Temporal variability in dissolved organic carbon and radiocarbon in the eastern North Pacific Ocean

. The factors regulating the steady state inventories and residence times of dissolved organic carbon (DEC) in the deep ocean are not well established. Previous studies of Dec have been limited to single time-point profiles that provide general information on the potential role of vertical advective-diffusive processes in controlling Dec distributions and mean apparent ages. We present results from a 2-year time series station in the eastern North Pacific (station M) where short-term (months) changes in inventories and A(cid:127)4C signatures of Dec as measured in deep profiles were examined in conjunction with changes in particulate organic carbon (Pec) pools. Significant long-term (i.e., months to years) changes in both Dec concentrations and A(cid:127)4C values were observed. These changes were especially evident at mesopelagic (-450 and 700 m) depths, close to the oxygen minimum. Both within the mixed layer and at mesopelagic depths, positive relationships were found between Dec A(cid:127)4C values and concentrations of station M, primarily reflecting diminishing vertical inputs of recent Dec throughout the main thermocline. At abyssal depths (-> 1600 m), however, A(cid:127)4C was inversely correlated with 14 Dec concentration. The A C signature of the less abundant suspended and sinking Pec pools has been observed to fluctuate over seasonal timescales at station M, presumably due in part to sorption of Dec to Pec However, the A(cid:127)4C values and concentrations of the correspondingly much larger Dec pool do not appear to be related to seasonal changes in either sinking Pec fluxes or suspended Pec abundances. Significantly elevated concentrations of Dec were observed at station M when compared with a previously occupied site in the north central Pacific (NCP) in all regions of the water column except mesopelagic depths, where concentrations were lower. The corresponding A(cid:127)4C values of Dec at all depths at station M were lower than in the NCP. We speculate that dissimilarities in the size and A(cid:127)4C signature of the Dec pools at seasonally productive station M and the oligotrophic NCP result from differences in Dec sources and sinks between the .two regions, as well as from the magnitude of interaction between Dec and Pec at these sites.


Introduction
Dissolved organic carbon (DEC) in seawater represents one of the largest pools of exchangeable organic matter at the earth's surface. The total size of the oceanic Dec reservoir has been estimated at 0.6-1.0 x 10 •8 g C [Williams and Druffel, 1987;Hedges, 1992] and far exceeds all other forms of marine organic carbon, both living and dead. As a result of its relatively low concentration in seawater (-40 •M in deep waters and up to -100/xM in surface waters) and its apparent relationship to the generation of particulate organic matter by primary and secondary production [Williams and Yentsch, 1976;Eppley et al., 1981;Lancelot, 1983 and Druffel, 1987;Bauer et al., 1992a] has been taken as an indication of the steady state and relative invariance in these parameters over large time and space scales. The present study was undertaken in an effort to evaluate the temporal variability in DOC concentrations and radiocarbon signatures as indicators of seasonal and longer-term sources and sinks of DOC to the deep ocean. A single time series station, having a typical maximum in upwelling-associated surface productivity and pelagic POC flux in early to midsummer  was occupied periodically for 2 years. This study was designed to examine (1) if DOC concentrations and radiocarbon signatures in the deep ocean are temporally variant and related to the concentrations and fluxes of sinking and/or suspended POM, and (2) how DOC concentrations and radiocarbon signatures in the eastern margin of the North Pacific compare with those in the central North Pacific gyre [Williams and Druffel, 1987]. To help evaluate these relationships, the concentrations, fluxes, and •4C natural abundances of both suspended and sinking POC were also measured during this 2year period; results of this aspect of the study are presented separately [Druffel et al., 1996]. During the course of this study, both normal and E1 Nifio-Southern Oscillation (ENSO) conditions were encountered, further allowing us to compare DOC concentrations and radiocarbon signatures during periods of typical and atypical meteorological and organic matter flux conditions.

Study Site
The area chosen for this study has been occupied every 3-4 months since 1987 by Smith et al. [1992bSmith et al. [ , 1994 for studies of deep ocean POC flux, benthic respiration and other parameters related to food sources and sinks in the benthic boundary layer. Station "M" is located at 34ø50'N, 123ø00'W, at the base of the Monterey Deep Sea Fan, and average water depth is approximately 4100 m (Figure 1).
Wind-driven upwelling occurs in late spring and summer along the California coast, and much of the biomass produced from this coastal upwelling is carried offshore in the form of discrete plumes or jets of chlorophyll Michaelson et al., 1988]. Independent long-term studies in this region [Hayward et al., 1994  . This production drives a seasonally varying vertical flux to the deep ocean at station M, with a primary peak occurring in early to mid summer and a secondary peak in the late fall .
Station M was occupied on six separate occasions from July 1991 to July 1993 (Table 1). The flux rate of POC at 50 m above bottom (mab) at this site ranges from less than ---2 mg C m -2 d-1 during nonupwelling periods to greater than 20 mg C m -2 d -• during the summer   (Table 1).
During the 1992 ENSO event, near-bottom POC flux rates were significantly lower than those measured during the summers of 1991 and 1993. As shown by Baldwin et al. [1998], climatic perturbations such as the 1992-1993 ENSO event caused significant interannual deviations from the normal seasonal pattern and greatly reduced the maximum POC flux typically observed in summer (Table 1). Thus station M waters and sediments experienced a protracted period of low POC flux, in contrast to the two annual maxima in flux rates normally observed at this site Baldwin et al., 1998].

Sampling Methods
All sampling preparations and sample processing procedures were conducted using organic-and isotopic-contamination-free protocols [Druffel e! al., 1992] sample collection for ---30 min prior to tripping. Bottles were kept closed and covered in the ship's lab at all times in order to reduce contamination from atmospheric and other sources.
Seawater for both total Dec concentrations and Ax4C measurements was gravity filtered directly from the sampling bottles through a precombusted (525øC) 147-mm-diameter GF/C (1-/•m nominal pore size) glass fiber filter. Samples were collected in precombusted 1-L glass bottles with Teflon-lined caps for measurements of A14C in Dec (DeC-A•4C). All samples were stored at -20øC in the dark until analysis. The sample collection, processing, and storage protocol used here was found not to contribute any measurable blank carbon to the samples. As part of the larger study at station M, concentrations and Ax4C signatures of suspended and sinking Pec were determined concurrently with Dec concentrations and A•4C, and sampling details and results are presented by Druffel et al. [1996]. In addition to Dec, samples were also collected at the same times and from the same sampling bottles for dissolved inorganic carbon (DIC) concentrations and A X4C signatures,  [Schultz, 1995] and salinity and oxygen.

Waters
During the occupation of station M in October 1992 (Pulse 15 cruise), seawater samples were collected from 85-m depth at 1-to 2-day intervals in order to document short-term fluctuations of DOC concentrations and A•4C in surface waters. This depth was selected because (1) it exhibited significant variability in DOC concentrations and A14C between early cruises and (2) it was located at or near the chlorophyll maximum [Bianchi et al., 1998]. It was not possible to perform these collections on all days or at the same time on each day. Samples were collected and stored as was described above. In addition to DOC and A•4C, dissolved oxygen concentrations were also monitored.

Analytical Methods
Concentrations of DOC were measured by high energy ultraviolet (UV) irradiation of seawater samples. The UV determinations of DOC concentration were conducted as a routine part of the measurement of •4C natural abundance of DOC. The DOC was oxidized by a modification of the method described by Williams and Druffel [1987] and Druffel et al. [1992]. Seawater samples (650 mL) were initially acidified to pH 2-2.5 with 85% H3PO 4 and sparged free of inorganic carbon for 45 min with ultrahigh-purity nitrogen or oxygen. All steps of this procedure were carried out following the transfer of the thawed sample from the sample bottle to an optically clear quartz vessel designed to interface directly to a vacuum extraction line. The quartz vessel containing the sparged seawater sample was irradiated at the focal point of a 2400-W, mediumpressure mercury arc UV lamp equipped with reflector assembly (Canrad-Hanovia Co., Newark, N.J.) for 120 min. Following irradiation, the CO2 evolved from the oxidation of DOC was purged from the quartz vessel using ultrahigh-purity nitrogen, transferred to the vacuum extraction line through a KIO3 trap to remove C12 gas, cryogenically purified, and quantified using an absolute pressure gauge (MKS Corporation). The sample was then split, with -90% of the volume being used for subsequent A•4C analysis and --•10% for /5•3C, and flame sealed into 6-mm Pyrex tubes.
The CO2 for A•4C measurements was converted to graphite targets in an atmosphere of H 2 over Co catalyst [Vogel et al., 1987] and analyzed at the Center for Accelerator Mass Spectrometry at Lawrence Livermore National Laboratory. Typical A•4C measurement errors for sample sizes in the range analyzed (-250-600 /•g C) were _+4-6%0. All reported A14C values were corrected for fractionation using the appropriate 15•3C value of each sample [Stuiver and Pollach, 1977], which was measured using either a VG Micromass 602E or a Finnegan Delta S isotope ratio mass spectrometer having analytical precisions of better than _+0.1%o. Irradiations of replicate seawater samples collected from the same cruise and depth were within approximately _+ 1 •M C (i.e., _+ lo-) of each other with respect to DOC concentrations calculated from CO2 yields. Hence the UV method has a higher precision and sensitivity than the high-temperature catalytic oxidation (HTCO) methods (_+lo-= -3 •M C) in common use. The A•4C values of these replicate samples had a precision that was within the A•4C measurement error of 3-6%o. The total blank for the UV oxidation procedure (including sample handling and sparging) was determined by reprocessing and reirradiating both seawater and double distilled water samples that had been previously processed and the DOC removed by UV oxidation. The total processing blank was found to be <1 /xmol CO2.
On two of the early cruises (Pulse 7 and 11) to station M, we compared the A14C of DOC extracted by UV oxidation to that extracted by continuous injection high-temperature catalytic oxidation (CI-HTCO) for selected samples (Table 2)

Seasonal DOC Concentrations in the Eastern North Pacific
All data from samples collected for water column profiles in this study are tabulated and presented in Table 3 (Table 1). The higher variability observed in station M surface waters may also in part be a result of greater "patchiness" in general there (see below and Figure 4). At --• 1600-m depth, a slightly greater range (6/aM) in subsurface DOC concentration was observed compared to all other depths sampled (average range = 4 +_ 1 /aM). Profiles of DOC concentration did not exhibit consistent season-to-season differences, but concentrations in July 1991, the period of highest sinking POC flux, were among the lowest observed throughout the entire water column (Figure 2a). In addition, there was no obvious relationship between the concentrations of DOC and suspended POC [Druffel et al., 1996] in the water column.  (Figures 2a and 2b). This was also true at the shallowest depth sampled in the present study ( Figure 2b). However, at -450-and 700-m depths, concentrations were consistently lower at station M than in the NCP (Figure 2b). These differences in DOC concentrations are likely related to variability in the relative strengths of DOC sources and sinks in various parts of the water column at these two sites (see Section 4).

Influence of POC Flux on DOC and A•4C-DOC
The major source of DOC to the deep ocean ultimately derives from soluble or solubilized forms of organic matter produced in the surface ocean by living organisms. However, the geochemical and microbial transformations leading to the formation of the large residual DOC pool observed in the oceans are not well understood. Therefore an interpretation of the •4C signature of the bulk DOC pool is complicated because the signature integrates the influence of numerous factors relating to DOC sources and sinks, and the isotopic signatures of DOC added to and removed from the standing pool.
At station M, seasonal profiles of DOC (Figures 2a and 2b) and A•4C of DOC (Figures 3a and 3b) Table  3) also suggests that organic matter respiration may be less efficient (i.e., oxygen limited) in this part of the water column.
An alternate hypothesis for the relationship between DOC,

A•4C-DOC, and POC flux is that the rate of change in DOC concentration and A•4C is related to flux. In this regard, the rise in DOC concentrations below 1600 m after June 1991 may be evidence for extensive POC solubilization following (in contrast to during) the high-flux period. Similarly, the decline in deep DOC concentrations after February 1992 (Pulse 11) may be related to the buildup of suspended POC and may be evidence for the conversion of DOC (or its sorption) to suspended POC during a relatively quiescent period between
high-flux events. If true, this conjecture could be shown unequivocally by performing a longer or more frequently sampled time series to further elucidate these trends.  Table 3). 1995] indicate that A•4C values are much greater (with correspondingly younger •4C ages) than deep ocean DOC. Therefore this mechanism for introducing greater amounts of •4Cdepleted DOC to station M waters may be limited or at least localized in nature. We speculate that selective adsorption of more hydrophobic, •4C-depleted DOC onto organic [Druffel et al., 1996] or inorganic [Mayer, 1994; particle surfaces, especially during high-particle-flux periods, could also lead to the observed relationship between A•4C and DOC in the deep ocean station M.

Values of the A•4C of DOC in the eastern North Pacific
were consistently and significantly lower than at a site in the north central Pacific examined previously by Druffel et al.
[1992] (Figures 3a and 3b). This offset in A•4C was accompanied by concomitantly higher DOC concentrations at nearly all depths at station M compared to the NCP (Figures 2a and 2b). Because some part of the offsets between station M and the NCP could be due to differences in isopycnal depths between the two sites, especially at depths above the main thermocline,  There are three possible explanations for the observed differences in the profiles from these two locations. First, there may be regional differences in the A14C of the source materials to these waters. Lateral movement of materials (i.e., Pec and Dec) outward from the margin toward the deep abyssal ocean at station M may impart a qualitative (i.e., isotopic) difference to the Dec or its precursor materials. However, because of the nonuniform offsets in Dec concentrations throughout the water column between these two sites (Figures 2a and 2b), the A14C signatures of material advected from the margin would need to vary widely as a function of depth in order to derive the observed offsets in the A14C profiles from the two locations (Figures 3a and 3b). Furthermore, surficial sediment (0-to 10-cm depth) porewater Dec as well as solid phase sediment organic carbon (sec) in this general region have been found to be uniformly enriched in 14C with respect to water column  Table 3) due to the absence of corresponding temperature and/or salinity measurements in many cases.

A14C-DIC and its transfer into both surface and deep water
Dec at station M.
A third possibility is that differences in the magnitude of sinking and/or suspended Pec fluxes between the NCP and station M help to control the observed average concentration and isotopic differences in Dec. Although low at most times of the year, deep sinking Pec fluxes at station M can also be quite high (up to 20 mg C m -2 d -1) on a seasonal basis (Table   1). This compares to typical sinking Pec fluxes for the oligotrophic NCP of -1-2 mg C m -2 d -1 [Smith, 1992;Smith et al., 1992]. Likewise, suspended Pec concentrations are up to 3-4 times higher at station M than at the NCP site in both surface (20-25 m) and deep (->900 m) waters [Druffel et al., 1992;1996 are to a first approximation controlled by vertical transport processes and that these are dominated by the strong summer flux of POC to the deep ocean at station M. Assuming that the primary means of organic matter transport is vertical (i.e., horizontal transport is insignificant), and the difference in the A•4C profiles of DOC in the NCP and at station M is due primarily to the difference in A•4C of surface DIC (and, consequently, surface organic matter) at the two locations, then we conclude that the surface A •4C signal is manifested in the deep DOC pool on timescales of at most 4-6 years (i.e., the time between NCP and station M sampling). If, on the other hand, horizontal transport from the shelf and slope to station M is more important, then temporal differences in A•4C of DIC could be manifested on shorter timescales. Time series measurements of the A•4C in suspended POC profiles show that significant short-term (months) shifts in these profiles occur [Druffel et al., 1996] and strongly suggest that lateral transport of slope and shelf derived material occurs during summer periods station M. However, such vertical and horizontal transport mechanisms may play an important role in the removal (i.e., via and adsorptive or "stripping" mechanism) of at least a fraction of DOC from the water column as well. sponding DOC pool. Instead, the deep ocean DOC pool appears to be controlled by longer-term inputs (POC dissolution, lateral advection) and losses (selective adsorption and/or "stripping out" of DOC during periods of elevated POC flux rates) resulting from the combined effects of interactions with organic and inorganic particles as well as other sources and sinks which have yet to be quantified. We hypothesize that differences in the DOC pools between the seasonally productive eastern North Pacific and the oligotrophic north Central Pacific arise from qualitative and quantitative differences in the sources and sinks of DOC between the two regions, as well as from the interactions between DOC and both organic and inorganic particles in the water column.