Sources of particulate organic matter in the Chukchi and Siberian shelves: clues from carbon and nitrogen isotopes

1. Introduction

As one of the most sensitive regions to global climate change, the Arctic Ocean is the freshest ocean with a large amount of freshwater input (Aagaard and Carmack, 1989). The intensified hydrologic cycles throughout the Arctic (Rawlins et al., 2010) have led to a decadal increase of discharge from Arctic rivers (McClelland et al., 2006; Déry et al., 2009), which has further increased the flux of dissolved inorganic carbon (Gordeev et al., 1996; Tank et al., 2012) and organic matter (Lobbes et al., 2000) to the Arctic Ocean. The enhanced river discharge, as well as the increasing atmosphere temperature and CO₂, are dramatically changing the Arctic Ocean ecosystem, which has been characterized by deepening of the nutricline (McLaughlin and Carmack, 2010), strengthening of acidification (Semiletov et al., 2016; Qi et al., 2017) and changing in marine biomass community (Li et al., 2009), etc. The stable carbon and nitrogen isotopes (δ¹³C and δ¹⁵N), and the C/N ratio of POM provide important insights into the source and fate of organic matter in the ocean, and thus have been widely used to reveal changes in planktonic production and carbon cycle in the Arctic (Kling et al., 1992; Naidu et al., 2000; Semiletov et al., 2005; Magen et al., 2010; Zhang et al., 2012; Pineault et al., 2013; Oxtoby et al., 2016).

The Chukchi Sea and the East Siberian Sea are the most productive waters in the western Arctic and the eastern Arctic, respectively, both of which receive a large amount of river water and terrestrial organic matter from the Arctic rivers (Arrigo and van Dijken, 2011). The terrestrial POM in the Chukchi Sea and the East Siberian Sea are mainly derived from the North American rivers (e.g., Yukon River and Mackenzie River) and the Russian rivers (e.g., Lena River, Kolyma River and Indigirka River) respectively. These terrestrial POM have similar C/N ratios (12–40, Lobbes et al., 2000; Guo and Macdonald, 2006; Guo et al., 2012) and isotopic compositions (δ¹³C values ranged from −28‰ to −26‰, McClelland et al., 2008; Vonk et al., 2010; Guo and Macdonald, 2006, and δ¹⁵N values were about 2.0‰, McClelland et al., 2008; Guo and Macdonald, 2006). Due to the complex effects of sea ice, stratification, dissolved inorganic carbon, nutrients and community species, the concentration and isotopic composition of marine POM derived from autochthonous production show large spatial variations in the Arctic Ocean (McLaughlin and Carmack, 2010; Arrigo et al., 2012; Zhang et al., 2012; Pineault et al., 2013). The δ¹³C value is as high as −11.4‰ in ice-algal POM (Tremblay et al., 2006; Pineault et al., 2013) and less than −27‰ in the ¹³C-depleted phytoplankton (Zhang et al., 2012; Pineault et al., 2013). The C/N ratio of POM also varies greatly. The high cellular C/N ratios were observed under N-limited conditions (Körtzinger et al., 2001; Vrede et al., 2002) and the low cellular C/N ratios occurred in small phytoplankton and light limitation conditions (Taguchi, 1976; Chalup and Laws, 1990; Crawford et al., 2015). The western and the eastern Arctic Ocean have different freshwater discharge and ecological characteristics. It is of significance to reveal the origin and variations of POM in the Chukchi Sea and the East Siberian Sea.

Polynyas, the areas of open water in seas that are usually ice-covered, are found more or less all around the Arctic Ocean, at least during some time of the year (Barber and Massom, 2007). Many coastal polynyas are open up during winter when strong wind blows off the coast, and disappear in summer. These polynyas are ice factories in which sea ice is produced while the brine is rejected and transported to depth. The brine-enriched water that is transported will be replaced by shelf water originating from adjacent oceans (Hoppema and Anderson, 2007). The polynyas are biologically productive regions in the Arctic Ocean due to their early bloom after the polar night (Tremblay et al., 2002; Arrigo, 2007). For example, the Wrangel Island polynya, located in the mixing area of Pacific waters and Atlantic waters, showed the highest production in the Chukchi and Siberian shelves (Anderson et al., 2011). The results of the size-fractionated biomass indicated that phytoplankton community was dominated by small cells in the Northeast Water polynya (Pesant et al., 1996), the North Water polynya (Klein et al., 2002) and the Arctic Ocean (Gosselin et al., 1997). Unfortunately, little is known about particulate organic matter and its isotopic composition in the Arctic polynyas. Tremblay (2006) found that the δ¹³C value of bottom-ice POM was as low as −24.9‰ in the initial stage of the early bloom in the North Water polynya, which was attributed to the ¹³C-depleted pelagic bloom. Considering the significant contribution of early bloom on primary production in the Arctic polynyas, the bloom-produced POM in the early stage may have an impact on temporal and spatial variations and characteristics of POM in summer.

The C/N ratio and δ¹³C have been widely used to distinguish organic matter between autochthonous and allochthonous (mostly terrestrial) sources, based on the fact that marine and terrestrial organic matter have distinctly different δ¹³C values. The accuracy of distinguishing between marine and terrestrial sources depends on the end-member characteristics of the C/N ratio and δ¹³C (Kuliński et al., 2014; Kumar et al., 2016). Based on the C/N ratio and δ¹³C of POM, a three endmember mixing model was used to trace the origin of POM in southern Beaufort Sea by further dividing the marine source POM into refractory and fresh POMs (Magen et al., 2010).

In this study, seawater samples were collected from the Chukchi Sea and East Siberian Sea for measurements of carbon and nitrogen content, C/N ratio, δ¹³C and δ¹⁵N of suspended particulate organic matter. The spatial distributions of C/N ratio, δ¹³C and δ¹⁵N were depicted, in which an anomalous characteristic was found in the Wrangel Island polynya. A three endmember mixing model of C/N ratio and δ¹³C value was used to quantify the contribution of POMs from different sources. This study will deepen our understanding of biogeochemical cycle of organic matter in the Chukchi Sea and East Siberian Sea.

2. Methods

2.1 Study area

The Chukchi Sea is connected to the Bering Sea through the Bering Strait in the south, and connects to the East Siberian Sea through the Long Strait in the west. The Chukchi and Siberian shelves are shallow, with a depth of less than 60 m and several shoals and canyons appearing. The circulations in the shelves are mainly restricted by the Pacific inflow, the Siberian Coastal Current (SCC, blue arrow) and the bathymetry (Fig. 1). The Pacific inflows are divided into three branches after leaving the Bering Strait, i.e., the Anadyr Water (AW) flows northwest to the East Siberian Sea through the Herald Canyon, the Bering Shelf Water (BSW) flows northward to the Chukchi Plateau and the Canada Basin through the Central Canyon, and the Alaska Coastal Water (ACW) flows northeast to the Canada Basin through the Barrow Canyon (Woodgate et al., 2005; Weingartner et al., 2005; Martin and Drucker, 1997). The SCC flows eastward from the Laptev Sea into the East Siberian Sea through the New Siberian Islands, then one branch of the SCC turns north entering the Makarov Basin and the other continues to flow eastward and mix with the Pacific water around the Wrangel Island (Talley et al., 2011). The pathway of the Siberian Slope Current (SSC) is uncertain and may sometimes flow eastward along the Siberian slope into the Canada Basin (Talley et al., 2011).

Figure  1.  The sampling locations, bathymetry and schematic circulations around the Chukchi Sea and the East Siberian Sea. The sampling locations during the Sino-Russian Joint Scientific Expedition and the 7th Chinese National Arctic Research Expedition were marked in squares and red circles, respectively. The grey squares represent stations where only surface samples have been collected. The bathymetric contours of 50 m, 100 m and 250 m are shown in blue, grey and red lines, respectively. Surface currents are marked with arrows of different colors, including Anadyr Water (AW, red arrows), Bering Shelf Water (BSW, brown arrows), Alaska Coastal Water (ACW, green arrows), Siberian Coastal Current (SCC, blue arrows), Siberian Slope Current (SSC, purple arrows) and Beaufort Gyre (BG, grey arrows). River runoff are marked in yellow arrows, including Lena River, Indigirka River, Kolyma River and Mackenzie River.

DownLoad: Full-Size Img PowerPoint

2.2 Sampling and measurements

Samples for measurements of δ¹⁸O in seawater and POC, PN, δ¹³C_POC and δ¹⁵C_PN of suspended particles were collected from July to September, 2016 in the Chukchi Sea and the Canada Basin (red dots in Fig. 1) onboard R/V Xuelong during the 7th Chinese National Arctic Research Expedition (CHINARE), and from August to September, 2016 in the East Siberian Sea (blue squares in Fig. 1) onboard R/V Academician LavrenJeev during the China−Russian Joint Cruise (LV-77 Cruise). A total of 69 stations were surveyed in the Chukchi Sea, the East Siberian Sea and the Canada Basin. Samples were collected from the entire water column, but only the data of the samples shallower than 200 m were analyzed here. Temperature and salinity were measured in-situ via a Sea-bird CTD sensor (SBE911 plus) during the 7th CHINARE. Temperature, salinity and Chl a concentration were measured onboard using an Alec CTD during the LV-77 Cruise.

Seawater samples for δ¹⁸O measurement were collected from Niskin bottles mounted on the CTD rosettes. Immediately after collection, the seawater sample was transferred to a 30 mL polyethylene bottle and kept in gas-tight, and brought back to the shore laboratory. In the shore laboratory, the sample was transferred into a 2 mL glass vial and sealed with Parafilm^®. The isotopic composition of oxygen was analyzed via a Wavelength-scanned Cavity Ring-down Spectroscopy (WS-CRDS) in a laser spectroscopy analyzer (Picarro L2140-i, USA) (Skrzypek and Ford, 2014). Samples were prepared and run in batches. Each sample was conducted by eight injections of 2 μL seawater. During the measurements, five internal standards with different δ¹⁸O values (−6.91‰, −4.86‰, −2.79‰, −0.23‰, and 2.15‰) were synchronously measured after several samples for calibration. These standards were periodically calibrated against the Vienna Standard Mean Ocean Water (VSMOW) and the Standard Light Antarctic Precipitation (SLAP) from the International Atomic Energy Agency (IAEA). The δ¹⁸O value, referenced to the VSMOW, was reported in per mil with a precision of better than ±0.03‰:

Seawater samples for POC, PN, δ¹³C_POC and δ¹⁵N_PN measurements were collected from Niskin bottles mounted on the CTD rosettes. The sample was filtered through a precombusted (450°C, 4 h) Whatman GF/F membrane onboard. The filters were kept frozen to shore laboratory for further treatments. After dried at 60°C, the membrane containing the particles was divided into subsamples. One quarter of the membrane was acid-fumed to remove inorganic carbon for subsequent measurements of particulate organic carbon (POC) and its isotopic composition (δ¹³C_POC). Half of the remained membrane was untreated to measure the particulate nitrogen (PN) and its isotopic composition (δ¹⁵N_PN). POC and PN contents as well as δ¹³C_POC and δ¹⁵N_PN were determined by a Thermo Finnigan Delta V mass spectrometer interfaced with an elemental analyzer (Carlo Erba NC 2500).

The internal standard (C₆H₆N₂O) was used to establish a standard curve for the calculation of C and N contents. The detection limits for carbon and nitrogen content were both 0.1 μmol. The isotopic reference standards for C and N are PeeDee Belemnite and atmospheric N₂, respectively. To monitor the instrument performance and data quality, certified standard samples, IAEA-C8 (δ¹³C=−18.3‰) and USGS40 (δ¹³C=−26.4‰), and IAEA-N2 (δ¹⁵N=20.3‰) and IAEA-N3 (δ¹⁵N=4.7‰) were used for δ¹³C and δ¹⁵N determination respectively. Both the analytical precisions of δ¹³C and δ¹⁵N are ±0.2‰. The values of δ¹³C and δ¹⁵N were defined as follows:

2.3 Quantification of freshwater components

Seawater in the Chukchi Sea, the East Siberian Sea and the Canada Basin is considered as a mixture of the Atlantic water (AW), meteoric water (MW, including runoff and precipitation) and sea ice meltwater (SIM). As a result, the fractions of these three components were calculated from the mass balance of salinity and δ¹⁸O as follows:

where f, δ and S refer to the fraction of each component, δ¹⁸O and salinity respectively. Subscript AW, MW, SIM and obs represent the Atlantic water, meteoric water, sea ice meltwater and the observed value of each sample respectively. The characteristic salinity and δ¹⁸O values of the AW, MW and SIM used in our calculations were adopted from Li et al. (2017) and listed in Table 1.

Table  1.  The characteristic δ¹⁸O and salinity of three endmembers used to quantify the freshwater components in the Chukchi Sea and the Siberian Sea

Endmember	δ18O/‰	S	References
Atlantic water	0.3	35	Frank (1996); Östlund and Hut (1984)
Meteoric water	−20	0	Cooper et al. (2005)
Sea ice meltwater	−2	4	Macdonald et al. (1999, 2002)

 | Show Table

DownLoad: CSV

3. Results

3.1 Hydrographic features

The scatter plot of temperature, salinity and δ¹⁸O was showed in Fig. 2. The waters in the Chukchi and Siberian shelves were divided into three parts, i.e., Pacific water in the Chukchi shelf; the East Siberian Shelf Water in the western Siberian shelf and the mixed water of the Pacific, and the East Siberian Shelf water in the central and eastern Siberian shelf. The Pacific water was characterized by high temperature (>2°C), high salinity (>30) and high δ¹⁸O values (>−2‰). The East Siberian Shelf Water was characterized by low temperature (<4°C), low salinity (<26) and low δ¹⁸O values (<−4‰). The hydrological characteristics of the mixed water were between the above two water masses. The temperature and salinity in the upper 200 m water column in the basins were lower than 0°C and ranged from 26 to 34 respectively. The δ¹⁸O values ranged from −4‰ to 0‰.

Figure  2.  The scatter plot of temperature, salinity and δ¹⁸O value in the upper 200 m around the Chukchi and East Siberian seas. The color bar shows δ¹⁸O values of sample. Samples influenced by the Pacific inflow, the East Siberian Shelf Water and the mixed shelf water are marked in different colored frames.

DownLoad: Full-Size Img PowerPoint

3.2 POC and PN

The concentrations of POC and PN varied from 0.18 μmol/L to 37.77 μmol/L, and from below detection to 4.40 μmol/L, with average values of (4.65±5.65) μmol/L and (0.77±0.77) μmol/L, respectively. The C/N ratios (mol/mol) ranged from 1.7 to 17.3 with an average of 6.1±2.6. The high C/N ratios (>8) appeared near the estuary, indicating the effect of terrestrial organic matter characterized by a high C/N ratio (>12) (Lobbes et al., 2000; McClelland et al., 2016). Although the average value of C/N ratio was close to the Redfield ratio (6.5 to 8.7) (Redfield, 1958; Takahashi et al., 1985), some C/N ratios were between 2 to 5, which was significantly lower than the average ratio of marine organic matter (5.6) (Copin-Montegut and Copin-Montegut, 1983) and that in the high latitude (6.0) (Martiny et al., 2013b). Considering that the shelves are significantly affected by terrestrial organic matter, and the C/N ratio of terrestrial organic matter is much higher than that of marine organic matter, the true C/N ratio of organic matter produced by marine organisms may be lower than the measured value. The measured C/N ratio in our study areas was as low as 2.0, which was consistent with the cellular C/N ratio of small phytoplankton Phaeodactylum tricornatum (2.7) (Taguchi, 1976) and the recently reported C/N ratios of natural POM and phytoplankton (Crawford et al., 2015). The physiological basis for the low C/N ratio was mainly related to light limitation as phytoplankton growing under low irradiance accumulate less carbon storage polymers (Chalup and Laws, 1990; Cronin and Lodge, 2003; Martiny et al., 2013a) and small cell size of phytoplankton (<5 μm) has a lower uptake ratio (ρ_C : ρ_N) than the larger one (>5 μm) (Crawford et al., 2015). Recent studies have shown that the dominance of picophytoplankton and bacterioplankton is increasing in the Arctic Ocean, which may produce organic matter with a low C/N ratio (Li et al., 2009; He et al., 2012).

3.3 δ¹³C and δ¹⁵N in particulate organic matter

The δ¹³C values of POM in our study areas varied from −29.5‰ to −17.5‰ (mean value is −25.9‰±2.0‰) and the δ¹⁵N values ranged from 3.9‰ to 13.1‰ (mean value is 8.0‰±1.6‰). The terrestrial organic matter in the Arctic was generally characterized by low δ¹³C values (from −30‰ to −25‰) (Lobbes et al., 2000; McClelland et al., 2008; Vonk et al., 2010), low δ¹⁵N values (about 2.0‰) (Guo and Macdonald, 2006; McClelland et al., 2008) and high C/N ratios (>12) (Lobbes et al., 2000; Guo and Macdonald, 2006; McClelland et al., 2016). The characteristics of marine biogenic organic matter in the Arctic Ocean were significantly different from those of terrestrial sources. Their δ¹³C and δ¹⁵N values ranged from −22‰ to −17‰ (Descolas-Gros and Fontugne, 1985; Tagliabue and Bopp, 2008) and from 6‰ to 9‰ (Tremblay et al., 2006; Pineault et al., 2013), respectively, and the C/N ratios were around 6 (Copin-Montegut and Copin-Montegut, 1983; Martiny et al., 2013b). The relationship between δ¹³C_POC and C/N ratio in our study areas showed that most of the data fall below the mixing zone of terrestrial and marine POM, indicating that other sources may cause a decrease in δ¹³C_POC and C/N ratio. In addition, with the decrease of δ¹³C_POC and C/N ratio, the δ¹⁵N_PN was generally increasing (Fig. 3). The decrease of the δ¹³C_POC and the C/N ratio may be attributed to a higher dissolved CO₂ concentration (Anderson et al., 2011; Zhang et al., 2012) and a large proportion of small phytoplankton in the East Siberian Sea and Canada Basin (Li et al., 2009; Coupel et al., 2015; Yun et al., 2015; Bhavya et al., 2018). Compared to large phytoplankton, small phytoplankton has a lower cellular C/N ratio and a higher carbon fractionation during assimilation (Taguchi, 1976; Burkhardt et al., 1999; Crawford et al., 2015; Oxtoby et al., 2016). The increase in δ¹⁵N_PN may be related to the early bloom of phytoplankton, a phenomenon often occurring in the Arctic polynyas (Arrigo and van Dijken, 2004; Tremblay et al., 2006). In addition to phytoplankton bloom, sediment resuspension (Lehmann et al., 2002), POM remineralization (Landrum et al., 2011) and denitrification (Buchwald, 2013) may also be the cause of δ¹⁵N_PN increase.

Figure  3.  The scatter plot of C/N ratio and δ¹³C_POC in the upper 100 m water column around the Chukchi Sea and East Siberian Sea. The color bar shows the δ¹⁵N value of sample. The average values of terrestrial and marine POM are marked in grey diamonds and their mixing zone is marked by grey shadow. The black arrow shows the decrease of δ¹³C_POC and C/N.

DownLoad: Full-Size Img PowerPoint

3.4 Surface distribution

The surface distributions of temperature (°C), salinity, δ¹⁸O (‰), fraction of meteoric water (f_MW, %) and sea ice meltwater (f_SIM, %), POC (μmol/L), PN (μmol/L), δ¹³C_POC (‰) and δ¹⁵N_PN (‰) were displayed in Fig. 4.

Figure  4.  Surface distribution of temperature (b, color), salinity (b, contour), δ¹⁸O (c), f_MW (d), f_SIM (e), POC (f), PN (g), C/N ratio (g), δ¹³C (h) and δ¹⁵N (i) around the Chukchi Sea and the East Siberian Sea. The simplified bathymetry is shown in Fig. 4a. Geographic locations marked with abbreviations are Bering Strait (BS), Chukchi shelf (CS), Wrangel Island (WI), Siberian shelf (SS), Makarov Basin (MB), Chukchi Plateau (CP) and Canada Basin (CB). The distribution of salinity and C/N ratio is shown in Figs 4b and g with the white contour respectively. The geographic locations marked in Fig. 4c are Siberian West Shelf (SWS), Siberian Middle Shelf (SMS) and Siberian East Shelf (SES).

DownLoad: Full-Size Img PowerPoint

The surface temperature and salinity decreased westward from the Chukchi Sea (T>6°C, S>30) to the East Siberian Sea (T<4°C, S<26), and a sharp front appeared near the Wrangel Island (180°E) (Figs 4a and b), indicating the combined effects of the warm, saline Pacific water and the cold, fresh Siberian shelf water (Woodgate et al., 2005; Talley et al., 2011). Surface temperature decreased to the north and reached below 0°C in the basins (Fig. 4a). The spatial variation of surface salinity was more complicated than temperature. Surface salinity decreased northward from the Chukchi Sea to the Chukchi Plateau and then to the Canada Basin (S<28), but increased northward from the East Siberian Sea to the Makarov Basin (Fig. 4b), reflecting the accumulation of freshwater in the Canada Basin (Alkire et al., 2015) and the influence of river runoff on the coast of the Eastern Siberian Sea (Lobbes et al., 2000). The maximum surface salinity appeared in the slope region off the Siberian, which may be due to the upwelling of subsurface water.

The spatial distribution of surface δ¹⁸O showed a pattern similar to salinity, with a decrease from the Chukchi Sea (>−2‰) to the East Siberian Sea (<−6‰). Whether in the Makarov Basin or in the Canada Basin, the δ¹⁸O values in surface seawater were stable at about −3‰ (Fig. 4c). According to the spatial variations of salinity and δ¹⁸O, surface waters in different regions in the Siberian shelf showed different characteristics. The Siberian West Shelf (SWS) has the lowest salinity and δ¹⁸O value, while the Siberian East Shelf (SES) has the highest salinity and δ¹⁸O value, and the Siberian Middle Shelf (SMS) shows the mediate salinity and δ¹⁸O value (Fig. 4c).

The variation pattern of f_MW was similar to δ¹⁸O, suggesting that meteoric water plays a dominant role in regulating δ¹⁸O distribution in the high Arctic (Li et al., 2017). The distribution of f_MW showed a continuous decrease from the SWS to the SMS, SES and the Chukchi shelf (CS), indicating that river water has a greater impact on the Siberian West Shelf and decreases eastward (Fig. 4d). The value of f_SIM and its distribution were affected by the formation and melting of sea ice. The negative f_SIM represents a net loss of sea ice, and the positive value represents a net accumulation of sea ice meltwater. The negative f_SIM was observed in the SWS (−10% to −5%) and Makarov Basin (−5% to 0), whereas the positive f_SIM was observed in the SMS, SES, CS (0 to 15%) and Canada Basin (0 to 5%), with the highest f_SIM (>10%) occurring in the northern Chukchi shelf (Fig. 4e), showing the loss of sea ice in the Eastern Arctic and accumulation of sea ice in the Western Arctic (Alkire et al., 2015). The high positive f_SIM values in the SMS and the northern Chukchi shelf were due to the regional accumulation of sea ice melting water.

The surface POC showed the highest concentrations (>6 μmol/L) in the SWS (Fig. 4f) due to the terrestrial influence from the Russian rivers (Anderson et al., 2011). Unlike the SWS, the high POC concentration observed in the CS (>4 μmol/L) (Fig. 4f) is mainly due to its high primary productivity (Arrigo et al., 2014). In general, the surface POC concentration decreased northward from the shelves (>2 μmol/L) to the basins (<2 μmol/L), reflecting the spatial variation of primary productivity.

The spatial distribution of surface PN concentration was very similar to that of POC, with high values in the SWS and the CS, and low values in the basins (Fig. 4g). The resultant C/N ratios in the shelves (<8, most of them <6) were generally lower than those in the basins (6 to 8). In the shelves, the SWS exhibited a high C/N ratio (6 to 8), while the SES showed a low value (<4) (Fig. 4g). It is worth noting that surface C/N ratio in the basins is close to the Redfield ratio and the average ratio of marine organic matter (Redfield, 1958; Martiny et al., 2013b).

The surface δ¹³C_POC ranged from −28‰ to −22‰, with high values in the CS (−26‰ to −22‰) and low values in the SMS, SES and Canada Basin (−28‰ to −26‰) (Fig. 4h). The surface δ¹³C_POC decreased by 0.8‰ on average from the Chukchi Sea to the Canada Basin, which is similar to the results of Zhang et al. (2012) and is the effect of increased dissolved CO₂ concentration in seawater. It is noted that the δ¹³C_POC value in the East Siberian Sea is significantly lower than that in the Chukchi Sea. The surface δ¹⁵N_PN ranged from 3.9‰ to 11.3‰, showing a general decrease from the shelf to the basin. In the shelves, the highest δ¹⁵N_PN was observed in the SES, while the lowest in the SWS (Fig. 4i). Considering that terrestrial organic matter is often depleted in ¹⁵N (Guo and Macdonald, 2006), it is a paradox that the low δ¹³C and high δ¹⁵N occur simultaneously in the SMS and the SES (Fig. 4i). In fact, this phenomenon has been found in other estuaries or shelves, as a result of complex biochemical processes (Kumar et al., 2016). Burkhardt et al. (1999) demonstrated that stable carbon isotope fractionation ε_p is inversely correlated with cell size among the diatoms. Thus the low-δ¹³C and high-δ¹⁵N POM in the SMS and SES might be the result of bloom of small size phytoplankton.

3.5 Latitudinal distribution in the Chukchi and Siberian shelves

The distributions of temperature (°C), salinity, δ¹⁸O (‰), f_MW (%), f_SIM (%), POC (μmol/L), PN (μmol/L), δ¹³C_POC (‰) and δ¹⁵N_PN (‰) along 148°E to 168°W in the Chukchi and Siberian shelves were shown in the Fig. 5. In the SWS and SMS, all parameters varied little vertically due to the shallow depth (20–30 m) and strong mixing. In contrast, the water column in the SES and CS was stratified to the upper warm layer (0–20 m) and the lower cold layer (20 m to bottom). Compared with the upper warm water, the cold water below the depth of about 20 m showed the higher salinity, more negative f_SIM, and the higher POC, PN, δ¹³C and δ¹⁵N values.

Figure  5.  The longitudinal distribution of temperature (a, color), salinity (a, contour), δ¹⁸O (b), f_MW (c), f_SIM (d), POC (e), PN (f, color), C/N ratio (f, contour), δ¹³C (g) and δ¹⁵N (h) in the Chukchi and the Siberian shelves. The distribution of salinity and C/N ratio is shown in Figs 5a and f with the white contour respectively. The geographic locations separated by white dashed lines are Siberian West Shelf (SWS), Siberian Middle Shelf (SMS), Siberian East Shelf (SES) and Chukchi Shelf (CS).

DownLoad: Full-Size Img PowerPoint

The sectional distribution of various parameters points to the following understandings: (1) The SES is a mixing zone for the Siberian shelf water and the Pacific inflow. The contours of salinity and δ¹⁸O showed that the light, fresh and ¹⁸O-depleted Siberian shelf water is mixed with the dense, saline and ¹⁸O-enriched Pacific inflow in the SES (Figs 5a and b). The mixing of seawater from the Atlantic and the Pacific promoted a pelagic bloom in the SES (Tremblay et al., 2006; Anderson et al., 2011). (2) The high concentrations of POC and PN observed at the bottom of the SES and the CS may be due to sediment resuspension. Resuspension of sediments is common in the Arctic shelves, which affects the POM and its isotopic composition in the water column (Lehmann et al., 2002). (3) Abnormal values of the f_SIM, C/N ratio, δ¹³C_POC and δ¹⁵N_PN were observed in the SES (Figs 5d, g, f and h). For example, the f_SIM in the SES was lower than the surrounding waters, indicating a significant loss of sea ice. The lowest C/N ratio, the lowest δ¹³C_POC and the highest δ¹⁵N_PN were all present in the SES. The simultaneous occurrence of these anomalies raises a question of what biogeochemical process plays a role in the SES, which is one of the issues that will be discussed in this study.

4. Discussion

4.1 Variation in isotopic composition

The POM in the Arctic seas is composed of terrestrial exogenous organic matter and marine authigenic organic matter, but contributions of these two sources will change in different regions. The covariation relationship between the averaged f_MW and the averaged POC or PN concentration indicated that the POM in the SWS and the SMS is more significantly affected by terrestrial source (a positive relationship between the f_MW and the POC or PN), while that in the SES, the CS and the basins (including the MB, CP and CB) may be mainly contributed by marine organisms (a negative relationship between the f_MW and the POC or PN) (Fig. 6a). Although the POMs in different regions have different sources, the δ¹³C_POC values of all regions were close to those of terrestrial organic matter, and the δ¹⁵N_PN values were close to marine organic matter (Figs 6b and c). It seems that δ¹³C_POC and δ¹⁵N_PN give inconsistent information, suggesting that the POC and PN are different in sources, turnover rates, and cycling pathways in the Arctic Ocean. Considering that δ¹³C and δ¹⁵N values of POM in the Arctic rivers are less changed than marine organic matter (Descolas-Gros and Fontugne, 1985; Tremblay et al., 2006; McClelland et al., 2008), this inconsistency was more likely due to the variation in isotopic composition of marine organic matter. As a result, we will focus below on possible factors that regulate the variation of δ¹³C_POC and δ¹⁵N_PN in the regions dominated by marine organic matter, i.e., the SES, the CS and the basins.

Figure  6.  The averaged POC and PN (a), f_MW (a), δ¹³C (b) and δ¹⁵N (c) in different regions. The solid and dashed lines represent the averaged δ¹³C and δ¹⁵N value of marine POM and terrestrial POM respectively. Regions shown in the x-axis include Siberian West Shelf (SWS), Siberian Middle Shelf (SMS), Siberian East Shelf (SES), Chukchi Shelf (CS), Makarov Basin (MB), Chukchi Plateau (CP) and Canada Basin (CB).

DownLoad: Full-Size Img PowerPoint

In the MB, CP and CB basins, the δ¹⁵N_PN was positively correlated with δ¹³C_POC, and the fitting equation was: δ¹⁵N=0.72δ¹³C+26.62, R²=0.36, p<0.000 1 (Fig. 7). Note that the δ¹⁵N_PN values at depths of 30 and 50 m at Station R14 are significantly lower and excluded from the fitting, although the reason for the lower δ¹⁵N was not clear. Previous studies have shown that the isotopic fractionation of carbon isotopes increases with the increasing CO₂ concentration and is negatively correlated with the growth rate and cell size of marine phytoplankton (Burkhardt et al., 1999). In the Arctic basins, low temperature results in high concentration of dissolved CO₂ and low growth rate of phytoplankton, which produces low δ¹³C_POC in the basins (Zhang et al., 2012). Unlike the case of carbon isotopes, nitrogen isotopic fractionation varied with phytoplankton species, and showed non-significant relationship with cell size or growth rate among most species of phytoplankton (Needoba et al., 2003), but a negative relationship with growth rate in the diatom Thalassiosira weissflogii (Montoya and McCarthy, 1995). Considering that diatom is an abundant species in the basins (Booth and Horner, 1997; Gosselin et al., 1997; Ardyna et al., 2017), it is reasonable to speculate that the positive correlation between δ¹³C_POC and δ¹⁵N_PN may be due to the slow growth rate of phytoplankton in the basins. Previous studies suggested that the growth rate of phytoplankton is controlled by light availability regulated by sea ice conditions (Yun et al., 2012). The synchronous decrease in δ¹³C_POC and δ¹⁵N_PN from the slope to the Chukchi plateau and then to the Canada Basin further supports this hypothesis (Figs 3h and i).

Figure  7.  Relationship between δ¹⁵N and δ¹³C of POM in the shelves and the basins. The samples collected from the Makarov Basin (MB), Chukchi Plateau (CP) and Canada Basin (CB) are marked in circles while the samples at depths of 30–50 m at station R14 are marked in black squares. The samples collected from the Siberian East Shelf (SES) and Chukchi Shelf (CS) are marked in grey triangles while the bottom samples from the SES and the CS are marked in black triangles. The dashed and the solid line represent the fitted lines of the basin and the shelf samples respectively.

DownLoad: Full-Size Img PowerPoint

Contrary to the situation in the basins, the δ¹⁵N_PN was inversely correlated with the δ¹³C_POC in the SES and the CS (δ¹⁵N=−0.28δ¹³C+1.73, R²=0.48, p<0.000 1, Fig. 7). Note that the bottom samples were not included in the regression as they were affected by sediment resuspension and remineralization (Figs 5e and f). Considering the high productivity in the Arctic shelves (Arrigo et al., 2014) and the low dissolved CO₂ concentration in the SES and CS (Anderson et al., 2011; Zhang et al., 2012), the low δ¹³C values in the SES and the CS could not be completely attributed to the CO₂ concentration and growth rate of phytoplankton. The possible reason was the contribution of ¹³C-depleted phytoplankton. First, a significantly lower C/N ratio was observed in the SES (Fig. 5f), indicating the presence of small phytoplankton (Taguchi, 1976; Crawford et al., 2015). Second, carbon isotope fractionation decreases as the cell size of phytoplankton increases, and small phytoplankton will produce ¹³C-depleted POM (Burkhardt et al., 1999). Therefore, the productivity of the low-cellular C/N ratio and ¹³C-depleted phytoplankton may explain the negative correlation between δ¹³C_POC and δ¹⁵N_PN in the SES and the CS.

In fact, the low C/N ratios and δ¹³C values observed in the SES were in consistent with previously reported in the Arctic (Crawford et al., 2015; Kumar et al., 2016), which was attributed to the contribution of small phytoplankton with low cellular C/N ratio and low δ¹³C. However, inconsistent with the low δ¹³C values, the POM in the SES exhibited an abnormally high δ¹⁵N_PN value (Fig. 4i). Therefore, it was necessary to clarify the mechanism by which the high δ¹⁵N_PN was produced in the SES. Here, we proposed that it is produced by nitrogen isotope fractionation during phytoplankton assimilation. The spatial distribution of the SIM inventory showed the negative values in the SES, but the positive values in its surrounding areas, which indicates the net loss of sea ice from the SES (Fig. 8). The net loss of sea ice is consistent with the fact that a polynya exists around the Wrangel Island in early summer (Stringer and Groves, 1991; Pickart et al., 2010; Moore and Pickart, 2012). In the spring and early summer, due to the disappearance of low irradiance and thick ices, the polynyas usually produce phytoplankton blooms earlier than the surrounding waters (Hsiao, 1992; Arrigo, 2007; Ambrose and Renaud, 1995; Arrigo and van Dijken, 2004; Tremblay et al., 2002). The biomass and production in the early bloom were dominated by small phytoplankton (<5 μm), and changed to large phytoplankton in late summer (Pesant et al., 1996). The Wrangel Island polynya is located in the mixed zone of the Pacific inflow and the Siberian Shelf Water, which provides nutrients for the early bloom of phytoplankton. For comparison, during the early bloom in the North Water polynya, nitrate concentration was as high as 10 μmol/L in the initial stage and decreased progressively in less than one month, which led to the exhausting of nitrate in the upper water column (>25 m) and increasing of δ¹⁵N_PN (Tremblay et al., 2006). Therefore, the early bloom of small phytoplankton was the most likely reason for the abnormally high δ¹⁵N_PN in the SES.

Figure  8.  Distribution of the inventory of sea ice meltwater in the Chukchi and the Siberian shelves.

DownLoad: Full-Size Img PowerPoint

Denitrification and sediment resuspension are two other possible factors for the high δ¹⁵N_PN in the SES. Denitrification was reported to be widespread in sediments and waters of the Chukchi and the Siberian shelves (Devol et al., 1997; Nitishinsky et al., 2007), and its isotope fractionation is as high as 20‰ (Granger et al., 2008). However, nitrogen isotopic composition of nitrate in the same voyage was similar in different regions (data not shown), which is difficult to explain the high δ¹⁵N observed only in the SES. For the effects of sediment resuspension, the δ¹⁵N_PN in surface sediments of the Siberian shelf ranged from 4.3‰ to 9.8‰ with an average of 8.0‰ (data not shown), which was significantly lower than those observed in the suspended particles of the SES (Fig. 5h). Obviously, the abnormally high δ¹⁵N_PN in the SES can’t be attributed to sediment resuspension. As for the effect of remineralization of particulate organic matter, it will simultaneously increase the δ¹⁵N_PN and δ¹³C_POC (Landrum et al., 2011). However, the negative correlation between the δ¹⁵N_PN and δ¹³C_POC in the SES ruled out this possibility (Fig. 7). In summary, the abnormally high δ¹⁵N_PN in the SES may be induced by the early bloom of small phytoplankton in the polynya.

4.2 Contribution of terrestrial, bloom-produced and non-bloom POMs

As stated above, the early bloom of small phytoplankton produced a distinctive POM characterized by low C/N ratio and low δ¹³C value, but high δ¹⁵N value. Here, the POM in the Chukchi and the Siberian shelves was thought to be composed of three different sources, namely terrestrial, bloom-produced and non-bloom biogenic. Based on the mass balance of C/N ratio and δ¹³C_POC, contribution fraction of these three sources was calculated as follows:

where f, R and δ¹³C refer to the source fraction, C/N ratio and δ¹³C value respectively. Subscript t, m, bm and obs represent the terrestrial POM, non-bloom biogenic POM, bloom-produced POM and the observed value respectively. The characteristic C/N ratio and δ¹³C value of each endmember were summarized in Table 2 and denoted in Fig. 9. The reasons for adopting these endmember values were described in the notes of Table 2.

Table  2.  The characteristic δ¹³C and C/N ratio of three endmembers used to calculate the contribution of terrestrial, bloom-produced and non-bloom biogenic POM in the Chukchi and the Siberian shelves

Endmember	δ13C/‰	C/N ratio	Reference
Terrestrial POM1)	−27.5±0.5	13±1	McClelland et al., 2008; Vonk et al., 2010; Lobbes et al., 2000; Guo and Macdonald, 2006; Guo et al., 2012
Non-bloom biogenic POM2)	−17±1	5.5±0.5	Descolas-Gros and Fontugne, 1985; Semiletov et al., 2005; Vonk et al., 2010; Copin-Montegut and Copin-Montegut, 1983; Martiny et al., 2013b
Bloom-produced POM3)	−30±1	3±1	Pineault et al., 2013; Kuliński et al., 2014; Kumar et al., 2016; Crawford et al., 2015
Notes: 1) Terrestrial POM: The reported δ13C value of terrestrial POM ranged from −22‰ to −30‰ with an average of −27‰ (Ehleringer et al., 1997) and its C/N ratio was higher than 12 (Lobbes et al., 2000; Guo et al., 2012; McClelland et al., 2016). The riverine POM dominated the source of terrestrial POM in the CS and the SES, while coastal erosion contributed nearly 60% to terrestrial POM in the SWS (Vonk et al., 2010). The weighted average of erosion-derived δ13CPOC was determined to be −27.5‰ in the SWS (McClelland et al., 2008; Vonk et al., 2010). Similar to the erosion-derived POM, the δ13CPOC value of riverine POM from the Russian rivers and the Yukon river was −25.6‰ to –27.4‰ (Lobbes et al., 2000) and about −26.2‰ respectively (Guo and Macdonald, 2006). In this study, the δ13C endmember value of terrestrial POM was assessed to be −27.5‰ ±0.5‰. As for the C/N ratio of terrestial POM, the reported value varied greatly, with around 12 in the Russian rivers (Lobbes et al., 2000) and from 14 to 40 in the Yukon River (Guo and Macdonald, 2006; Guo et al., 2012). Considering that the Russian rivers play a dominate role in our study areas, a value of 13±1 was adopted as the endmember C/N ratio. 2) Non-bloom biogenic POM: The δ13C value of biogenic POM ranged from −10‰ to −31‰ with an average of −20‰, and most of which was constrained between −17‰ to −22‰ (Descolas-Gros and Fontugne, 1985; Semiletov et al., 2005; Vonk et al., 2010). In the Arctic Ocean, the δ13C values of POM exhibit a wide range and vary from −16.7‰ to −30.4‰ (Goericke and Fry, 1994). In the Chukchi shelf, dissolved CO2 concentration is less than 20 μmol/L (Zhang et al., 2012), which corresponds to an isotope fractionation factor (εp) of 17‰ predicted by the Henry’s Law (Goericke and Fry, 1994). The highest value of δ13CPOC in the Chukchi and the Siberian shelves is −17.5‰ in this study. Obviously, the measured, the reported and the calculated δ13C values (by εp) in Arctic shelves are consistent, which supports the use of −17‰±1‰ as the δ13C endmember of non-bloom biogenic POM. The average C/N ratio of marine organic matter was 5.6 (Copin-Montegut and Copin-Montegut, 1983), and that in high-latitude cold waters was 6.0 (Martiny et al., 2013b). Here, the endmember C/N ratio of non-bloom biogenic POM was assessed to be 5.5±0.5. 3) Bloom-produced POM: The reported δ13C value of POM during phytoplankton bloom was as low as −24.5‰ to −31‰, which is significantly lower than those predicted by εp (Pineault et al., 2013; Kuliński et al., 2014; Kumar et al., 2016). The low δ13C value may be related to the high εp for small phytoplankton. For example, the εp of P. tricornutum is higher than 25‰ (Burkhardt et al., 1999). The lowest value of δ13CPOC measured in the Chukchi Sea and the Siberian Sea in this study was −29.5‰. Considering the contribution of 13C-depleted POM from the early bloom and the low δ13CPOC values measured, a value of −30‰±1‰ was adopted as the δ13CPOC endmember of bloom-produced POM. The C/N ratio of biogenic POM was as low as 2 to 4 in the Chukchi Sea and the Canada Basin (Crawford et al., 2015). The C/N ratio in the SES ranged from 2 to 5 in this study (Fig. 5f). The cellular C/N ratio of small diatoms (volume<300 μm3) was reported to be in a range of 2.8 to 3.8 (Taguchi, 1976). Therefore, a value of 3±1 was adopted as the C/N ratio endmember.

 | Show Table

DownLoad: CSV

Figure  9.  The scatter plot of C/N ratio and δ¹³C_POC in the Chukchi and the Siberian shelves. The endmember values of terrestrial, non-bloom biogenic and bloom-produces POM are marked in black diamonds.

DownLoad: Full-Size Img PowerPoint

After clarifying the C/N ratio and δ¹³C endmembers, most of the data we measured in the Chukchi and the Siberian shelves were located in the mixed area of these three sources (Fig. 9). Although a few data are below the mixing line of the bloom-produced and the non-bloom biogenic POM, they were still within 95% of the prediction interval (Fig. 9). This deviation reflects the impact of endmember fluctuations. The values of f_t, f_m and f_bm were calculated using Monte Carlo analysis with an average deviation of 7% for each source.

The averaged f_t in the water column varied from <10% to 60%, and presented an eastward decline in the Chukchi and the Siberian shelves (Fig. 10a). The higher f_t value (>30%) occurred in the SWS and decreased slightly to the north, indicating a strong source of terrestrial POM to the Siberian shelf. The f_t value was the lowest (<20%) in the SES and the CS, corresponding to the high biological production in these areas (Anderson et al., 2011). The f_t value in the northern CS (20%<f_t<30%) was higher than that in the southern shelf, probably due to the terrestrial input from the Mackenzie River on the east side. Note that a significant positive correlation between the f_t and the f_MW is observed (p<0.000 1), further supporting the reliability of the calculated f_t (Fig. 11).

Figure  10.  Distribution of depth-averaged f_t (a), f_m (b) and f_bm (c) in the Chukchi and the Siberian shelves.

DownLoad: Full-Size Img PowerPoint

Figure  11.  Relationship between the depth-averaged f_t and the depth-averaged f_MW in the Chukchi and the Siberian shelves.

DownLoad: Full-Size Img PowerPoint

The averaged f_m in the water column varied from 10% to 70%, with values being >30% and ≤20% in the Chukchi shelf and the Siberian shelf respectively (Fig. 10b). The highest f_m (>60%) appeared near the Bering Strait, confirming that the Pacific inflow stimulated biological production in the Chukchi shelf (Zhuang et al., 2016).

The averaged f_bm ranged from 15% to 90% and exhibited a radial decrease around the SES (Fig. 10c). The particularly high f_bm value (>60%) in the SES indicated that the POM was dominated by bloom-produced organic matter, which was consistent with the early bloom of phytoplankton in the polynya. The f_bm value was higher than the f_m value in most of the Chukchi and the Siberian shelves, which emphasizes the role of small phytoplankton in the ecosystem. With the exception of the SWS, the spatial variation of f_bm in other regions was mirrored to that of sea ice meltwater inventory with the highest f_bm and the lowest I_SIM in the polynya region (Figs 8 and 10c). The inventory of bloom-produced POC (POC_bm) in the water column, calculated by depth integration (excluding the bottom samples affected by sediment resuspension as shown in Figs 5e and f), was inversely correlated to the inventory of sea ice meltwater in the CS, SES and SMS (Fig. 12). This relationship suggested that the early bloom of phytoplankton benefits from the loss of sea ice. The loss of sea ice will increase irradiance in the waters, thus promotes the early bloom of phytoplankton shortly after the end of polar night (Hsiao, 1992; Tremblay et al., 2002). Since utilization efficiency of light by small phytoplankton (small chloroplast in cell) is higher than that of large phytoplankton (large chloroplast in cell) (Taguchi, 1976; Pesant et al., 1996), the early bloom was dominated by small phytoplankton, which leads to the low C/N ratio and high δ¹⁵N_PN, and thus the high f_bm in our study areas. The reasons that the POM signals generated by the early bloom of small phytoplankton persist until summer may have two aspects, one is the high buoyancy of diatoms (Pesant et al., 1996), which are dominate species in the Arctic polynyas (Pesant et al., 1996; Lewis et al., 1996; Oxtoby et al., 2016), and the other is the weak hydrodynamic conditions. The SES is a typical region where water exchange is limited by the Pacific inflow in the east and the Siberian Coastal Current in the west (Fig. 1). The data from the SWS fell below the fitted line of POC_bm and I_SIM in Fig. 12, probably due to the nutrient deficiencies that limit primary production.

Figure  12.  Relationship between POC_bm inventory and I_SIM in the Chukchi and Siberian shelves.

DownLoad: Full-Size Img PowerPoint

5. Conclusions

The POM in the Chukchi Sea and the East Siberian Sea was generally characterized by low C/N ratio, low δ¹³C and high δ¹⁵N in summer. A positive relationship between δ¹³C and δ¹⁵N in the basins and a negative one in the Chukchi and Siberian shelves were observed, which indicated that the POC and PN are different in sources, turnover rates, and cycling pathways. The low δ¹³C values and a positive correlation between δ¹³C and δ¹⁵N in the basins are due to high dissolved concentration of CO₂ and slow growth rate of phytoplankton, while the low C/N ratios, the low δ¹³C values, the high δ¹⁵N values and a negative correlation between δ¹³C and δ¹⁵N in the shelves are attributed to the early bloom of small phytoplankton in the polynya. The contributions of terrestrial, bloom-produced and non-bloom biogenic components in bulk POM were estimated by the mass balance of C/N ratio and δ¹³C value. The terrestrial POM in the SWS accounted for more than 30% of the bulk POM, and was reduced to less than 20% in the CS and the SES, indicating that the POM in the SWS is greatly affected by terrestrial source, while that in the CS and the SES is dominated by biological source. The non-bloom biogenic POM occupied more than 60% near the Bering Strait and less than 20% in the Siberian Sea. The bloom-produced POM was the most abundant component in bulk POM, with the highest fraction (>70%) in the Wrangel polynya and about 40% in most of the Chukchi and the Siberian shelves. A negative relationship was observed between the inventory of bloom-produced POM and the inventory of sea ice meltwater in the CS, SES and the SMS, indicating the early bloom of small phytoplankton benefits from the loss of sea ice in the polynya. Our results demonstrate that phytoplankton blooms in spring or early summer may have residual effects on summer ecosystem. It is necessary to have a deeper understanding of seasonal variations in the Arctic Ocean in future.

Acknowledgements

We thank Xiaoyu Wang for generously providing temperature, salinity and Chl a data in the LV-77 Cruise.