The sediment core St10-PC (29°28′N, 133°37′E) (length = 647 cm) was collected at a water depth of 2670 m during the R/V Hakuho-maru cruise of KH16-6 in the northern Kyushu-Palau Ridge of the Northwest Pacific (Figure 1A). The station is located at approximately 600 km east of Kyushu Island and experiences frequent volcanic eruptions. The sediments of core St10-PC mainly consist of homogeneous brownish-gray silty clay with several volcanic ash layers. A total of 284 samples were delineated at approximately 2 cm intervals for the analyses and stored at −20°C until analysis.

The mass accumulation rate (MAR), an index to refer to the sediment fluxes, was calculated by the linear sedimentation rates (LSR) and the dry bulk density (DBD) of the sediment using the following Eqs 1, 2, then the burial fluxes of TOC were determined using Eq. 3: L S R   ( c m / k y r ) = Δ h / Δ t (1) M A R   ( m g / c m 2 / k y r ) = L S R ( c m / k y r ) ∗ D B D ( g / c m 3 ) ∗ 1000 (2) T O C f l u x e s ( m g / c m 2 / k y r ) = T O C   ( % ) ∗ M A R (3) where Δh and Δt are the depth and age differences, respectively, between the two age control points.

The biogenic calcium carbonate (CaCO₃) content as determined by inorganic carbon was calculated using Eq. 4: C a C O 3   ( % ) = ( T C − T O C ) ∗ 8.33 ∗ 100 % (4)

The biogenic barium ( B a b i o ) content as determined by the major element barium was calculated using Eq. 5: w ( B a b i o ) = B a t o t a l − A l t o t a l ∗ ( B a / A l ) t e r r (5) where w ( B a b i o ) is the mass fraction of biogenic barium in the sediments (μg/g); Ba_total, Al_total, and (Ba/Al)_terr are the total barium and aluminum contents in the sediments (μg/g) and terrestrial Ba/Al ratio (0.0041, Shao et al., 2016), respectively.

The end-member statistical analysis employed a MATLAB modeling algorithm (Dietze et al., 2012) to analyze individual grain-size populations (end-member loadings, volume %) and their contributions to bulk grain-size composition (scores, %) (Zhou et al., 2015).

In the Northwest Pacific, definite age models of long-term sedimentary cores generally use a comparison between the δ¹⁸O of planktonic foraminifera and the LR04 δ¹⁸O stack of benthic foraminifera in a specific sediment core (Ikehara et al., 2009; Sun et al., 2011; Ujiié et al., 2016). An age model for core St10-PC was reconstructed based on 11 age control points (Table 1) provided by the AMS¹⁴C dating results and the G. inflata-δ¹⁸O records in this core (Figure 2C, data are available in the supplementary dataset), and correlating this G. inflata-δ¹⁸O records to the standard deep-sea oxygen isotopic LR04-δ¹⁸O stack (Figure 2E; Lisiecki and Raymo, 2005) and G. ruber-δ¹⁸O records in core MD06-3050 from near Luzon Island in the Philippine Sea (Figure 2D, Sun et al., 2011). This age model indicates that core St10-PC covers past ∼380 kyr (Figure 2), spanning marine isotope stages (MIS) 1–10, which cover nearly five complete glacial-interglacial cycles (Figure 2).

According to this chronology, the average LSR was lower than those in other sediment cores from marginal seas (Shao et al., 2016), varying over the range between 0.5 (in Holocene) and 3.6 (in MIS2) cm/kyr, with an average value of 1.8 cm/kyr (Table 1; Figure 2A). The MAR varied from 353.6 to 3121.8 mg/cm²/kyr, with average value of 1484.5 mg/cm²/kyr (Table 1; Figure 2B). Glacial-interglacial differences in LSR and MAR were evident with higher average values in glacial periods than that in interglacial periods (Figures 2A,B), which is thought to primarily reflect the increased terrestrial input (dust input and fluvial discharge) when sea level declined (Figure 2F, Waelbroeck et al., 2002).

To estimate the TOC-fluxes on orbital scale, we combined the MAR data in this study with the previously reported TOC abundances (Zhang et al., 2020). The TOC-fluxes varied from 0.8 to 11.7 mg/cm²/kyr (Figure 3F), with a mean value of 3.2 mg/cm²/kyr. Furthermore, the peaks in TOC contents correlated well with their fluxes (Figures 3F,G), showing significant cyclic variations that were almost two times higher during glacial periods (particularly in MIS2, p < 0.01, Supplementary Table S1), which is consistent with hundreds of sedimentary OC burial records around the world (Cartapanis et al., 2016). The TN contents and fluxes (Supplementary Figures S1B,C) varied in the similar tendencies as TOC. However, the TOC vs. TN scatter plot (Supplementary Figure S2) shows an intercept of 0.026% inorganic nitrogen, which accounts for ∼58% of the mean total nitrogen and results in low TOC/TN values downcore, with a range of 2.34–8.47 (Supplementary Figure S1A). δ¹³C_org in core St10-PC varied between -25.24‰ and −21.04‰ (mean: 22.59‰) (Figure 3E), trending to more positive during the glacials (p < 0.01, Supplementary Table S1). The overall more positive δ¹³C_org values may indicate sedimentary OC mainly derived from marine-sourced contributions (Ikehara et al., 2009; Li et al., 2018).

The abundances of CaCO₃ and B a b i o in marine sediments have been widely used as paleo-productivity proxies (Agnihotri et al., 2003; Shao et al., 2016). In this study, CaCO₃ and B a b i o contents in core St10-PC ranged from 8.90 to 57.46% and 176.58–512.09 μg/g respectively, with average of 35.56% and 345.68 μg/g (Figures 3B,D). Downcore CaCO₃ (p < 0.05, Supplementary Table S1) and B a b i o (p < 0.01, Supplementary Table S1) were significantly higher during interglacials than glacials, which is consistent with the Ca/Ti and Ba/Al ratios (Figures 3A,C), but reversed compared to the OC fluxes (Figure 3F).

In this study, the concentrations of trace elements sensitive to the redox conditions of the hydrological regime, such as uranium (U), vanadium (V), molybdenum (Mo), copper (Cu), thorium (Th), and gallium (Ga) (Agnihotri et al., 2003; Tribovillard et al., 2006), had significant cyclic characteristics during glacial-interglacial cycles. To normalize these trace element concentrations, we used the standard approach of dividing trace element concentrations by aluminum (Al) concentrations as an index for the relative enrichment (Bennett and Canfield, 2020). Accordingly, these elements were divided into three categories.

First, the variation range of Mo showed relatively more fluctuation, with value ranges of 0.08–0.61 μg/g. Al (Figure 4F). Its distinctly higher values during the Holocene were several times higher than the other peak values in the glacial stages MIS 10, 8, and 6. Second, U varied from 0.19 to 0.67 μg/g. Al (Figure 4E), showing relatively higher values during the glacial stages MIS10, 8, and 6, and a stable tendency since the last 130 kyr. The last category included Cu, Th, Ga, and V, whose ranges were relatively smaller but with obvious cyclic features over glacial-interglacial cycles (Figures 4A–D). Most RSEs exhibit a significant enrichment during glacial periods (p < 0.01, Supplementary Table S1), but Mo and Cu show little enrichment (p > 0.05, Supplementary Table S1).

Detailed sediment grain size information and original data were derived from Zhang et al. (2020). The detrital distribution of the mean grain size of the sediments varied between 4.49 and 213.16 µm, with a mean value of 17.86 µm to the maximum values of volcanic ash layers (Supplementary Figure S3A). Overall, the cyclic variation in mean grain size was finer during glacial periods than during interglacial periods (p < 0.01, Supplementary Table S1).

The end-member analysis results of grain size showed three end members, with peak modes in the fine (EM1, 5.23 µm), fine to medium (EM2, 19.87 µm), and coarse (EM3, 93.51 µm) particles (Supplementary Figures S3B–D). The downcore tendencies of these end-members show different cyclic variations. EM1 variation was consistent with that of the clay content; both were higher during glacial periods and lower during interglacial periods.

The decoupling between marine productivity and OC burial, as noted above, could indicate other important constraints on OC burial over glacial-interglacial cycles. It has been suggested that marine biological productivity may contribute less than half of the observed glacial-interglacial variations in OC burial, while the physical vertical processes with hydrological constraints may be more responsible for oceanic OC burial (Kohfeld et al., 2005). Hydrological redox conditions control not only microbial growth and activities (Belanger et al., 2020), but also microbial oxidation to OC (Anderson et al., 2019). The first-order impact of microbial oxidization of OC burial depends on the concentration and extent of dissolved oxygen with water depth (Hartnett et al., 1998; Galbraith and Skinner, 2020). Ocean circulation in the Northwest Pacific is complex (Figure 1B), with several current systems with different properties at different water depths, including the KC in the upper ocean (Gallagher et al., 2015), the NPIW in the middle (Rippert et al., 2017) and the AABW in deeper water (Kender et al., 2018), which provide different constraints for the dissolved oxygen in the water column over the glacial-interglacial cycles. Therefore, the impact of paleo-redox conditions varying from the upper ocean to the sediment-water interface on OC burial was examined between glacial and interglacial stages.

KC intensity has been shown to have an important impact on OC burial efficiency in the middle Okinawa Trough through changing bottom water O₂ concentration (Li et al., 2018; Zou et al., 2021). As discussed above, the KC was enhanced during the warmer interglacial periods (Figures 5C–E, Vats et al., 2020; Zou et al., 2021), which provided higher dissolved oxygen in the upper ocean and had a negative effect on OC preservation (Li et al., 2018). On the other hand, the NPIW is mainly restricted to subtropical North Pacific regions between ∼20°N and 40°N and mainly circulates within 300–1000 m of water depth, with the property of low dissolved oxygen (Talley, 1993). The Bering Sea, as the additional source of the Glacial NPIW (Horikawa et al., 2010), resulted in the enhancement of the NPIW formation during the glacial periods (Figure 5B). The Glacial NPIW penetrated to deeper range with relatively low oxygen and thus reduced the oxidative decomposition of OC by microorganisms (Worne et al., 2019). Finally, in this area, the extent of dissolved oxygen concentrations in the deep North Pacific is mainly determined by the deeper waters from Southern Ocean (Kender et al., 2018), such as Antarctic bottom water (AABW), which is oxygen-rich (Rippert et al., 2017). Therefore, AABW may have affected OC burial and preservation by changing the redox conditions of the deep ocean. It has been shown that less dissolved oxygen-rich water, indicated by higher authigenic uranium values (Figure 5A, Jaccard et al., 2016), moving northward in the glacial periods and controbuting to better OC preservation in this stage.

In bottom water, the efficiency of OC degradation at the sediment-water interface, with a loss rate of OC as high as 89.8%, is much higher than that of the transportation in the water column (Ordoñez et al., 2015). Therefore, the effect of redox conditions on OC burial at the sediment-water interface cannot be ignored. The variations in these redox sensitive elements (RSEs) have been used extensively as geochemical proxies to infer the accessibility of dissolved oxygen in the overlying water during the deposition (Bennett and Canfield, 2020). To contextualize the redox conditions through RSE enrichment, we compared the data of this study to a variety of modern depositional environments characterized by the enrichment of V, U, and Mo (Table 3, Bennett and Canfield, 2020). From these observational data at all depths in the global ocean, it can be seen that there was a relatively high variability of enrichment with depth in perennial OMZ (oxygen minimum zone) settings (Scholz et al., 2011), suggesting that water depth may strongly influence redox conditions. The RSE enrichments in core St10-PC characterize a notable increasing tendency with relatively reducing conditions during glacial periods, especially the third category of Th, Ga, V, and Cu, which are in line with the higher sedimentary OC burial during those times (Figure 4). Most of these RSEs exist stably in seawater in the form of organometallic ligands (Tribovillard et al., 2006). In the process of sediment settling from the surface to the bottom water, organic matter is easily degraded by microorganisms, during which the complex of RSEs and organic matter is decomposed, and the elements are released back into the water (Morford and Emerson, 1999). Therefore, this noted enrichment of RSEs shows that the sediment-bottom water interface was in a relative reducing state during the glacial periods, with a relatively lower oxygen content in the overlying water, which was more conducive to the burial and preservation of organic matter.

To better understand the different patterns of OC burial with different drivers between glacial and interglacial periods, we constructed a conceptual schematic of OC burial and preservation in the mid-latitude Northwest Pacific (Figure 6). During glacial periods, despite the increase in dust input and terrestrial matter, the strengthening barrier of glacial NPIW inhibited vertical mixing, slowing transport of nutrient-rich deeper water to the surface (Worne et al., 2019), and thus biological productivity was relatively lower (Figure 6A). However, dust input and terrestrial fine-grained materials adsorbed with organic matter to protect and accelerate the deposition of organic matter to the seafloor, where there was a relative reducing condition with less dissolved oxygen content to facilitate OC preservation (Carapanis et al., 2016). In contrast, during interglacial periods (Figure 6B), induced by enhanced water column vertical overturning (Worne et al., 2019), the upwelling water brought more nutrients and dissolved oxygen to the surface, which was conducive to phytoplankton photosynthesis. However, under the impact of relative oxidation conditions, microorganisms were relatively active and performed respiration to release organic matter back into the water column (Bennett and Canfield, 2020). Therefore, OC burial was significantly lower in the interglacial periods than in the glacial periods.

Compared to the Northwest Pacific, recent observational records in other areas suggest that the glacial-interglacial regime in OC burial and preservation is likely to be regionally heterogeneous (Carapanis et al., 2016; LaRowe et al., 2020). Carapanis et al. (2016) concluded from hundreds of sediment core records from the global oceans that OC burial pulses derived from higher export production from the surface ocean, higher transfer efficiency to the deep sea and/or enhanced preservation of organic matter in glacial sediment. Our record of OC burial decoupled from marine productivity and mainly forced by the preservation mechanism and redox conditions, is consistent with observations not only in the Northwest Pacific (Li et al., 2017) but also in the Northeast Pacific (Cristina et al., 2014), where OC burial has been shown to be decoupled from productivity and controlled by the redox conditions and/or other forces. The most controversial and noteworthy issue is the effect of dust input on marine productivity, which may determine how much production in surface waters is exported to the deeper ocean for burial in sediments. Multiple studies in global areas, including the North Pacific (Kienast et al., 2004; Li et al., 2017; Chen et al., 2020) and even the Antarctic Zone (Parekh et al., 2008) and sub-Antarctic Zone (Thöle et al., 2019), have provided evidence for the uncertainties and the limited effects of dust input on primary production, depending on the Fe bioavailability and nutrient availability in different areas (Li et al., 2017; Chen et al., 2020). Our results support that the evolution of marine productivity is not a simple function of OC burial, which reminds us to be cautious in evaluating biological pumps using geological OC records (Li et al., 2018).

In the present study, the water redox conditions could have been a more important driver than marine productivity of OC burial on orbital scale. As mentioned before, the sedimentary redox conditions were oxygen-rich in the Northwest Pacific during interglacial periods relative to the glacial periods, which is similar to the records in the North Pacific (Costa et al., 2018b; Li et al., 2018). Compared to the carbon sink in the surface ocean, carbon preservation in the deeper ocean plays a more prominent role in global climate change in the Northwest Pacific, where extensive and intense circulation in the water column affects carbon degradation (Kienast et al., 2004). Under the influence of such a regime, as mentioned above, our observation of glacial OC burial pulses over the past 380 kyr provides an improved understanding of the global carbon cycle on glacial-interglacial timescales. Its variations throughout Earth’s history were highly sensitive to atmospheric CO₂ concentrations (Figures 3G,H), implying a notable response of deep-sea OC burial to paleoclimate evolution (Carapanis et al., 2016).