Age estimation of marker dark layers
In this study, we first constructed perfectly continuous composite columns for the entire Quaternary sequences at the six deeper sites studied, and using these composite columns, we correlated distinct dark layers among the sites for the interval covering the last 1.5 My corresponding to the lithological Subunit IA (Additional file 1: Figure S1). Then, we constructed a high-resolution and high-precision age model at Site U1424 based on correlation of GRA and NGR with LR04 δ18O stack (Additional file 2: Figure S2). By using this age model, it is possible to estimate the ages of individual marker dark layers and thus project their ages to other sites.
Coding of dark layers
As is described above, dark layers are useful markers to make a high-resolution correlation between sites and assign precise ages to the sedimentary sequences in the deeper part of the Japan Sea. Consequently, it is useful to give them code numbers so that we can easily identify dark layers of our interest and estimate their ages. As to the last glacial period, thinly laminated (TL) layer numbering was conducted in the way so that TL numbers agree with interstadial numbers of DOC (Tada et al. 1999; Dansgaard et al. 1993). No numbering has been made on dark layers deposited before Marine isotope stage (MIS) 6 except by Tada et al. (1992) who made tentative naming of distinct dark layers at ODP Sites 794, 795, and 797. However, their naming was not systematic and went back to only ca. 0.7 Ma.
In this study, we conduct numbering of the marker dark layers in Subunit IA covering the last 1.45 Ma. We first identify relatively thick and distinct dark layers as marker layers for the purpose of inter-site correlation as is described above. Then, using these marker dark layers and tephra layers as guides, we further identify thinner but distinct dark layers that can be correlated among the deeper six sites and can be used as second order marker layers. The results are listed in Additional file 3: Table S1.
The relatively thick and distinct dark layers occur every several meters and they tend to be associated with quasi-cyclic changes in amplitude and frequency of sediment lightness (L*) in Subunit IA, which seems to be associated with glacial-interglacial sea level changes as will be discussed later. In brief, thicker dark layers tend to occur during glacial maxima especially when sea level was lowered by more than 100 m and euxinic deep water emerged due to development of low-salinity surface water (Tada et al. 1999; Kido et al. 2007; also Seki 2017). Thick dark layers accumulated under euxinic deep water were found during MIS 2, 6, 10, 12, 16, 20, and 22 (Seki 2017). Such dark layers are overlain by thick light layers which were deposited during interglacial maxima. These characteristics are less clear before MIS 22 (~ 0.9 Ma) when thick diatomaceous dark layers frequently deposited during interglacial maxima.
We define a cycle of such several meter-scale change in amplitude and frequency of lightness as starting from the top of a relatively thick and distinct dark layer downward to the base of the relatively thick light layer so that one cycle basically corresponds to one glacial-interglacial cycle. From the core top downward, cycle number starts from 0 and increases downward. Because we tune changes in sea level proxies such as GRA and NGR to LR04 δ18O stack so as to construct high-resolution age model for Unit I as described above, and because one cycle basically represents one glacial-interglacial cycle, there is a relation between a cycle number and MIS number. Namely,
where X is a cycle number and Y is an MIS number when the thick dark layer at the top of cycle X deposited. Exceptions are cycle 0 that corresponds to the Holocene when no thick dark layer exists at the top, cycle 1 whose top thick dark layer corresponds to MIS 2 and thus X = Y/2, cycle 20 whose thick and dark layer deposited during MIS 45 (interglacial period) and thus X = Y/2 − 2.5, and cycle 21 whose thick dark layer deposited at MIS 46 and thus X = Y/2 − 2. A relatively thick and distinct dark layer at a top of cycle X is named Dark Layer (DL)-X-1 (Additional file 1: Figure S1; Additional file 3: Table S1). Every cycle contains several thinner but still distinct dark layers that are possible to correlate in-between sites. We put numbers 2, 3, … on them in descending order and described as suffixes to the cycle number. Namely, Zth dark layer in cycle X is DL-X-Z. In Additional file 3: Table S1, we list ages at the base of each dark layer based on tephra-p-mag-tuned and LR04-tuned age models for the last ~ 3 My. We also list depths and ages at the top of marker dark layer that define the top of each cycle, although it should be noted that their depths and ages may be less well defined and could be slightly diachronous due to the result of bioturbation.
Temporal and spatial changes in linear sedimentation rates (LSRs) in the Japan Sea during the last 1.45 My
As is described in the previous section, we estimated the age at the base of each marker dark layer at Site 1424 and projected its age to other sites based on inter-site correlation described above (Table 2). By using these data, it is possible to calculate linear sedimentation rates (LSRs) at all the six sites with high time resolution (Additional file 4: Table S2). Additional file 5: Figure S3 shows temporal changes in LSRs at all sites. LSRs are more or less constant at ~ 4 and ~ 3 cm/ky at Sites U1422 and U1424 (deeper sites); gradually increasing from ~ 3 to ~ 10 cm/ky, from ~ 2 to ~ 7 cm/ky, and from ~ 2 to ~ 8 cm/ky at Sites U1423, U1425, and U1430 (intermediate depths sites); and relatively high and more or less constant at ~ 10 cm/ky at Site U1426 (shallow site), respectively. Exceptions are between ~ 850 and ~ 1200 ka when LSRs are slightly higher and variable and after ~ 250 ka when LSRs increase upward. At Site U1430, extremely low LSR is observed at ~ 800 ka.
In general, LSR tends to be lower at deeper sites, which probably reflects dissolution of carbonate (and biogenic silica to some extent). The upward increase in LSRs after ~ 250 ka is probably because near-surface sediments are less compacted. The gradually increasing trend of LSR at intermediate depth sites can be partly due to lesser degree of compaction, but the increasing trends started from deeper parts. The upward increasing trend of LSRs could be explained by the upward increase in biogenic carbonate content due to deepening of calcium carbonate compensation depth (CCD) that crossed the intermediate depths of 1000 to 2000 m at ~ 1.0 Ma. Variable LSRs between ~ 850 and ~ 1200 ka could be related to large amplitude changes of CCD during transition from shallower CCD before 1200 ka to deeper CCD after 850 ka. Mass accumulation rate (MAR) calculation of carbonate at these sites based on high-resolution XRF core scanner and gamma ray attenuation (GRA) porosity data combined with LSR data will be necessary to test this possibility.
Establishment of paleo-observatory network
Reconstruction of material cycling such as carbon, sulfur, and phosphorous within the earth surface is critical to understand dynamics and controlling mechanisms of climatic changes in the past and preparing for the future. Especially, changes in centennial to millennial time scales are of prime importance since they are beyond the coverage of observational records, but still, climatic changes of such time scales are relevant to the society. Spatial coverage is also important since spatial changes in mass flux often give clues to understand the processes within the system of interest and their controlling factors. However, it is generally difficult to correlate geological records from different localities with high time precision and resolution. In this respect, Quaternary hemipelagic sediments of the Japan Sea provide a rare opportunity to study spatiotemporal changes of mass flux with high time precision and resolution. Namely, marker dark layers and tephra layers can be correlated with the time precision of less than 100 years, and approximately 250 marker layers are defined during the last 1.45 Ma, providing ~ 6 ky resolution in average. These time precision and resolution are an order of magnitude higher than those for other correlation methods such as biostratigraphy, paleomagnetism, and radiometric dating. In addition, dry bulk density of the sediments can be estimated from GRA data measured onboard. Thus, it is possible to reconstruct MARs with an average time resolution of ~ 6 ky at the six deeper IODP sites that cover the area more than a half of the Japan Sea. Moreover, we can calculate MAR of individual chemical element and/or mineral by combining high-resolution MAR data with high-resolution chemical and/or mineralogical analyses data.
In this study, we propose to utilize the six IODP sites in the Japan Sea as paleo-observatory network since high-precision age data and high-resolution LSRs and GRA data are available; the sites cover wide latitude, longitude, and water depth ranges; proxy data at each site can be correlated with precision of less than 100 years; and good amount of samples are still available from IODP.