Journal of Geophysical Research: Oceans

Ocean acidification and biologically induced seasonality of carbonate mineral saturation states in the western Arctic Ocean

Authors


Abstract

[1] Calcium carbonate (CaCO3) mineral saturation states for aragonite (Ωaragonite) and calcite (Ωcalcite) are calculated for waters of the Chukchi Sea shelf and Canada Basin of the western Arctic Ocean during the Shelf-Basin Interactions project from 2002 to 2004. On the Chukchi Sea shelf, a strong seasonality and vertical differentiation of aragonite and calcite saturation states was observed. During the summertime sea ice retreat period, high rates of phytoplankton primary production and net community production act to increase the Ωaragonite and Ωcalcite of surface waters, while subsurface waters become undersaturated with respect to aragonite due primarily to remineralization of organic matter to CO2. This seasonal “phytoplankton-carbonate saturation state” interaction induces strong undersaturation of aragonite (Ωaragonite = <0.7–1) at ∼40–150 m depth in the northern Chukchi Sea and in the Canada Basin within upper halocline waters at ∼100–200 m depth. Patches of aragonite undersaturated surface water were also found in the Canada Basin resulting from significant sea ice melt contributions (>10%). The seasonal aragonite undersaturation of waters observed on the Chukchi Sea shelf is likely a recent phenomenon that results from the uptake of anthropogenic CO2 and subsequent ocean acidification, with seasonality of saturation states superimposed by biological processes. These undersaturated waters are potentially highly corrosive to calcifying benthic fauna (e.g., bivalves and echinoderms) found on the shelf, with implications for the food sources of large benthic feeding mammals (e.g., walrus, gray whales, and bearded seals). The benthic ecosystem of the Chukchi Sea (and other Arctic Ocean shelves) is thus potentially vulnerable to future ocean acidification and suppression of CaCO3 saturation states.

1. Introduction

[2] Over the last several centuries, anthropogenic activities have released large quantities of carbon dioxide (CO2) into the atmosphere [Intergovernmental Panel on Climate Change (IPCC), 2007]. Anthropogenic CO2, emitted primarily from fossil fuel use, cement manufacture, and land use changes has not only accumulated in the atmosphere, but it has also been taken up by the terrestrial biosphere and oceans [Sabine et al., 2004]. As a consequence of the uptake of anthropogenic CO2, dissolved inorganic carbon (DIC) and partial pressures of CO2 (pCO2) have increased while the pH has decreased in the upper ocean over the last few decades [e.g., Bates, 2007; Bates and Peters, 2007; Takahashi et al., 2009]. This gradual process of ocean acidification has long been recognized [e.g., Broecker and Takahashi, 1966; Broecker et al., 1971; Bacastow and Keeling, 1973]. Estimates based on the IPCC projections of future CO2 emission and subsequent ocean absorption of anthropogenic CO2 [Bindoff et al., 2007] predict that upper ocean pH will decrease by 0.3–0.5 units over the next century and beyond [Caldeira and Wickett, 2003, 2005].

[3] Ocean acidification also acts to decrease seawater carbonate ion concentration [CO32−] and the saturation state (Ω) of calcium carbonate (CaCO3) minerals such as calcite and aragonite. Recent estimates suggest that global upper ocean [CO32−] concentrations will decrease by 50% by the end of the century [Orr et al., 2005, 2006]. In many oceanic basins, the oceanic uptake of anthropogenic CO2 has resulted in the shoaling of the aragonite saturation horizon (i.e., Ωaragonite = 1) by 40 to 200 m [Feely et al., 2004] over the past century and recent observations show potentially corrosive waters impinging onto the continental shelf of western North America [Feely et al., 2008]. At Ω values of <1, waters become potentially corrosive to unprotected shells or skeletons of calcium carbonate, as well as carbonate sediments [Andersson et al., 2003, 2005, 2006]. The decrease in Ω has profound negative implications for many (but not all) marine organisms in both high- and low-latitude oceans that produce shells, tests or skeletons made of CaCO3 minerals, such as corals, coralline algae, pteropods, coccolithophorids, mollusks, echinoderms, and foraminifera [e.g., Buddemeier et al., 2004; Royal Society, 2005; Kuffner et al., 2008; Fabry et al., 2008; Doney et al., 2009].

[4] The effects of ocean acidification are potentially far-reaching throughout the global ocean, with models indicating that Arctic Ocean will be impacted before other regions [Orr et al., 2005, 2006; Steinacher et al., 2009]. The future decrease in Ω will have implications for CaCO3 producing marine organisms such as cold-water corals and bivalves, and high Mg calcite producing echinoderms in the Arctic. Jutterström and Anderson [2005] recently observed that the saturation horizon (i.e., Ω = 1) for calcite (Ωcalcite) and aragonite (Ωaragonite) was 4000 m and 2500 m deep, respectively, in the central basin of the Arctic Ocean. However, undersaturation of aragonite was also observed in the subsurface halocline of the central basin. Previous studies have found good preservation of aragonitic pteropod tests in the sediments of the Arctic Ocean [Belyaeva and Khusid, 1990; Bergsten, 1994; Stein et al., 1994], while evidence of CaCO3 dissolution in seasonally ice-free areas of the Laptev Sea has also been observed [Wollenburg and Kuhnt, 2000].

[5] In this study, we report CaCO3 saturation states for calcite (Ωcalcite) and aragonite (Ωaragonite) for surface and subsurface waters of the Western Arctic Ocean including the Chukchi and Beaufort Sea shelves and the adjacent deep Canada Basin. In the shallow coastal waters of the Chukchi and Beaufort seas, the inflow of nutrient-rich Pacific Ocean water coupled with the seasonal retreat of sea ice and surface warming, sustains high rates of phytoplankton primary production (PP) which in turn supports rich and diverse open water and benthic ecosystems [Feder et al., 2005, 2007; Grebmeier et al., 2008]. Our seasonal field observations, collected during the Shelf-Basin Interactions (SBI) program in 2002 and 2004 [Bates, 2006] show distinct seasonal changes of Ωcalcite and Ωaragonite, with significant undersaturation for aragonite minerals (Ωaragonite = <1) developing in the halocline on the northern Chukchi Sea shelf, and in surface and upper halocline layer waters of the Canada Basin. Similar seasonal CaCO3 saturation state changes have been observed in the open ocean of the North Pacific Ocean [Feely et al., 1988]. In the Chukchi Sea, the high rates of phytoplankton PP act to seasonally increase CaCO3 saturation states in the surface layer while the vertical export of organic carbon and subsequent remineralization decrease saturation states in subsurface waters, which amplifies existing ocean acidification due to the global uptake of anthropogenic CO2. We describe this seasonal “phytoplankton-carbonate saturation state” (PhyCASS) interaction as a case where ocean biology drives divergent trajectories for carbonate chemistry in surface and subsurface waters over seasonal timescales on this Arctic shelf but as well as potentially elsewhere in the global ocean. We also briefly discuss the biological implications of these observations for calcifying marine organisms such as bivalves and echinoderms that are significant components of the benthos of the Chukchi Sea shelf.

2. Arctic Ocean Water Masses and Circulation

[6] There is a general northward flow of ∼0.8–1.0 Sv of Pacific Ocean water through Bering Strait [Coachman and Barnes, 1961; Roach et al., 1995; Woodgate and Aagaard, 2005; Woodgate et al., 2005a, 2005b] across the shallow Chukchi Sea shelf and outflow from the shelf through Herald Valley and Barrow Canyon [Aagaard, 1989; Pickart et al., 2005; Winsor and Chapman, 2004]. A narrow eastward boundary current is present along the shelf break of the Chukchi and Beaufort seas [Carmack and Macdonald, 2002; Nikolopoulos et al., 2009], while wind-driven, clockwise rotation of the Beaufort Gyre controls regional scale sea ice and upper ocean circulation in the Canada Basin. Shelf-basin exchanges are facilitated by wind-driven air-sea interaction [Carmack and Chapman, 2003], generation of mesoscale eddies [Manley and Hunkins, 1985; Muench et al., 2000; Spall et al., 2008; Mathis et al., 2007a], and dense water formation through winter brine rejection [Weingartner et al., 1998]. The upper water masses of the western Arctic Ocean are formed from a mixture of Pacific Ocean, sea ice melt and runoff contributions [Macdonald et al., 2002; Kadko and Swart, 2004; Cooper et al., 2008] that are highly modified on the adjacent continental shelves of the Arctic during sea ice retreat and advance [Aagaard and Carmack, 1989; Aagaard et al., 1981; Bates, 2006].

3. Methods and Data Sources

3.1. Cruise Information and Sampling

[7] Seasonal physical and biogeochemical observations on the Chukchi and Beaufort Sea shelves and adjacent Canada Basin (Figure 1) were collected in 2002 and 2004 from the icebreaker U.S. Coast Guard Cutter Healy, as part of the Western Arctic SBI project (Figure 2) [Grebmeier and Harvey, 2005]. In each year of the field campaign wintertime conditions (∼95–100% sea ice cover) were sampled during early season cruises (May–June; cruises HLY-02-01 and HLY-04-02), while summertime conditions were sampled later during the period of maximum sea ice loss (July–August; cruises HLY-02-02 and HLY-04-03).

Figure 1.

Generalized circulation pattern of surface waters in the Chukchi Sea, Beaufort Sea, and Canada Basin region. Water transiting through Bering Strait is composed of warmer, fresher Alaskan Coastal Current (ACC) waters in the east [Woodgate et al., 2005a]; Bering Shelf Water (BSW) in central Bering Strait; and colder, saltier, more nutrient-rich Anadyr Water (AW) to the west. Bering Shelf and Anadyr Water merge on the Chukchi Sea [Roach et al., 1995; Woodgate et al., 2005a]. The East Siberian Coastal Current (ESC) flows intermittently into the Chukchi Sea from the East Siberian Sea through Long Strait [Weingartner et al., 1999].

Figure 2.

Location map of cruises in the Chukchi Sea and Canada Basin of the Arctic Ocean. (a) Winter/spring 2002 CTD/hydrocast stations (HLY-02-01 cruise; 5 May to 15 June 2002), (b) winter/spring 2004 CTD/rosette stations (HLY-04-02 cruise; 17 May to 21 June 2004), (c) summer 2002 CTD/rosette stations (HLY-02-03 cruise; 17 July to 26 August 2002), and (d) summer 2004 CTD/rosette stations (HLY-04-03 cruise; 16 July to 26 August 2002). Insets are of CTD/hydrocast stations at Bering Strait.

[8] At each conductivity-temperature-depth probe (CTD)/hydrocast station, the following hydrographic and biogeochemical sampling was conducted, including: inorganic nutrients and dissolved oxygen [Codispoti et al., 2005], suspended particulate organic matter [Bates et al., 2005a], dissolved organic matter [Mathis et al., 2005, 2007b], DIC and total alkalinity (TA) [Bates et al., 2005b], δ18O values [Cooper et al., 2005], and biological data (e.g., chlorophyll a [Hill and Cota, 2005]). Rate measurements such as in situ 14C primary production [Hill and Cota, 2005] and 234Th/238U export production [Moran et al., 2005; Lepore et al., 2007] were also determined. CTD, bottle and rate measurement data are available at National Center for Atmospheric Research (NCAR)/Earth Observing Laboratory (EOL) website for SBI (http://www.eol.ucar.edu/projects/sbi/) and also archived at the National Snow and Ice Data Center (NSIDC; http://nsidc.org/).

[9] Shelf-basin hydrographic and biogeochemical sections were sampled between the Chukchi/Beaufort Sea shelves and adjacent deep Canada Basin of the Arctic Ocean during the early season and summertime cruises. These sections included the following: (1) West Hanna Shoal (WHS); (2) East Hanna Shoal (EHS); (3) Barrow Canyon (BC), and; (4) East of Pt. Barrow (EB) (Figure 2). A total of fourteen hydrographic and biogeochemical shelf-basin sections were sampled in 2002 and 2004. In addition, three sections across Bering Strait were also undertaken in 2002 and 2004.

3.2. DIC and TA Definitions and Sample Analyses

[10] Seawater carbonate chemistry is governed by a series of chemical uptake, dissolution and precipitation reactions, such that

equation image

The four directly measurable carbonate system parameters are DIC, TA, pCO2, and pH. DIC is defined as [Dickson and Goyet, 1994; Dickson et al., 2007]

equation image

where [CO2*] represents the total concentration of carbon dioxide species, whether present as H2CO3 or as CO2. The TA of seawater is defined as

equation image

where [HCO3] + 2[CO32−] + B(OH)4 are the major components of seawater TA, while other species are minor components having negligible impact on TA. DIC and TA are expressed as μmoles kg−1. pCO2 is the partial pressure of CO2 in equilibrium with seawater expressed as μatm, while pH is the negative log of the H+ activity on the total seawater pH scale.

[11] Seawater samples for DIC and TA were drawn from the Niskin samplers into precleaned ∼300 ml Pyrex bottles, poisoned with HgCl2 to halt biological activity, sealed, and returned to the Bermuda Institute of Ocean Sciences [BIOS] for analysis. DIC samples were analyzed using a highly precise (∼0.025%; <0.5 μmoles kg−1) gas extraction and coulometric detection system [Bates et al., 1996]. TA samples were analyzed using an automated potentiometric method with a precision of <∼1 μmoles kg−1 [Bates, 2006]. Both DIC and TA data were compared against certified reference materials to ensure that the inaccuracy of the measurements were less than ∼0.1% (∼2 μmoles kg−1). The total number of DIC and TA samples collected for use in this study was 1665, including waters from the Bering Strait, across the Chukchi Sea and Beaufort Sea shelves and from the deep waters of the Canada Basin.

3.3. Carbonic Acid System Calculations and Impacts of Biological Processes

[12] The complete seawater carbonic acid system (i.e., pCO2, [HCO3], [CO32−] and [H+], CaCO3 saturation states of aragonite, Ωaragonite, and calcite, Ωcalcite) were calculated from DIC, TA, temperature and salinity data using the equations of Zeebe and Wolf-Gladrow [2001]. The carbonic acid dissociation constants of Mehrbach et al. [1973] (as refit by Dickson and Millero [1987], i.e., pK1 and pK2) were used to determine seawater pCO2 and other carbonate parameters. The equations governing the calculation of pK1 and pK2 have lower temperature limits of −1°C or 0°C, but here, these equations were extended to temperatures lower than −1°C. The CO2 solubility equations of Weiss [1974], and dissociation constants for borate [Dickson, 1990], silicate and phosphate [Dickson and Goyet, 1994] were used as part of these calculations.

[13] CaCO3 saturation states for both aragonite (Ωaragonite) and calcite (Ωcalcite) were calculated as the ion product of carbonate and calcium concentrations (e.g., aragonite saturation state, Ωaragonite = [Ca2+] [CO32−]/Ksp*, aragonite), using the solubility products, Ksp*, experimentally determined by Mucci [1983] as functions of temperature, salinity and pressure. We calculate Ω values for low Mg calcite (LMC) rather than high Mg calcite (HMC). HMC is considered a more metastable CaCO3 mineral relative to aragonite, and consequently HMC Ω values are lower than Ωaragonite under equivalent DIC, TA, temperature (T), salinity (S) and pressure. The difference in Ω between aragonite and HMC ranges from ∼<0.2 to >1 and depends on the solubility constants used [e.g., Plummer and Mackenzie, 1974; Bischoff et al., 1987, 1993] and mol % Mg content of HMC [e.g., Andersson et al., 2008].

[14] The uncertainty of seawater pCO2 calculations has been shown previously to be <10 μatm [Bates, 2006], which is small compared to the large natural variability of seawater pCO2 observed in the region (∼50–500 μatm) [Bates et al., 2006]. Uncertainty in the estimation of [HCO3], [CO32−], Ωaragonite and Ωcalcite was ∼5 μmoles kg−1, ∼5 μmoles kg−1, 0.02 and 0.02, respectively. In our calculations, when required, DIC and TA data were normalized to a constant salinity of 33.1, the core salinity of upper halocline layer (UHL) water. We assumed that normalizing DIC and TA data to a constant salinity of 33.1 does not introduce potential biases [Friis et al., 2003; Friis, 2006].

[15] The effect of photosynthesis and respiration on the carbonate system is governed by the following general chemical reaction:

equation image

Photosynthesis and respiration (or remineralization of organic carbon) acts to decrease/increase DIC and pCO2, respectively, while TA remains essentially constant (except for minor effects due to nitrate uptake/release [Brewer and Goldman, 1976]). Thus, phytoplankton photosynthesis acts to increase pH, Ωaragonite and Ωcalcite, while respiration/remineralization acts to decrease pH, Ωaragonite and Ωcalcite. Calcium carbonate (CaCO3) mineral production and dissolution processes are governed by the following general chemical reaction:

equation image

Typically, CaCO3 production and dissolution rates vary as a function of saturation state for aragonite (Ωaragonite) or calcite (Ωcalcite). For example, dissolution of aragonite is generally favored when Ωaragonite values are <1, while aragonite formation generally occurs at Ωaragonite values of >1. CaCO3 production/dissolution act to decrease/increase DIC and TA in the ratio of 1:2 generally, thereby increasing/decreasing pCO2, Ωaragonite and Ωcalcite.

3.4. Water Mass Identification and Estimation of the Seawater, Freshwater, and Sea Ice Melt Contributions to the Western Arctic

[16] Wintertime shelf waters of the Chukchi and Beaufort seas have a narrow range of temperature, salinity, and density properties [Woodgate et al., 2005b]. Subsequently, during summertime sea ice retreat, shelf surface water T, S and density properties develops large variability and the mixed layer typically extends from 10 to 50 m deep, overlying remnant winter water of Pacific Ocean origin [Codispoti et al., 2005]. In the Canada Basin, a strong vertical density gradient separates nutrient poor surface waters or polar mixed layer (PML) from underlying nutrient rich halocline waters that can be subdivided into UHL and a lower halocline layer (LHL). The UHL, predominantly of Pacific Ocean origin, is typically found at ∼80–120 m deep, with a core salinity of 33.1–33.2, nitrate and phosphate concentrations of ∼14 ± 2 μmoles kg−1 and ∼1.8 ± 0.2 μmoles kg−1, respectively. Waters of the LHL originate from Atlantic waters that have been modified during transit over the continental shelves of the eastern Arctic Ocean. The core layer of the LHL is found at ∼120–230 m deep [Kinney et al., 1970; Jones and Anderson, 1986; Jones et al., 2003] with a characteristic salinity range of 34.2–34.6 and nitrate and phosphate concentrations of ∼12 ± 1 μmoles kg−1 and ∼0.8 ± 0.2 μmoles kg−1, respectively. At deeper depths of the Canada Basin, two additional Atlantic layers are observed, Atlantic Water Layer (AWL; ∼200–800 m deep) with a core salinity of ∼34.8–34.9, and below that Arctic Ocean Deep Water (AODW) that fills the depths of both Canada and Eurasian Basins.

[17] The upper waters of the Arctic Ocean are highly influenced by river runoff and sea ice melt contributions. For example, Macdonald et al. [2002] estimated that ∼8 m of the upper 40 m of the PML had freshwater origins from runoff at the SHEBA site in the Canada Basin, while Cooper et al. [2005] report that the river fraction in surface waters of the Canada Basin and Chukchi Sea can be up to 20%. In the upper 10 m of the PML, the sea ice melt fraction can also exceed 25% [Cooper et al., 2005] during the summertime retreat of sea ice. Rivers entering the Arctic Ocean have high levels of total alkalinity that contribute significantly to the TA of the PML [e.g., Ludwig et al., 1996; Anderson et al., 2004; Yamamoto-Kawai et al., 2005; Cooper et al., 2008].

[18] Oxygen isotope ratios (δ18O) and salinity distributions have been used as tracers of river runoff and sea ice melt in the Arctic Ocean [e.g., Macdonald et al., 1989, 1995, 2002; Cooper et al., 1997, 1999, 2005; Ekwurzel et al., 2001; Schlosser et al., 2002]. Since atmospheric water is depleted in the heavier oxygen isotope with a well-defined latitudinal dependence [Dansgaard, 1964], δ18O can be used as a tracer of meteoric input to seawater. The δ18O values of Arctic Ocean seawater thus reflect the contributions from seawater, runoff, and freshwater derived from sea ice [Östlund and Hut, 1984; Macdonald et al., 1989; Cooper et al., 2005]. Following the method of Cooper et al. [2005], the estimated river freshwater runoff and sea ice melt components of each water sample taken during the SBI surveys of the Chukchi Sea and Canada Basin in 2002 (n = 426) and 2004 (n = 174) was solved using three-component mixing equations of δ18O and salinity data (Figure 3). The core value of each of the components was as follows: (1) freshwater δ18O value, including local precipitation (δ18O = −20.3‰, S = 0; a slightly heavier value than the −21.0‰ of Östlund and Hut [1984]); (2) sea ice δ18O value (δ18O = −1.9‰, S = 4.5) [Eicken et al., 2002]; and (3) a seawater component, in this case from below the fall and remnant spring mixed layers representing the upper halocline of the southern Beaufort sea (δ18O = +0.3‰ and S = 33.5 [Kadko and Swart, 2004]). The freshwater δ18O signature of Arctic rivers spans a wide range, but we used flow-weighted estimates of Mackenzie River values for our analysis rather than other Arctic rivers due to the dominance of the Mackenzie River freshwater contributions to the upper waters of the Canada Basin [Cooper et al., 2005].

Figure 3.

Relationship between δ18O values and salinity for SBI water samples collected in the Canada Basin (and Chukchi and Beaufort seas) in 2002 (data from Cooper et al. [2005]). The estimated fraction of freshwater derived from river runoff (black dashed lines) and sea ice melt (solid gray lines) is shown. Further details about the δ18O sample analysis are given by Cooper et al. [2005].

4. Results and Discussion

[19] In this section, the seasonality of physics and biology (section 4.1), freshwater and sea ice melt end-members for DIC and TA (section 4.2), and CaCO3 mineral saturation states (section 4.3) are reported for the Chukchi Sea shelf and Canada Basin waters. Aragonite undersaturation in Canada Basin surface waters that are highly influenced by sea ice melt and runoff contributions, in the subsurface UHL and deep Canada Basin, as well as the impact on air-sea CO2 gas exchange are discussed in section 4.4. In section 4.5, we discuss the biologically induced seasonality of CaCO3 saturation states for aragonite in surface and subsurface waters of the Chukchi Sea shelf.

4.1. Physical and Biological Seasonality of the Chukchi/Beaufort Sea Shelf and Adjacent Canada Basin

[20] The Chukchi Sea/Beaufort Sea shelves and adjacent Canada Basin are highly influenced by seasonality in sea ice conditions (sea ice retreat and advance), warming and freshening of surface waters, and seasonal availability of light and nutrients. During the SBI project, early season cruises sampled wintertime conditions (∼95–100% sea ice cover) and later summertime conditions reflecting warming and sea ice retreat. In the mixed layer waters of the Chukchi Sea and Beaufort Sea shelves (∼0–30 m) and Canada Basin (∼0–50 m), salinity typically ranged from ∼25 to 33 with freshening in summer due to sea ice melt in 2002 (Figure 4) and 2004 (Figure 5). Warming of mixed-layer temperatures from spring (typically ∼−1.7°C to −1.5°C for Canada Basin PML; 95–100% sea ice covered) to summer (typically ∼−1.5 to 0°C for Canada Basin PML; ∼75–95% sea ice cover [Bates, 2006]) was observed in both years.

Figure 4.

CaCO3 mineral saturation states superimposed on temperature and salinity observed in 2002 in the Chukchi Sea and Canada Basin of the Arctic Ocean. (a) Ωaragonite in spring 2002, (b) Ωcalcite in spring 2002, (c) Ωaragonite in summer 2002, and (d) Ωcalcite in summer 2002. Water masses are differentiated into PML, UHL, LHL, and AODW (with the LHL and AODW only occurring in the Canada Basin).

Figure 5.

CaCO3 mineral saturation states superimposed on temperature and salinity observed in 2004 in the Chukchi Sea and Canada Basin of the Arctic Ocean. (a) Ωaragonite in spring 2004, (b) Ωcalcite in spring 2004, (c) Ωaragonite in summer 2004, and (d) Ωcalcite in summer 2004. Water masses are differentiated into PML, UHL, LHL, and AODW (with the LHL and AODW only occurring in the Canada Basin).

[21] In summertime, the loss of sea ice, warming and increase in light prompts a seasonal biological response in the water column and in the benthos. In surface waters of the Chukchi Sea shelf, high rates of phytoplankton PP (>1 g C m−2 d−1) and NCP were observed in 2002 and 2004 [Hill and Cota, 2005; Bates et al., 2005a; Mathis et al., 2009] during the summertime, open water sea ice free period. At the same time, there was considerable vertical export of organic matter from the mixed layer to the underlying halocline and benthos [Moran et al., 2005; Lepore et al., 2007] and horizontal export of suspended particulate organic matter (SPOM) at ∼50–150 m deep from the Chukchi Sea shelf into the UHL of the adjacent Canada Basin [Bates et al., 2005b; Bates, 2006]. On the shelf, the vertical export of organic matter supports a large benthic ecosystem of shelled fauna, echinoderms and crustacea [Feder et al., 2005; Grebmeier et al., 2006a], with a large oxygen demand generated in the sediments [Grebmeier and Harvey, 2005; Grebmeier et al., 2008].

[22] In contrast to the Chukchi Sea shelf, the adjacent Canada Basin experiences an attenuated seasonality in sea ice conditions, warming and freshening. In the PML of the deep Canada Basin, summertime sea ice loss was modest (∼75–95% sea ice cover), with little warming (<1°C) and freshening. Low rates of PP (<10 mg C m−2 d−1) and NCP were observed in both years and seasons [Hill and Cota, 2005; Bates et al., 2005a; Mathis et al., 2009].

4.2. Freshwater and Sea Ice Melt End-Members for DIC and TA

[23] The freshwater component due to runoff (which includes rain and snow falling on the ocean surface) varied between a few percent and ∼17% (Figure 3). In the PML layer, river runoff accounted for ∼7–17% of each seawater sample. The UHL, which like the PML has a Pacific Ocean origin, was also influenced by freshwater from runoff, with the runoff fraction estimated at ∼4–8%. In deeper water of the LHL, AWL and AODW (all with Atlantic Ocean origins), the freshwater component due to river runoff was close to zero. Even within the LHL, the runoff fraction was small (<1–3%) indicating low vertical diffusion mixing rates [Wallace et al., 1987] between the two layers of the halocline in the Arctic Ocean.

[24] Across the Chukchi Sea and Canada Basin, the freshwater component due to melted sea ice was only greater than 5% in a few surface samples (upper 5–10 m) of the PML and only during the summertime SBI cruises. In deeper waters of the PML, halocline layers (UHL and LHL), and Atlantic Layer the sea ice melt fraction was close to zero.

[25] The runoff and sea ice melt DIC and TA end-members components (in the Canada Basin primarily) were estimated from the δ18O − salinity and nDIC/nTA + NO3 relationships. For the four SBI cruises, the runoff end-member TA ranged from ∼830–970 μmoles kg−1, while DIC ranged from ∼320–430 μmoles kg−1, with higher end-member values estimated for the summertime. The runoff end-member TA values calculated here were comparable to the 1000 μmoles kg−1 reported for Arctic Ocean river sources [Anderson et al., 1988; Olsson and Anderson, 1997], but lower than the flow weighted 1618 μmoles kg−1 reported for the Mackenzie River [Cooper et al., 2008]. Cooper et al. [2008] also report that the flow-weighted average TA for the six major Arctic rivers (i.e., 'Ob, Yenisey, Lena, Kolyma, Yukon and Mackenzie Rivers) was 1048 μmoles kg−1, similar to the river freshwater end-member calculated here. The Mackenzie and Yukon Rivers typically have higher TA (freshwater) end-members than Eurasian Arctic rivers draining the Siberian tundra (e.g., Lena, Ob', Yenisey, Kolyma), reflecting a chemistry-influenced watershed with a higher proportion of carbonate rock lithologies and sediments [Telang et al., 1991; Amiotte Suchet et al., 2003]. We estimate that the sea ice melt TA end-member is ∼304 ± 120 μmoles kg−1 comparable to the 300 μmoles kg−1 reported by Anderson and Jones [1985].

4.3. CaCO3 Saturation States in the Chukchi Sea and Canada Basin

[26] CaCO3 mineral saturation states for aragonite and calcite in waters of the Chukchi Sea shelf and adjacent Canada Basin are shown in a composite plot of Ωaragonite and Ωcalcite values against depth (Figure 6). The differentiation in CaCO3 saturation states between water masses of the Arctic Ocean is also evident in Ω, temperature and salinity plots for Chukchi Sea shelf and adjacent Canada Basin waters in 2002 (Figure 4) and 2004 (Figure 5).

Figure 6.

CaCO3 mineral saturation states against depth observed in 2002 and 2004 in the Chukchi Sea and Canada Basin of the Arctic Ocean. (a) Ωaragonite against depth and (b) Ωcalcite against depth. The black vertical dashed lines represent the saturation depth where Ω = 1. The gray horizontal lines differentiate the approximate depths of the PML, UHL, LHL, AWL, and AODW in the deep Canada Basin. Here diamonds represent data for spring 2002, circles represent data for summer 2002, squares represent data for spring 2004, and triangles represent data for summer 2004.

[27] Across the Chukchi Sea and Canada Basin region, CaCO3 saturation states were highly variable ranging from ∼0.8–6 for Ωcalcite and 0.6–4.0 for Ωaragonite, respectively (Figures 46). With notable exceptions (discussed in section 4.4), the surface polar mixed layer waters across the Chukchi Sea shelf and adjacent Canada Basin were clearly saturated with respect to CaCO3 minerals in both years (2002 and 2004) and seasons of observation (winter/spring period of sea ice cover; and summer period of sea ice extent minima). Although the data are not shown here, as discussed in the methods, the Ω values for HMC were lower than for aragonite.

4.4. Ωaragonite Variability in Waters of the Canada Basin

[28] In the Canada Basin, Ωaragonite values of the PML (i.e., 0–50 m) were less variable compared to the Chukchi Sea shelf. In general, most surface and lower halocline layer water samples had relatively saturated Ωaragonite values of ∼1.4 to 2.1 (Figure 7). However, undersaturation of aragonite (and calcite) was observed in some surface samples, in the UHL and in deep waters of the Canada Basin.

Figure 7.

Aragonite mineral saturation state (Ωaragonite) of PML and UHL waters plotted against depth observed in the Chukchi Sea of the Arctic Ocean. Most of the data represent mixed layer water (∼0–30 m) and underlying halocline water (∼30 m to bottom depth of ∼60–100 m) from the Chukchi Sea shelf. Also included are data from the Chukchi Sea shelf slope where water depths varied from ∼60 to ∼500 m. In these locations PML, UHL, and LHL waters were present. (a) The 2002 spring and summer data and (b) the 2004 spring and summer data. Open and closed diamonds represent waters of the PML for spring and summer, respectively. The black vertical dashed lines represent the saturation depth where Ω = 1. The gray horizontal lines differentiate the approximate depths of the PML and UHL. Open squares and squares with pluses represent waters sampled at Bering Strait for spring and summer, respectively.

4.4.1. Undersaturation of Aragonite in Surface Waters Influenced by Sea Ice Melt and River Runoff

[29] A few summertime surface samples in the upper 5 m of the PML of the Canada Basin had undersaturated Ωaragonite values as low as 0.6 (Figure 8a). The low Ωaragonite values occurred at stations in the Canada Basin adjacent to Barrow Canyon and east of Point Barrow on the Beaufort Sea shelf (Figure 2). Although sampling for δ18O was limited in scope (particularly in 2004), aragonite undersaturation increased with increasing sea ice fraction (>15%; Figure 8a) in the few samples taken in the Canada Basin. In waters with the highest sea ice melt fraction, calcite undersaturation was also observed (data not shown in Figure 8). These observations indicate that meltwater from sea ice had high pCO2 values (and low Ω) compared to the PML of the Canada Basin. Elsewhere, sea ice phytoplankton communities have been observed to significantly decrease pCO2 and enhance CaCO3 mineral saturation states due to photosynthetic uptake of CO2 and high rates of sea ice phytoplankton PP [Delille et al., 2007]. Our observations in the later part of the summertime suggest that sea ice in Canada Basin had transitioned from net autotrophic to net heterotrophic conditions [Chen and Borges, 2009], with sea ice melt pCO2 increasing and CaCO3 mineral saturation states decreasing as a result.

Figure 8.

Aragonite mineral saturation states, runoff, DIC, and TA relationships for PML and UHL waters of the Canada Basin of the Arctic Ocean. (a) Ωaragonite versus runoff fraction (0.05 = 5%; 1.0 = 100%) for PML and UHL waters of the Canada Basin. Samples with >8% sea ice melt fraction are also identified (i.e., summer 2002 (solid circles) and summer 2004 (solid triangles)). (b) Ωaragonite versus seawater pCO2 for PML and UHL waters of the Canada Basin. (c) DIC versus TA for PML and UHL waters of the Canada Basin. (d) The nDIC (normalized to a salinity of 33.1) versus nTA + NO3 (normalized to a salinity of 33.1) for PML and UHL waters of the Canada Basin. Water masses are differentiated into PML, UHL, LHL, and AODW (with the LHL and AODW only occurring in the Canada Basin). Open diamonds represent waters of the PML for spring 2002, open circles represent waters of the PML for summer 2002, open squares represent waters of the PML for spring 2004, and open triangles represent waters of the PML for summer 2004. Open pluses represent waters of the core of the UHL for spring 2002, solid pluses represent waters of the core of the UHL for summer 2002, crosses represent waters of the core of the UHL for spring 2004, and asterisks represent waters of the core of the UHL for summer 2004.

[30] The PML of the Canada Basin had relatively high proportions of runoff (∼6–18%) (Figure 8a). Although PML waters with the highest fraction of river (freshwater) were saturated with respect to aragonite, Ωaragonite values decreased as runoff fraction increased (Figure 8a). Salisbury et al. [2008] have also shown that the Ω values of major Arctic rivers are close to zero for the freshwater end-member. These findings are consistent with observations that a wide variety of Arctic rivers on several shelves are net heterotrophic [e.g., Chen and Borges, 2009] and carry high pCO2 contents (with corresponding low Ω values) [Kelley, 1970; Makkaveev, 1994; Semiletov, 1999; Semiletov et al., 2007; Nitishinsky et al., 2007].

4.4.2. Undersaturation of Aragonite in the UHL of the Canada Basin

[31] In the UHL of the Canada Basin, CaCO3 saturation states were very different from the overlying PML. In the core of the UHL at a salinity of ∼33.1 ± 0.2 and temperature of ∼−1.8°C to −1.5°C, CaCO3 saturation states ranged from 1.6 to 2.5 for Ωcalcite and ∼0.9–1.5 for Ωaragonite, respectively (Figure 7). The lower values of Ωaragonite occur close to the Chukchi Sea shelf break and it appears that these waters were potentially corrosive to biogenic aragonite minerals and sediments. The core of the UHL observed in the Canada Basin has relatively little runoff (<5%) or sea ice melt (<1%) fractions (Figure 8a). The seawater pCO2 levels (∼350–500 μatm) (Figure 8b), DIC and TA values (Figures 8c and 8d) were high relative to the overlying PML.

Figure 9.

Aragonite mineral saturation state (Ωaragonite) of PML and UHL waters plotted against depth observed in 2002 and 2004 in the Canada Basin of the Arctic Ocean. These data are superimposed on all data from the Chukchi Sea observed during the SBI project. The black vertical dashed lines represent the saturation depth where Ω = 1. The gray horizontal lines differentiate the approximate depths of the PML, UHL, and LHL in the deep Canada Basin. Open diamonds represent waters of the PML for spring 2002, open circles represent waters of the PML for summer 2002, open squares represent waters of the PML for spring 2004, and open triangles represent waters of the PML for summer 2004. Open pluses represent waters of the core of the UHL for spring 2002, closed pluses represent waters of the core of the UHL for summer 2002, crosses represent waters of the core of the UHL for spring 2004, and asterisks represent waters of the core of the UHL for summer 2004.

[32] The low Ωaragonite values and high seawater pCO2 levels in the UHL on the Canada Basin presumably reflect the contribution of CO2 remineralized from organic matter produced on the Chukchi Sea shelf and transferred by shelf-basin exchanges. Jutterström and Anderson [2005] reported CaCO3 saturation states calculated from DIC and TA data collected during different trans-Arctic Ocean cruises. They show undersaturated conditions for aragonite (and even calcite) in the UHL across much of the Canada Basin. Thus, low saturation states for aragonite observed in the Canada Basin adjacent to the Chukchi Sea during the SBI project appear ubiquitous across the Canada Basin.

4.4.3. Lysocline Depths for Aragonite and Calcite in the Canada Basin

[33] Beneath the UHL of the Canada Basin, considerable vertical differentiation in CaCO3 saturation states was observed (Figure 4). In the LHL, CaCO3 saturation states ranged from 2 to 3 for Ωcalcite and 1.5–1.4 for Ωaragonite, respectively (Figures 5 and 6). Unlike the UHL, the LHL is thought to have distinct Eurasian Basin and Atlantic Ocean origins [Kinney et al., 1970; Jones and Anderson, 1986; Jones et al., 2003] with a residence time of 9.6 ± 4.6 years [Ekwurzel et al., 2001]. In the deep Canada Basin of the Arctic Ocean, CaCO3 saturation states decreased in waters of the AWL and AODW (Figure 4).

[34] In a previous study, Jutterström and Anderson [2005] used an Ω value of 0.8 from Milliman et al. [1999] to define the lysocline for CaCO3 minerals in the Arctic Ocean. Our data indicate that the saturation horizon (i.e., Ω = 1) and lysocline (i.e., Ω = 0.8) for aragonite at depths of ∼2800–3000 m and ∼3500 m, respectively (Figure 4). For calcite, we find calcite oversaturation (i.e., Ω > 1) at the deepest depths. These findings are similar to those reported by Jutterström and Anderson [2005], confirming that calcite and aragonite minerals, shells and tests are likely not to be dissolved through the water column and will be preserved in the deepest portions of the Canada Basin. This presumption is confirmed by the presence of aragonitic pteropod shells in sediments at ∼3500 m deep in the Canada Basin [Belyaeva and Khusid, 1990].

[35] In the deep waters of the Canada Basin, CaCO3 saturation states are similar to source waters in the North Atlantic Ocean [Feely et al., 2004]. The residence time of AWL in the Arctic Ocean is thought to be ∼20 ± 10 years [Anderson et al., 1999; Macdonald et al., 2009] with evidence of recent warming and salinification of the AWL [e.g., Polyakov et al., 2004]. Thus, in the future, CaCO3 saturation states of the LHL and AWL are likely to be strongly influenced by variability in the inflow of Atlantic water into the Arctic Ocean [e.g., Polyakov et al., 2004]. In the deepest parts of the Arctic Ocean (i.e., >1500 m depth; AODW and bottom water), Ωaragonite values are low and close to undersaturation. The low saturation states observed in deep water of the Arctic is not surprising due to relative isolation and slow ventilation of the deep waters, which have a residence time of ∼350–450 years [e.g., Schlosser et al., 1994, 1997; Becker and Bjork, 1996; Timmermans et al., 2005].

4.4.4. Implications for Air-Sea CO2 Exchange in the Canada Basin

[36] Our observations from 2002 to 2004 suggest that the surface waters (except for localized areas of sea ice melt) of the Canada Basin were generally oversaturated with respect to CaCO3 minerals during the summer sea ice retreat and melt period. This observation is consistent with the finding that seawater pCO2 levels were in most cases low (∼<200–300 μatm) and undersaturated with respect to atmospheric CO2 levels. As shown by Bates et al. [2006], these low seawater pCO2 levels give surface waters of the Canada Basin considerable potential to absorb CO2 from the atmosphere (while enhancing Ω). This feature relates to the length of time surface layer waters are exposed to the atmosphere and the buffering capacity (or Revelle Factor) of seawater [Takahashi et al., 1993; Sabine et al., 2004]. Surface waters of the Canada Basin and Chukchi Sea shelf have unusually low Revelle Factors (∼3.5–6.5 [Bates, 2006]) compared to the surface waters of the North Pacific and North Atlantic Oceans.

[37] Due to carbonate chemistry thermodynamics, CaCO3 saturation states are primarily controlled by the ratio of DIC to TA, with Ω tending to decrease at higher DIC to TA ratios. Similarly, lower DIC to TA ratios decrease the Revelle Factor thereby increasing the capacity of seawater to absorb CO2. In the Canada Basin, PML waters have a high salinity normalized TA compared to UHL waters (Figure 8d).

[38] For PML and UHL waters of the Canada Basin, the ratio of DIC to TA decreases with the increasing runoff fraction (Figure 8d). Although the PML and UHL have similar Pacific Ocean origins (note that some fraction of PML is generated in the Arctic from sea ice melt and river runoff), the lower DIC to TA ratios observed in the PML compared to the UHL tends to increase Ω and decrease the Revelle Factor enhancing the capacity of Canada Basin surface waters to absorb CO2.

4.5. Seasonality of Ω in Surface and Subsurface Waters of the Chukchi Sea Shelf

[39] In section 4.5.1, we show that seasonality of aragonite and calcite saturation observed in surface and subsurface of the Chukchi Sea was highly influenced by biological processes. As discussed earlier in section 4.1, the Chukchi Sea experiences seasonal changes in sea ice conditions, nutrient supply, temperature, salinity, and phytoplankton PP. Biological processes act upon the carbonate system and CaCO3 saturation state in several ways. The photosynthetic uptake of CO2 through PP or NCP by noncalcifying pelagic phytoplankton acts to decrease DIC and the ratio of DIC to TA, thereby decreasing seawater pCO2, and increasing pH, and Ωaragonite and Ωcalcite (section 3.3). In contrast, respiration or remineralization of organic matter to CO2 acts to increase DIC, and the ratio of DIC to TA. This increases seawater pCO2, and lowers pH and Ω. Thermodynamically, temperature changes also act on Ω although their relative influence on Chukchi Sea and Canada Basin waters was minor (see section 4.5.1).

4.5.1. Biological Enhancement of Ω in Surface Waters of the Chukchi Sea Shelf

[40] Across the shallow depths of the Chukchi Sea shelf, saturated conditions were observed for calcite (Ωcalcite = ∼1.5–5) and aragonite (Ωaragonite = ∼1.3–4) in the surface layer (∼0–30 m) (Figure 9). In both years of observation, there was a distinct seasonality of CaCO3 saturation states, with summertime Ωcalcite and Ωaragonite values higher than the wintertime conditions sampled during the spring cruises (particularly in the upper 50 m; Table 1). The high CaCO3 saturation states of the surface layer on the Chukchi Sea shelf resulted from two factors primarily.

Table 1. Averages and Standard Deviation of Ωaragonite Values for Water Masses Located at Bering Strait and on the Chukchi Sea Shelf and Slopea
 0–50 m50–100 m100–150 m150–200 mBering Strait
  • a

    Bottom depth of 500 m. Ωaragonite values are binned into 50 m layers in the upper 200 m.

2002
Spring1.54 ± 0.281.53 ± 0.221.43 ± 0.211.43 ± 0.251.49 ± 0.13
Summer1.92 ± 0.711.31 ± 0.241.09 ± 0.111.16 ± 0.142.70 ± 0.56
2004
Spring1.68 ± 0.401.46 ± 0.311.43 ± 0.311.30 ± 0.122.08 ± 0.26
Summer1.95 ± 0.551.41 ± 0.411.32 ± 0.261.30 ± 0.222.16 ± 0.41

[41] First, the Ωcalcite and Ωaragonite values of Pacific Ocean waters entering the Chukchi Sea through Bering Strait tend to cluster at high values. For example, the Ωaragonite values of Alaskan Coastal Current (ACC) and Anadyr/Bering Sea Shelf waters across the eastern section of Bering Strait (see Figure 1) were 1.5–2.0 and 2.5–3.0, respectively (Figure 9). The average Ωaragonite values of the Bering Strait inflow were greater than 2 (Table 1). Since the circulation of the Chukchi Sea shelf is dominated by the northward flow of Pacific Ocean waters through Bering Strait, it is likely that the CaCO3 saturation states of Chukchi Sea surface water are preconditioned by the Bering Sea.

[42] The eastern Bering Sea has an extensive (>1000 km length) and broad (>500 km) shelf, with surface circulation “funneling” waters toward Bering Strait [e.g., Woodgate et al., 2005a, 2005b; Dunton et al., 2005; Grebmeier et al., 2006a]. Nutrient rich waters, sea ice formation and melt, and long hours of irradiance make the Bering Sea shelf a highly productive marine ecosystem [Wyllie-Echeverria and Ohtani, 1999]. At the shelf break, a “green belt” extends northward along the continental shelf, supporting a large population of seabirds, and marine mammals [e.g., Kinder and Coachman, 1978; Hansell et al., 1989; Springer and McRoy, 1993; Springer et al., 1996; Okkonen et al., 2004; Grebmeier et al., 2006b]. Given these conditions, and although supporting carbonate chemistry data is lacking, it is highly likely that Bering Sea phytoplankton PP preconditions the Chukchi Sea by enhancing the saturation states of surface waters flowing from the Bering Sea into the Chukchi Sea.

[43] Second, during the summertime on the Chukchi Sea shelf, when sea ice cover reached minimal levels (i.e., 0–5% sea ice), high rates of phytoplankton PP (>1 g C d−1) and NCP [e.g., Hill and Cota, 2005; Bates et al., 2005a; Mathis et al., 2009] imparted a signal on the carbonate chemistry of surface waters of the Chukchi Sea in 2002 and 2004. Both DIC and pCO2 content of the surface layer decreased in response to high rates of PP and NCP [Bates et al., 2005a, 2006], thereby increasing the Ω of aragonite and calcite. This is shown in Table 1 where the average Ωaragonite values in the upper 50 m increased from spring to summer by 0.36 and 0.27, in 2002 and 2004, respectively (Table 1). The seasonal changes in carbonate chemistry and increase in Ωaragonite values for surface waters can also be shown in the chemical sections between the Chukchi Sea shelf and the deep Canada Basin of the Arctic Ocean. One representative section at Barrow Canyon (summer of 2002) shows relatively low DIC, TA, and pCO2 values, and high Ωaragonite values (Ωaragonite = ∼1.5–4) in the surface layer (0–∼30 m) of the Chukchi Sea shelf (Figure 10). These general features were shared by each of the 14 sections occupied during the SBI program.

Figure 10.

Representative section of (a) DIC, (b) TA, (c) pCO2, and (d) Ωaragonite across the Chukchi Sea shelf into the deep Canada Basin of the Arctic Ocean at Barrow Canyon for summer 2002. Note that the pink color represents regions of aragonite undersaturation.

[44] The seasonal increase in Ωaragonite values for surface waters in the Chukchi Sea results primarily from the uptake of DIC due to high rates of PP and NCP. Similar seasonal changes in Ωaragonite and Ωcalcite values have been demonstrated in the North Pacific Ocean where values increased by ∼0.5 ± 0.2 due to euphotic zone summertime phytoplankton PP and concomitant decrease in DIC [Feely et al., 1988].

[45] In addition to biological processes, temperature also had an influence on carbonate mineral saturation states although it is minor and bidirectional. As warmer Bering Sea surface waters flow northward into the Chukchi Sea, CaCO3 saturation states decrease (ΔΩ of ∼0.1–0.2) due to cooling (∼3–8°C). However, the seasonality in surface temperatures (∼0.5–2°C) in the northern Chukchi Sea will also impart a small change in saturation states (i.e., ΔΩ of ∼<0.1). Salinity and pressure (depth) changes will have very minor impact on Ω in this region.

4.5.2. Biological Suppression of Ω in Subsurface Waters of the Chukchi Sea

[46] Biological processes imparted a large seasonal signature on CaCO3 mineral saturations states of subsurface waters of the Chukchi Sea. Beneath the surface waters (0–∼30 m) of the Chukchi Sea shelf, halocline waters (∼30–100 m) overlying the shelf sediments and benthos had relatively high DIC, TA, and pCO2, and low Ωaragonite values compared to surface values (Figure 10 and Table 1). In the subsurface, it is evident that summertime remineralization of organic matter produced and vertically exported from surface waters of the Chukchi Sea liberates CO2 and decreased [CO32−] content and the CaCO3 saturation states of aragonite and calcite. The biological influences on saturation states of subsurface waters can also be seen at depths of 100–150 m where average Ωaragonite decreased from spring to summer by as much as 0.34 in 2002 (Table 1). These patterns are consistent with similar mechanisms lowering Ω via subsurface remineralization of vertically exported organic matter that have been observed in the North Pacific Ocean [Feely et al., 1988]. In the Chukchi Sea, seasonal changes in temperature had minor impact on subsurface CaCO3 mineral saturation states since temperature changes were small (<0.5°C).

[47] During the SBI project, hydrographic and chemical transitions between waters of the Chukchi Sea shelf and Canada Basin were also observed. For example, the low DIC and pCO2 values, and high Ωaragonite values observed in the surface layer (0–∼30 m) on the Chukchi Sea shelf continue and merge with the PML of the adjacent Canada Basin (0–∼50 m) (Figure 10).

[48] The high DIC, TA, and pCO2, and low Ωaragonite values of the halocline layer on the Chukchi Sea shelf also continue and merge with the UHL of the adjacent Canada Basin (∼100–200 m). For example, Ωaragonite values ranged from ∼<1.0–∼1.5 in the core of the UHL, while Ωaragonite values were higher in the PML and LHL (Figure 10). In the waters of the LHL adjacent to the Chukchi Sea shelf and beneath the UHL, DIC and pCO2 had lower values than the overlying UHL. Ωaragonite values were higher in the LHL (∼1.3–2.0) adjacent to the Chukchi Sea shelf compared to the UHL.

[49] At the northern periphery of the Chukchi continental shelf (typically between 100 and 500 m bottom depth), distinct seasonality was observed in CaCO3 mineral saturation states. For example, during the winter/spring cruise of 2002 and 2004, Ωcalcite (not plotted) and Ωaragonite were saturated in the PML and UHL (Figure 11). However, a couple of months later, during the summertime, undersaturated conditions were observed for aragonite (Ωaragonite = ∼0.6–>1) in the shelf slope region of the Chukchi Sea (Figure 11). In 2002, slightly undersaturated conditions for aragonite (Ωaragonite = ∼0.9–1.0) were observed at depths of ∼80–160 m in the region of Barrow Canyon (stations 12, 13 and 14; bottom depths ∼200–500 m). Jutterström and Anderson [2005; Figure 3] reported slightly undersaturated conditions for aragonite at ∼50–120 m deep on the shelf slope of the Chukchi Sea in 1994. In 2004, the undersaturation for aragonite was stronger (Ωaragonite = ∼0.72–1.0) at similar depths (∼60–150 m) in Barrow Canyon (stations 23; bottom depth of ∼200–500 m) (Figure 11). In addition, to the west on the Chukchi Sea shelf, in the vicinity of Hanna Shoals, similar undersaturation (Ωaragonite = ∼0.74–1.0) for aragonite was also observed at the shelf break (bottom depth of ∼100–200 m) at depths of ∼40–125 m.

Figure 11.

Seasonal sections of Ωaragonite across the Chukchi Sea shelf into the deep Canada Basin of the Arctic Ocean at Barrow Canyon for (a) spring 2002, (b) summer 2002, (c) spring 2004, and (d) summer 2004. Note that the pink color represents regions of aragonite undersaturation.

[50] The summertime subsurface undersaturation for aragonite observed in 2002 and 2004 at the Chukchi Sea shelf break appears to be related to biological forcing. The northward and offshelf water transport of shelf waters tends to be focused through Barrow Canyon [Bates, 2006], and the area and depth of aragonite undersaturation tends to be colocated with regenerated nutrients (e.g., NH4, NO3 + NO2 [Codispoti et al., 2005]), high concentrations (>30 μM) of suspended particulate organic matter [Bates et al., 2005b; Lepore et al., 2007] and dissolved organic carbon [Mathis et al., 2005, 2007b]. Similar to the shallow Chukchi Sea shelf, summertime remineralization of organic matter to CO2 acts to decrease [CO32−] content and the saturation states of aragonite and calcite in halocline waters exported from the Chukchi Sea shelf into the Canada Basin. Mathis et al. [2009] have also shown that rates of NCP on the Chukchi Sea shelf were higher in 2004 compared to 2002. The widespread subsurface undersaturation of aragonite observed in 2004 at Barrow Canyon and across the Chukchi Sea shelf break appears appears related to the higher rates of shelf NCP, and subsequent remineralization of organic matter to CO2 at depth.

4.5.3. Seasonal Phytoplankton-Carbonate Saturation State Interaction Amplifies the Impact of Ocean Acidification

[51] The seasonality of phytoplankton PP and NCP induces divergent directions for CaCO3 saturation states for surface and subsurface waters of the Chukchi Sea shelf. In surface waters, brief but high rates of phytoplankton PP during sea ice retreat seasonally increases Ω while vertical export and subsequent remineralization decreases Ω in subsurface waters. The seasonal changes in Ω induced by biology can be described as a seasonal phytoplankton-carbonate saturation state (PhyCASS) interaction that drives divergent trajectories for carbonate chemistry in surface and subsurface waters of Arctic shelves like the Chukchi Sea. Unlike the open ocean North Pacific Ocean, where similar seasonality of Ω has been observed in surface and subsurface waters previously [Feely et al., 1988], the shallow nature of the Chukchi Sea shelf may amplify the impact of phytoplankton-carbonate saturation state interactions. Similar phytoplankton-carbonate saturation state interactions can also occur on very small scales (<100 m2 area) such as within Devil's Hole, Bermuda [Andersson et al., 2007]. In the case of the Chukchi Sea shelf, the trapping of organic matter within a narrow subsurface zone (from ∼30 m to 50–100 m) and its subsequent remineralization to CO2 by the microbial community as well as contributions from zooplankton and benthic respiration are likely to amplify the suppression of Ω compared to other shelves.

[52] We anticipate that the seasonal changes in surface and subsurface Ω on the Chukchi Sea are seasonally moderated over the wintertime by vertical mixing and homogenization of the entire water column during sea ice and brine formation [Weingartner et al., 1998]. Another factor is the short residence time of water on the Chukchi Sea shelf of a few months [Macdonald et al., 2009], with transport and inflow of Bering Sea water through Bering Strait and across the shelf. It is likely that other polar and subpolar shelves may experience similar seasonality of Ω, and phytoplankton-carbonate saturation state interactions, such as the inflow shelf of the Barents Sea [Carmack and Wassmann, 2006] or the subpolar Bering Sea shelf, where sea ice retreat, light and nutrient supply support brief but high rates of phytoplankton PP and organic matter export. In contrast, the Siberian shelves and Beaufort Sea shelf probably experience attenuated seasonality of surface and subsurface Ω due to much lower rates of surface seasonal phytoplankton PP and organic matter export [Macdonald et al., 2009].

[53] Seasonal phytoplankton-carbonate saturation state interactions amplify the existing impact of ocean acidification on the Arctic shelves. Our observations from SBI span a relatively short period of time (2002–2004) and it remains unclear if they represent conditions present each year or if this feature is a relatively recent phenomenon. Ω conditions can be estimated for the Chukchi Sea without the addition of anthropogenic CO2, thereby reflecting preindustrial conditions. At present, surface waters of the temperate and subpolar Northern Pacific Ocean have absorbed approximately 30–40 μmoles kg−1 of anthropogenic CO2 [Sabine et al., 2004]. Since Chukchi Sea shelf surface and subsurface waters primarily have a Pacific Ocean origin and recent exposure to the atmosphere, we estimate that these water masses contain ∼40 ± 5 μmoles kg−1 of anthropogenic CO2. In the Canada Basin and across the Arctic Ocean, recent estimates indicate that anthropogenic CO2 contributes ∼40 μmoles kg−1 to the DIC content of surface waters in the Canada Basin [Tanhua et al., 2009].

[54] Given the typical inorganic carbon properties of Chukchi Sea shelf waters, the addition of 10 μmoles kg−1 of anthropogenic CO2 decreases Ω by ∼0.1 ± 0.02. If the anthropogenic CO2 component to DIC observed in the Chukchi Sea and adjacent Canada Basin is removed (i.e., 40 ± 5 μmoles kg−1), Ωaragonite and Ωcalcite values would be approximately 0.4 higher. Thus, in preindustrial times, Ωaragonite undersaturation would probably not have occurred in subsurface waters of the Chukchi Sea shelf or in the UHL of the Canada Basin (if all other factors such as temperature, salinity, nutrient supply, phytoplankton PP, and organic matter production/export were considered equal). These projections lead us to conclude that aragonite undersaturation is likely to be a recent phenomenon (over the last ∼50 years) with biologically induced seasonality of Ω superimposed on a decline in Ω due to long-term ocean acidification. In other regions, the large accumulation of anthropogenic CO2 has had significant impact on carbonate chemistry and induced a rapid upward movement of the aragonite and calcite saturation horizons onto continental shelves [Park et al., 2006; Feely et al., 2008].

4.5.4. Ocean Acidification and Ω Seasonality Impacts for Benthic Fauna on the Chukchi Sea Shelf

[55] The seasonal phytoplankton-carbonate saturation state interaction evident on the Chukchi Sea shelf has potentially both positive and negative impacts on pelagic and benthic communities on the region. The benthos of the northern Chukchi Sea is dominated by echinoderms (e.g., Ophiura sarsi, Gorgonocephalus caryi; Strongylocentratus droebachiensis) that secrete HMC [Feder et al., 1994; Piepenburg, 2005], but also includes calcifying epifaunal and infaunal bivalves and mollusks [e.g., Feder et al., 1994, 2005, 2007; Dunton et al., 2005; Grebmeier et al., 2006a] that secrete bimineralic shells of aragonite and calcite. In the region of Barrow Canyon, integrated chlorophyll and rates of organic matter (OM) production are highest, and coincide with the highest benthic biomass and sediment oxygen consumption rates in the Chukchi Sea [e.g., Grebmeier et al., 2006a]. The high benthic biomass contributes food sources for large benthic-feeding mammal populations (including walrus Odobenus rosmarus and gray whale Eschrichtius robustus) [Feder et al., 1994].

[56] In the surface waters (∼0–30 m) of the Chukchi Sea, high rates of PP and NCP maintain and induce oversaturated Ω conditions. At present, this has little impact on pelagic calcifying organisms such as coccolithophorids since they are typically absent from the Chukchi Sea [Ashjian et al., 2005]. For the shallow nearshore of the northern Chukchi Sea, oversaturation of CaCO3 minerals provides suitable conditions for the calcifying components of the benthos (e.g., mollusks, bivalves, echinoderms [Feder et al., 2005]).

[57] In contrast, in the subsurface waters (<30 m deep) that are closer to sediments on the Chukchi Sea shelf, remineralization of OM induces undersaturated Ω conditions for aragonite (and HMC). At the present time, these waters are seasonally corrosive to aragonite and HMC mineral components of the Chukchi Sea sediment. More importantly, calcifying organisms such as mollusks, bivalves and echinoderms (which have a low mol% Mg HMC) are negatively impacted in most ocean acidification scenarios [e.g., Kuffner et al., 2008]. Depending on the mol% Mg for HMC [Andersson et al., 2008], undersaturated conditions for HMC may be present year-round. In some taxa (but not all), the reduction of Ω due to the reduction in pH decreases the rate of calcification of calcium carbonate shells and tests [Ries et al., 2008], and juvenile recruitment into the sessile phase [Gazeau et al., 2007]. If some calcifying components of the benthos are negatively impacted, then the community structure and support for higher trophic levels may change. Along with other environmental pressures on the benthos such as sea ice loss, warming, changes in stratification and nutrient supply, and response to resource utilization, our observations indicate that ocean acidification and subsurface suppression of Ω should be added to the probable list of environmental stressors of the Chukchi Sea.

5. Concluding Statements

[58] Our observations indicate that subsurface waters of the Chukchi Sea, localized areas of surface water highly influenced by sea ice melt and the upper halocline of the adjacent Canada basin exhibit aragonite undersaturation. Seasonal phytoplankton-carbonate saturation state interactions enhance subsurface aragonite undersaturation that is already imparted by ocean acidification. Under business-as-usual scenarios for CO2 emissions, it is expected that an additional 50–100 μmoles kg−1 of anthropogenic CO2 will be absorbed by the surface ocean by the year 2100 [IPCC, 2007]. Given these emission scenarios, Ω would decline over the next century by 0.5 ± 0.2 due to ocean acidification. Considering Ω only, it might be expected that by 2100, strong undersaturation of aragonite, HMC and perhaps even low Mg calcite, would be present on the Chukchi Sea shelf (and other Arctic shelves) for most of the year.

[59] Over the last several years, the pace of decline of Arctic Ocean sea ice has accelerated beyond model predictions [Winton, 2006; Holland et al., 2006; Overland and Wang, 2007; Stroeve et al., 2007; Maslanik et al., 2007; Shimada et al., 2006; Giles et al., 2008]. In summer 2007 and 2008, sea ice extent reached a seasonal minima 25% lower than any previously observed in the satellite record [Maslanik et al., 2007; Comiso et al., 2008] comprising an additional loss of ∼1.2 million km2. As a result of increased open water, recent studies have also indicated that the phytoplankton-growing season has lengthened and that phytoplankton PP has increased by 10–40%, especially in the Chukchi and Beaufort seas during the last decade [Arrigo et al., 2008]. Consequently it is probable that the seasonal phytoplankton-carbonate saturation state interaction has increased recently with surface increases in Ω (which mitigates the effect of ocean acidification) and suppression of subsurface Ω (which amplifies the effect of ocean acidification). However, the future trajectory of Ω in the Chukchi Sea, Canada Basin and broader Arctic Ocean is best summarized as uncertain due to the challenges in predicting future changes in sea ice cover, temperature, stratification and nutrient supply, freshwater inputs of water and organic carbon, and complex physical and biological feedbacks in the region.

Acknowledgments

[60] We thank the captain, crew, and all scientific participants on the four SBI cruises. In particular, our thanks to Charlie Farmer (RSMAS, University of Miami), Cindy Moore (RSMAS, University of Miami), Christine Pequignet (University of Hawaii), and Paul Lethaby (University of Hawaii) for their participation in the field activities of SBI program. Christine Pequignet and Julian Mitchell (BIOS) are thanked for their organization of precruise and postcruise logistics. Margaret H. P. Best and Julian Mitchell (BIOS) are thanked for the analyses of DIC and TA. Marlene Jeffries (BIOS) is thanked for data management, data visualization, and computation of carbonate parameters. Dave Kadko (RSMAS) is thanked for discussions about estimating freshwater components from δ18O and salinity data. NSF grants OPP-0124868 (N.R.B.), OPP-0732093 (N.R.B. and J.T.M.), and OPP-0125082 (L.W.C.) supported this research.