Ventilation changes in the western North Paciﬁc since the last glacial period

. We reconstructed the ventilation record of deep water at 2100 m depth in the mid-latitude western North Paciﬁc over the past 25 kyr from radiocarbon measurements of coexisting planktic and benthic foraminiferal shells in sediment with a high sedimentation rate. The 14 C data on fragile and robust planktic foraminiferal shells were concordant with each other, ensuring high quality of the reconstructed ventilation record. The radiocarbon activity changes were consistent with the atmospheric record, suggesting that no massive mixing of old carbon from the abyssal reservoir oc-curred throughout the glacial to deglacial periods.


Introduction
The atmospheric CO 2 content during glacial periods was about 80 ppm lower than the pre-industrial level (Monnin et al., 2001). During the early phase of the deglacial period between 17.5 and 15 kyr BP, the ratio of the radionuclides 231 Pa and 230 Th in northern Atlantic sediments suggests a slowdown of the Atlantic Meridional Overturning Circulation (AMOC), triggered by the massive discharge of fresh water to the North Atlantic known as Heinrich Event 1 (H1) (McManus et al., 2004). Because of a 190 ‰ drop in the 14 C/ 12 C ratio in the atmosphere and an atmospheric CO 2 rise of 40 ppm during H1, renewal of the isolated carbon reservoir in deep water is thought to be linked to reorganizations in AMOC (Denton et al., 2006;Broecker and Barker, 2007).
Recently,  examined a compilation of radiocarbon records and modelling simulations and suggested that deep water was formed in the North Pacific extending to a depth of ∼2500 m during H1. The main simulated pathway of deepwater spreading is along the western margin of the North Pacific, in a deep western boundary current analogous to the present one in the North Atlantic. However, our knowledge of paleo-ventilation, particularly in water deeper than 2000 m in the western North Pacific, is limited because of poor preservation of foraminiferal shells in sediment. Here we present a detailed account of ventilation changes in the mid-latitude western North Pacific based on radiocarbon records from coexisting planktic and benthic foraminifera in sediment with high sedimentation rates. Because our ventilation reconstruction is based on radiocarbon data from multiple planktic species in the mid to highlatitude western North Pacific, our record provides robust evidence for the ventilation history in the region.

Sediment samples
Giant piston core MD01-2420 was obtained from the western North Pacific off Japan (36 • 04 N, 141 • 49 E; water depth 2101 m; Fig. 1; Table 1) during the International Marine Past Global Change Study (IMAGES, http://www.images-pages. org/), Western Pacific Margin (WEPAMA) 2001 Cruise of the R/V Marion Defresne. Stormy conditions led to limited core recovery of 8.99 m. Sediments in the core are composed mainly of dark olive-grey homogeneous silty clay, and no sediment disturbance is apparent. In order to evaluate the core quality, Sagawa et al. (2006) confirmed the high quality of the core by comparing the sediment colour and planktic foraminiferal δ 18 O between this core and MD01-2421, another IMAGES piston core adjacent to the MD01-2420 site with well-established age control (Oba and Murayama, 2004;Oba et al., 2006).

Radiocarbon measurements
We extracted sediment samples 2 cm thick at 15 horizons from the upper 6 m of core MD01-2420. Samples were washed on a 63 µm mesh sieve and dried in an oven at 60 • C. Coexisting planktic foraminifera (Globigerina bulloides and Globorotalia inflata) and benthic foraminifera (Uvigerina spp. and mixed species) were used for radiocarbon dating. We picked foraminiferal shells from the >250 µm fraction of each sample under a stereomicroscope. When the numbers were insufficient, we also picked shells from 125-250 µm fractions. If the amount of shells was still insufficient, we picked additional shells from adjoining sediment samples.
The foraminiferal shells were cleaned by soaking them in 99.5 % methyl alcohol, followed by ultrasonication until all chambers were open. After confirming that all dirt had been removed, we washed the shells in Milli-Q water and dried them in an oven at 40 • C. 14 C ages were measured by accelerator mass spectrometry (AMS) at the National Ocean Sciences AMS facility (NOSAMS) at Woods Hole Oceanographic Institution (Table 2). Three 14 C ages of G. inflata were measured at the Center for Chronological Research, Nagoya University, Japan (Table 2; Sagawa et al., 2006).
We converted radiocarbon ages of the planktic foraminifera samples to calendar ages by using the Calib 6.0 program with the Marine09 calibration dataset (Reimer et al., 2009). The regional marine reservoir age ( R) is defined as the deviation of the local radiocarbon age from the globally averaged reservoir age (∼400 yr). For core MD01-2420, we chose R = 100 ± 200 yr based on Shishikura et al. (2007) and Yoneda et al. (2007) where 8033 and 8266 are the Libby and true radiocarbon mean-life in years, respectively (Adkins and Boyle, 1997).

Evaluation of radiocarbon data
Bioturbation creates major biases when we try to reconstruct past ocean ventilation based on radiocarbon age differences between co-existing planktic and benthic foraminiferal shells (Broecker et al., 1984). Hence, samples from sediments of high sedimentation rate are required. Because such sediments are found near the continental slopes, we must rule out the presence of reworked materials (Broecker et al., 2004a). To test these biases, radiocarbon measurements on multiple planktic foraminiferal species employing fragile and robust species are effective (Broecker et al., 2004a. In core MD01-2420, we used two planktic species: Globigerina bulloides with relatively fragile shells and Globorotalia inflata with robust shells from eight intervals ( Table 2). The planktic species yielded closely matching 14 C ages (except in the 390.1 cm interval) and the sedimentation rate was high (∼30 cm kyr −1 ) without age reversal, suggesting that core MD01-2420 was appropriate for reconstruction of past ventilation (Tables 2 and 3; Fig. 2). In the sample from the 390.1 cm interval, the 14 C age of G. bulloides was 600 yr older than that of G. inflata, implying considerable contamination by reworking. We selected the 14 C age of G. inflata for the 390.1 cm interval because the younger radiocarbon age was closest to the true age of the sediment (Broecker et al., 2004a).

Ventilation changes in core MD01-2420
Radiocarbon age differences between co-existing planktic and benthic foraminiferal shells (B-P age) indicate apparent    ventilation ages in the past. B-P ages in core MD01-2420 ranged from 1150 to 1550 yr during the last 25 kyr (Fig. 3). Because the weighted average variance was 1360 ± 140 yr, the apparent ventilation ages in core MD01-2420 showed no significant changes within the measurement uncertainties. Thus, there is no sign of intrusions of anomalously old water masses at the MD01-2420 site throughout the last 25 kyr, which is consistent with previous studies (Broecker et al., 2004b(Broecker et al., , 2008.
Because we had to take into account large atmospheric 14 C changes, including a 190 ‰ drop between 14.5 and 17.5 kyr BP (Broecker and Barker, 2007), we calculated 14 C of benthic foraminiferal shells in core MD01-2420 using Eq. (1) ( Table 4; Fig. 4). Our 14 C record changed in concert with previously published atmospheric (Intcal09) and tropical surface ocean (Marine09) curves (Reimer et al., 2009) throughout the last 25 kyr. We compiled western North Pacific 14 C records calculated by using published radiocarbon ages from eight cores ranging in water depth from 900 to 2800 m (Table 1; Fig. 1; Duplessy et al., 1989;Murayama et al., 1992;Ahagon et al., 2003;Broecker et al., 2004bBroecker et al., , 2008Minoshima et al., 2007;Sagawa and Ikehara, 2008). These records were consistent with the 14 C record in MD01-2420 (Fig. 4). The western North Pacific 14 C records co-varied with atmospheric 14 C changes during the last glacial to deglacial periods, which is in clear contrast with data from the eastern Pacific showing that very old intermediate water masses observed at two sites off Baja California (Marchitto et al., 2007) and near the Galapagos Islands (Stott et al., 2009) during the last deglacial period.   Figure 5 shows a comparison of glacial-deglacial changes in 14 C records at 2101 m at core MD01-2420, at 3647 m water depth in the Gulf of Alaska at ODP Site 887 (Galbraith et al., 2007), and at 2710 m water depth in the eastern North Pacific off the Oregon coast at W8709A-13PC (Mix et al., 1999;Lund et al., 2011). During the last glacial maximum between 18 and 21 kyr BP, the difference in 14 C between MD01-2420 and ODP 887 was 76 ± 34 ‰ (weighed average). During early H1 between 17 and 17.5 kyr BP, the difference increased to 142 ± 47 ‰. Note, however, that the comparison is based on a single data point in core ODP 887. After the Bølling interstadial (∼14.5 kyr BP), there was no significant 14 C difference between ODP 887 and MD01-2420, which was comparable to the present 14 C distribution in the North Pacific (Key et al., 2004). 14 C in core W8709A-13PC were lower than that in core MD01-2420 throughout the glacial-deglacial period (Fig. 5). This feature is consistent with the present 14 C distribution in the North Pacific (Key et al., 2004). During the glacial to early H1 interval, the 14 C in core W8709A-13PC changed consistently with that in core ODP 887. On the other hand, 14 C in core W8709A-13PC appears to be lower than that of core ODP 887 during the Bølling-Allerød (∼13.0-14.5 kyr BP). These 14 C 14 C change in core MD01-2420 (this study), core ODP 887 in the Gulf of Alaska (3647 m, Galbraith et al., 2007) and core W8709A-13PC in the eastern North Pacific (2710 m, Mix et al., 1999;Lund et al., 2011) between 10 and 23 kyr BP along with Int-cal09 and Marine09 curves (Reimer et al., 2009). variations suggest a reorganization of water-mass structure in the North Pacific during the deglacial period from a stratified glacial mode to an upwelling interglacial mode. The glacial Pacific Ocean had two water masses: well-ventilated and nutrient-depleted glacial North Pacific Intermediate Water (GNPIW) above ∼2000 m and less-ventilated and nutrientenriched deep water below ∼2000 m (Keigwin, 1998;Matsumoto et al., 2002). GNPIW is a thicker and more deeply penetrating water mass than today's North Pacific Intermediate Water (NPIW). Although a water mass extending to 2000 m should not be called an intermediate-water in the sense of physical oceanography (Matsumoto et al., 2002), we use the term "GNPIW" to refer to the well-ventilated water mass following Matsumoto et al. (2002). The source of GNPIW was possibly in the Bering Sea, given microfossil  and neodymium isotope evidence (Horikawa et al., 2010). During H1 in the early deglacial period, deep water extending to a depth of ∼2500 m formed in the North Pacific, regarded as an intensified NPIW (Ohkouchi et al., 1994;. This deep water may have yielded the large 14 C differences during H1 between MD01-2420 (2101 m) and ODP 887 (3647 m) by enhancing ventilation in the intensified GNPIW during the early H1 between 17 and 17.5 kyr BP (Fig. 5). Since the Bølling-Allerød period, ocean circulation in the North Pacific has been in an interglacial mode, essentially the same as the present one. The present abyssal circulation from the south flows into the North Pacific, and upwells to mid-depth and returns south as the Pacific Deep Water (PDW) (Schmitz, 1996). Above the PDW, the NPIW with a salinity minimum lies at depths of 300 to 800 m (Talley, 1993). The main pathway of deepwater is along the western margin of the ocean basins, as a deep western boundary current. In the North Pacific, the western boundary flow is a principal factor in establishing the east-west gradient of deep-Pacific ventilation, which is found in sedimentary 14 C records in cores MD01-2420 and W8709A-13PC. However, this gradient is still too weak to explain the very old intermediate water in the eastern Pacific during the last deglacial period (Marchitto et al., 2007;Stott et al., 2009). Obviously, more sedimentary ventilation records are required to reconstruct the spatial and temporal ventilation change in the North Pacific. In particular, the following three regions are potential candidates to be tackled:

Ventilation history and water mass structure change in the North Pacific
(1) the Bering Sea, as a possible ventilation source during glacial-deglacial periods; (2) the western subarctic Pacific off the Kamchatka Peninsula, principal pathway of deep water flow; and (3) mid-depth of the eastern North Pacific to fill the gap between 1000 and 2700 m water depths.

Implications for the release of old carbon from the deep sea during the last glacial termination
During the H1 period, old carbon must have been released from the abyssal reservoir (Broecker and Barker, 2007), probably from the Southern Ocean (Skinner et al., 2010). At the same time, very old intermediate water masses existed in the eastern Pacific (Marchitto et al., 2007;Stott et al., 2009). However, there is no sign of an old carbon release in the Antarctic Intermediate Water pathway (De Pol-Holz et al., 2010;Rose et al., 2010). Recently, Hain et al. (2011) pointed out that the 14 C anomalies in the intermediate water are not basin-scale but local phenomena. In the western North Pacific, there is no 14 C anomaly between 900 and 2800 m relative to the atmospheric 14 C change. During the early H1,  suggested that well ventilated deepwater extending to ∼2500 m, originated from the North Pacific, flowed southward along the western margin of the North Pacific. We suggest that the deepwater from the North Pacific may have helped produce the local mid-depth 14 C anomalies in the eastern equatorial Pacific by flushing a part of old deep Pacific water. From the abyssal North Pacific, old carbon release was suggested at the beginning of the Bølling-Allerød (Galbraith et al., 2007). Relatively low 14 C at core W8709A-13PC during the Bølling-Allerød might be caused by an influence of the aged abyssal water. Major reorganization of ocean circulation in the North Pacific during the glacial-deglacial period affected productivity through upwelling. During the last glacial maximum, primary productivity in the subarctic Pacific was low because of stratification (Narita et al., 2002;Jaccard et al., 2005;Galbraith et al., 2007;Brunelle et al., 2010). Thick GNPIW with low nutrient concentrations down to ∼2000 m suppressed biological productivity in the euphotic layer. This stratification temporarily intensified during H1 because of the expansion of the nutrient-depleted water mass down to ∼2500 m in the North Pacific . At the beginning of the Bølling, productivity in the subarctic Pacific rose rapidly (Crusius et al., 2004;Galbraith et al., 2007;Jaccard et al., 2009;Brunelle et al., 2010;Davies et al., 2011) in association with enhanced upwelling by breakdown of the glacial stratification.

Conclusions
We measured radiocarbon ages of multiple planktic and benthic foraminiferal species in sediment core MD01-2420 obtained at 2100 m from an area with a high sedimentation rate in the western North Pacific. The reconstructed ventilation history of the western North Pacific demonstrates changes consistent with the radiocarbon activity of the atmosphere, suggesting no sign of massive mixing of old carbon from the abyssal reservoir throughout the glacial to deglacial period. Comparison of 14 C records between cores MD01-2420, ODP 887 (Gulf of Alaska, 3647 m, Galbraith et al., 2007) and W8709A-13PC (eastern North Pacific, 2710 m, Mix et al., 1999;Lund et al., 2011) suggests a reorganization of water-mass structure in the North Pacific during the deglacial period from a stratified glacial mode, with two water masses bounded at 2000 m, to an upwelling interglacial mode to an upwelling interglacial mode during the last deglacial period. The western boundary flow appears to be a principal factor for the east-west gradient of the North Pacific ventilation, yielding horizontal 14 C anomalies during the deglacial reorganization.