Based on measurements from the WOCE/JGOFS global CO2 survey, the CLIVAR/CO2 Repeat Hydrography Program and the Canadian Line P survey, we have observed an average decrease of 0.34% yr−1 in the saturation state of surface seawater in the Pacific Ocean with respect to aragonite and calcite. The upward migrations of the aragonite and calcite saturation horizons, averaging about 1 to 2 m yr−1, are the direct result of the uptake of anthropogenic CO2 by the oceans and regional changes in circulation and biogeochemical processes. The shoaling of the saturation horizon is regionally variable, with more rapid shoaling in the South Pacific where there is a larger uptake of anthropogenic CO2. In some locations, particularly in the North Pacific Subtropical Gyre and in the California Current, the decadal changes in circulation can be the dominant factor in controlling the migration of the saturation horizon. If CO2 emissions continue as projected over the rest of this century, the resulting changes in the marine carbonate system would mean that many coral reef systems in the Pacific would no longer be able to sustain a sufficiently high rate of calcification to maintain the viability of these ecosystems as a whole, and these changes perhaps could seriously impact the thousands of marine species that depend on them for survival.
 Carbon dioxide is one of the most important of the “green-house” gases in the atmosphere, contributing to the heat balance of the earth as well as affecting the calcium carbonate (CaCO3) equilibrium in the oceans. As a result of the industrial and agricultural activities of humans since the beginning of the industrial era, atmospheric CO2 concentrations have increased about 40% [Prentice et al., 2001; Royal Society, 2005; Solomon et al., 2007; Sabine and Feely, 2007]. Atmospheric CO2 concentrations are now higher than has been experienced for at least the last 800,000 years [Keeling and Whorf, 2004; Lüthi et al., 2008]. The global oceans are the largest natural long-term reservoir for the excess CO2, currently absorbing 26% of the combined carbon released from deforestation and fossil fuel combustion up to the present and could absorb as much as 90% of the excess CO2 over the next several millennia [Archer et al., 1998; Canadell et al., 2007; Le Quéré et al., 2009; Sabine et al., 2011]. Seawater chemistry is now changing in response to continually rising atmospheric CO2 levels. For example, the mean surface ocean pH has decreased by about 0.1 units since the beginning of the industrial revolution [Caldeira and Wickett, 2003, 2005; Feely et al., 2004, 2009a; Orr et al., 2005]. If current carbon dioxide emission trends continue this process, commonly known as “ocean acidification,” will occur at rates and extents that have not been observed for tens of millions of years [Feely et al., 2004, 2009a; Kump et al., 2009]. A doubling of atmospheric carbon dioxide concentration from pre-industrial levels, which could occur in as little as 50 years, is predicted to correspond with an average sea surface pH decrease of about 0.25 [Caldeira and Wickett, 2005].
where Ωarag and Ωcal are calculated using the CO2SYS program developed by Lewis and Wallace . As atmospheric pCO2 increases and equilibrates with seawater, carbonate ion is consumed via a series of reactions:
where the reaction of CO2(aq) with H2O (3) leads to an initial increase in dissolved H2CO3 from the gas exchange process. These reactions are reversible and the thermodynamics of these reactions in seawater are well established [Millero et al., 2002, and references therein]. It is these reactions in combination with the slow circulation and primary production throughout the global oceans that control pH over timescales of hundreds to thousands of years. By the end of this century, ocean acidification could decrease surface ocean pH by as much as 0.4 pH units relative to pre-industrial values [Orr et al., 2005; Meehl et al., 2007; Joos et al., 2011]. The corresponding carbonate ion decrease in the surface waters would be approximately 50% [Kleypas et al., 2006; Feely et al., 2009b]. Thus, an increase in pCO2 and a corresponding reduction in CO32– concentration will result in a reduction of saturation state with respect to CaCO3 phases. In this paper, we document the decreases in the aragonite and calcite saturation state of the Pacific Ocean between the time of the WOCE/JGOFS Global CO2 Survey and CLIVAR/CO2 Repeat Hydrography Program. We will show how the changes in saturation state are affected by both the influx of anthropogenic CO2 as well as changes in overturning circulation over the time interval between the observations.
 For the CLIVAR/CO2 Repeat Hydrography cruises, samples were collected and analyzed for DIC, TA, oxygen, nutrient, tracer, and hydrographic data at repeat sections along the P02 (from Japan to the United States along 30°N; 2004), P16S (152°W from Tahiti to 71°S; 2005), P16N (along 152°W from Tahiti to Alaska; 2006), P6 (along 30°S from Australia to Chile; 2003), and P18 (along 110°W; 2007–2008 and 1994) transects (Figure 1). As a minimum, DIC and TA were measured on all the cruises. The CLIVAR/CO2 data quality was confirmed by daily analyses of Certified Reference Materials [Dickson, 2001; Dickson et al., 2007]. The consistency of the individual cruises was checked by comparing deepwater (>2000 m) values at stations that overlapped on P16S and P16N, and at the intersection of P16N and P02 [Sabine et al., 2008]. These quality checks suggest that the DIC data are accurate within ∼1 μmol kg−1 and the TA data are accurate within ∼3 μmol kg−1. The P16 pH measurements were made spectrophotometrically with an overall precision of ±0.0015 [Byrne et al., 2010]. The CLIVAR/CO2 Repeat Hydrography physical and chemical data were compared to the 1990s WOCE/JGOFS data along the two P16N and P02 sections by examining values on isopycnal surfaces in deep water. The only observed offsets were found in the 1994 P02 TA data that required an adjustment of +10 μmol kg−1.
2.3. Line P
 For the 2004 February Line P data [Miller et al., 2009], samples were collected and analyzed for DIC, TA, oxygen, nutrient, and hydrographic data along Line P (Figure 1). DIC and TA were measured employing the CLIVAR/CO2 methodology (Section 2.1 above), and were confirmed by daily analyses of Certified Reference Materials and secondary standards directly calibrated against the certified materials [Dickson, 2001; Dickson et al., 2007]. The consistency of the data were checked by comparing deepwater (>2000 m) values at stations that were close to the P16N section. These quality checks suggest that the DIC data are accurate within ∼2 μmol kg−1 and the TA data are accurate within ∼3 μmol kg−1.
3. Data Analysis
3.1. Total Change in Aragonite/Calcite Saturation State Between Cruises
 Total changes in aragonite and calcite saturation levels were calculated using the CO2SYS program developed by Lewis and Wallace . The in situ degree of seawater saturation with respect to aragonite and calcite calculated from equations (1) and (2), where the Ca+2 concentrations are estimated from salinity and carbonate ion concentrations, are calculated from the dissolved inorganic carbon (DIC) and total alkalinity (TA) data. The pressure effect on the solubility is estimated from the equation of Mucci  that includes the adjustments to the constants recommended by Millero . The overall uncertainties of aragonite and calcite saturation state are on the order of ±0.03 and ±0.05, respectively. The aragonite solubility calculations are in agreement with field experiments of the first instance of aragonite dissolution based on freshly collected pteropod shells placed into a spectrophotometer under conditions of ambient temperature and pressure [Feely et al., 1988]. The total changes in aragonite or calcite saturation state were calculated as the gridded and interpolated differences between saturation values estimated for each pair of repeat cruises.
3.2. Changes in Aragonite/Calcite Saturation State Due to Changes in Anthropogenic CO2 and Changes in Circulation and Mixing Processes
 The extended multiple linear regression (e-MLR) approach developed by Friis et al.  was used for this analysis. In this procedure, the observed DIC and TA data are fitted as a function of physical (e.g., temperature, salinity) and chemical (e.g., phosphate, nitrate, silicate) parameters. Multiple linear regression fits are determined for each cruise using the same set of independent physical and chemical parameters [Sabine et al., 2008]. The coefficients of these two fits are then subtracted, such that the resulting equation directly determines the net DIC and TA change between the two cruises. Using this method, much of the random variability in the independent parameter measurements is minimized for both cruises, and the propagation of errors that results from a particular independent parameter's inability to describe completely the dependent parameter are partially canceled out when the coefficients are subtracted [Friis et al., 2005; Sabine et al., 2008; Wanninkhof et al., 2010; Goodkin et al., 2011]. The determination of which parameters are selected for use is based on the statistical fits of the field data. Almost all studies to date have used salinity (S) and potential temperature (θ) to characterize the conservative characteristics. For our study, all repeat cruises were fit as a function of: S, θ, potential density, phosphate (P) and silicate (Si). Oxygen, or apparent oxygen utilization (AOU), was specifically not used for the fit in anticipation of using AOU to characterize the difference in the circulation changes as described below. The P16 and P02 lines were subdivided into three segments and fit with three independent e-MLR functions [Sabine et al., 2008]. The divisions were chosen by first fitting all of the data along a section with a single function and plotting the residuals as a function of latitude (P16 and P18) or longitude (P02). The residuals resulted in a pattern, which indicated that the tropical Pacific (15°S–15°N) had a different pattern of residuals, and that of P02 should be divided at 145°E and 140°W. To estimate the circulation effects, AOU was fit using the same e-MLR approach that was applied to the carbon data. The coefficients and standard errors for the AOU fits are given by Sabine et al. , employing a carbon to oxygen stoichiometric ratio of 117/170 [Anderson and Sarmiento, 1994]. Sabine et al.  determined that approximately 80% of the DIC change in the North Pacific over the last decade is the result of circulation/ventilation changes and that circulation effects resulted in almost no change in the South Pacific [Sabine et al., 2008, Figure 2]. The central core of the maximum AOU change appears to be associated with the 26.6 potential density surface in the subtropical North Pacific, consistent with previous studies [Emerson et al., 2001; Ono et al., 2001; Deutsch et al., 2006]. The change in aragonite/calcite saturation state that is due to the uptake of anthropogenic CO2 is then determined by subtracting this circulation and/or ventilation change in DIC from the total change in DIC and getting the anthropogenic CO2 change by difference. These new values are then used to estimate the change in saturation state that is due to the anthropogenic component. The technique is based on the assumption of constancy of the processes controlling the coefficients in the e-MLR equations. Significant changes in elemental ratios over time could lead to an underestimation of anthropogenic CO2 contribution [Wanninkhof et al., 2010].
4.1. The P16 South-North Transects (2005–06 Versus 1991)
 There is significant shoaling of the aragonite and calcite saturation horizons from south to north in the Pacific because of the higher DIC concentrations relative to TA at shallower depths in the northern hemisphere that result from enhanced upwelling at the equator, at 10°N, and north of about 40°N in the subarctic North Pacific (Figures 2 and 3, respectively). The 2005–06 CLIVAR/CO2 Repeat Hydrography cruise data show a general upward migration of the aragonite saturation horizon of about 1–2 m yr−1 along the entire cruise track and an average decrease in the overall aragonite saturation state of about 4.5% in near-surface waters over the 14-year interval between the two cruises. Most of the change in the aragonite saturation state, ranging from +0.3 to −0.5, occurred in the upper 600 m of the water column (Figure 2). The positive changes in saturation state are observed near frontal zones associated with the North Equatorial Current at about 12–14°N where mesoscale changes in salinity and temperature predominate [Fine et al., 2001]. The calcite saturation horizon (Ωcal = 1.0) rose from depths greater than 2800 m in the South Pacific and shoaled to depths less than 200 m between 40°N and 50°N (Figure 3). On average, the calcite saturation horizon in the Pacific shoaled about 1 m yr−1 from 1991 to 2006.
4.2. The P02 West-East Sections (2004 Versus 1994) in the North Pacific
 The west-to-east shoaling of the aragonite and calcite saturation horizons (Figures 4 and 5) is consistent with the shoaling of the TA concentrations shown in Feely et al. . This shoaling is the result of the deep ventilation in the western Pacific and anticyclonic circulation in the North Pacific. The 2004 P02 aragonite saturation horizon (Ωarag = 1.0) shoaled from depths of about 750 m near 140°E to 150 m near 122°W. From there, it deepened slightly to about 300 m near the North American coast. The data indicate a distinct upward migration of the saturation horizon along portions of the section. In particular, for the California Current region between 135°W and 120°W, the saturation horizon in the eastern subtropical Pacific west of the continental shelf has risen more than 100 m since the previous WOCE cruise in 1994. Most of the changes of the saturation state (ranging from +0.5 to −0.5 for aragonite and +0.8 to −0.8 for calcite) occurred in the upper 800 m of the water column, deepening toward the west (Figures 4 and 5). The decreases in saturation state occurred between 100 and 600 m and are located in regions where changes in ventilation and uptake of anthropogenic CO2 have major negative impacts on saturation state.
4.3. P18 South-North Transects Along 110°W (2007–2008 Versus 1994)
 The P18 sections show similar shoaling of the aragonite and calcite saturation horizons from south to north compared with the P16 sections (Figures 6 and 7). The 2007–2008 aragonite saturation depth shoaled from about 1000 m near 40°S to <250 m near 10°S, deepened to ∼300 m at the equator, and shoaled to ∼150 m near 10°N. The general pattern of aragonite and calcite saturation along the 2007–2008 P18 transect is generally consistent with previous results for 1994 along the same section (Figure 6). The 2007–2008 P18 cruise data show a general upward migration of the aragonite saturation horizon of about 1–2 m yr−1 along the cruise track and an average decrease in the overall aragonite saturation state of ∼4.5% in near-surface waters over the 14-year period between the two cruises. Most of the change in the aragonite saturation state, ranging from +0.2 to −0.8, occurred in the upper 500 m of the water column (Figure 6). The calcite saturation horizon rose from depths greater than 2800 m in the South Pacific and shoaled to about 550 m near 10°N in the North Pacific (Figure 7).
4.4. The P06 West-East Section (2003) in the South Pacific
 The distributions of temperature, salinity, DIC, and aragonite saturation for the P06 west-east 2003 section in the South Pacific are plotted in Figure 8. Since insufficient alkalinity data were collected for the previous WOCE section, it is not possible to make aragonite saturation state comparisons with the earlier data. Nevertheless, this is the first complete east-west aragonite saturation section for the South Pacific. As with the North Pacific, the aragonite saturation horizon shoals from west to east in the South Pacific, starting from depths around 1200 m near 160°W to 800 m near 80°W in the deeper waters. In the eastern South Pacific, there is a large mass of eastward increasing high-DIC, high-salinity, undersaturated water ranging in depths from about 100 to 600 m near the coast of South America. This extremely shallow undersaturated water is probably the result of the uptake of anthropogenic CO2 combined with high amounts of remineralized organic matter along the South American coast, leading to an unusual, heretofore unrecognized, ocean acidification site close to the coast. This site is similar to what has been observed off the west coast of Africa in the South Atlantic [Chung et al., 2003; Feely et al., 2004] and off the west coast of North America [Feely et al., 2008]. In those cases, the oxidation of organic matter augmented by the uptake of anthropogenic CO2 accounted for the observed local reduction in the aragonite saturation state.
4.5. Line P
 The Line P aragonite and calcite saturation sections for February 2004 are shown in Figures 9a and 9b. At that time the aragonite saturation horizon (Ωarag = 1.0) was shallowest (∼160 m) at the westernmost Station “P” From there, it deepened slightly to about 190 m near the North American coast off Vancouver Island. The calcite saturation data also suggested a distinct shoaling of the Ωcal = 1 horizon to a depth of about 200 m near station “P” and a gradual deepening of the saturation horizon toward the east. This is consistent with the predominance of downwelling of water properties and carbon system parameters along the coast during the winter months [Ianson and Allen, 2002; Ianson et al., 2009]. The shallow depths of the aragonite and calcite saturation horizons occur in regions where uptake of anthropogenic CO2 produces a significant shoaling effect on saturation state (see Section 5 below).
5.1. Estimates of the Relative Role of Anthropogenic CO2 and Circulation Changes on the Vertical Migration of the Aragonite/Calcite Saturation Horizons
 The repeat sections allow us to determine the changes in saturation state and upward migration of the saturation horizons over the time intervals of the cruises. These changes can be caused by: 1) uptake of anthropogenic CO2; 2) changes in circulation and/or ventilation; and 3) changes in biogeochemical processes. The calculated change in aragonite saturation state between the decadal cruises for each of the transects, which is based on the increase in anthropogenic CO2 as calculated by Sabine et al. , is shown in Figure 10. The decrease of the saturation state is regionally variable, with much deeper changes in the South Pacific and western North Pacific than the eastern North Pacific. This is largely due to the fact that the South Pacific and western North Pacific take up more anthropogenic CO2 than the eastern North Pacific [Sabine et al., 2004, 2008]. The situation in the North Pacific is complicated by the fact that in the subpolar North Pacific most of the DIC increase was caused by a decrease in the overturning circulation due to reduced winds since the 1970s, causing an increase in apparent oxygen utilization (AOU) rather than uptake of anthropogenic CO2 [McPhaden and Zhang, 2002; Deutsch et al., 2006; Mecking et al., 2008; Sabine et al., 2008; Sabine and Tanhua, 2010].
 The average shoaling rate of the Ωarag = 1.0 horizon and the average change in saturation state in the upper 100 m due to uptake of anthropogenic CO2 are given in Table 1. The results indicate an upward migration of the aragonite saturation horizon on the order of 1–2 m yr−1, with higher shoaling rates in the South Pacific than in the North Pacific because of the higher anthropogenic CO2 uptake in the South Pacific. These rates are roughly consistent with the model estimates for the Pacific Ocean given in Orr et al.  for an IPCC IS92a “business as usual” scenario.
Table 1. Average Shoaling of the Aragonite Saturation Horizon (Ωarag = 1.0) and Decrease in the Aragonite Saturation State in the Upper 100 m in the South and North Pacific Due to the Uptake of Anthropogenic CO2
 The far eastern Pacific data indicate an average upward migration of the aragonite saturation horizon of more than 5 m yr−1 in the California Current (Figure 4). This large and unexpected change is primarily due to a significant change in the circulation and water mass properties of the California Current since 1998. A careful analysis of the CalCOFI data [Di Lorenzo et al., 2005] noted large-scale cooling and freshening of the water from 50 to 200 m within the California Current. The authors interpreted these observations as indicating an enhancement of the southward advection of cool, lower-salinity subarctic water coming from the subarctic North Pacific. This interpretation is consistent with our observations of increased amounts of cool, CO2-rich water at the same depths in the 2004 data set compared with the 1994 data. Physical and chemical changes of the water mass properties in the California Current were apparently the major factors controlling the upward migration of the aragonite saturation horizon and probably played a significant role in controlling the seasonal upwelling of corrosive “ocean-acidified” water onto the continental shelf [Feely et al., 2008]. These results suggest that large-scale changes in circulation can be as important as, or in some cases, more important than, the direct effects of anthropogenic CO2 uptake in affecting the location of corrosive water in some parts of the eastern Pacific. More detailed information on the temporal variability of the physical and chemical properties of the California Current is required before we can accurately predict how these long-term changes will affect our coastal ecosystems. In particular, recent modeling of primary production and nitrate transport processes in the California Current ecosystem suggests that increased nitrate supply and upwelling of lower pH source waters resulting from increased stratification of the North Pacific will cause an enhanced intensification of the acidification in this region over the next century [Rykaczewski and Dunne, 2010].
5.2. Potential Impacts of the Changes in Aragonite and Calcite Saturation State in the Pacific Ocean
 Over the past 250 years, since the beginning of the industrial revolution, there has been about a 16% decrease in aragonite and calcite saturation state in the Pacific Ocean. From repeat oceanographic surveys, we have observed an average 0.34% yr−1 decrease in the saturation state of surface seawater with respect to aragonite and calcite over a 14-year period. This has caused an upward migration of the aragonite and calcite saturation horizons toward the ocean surface on the order of 1–2 m yr−1. These changes are the result of the uptake of anthropogenic CO2 by the oceans, as well as other smaller scale regional changes in circulation over decadal time scales. If CO2 emissions continue as projected out to the end this century, the resulting changes in the marine carbonate system would mean that many coral reef systems in the Pacific would probably no longer be able to maintain the necessary rate of calcification required to sustain their vitality.
 This work was sponsored the National Oceanic and Atmospheric Administration and the National Science Foundation. We specifically acknowledge Joel Levy of the NOAA Climate Program Office Ocean Climate Observation Program, and Eric Itsweire and Don Rice of the National Science Foundation for their support. We also want to thank all the officers, crew, and scientists of the WOCE/JGOFS, CLIVAR/CO2 Repeat Hydrography, and Line P cruises in the Pacific for providing this valuable data to the ocean community.