Dissolved inorganic radiocarbon in the North Paciﬁc Ocean and Sargasso Sea

We present radiocarbon measurements of dissolved inorganic carbon (DIC) in depth proﬁles from reoccupations of our central North Paciﬁc Ocean and Sargasso Sea (SS) sites. From 1985 to 1999, an increase in D 14 C values of 6–17 % was measured between 1150 and 2400 m depth in the North Central Paciﬁc. Natural changes in deep circulation are likely responsible for variability of D 14 C signatures in the deep ocean, though we cannot rule out the presence of bomb 14 C at this depth range. Bomb 14 C had increased in the deep SS from 1989 to 2000; this is the result of southward transport of North Atlantic Deep Water (NADW) to this site.


Introduction
Oceanic dissolved inorganic carbon (DIC) is the largest pool of exchangeable carbon on Earth (36,000 GtC). It is important to know the timescales of mixing and ventilation of deep ocean waters because most of the excess CO 2 produced by fossil fuel and biomass burning will eventually be stored in the oceans. Early radiocarbon measurements in seawater revealed several hundred-year transit times of deep water masses in the world's oceans (Bien et al., 1963;Broecker et al., 1960;Stuiver et al., 1983). Global data sets amassed during programs such as the Geochemical Ocean Sections Study (GEOSECS) and the World Ocean Circulation Experiment (WOCE) have been used for quantitative studies of the oceanic carbon cycle (Key, 1996;Stuiver et al., 1983).
Since the production of bomb 14 C in the late 1950s and early 1960s, it has been possible to measure short-term exchange of carbon, e.g. the transfer of CO 2 across the air-sea interface. The level of bomb 14 C in a given carbon pool is a reflection of the turnover time of the carbon with respect to exchange with the atmosphere. By the 1970s, bomb 14 C was detected to depths of several hundred meters in the non-polar oceans Stuiver and Ostlund, 1980). The deep northern North Atlantic contained bomb 14 C throughout the entire water column, introduced during North Atlantic Deep Water (NADW) formation. By 1991, penetration of bomb 14 C extended to 1000 m depth in the Pacific (Key, 1997;Stuiver et al., 1996).
The D 14 C data reported here show an increase in values in the deep waters of both the North-central Pacific (NCP) and the Sargasso Sea (SS) . Change in circulation is the likely mechanism responsible for the higher D 14 C values in the upper part of the deep NCP. Remineralization of bomb-laden particulate organic matter from the surface is unlikely to have caused a measurable increase in the D 14 C values of DIC in the deep North Pacific or North Atlantic.

Methods
Water samples were collected from the NCP site (311N, 1591W, bottom depth 5820 m) on the Avon cruise from 26 May to 12 June 1999 and from the SS site (31150 0 N, 63130 0 W, bottom depth at 4500 m) on the SarC cruise from 14 to 29 June 2000. The NCP site is 1000 km north of Hawaii and the SS site is 100 km southeast of Bermuda. Radiocarbon and d 13 C measurements of suspended particulate organic carbon collected during these cruises were reported previously (Druffel et al., 2003), and radiocarbon and abundance data of dissolved organic carbon will be reported separately (Bauer et al., in preparation;Loh et al., 2004). The NCP and SS sites were occupied earlier, from 6 June to 3 July 1987 (Eve cruise) and from 29 May to 20 June 1989 (Hydros-6 cruise), respectively, and isotope results were reported separately .
Seawater samples were collected in 1999 and 2000 using 12-L or 30-L Go-flo bottles deployed on a hydrowire. Samples for DIC D 14 C and d 13 C analyses were filtered through glass fiber filters (1 mm effective pore size) directly into 1-L glass containers and poisoned with a saturated solution of mercuric chloride. Separate samples were filtered and poisoned in the same manner for alkalinity and total CO 2 ([DIC]) analyses. For the DIC D 14 C and d 13 C analyses, the samples were acidified and sparged of CO 2 gas according to published techniques (McNichol et al., 1994). Carbon dioxide was converted to graphite using hydrogen gas and cobalt metal at 550 1C (Vogel et al., 1987).
The D 14 C measurements from the Avon 1999 cruise were made at the National Ocean Sciences Accelerator Mass Spectrometry Facility (NOSAMS) of the Woods Hole Oceanographic Institution (WHOI), and those from the SarC 2000 cruise were made at the W.M. Keck Carbon Cycle AMS Laboratory at the University of California, Irvine (UCI). The DIC D 14 C values were reported according to standard techniques (Stuiver and Polach, 1977) and have a total uncertainty determined from replicate seawater analyses of73.9%. The d 13 C measurements were performed at NOSAMS with a total uncertainty of 70.1%. Alkalinity and [DIC] measurements were obtained by closed vessel titration of large volume ($100 ml) samples using an automated titration system (Bradshaw et al., 1981;Brewer et al., 1986) in the laboratory of D. McCorkle (WHOI). Alkalinity and [DIC] measurements were determined using a nonlinear curve fitting approach (DOE, 1994) and standardized using certified reference materials obtained from Andrew Dickson (Scripps Institution of Oceanography). The standard deviation of pairs of replicate analyses of culture water was 4 meq/kg for alkalinity and 6 mmol/kg for [DIC]. The alkalinity results were consistently high by 25-40 meq/kg compared with other cruises, which we attribute to the unusually long storage times of the samples prior to analysis (49 mos); for this reason, the alkalinity results were not reported.
The DIC D 14 C values obtained for samples from the SS site in 2000 (Fig. 2) decreased from 81% at 3 m to a low of À72% at 1005 m (Antarctic Intermediate Water (AAIW)); values increased to a secondary maximum (À32%) at 1513 m, and decreased to À84% at 50 m above bottom. The [DIC] values were low in the upper 50 m (2099-2107 mmol/kg), highest in AAIW (2203-2221 mmol/kg between 850 and 1300 m) and slightly lower in NADW (2182-2201 mmol/kg at 1500-3600 m) ( Table 2).
The DIC d 13 C values in the NCP were highest in surface waters (1.08% at 3 m), decreased to a minimum of À0.84% at 1141 m, and increased with depth to 5770 m (0.22%) (Fig. 3a). At the SS site, the DIC d 13 C values were highest in surface water (1.09% at 20 m), decreased to a minimum (À0.21%) at 1005 m, and increased to 0.7-0.9% in deep water ( Fig. 3b).

Discussion
In the following discussion, we compare the isotopic measurements of samples reported for the recent cruises to the NCP and SS sites with those obtained from earlier cruises.

North Central Pacific
Surface D 14 C values in June 1999 were about 40-50% lower than those measured in samples from the Eve cruise in June 1987   (Fig. 1a). This reduction reflects the decrease of bomb 14 C in DIC in subtropical surface waters since the early 1970s (Druffel, 1987). In contrast, D 14 C values between 600 and 2400 m depth were sig-nificantly higher in 1999 than those in 1987 ( Fig. 1a, b). The average D 14 C difference at depths between 600 and 900 m (38721%, n ¼ 3) was greater than that between 1150 and 2400 m (871%, n ¼ 4). Comparison of our DIC D 14 C data with profiles obtained during GEOSECS in September 1973 (Stn 204, 311N, 1501W)  and WOCE in March 1991 (Stn 31, 301N, 1521W) (Key, 1996;Stuiver et al., 1996) is also plotted with our data in Fig. 1a, b. These sites are located 860 and 680 km east of our NCP site, respectively, and some spatial variability of D 14 C may be expected. In the upper 100 m, D 14 C values were highest during 1973 (146-178%) and lowest during the 1999 cruise (77-95%). Between 400 and 2400 m, the 1999 values are higher than D 14 C values from similar depths for any of the previous cruises (Fig. 1a, b). Additionally, the depth of the D 14 C minimum appears to have deepened with time, from about 1700-2400 m in 1973, to 2200-2400 m in 1987-1991, to 3200 m in 1999 (Fig. 1a, b).
To more clearly illustrate the change of D 14 C with time, D 14 C measurements from nine depth ranges (50-2400 m) are plotted versus time of collection (Fig. 4). Data from an earlier cruise (Alcyone October 1985) to the NCP site (Druffel et al., 1989) are also included. At $50 m depth, D 14 C   and WOCE (March 1991, Stn 31, 301N, 1521W) (Key, 1996;Stuiver et al., 1996). values decrease from 1973 to 1999, and at 150 m a maximum in 1985 is apparent. At 300 m and deeper, D 14 C values generally increase with time. Least squares fits of the data from 900, 1150, 1800 and 2400 m depth (Fig. 4 inset) reveal increases of 35%, 17%, 14% and 12%, respectively, during the 26-year period. These increases in D 14 C are 43 times the 1-s uncertainty of our measurements (3.9%), and thus statistically significant. Could this increase be due to penetration of bomb 14 C below the main thermocline?

ARTICLE IN PRESS
From data compiled as a part of the Global Ocean Data Analysis Project, Key et al. (2004) showed that bomb radiocarbon had penetrated no deeper than 1000 m in the North Pacific by 1991. They found a weak bomb 14 C signal at 1000 m only in the region of intermediate water formation in the far northwest Pacific. The data presented here (Fig. 4) suggest that, by 1999, D 14 C values were higher in the North central Pacific basin, as deep as 2400 m.
Is it possible that the 12-17% increase in the deep ocean DIC D 14 C values represents natural variability? The ranges of D 14 C values taken over a 2-year period at 1600 and 2500 m depth at Stn M in the NE Pacific were 16% (n ¼ 7) and 18% (n ¼ 6), respectively (Masiello et al., 1998). The ranges of D 14 C values measured in samples from our cruises to the NCP site at 1800 and 2400 m in 1985 and 1987 were 8% (n ¼ 2) and 4% (n ¼ 4), respectively. The Stn M site is a more productive, coastal site, and comparison with the midgyre NCP site may not be valid. Roussenov et al. (2004) used an isopycnic circulation model to show that DIC D 14 C values in the deep Pacific are controlled by lateral transport of bottom water from the south and balance between advection-diffusion and decay of 14 C in the vertical. They found that a strengthening of diapycnic mixing causes an overall increase in D 14 C values over the North Pacific basin similar to the increase that we see at the NCP site. Thus, it is possible that shifts in deep circulation could cause changes in D 14 C of the order that we observe at 1150-2400 m depth. Is it possible that the increase in D 14 C is due to input of bomb 14 C? The two possible sources of bomb 14 C would be from remineralization of surface-derived particulate organic carbon (POC) to DIC, and physical mixing of bomb-laden upper waters into the deeper layers of the ocean. First, we calculate the change in deep DIC D 14 C expected from the input of remineralized bomb-laden POC from surface waters. The amount of DIC in the water column from 1000 to 2000 m with 1 m 2 area is 2.93 Â 10 4 gC/m 2 , assuming an average [DIC] of 2370 mmol/kg, and an average density of 1030 kg/m 3 . The potential change of D 14 C by remineralization of POC using the average global flux rate of sinking POC at 1000 m depth of 2.8 gC/m 2 (Martin et al. (1987) for 26 years ¼ 73 gC/m 2 ), with a D 14 C of 89%, is only 0.75%, which is much smaller than the observed 12-17% difference. This calculation assumes only vertical transport of remineralized POC and does not include lateral transport of material. As a note, we could neither use the potential alkalinity method (Rubin and Key, 2002) to estimate the actual bomb 14 C concentration (alkalinity measurements were too high), nor could we use Broecker's silica method (Broecker et al.,    and TTO in 1981 (Stn 237, 3318 0 N, 56130 0 W) cruises (Ostlund, 1981).  (Stuiver et al., 1996;Key et al., 2004). Therefore, we believe that the increase in the D 14 C values with time represents changes in circulation and mixing in the deep waters, causing shifts in the baseline D 14 C signature.

ARTICLE IN PRESS
Important processes contributing to d 13 C of DIC are (1) mixing, (2) transfer of CO 2 between air and sea, (3) remineralization of organic matter to CO 2 , and (4) removal of CO 2 during photosynthesis in the surface water. At the NCP site, DIC d 13 C values in the upper 1100 m were lower in 1999 than during any of the earlier cruises (Fig. 3a). This depletion is evidence of the 13 C Suess Effect, which is the input of 13 C-depleted CO 2 from fossil fuels and terrestrial biomass to the main thermocline (McNichol and Druffel, 1992;Quay et al., 1992). The d 13 C values in the surface decreased by 0.8% from 1973 (Kroopnick, 1985) to 1999 and at 1100 m by about 0.3% (Fig. 3a). The decrease of d 13 C in the surface of the North Pacific was about 0.02% y À1 from 1970 to 1990 (Quay et al., 1992). This is similar to our results (0.8%/26 yr ¼ 0.03% y À1 ) from 1973 to 1999, and is consistent with an increased rate as anthropogenic CO 2 input increases with time.

Sargasso Sea
The D 14 C value of SS surface water sampled in 2000 was about 40% lower than that obtained 11 years earlier at this site (Fig. 2). In contrast, at 1800 and 2200 m, D 14 C values were significantly higher (by 1974%) during 2000 than those obtained in 1989, and the same or slightly higher (773%, n ¼ 4) below 2200 m depth (Fig. 2).
A comparison of our data in 2000 with a profile obtained during GEOSECS in March 1973 (Stn 120, 33116 0 N, 56133 0 W)  and one depth (1994 m) from TTO in October 1981 (Stn 237, 3318 0 N, 56130 0 W) (Ostlund, 1981) are also shown in Fig. 2  incorporated into NADW and transported south to the subtropics via deep thermohaline circulation (Ostlund and Rooth, 1990;Smethie et al., 1986). As shown above for the NCP site, remineralization of surface-derived POC to DIC is an unlikely source of bomb 14 C to the deep SS, unless POC fluxes and dissolution rates are seriously underestimated. At the SS site, DIC d 13 C values in the upper 1500 m were lower during 2000 than those obtained from Stn 31 (2710 0 N 53132 0 W) during the 1972 GEOSECS cruise (Kroopnick, 1985) (Fig. 3b) (no d 13 C measurements were available from Stn 120). From 1972 to 2000, the d 13 C values had decreased by about 0.8% in the surface and by about 0.5% between 200 and 700 m. At 1000 m, the 2000 value appears anomalously low (À0.2%), though there is no obvious reason to discount this measurement. As in the NCP, the decrease of d 13 C values indicates the presence of 13 C-depleted atmospheric CO 2 from anthropogenic sources. Kortzinger et al. (2003) reported that the mean d 13 C decrease in the upper 1000 m of the North Atlantic from 1950 to 1993 was 0.02670.002% y À1 . Data from Fig. 3b show a similar decrease from 1972 to 2000 of 0.015-0.026% y À1 in the upper 700 m.

Conclusion
Bomb 14 C has continued to increase in the deep waters of the SS from 1973 to 2000. What is surprising is the increase in D 14 C to a depth of $2400 m in the NCP, because anthropogenic tracers (e.g., bomb 14 C, tritium and CFCs) have been found only to a depth of about 1000 m in the North Pacific by the 1990s . It does not seem feasible that remineralization of POC from the surface could provide a sufficient source of bomb 14 C to the deep North Pacific. Instead, we conclude that changes in deep water circulation patterns offer a plausible explanation of the increase in D 14 C that we observe in the deep waters of the North Pacific Ocean. with the analyses. We thank Bob Key for his advice and guidance, and Robbie Toggweiler, Mike Bacon and two anonymous reviewers for valuable comments on the manuscript. We gratefully acknowledge funding from the National Science Foundation  Fig. 2 caption) plotted as a function of time of sample collection. Results are binned in depth ranges defined in legend. Inset shows D 14 C results for the depth bins at 1800, 2700, 3600 and 4400 m with least squares fits (dashed lines) shown. Linear correlations, r, for these four depth bins are 0.97, 0.96, 0.99 and 0.96, which represent correlations significant to the 95%, 95%, 99% and 95% confidence levels, respectively.

ARTICLE IN PRESS
University of California Office of the President Marine Council Research Fellowship (to JH).