs and (B)

Fig. 3 .Fig. 1 .Fig. 3 .Fig. 5 .
Fig. 3. Age-depth distribution of calibrated 14 C ages in cores under study Further south in the Chukchi Sea, these indices show a more Biogeosciences Discuss., doi:10.5194/bg-2016-529,2016 Manuscript under review for journal Biogeosciences Published: December 2016 c Author(s) 2016.CC-BY 3.0 License.reasonable relation to SST, although still off the global core top calibration curve (Kim et al., 2010).Therefore, one must be cautious about translating TEX 86 and TEX 86 L data into absolute SST values in the study area.Nevertheless, relative downcore changes in TEX 86 and