Detecting anthropogenic carbon dioxide uptake and ocean acidification in the North Atlantic Ocean

Abstract. Fossil fuel use, cement manufacture and land-use changes are the primary sources of anthropogenic carbon dioxide (CO2) to the atmosphere, with the ocean absorbing approximately 30% (Sabine et al., 2004). Ocean uptake and chemical equilibration of anthropogenic CO2 with seawater results in a gradual reduction in seawater pH and saturation states (Ω) for calcium carbonate (CaCO3) minerals in a process termed ocean acidification. Assessing the present and future impact of ocean acidification on marine ecosystems requires detection of the multi-decadal rate of change across ocean basins and at ocean time-series sites. Here, we show the longest continuous record of ocean CO2 changes and ocean acidification in the North Atlantic subtropical gyre near Bermuda from 1983–2011. Dissolved inorganic carbon (DIC) and partial pressure of CO2 (pCO2) increased in surface seawater by ~40 μmol kg−1 and ~50 μatm (~20%), respectively. Increasing Revelle factor (β) values imply that the capacity of North Atlantic surface waters to absorb CO2 has also diminished. As indicators of ocean acidification, seawater pH decreased by ~0.05 (0.0017 yr−1) and ω values by ~7–8%. Such data provide critically needed multi-decadal information for assessing the North Atlantic Ocean CO2 sink and the pH changes that determine marine ecosystem responses to ocean acidification.


Introduction
The emissions of anthropogenic CO 2 to the atmosphere due to fossil fuel use, cement manufacture and land-use changes (Houghton, 2008) have increased rapidly over the last decade (Friedlingstein et al., 2010).While anthropogenic CO 2 accumulates in the atmosphere, it is also taken up by the terrestrial biosphere and oceans.The annual global ocean uptake is estimated at ∼ 1.4 to 2.5 Pg C yr −1 (e.g., Takahashi et al., 2002;Manning and Keeling, 2006;Takahashi et al., 2009;McKinley et al., 2011; Pg = 10 15 g), with annual rates of CO 2 uptake increasing with time (Le Qu ér é et al. is estimated at ∼ 120-140 Pg C (Sabine et al., 2004;Khatiwala et al., 2009).Ocean uptake of anthropogenic CO 2 and seawater chemistry changes such as reduction in seawater pH and saturation states for calcium carbonate (CaCO 3 ) minerals such as calcite (Ω calcite ) and aragonite (Ω aragonite ) is termed ocean acidification (e.g., Calderia andWickett, 2003, 2005;Orr et al., 2005;Doney et al., 2009;Feely et al., 2009), and has likely problematic consequences for marine organisms and ecosystems that are, as yet, poorly understood.Assessments of rates of ocean CO 2 uptake over multi-decadal timeperiods, changes in the capacity of the ocean to absorb CO 2 and ocean acidification impacts are important for predicting future climate change and marine ecosystem responses (e.g., Fabry et al., 2009).These ocean carbon cycle and ocean acidification data provide critically needed observational tests of global coupled ocean-atmosphere models, and allow attribution of changes to both anthropogenic and natural causes (e.g., Le Qu ér é et al., 2010).Repeat hydrographic sections provide a means of quantifying basinwide ocean uptake of anthropogenic CO 2 (e.g., Friis et al., 2005;Brown et al., 2010;Wanninkhof et al., 2010).Alternatively, higher frequency observations of the changes in seawater chemistry due to uptake of anthropogenic CO 2 have been collected at a few ocean time-series near various islands, including near Bermuda (Bates, 2001(Bates, , 2007;;Bates and Peters, 2007), Hawaii (e.g., Dore et al., 2003;Brix et al., 2004;Dore et al., 2009), the Canary Islands (Santana-Casiano et al., 2007;Gonzalez-Davila et al., 2010) and Iceland (Olafsson et al., 2010).Here, we examine direct observations of the seawater carbonate chemistry changes resulting from ocean uptake of CO 2 over the last 30 yr in the subtropical gyre of the North Atlantic Ocean near Bermuda from 1983-2011.Seawater samples were collected from two time-series sites near Bermuda (i.e., Bermuda Atlantic Time-series Study; BATS; 31 • 40 N, 64 • 10 W (Michaels and Knap, 1998;Steinberg et al., 2001); and Hydrostation S, 32 • 10 N, 64 • 30 W; Keeling, 1993).Samples for dissolved inorganic carbon (DIC) and total alkalinity (TA) were analyzed (Bates et al., 1996a;Dickson et al., 2007), multi-decadal checks on accuracy of data are made and computations of other components of the seawater carbonate system computed Introduction

Conclusions References
Tables Figures

Back Close
Full  (Robbins et al., 2010).These data provide directly observed indicators of ocean acidification and include pH, carbonate ion concentration ([CO 2− 3 ]) and the saturation state (Ω) for calcium carbonate (CaCO 3 ) minerals.Finally, these data are combined with a few earlier measurements in the North Atlantic Ocean (from the Geochemical Ocean Section Study, GEOSECS (Kroopnick et al., 1972) and Transient Tracers in the Ocean, TTO projects; Brewer et al., 1985) extending these trends over the past 40 years.

Seawater carbonate chemistry sampling at BATS
A time-series of observations of seawater carbonate chemistry observations in the upper ocean have been collected in the subtropical gyre of the North Atlantic Ocean near Bermuda since 1983 at the Bermuda Atlantic Time-series Study (BATS) and Hydrostation S sites (Bates, 2007;Bates and Peters, 2007).The combined ocean timeseries data represent monthly water column sampling for DIC and total alkalinity (TA) at BATS, 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, b;Bates, 2001).Addtiional surface samples were collected at Hydrostation S and analyzed for DIC and TA using manometric and potentiometric methods, respectively, at Scripps Institution of Oceanography (Keeling, 1993).Here, DIC is defined as (Dickson et al., 2007): Introduction

Conclusions References
Tables Figures

Back Close
Full

Sampling frequency at BATS
The sampling frequency of the combined dataset from Hydrostation S and BATS was not uniform in time.In the 1980s, samples were collected 9-12 times a year, while since 1992, sampling increased to 14-15 times a year (Fig. 1).The increase in sampling since the middle 1990s was due to supplemental BATS bloom cruises (1 to 4 in number) conducted during the January to April period in addition to BATS core cruises.The increase in sampling frequency over time weights the time-series to springtime which has to be addressed when determining non-seasonally aliased trends.

Sampling methods
At Hydrostation S, samples were collected into 500 ml Pyrex bottles, poisoned with HgCl 2 , sealed with ground glass stoppers and then shipped to Scripps Institution of Oceanography (SIO) for analysis (Keeling, 1993;Brix et al., 2004).Storage time for samples before analysis ranged from a few months to several years.Similar sampling protocols were established at BIOS for sampling at the BATS (Bates et al., 1996a,b;Bates, 2001) but in the early 2000s, smaller Pyrex bottles (∼ 350 ml) were used.Samples were typically analyzed within a few months of collection.

Analytical methods
Hydrostation S samples were analyzed for DIC at SIO using manometric methods (Keeling, 1993) and potentiometric titration methods were used for determination of TA (Keeling, 1993).Analytical precision for both DIC and TA at SIO was typically < 0.2 %.At BIOS, DIC was determined using coulometric methods with a SOMMA system (Johnson et al., 1993;Bates et al., 1996a;Dickson et al., 2007).During the Introduction

Conclusions References
Tables Figures

Back Close
Full first two years of sampling, DIC samples were analyzed at WHOI (e.g., BATS cruise 1 to 21), and subsequently at BIOS.DIC measurements were calibrated with known volumes of pure CO 2 gas while certified reference materials (CRM's; Dickson et al., 2007) were routinely analyzed each day of analysis.Potentiometric titration methods were also used for determination of TA at BIOS (Bates et al., 1996b).At the beginning of the 1990's, a manual alkalinity titrator was used for determination of TA.This was replaced by an automated VINDTA 2S (Versatile Instrument for the Determination of Titration Alkalinity) in the early 2000s.For both manual and automated TA systems, 15-20 titration points past the carbonic acid end point were determined for each sample, with TA computed from these titration data using non-linear least squares methods (Dickson et al., 2007).Surface samples of Sargasso Sea water was also analyzed each day prior to sample analyses and CRM's were used routinely to calibrate the TA measurements.Analytical precision for both DIC and TA at BIOS was typically < 0.2 % for within bottle and between bottle replicate analyses of more than 2000 samples.

Comparison of replicate samples analyzed at BIOS and SIO
From 1989-2010, selected replicate surface DIC and TA samples at BATS were analyzed independently at BIOS and Scripps using different analytical techniques.Replicate surface and 10 m depth samples were collected on BATS cruises over a period of 20 yr from 1990 to 2010.Comparison of DIC samples analyzed independently at BIOS and SIO indicate that the mean difference was 1.24 ± 3.35 µmol kg −1 (Fig. 2 , 1993) was used as water-column samples were not collected for the period July-December 1990.For BATS cruises 21-27 and 30-36, TA samples analyzed at SIO were used (Keeling, 1993).For Hydrostation S cruise 605698101, and BATS cruises 1-3, 4-20, 79A, 90A, 91A, 100A, 101A, 102A, 113A, 114A, 138, 185A, and 186A, TA was calculated from salinity (with an error of ∼ 2.8 µmol kg −1 ; Bates et al., 1996a).Exclusion of these samples from trend analysis did not change results significantly.The locations of Hydrostation S and BATS are separated in space by ∼ 50 km, but analysis of both Hydrostation S and BATS data at SIO suggests that there is no statistical difference between the two locations.Surface data are shown here but no statistical diffference was found if mean DIC and TA were determined for different depth intervals in the mixed layer (using a 0.5 • C temperature criterion to define mixed layer depth).

Computation of seawater carbonate chemistry
Seawater pCO 2 , pH, [CO 2− 3 ], mineral saturation states for calcite (Ω calcite ) and aragonite (Ω aragonite ), and the Revelle factor (β) were computed from DIC, TA, temperature and salinity data using the program CO2calc (Robbins et al., 2010).Carbonic acid dissociation constants (i.e., pK 1 and pK 2 ) of Mehrbach et al. (1973), as refit by Dickson and Millero (1987) were used for the computation, as well as dissociation constants for HSO − 4 (Dickson, 1990).GEOSECS and Transient Tracers in the Ocean (TTO) DIC, TA, temperature and salinity data were taken from the CDIAC site (http://cidiac.ornl.gov).We used the same dissociation constants to compute surface pCO 2 , pH and Ω aragonite from GEOSECS and Transient Tracers in the Ocean (TTO) DIC, TA, temperature and salinity data.GEOECS and TTO data available through WAVES (Web-Accessible Visualization and Extraction System) that allows access to discrete data that are part of Introduction

Conclusions References
Tables Figures

Back Close
Full the Global Ocean Data Analysis Project (GLODAP) and Carbon in the North Atlantic (CARINA) databases.

Computation of seawater carbonate chemistry
The computation of seawater pCO 2 was compared to direct measurements of pCO 2 from the R/V Atlantic Explorer (AE) and Bermuda Testbed Mooring (BTM) from 2005 to 2010 (Fig. 3).During this period, 42 direct comparisons were made.Observed seawater pCO 2 was collected from BIOS ship R/V Atlantic Explorer at the BATS site (31 • 43 N, 64 • 10 W) using a pCO 2 system calibrated with 4 CO 2 -in-air standards similar to previous methods used at BIOS (Bates et al., 1998).Seawater pCO 2 measurements were also made using a pCO 2 sensor attached to the Bermuda Testbed Mooring (BTM; 31 • 41.77 N, 64 • 10.52 W; Dickey et al., 2009, http://cidiac.ornl.gov).This system was calibrated with one CO 2 -in-air standards and deployed from 2005 to 2007.Both seawater pCO 2 systems were calibrated with CO 2 -in-air standards calibrated at CMDL (NOAA).At the time of sampling for DIC and TA at the BATS site (i.e., time that Niskin sampler was tripped and filled), mean observed R/V Atlantic Explorer and BTM seawater pCO 2 were averaged over 1 h with the mid-point exactly contemporaneous with the time of sampling.Compared to the observed R/V Atlantic Explorer and BTM seawater pCO 2 data, computed pCO 2 from DIC and TA was lower by a mean of −4.7 ± 13.6 µatm.However, there were slight differences in sampling temperatures, and thus, R/V Atlantic Explorer and BTM seawater pCO 2 data were corrected to the SST's at time of DIC/TA sampling at BATS.The mean difference between temperature corrected seawater pCO 2 computed at BATS, and directly observed from the R/V Atlantic Explorer and BTM was small (−3.1 ± 10.8 µatm).No systematic bias in the computation of seawater pCO 2 was apparent.Introduction

Conclusions References
Tables Figures

Back Close
Full

Trend analysis and statistics
Trend analyses of the time-series of surface temperature and salinity, seawater carbonate chemistry (DIC, TA, pCO 2 , Revelle factor, β) and ocean acidification indicators (pH, [CO 2− 3 ], Ω calcite and Ω aragonite ) was conducted.Here, trend analysis of salinity normalized DIC (nDIC) and TA (nTA) data were also made in order to account for local evaporation and precipitation changes.These data were normalized to a salinity of 36.6, as this represents the mean salinity observed at the BATS site (Bates, 2007).Trend analysis was performed with observed data (Table 1) and seasonally detrended data (Table 2).Regression statistics given were slope, error, r 2 , p-value and n.Trends with p-values greater than 0.01 were deemed statistically not significant at the 99 % confidence level.

Seasonal detrending of data
Trends analyses with observed data exhibits seasonal aliasing due to sampling weighting to spring conditions.For example, sea surface temperature (SST) apparently cooled during the 1983-2011 period at a rate of −0.075 decade −1 , but, this largely reflects a sampling bias that is weighted to springtime conditions.To account for seasonal weighting, the data were also seasonally detrended.Seasonal detrending of the BATS/Hydrostation S data was accomplished by binning data into appropriate month, with mean values calculated from 2 or more cruises conducted within a representative month each year.This provides a uniform timestep of approximately 1 month (i.e., 365 or 366 days/12) throughout the time-series, thereby removing any potentially seasonal weighting especially to springtime conditions.The mean seasonality of all parameters are shown in Fig. 4. Second, a mean and standard deviation is then determined each month for the 1983-2011 (Fig. 4), and anomalies computed from monthly data minus mean values.Trends and regression statistics for seasonally detrended data are given in Table 2. Removing any potentially seasonal weighting allowed trend analysis of data that had any non-temporal uniformity reduced as much as possible.The caveat to this Introduction

Conclusions References
Tables Figures

Back Close
Full approach and other approaches (e.g., using a 12 month harmonic fit to the data) is that energy may be lost or gained from this time-series data.

Seawater carbonate chemistry changes in surface waters
The time-series of surface temperature and salinity, seawater carbonate chemistry (DIC, TA, pCO 2 , Revelle factor, β) and ocean acidification indicators (pH, [CO 2− 3 ], Ω calcite and Ω aragonite ) off Bermuda in the North Atlantic Ocean are shown in Fig. 5.However, these data are subject to seasonal weighting due to additonal sampling in springtime and thus, seasonal detrended data are shown in Fig. 6.Trends and regression statistics for seasonally detrended data are given in Table 2.

Warming and salinity increases in surface waters
There is evidence for warming and increased salinity of surface waters observed in the North Atlantic subtropical gyre near Bermuda.Time series analysis of data at BATS reveal significant long-term trends in the temperature and salinity of the surface ocean increasing at rates of ∼ 0.011 ± 0.006 • C yr −1 (∼ 0.11 • C decade −1 ) and 0.0054 ± 0.0001 yr −1 , respectively (Table 2; Fig. 6).However, it should be noted that the surface temperature trend is not statistically significant at the 95 % level (p-value > 0.05), but similar temperature trends that are statistically significant have been observed within the upper 400 m at BATS and Hydrostation S over the last 55 yr with temperature and salinity increasing at rates of ∼ 0.01 • C yr −1 and 0.002 yr −1 , respectively (Joyce et al., 1999).Similar long-term changes in surface temperatures have been observed across the subtropical gyre of the North Atlantic Ocean (Grist et al., 2010)  ), that appears related to coordinated changes in freshwater fluxes and modes of climate variability such as the North Atlantic Oscillation (Hurrell and Deser, 2010).

Trends in seawater carbonate chemistry of surface waters
Our observations near Bermuda also show multi-decadal changes in seawater carbonate chemistry in response to anthropogenic CO 2 uptake by surface waters.Direct observations of DIC, computed seawater pCO 2 and Revelle factor (β) exhibit significant increases over the last 3 decades (Fig. 6; Table 2).Surface DIC and salinity normalized DIC (i.e., nDIC) have increased by 1.39 ± 0.06 and 1.08 ± 0.06 µmol kg −1 yr −1 , respectively, a change of nearly 40 µmol kg −1 , or ∼ 2 % from 1983 to 2011.Similar changes were observed previously at BATS (Bates, 2007) and attributed to the uptake of anthropogenic CO 2 from the atmosphere (Gruber et al., 2002;Bates et al., 2002).Similar changes in DIC over shorter timescales have been observed elsewhere off Hawaii (Dore et al., 2003(Dore et al., , 2009) ) and the Canary Islands (Santana-Casiano et al., 2007;Gonzalez-Davila et al., 2010).For example, total alkalinity, TA increased slightly by 0.48 ± 0.07 µmol kg −1 yr −1 , attributed here to gradual increase in the salinity of surface subtropical gyre waters.Salinity normalized TA (nTA) increased slightly at a rate of 0.10 ± 0.03 µmol kg −1 yr −1 but since this trend was not statistically significant, there was no definitive evidence that nTA has changed during the last 3 decades (Table 2).

Changes in seawater pCO 2 with time and its attribution
Surface seawater pCO 2 exhibited significant increases over the last 3 decades, increasing at a rate of 1.80 ± 0.09 µatm kg −1 yr −1 and representing an increase of nearly 55 µatm or 20 % from 1983 to 2011 (Fig. 6; Table 2).The increasing trend in seawater pCO 2 can be attributed to changes in DIC, TA, temperature and salinity (Fig. 7).The Introduction

Conclusions References
Tables Figures

Back Close
Full trend in observed DIC would increase seawater pCO 2 by +122 %, but this effect is counteracted by a small increase in TA that decreases pCO 2 (−33 %).Trends in temperature (+8 %) and salinity (+3 %) have minor impacts on calculated seawater pCO 2 .Similar trends and attribution of trends have been shown for observed pCO 2 across the North Atlantic in the seasonally stratified subtropical gyre (Bates, 2011) where the increase in seawater pCO 2 can be attributed to non-temperature effects (i.e., sum of DIC, TA and salinity changes).Over shorter timescales, similar trends in surface seawater pCO 2 have been observed elsewhere off Hawaii (Dore et al., 2003(Dore et al., , 2009) ) and the Canary Islands (Santana-Casiano et al., 2007;Gonzalez-Davila et al., 2010).

Changes in the ocean CO 2 sink?
These observations infer that the ocean CO 2 sink in the subtropical gyre has not changed significantly over the last three decades.Similar to previous observations at BATS, seawater pCO 2 has increased at a comparable rate to atmospheric pCO 2 (Bates, 2007;Takahashi et al., 2009; Table 2, 1.72 ± 0.01 µatm kg −1 yr −1 ) from 1983 to present.These trends indicate that the driving force for air-sea CO 2 gas exchange (i.e., ∆pCO 2 ; pCO 2 difference between atmosphere and seawater) has not changed significantly over the last 3 decades.One might therefore think that the ocean CO 2 sink near Bermuda has not changed significantly over time.However, it should be noted that windspeed also contributes in addition to ∆pCO 2 to the magnitude of air-sea CO 2 gas exchange.There is some evidence that the annual CO 2 sink has increased slightly due to higher annual windspeeds in the 2000's observed near Bermuda (Bates, 2007) perhaps in response to shift in the winter North Atlantic Oscillation from positive to neutral/negative over the last 2 decades (Hurrell and Deser, 2010;Bates, 2011).Other studies have also shown that with observations conducted over time periods longer than 2 decades, seawater pCO 2 has increased at the same rate as the atmosphere (Bates et al., 2002;McKinley et al., 2011) with longer term observations smoothing out shorter term variability.Such studies suggest that analysis of the CO 2 sink or source status using data that has a relatively short duration (<10 yr) may not be sufficient to 1000 Introduction

Conclusions References
Tables Figures

Back Close
Full determine whether the North Atlantic Ocean CO 2 sink has decreased over time (e.g., Schuster and Watson, 2007;Schuster et al., 2009;Watson et al., 2009).Such assessments may require a longer term view especially in light of this and other studies (McKinley et al., 2011); and especially when deciphering anthropogenic secular trends from natural variability imparted by such phenomena as the NAO (Joyce et al., 1999;Gruber et al., 2002;Hurrell and Deser, 2010) and Atlantic Multidecadal Oscillation (McKinley et al., 2011).In summary, it appears that ∆pCO 2 values have not changed significantly over the last 30 yr, and there is little evidence of any change in the CO 2 sink in the subtropical gyre of the North Atlantic near Bermuda.
In addition, over the 1983-2011 period, the Revelle factor (β) has increased at a rate of 0.0137±0.0008yr −1 or ∼ 0.41 over the last 3 decades.This indicates that the buffer capacity of subtropical gyre surface waters to absorb CO 2 has gradually reduced over time, confirming previous model studies that predict an increasing trend in β at BATS and for the North Atlantic Ocean in response to the ocean uptake of anthropogenic CO 2 from the atmosphere (Thomas et al., 2007).

The signal of ocean acidification in the North Atlantic Ocean
The BATS/Hydrostation S time-series data allow direct detection of the signal of ocean acidification in surface waters of the North Atlantic.The uptake of anthropogenic CO 2 from the atmosphere by the ocean changes seawater chemistry through chemical equilibrium of CO 2 with seawater.Dissolved CO 2 forms a weak acid and the pH and [CO 2− 3 ] decrease as seawater absorbs CO 2 , a process termed ocean acidification (Caldeira andWickett, 2003, 2005).The effects of ocean acidification are potentially far-reaching in the global ocean, particularly for organisms that secrete CaCO 3 skeletons, tests or shells and for marine ecosystems where calcification and pH controls on biogeochemical processes are important factors (e.g., Fabry et al., 2009).present.At BATS, the mean pH is 8.094 with a range of 8.00-8.18(Fig. 5).
The BATS site near Bermuda constitutes the longest time-series record of ocean acidification anywhere in the global ocean (Fig. 6; Bates, 2007).Trend analysis shows that the primary indicator of ocean acidification, seawater pH, has decreased at a rate of −0.0017 ± 0.0001 yr −1 , a total decline in seawater pH of ∼ 0.05 over the past 3 decades (Fig. 6; Table 2).This represents a ∼ 12 % increase in hydrogen ion concentration since 1983.Other indicators of ocean acidification at BATS such as [CO 2− 3 ], Ω calcite and Ω aragonite have also decreased at a rate of −0.58 ± 0.041 µmol kg −1 yr −1 , −0.0141±0.0009yr −1 , and −0.0091±0.0006yr −1 , respectively (Table 2).Trend analysis at BATS, and other published trends for three other ocean time-series sites indicate that surface seawater pH has decreased at a rate of −0.0014 to −0.0019 yr −1 (e.g., Bates and Peters, 2007;Bates, 2007;Dore et al., 2009;Gonzalez-Davila et al., 2010;Byrne et al., 2010).

Conclusions
The long term ocean observations near Bermuda constitute the longest running timeseries of changes in seawater carbonate chemistry due to the uptake of anthropogenic CO 2 and resulting ocean acidification impacts.Such records and those from other time-series sites (Dore et al., 2009;Gonzalez-Davila et al., 2010;Olafsson et al., 2010) provide critically needed data showing that such changes are due to anthropogenic CO 2 release and absorption by the global ocean and to test ocean-atmosphere models (e.g., Bates, 2007;McKinley et al., 2011)  .Combined with BATS/Hydrostation S data, this extends the time-series records of computed surface pCO 2 , pH and Ω aragonite back to the early 1970s (Fig. 8; Supplementary Information).These data were sampled within 200 km of Bermuda (but not at BATS/Hydrostation) and both surface pCO 2 , pH and Ω aragonite data from GEOSECS and TTO fall close to the regression lines for these parameters (from Table 1, Fig. 2).
Thus, the trends established at BATS/Hydrostation S appear to extend back to the early 1970s, constituting a nearly continuous 40 yr record of changing seawater carbonate chemistry and ocean acidification indicators.Introduction

Conclusions References
Tables Figures

Salinity
, 2009).The cumulative total global ocean uptake of anthropogenic CO 2 since pre-industrial time Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | principal components of seawater TA(Dickson et al., 2007).
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | indicating a system level change in the marine environment.The observed increase in surface salinity near Bermuda is also evident across the North Atlantic Ocean subtropical gyre (Zhang et al., Discussion Paper | Discussion Paper | Discussion Paper | 2011 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Relevant indicators of ocean acidification include seawater pH, but also [CO 2Discussion Paper | Discussion Paper | Discussion Paper | that are used for predictive climatechange purposes.Before regular sampling began at the Hydrostation S in 1983 and later at the BATS site in 1988, there was occasional sampling in the North Atlantic subtropical gyre near Bermuda through the GEOSECS and Transient Tracers in the Ocean (TTO) projects.Here, we compute surface pCO 2 , pH and Ω aragonite from surface GEOSECS and TTO DIC and total alkalinity data(Kroopnick et al., 1972;Brewer et al., Discussion Paper | Discussion Paper | Discussion Paper | 1985) Discussion Paper | Discussion Paper | Discussion Paper | doi:10.1126/science.1177394,2009.Zhang, L. P., Wu, L. X., and Zhang, J. X.: Freshwater loss/salinification: Simulated response to recent freshwater flux change over the Gulf Stream and it's extension: coupled oceanatmosphere adjustment and Atlantic-Pacific teleconnection, J. Climate, 24(15), 3971-3988, doi:10.1175/2011JCL14020.1DiscussionPaper | Discussion Paper | Discussion Paper | Fig. 1

Fig. 1 .
Fig. 1.Sampling frequency or number of cruises each year when seawater carbonate chemistry and other parameters were collected.From September 1988, only BATS cruises are shown with Hydrostation S cruise beforehand.The * symbol denotes years when the full year was not sampled.

Fig. 7 .
Fig. 7. Long-term trend in seawater pCO 2 (purple) and its attribution to long-term changes in salinity, temperature, TA and DIC (green).The relative attribution of change was computed using CO2calc (Robbins et al., 2010) using mean temperature, salinity, TA and DIC values observed at BATS.Mean values and trends taken from Table 2.

Fig. 8 .
Figure 8 Fig. 8. Time-series of atmospheric and ocean pCO 2 , pH and aragonite saturation states.(A) time-series of atmospheric pCO 2 (ppm) from Mauna Loa, Hawaii (red line), and Bermuda (pink symbol), and surface ocean seawater pCO 2 (µatm) at the Bermuda Atlantic Time-series Study (BATS) site off Bermuda.Observed (grey) and seasonally detrended (purple) surface ocean seawater pCO 2 levels are shown.Earlier seawater data from the GEOSECS and TTO expeditions in the North Atlantic Ocean are also shown in this and following panels.(B) time-series of surface ocean seawater pH at the BATS site off Bermuda.Observed (grey) and seasonally detrended (orange) seawater pH are shown.(C) time-series of surface ocean aragonite saturation state (Ω aragonite ) for calcium carbonate at the BATS site off Bermuda.Observed (purple) and seasonally detrended (purple line) seawater Ω aragonite are shown.

2.6 Compilation of a combined BATS/Hydrostation S record
.