Sediment core R07 (28 cm long) was collected from the Chukchi Sea shelf (73.00°N, 168.97°W, 73 m water depth; Figure 1 ) with a multiple-core sampler during the sixth Chinese National Arctic Research Expeditions in 2014. The lithological characteristics have been described in Zhang et al. (2018). Briefly, sediment core R07 was dominated by silt (avg. = 65%) and clay (avg. = 25%). The very top layer (0-1 cm) was used for other research, with the remaining core (1-28 cm) sampled every cm, and stored at –20°C until analysis.

²¹⁰Pb concentrations were measured to determine the chronology of Core R07 using ORTEC HPGe detectors (GEM, Lo-Ax and GMX) at the Nanjing Institute of Geography and Limnology, and had been published by Zhang et al. (2018). A sedimentation rate of 0.41 cm yr^–1 (Zhang et al., 2018; Figure S1 ) was calculated using the constant initial concentration (CIC) model (Oldfield et al., 1978) for Core R07 (see supplementary material for details), which is similar to the sedimentation rate reported from an adjacent station (71.49°N, 167.78°W, 0.37 cm yr^–1, Cooper and Grebmeier, 2018) and also within the estimated range of 0.20-0.70 cm yr^–1 sedimentation rate of the Chukchi Sea shelf (Huh et al., 1997). The ²¹⁰Pb results indicate the Core R07 contained an ca. 70-years sediment record (from ca. 1946 to ca. 2011).

Methods for biomarker analysis were modified from Yang et al. (2014). Briefly, an aliquot of each freeze-dried and homogenized sediment-sample (ca. 5 g) was ultrasonically extracted (×7) using a mixture of dichloromethane (DCM) and methanol (MeOH) (9:1, v:v). Dry sediments were spiked with known amounts of internal standards (androstane, cholan-24-oic acid, and androstan-3-ol) for biomarker quantification before lipid extraction. The total lipid extracts were separated into non-polar and polar fractions on silica gel chromatography columns by eluting with n-hexane and MeOH, respectively. After drying under a gentle stream of nitrogen gas, the non-polar fraction containing n-alkanes was redissolved for instrumental analysis, while the polar fraction was saponified with 5 mL of 1-mol/L KOH/MeOH (5% H₂O) solution for 2 h at 80 °C. After cooling, the solution was extracted using n-hexane (2 mL × 6) to obtain the neutral fraction, containing n-alkanols, sterols, diols, etc. The remaining solution was then acidified using HCl (pH< 2), and then extracted with n-hexane (2 mL × 6) to obtain the fatty acids. All fractions were dried under nitrogen gas, and the neutral fraction was derivatized using N, O-bis (trimethylsilyl)-trifluoroacetamide (BSTFA) at 70°C for 1.5 h. The fatty acids were esterified using a 14% BF₃/MeOH solution at 70°C for 1.5 h, and then extracted with n-hexane (2 mL × 6).

The alkanes, derivatized neutral compounds, and fatty acid methyl esters were analyzed using a gas chromatography-mass spectrometer (GC-MS; Agilent 7890A/5975C) equipped with a DB-5MS capillary column (60 m × 0.25 mm × 0.25 μm). The GC oven was heated from 70 to 210°C at 10°C/min, and from 210 to 310°C at 3°C/min, and then held at 310°C for 36 min, with helium as the carrier gas. The inlet temperature was 310°C and the EI energy was 70 eV. Samples were scanned in the full scan mode from 50 to 550 m/z.

Compounds were identified by comparing their mass spectra with literature (Boon et al., 1979; Volkman, 1986; Versteegh et al., 1997; Rampen et al., 2012) and database (NIST08 library data). Diols were quantified in the extracted ion chromatogram according to Rampen et al. (2012). Concentrations of different compounds were calculated by comparing peak areas with those of internal standards and then normalized to the dry weights of sediments. It needs to be pointed out that since the potential response factor differences between compounds of interests and internal standards were not determined, the quantification results are considered semi-quantitative. However, this limitation should not affect our conclusions since we focused on variations of biomarker concentrations over time as well as their ratios.

The Carbon Preference Index (CPI) of n-alkanes was calculated using the following formula:

The CPI for n-alkanols and n-alkanoic acids were calculated using the following equation:

The original data of the monthly Arctic Oscillation (AO) index were obtained from National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center (https://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao.shtml). Since phytoplankton growth in the Chukchi Sea shelf mainly occurs in summer (Hill et al., 2018; Bai et al., 2019), the summer AO index was calculated, which is an average of the AO index in June, July and August.

The historical SSTs at core R07 location were obtained from the NOAA Extended Reconstructed Sea Surface Temperature (ERSST) V5 (Huang et al., 2017), which is provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA (at https://www.esrl.noaa.gov/psd/data/gridded/data.noaa.ersst.v5.html). The SST calculation is described in detail by Gao et al. (2021). Considering the SST in winter (December, January and February, DJF) and spring (March, April and May, MAM) is –1.8°C, the SST of an average of summer (June, July and August, JJA) and fall (September, October and November, SON) is used in this paper.

Monthly satellite sea ice concentration (SIC) data (1979–2020) were downloaded from Nimbus-7 SMMR and DMSP SSM/I-SSMIS Passive Microwave Data at the National Snow and Ice Data Center (NSIDC, http://nsidc.org/data, grid cell size 25 × 25 km; Cavalieri et al., 1996). As the area where R07 core is located is covered by sea ice in winter (DJF) and spring (MAM) (SIC=100%) ( Figure S2 ), which corresponds to the SST –1.8°C, an average SIC of summer (JJA) and fall (SON) were used in this study. Although satellite SIC data before 1979 is absent, we can get the SIC record before 1979 from the SST data the significant correlation between SIC and SST during 1979 to 2020 ( Figure S3 ; see supplementary material for details).

Sea water temperature, salinity and nutrients (nitrate, silicate and phosphate) at different depth were downloaded from World Ocean Atlas (WOA) 2018 in the National Centers for Environmental Information (NCEI) on a 1° grid resolution (https://www.ncei.noaa.gov/access/world-ocean-atlas-2018/). These databases provide standardized vertical intervals from the surface (0 m) to the seafloor (5500 m) with annual mean parameters between 1955 and 2017. Parameter values at 0, 10, 20, 30, 50, 75 and 100 m were used in this paper. The data for the 73°N transect where R07 core is located were calculated by averaging the 72.5°N and 73.5°N transect ( Figure 2 ).

SST and SIC in this paper represent the average values of SST and SIC in summer (JJA) and fall (SON), respectively.

Long-chain n-alkanes, n-alkanols, and n-alkanoic acids are ubiquitous in marine sediments, which have been widely employed as an indicator of terrestrial input in marine environments (Eglinton and Eglinton, 2008; Naeher et al., 2022). In Core R07, the distribution characteristics (high CPI) of long-chain n-alkanes, n-alkanols, and n-alkanoic acids indicate that these compounds are derived from terrestrial plants (Eglinton and Hamilton, 1967; Rielley et al., 1991; Meyers, 1997; Eglinton and Eglinton, 2008). Some bacteria could also contain long-chain n-alkanes; however, these n-alkanes lack an odd-over-even carbon predominance, with low CPI values of approximately 1.0 (Han and Calvin, 1969; Cranwell et al., 1987). In Arctic Ocean, these compounds have been used to reconstruct terrigenous input into sediments (Belicka et al., 2004; Yunker et al., 2005; Faux et al., 2011; Yunker et al., 2011) and sediment trap material (Bai et al., 2019).

C₃₂ 1,15-diol, a tracer for the riverine input (Lattaud et al., 2017a; Lattaud et al., 2017b; Lattaud et al., 2019), was also detected in Core R07. C₃₂ 1,15-diol is mainly produced by freshwater eustigmatophytes in rivers (Lattaud et al., 2017a) and transported to the continental shelf by currents. Previous studies have shown that the Chukchi Sea shelf receives high amounts of sediments from rivers (especially Yukon River) (Stein, 2008; Darby et al., 2009), which probably explains the presence of C₃₂ 1,15-diol in Core R07.

Interestingly, the concentrations of terrigenous biomarkers, including long-chain n-alkanes, n-alkanols, n-alkanoic acids, and C₃₂ 1,15-diol, exhibit similar trends in Core R07 ( Figure 3 ). Previous studies showed that the terrestrial materials in the Arctic Ocean sediment were transported to this area mainly through sea ice and water mass (or current) (Belicka et al., 2002; Belicka et al., 2009; Darby et al., 2009; Stein, 2019). Sea ice contains terrigenous material and release it to the area of sea-ice melting (Stein, 2019). But, there is no significant change in SIC before and after 1983 ( Figure 4 ), indicated by SST ( Figure S4 ; mean SST of –0.92°C for 1974~1983 and –0.78°C for 1984~1993). Thus, this synchronous change in terrigenous input is unlikely to be caused by the sea ice. Considering that C₃₂ 1,15-diol is derived from riverine input, the similar trend of terrigenous compounds indicates that changes in the contribution of riverine input, which was transported mainly by water mass to this area, may primarily account for the variations in concentrations of these biomarkers in Core R07. This was supported by the year-round sediment trap from the Chukchi Sea. Their results showed that the mass fluxes of long-chain n-alkanes started increase with the sea ice melt, and the highest occurred during the ice free period (Bai et al., 2019), when there is the highest river discharge (Ge et al., 2013).

Abundant marine ecosystem-derived compounds were also detected in Core R07, including brassicasterol (24-methylcholesta-5,22E-dien-3β-ol), dinosterol (4α,23,24-trimethyl-5α-cholest-22-en-3β-ol), and C₃₀ diols, which are thought to be biomarkers for diatoms, dinoflagellates, and eustigmatophytes, respectively, and their concentrations can reflect the corresponding phytoplankton productivity (Schubert et al., 1998; He et al., 2013). Although brassicasterol can be produced by algae other than diatoms, such as dinoflagellates and haptophytes (Volkman, 1986; Ding et al., 2019), in an ecosystem dominated by diatoms, this compound is mainly produced by diatoms and could be used for reconstructing the diatom productivity (Wu et al., 2016; Wang et al., 2019). In the Chukchi Sea shelf, the phytoplankton community was dominated by diatoms, which account for at least 60% of the total population (Sergeeva et al., 2010; Lewis et al., 2019). Although brassicasterol and dinosterol also have terrigenous sources, they are thought to be generally minor in open ocean waters (Volkman, 1986; Fahl and Stein, 1999). In addition, considering the small percentage of terrestrial organic matter as indicated by low branched and isoprenoid tetraethers (BIT) index values (avg. = 0.07; Gao et al., 2021) and high δ ¹³C_org (avg. = –22.25‰; Zhang et al., 2018), these brassicasteol and dinosterol in R07 should be mainly from marine algae. Thus, brassicasterol, dinosterol and C₃₀ diols can represent the production of diatom, dinoflagellate, and eustigmatophytes, respectively. In Core R07, concentrations of these biomarkers showed similar trend and were roughly consistent with the changes in TOC ( Figure 4 ).

Although quantifying the contribution of individual phytoplankton productivity with biomarkers is difficult at present due to lipid biosynthesis per cell differing among species (Ding et al., 2019), source-specific compounds (biomarker) are useful for evaluating relative changes in the biomass of different phytoplankton (Bi et al., 2021). For example, the B/D ratio enables assessment of spatiotemporal changes in the relative concentrations of diatoms and dinoflagellates, and subsequent, paleoenvironmental reconstruction at different timescales (Schubert et al., 1998; He et al., 2013; Wu et al., 2016 and references therein).

Degradation can influence biomarker concentrations in marine sediments, thereby complicating their utilization in paleoenvironment studies (Hernández-Sánchez et al., 2014; Naeher et al., 2022). However, the shallow water depth (73 m), high primary production, and high deposition rate in this area (Grebmeier et al., 2006) enable a high fraction of the algae-derived organic matters to reach the sediments (Belicka et al., 2002). In addition, the temperature in Chukchi Sea shelf water are relatively low, which is beneficial for lipid preservation. In Core R07, lipid concentrations do not exhibit a single decreasing trend from the top to bottom ( Figures 3 , 4 ), which is different from the typical degradation distribution. Therefore, biomarkers in sediments from Core R07 are unlikely to be considerably affected by degradation or diagenetic processes, and can, therefore, reflect the original input.

In Core R07, the B/D ratios showed different pattern, with an average value of 0.99 before ca. 1983, and of 1.53 after then ( Figure 3 A ), indicating a change in the phytoplankton community (increased proportion of diatoms relative to dinoflagellates). Owing to differences in the optimal growth environments of phytoplankton, the phytoplankton community structure was influenced by water physical conditions and nutrient concentrations, which were significantly affected by the location and structure of water mass ( Figure 2 ; Springer and McRoy, 1993; Lee et al., 2007). Therefore, phytoplankton community is considered to be water mass-specific (Hamilton et al., 2008; Sergeeva et al., 2010; Zhuang et al., 2016; Martini et al., 2016; Giesbrecht et al., 2018; Lee et al., 2019). According to previous study in the western Chukchi Sea shelf, there is more dinoflagellates relative to diatoms in the area close to ACW than that close to BSW (Sergeeva et al., 2010). In particular, observed data showed that spread of the ACW reduced the diatom biomass in the Chukchi Sea (Zhuang et al., 2016). Thus, the change in phytoplankton community recorded by the elevated B/D ratios in Core R07, which indicates an increased diatoms relative to dinoflagellates ( Figure 3 A ) highlight water mass structure change in this area, i.e., reduced ACW and increased BSW.

Relative to the ACW, the BSW is colder, saltier, and richer in nutrients ( Figure 2 ; Woodgate, 2018), and these properties are beneficial for water mixing. It has been reported that, compared with dinoflagellates, diatoms are reported to bloom at colder temperatures, higher nutrients, and intense mixing conditions (Margalef, 1978; Winder and Sommer, 2012; Xiao et al., 2018). Laboratory experiments (diatom-dinoflagellate competition experiments) also showed that there is a competitive superiority of the diatoms at high nutrient concentration, low temperature and low N:P ratios (Bi et al., 2021). In addition, as essential substance for diatom growth, high silicate favors the diatom growth has been reported in natural environments (Ragueneau et al., 2002; Allen et al., 2005) and incubation experiment (Lü et al., 2020). As the WOA data showed, in the area that close to the ACW (east of the R07), there is lower nitrate, lower silicate and higher temperature ( Figure 2 ). Thus, the altered water condition associated with reduced ACW enhanced the growth of diatoms relative to dinoflagellates and thereby increasing the B/D ratio ( Figure 3 A ).

Change in seasonal sea ice is another important factor affecting phytoplankton community structure, which has been reported in different area in Arctic Ocean, such as Chukchi Sea (Fujiwara et al., 2016; Neeley et al., 2018), Bering Sea (Fujiwara et al., 2016), Canada Basin (Coupel et al., 2012) and East Siberian Seas (Lee et al., 2019). However, the constant SIC before and after 1983 in the study area ( Figure 4 ; Figure S4 ) indicates that change in phytoplankton community structure in R07 is unlikely to be caused by sea ice change.

The change in water mass structure is also reflected in the terrigenous input variations in Core R07, which reveals synchronous change with the phytoplankton community ( Figure 3 ). The concentrations of terrestrial biomarkers (i.e., long-chain n-alkanes, n-alkanols, n-alkanoic acids, C₃₂ 1,15-diol) were used to trace the terrestrial input into the Chukchi Sea, considering that end-member proxies, such as BIT (Hopmans et al., 2004), δ ¹³C_org (Naidu et al., 1993) and C/N (Middelburg and Nieuwenhuize, 1998), are influenced by changes in marine organic matters. In R07, these terrigenous biomarkers showed increasing trends since ca. 1946 ( Figure 3 ), which indicate the increased terrestrial input. As indicated in Section 5.1, variations in the biomarker concentrations in Core R07 are primarily caused by changes in riverine input that was transported to this area. The Yukon River is among the six largest rivers in the Arctic region (Cooper et al., 2008), and plays a particularly important role in the supply of terrestrial organic matter into the Chukchi Sea (Stein, 2008; Darby et al., 2009; Harvey and Taylor, 2017). The estimated annual freshwater discharge of the Yukon River is approximately 205 km³, with a suspended matter flux of 5.4 × 10⁷ t (Stein, 2008). According to previous studies, river discharge to the Arctic Ocean is increasing since 1940s (Peterson et al., 2002; Wu et al., 2005), in particular, the discharge records at Pilot Station showed an increasing trend of Yukon River discharge (Ge et al., 2013). Increased river discharge from the Yukon River and others along the Alaska coast is entrained in the ACW and transported northward into the Chukchi Sea (Mathis et al., 2005; Iken et al., 2010), thereby elevating the terrestrial organic matter input, as depicted in Core R07 before ca. 1983 ( Figure 3 ).

However, under the increasing riverine input, terrigenous input indicated by terrestrial biomarkers showed a decreasing trend after ca. 1983 ( Figure 3 ), which may be attributed to changes in the structure of the water mass. Previous studies showed that the ACW plays a central role in transporting terrestrial materials into the Chukchi Sea shelf (Naidu et al., 2004; Harvey and Taylor, 2017; Zhang et al., 2021). With the change in water mass structure, i.e., reduced ACW in this area, the terrestrial organic matter input transported by the ACW to the study area significantly decreased. Consequently, the concentrations of terrestrial-derived organic matter reduced since ca. 1983 ( Figure 3 ).

The change in water mass structure is likely to be associated with a phase shift in the AO and/or unsynchronized flow changes in the ACW and BSW. The AO is an annular mode of atmospheric circulation, with an annular (circled) center covering the entire Arctic region (Thompson and Wallace, 1998; Wang et al., 2014). AO-derived wind anomaly is cyclonic and anticyclonic during the positive and negative phases, respectively ( Figure S5 ; Proshutinsky and Johnson, 1997; Wu et al., 2006; Wang et al., 2014). It has been reported that wind can change the surface distribution of water masses in the Chukchi Sea (Winsor and Chapman, 2004). With the summer AO shift to a predominantly positive phase at approximately 1983 ( Figure 3 A ), a cyclonic wind anomaly occurred and caused a westerly wind anomaly in the Chukchi Sea shelf ( Figure S5 ). Derived by the westerly wind anomaly, the ACW moved eastward since ca. 1983, thereby reducing its impact on the Core R07 area ( Figure 2 ).

However, after the predominant phase of AO shift from positive to negative after ca. 2003, there is still elevated B/D ratios and decreasing trend in terrigenous biomarker ( Figure 3 ), which may be attributed to the unsynchronized changes in the ACW and BSW. Mooring data showed that there was an increase in the annual mean transport of BSW into the Arctic at least since 2002, whereas no evident trend for the annual mean flow of the ACW (Woodgate, 2018). The unsynchronized changes in water masses also can cause the elevated proportion of BSW and the reduced impact of ACW in the Core R07 area. However, since these mooring data are only available from 2002, it is uncertain whether the structure change of water mass at ca. 1983 was caused by the asynchronous change. In addition, considering the complexity of this area, more research is needed to explore the exact processes and factors of changes in water mass structure.

In the Arctic Ocean, phytoplankton production was controlled by the extreme seasonality of the polar environment (Nelson et al., 2014). Stimulated by increasing irradiance, the annual phytoplankton growth starts with ice algae at the ice-water interface (Gradinger, 2009). As the sea ice thins and retreats in the spring or summer, light availability gradually increases to be sufficient to trigger the blooms of under-ice phytoplankton, which has been observed beneath the 0.8- to 1.3-m-thick sea ice in the Chukchi Sea shelf (Arrigo et al., 2012). This under-ice phytoplankton bloom was also found in Core R07. Low but significant amounts of phytoplankton biomarkers (avg. of 0.47 μg/g, 0.51μg/g and 178.63 ng/g for barssicasterol, dinosterol and C30 diols, respectively) were detected in sediments deposited during ca. 1946 to ca. 1953 when SST was –1.8°C and SIC was almost 100% (i.e. sea ice cover) throughout the year ( Figure 4 ). As sea ice continues to melt, high phytoplankton productivity can last until the nutrient in water is depleted or the ice cover forms. A one-year sediment trap experiment conducted in the Chukchi Sea (DM: 74.6°N, 158.23°W) showed that phytoplankton biomarkers (brassicasterol and dinosterol) fluxes begin to rise as the sea ice melts and fall back to low levels before nutrient depletion or ice cover formation (Bai et al., 2019). Annual sea ice pattern in the area of R07 is similar with that of DM, i.e. sea ice starts to melt in June, and is completely formed in December ( Figure S1 ), indicating similar growth patterns of phytoplankton between areas of R07 and DM. Besides, the subsurface phytoplankton blooms and near-bottom blooms have been observed in the Chukchi Sea shelf (Lowry et al., 2015; Stabeno et al., 2020), which might also occur in the area of Core R07.

In recent decades, the Arctic Ocean has undergone drastic changes, including an unprecedented warming (Serreze and Barry, 2011) and accelerated seasonal sea ice retreat (Stroeve et al., 2012). In the study area, SST showed an increasing trend and SIC was decreasing from 1952 ( Figure 4 ), which indicated an earlier sea ice melting and, therefore, longer growth season for phytoplankton (Arrigo et al., 2008; Arrigo and van Dijken, 2015). As discussed in Section 5.2, the discharge from the Yukon River into the ACW increased in the past decades (Peterson et al., 2002; Ge et al., 2013). The increased terrestrial material transported by ACW elevated the terrigenous organic matter and nutrients into the study area ( Figure 3 ), which are favorable for phytoplankton growth (Terhaar et al., 2021). Consequently, under the influence of increased temperature, longer growth season and increased terrigenous input, phytoplankton primary productivity in the Chukchi Sea shelf significantly increased, as reflected by the dramatic increase in marine-sourced lipids in Core R07 from ca. 1946 ( Figure 4 ).

However, with the constant or slightly increased temperature between ca. 1983 to ca. 2000, there is a significant decrease in marine-sourced lipids, which is approximately half of the increase from ca. 1946 to ca. 1983 ( Figures 4 , S6 ), indicating a nonnegligible decrease in phytoplankton primary productivity. Similar changes also were found in a sediment core collected approximately 150 km south of Core R07, in which cyst density of a dinoflagellate species exhibited a significant decrease after a period of increase (Natsuike et al., 2013). We speculate that this decrease in phytoplankton productivity is likely caused by the change in the water mass structure, which has been suggested can impacts both the phytoplankton community and phytoplankton productivity (Grebmeier et al., 2006). As proposed by previous studies, primary productivity is controlled by the interaction of light and nutrients (Popova et al., 2010; Terhaar et al., 2021). As indicated in Section 5.2, there was a change in water mass structure at ca. 1983; more specifically, the contribution from the ACW decreased, while that from BSW increased. Compared to the ACW, BSW is colder and saltier (Grebmeier et al., 2006), and these characteristics of cold and salt accelerate water mixing, and consequently, increase turbidity. Light limitation induced by high turbidity has been reported to decrease productivity in the East China Sea and South Yellow Sea (Wu et al., 2016), Southern Ocean MIZ (Fitch and Moore, 2007), and coastal waters surrounding Qatar (Arabian Gulf) (Quigg et al., 2013). A similar phenomenon may also exist in the Chukchi Sea, as the enhancement experiments have showed that light is an important controlling factor for the phytoplankton productivity rates in this area (Lee et al., 2013). Notwithstanding that the AO shift led to an increase in nutrient-rich BSW in this area ( Figure 2 ), nutrients might not be the main limiting factor of primary productivity as its minimal effect on primary productivity rates under high nutrient concentrations (Lee et al., 2013). In addition, high turbidity can reduce water transparency and thus diminish the light intensity reaching deep water layers, which may hinder or attenuate subsurface and near-bottom blooms. Thus, light limitation caused by change in water mass structure may decrease the phytoplankton productivity in the water column, as reflected by the decreased concentrations of marine biomarkers in Core R07 ( Figure 4 ). This result suggests that, in addition to global warming and sea ice retreat, other factors (such as change in water mass structure) significantly affect phytoplankton primary productivity in the Chukchi Sea shelf.

Following the decrease, the biomarker concentrations in R07 showed significant increase in phytoplankton primary productivity since ca. 2000, which may be attributed to the warming and accelerated sea ice loss ( Figure 4 ; Budikova, 2009; Stroeve et al., 2012). Warming enhances stratification and lowers the mixing depth, and both conditions increase light availability (Winder and Sommer, 2012). The higher SST and decreased SIC indicate a longer growth season for phytoplankton. In general, the primary production rates in open waters are significantly higher than those below snow-covered sea ice because of higher light intensities (Mathis et al., 2014). Longer phytoplankton growing season, increased light availability, and high nutrient content have been reported to increase primary production in the Arctic (Arrigo et al., 2008; Arrigo and van Dijken, 2015). Owing to these changes, the productivity of diatoms, dinoflagellates, and eustigmatophytes based on biomarkers increased from ca. 2000 to ca. 2011 in the Core R07 ( Figure 4 ). The result is consistent with the increase in the annual net primary production (42%) in the Chukchi Sea shelf from 1998 to 2012 (Arrigo and van Dijken, 2015).