Multi-decadal uptake of carbon dioxide into subtropical mode water of the North Atlantic Ocean

Introduction Conclusions References


Introduction
The North Atlantic Ocean contributes approximately 25 % to the global ocean sink of anthropogenic carbon dioxide (CO 2 ) from the atmosphere (Takahashi et al., 2002(Takahashi et al., , 2009)).Recent studies have suggested that the basinwide CO 2 sink in the North Atlantic has decreased over the last decade (Schuster and Watson, 2007;Watson et al., 2009).However, ocean observations conducted over longer timescales indicate that surface ocean pCO 2 (partial pressure of CO 2 ) has increased at the same rate as the atmosphere, with the implication that the North Atlantic Ocean CO 2 sink has not changed significantly over multi-decadal timescales (McKinley et al., 2004;Thomas et al., 2008;McKinley et al., 2011;Bates et al., 2012).Complicating any assessment of change in the CO 2 sink-source strength of the North Atlantic Ocean is a recognition that natural interannual variability imparted by modes of climate variability -such as the North Atlantic Oscillation (NAO; Hurrell and Deser, 2009), Atlantic Multidecadal Variability (AMV; McKinley et al., 2004;Ullman et al., 2009;Metzl et al., 2010;McKinley et al., 2011) and El Niño-Southern Oscillation (ENSO) -play important roles in controlling air-sea CO 2 fluxes in the subtropical and subpolar gyres of the basin (Gruber et al., 2002;Bates et al., 2002;Thomas et al., 2008;Gruber, 2009;Levine et al., 2011).
In the subtropical gyre of the North Atlantic Ocean, the formation of subtropical mode water (STMW) or Eighteen Degree Water (EDW; Worthington, 1959Worthington, , 1976;;Hanawa and Talley, 2001) in winter just south of the Gulf Stream (Marshall et al., 2009;Fig. 1a) may contribute substantively to variability in the CO 2 sink in the North Atlantic (Gruber et al., 2002;Bates et al., 2002;Gruber, 2009).These studies indicate that the STMW absorbed 0.030 to 0.240 Pg C yr −1 in the 1990s (Gruber et al., 2002;Bates et al., 2002), while model studies suggest that the wintertime replenishment of CO 2 to the storage of anthropogenic CO 2 in STMW varies by approximately 0.100 Pg C yr −1 , depending on the state of the NAO (Levine et al., 2011).More recently, studies have focussed on determining the rate of CO 2 uptake at Published by Copernicus Publications on behalf of the European Geosciences Union.N. R. Bates: Multi-decadal uptake of carbon dioxide into subtropical mode water Manuscript bg-2011-448. Bates, N.R., 2011, submitted Marshall et al. (2009).This figure is modified from Fig. 2a and Fig. 12a of Marshall et al. (2009).(B) Location of the BATS (Bermuda Atlantic Time-series Study; 31 • 40 N, 64 • 10 W) site as well as the long-term Hydrostation S ocean time-series.
the site of STMW outcrop during its winter formation (Andersson et al., 2012) as part of the CLIMODE (Climate of Mode Water) project (Marshall et al., 2009).In this paper, multi-decadal changes from 1988 to 2011 in the dissolved inorganic carbon (DIC) content of the STMW observed at the Bermuda Atlantic Time-series Study (BATS; 31 • 40 N, 64 • 10 W; Fig. 1b) site near Bermuda are evaluated.Estimates of how much CO 2 has been taken up in STMW over the last 2 decades are made.This is viewed in context with rate of equilibration of the ocean with anthropogenic CO 2 increase in the atmosphere, observed increases in the rates of phytoplankton primary production, and vertical export of sinking organic carbon in the North Atlantic sub-tropical gyre (Lomas et al., 2010).Finally, relationships between the DIC content of STMW and NAO state are examined to determine whether the uptake and storage of CO 2 in STMW has changed contemporaneously with the NAO.STMW forms during winter just south of the Gulf Stream in a region of very high heat loss from the ocean, which triggers intense convective mixing (Marshall et al., 2009) and subsequent isopycnal transfer into the thermocline of the subtropical gyre.STMW is a classically identified water mass that has a temperature and salinity range of 17.8-18.4• C and 36.5 ± 0.05, respectively, and is commonly identified by a characteristic minimum in the vertical gradient of potential density or potential isopycnic vorticity (Jenkins, 1982).The volume of STMW formed each winter ranges from about 4.6 to 8 Svy (where 1 Svy = 3.15 × 10 13 m 3 and corresponds to a 1 Sv (10 6 m 3 s −1 ) flow sustained for 1 yr; Speer and Tziperman, 1992;Kwon and Riser, 2004;Maze et al., 2009;Forget et al., 2011).The volume of STMW in the subtropical gyre is approximately 75 to 80 Svy (Worthington, 1976;Forget et al., 2011) with seasonal and annual fluctuations ranging from 3.5 to 8.6 Svy (Kwon and Riser, 2004;Forget et al., 2011) and annual fluctuations at 1.4 Svy, respectively.The latter rate implies that the residence time of STMW is 50 yr, but recent studies indicate that approximately 8 yr is sufficient to ventilate STMW (Forget et al., 2011).
The outcrop and volume of STMW formation is coupled to the NAO, which is the dominant climate mode that influences interannual and multi-decadal variability in the North Atlantic Ocean (Hurrell and Deser, 2009;Jenkins, 1982;Hurrell, 1995).During NAO positive phases, stronger atmospheric pressure gradients between the subpolar and subtropical region increase winter storm frequency and shift the Gulf Stream northward (Hurrell, 1995;Hurrell et al., 2001;Marshall et al., 2001) with a lag of about 1 to 2 yr.During NAO negative phases, the Icelandic atmospheric low pressure shifts winter storm tracks southward, winter storms tend to be fewer in number, and the Gulf Stream shifts southward.The outcrop area, extent and formation rate of STMW south of the Gulf Stream also shift in concert with NAO variability (Alfutis and Cornillon, 2001;Joyce et al., 2000;Levine et al., 2011).

Determining long-term trends
A time-series of observations of seawater carbonate chemistry in the upper ocean have been collected in the subtropical gyre of the North Atlantic Ocean near Bermuda since 1988 at the Bermuda Atlantic Time-series Study (BATS) site (Bates, 2007;Bates and Peters, 2007).This includes monthly water column sampling for DIC and total alkalinity (TA), with analysis of samples at the Bermuda Institute of Ocean Sciences (BIOS) using highly precise and accurate coulometric and potentiometric techniques, respectively (Bates et al., 1996a(Bates et al., , b, 2012;;Bates, 2001).DIC is defined as (Dickson et al., 2007): The frequency of sampling at BATS was not uniform in time.
Beginning in 1988, samples were collected 9-12 times a year but not always at the depth of the STMW, but the sampling frequency increased to 14-15 times a year over the last 15 years.The time-series of DIC in STMW is not weighted to spring conditions, unlike surface observations where seasonality exerts significant influence on trends in seawater carbonate chemistry (Bates et al., 2012).

Sampling methods
Samples from the BATS site were collected into 500 ml Pyrex bottles, poisoned with HgCl 2 , sealed with ground glass stoppers and subsequently analyzed at BIOS (Bates et al., 1996a;Bates, 2007).Since the early 2000s, smaller Pyrex bottles (350 ml) were used.Samples were typically analyzed within a few months of collection.

Analytical methods
At BIOS, DIC was determined using coulometric methods with a SOMMA system (Bates et al., 1996a;Bates, 2007).
During the first two years of sampling (1988)(1989)(1990), DIC samples from BATS cruises 1 to 21 were analyzed at WHOI, and thereafter at BIOS.Each day of analysis, DIC was calibrated with known volumes of pure CO 2 gas, and certified reference materials (CRMs; Dickson et al., 2007) were routinely analyzed each day of analysis.Potentiometric titration methods were also used for determination of alkalinity at BIOS (Bates et al., 1996b).In the 1990s, a manual alkalinity titrator was used, replaced by an automated VINDTA 2S (Versatile Instrument for the Determination of Titration Alkalinity; manufactured by Marianda Co.) in the early 2000s.
In both systems, at least 15-20 titration points past the carbonic acid end point were determined for each sample, and TA was then computed using non-linear least squares methods (Dickson et al., 2007).Surface water from Sargasso Sea was typically used to condition the titrator while CRMs were routinely used.Analytical precision for both DIC and TA at BIOS was typically < 0.2 % for more than 1000 withinbottle and between-bottle replicate analyses.

Identifying changes in STMW CO 2 content
The time-series of seawater carbonate chemistry for STMW was compiled using temperature criteria (i.e.17.8-18.4• C) to identify STMW water, as in previous studies (Bates et al., 2002).During a few cruises, total alkalinity was not determined analytically, but it was computed from salinity with an error of 2.8 µmoles kg −1 (Bates et al., 1996b; TA = −47.155+ 66.576 × S).

Seawater carbonate chemistry computations
The complete seawater carbonic acid system (i.e.CO 2 , H 2 CO 3 , HCO − 3 , CO 2− 3 , and H + ) can be calculated from a combination of two carbonate system parameters (i.e.DIC, TA, pCO 2 and/or pH) and temperature and salinity.Seawater pCO 2 , pH, [CO 2− 3 ], mineral saturation states for calcite ( calcite ) and aragonite ( aragonite ), and the Revelle factor (β; = ∂ ln pCO 2 /∂ ln DIC) were computed from DIC, TA, temperature and salinity data using the program CO2calc (Robbins et al., 2010).The carbonic acid dissociation constants of Mehrbach et al. (1973) as refit by Dickson and Millero (1987) was used for the computation, as well as dissociation constants for HSO − 4 (Dickson, 1990).Here, pCO 2 is the partial pressure of CO 2 in equilibrium with seawater, while pH is expressed on the total seawater scale.
Salinity normalized DIC (i.e.nDIC) and TA (i.e.nTA) data were also determined to account for local evaporation and precipitation changes.Trends and regression statistics for CO 2 changes in the STMW are given in Table 1 (using a 17.8-18.4• C temperature criterion to define STMW; Fig. 2; Table 1).A density criterion (i.e.26.4 σ isopycnal surface) can also be used to define STMW but does not give significantly different results.Surface trends are also discussed here for comparative purposes, but more detailed information about surface seawater carbonate chemistry changes are given elsewhere (Bates et al., 2012).

Trend analysis and statistics
Trend analysis was performed with observed data for STMW (Table 1).Trends were also computed for the two time periods (1988-2000; 2000-2011; Table 2).Regression statistics given were slope, error, r 2 , p-value and n.Trends with pvalues greater than 0.01 were deemed statistically not significant.Monthly and winter JFM (January, February, March) mean NAO index data was obtained from the NOAA Climate Prediction Office (http://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/nao index.html).Correlations between winter NAO, mixed layer depth, and surface and STMW DIC contents are shown in Table 3. Mixed layer depth (MLD) was computed from the depth at which temperature was less than 0.5 • C cooler than surface temperatures.

Detecting rates of ocean acidification in STMW
Trend analysis indicates that relevant indicators of ocean acidification in STMW such as pH have also significantly decreased.The pH of STMW decreased at a rate of −0.0022 ± 0.0002 yr −1 , a seawater pH decline of 0.05 over the past 2 decades (Fig. 2d; Table 1).This rate of pH decrease is higher than that for surface waters at BATS (−0.0017 ± 0.0001 yr −1 ) as well as off Hawaii and the Canary Islands (−0.0014 to −0.0019 yr −1 ) (Dore et al., 2009;González-Dávila et al., 2010).This change in pH of the STMW represents a 12 % increase in hydrogen ions (H + ) since 1988.Other indicators of ocean acidification at BATS such as [CO 2− 3 ], calcite and aragonite have also decreased by −0.87 ± 0.07 µmol kg −1 yr −1 , −0.022 ± 0.002 yr −1 , and −0.014 ± 0.001 yr −1 , respectively (Fig. 2d and e; Table 1).The ecosystem response in the mesopelagic (i.e. at the depth of STMW within the subtropical gyre) to ocean acidification is unknown at present.

Determining the rates of CO 2 uptake into STMW over the last 2 decades
Trend analyses reveal that DIC STMW has increased at a significantly higher rate than observed for surface waters in the subtropical gyre .For example, the DIC of surface waters (i.e.DIC surface ), expressed here as δDIC surface /δt 1988−2011 , increased at a rate of +1.08 ± 0.05 µmol kg −1 yr −1 (Bates et al., 2012;Fig. 3a).This trend is similar to that expected from equilibration of STMW with increasing atmospheric CO 2 due to the anthropogenic CO 2 transient.with the observed atmospheric pCO 2 increases (due to anthropogenic CO 2 ) over the same time 1988-2011 period (i.e.+1.72 ± 0.01 ppm yr −1 ).The trend for δDIC anthropogenic /δt 1988−2011 is +1.06 µmol kg −1 yr −1 , and this effectively represents the uptake rate of anthropogenic CO 2 (C ant ) from the atmosphere.Examination of trends in seawater carbonate chemistry indicate that the DIC content of surface and STMW waters observed at the BATS site have increased at divergent rates over the last 2 decades (Fig. 3a).Once STMW is formed during winter, isopycnal mixing and subduction adds to the existing pool of STMW found between surface/seasonal thermocline waters and the permanent thermocline in the subtropical gyre (approximately 250-450 m deep).STMW is ultimately subducted to deeper depths with time (Behringer and Stommel, 1980;Spall, 1992;Fig. 4a).Using the δDIC STMW /δt 1988−2011 trend of +1.51 ± 0.08 µmol kg −1 yr −1 , and given that the volume of STMW in the North Atlantic subtropical gyre is about 75 Svy (Forget et al., 2011), it is estimated that the STMW has accumulated 0.985 ± 0.018 Pg C since 1988 (Fig. 4b).The error in the amount of CO 2 taken up into STMW relates to that propagated from the trend error. Howver, the volume of STMW could be underestimated by 10 % (Forget et al., 2011), which implies that the estimate of 0.985 ± 0.018 Pg C, and other estimations discussed later in the paper, may be higher by 10 %.The separation of STMW from contact with the atmosphere within the subtropical gyre (with exceptions discussed later) and eventual subduction to deeper waters constitutes a mechanism for transferring CO 2 from the airsea interface to the deep ocean, and thus constitutes a sink for CO 2 .The sink of CO 2 into the STMW, averaged annually for the period 1988-2011 (0.042 Pg C yr −1 ), is approxiamtely 20 % of the CO 2 uptake in the North Atlantic Ocean in the zone between 14 • and 50 • N (0.220 Pg C yr −1 ; Takahashi et al., 2009).Thus, over the last 2 decades, the STMW has been a quantitatively important additional CO 2 sink term in the sink-source CO 2 budget of the North Atlantic Ocean.

The STMW CO 2 sink and its attribution
The change in DIC content of the STMW (i.e.δDIC STMW /t 1988−2011 ) potentially results from a number of factors, including: (1) equilibration at the STMW outcrop with the anthropogenic CO 2 increase in the atmosphere (i.e.δDIC anthropogenic /t 1988−2011 ); (2) physical changes at the outcrop site of STMW winter formation encompasses the summed changes due to variability in air-sea gas exchange of CO 2 , the preformed DIC content of waters that contribute to STMW formation during winter, formation rate of STMW and its subsequent dispersion rates in the subtropical gyre (here termed δDIC outcrop /t (3) The δDIC anthropogenic /t 1988−2011 term (Eq. 3) was estimated earlier at 1.06 µmol kg −1 yr −1 , with the implication that the uptake of anthropogenic CO 2 constitutes approximately 70 % of the observed increase in DIC STMW .The rate of increase in atmopsheric CO 2 at Bermuda has not changed significantly over the past couple of decades (Bates, 2007;Bates et al., 2012), and thus it is assumed that the equilibration rate of anthropogenic CO 2 with surface waters   Speer and Tziperman, 1992;Kwon and Riser, 2004;Forget et al., 2011), volume of STMW (Svy; Maze et al., 2009;Forget et al., 2011), and seasonal/annual fluctuations in the volume of STMW (Kwon and Riser, 2004;Forget et al., 2011)  (i.e.δDIC anthropogenic /t 1988−2011 ) has also not changed significantly.It is estimated that the uptake of anthropogenic CO 2 into STMW since 1988 was 0.692 ± 0.018 Pg C or approximately 70 ± 1.8 % of the total STMW CO 2 sink (0.985 ± 0.018 Pg C; Fig. 3b).The remaining terms in Eq. ( 1) comprise the remaining 30 % of the observed changes.
Term two of Eq. (3) (i.e.δDIC biology /t 1988−2011 ) can be approximated by examining changes in organic carbon export and remineralization and/or biogeochemical properties (e.g.dissolved oxygen and inorganic nutrients).Term three of Eq. (3) (i.e.δDIC outcrop /t 1988−2011 ) is difficult to determine directly due to the lack of sustained observations at the STMW site of formation, but some field (Andersson et al., 2012) and model studies (Levine et al., 2011) provide some inferences on this term.Thus, Eq. ( 3) can be solved by difference, with the caveat that there are many uncertainties underlying this approach: Since 1990, integrated rates of in situ primary production (0-140 m) have increased at BATS by 44 % (+0.85 mmol C m −2 d −1 yr −1 ), and since 1996 by 98 % (Lomas et al., 2010).During the 1990-2007 period, the sinking particulate organic carbon (POC) flux record at 150 m increased by 71 % (+0.12 mmol C m −2 d −1 yr −1 ).However, at 300 m, no long-term change was observed in POC flux (Lomas et al., 2010), with the implication that the observed increase in ocean carbon export is attenuated www.biogeosciences.net/9/2649/2012/by mesopelagic remineralization of sinking organic carbon back to CO 2 in the 150 m to 300 m depth zone.Since this depth zone coincides approximately with the upper/middle portion of the STMW, enhanced mesopelagic remineralization likely contributes to the observed increase in DIC STMW .Here, the increase in POC flux (i.e.+0.12 mmol C m −2 d −1 yr −1 ) observed at BATS is converted to DIC change over time, assuming a 150 m thickness of STMW at BATS.The increase in DIC due to enhanced remineralization is estimated at +0.29 ± 0.05 µmol kg −1 yr −1 (i.e.

A multi-decadal change in the STMW CO 2 sink and its relation to NAO variability
A closer inspection of the trends in DIC STMW reveals considerable difference between the 1990s and 2000s.The trend for DIC STMW determined for 1988 to 2001 was much higher at 2.06 ± 0.26 µmol kg −1 yr −1 (Bates et al., 2002, and recalculated here) compared to that observed for the 2001 to 2011 period at 0.56 ± 0.21 µmol kg −1 yr −1 (Fig. 3a; Table 2).Over the last 2 decades, the NAO (JFM) winter state has transitioned from mostly positive values in the 1990s to neutral and negative values in the 2000s (Fig. 3b).Contemporaneous with the change in winter NAO, the mean winter mixed layer depth (MLD) has also deepened significantly from 110 to 150 m (Fig. 3b).Notably, a significant decline in DIC STMW was observed in 2011 that coincides with a strongly negative NAO (JFM) winter phase (Figs. 2f and 3b).If data from 2011 is not included in the trend analysis, the δDIC STMW /t 1988−2010 was significantly higher at 1.74 ± 0.08 µmol kg −1 yr −1 (see Table 1).
In contrast, winter DIC STMW appears to lag NAO by up to 1-2 yr (Levine et al., 2011), with the lag observed at BATS greater during NAO positive phases.NAO, mixed layer depth and the difference between surface and STMW DIC also appear well correlated with each other (Fig. 3e and f; r 2 values of 0.25-0.45;Table 3).
The two periods of observation, 1988-2001 and 2001-2011, correspond approximately to a decade of positive winter NAO (mean of +0.73, 1989NAO (mean of +0.73, -2001) ) followed by a decade of neutral/negative NAO (mean of +0.10, 2001-2011).Levine et al. (2011) suggest that STMW takes up more CO 2 during NAO positive phase as the Gulf Stream and STMW outcrop move northward in response to NAO.During an NAO neutral to negative phase, the Gulf Stream and STMW outcrop move south with less CO 2 taken up by STMW.Levine et al. (2011) show a STMW CO 2 uptake variability of 0.100 Pg C yr −1 , depending on the state of the NAO.
In the 1990s, trend analyses and STMW sink CO 2 estimates indicate that the accumulation of CO 2 into STMW was large during the NAO positive phase, with C ant computed to be 0.391 ± 0.077 Pg C or approximately 51 ± 10 % of the total 0.760 Pg C sink in the 1990s (Fig. 3c).Furthermore, it is estimated that enhanced remineralization of sinking organic carbon in STMW (Lomas et al., 2010) may have contributed 0.109 ± 0.019 Pg C or 14 ± 2 % to the total sink.The remainder (i.e.0.262 Pg C or 34 % of the total sink) must be due to changes in DIC contents during formation of the STMW at the outcrop due to a combination of changes in air-sea gas exchange rates and water mass contributions to STMW.With movement northward of the Gulf Stream and STMW outcrop during an NAO positive period, entrainment of high DIC from deeper water (as shown in DIC sections across the STMW outcrop, Marshall et al., 2009;Andersson et al., 2012) and less mixing with relatively low DIC subtropical gyre waters during winter STMW formation potentially contribute to higher DIC input into the STMW (Fig. 5e).
In the 2000s, the observed uptake was much reduced compared to the 1990s at 0.159 Pg C (Fig. 5f).Indeed, this is much less than predicted from equilibration with anthropogenic CO 2 in the atmosphere (i.e.0.301 ± 0.030 Pg C).Furthermore, an additional 0.082 ± 0.014 Pg C may have accumulated in STMW during this period due to enhanced remineralization of organic carbon in STMW.These two terms have to be balanced by reduced input of DIC to the STMW during winter formation of STMW (here computed at −0.025 Pg C; Fig. 5f) and loss from STMW due to mixing with surface waters/seasonal thermocline and/or enhanced diapycnal mixing with waters deeper than STMW.Given that the winter DIC surface increases by 0.70 µmol kg −1 yr −1 , we estimate that STMW may have lost 0.198 Pg C (Fig. 5f) due to mixing with surface waters.The implication is that much of the anthropogenic CO 2 uptake into STMW during the 2000s will have been lost to surface waters (in contact with the atmosphere) and thus ventilated over a shorter timeframe.Thus, in contrast to the 1990s, the STMW appears not to have been a sink for anthropogenic CO 2 in the 2000s.Indeed, without the potential increase in DIC STMW due to enhanced remineralization of organic carbon, the DIC content of STMW would have declined during the 2000s.
The observations at BATS thus indicate that the phase of the NAO is important for determining the long-term and short-term sinks of CO 2 into STMW.During NAO positive phases, the CO 2 sink into STMW appears to be long-term (i.e. more than 10 yr), whilst it is short-term (i.e. less than 10 yr) during the NAO neutral/negative period of the 2000s.
As such, the reduced potential for uptake and long-term storage of CO 2 in STMW during the 2000s likely contributed to the observed decline in the North Atlantic Ocean CO 2 sink during this decade (Schuster and Watson, 2007;Watson et al., 2009).It may be that the 1990s uptake and storage of CO 2 into the STMW reflects an anomalous decade since the winter NAO was predominantly neutral/negative from the 1950s to 1980s, with a return to this state in the 2000s.This study supports the findings of other studies (Gruber et al., 2002;Bates et al., 2002;McKinley et al., 2004;Thomas et al., 2008;Gruber, 2009;Ullman et al., 2009;McKinley et al., 2011) that natural climate variability such as NAO imparts a significant influence on CO 2 sinks in the North Atlantic Ocean.

Conclusions
The observations at BATS thus indicate that the phase of the NAO is important for determining the long-term and shortterm sinks of CO 2 into STMW.During NAO positive phases, the CO 2 sink into STMW appears to be long-term (i.e. more than 10 yr), whilst it is short-term (i.e. less than 10 yr) during the NAO neutral/negative period of the 2000s.As such, the reduced potential for uptake and long-term storage of CO 2 in STMW during the 2000s likely contributed to the observed decline in the North Atlantic Ocean CO 2 sink during this decade.It may be that the 1990s uptake and storage of CO 2 into the STMW reflects an anomalous decade since the winter NAO was predominantly neutral/negative from the 1950s to 1980s, with a return to this state in the 2000s.Thus, the NAO imparts a significant influence on CO 2 sinks in the North Atlantic Ocean.

Fig. 1 .
Figure 1 Fig. 1.Location map of general area of STMW outctop and the BATS site near Bermuda in the NW Atlantic Ocean.(A) General outcrop area of STMW in the NW Atlantic Ocean, showing locations of climatological 17 • C and 19 • C isotherms and mean STMW formation rate area (orange and purple) for the 2004-2006 period from Marshall et al. (2009).This figure is modified from Fig. 2a and Fig. 12a of Marshall et al. (2009).(B) Location of the BATS (Bermuda Atlantic Time-series Study; 31 • 40 N, 64 • 10 W) site as well as the long-term Hydrostation S ocean time-series.

Fig. 3 .
Fig. 3. Time-series and relationships between DIC, mixed layer depths (MLD) and NAO index from 1988 to 2011 at the BATS (Bermuda Atlantic Time-series Study; 31 • 40 N, 64 • 10 W) site located near Bermuda in the NW Atlantic Ocean.Slopes and statistics of regressions are listed in Table 1.(A) Time-series of DIC in surface (DIC surface ; µmol kg −1 ; light green symbols) and STMW (DIC STMW ; µmol kg −1 ; dark green symbols) waters.(B) Timeseries of winter NAO (JFM; red symbols and line) and mixed layer depth (m; blue symbols and line).(C) Time-series of winter (JFM) DIC in surface (µmol kg −1 ; light green symbols and line) and STMW (µmol kg −1 ; dark green symbols and line) waters.(D) Time series of winter (JFM) DIC in surface (µmol kg −1 ; light green line)and STMW (µmol kg −1 ; dark green line) waters and winter NAO (JFM; red line).The surface and STMW DIC data are corrected for long-term trends and so represent anomalies relative to longterm trends.(E).Scatterplot of winter (JFM) NAO (red symbols) and DIC difference between STMW and surface (µmol kg −1 ; blue symbols) against mixed layer depth (m).(F).Scatterplot of winter (JFM) NAO against DIC difference between surface and STMW (µmol kg −1 ; green symbols).

Table 3 .
Correlation table of selected mean winter (JFM) variables.Within each cell, the first row is the correlation coefficient and second row is the p-value.For each correlation, N = 22-23.Only those correlations with P ≤ 0.1 are shown, with P ≤ 0.01 in bold.
periods.The trend and CO 2 sink in STMW is given in row 1 in each panel (δDIC STMW /t; black).In each panel, row 2 denotes potential uptake due to anthropogenic CO 2 (C ant or δDIC anthropogenic /t; red) row 3 denotes increase due to remineralization of organic carbon in STMW to CO 2 (i.e.δDIC biology /t; green); row 4 denotes increase (decrease) at the STMW outcrop (i.e.δDIC outcrop /t; blue); and row 5 denotes the potential change due to mixing of STMW with surface/seasonal thermocline waters (i.e.dDIC mixing /t).