A 6-year time series of Δ14C, δ13C, and C/N measurements in deep sinking particulate organic matter (POM) is presented for an abyssal site, Station M in the northeast Pacific Ocean. The Δ14C values revealed that sinking POM at 3450 m depth (650 m above bottom) contained old carbon despite its presumed short transit time in the water column. The isotopic and chemical properties of the sinking POM varied with time and appear to be controlled by more than one major process. In 1993, 1994, and late 1996, isotopic signatures and C/N molar ratios indicate negligible or vertically homogeneous influence of resuspended particles from the bottom or particles laterally transported from the margin to the study site. However, during early 1995 and 1998, Δ14C values were lower than those during other periods and C/N values at three deep depths were not equal, indicating that the study site was influenced by resuspended sediments more severely than during other periods. During mid-1995 to mid-1996, δ13C values decreased abruptly while Δ14C values increased slightly, and C/N values were extremely high (up to ∼80) at 50 and 600 m above bottom; these results suggest input of degraded, modern, terrestrial organic matter. The periods of anomalous isotopic signatures, as well as vertically heterogeneous C/N values [Smith et al., 2001], were correlated with high discharge periods of California rivers with a time lag of 2 to 4 months. The correlation suggests that regional meteorological events are important in controlling the biogeochemical properties of particles at Station M by varying the intensity of resuspension and transport of organic matter from the continental margin.
 Comprehensive studies of biological, physical, and chemical properties in the water column and the sediment have been carried out at Station M since 1989 [Druffel et al., 1996; Smith et al., 2001]. Here, we report a 6-year time series of Δ14C, δ13C and C/N measurements of sinking POM collected as a part of the larger Station M studies [Smith et al., 2001]. We used the time series data to study carbon sources to sinking POM and processes that control the properties of sinking POM.
2. Study Site Characteristics and Sample Analyses
 Station M (34°50′N, 123°00′W, water depth of 4100 m) is located 50 km west of the base of the continental rise, 220 km off the coast of central California (Figure 1). The site lies within the southward flowing California Current in the surface. Near the coast, both the seasonal countercurrent at the surface in fall and winter and the subsurface undercurrent, mainly confined on the continental slope, flow northward along the California coast [Lynn and Simpson, 1987, and references therein].
 The flux of particulate mass at 600 m above bottom (mab) at Station M is positively correlated with an upwelling index [Bakun, 1973] that is a proxy for primary production, with a time lag of approximately 50 days [Baldwin et al., 1998]. This indicates that primary production in the overlying surface waters is the major source of the sinking POM. Input of resuspended SOM to POM at Station M is apparent from Δ14C measurements [Druffel et al., 1998], pyrophaeophorbide-a, a degradation product of chlorophyll-a [Bianchi et al., 1998], and the aluminum content in suspended POM [Sherrell et al., 1998]. The high C/N values of sinking POM observed at 50 mab suggest that terrestrial OC reaches Station M [Smith et al., 2001], despite its distance from the coast.
 Sinking particles were collected using a single conical sediment trap deployed at 650 mab along with two other sediment traps (600 mab and 50 mab [Smith et al., 2001]) on a single mooring. The benthic mixed layer has been reported to reach up to 80 mab with an average height of 45 mab at Station M [Beaulieu and Baldwin, 1998]. The 0.25 m2 opening of the cone was covered with a 1 × 1 cm baffle to reduce turbulent flow. Polyethylene collection bottles were attached to a sequencer at the bottom of the cone and programmed to sample in 10-day increments. The bottles were filled with seawater collected from the same depth that the sediment trap was moored. The water was filtered with a precombusted quartz filter (142 mm diameter Whatman ultrapure QMA, 0.8 μm pore diameter) and poisoned with mercuric chloride to a final concentration of 3 mM. The sediment traps were retrieved and redeployed every 13–14 weeks. Samples for carbon isotopic measurements were collected from July 1993 to the end of 1998 with a hiatus from October 1996 to November 1997. The sediment trap failed to collect samples from mid-June to late October 1994 because of presumed clogging due to extremely high flux [Baldwin et al., 1998].
 Retrieved sample bottles were kept on ice until processing. Swimmers, organisms that were alive when trapped, were removed manually using forceps from the sample under an illuminated magnifying glass. Each sample was filtered on precombusted quartz filters (Whatman QMA, 47 mm diameter, 0.8 μm pore size) by applying gentle vacuum. Multiple filter pads were used for large samples. Filtered samples were kept frozen at −20°C. The entire layer from a portion of the filter was scraped off to minimize sampling bias. Each sample was dried in an oven (<55°C for 12–24 hours), crushed in a glass vial with a glass rod and frozen until analysis. In order to determine the flux of sinking POC from our samples, it would have been necessary to dry all of the filter pads from a given sample; to facilitate future sampling, we instead left the remainder frozen. We used POC flux into the 600 mab trap published by Smith et al. .
 A CHN analyzer (Carlo Erba) was used for total carbon (organic plus inorganic carbon) and total nitrogen content determination. For organic carbon content, a few milligrams of each sample were acidified in a silver boat with incremental amounts of sulfurous acid (H2SO3, certified grade) until the total volume of acid was 400 μL [Verardo et al., 1989]. The acidified sample was dried after each addition of acid. After addition of about 150 μL, bubbles were not observed. Organic carbon content of each acidified sample was determined using the same CHN analyzer. The relative standard deviations (i.e., coefficient of variation = [1 standard deviation × 100]/mean) of duplicate sample analyses were 2% for organic carbon and 4% for total nitrogen.
 Organic carbon content of the 650 mab trap determined by this method agreed with that determined by manometric measurement of CO2 gas produced by combustion of samples for isotope measurements (the difference between two methods is 0.0 ± 0.3%, n = 125). The organic carbon contents of samples from 600 and 50 mab (previously published by Smith et al. ) were determined from the difference between total carbon content measured using a CHN analyzer and inorganic carbon content measured using a Coulometrics Carbon Analyzer. Despite the fact that organic carbon content was determined by two different methods, organic carbon to total nitrogen molar ratios (C/N) from the three trap depths are virtually identical during most of the study, with the exception of short periods during 1995 and 1996. Four samples from the 650 mab trap, including two samples collected when C/N values of 50 mab trap samples were the highest, were analyzed for total carbon and total nitrogen contents independently at the Virginia Institute of Marine Science. The C/N values calculated using these total nitrogen contents were not different from the University of California, Irvine, results within the analysis error (the difference was 0.6 ± 0.2).
 For isotope ratio determination of organic carbon, about 30 mg of each sample was weighed in a silver boat, put into a 9-mm OD quartz tube (Vycor brand), acidified with 1 mL of 3% phosphoric acid (certified grade), and allowed to stand overnight. The sample was then dried under vacuum with CuO, flame-sealed, and combusted at 850°C for 2 hours [Druffel et al., 1992]. Water vapor was removed from resultant CO2 gas cryogenically using isopropyl alcohol/dry ice traps. The volume of the gas was determined by measuring the pressure in a known volume. About 750 μgC of each CO2 sample was used for a Δ14C measurement, and about 100 μgC was used for a δ13C measurement. The CO2 gas was reduced to graphite on a cobalt catalyst using H2 gas at 580°C for 8 hours [Vogel et al., 1987]. Isotope ratio analyses were performed either at the National Ocean Sciences AMS Facility at Woods Hole Oceanographic Institution or the Keck Carbon Cycle AMS Laboratory at the University of California, Irvine. Reproducibility (1 standard deviation) of duplicate POM analyses was ±4–8‰ for Δ14C and ±0.1‰ for δ13C. The Δ14C and δ13C values of three samples from the 50 mab trap were also analyzed for comparison with the results from the 650 mab trap.1
3. Results and Discussion
3.1. POC Flux
 The POC fluxes into the 50 mab and 600 mab traps were reported by Smith et al. , and were temporally correlated with each other (r = 0.80, P < 0.001). For this study, we used the POC flux into the 600 mab trap as a proxy for the flux into our 650 mab trap to compare with our isotopic and chemical data.
 Briefly, POC flux reached a maximum of 26 mgC m−2 d−1 in June 1993 (Figure 2a). In summer 1994, a POC flux as high as in 1993 was observed in the 50 mab trap; unfortunately, the sediment traps at 650 mab and 600 mab failed to collect samples because of assumed “clogging” of the baffle by large-sized aggregated particle masses [Baldwin et al., 1998]. In 1995 and 1996, there were no prominent peaks in POC flux. In late 1997 and early 1998, POC flux remained low (<5 mgC m−2 d−1), likely due to a strong El Niño event that caused low productivity in the northeast Pacific [Chavez et al., 2002]. In late 1998, POC flux was high (11 mgC m−2 d−1) and was associated with a La Niña event [Chavez et al., 2002].
3.2. Δ14C Measurements of Sinking POM: Sources of Carbon
 The Δ14C values obtained for samples collected in 10-day intervals from June 1993 through December 1998 ranged from about −80 to +40‰ (Figure 2b). The Δ14C values in early 1995 and during 1998 were lower than for other periods. Large and rapid temporal variability was reflected at times with variations of larger than 16‰ (2 sigma) between two consecutive samples; for example, in April 1995, Δ14C values decreased by 70‰ from the previous 10-day sample.
 Most sinking POM samples had post-bomb Δ14C values (>−50‰) confirming a short turnover time (<30 years). However, Δ14C values of sinking POM are lower than those of plankton (59–81‰ [Wang et al., 1998]) and surface DIC measured at Station M from 1992 to 1998 (40–80‰ [Masiello et al., 1998; E. R. M. Druffel and S. Griffin, unpublished data, 2004]). These data indicate that plankton from the overlying water was not the only source of sinking POM. Sinking POM must have acquired old carbon, because aging during its vertical transport is far too small to account for this vertical gradient in Δ14C. A vertical gradient was also observed in Δ14C of suspended POM at Station M [Druffel et al., 1996] and at very remote locations where the influence of resuspended sediments was likely to be negligible [Druffel and Griffin, 1998]. The mechanisms suggested to explain this vertical gradient included sorption of DOM onto POM, incorporation of DOM by bacterial heterotrophy, and incorporation of DIC via anapleurotic reactions that produce organic compounds. Incorporation of DOM is likely responsible, at least partly, for low Δ14C values of sinking POM at Station M [Hwang, 2004, chap. 6].
 If Δ14C of sinking POM reflects mainly the temporal variability of Δ14C of DIC in surface waters, then POM Δ14C is expected to be high in El Niño years and low in non-El-Niño years. Upwelling of low Δ14C subsurface water is suppressed during El Niño years, resulting in surface waters with Δ14C values that are higher than during non-El-Niño years. However, a correlation between the Δ14C values of sinking POM and El Niño-Southern Oscillation at this site is not apparent from these data. In contrast, sinking POM Δ14C values were lower in 1998, a strong El Niño year, than during earlier years. This indicates that the original Δ14C signatures were altered significantly after POM left the surface. The variability of sinking POM Δ14C values (about 80‰) is greater compared to those of DIC in surface waters (∼40‰) [Masiello et al., 1998; E. R. M. Druffel and S. Griffin, unpublished data]. This suggests that processes exist to cause larger variability in sinking POM Δ14C other than physical processes at the surface, associated with changes in water masses.
 One potential carbon source to sinking POM is organic matter that is advected laterally to the Station M water column. Lateral advection of resuspended sediment from the California margin has been documented by several investigators [Druffel et al., 1998; Sherrell et al., 1998; Bianchi et al., 1998; Smith et al., 2001]. The Δ14C values of the surface sediments from the continental rise and slope are around −250‰ [Druffel et al., 1998]. The 80‰ variability in sinking POM Δ14C (−40 to 40‰) can be accounted for by incorporation of less than 30% resuspended SOM into sinking POM using simple isotopic mass balance (40‰ × (1 − x) + (−250‰) × x = −40‰, where x = the fraction of carbon from DOM in sinking POM = 0.28).
 Laterally transported material may contain young terrestrial OC and/or old carbon from eroded bedrock from rivers. Rivers near Station M (except the San Joaquin River) do not have well-developed estuaries and therefore export particles directly into the coastal ocean. A fraction of riverine POM may be exported as suspended POM beyond the continental shelf and then incorporated into sinking POM. During the flood of the Eel River in January 1995, only 25% of the exported sediment formed a recognizable deposit on the shelf [Wheatcroft et al., 1997], implying that the other portion may have been dispersed over other parts of the continental margin. The Salinas, the Santa Ynez, and the Santa Clara Rivers are characterized as small mountainous rivers [Inman and Jenkins, 1999] and expected to export low Δ14C OM because of a large inclusion of relict OM [Kao and Liu, 1996; Masiello and Druffel, 2001; Blair et al., 2003; Komada et al., 2004]. As an example, Komada et al.  showed that suspended POM (Δ14C values ranging from −232 to −544‰) in the Santa Clara River stream near the ocean was a mixture of ancient shale OM (Δ14C values <−885‰) and modern soil OM (Δ14C ranging from 45 to 122‰ at the surface). Sediments on the California shelf near the Santa Clara and the Santa Ynez Rivers appear to contain a significant amount (>20% of lipids and the acid-insoluble organic fraction) of old carbon from those rivers [Hwang, 2004]. Resuspended SOM from the shelf may also contain petroleum from natural seeps as sources of low Δ14C OM [Wilson et al., 1974; Bauer et al., 1990].
3.3. Potential Causes of Variability in δ13C Values for Sinking POM
 The δ13C values of sinking POM ranged from −23.5 to −20.5‰ (Figure 2c). Except for the two short periods in mid-1993 and mid-1995, δ13C values varied by less than 2‰. The degree of δ13C fractionation during photosynthesis is a function of environmental factors, primarily pCO2 in seawater and available light [Rau et al., 1992; Thompson and Calvert, 1994; Laws et al., 1995; Rost et al., 2002]. Because primary production affects pCO2 levels near the producers, there is a positive correlation between δ13C values of the phytoplankton and primary production (and therefore, sinking POC flux) [Deuser et al., 1968; Deuser, 1970; Rau et al., 1992]. The Station M data show a correlation (r = 0.76, P < 0.001, n = 69) between sinking POC flux and δ13C values only during 1993, late 1994, and late 1996; this implies that during other periods, either pCO2 was not depleted enough to cause detectable enrichment of 13C or δ13C values were modified by other processes. An abrupt decrease in δ13C in summer 1995 that did not accompany significant change in POC flux is not likely to be caused by biological activity in surface waters.
 Sources other than plankton-derived POM from the overlying surface waters are considered to explain the variability of δ13C values that is not correlated with POC flux. Unlike Δ14C values, input of organic carbon from DOM, resuspended SOM, and relict OM from rivers will not cause significant change in POM δ13C, because their δ13C signatures are similar to those of sinking POM (−21 to −23‰ [Bauer et al., 1998; Druffel et al., 1998]). However, terrestrial OC from C-3 plants has significantly lower δ13C values (−22 to −34‰ [Trumbore and Druffel, 1995; Meyers and Lallier-Vergès, 1999]); therefore incorporation of terrestrial organic carbon may be a cause of the variability.
3.4. Molar Ratio of Organic Carbon to Total Nitrogen (C/N)
 Molar C/N ratios of sinking POM at 650 mab at Station M were constant, ∼10 during all sampling periods (Figure 2d). Generally, C/N values of sinking POM in the deep ocean are higher than the Redfield ratio for phytoplankton [Knauer et al., 1979; Loh and Bauer, 2000], and our data are consistent with this observation. Selective degradation of N-rich proteins in the water column may be a reason for this difference.
 The C/N values at 650 mab (this study) were compared with those of Smith et al.  for 600 and 50 mab (Figure 2d). In 1993 and 1994, C/N values from all depths (10 ± 2) were similar, indicating that input of allochthonous OM was either negligible or affected all three traps equally. However, C/N values at 600 mab and 50 mab show dramatically different patterns from those at 650 mab in other periods. In early 1995, C/N values at 600 mab (9 ± 2) and 50 mab (7 ± 1) were slightly lower than those at 650 mab (11 ± 1). In July–November 1995 and January–May 1996, C/N values of POM at 50 mab and 600 mab were extremely high (25–80), while those at 650 mab remained unchanged (9 ± 1). Conversely in 1998, C/N values at 600 and 50 mab (12 ± 1) were slightly higher than those at 650 mab (10 ± 1). Extremely high C/N values also suggest that there were sources of OC other than plankton-derived OC from the overlying waters. The variation of the difference in C/N values among the traps suggests that the height of vertical reach of the allochthonous OC varied with time.
3.5. Periods of Anomalous Properties of Sinking POM
 The periods during which C/N values at three depths disagree coincide with those when Δ14C and/or δ13C values at 650 mab are anomalous. In accordance with the deviation of Δ14C, δ13C, and C/N values, the time series was divided into four periods (Figures 2 and 3: (1) normal periods: March 1993 to June 1994, November 1994 to February 1995, and May 1996 to October 1996; (2) period A: February 1995 to May 1995; (3) period B: May 1995 to May 1996; and (4) period C: December 1997 to December 1998. The normal period is defined as that when δ13C values are linearly correlated with sinking POC flux, and C/N values at all three depths are equal (Figure 3). The Δ14C and δ13C values for the 50 mab samples (red squares in Figures 2b and 2c) were similar to those of 650 mab samples only in the normal period (July 1996). Sinking POM in this period appears minimally influenced by allochthonous OC, and is therefore assumed to be representative of open ocean POM. Period A is defined by low and variable Δ14C values (−25 ± 33‰) and significant variability in C/N among samples at three depths (Figure 3b). Period B is defined by low δ13C values (−22.6 ± 0.5‰), normal or slightly elevated Δ14C (18 ± 15‰), and large differences in C/N values among samples at three depths. Period C is similar to period A except that C/N values at 650 mab were lower than those at 600 and 50 mab (Figure 2d).
 We considered three processes that might be responsible for the variability of the isotopic signatures and the C/N values of sinking POM during these periods: (1) input of resuspended SOM from local or shelf/slope sources; (2) input of old OM from land via rivers; and (3) input of young OM from terrestrial plants via rivers. Resuspended SOM may also contain OC from small mountainous rivers as discussed in section 3.2. Previously reported Δ14C, δ13C, and C/N values of the potential organic carbon sources at locations close to the study site, and the direction of change of these properties when sinking POM includes organic matter from these sources, are listed in Table 1.
Table 1. Previously Reported Δ14C, δ13C, and C/N Values of the Organic Matter Sources at Locations Near the Study Site and the Direction of Change (in Parentheses) of These Properties When Sinking POM Incorporates Organic Matter From These Sources
 The rapid decrease in Δ14C values during period A requires one or more processes that can occur over a 10-day time period. Incorporation of resuspended SOM is one possible mechanism, because of the low Δ14C values (−250‰) for SOM but similar δ13C values in SOM relative to sinking POM. The C/N value for SOM was around 10 in surface sediment at Station M in 1995 [Smith et al., 2001] and around 8 in a slope site in 2001 [Hwang, 2004].
 During period B, C/N values at 50 and 600 mab were extremely high, while those at 650 mab remained unchanged (9 ± 1). In the mid-July 1995 sample, when the C/N value at 50 mab was the highest (85), the δ13C value decreased from −21.9 to −23.6‰, while Δ14C values increased by 20‰. This δ13C value is the lowest observed during the entire study period. High C/N and high/normal Δ14C values combined with low δ13C values indicate the presence of modern, terrestrial organic matter. The isotopic signatures of sinking POM from the 50 mab trap when C/N values were the highest (June 1995) showed an even stronger signal (red squares in Figures 2b and 2c). The Δ14C and δ13C values were 66‰ and −24.0‰, respectively.
 The dominant terrestrial vegetation (>99%) in most California watersheds is C-3 plants [Still et al., 2003]. The δ13C values of surface soil OC in the Santa Ynez basin ranged from −27 to −28‰ [Komada et al., 2004]; thus incorporation of terrestrial OC into sinking POM will lower δ13C values. The small mountainous rivers are not likely the source of the modern terrestrial OC, unless plant debris is separated from relict OC during transport due to the differences in their physical properties. However, the San Joaquin River has a relatively large drainage basin (19,024 km2), 32% of which is cropland and pasture [Saleh et al., 2003] and is likely to export modern plant material during the rainy seasons.
 It is interesting that the C/N values of POM at 650 mab remained constant (∼9) during Event B, and did not appear to reflect the input of terrestrial OC as did the isotopic signatures and the C/N values of the samples from the deeper traps (Figure 2d). However, C/N is not a conservative property in the sense that C and N can be degraded at different rates. In addition, C/N values cannot be treated linearly as in a simple mass balance equation. For example, a 50:50 mixture by weight of two POM pools with the same OC content but different C/N values (80C/1N versus 80C/10N) will have a C/N value of 14.5 instead of 44. Therefore, unlike isotopic signatures, the change of C/N by mixing with other POM may be small in comparison.
 Period C was similar to period A, except that C/N values at 50 and 600 mab were higher than those at 650 mab during this period. Low Δ14C (−20‰) and normal δ13C values of sinking POM may suggest incorporation of resuspended SOM into POM. The C/N values of resuspended SOM may also vary with time. For example, Smith et al.  reported an increase in C/N from 8 in 1990 to about 12 in 1996, and the highest value (∼19) was observed in 1998 in SOM at Station M. Therefore the same process of SOM incorporation as in period A may have caused the observed vertical distribution of C/N values.
3.6. Potential Mechanisms of Lateral Transport of Resuspended SOM
Smith et al.  reported previously that temporal peaks in C/N at 50 mab were correlated with the San Joaquin River discharge with a time lag of about 4 months. We compared periods of high discharge rates of four rivers close to the study site with the periods of anomalous properties of sinking POM (Figure 2e). Discharge rates in 1992/1993 winter were relatively low, and only a small difference in C/N values at 650 and 600 mab was observed in early 1993. In winter 1993/1994, river discharge rates were very low, and our data show negligible input of resuspended SOM. The high river discharge rates in 1995 and in 1998 correlate with anomalous properties of sinking POM with a time lag of 2 to 4 months. Although a cause-effect relationship between river discharge and anomalous properties of sinking POM cannot be confirmed at this point, river discharge is apparently related with the driving force of lateral transport of resuspended SOM and/or riverine POM. Regional weather may thus be an important controlling factor influencing the properties of sinking POM at Station M. Winter storms may enhance resuspension of sediments; however, wind speed recorded at a buoy station (National Data Buoy Center, station 46023, data not shown) during the study period did not show a clear pattern of high wind in winter.
 The mechanism of lateral transport, especially to the deep water column, in the study area is not well understood. One potential mode of lateral transport to Station M is lateral dispersion along isopycnal surfaces, and then vertical transport via gravity. Another potential mode is lateral flow along the ocean floor as a thick nepheloid layer. When river water contains a high enough particle load its density may exceed that of seawater, and a hyperpycnal plume or margin plume can develop [Imran and Syvitski, 2000; Wheatcroft, 2000]. Hyperpycnal plumes have been observed from small mountainous rivers such as the Eel River and the Santa Clara River [Wheatcroft et al., 1997; Mertes and Warrick, 2001; Warrick, 2002]. Warrick  showed that over 99% of the discharge by the Santa Clara River might be hyperpycnal. He also showed schematically that thick bottom nepheloid layers initiated by hyperpycnal plumes could move farther to the outer shelf than the hyperpycnal plumes [Warrick, 2002]. The signal of resuspended SOM or terrestrial OM was strongest in the 50 mab trap, suggesting that this mode of lateral advection is a more plausible mechanism of OM transport to our site.
4. Implications for the Carbon Cycle
 It has been suggested that a large fraction of OM produced on continental shelves may be transported to slope waters and sediments [Walsh et al., 1981; Bauer and Druffel, 1998; Bauer et al., 2001]. The results of programs such as SEEP I and II suggested that a very small fraction (much smaller than 5%) of the primary production on the eastern U.S. shelf was exported to the continental slope [Biscaye et al., 1994]. However, other reports suggest that a significant amount of OC (9 ± 6% of primary production of the coastal ocean) is transported from the continental margins to the ocean basin [Bauer and Druffel, 1998; Bauer et al., 2001; Liu et al., 2000, and references therein]. Sediment reworking by resuspension on the Southern California Shelf was indicated by low Δ14C values of SOM in spite of high sedimentation rates [Santschi et al., 2001; Hwang, 2004]. Higher OC content (3%) in the sediment was observed at a continental rise site (3800 m water depth) than at other sites of different water depths between Station M and the California coast (4100, 2000, 500, 200, and 100 m water depths [Druffel et al., 1998; Hwang, 2004]). This site may be a deposition center of OC, implying that OC transport from the shelf is a persistent phenomenon in the continental slope near Station M.
 Nearly 20% of oceanic primary production takes place in coastal oceans [Martin et al., 1987; Hedges, 1992; Liu et al., 2000], and this value is expected to increase because of human activities, such as enhanced input of nutrients. Sediment resuspension and transport appear to be enhanced by coastal storms and river discharge in winter seasons in the California coast. The correlation between OC transport and weather implies that OC transport may be different in the future as a response to the global climate change. Our results imply that change in OC transport from the margins to the deep ocean in relation with regional weather should be considered for better prediction of future change in global carbon cycling.
 The transport of OC has an implication for paleoclimatology as well. Interpretation of paleo-temperature proxies requires that the compounds in a sediment horizon have the same temporal and spatial origin. However, our results suggest that care should be taken in interpreting the paleochronological data. Displacement by resuspension, advective transport, and then deposition of pre-aged organic compounds can cause large offset in the ages between time proxies and temperature proxies [Ohkouchi et al., 2002].
 We thank John Southon, Guaciara dos Santos, and Xiaomei Xu at the Keck Carbon Cycle AMS Laboratory at UC Irvine, and Ann McNichol, Al Gagnon, and John Hayes at NOSAMS, WHOI, for Δ14C and δ13C measurements; Carrie Masiello, Steven Beaupré, and Ed Keesee for CHN analyses; Sue Trumbore and Shuhui Zheng for shared resources; Carrie Masiello, Rob Glatts, Fred Uhlman, Bob Wilson, and the resident technician group at SIO for help with sampling; Sue Trumbore, Tomoko Komada, Angelo Carlucci, and three anonymous reviewers for helpful comments on the manuscript; Captains and crew members of the R/V New Horizon and the R/V Atlantis II. This research was supported by NSF OCE Chemical Oceanography Program (to E. R. M. D. and J. E. B.), Biological Oceanography Program (to K. L. S.), and the UCOP Marine Science Fellowship Program (to J. H.).