Constraining the propagation of bomb-radiocarbon through the dissolved organic carbon (DOC) pool in the northeast Paciﬁc Ocean

This study extends the 1991-1995 records of marine dissolved organic carbon (DOC) concentrations and D 14 C values at hydrographic Station M (34 1 50 0 N, 123 1 00 0 W) with new measurements from a frozen (-20 1 C) archive of samples collected between April 1998 and October 2004. The magnitudes and synchronicity of major D 14 C anomalies throughout the time-series imply transport of DOC from the surface ocean to depths of at least 450m on the timescale of months. Keeling plots of all measurements at Station M predict a continuum of possible background DOC compositions containing at least 21 m M of -1000 % (i.e., Z 57,000 14 C years) DOC, but are more consistent with mean deep DOC (38 m M, -549 % ; i.e., (cid:2) 6,400 14 C years). These results and coral records of surface dissolved inorganic carbon (DIC) D 14 C were used to estimate pre-bomb DOC D 14 C depth proﬁles. The combined results indicate that bomb- 14 C has penetrated the DOC pool to depths of Z 450m, though the signal at that depth is obscured by short-term variability.


Introduction
Dissolved organic carbon (DOC) is the largest reservoir of reduced carbon in seawater with a global abundance of $685 Â 10 15 g C (Hansell and Carlson, 1998) and is approximately equal in size to atmospheric CO 2 (Hedges, 1992). Despite its importance in the oceanic carbon cycle and more than a century of research (Natterer, 1892), DOC biogeochemical fluxes are poorly constrained. For example, the estimated global riverine input of $0.25 Gt DOC per year (Cauwet, 2002;Meybeck, 1982) is sufficient to support its radiocarbon ( 14 C) based oceanic residence time. Yet molecular composition (Meyers-Schulte and Hedges, 1986;Opsahl and Benner, 1997) and stable carbon (d 13 C) isotopic (Williams and Gordon, 1970) signatures indicate that the majority of marine DOC is autochthonous, ultimately originating from primary production in the euphotic zone. Furthermore, while 4,000-6,000 year 14 C ages reported for deep DOC suggests a major proportion cycles on long time scales and ages during deep water transit Williams and Druffel, 1987), the removal processes necessary to produce these ages remain unknown.
As a tracer of time Libby, 1955Libby, , 1961Libby et al., 1949) and carbon sources Mortazavi and Chanton, 2004;Trumbore and Druffel, 1995), the 14 C content of marine DOC is a powerful tool for potentially constraining many of these uncertainties. However, significant analytical challenges have limited the number of bulk marine DOC D 14 C observations to five regions globally (Fig. 1). Long-term measurements of DOC D 14 C have been reported for only two of these locations: the central North Pacific (CNP) in 1985 and 1987 Williams and Druffel, 1987), and Hydrographic Station M off the coast of California from July 1991 to June 1995 (Bauer et al., 1998a(Bauer et al., , 1998b. Therefore, the global data set of DOC D 14 C observations is significantly limited in spatiotemporal range and resolution.  Williams and Druffel, 1987) created by thermonuclear weapons testing in the 1950s and 1960s (Nydal, 1963;Nydal et al., 1980). Since bulk DOC D 14 C values are weighted averages of the radiocarbon contents of DOC constituents, knowledge of bomb-DO 14 C distributions would constrain the spatiotemporal scales over which DOC is reactive. The long time-series at Station M is the most promising data set for observing such changes in the marine DO 14 C inventory.
This study extends the previously published record of DOC concentrations and D 14 C values at Station M (Bauer et al., 1998a(Bauer et al., , 1998b with new measurements of archived samples from April 1998 to October 2004. In addition, Keeling plots (Keeling, 1958;Mortazavi and Chanton, 2004) from each depth profile throughout this time-series were used to constrain the isotopic composition of bulk marine DOC at Station M. These analyses were combined with coral records of surface DIC D 14 C (Druffel, 1987) to construct pre-bomb DOC D 14 C depth profiles and constrain the propagation of bomb-carbon through the DOC pool at Station M.

Materials and procedures
Seawater samples were collected during a series of cruises 37,38,40,and 45; Table 1) to hydrographic Station M (34150 0 N, 123100 0 W), a long-term abyssal study site (maximum depth of $4100 m) located 220 km west of Point Conception, CA (Smith and Druffel, 1998). Samples from $25 m, $85 m, and $450 m depth were analyzed for the period of April 1998 to June 2002. Additional samples from seven depths (one surface, three mesopelagic, and three 41000 m depth) were collected during October 2004. Samples collected from the Southern Ocean (541S, 1761W) in December 1995 (Druffel and Bauer, 2000) were also analyzed as a test of isotopic fidelity during long-term frozen storage. Except where noted, the time-series (Table 1) represents single measurements of individual samples from discrete bottle or CTD rosette casts.
All samples from April 1998 to June 2002 were gravity filtered directly from Go-Flo bottles into 1-liter glass bottles using pre-combusted (525 1C) GF/C glass fiber filters (1 mm nominal pore size). Samples from October 2004 were gravity filtered directly from Go-Flo bottles into 3.8-liter glass jugs using Whatman Polycap AS filter capsules (0.2 mm nominal pore size) that were pre-rinsed with acetonitrile and Milli-Q water. Filters were connected to the bottles using Nalgene 50 platinized silicone tubing that was previously washed with 10% HCl and Milli-Q water. All filters and silicone tubing were flushed with sample for several minutes prior to dispensing into glass bottles. The bottles were then sealed with polytetrafluoroethylene (PTFE) Teflon lined caps that were pre-rinsed with 10% HCl, Milli-Q water, and seawater sample. Lastly, the sealed sample bottles were wrapped in clean polypropylene bags, and stored frozen at -20 1C until laboratory analysis. DOC concentrations and D 14 C values were measured via an ultraviolet-oxidation procedure described in detail elsewhere . Briefly, each seawater sample was transferred into a quartz reaction vessel, then acidified with 1 ml of 85% H 3 PO 4 , sparged with ultra high purity (UHP) helium to remove DIC, and irradiated with UV-light from a 1200 Watt medium pressure mercury arc lamp for a total of 4 hours. Samples archived before October 2004 ranged in volume from 347 -552 ml and were diluted to $1 liter with pre-irradiated Milli-Q water prior to UV-oxidation. The resulting CO 2 was then extracted from the residual seawater via sparging with UHP helium and collected in a dedicated vacuum line. DOC concentrations were calculated via manometric quantification of the CO 2 collected and measurements of the seawater volumes that were irradiated.
Following reduction of CO 2 to graphite (Vogel et al., 1987), DOC D 14 C and d 13 C values were concurrently ig. 1. Map of locations of previously published bulk marine DOC D 14 C measurements Bauer et al., 1998aBauer et al., , 2002Bauer et al., , 1992Beaupré et al., 2007;Druffel and Bauer, 2000;Druffel et al., 1992Druffel et al., , 1989Williams and Druffel, 1987;Williams et al., 1969Williams et al., , 1978 (Santos et al., 2007). Since DOC abundances in the smaller volume samples (ca. o0.5 L) were insufficient for complimentary d 13 C measurements by isotope ratio mass spectrometry (IRMS), d 13 C measurements are not reported. However, D 14 C values are still reported according to the conventions of Stuiver and Polach (1977).
All concentrations and D 14 C values were corrected for the methodological blank of the UV-extraction procedure.
The mass of blank carbon, 271 mg C, was determined by reanalysis of previously oxidized seawater and Milli-Q water. The small mass of carbon in the blank precluded direct analysis by AMS. Therefore, the radiocarbon content of the methodological blank was determined by observing the deviations in D 14 C measurements of UV oxidized 14 C standards [IAEA C-6 (Rozanski et al., 1992), IAEA C-7, and IAEA C-8 (Le Clercq et al., 1998)] from their consensus values.

Sample storage
The effects of long-term storage (r12 years) on DOC concentrations and D 14 C values were determined by measuring Southern Ocean samples collected and frozen  (Bauer et al., 1998a(Bauer et al., , 1998bMasiello et al., 1998). in December 1995. These were the only available samples stored prior to the onset of the Station M DOC archive and quantified prior to frozen storage (Druffel and Bauer, 2000). Seven samples containing less than 100 ml of seawater were available from 25 m, 85 m, 250 m, 450 m, and 4200 m depth, and required dilution to $1 liter for UV-oxidation of DOC. These volumes and the associated total masses of DOC extracted from each sample were nearly an order of magnitude smaller than the undiluted seawater typically used in this extraction procedure, resulting in significantly larger uncertainties for both concentration and D 14 C measurements. Upon thawing, five of the seven samples revealed the presence of a persistent white crystalline precipitate, as well as a crystalline film on the inner walls of their glass bottles. Elemental analysis (EA) coupled with IRMS of lyophilized crystals revealed that the precipitate was 6.470.1%

ARTICLE IN PRESS
carbon by mass with a d 13 C value of -0.670.2%. These values are consistent with the precipitation of barium carbonate but may contain other constituents. The limited mass of crystals did not permit further elemental characterization.
Concentrations and D 14 C values of the two samples that did not possess crystals were within 2 standard deviations of previously published values ( Fig. 2, gray data points), whereas those of the five samples with precipitated crystals were not ( Fig. 2, white data points).
Since the crystals did not dissolve prior to analysis, both high and low yields could have resulted from transferring non-uniform slurries into the reactor. Despite originating from the same Go-Flo bottle, two out of three 250 m deep samples developed a precipitate and exhibited concentration and isotopic inconstancy (Fig. 2,  triangles). Therefore, the crystallization observed here might be an artifact of sampling procedures coupled with frozen storage rather than an intrinsic property of the seawater samples. Only one of the 24 archived Station M time-series samples (UCID 9288, Pulse-34 cruise, April 1998, 450 m) developed persistent crystals and possessed a DOC concentration (58.9 mM) that was inconsistent at this depth of the time-series. This sample was presumed compromised and therefore excluded from the present analysis. Fine white precipitates developed in the three samples from Pulse-34 and the 450 m sample from Pulse-38 that are presented in Table 1. However, these crystals dissolved readily upon shaking and were attributed to turbidity in the thawed samples. All measurements reported here (Table 1) were performed on samples that did not develop a persistent crystalline precipitate during storage.
Although the physicochemical connection between the precipitate and DOC fidelity is not yet understood, their association has several practical implications. Analytically, samples can be frozen for at least 12 years with DOC concentration and isotopic fidelity within the present methodological uncertainty as long as precipitation has not occurred. However, the unpredictable occurrence of precipitation in this Southern Ocean sample set preliminarily limits the viability of long term frozen storage of seawater for DOC analyses. Geochemically, if an inverse relationship exists between fidelity and crystal formation in frozen samples, then a natural mechanism may exist for altering the native characteristics of DOC. Additional analyses are required to confirm this relationship and its potential for coupling the biogeochemistry of DOC to seaice formation.

DOC time series data
This work extends the previously published Station M DOC time-series that began July 1991 and ended July 1993 (Bauer et al., 1998a). An additional depth profile of DOC concentration and D 14 C at Station M from June 1995 (Bauer et al., 1998b) has been included in this assimilation (Table 1, Fig. 3). The literature values were reported with precisions of 3-6% and approximately 71 mM for replicate D 14 C and concentration measurements (Bauer et al., 1998a), respectively, and are comparable with the method employed here . Therefore, the complete time-series (Fig. 3) is assumed to be free of instrumental artifacts.
The collection depths of previously published (Bauer et al., 1998a(Bauer et al., , 1998b and archived DOC samples varied with each cruise. Thus, concentrations and D 14 C values were grouped into ''depth bins'' (Table 1) (Druffel and Bauer, 2000). Percent yield is the ratio of concentration measurements performed in 2007 to those performed at the time of collection in 1995 (Druffel and Bauer, 2000), expressed as a percentage.
The concentration and D 14 C value of the previously published 4200 m sample was taken as the average of all measurements below 1000 m to accommodate the large scatter in the original data. Error bars on the data points represent 72 standard deviations. The gray data points correspond to the only samples in this test that did not contain a crystalline precipitate when thawed in 2007. not positively correlated with the number of samples or range of depths per bin (Beaupré , 2007). Therefore, variability within each depth bin throughout the timeseries was not an artifact of data organization.
On average, depth profiles of DOC concentrations and D 14 C values from each depth bin were typical of marine DOC (Fig. 4a, b). Both were enriched in surface waters and decreased monotonically until becoming practically invariant below $1000 m. DOC concentrations varied significantly in the upper $450 m (Fig. 3a), with maximum variability (1s ¼ 74.8 mM) at 24 m and 85 m depth (Table 1). With the exception of $1600 m (72.4 mM) and 50 m above bottom (72.9 mM) the magnitudes of variability at all other depths (r71.6 mM) were on the order of methodological precision (r71.5 mM) for concentration measurements via UV-oxidation .   (Bauer et al., 1998a(Bauer et al., , 1998b. Major tick marks on the ''Year'' axis denote the first day of January. propagation of the near-surface D 14 C signal to depth. One possible mechanism may be seasonal convective overturn, as has been observed to vertically redistribute DOC in the Sargasso Sea (Carlson et al., 1994). However, limited temporal resolution in the time-series did not permit conclusive analyses of seasonal or climatological influences on these events. The strong signal at 455 m during various times of year (Figs. 3 and 5) may also have developed hydrographically via water movement and spatial gradients near Station M (e.g., Hansell and Carlson, 2009;Hansell et al., 2002), or geochemically by any processes that transport 14 C-enriched organic carbon through the operationally-defined DOC pool. For example, sinking particulate organic carbon (POC) dissolution (Repeta and Aluwihare, 2006) or POC mediated transport of DOC from the surface ocean (Beaupré , 2007;Druffel et al., 1996) may have influenced the D 14 C values at depth.
The vertical distribution of DOC D 14 C was directly related to DIC D 14 C throughout the entire water column (r 2 ¼ 0.952; Fig. 6) suggesting tightly coupled long-term (4 months) processes that redistribute both carbon pools at Station M. However, none of the individual depth bins exhibited robust correlations between DOC and DIC D 14 C values (Fig. 6). The highest correlation, r 2 ¼ 0.635, was observed at 24 m, while coefficients of determination from all other depths were r0.228. Therefore, the physical controls on short-term DIC variability below $85 m (Masiello et al., 1998) did not likely dominate the observed short-term variability in DOC D 14 C. Mortazavi and Chanton (2004) have demonstrated that covariance between DOC concentrations and D 14 C values with depth (Fig. 4) can be explained by Keeling plot models. Under this model, DOC consists of a background (bg) component of constant concentration and isotopic composition throughout the water column to which a second, isotopically distinct component is added in excess (xs) (Keeling, 1958;Pataki et al., 2003). This model makes no assumptions about the composition of background DOC and is consistent with previous descriptions of an isotopically-depleted DOC fraction that is uniformly distributed with depth (Williams and Druffel, 1987). The model can be expressed by the following equations for conservation of mass, (1) and isotopic mass-balance  values determined by Mortazavi and Chanton (2004) for the period of July 1991 to July 1993.

Concentration and isotopic composition of background DOC
The Keeling plot model does not presume a concentration or isotopic signature for the background component of DOC. The only constraint is conservation of mass. That is, the background DOC must not exceed the minimum observed concentration and D 14 C values. Uniform concentrations and D 14 C values below $1000 m (Figs. 4a, b) suggest an upper limit equal to mean values for deep DOC (3872 mM and -549720%). Since all D 14 C values must be greater than -1000%, the D 14 C value of background DOC must lie between -1000% and -549720%. Based on the Keeling plot of all time-series data (Table 2), this corresponds to a continuum of potential background compositions that range from 2171 mM with a D 14 C of -1000% to 3872 mM with a D 14 C of -549720% (Fig. 8a).
Therefore, the concentration of background DOC present throughout the water column is at least 21 mM. Likewise, the concentration of surface-derived excess DOC in deep water at Station M is at most 17 mM (i.e., 38 mM -21 mM).
In addition, the significant variability in mean deep DOC (72 mM,720%) may be explained by assuming a background composition along this continuum and propagating the variabilities in Eqs. (1) and (2) introduced by the presence of DOC xs in the deep ocean. For example, adding 272 mM of 53720% DOC xs (e.g., mean 24 m DIC, Table 1) to a background of 3670 mM and -56470% would lead to mean bulk values of 3872 mM and -548732%.
Since the Keeling plot model (Eq. (3) For example, varying DOC D 14 C xs values faster than DOC xs can be redistributed and removed is one possible mechanism by which the water column could be populated with multiple, unique components. Although the persistently high D 14 C xs and rising r 2 values for Keeling plots from Feb 1991 to Feb 1993 are consistent with this hypothesis, finer temporal resolution and measurement uncertainty are required to confirm this mechanism.
3.6. Pre-bomb DOC D 14 C depth profiles Although propagation of the bomb-transient through the DOC pool has been modeled (Repeta and Aluwihare, 2006), there are currently no known proxies for reconstructing historical marine DOC. However, the strong correlation between DOC and DIC D 14 C (Fig. 6), coupled with proxy DIC data (Druffel, 1989;Druffel and Griffin, 1995;Druffel et al., 2004) and Keeling plot analyses, can be used to estimate historical profiles of DOC D 14 C. Without additional information, depth-dependent reconstructions must assume the historical marine carbon cycle did not differ significantly from contemporary observations. Therefore, the average vertical distribution of DOC concentrations immediately prior to thermonuclear weapons testing is assumed to have been the same as observed during this time series (Fig. 4a). Since the conventional radiocarbon ages of deep DOC in the Eastern North Pacific (6,400 years) are orders of magnitude greater than the time elapsed since thermonuclear weapons testing ($50 years), we further assume that the pre-bomb composition of background DOC has not changed and is equal to mean, deep DOC (38 mM and -549%). Finally, assuming that D 14 C xs is equal to prenuclear estimates of surface DIC at Station M [ca. -60%, (Berger et al., 1966;Druffel and Williams, 1991)] the pre-bomb D 14 C profile is calculated by substitution into Eq. (3):

ARTICLE IN PRESS
The depths at which D 14 C values are assigned in this prebomb profile (Fig. 9), gray diamonds) are based on the depth profile of contemporary DOC concentrations (Fig. 4a). The pre-bomb profile is isotopically depleted in the upper $1000 m compared to the contemporary profile and suggests that significant amounts of bomb-14 C have not penetrated the DOC pool below this depth. Representing background carbon as mean, deep DOC therefore generates a conservative estimate of the persistence and re-distribution of bomb-14 C throughout the water column. Alternatively, assuming background DOC is compositionally similar to the Keeling limit (21 mM and -1000%, -900 -800 -700 -600 -500 -35000 -30000 -25000 -20000 -15000 -10000  Table 2). The dashed gray line is the relationship predicted by Eq. (5) assuming the background component of DOC has a concentration (21 mM) and D 14 C value (-1000%) equal to the Keeling plot limit (i.e., Fig. 8a). The point lying outside the gray shaded area (-14700 mM%, -59%) is from the April 1998 depth profile.  Fig. 4b). Prebomb D 14 C values were calculated using a pre-bomb estimate of the surface DIC D 14 C value [ca. -60%, (Berger et al., 1966;Druffel and Williams, 1991)] and assuming DOC bg is compositionally similar to mean deep DOC (38 mM, -549%; gray diamonds) and the Keeling limit (21 mM, Eq. (7)) implies significant bomb carbon has been distributed to the deep waters of Station M.

Conclusions
The synchronicity, magnitude, and rate of variability in strained. Continued observations with increased spatiotemporal resolution and precision may refine the relative importance of particle transport, hydrography, and climatology in this system. In addition, an augmented, highprecision time-series will help constrain the isotopic composition of DOC and validate the two-component model at this site. The global DOC data set ( Fig. 1) may be similarly augmented via analyses of modern proxies (e.g., DIC and banded corals), contemporaneous DOC D 14 C measurements, and quantification of regional variability in the relationship between deep DIC and DOC D 14 C.
Coupled with simple-box models, these reconstructions may be used to determine the global impact of bomb-14 C on marine DOC, the turnover time and fate of long-lived deep DOC, and the fluxes between oceanic carbon reservoirs.