Carbon cycling in the Arctic Archipelago : the export of Pacific carbon to the North Atlantic

Carbon cycling in the Arctic Archipelago: the export of Pacific carbon to the North Atlantic E. H. Shadwick, T. Papakyriakou, A. E. F. Prowe, D. Leong, S. A. Moore, and H. Thomas Department of Oceanography, Dalhousie University, Halifax NS, Canada Center for Earth Observation Science, University of Manitoba, Winnipeg MB, Canada Leibniz-Institut für Meereswissenschaften, IFM-GEOMAR, 24105 Kiel, Germany NSERC Canada Research Chair in Marine Biogeochemistry, Canada Received: 19 November 2008 – Accepted: 19 November 2008 – Published: 16 January 2009 Correspondence to: E. H. Shadwick (elizabeth.shadwick@dal.ca) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
In recent years much research has been devoted to understanding the ocean carbon cycle because of its prominent role in controlling global climate.More than 98% of the carbon in the atmosphere-ocean system is in the ocean; small changes in the ocean carbon cycle can therefore greatly influence atmospheric carbon dioxide (CO 2 ) concentration.
The Arctic Ocean is expected to be disproportionately responsive to climate changes.High latitude oceans are ecologically sensitive areas where the early detection of climactic changes may be possible.Much attention has been given to the oceans' ability to sequester anthropogenic carbon dioxide, and regions of deep water Figures

Back Close Full Screen / Esc
Printer-friendly Version Interactive Discussion formation are of particular importance for this process.Change sentence to: Seasonally ice-covered waters are sites of exceptionally high heat exchange, and may therefore potentially transfer carbon dioxide into the deep ocean with greater efficiency than their lower latitude counterparts (Smith and Gordon, 1997;Miller et al., 2002).
As such, the Arctic Ocean, and its carbon cycle in particular have been the subject of much recent study (Anderson et al., 1998a, b;Fransson et al., 2001;Miller et al., 2002;Bates et al., 2005).
A greater understanding of the linkage between the export of freshwater from the Arctic Ocean to the North Atlantic and the North Atlantic Oscillation (NAO) has emerged in recent years (Kwok, 2000;Peterson et al., 2002;Anderson et al., 2004;Peterson et al., 2006;Greene and Pershing, 2007).A high NAO index enhances and diverts the prevailing westerly winds, causing an eastward shift of the Eurasian river discharge and greater precipitation over Northern Europe (Thompson and Wallace, 2001).This eastward diversion of Eurasian runoff has been linked to a freshening of the Canada Basin (Macdonald et al., 2002).The export of freshwater from the Arctic to the North Atlantic is dominated by transport through the Canadian Arctic Archipelago and through Fram Strait (Serreze et al., 2006).Changes in this freshwater export are expected to have consequences for North Atlantic Deep Water formation (Rahmstorf, 2002).
It has been suggested that the decline in CO 2 uptake in the subpolar and temperate North Atlantic Ocean in the last decade is linked to the low or neutral NAO index over the same period.The NAO driven export of water high in dissolved inorganic carbon (DIC) from the Arctic Ocean has implications for the uptake of atmospheric CO 2 in the North Atlantic, particularly in the northwestern region (Thomas et al., 2008).An understanding of the distribution and cycling of DIC in the waters of the Canadian Arctic Archipelago is therefore of great importance to an understanding of carbon cycling in the Arctic Ocean as a whole.In the present paper, observed concentrations of DIC from west of Banks Island, eastward through the Archipelago to Jones and Lancaster Sounds, are exploited to reveal the mixing between five dominant water masses.

Sample collection and analysis
Samples were collected at 31 stations throughout the Canadian Archipelago from Baffin Bay westward to Banks Island (Fig. 1) during September and October, 2007.Approximately 500 samples were collected from the entire water column, with higher vertical resolution within the euphotic zone, at all stations shown in Fig. 1.Dissolved inorganic carbon (DIC) samples were tapped from 12 l Niskin bottles mounted on a General Oceanic 24-bottle rosette fitted with SeaBird CTD such that all chemical data are associated with high precision in-situ temperature, salinity and oxygen data.Following water collection, DIC samples were poisoned with a solution of supersaturated HgCl 2 to halt biological activity, and stored in the dark, at 4 • C, to await analysis.
All samples were analyzed on board by coulometric titration using a VINDTA 3C (Versatile Instrument for the Determination of Titration Alkalinity, by Marianda).Analytical methods for determination of DIC have been fully described elsewhere (Johnson et al., 1993;Fransson et al., 2001;Bates et al., 2005).Routine analyses of Certified Reference Materials (provided by A. G. Dickson, Scripps Institution of Oceanography) ensured that the uncertainty of the DIC measurements was less than 1.5%.River inflow to the Arctic is dominated by Eurasian rivers, with their North American counterparts playing a smaller, though still significant, role.Atlantic water enters the Canadian Archipelago, in relatively small quantities, at the surface from the East via Nares Strait, and more significantly, as deep water from the west via the Beaufort Sea and the Canada Basin.Pacific water enters from the West, and river runoff enters the Archipelago from the numerous terrestrial sources, the largest of these being the Yukon and Mackenzie Rivers.
Previous studies have reported various classifications and names for the portions of the Arctic Ocean water column (Anderson et al., 1994;Jones et al., 1998;Ekwurzel et al., 2001;Ingram et al., 2002;McLaughlin et al., 2004).In the data set reported here, salinity and DIC concentrations of discrete samples revealed a mixing of four source waters: river runoff or terrestrial water, Arctic Ocean water, Pacific Ocean water, and Atlantic Ocean water (Fig. 2).We here define river runoff (or water of terrestrial origin) as having S<29, Arctic water as having 28<S<32 (following the range of salinity values for the Canada Basin reported by Yamamoto-Kawai and Tanaka, 2005), Pacific water as having salinity S=33.1 (following Macdonald et al., 2002 andMcLaughlin et al., 2004), and Atlantic water as having salinity of S>34.
The DIC value for our river runoff end member, (denoted by subscript rro), of salinity S rro =0 was determined by fitting a regression line for all DIC samples with S<29 and taking the intercept at S=0.The resulting DIC value for our zero-salinity end member is in agreement with the estimate of Hansell et al. (2004) for the Beaufort Sea.The other end member properties are the mean value of several samples with characteristic salinities of Arctic, Pacific, and Atlantic water.End member properties are given in Table 1.
While there has been some debate as to the utility of the salinity normalization in the analysis of water mass properties, particularly in the coastal zone (Friis et al., 2003;Robbins, 2001), we here find that salinity normalized DIC, (DIC norm ), is useful parameter in the identification of water masses and conservative mixing relationships.
In the subsequent analysis, DIC normalized to a regional mean salinity of S mean =32.5, Introduction

Conclusions References
Tables Figures

Back Close
Full (i.e.DIC norm =32.5DIC/S) is reported (Figs. 2 and 4).Data with salinity S<29 were normalized using a non-zero freshwater end-member using Eq. ( 1) and following Friis et al. (2003): where the superscript "obs" refers to the observed value.

Mixing analysis
In general, knowing the salinity and DIC of the four sources: river runoff, Arctic, Pacific, and Atlantic water, one can compute their relative fractions in a seawater sample with known salinity and DIC.Using the aforementioned end-member properties and assuming three end member mixing, a version of the following equations may be used to compute the relative fractions of river runoff (f rro ), Arctic (f arc ), Pacific (f pac ) and Atlantic water (f atl ) in a given sample of known salinity (S) and DIC concentration (DIC).Not all waters have a third source water component.The following three component mass balance equations were solved for all samples to determine which waters had, for example, an Arctic, Pacific and Atlantic component.
If the three-component solution results in a negative value of, for example, runoff fraction (f rro ), then river runoff was not present, and in this case the two component mass balance system would be solved for this sample.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion

Profiles of dissolved inorganic carbon
The distribution of DIC with depth in three regions Atlantic water dominates the water column at depths greater than 200 m; the highest fraction is found at the westernmost stations, entering the Canadian Archipelago as deep water via the Beaufort Sea.The DIC concentration is notably higher at the depth of 200 m in centrally located station than is seen at the same depth in the more western station.We propose that organic matter remineralization over the shallow sill at the central location is the reason for the increased concentrations seen there.Atlantic water was also found at stations located in Lancaster and Jones Sounds, comprising between 32% and 54% of the water column (see Table 2).However, this is Atlantic water coming via the West Greenland Current (WGC) in the Labrador Sea, and not Atlantic water leaving the Arctic ocean as deep water from the Beaufort Sea.

Mixing of water masses
In the plot of salinity normalized DIC versus salinity, five mixing regimes are revealed (Fig. 4).Mixing of river runoff, or water of terrestrial origin, with Arctic water was primarily observed in the surface waters of the westernmost stations and throughout the water column at those few stations located very close to the North American continent.Mixing of Arctic and Pacific waters, with, in some cases a third source of river runoff, was observed at most stations in the depth range of roughly 30 to 100 m.None of the stations located east of 80 • W had samples belonging to this regime.Mixing of Atlantic water, entering the Arctic via the Beaufort Sea, and Pacific water, primarily found in the deepest waters of the westernmost stations, (i.e.west of 100 • W), displays the least amount of scatter about the proposed idealized "mixing line".DIC data at S>36, taken from a model simulation of Gulf Stream water (Thomas et al., 2008), though not plotted here, was used to validate the slope of this proposed line.
The fourth mixing regime has Atlantic source water of slightly lower salinity and higher DIC than the Atlantic-Pacific mixing end member.This is Atlantic water from the West Greenland Current in the Labrador Sea.The second end member for this mixing is a water mass which is here referred to as "diluted Pacific water".It has lower salinity and DIC than the Pacific water, the result of mixing of river runoff and Arctic waters.
Salinity, DIC, and total alkalinity (TA) data from the Labrador Sea in 1998 (Jones, 1998) were used to identify Atlantic water from the WGC.These data were corrected for the increased atmospheric, and subsequently oceanic, CO 2 concentration over the 10 year period since the data was collected.DIC and TA data from the 1998 Labrador Sea data set were used to compute pCO 2 (Millero, 1994), which was then scaled up according to an increase of 1.7 ppm per year.Assuming that TA remained constant over returned to Baffin Bay.The contribution of Atlantic water seen at stations in Jones and Lancaster Sounds has not come via the Archipelago, but from Baffin Bay.This seems to be in agreement with Jones et al. (2003) who, using nutrient ratios as water mass tracers, found no Atlantic water, of Arctic origin, in the outflow from either Lancaster or Jones Sound.
We therefore see two distinct mixing regimes of Atlantic-Pacific water, which are separated not only in terms of water mass properties but also geography.The upper water layer at Lancaster and Jones Sound flows out of the Archipelago and into Baffin Bay with mixing along the diluted Pacific-Baffin Bay waters (shown in green in Fig. 4), and between diluted Pacific-Labrador Sea waters (shown in red in Fig. 4).However, the inflow to the Archipelago from the West (via the Beaufort Sea) is dominated by mixing between deeper Atlantic and Pacific water (shown in blue).

Deep water carbon storage
If the distribution of DIC concentration of samples resulting from the mixing of Atlantic and Pacific waters was the result of conservative mixing alone, all samples should fall along the proposed "mixing line" (shown in blue in Fig. 4).The bowed structure of the DIC distribution (Figs. 2 and 4, in blue) implies that another, biological, process is at play.Further confirmation that this is indeed a biological process, is the absence of a water mass with high salinity (S≥34), and high DIC (DIC>2250 µmol kg −1 ), that would act to pull our samples away from the inferred mixing line, attributing this bowed structure to a physical process.Nonconservative variations in DIC are thus attributed to biology (∆DIC bio ), and are then the differences between the expected (DIC ex ) (i.e. the idealized mixing) and the observed (DIC obs ) values: Negative values of ∆DIC bio correspond to inorganic carbon deficits implying that production exceeds respiration in the system.Conversely, positive values of ∆DIC bio imply that respiration exceeds production and correspond to a surplus of inorganic carbon.979 Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
The samples associated with the greatest deviation from expected DIC values were in the range of 10 µmol kg −1 <∆DIC bio <50 µmol kg −1 .Integrating the ∆DIC bio over the salinity range of Pacific (S=33.1) and Atlantic (S=34.88)waters yields a total carbon surplus, or respiration inventory of 5.8±0.2mol C m −2 .Further confirmation that biology is driving the increased DIC concentrations seen in our Pacific-Atlantic water masses is evident from a plot of apparent oxygen utilization (AOU) for the same discrete samples (Fig. 5).The bowed structure of our DIC signal is mirrored in the AOU distribution.As in the case of DIC, AOU is integrated over the same range of salinity, and we find a total oxygen utilization of 7.7±0.5 mol m −2 , yielding a Ratio of AOU:DIC=1.22.If we consider a Redfield ratio of O 2 :CO 2 =138:106=1.30,our estimates of carbon surplus and oxygen utilization are quite reasonable (Millero, 2006).Thus primary production in the archipelago results in a drawdown of carbon to the deep water.If we now assume that the residence time of Pacific (and Atlantic) water in the Arctic Ocean is approximately 10 years (Yamamoto-Kawai et al., 2008), and the size of the study region in the Archipelago is roughly 1.34×10 5 km 2 , we arrive at a carbon draw down of 9.5±1.5×10 11g C yr −1 .If we now estimate the size of the study region in the Canadian Archipelago at roughly 0.4±0.01×10 6 km 2 we arrive at a carbon drawdown of 2.21±0.04×10 9g C yr −1 .Bulk carbon inventories such as this can provide reliable estimates of system export production (Thomas et al., 2005).The estimate made here is in general agreement with the rather sparse primary production data for this region, which ranges from 25 to 50 g C m −2 yr −1 , and provides a reasonable constraint on this estimate (Cota et al., 1996;Pomeroy, 1997).Using a value of primary production of roughly 34 g C m −2 yr −1 , following Cota et al. (1996), for a region with similar geographic location and water column depth, our calculated carbon surplus corresponds to an export production of roughly 20%.

Carbon export to the North Atlantic
The Beaufort Sea communicates directly with the Atlantic Ocean through those passages in the Arctic Archipelago that enter Baffin Bay. and Jones Sound all exit the Archipelago to Baffin Bay.Having identified the composition of water flowing out of the Canadian Archipelago, in this case via Jones and Lancaster Sound, we now estimate the carbon transport from the Arctic into this region.We find that water entering Baffin Bay from Jones Sound is composed of roughly 27% Arctic, 40% Pacific and 33% Atlantic water, while water from Lancaster Sound is composed of roughly 15% Arctic, 42% Pacific and 43% Atlantic water (Fig. 6, Table 2).The contribution from river runoff at both locations was found to be negligible for the purposes of the present discussion.
Several estimates of volume transport from the Archipelago, from Jones and Lancaster Sound have been made (Rudels, 1986;Ingram et al., 2002;Jones et al., 2003;Cuny et al., 2005;Prinsenberg and Hamilton, 2005).We applied the estimate of Ingram et al. (2002), which assumes that the net Eastward outflow from Jones Sound and Lancaster Sound are 0.3×10 6 m −3 s −1 , and 1.1×10 6 m −3 s −1 , respectively.Multiplying this volume transport by the average DIC concentration in a specified depth range gives an estimated value of carbon exported from the Archipelago to Baffin Bay at each location.The resulting export of carbon from Jones Sound to Baffin Bay is roughly 0.77×10 14 g C yr −1 , while that from Lancaster Sound is roughly 2.31×10 14 g C yr −1 (Table 2).The total transport by the Baffin Current has been estimated as 1.7×10 6 m −3 s −1 , the outflow from Jones Sound and Lancaster Sound then supply roughly 18% and 65% of this water, respectively, making the Arctic Archipelago an important source of both water and carbon to Baffin Bay and subsequently to the North Atlantic.
An alternate approach is to compare the DIC concentration of water coming into Baffin Bay via the West Greenland current with that of the outflow from the Archipelago.In this way we can estimate the carbon enrichment resulting from the outflow of Arctic and Pacific waters via the Archipelago.We assume that the North-easternmost station of the Labrador Sea transect (Fig. 1) is representative of WGC water with a mean DIC concentration of 2130 µmol kg  2).The difference between these values allows us to estimate the carbon enrichment at roughly 18 µmol kg −1 , or 0.22 g C m −3 .

Conclusions
Previous studies have relied on nutrient relationships for water mass identification in the Arctic Ocean (Anderson et al., 1994;Jones et al., 1998Jones et al., , 2003;;McLaughlin et al., 2004).We relied instead on measurements of dissolved inorganic carbon, and found it to be a useful parameter in the identification of water masses and conservative mixing relationships.Four water masses were found to dominate the water column in the Canadian Archipelago, with conservative mixing occurring between them.We were able to isolate deeper Pacific and Atlantic water entering the Arctic Ocean from the western side via the Beaufort Sea, from Pacific and water flowing eastward out of the Archipelago, and Atlantic water from the Labrador Sea encountering this outflow.An estimate of carbon transfer from the surface waters to the deep sea was made by computing the difference between observed DIC concentrations and those expected from conservative mixing alone.An estimate of oxygen utilization produces the expected Redfield ratio with exported carbon.The carbon export to Baffin Bay via Jones sound was also estimated and found to make a significant contribution with respect to both volume transport and DIC delivery to the Baffin Current, which feeds the North Atlantic.An estimate of the carbon enrichment resulting from the outflow of Arctic and Pacific waters via the Arctic Archipelago was made.
While there remains much uncertainty with respect to carbon cycling in the Arctic, this investigation into the Canadian Archipelago sheds some light on the water mass properties in this area.Export of high carbon Pacific water to the North Atlantic has implications for the uptake of anthropogenic carbon, and hence ocean acidification in this area.However, from the perspective of natural variablity, NAO driven changes in composition of waters leaving the Arctic may also influence carbon uptake in the subpolar and temperate North Atlantic.As one of the gateways from the Pacific to the 982

2. 2
Water mass definitions Atlantic water enters the Arctic Ocean through Fram Strait and the Barents Sea.Pacific water enters the Arctic Ocean through the Bering Strait and via the Chukchi Sea.

Fig. 3 :
(a) West of Banks Island, near the Canada Basin; (b) centrally located in the Canadian Archipelago; and (c) near the outflow of Lancaster Sound to Baffin Bay.River runoff was found only in near surface waters at those stations located very close to land.Arctic water is predominantly seen in the depth range of 30 to 100 m, and is fairly evenly distributed longitudinally, though the depth of penetration decreases eastward through the Archipelago.High contributions of Pacific water are seen in the range of 80 to 200 m at the westernmost stations, while the more easterly stations show slightly lesser contributions with a fairly significant fraction of Pacific water penetrating down to depths of greater than 400 m.
this 10 year period, TA and the augmented pCO 2 values were then used to compute the present day DIC concentrations.The water masses observed at stations in Lancaster and Jones Sounds are the result of mixing between the diluted Pacific and Atlantic water masses, with the Atlantic component seen at these locations composed of WGC water from the Labrador Sea.Mixing of diluted Pacific water with Atlantic water occurs in the depth range of roughly 80 to 200 m with the mixing of deeper waters (from 200-500 m) occurring between the diluted Pacific and Labrador Sea water masses (shown in red in Fig.4).WGC water from Baffin Bay enters Jones and Lancaster Sounds, encounters the sill and is −1 .Ignoring the contribution from the Hudson Strait outflow, the outflow from the Archipelago via Jones and Lancaster Sounds has a DIC

Fig. 1 .Fig. 2 .
Fig. 1.Locations of stations sampled for dissolved inorganic carbon (DIC) and hydrographic data.Red squares indicate stations occupied in 2007, blue circles indicate stations from which an older data set from the Labrador Sea was used.

Fig. 3 .Fig. 4 .Fig. 5 .Fig. 6 .
Fig. 3. Profiles of dissolved inorganic carbon (DIC) for (a) a station located west of Banks Island, (b) a station centrally located in the Canadian Archipelago, and (c) a station located near the outflow of Lancaster Sound to Baffin Bay.The coloured lines indicated the dominance of mixing between: (pink) fresh and Arctic water; (orange) Arctic and Pacific water; (blue) Pacific and deep Atlantic water coming via the Beaufort Sea, and (green) diluted Pacific and Atlantic water, coming via the Labrador Sea, respectively.
Arctic, Pacific, and two sources of Atlantic water: from the West via the Beaufort Sea, and from Baffin Bay via the West Greenland Current, flowing out of the Archipelago and into Baffin Bay are determined and the corresponding eastward transport of carbon is estimated.

Table 1 .
Arctic Archipelago plays an important role in the carbon chemistry of the Arctic ocean, which in turn plays a crucial role in the global carbon cycle.Figures Prowe, F., Lima, I. D., Doney, S. C., Wanninkhof, R., Greatbatch, R. J., Schuster, U., and Corbiere, A.: Changes in the North Atlantic Oscillation influence CO 2 uptake in the North Atlantic over the past 2 decades, Glob.Biogeochem.Cy., 22, GB4027,  doi:10.1029/2007GB003167,2008.973, 978 Thompson, D. W. and Wallace, J. M.: Regional Climate Impacts of the Northern Hemisphere End member properties used in the two and three component mass balance equations.

Table 2 .
Water mass composition for flow out of the Arctic Archipelago into Baffin Bay via Jones Sound (top), and Lancaster Sound (bottom) where volume transport is given in ×10 6 m 3 s −1 , and the total carbon transport in units of ×1014 g C yr −1 .DIC is given in units of µmol kg −1 , and f xxx , and T xxx , are the water mass fraction and corresponding carbon transport for Arctic (arc), Pacific (pac), and Atlantic (atl) water.The volume and carbon transport from both Jones and Lancaster Sounds are given in the last row.