Stratigraphy was developed by piecing together a continuous splice from the holes at each site, using the CaCO3 and Corg time series to correlate between sites, and using benthic oxygen isotope time series to confirm that the correlations are indeed chronostratigraphic. We followed this method because we were beginning from reconnaissance-scale oxygen isotope records and because both the CaCO3 and Corg time series have more bandwidth and thus a potential for a more constrained correlation. Any systematic problems between the "carbon" stratigraphy and the oxygen isotope stratigraphy would indicate a problem with the assumption that the CaCO3 and Corg time series are chronostratigraphic.
We have established an age model for the 0- to 40-ka time period where we have adequate to excellent radiocarbon and oxygen isotope control. From 140 to 574 ka we have constructed a preliminary age model and warn readers that this age model will probably be revised as more data become available. We used the 0- to 140-ka ages to calculate MARs for this time period, as explained later.
For the Leg 167 drill sites we began by developing a continuous sediment column based initially on the shipboard splice (Lyle, Koizumi, Richter, et al., 1997). We revised the shipboard splice using our carbon records. For the most part the adjustments to the shipboard splice were minor, with the exception of the lower parts of Site 1019, as discussed in more detail below. Note that our revised offsets listed in Table 4, Table 5, Table 6, Table 7, Table 8, Table 9, Table 10, and Table 11 are intended to be added to the meters below seafloor (mbsf) depth scale, not to the shipboard meters composite depth (mcd) scale, to obtain the revised scale (rmcd).
We then chose a few significant tie points between drill sites, based upon important oxygen isotope events (e.g., MIS 6/5 boundary) and any radiocarbon data we had, to begin correlation of the CaCO3 and Corg time series. We used the program Analyseries 1.1 (Paillard et al., 1996) to correlate all records to a master site, Site 1020, in the depth domain. We alternately used each of the two carbon records to develop the correlation because they have significantly different time series (Fig. 3, Fig. 4). The final correlations shown in Table 4, Table 5, Table 6, Table 7, Table 8, Table 9, Table 10, and Table 11 maximize the coherence from both records with Site 1020. By using both CaCO3 and Corg records it was possible to spot correlations even if the amplitudes of the events in different cores were significantly different. We then used the depth correlation to Site 1020 to transfer an age scale from Site 1020 to the other cores. Finally we compared isotope records to confirm that the CaCO3 and Corg time series are chronostratigraphic, and in some cases repeated the process when the carbon time series and isotopes disagreed.
We constructed the age model for 0-140 ka by correlations of the two piston Cores W8709-13 and EW9504-17 to Site 1020. Core W8709-13 was taken as part of the Multitracers transect (Table 1; Fig. 1; Lyle et al., 1992) and has a detailed chronostratigraphy in calendar ka to 57 ka (Lund and Mix, 1998). This chronostratigraphy is based upon 22 AMS radiocarbon dates on both planktic and benthic foraminifers to 35 ka and then upon correlation of millennial-scale oxygen isotope peaks to the north Atlantic Core V23-81 for the interval 36-57 ka. A recent radiocarbon compilation with some new dates has also revised the Core W8709-13 age model somewhat (Mix et al., 1999). In this paper, we have chosen not to revise the 0-20 ka part of the time scale because the changes are relatively small. Instead we refer the readers to Mix et al. (1999) if they need the latest age model for this interval. CaCO3 data used to correlate Cores W8709-13 and EW9504-17 are given in Table 5.
To continue the age model to 140 ka, we correlated Core W8709-13 to Core EW9504-17 taken about 15 km to the northwest on the site-survey cruise for Leg 167 (Table 1; Fig. 1;). This 15-m piston core reached through MIS 5 to the end of MIS 6. It was measured at 5-cm intervals for CaCO3 and Corg (Table 6). Benthic oxygen isotope time series of similar detail were also developed prior to the late Holocene by A.C. Mix (unpubl. data, Table 6). The three time series were used to correlate the two cores in detail and to transfer the age scale to Core EW9504-17. Below 57 ka, we used the isotope stratigraphy to develop an age model to 140 ka. We used the MIS time scale of Martinson et al. (1987) for this age model. For ages greater than 140 ka, we used the time scale of Martinson et al. (1987) to 300 ka, and used the Imbrie et al. (1984) time scale for older sediments.
We chose Site 1020 as our "master" site for the correlation, because we had sampled a record that spans >500 k.y. and sedimentation rates are sufficiently high to record high-frequency events, averaging 10 cm/k.y. for the entire Pleistocene. Unfortunately, Site 1020 has a few small turbidites most noticeably on the MIS 2/1 deglaciation (Table 4), which makes this a poor site to study the last ~15 k.y. in detail. To obtain the age model for 0-140 ka for Site 1020 we carefully correlated the carbon records with Core EW9504-17 and used reconnaissance-scale isotope measurements reported by A.C. Mix (unpubl. data) to confirm that the age model was reasonable. Again, we used the strategy to iterate between Corg and CaCO3 correlation to maximize coherence in both time series and better constrain our stratigraphic model. Once we had an initial age model, we calculated sedimentation rates and reiterated the model to minimize sedimentation rate changes while still maintaining the high coherence between records (Fig. 6).
For sediments older than 140 ka (16.69 rmcd), we have a more difficult problem assigning age, and so we consider the age model from 148 to 574 ka preliminary. We have some oxygen isotope control to MIS 7 (Fig. 3) but little more is available now. Our other age control is based upon the last occurrence of the nannofossil Pseudoemiliania lacunosa (460 ka) at about 51 rmcd, the Brunhes/Matuyama paleomagnetic boundary at 84.83 mcd (780 ka; Lyle, Koizumi, Richter, et al., 1997), and pollen stratigraphy (Heusser et al., Chap. 17, this volume). One of the notable features of deglaciations at the MIS 5/6 and MIS 2/1 boundaries is an initial spike in alder (Alnus) pollen followed by oak (Quercus) and redwood (Sequoia). If we assume that this pattern is typical of deglaciations, we can date previous deglaciations below our oxygen isotope control (Fig. 7). The age model thus developed is in harmony with an age model developed by Kreitz et al. (Chap. 10, this volume) using alkenone SST estimates and assuming that cold SST marks glacials.
We had no isotope control for Site 1021 but established an initial age model from the P. lacunosa last occurrence datum (15.41-18.41 mcd; Fornaciari, Chap. 1, this volume) and the Brunhes/Matuyama paleomagnetic boundary (27.16 mcd; Lyle, Koizumi, Richter, et al., 1997). The final age model in Table 7 was refined by correlating the Site 1021 carbon records to Site 1020. After we completed our age model, we learned that Guyodo et al. (1999) had independently estimated ages using paleomagnetic intensity. Our age models agree well (±5 k.y.), except for the interval 360-480 ka, where we have an offset of about 40 k.y. between the two independent age models.
Site 1018 was relatively straightforward to correlate to Site 1020 (Table 8), and the benthic oxygen isotope time series (Andreasen et al., Chap. 8, this volume) is in agreement with the age model in Table 8. The main difficulties arose because Site 1018 is missing much of the Holocene, which is confirmed by the lack of a Holocene Sequoia pollen peak (L.E. Heusser, unpubl. data). A Holocene section was found in the site survey core (Core EW9504-13), so the missing section at Site 1018 is probably a recovery problem of the soft uppermost sediments, not a hiatus. The CaCO3 and Corg correlations also presented a challenge. The peaks in Site 1018 were often significantly different in amplitude than Site 1020, and more iterations between the Corg and CaCO3 time series were needed to achieve an optimum correlation. The differences in amplitude reflect some regionality in sedimentation, because this drill site is more than 500 km SSE of Site 1020.
Site 1019 is our problem child, with the promise of extremely high-resolution sections at the MIS 2/1 deglaciation (e.g., Mix et al., 1999) and at the MIS 5/6 boundary but with the difficulty of highly variable sedimentation. In addition, isotope measurements were not as clean to interpret as at other drill sites; we had problems with mislabeled samples, and we sampled at a coarser resolution (30-cm spacing) than the other drill sites. Despite these difficulties, we have developed an age model for about 0-250 ka, with an average sedimentation rate near 30 cm/k.y.
We are in the process of rerunning the carbon analyses on this core at a sample spacing of 5 cm. We report data at this spacing from 0 to 17.5 rmcd, or for the time interval 0-44 ka. The more detailed sampling has improved our confidence in correlations to other Leg 167 drill sites.
Site 1019 is the only site we studied that has major differences between the shipboard composite section and the revised section (Table 9). We redid the composite section on Holes 1019C and 1019E using magnetic susceptibility, color reflectance, CaCO3, and Corg. We did not use the gamma-ray attenuation porosity evaluator (GRAPE) bulk density data, because the site was very gassy and we were concerned that the GRAPE data had many coring artifacts. The additional time series resolved several ambiguities in creating the composite section. For example, a dolomitic sediment layer appears in both holes as a carbonate spike at about 52 rmcd and provided an important tie point between the two holes. Note that the revised offsets in Table 9 are significantly different than the shipboard composite splice, especially from Core 167-1019C-4H and below.
We have good AMS radiocarbon age control from about 6 to 24 ka (2.9-12.1 rmcd; Mix et al., 1999) from 10 mixed planktic foraminifer samples, 10 mixed benthic foraminifer samples, and 1 piece of bark. We used these ages to constrain the Site 1019 age model for this time period. The radiocarbon dates indicate that early deglacial sedimentation rates rose to greater than 65 cm/k.y. The 5-cm depth resolution of our samples through this time period converts to a time resolution of 75 yr.
From the middle of MIS 3 to MIS 5d (~16-27 rmcd) sedimentation rates were significantly slower, as low as 14 cm/k.y. The typical 30-cm sample spacing in this depth interval spans ~2 k.y., and part of the problem we experience in correlating may occur because we have aliased peaks. We expected some slowdown in sedimentation in this interval because the 3.5-kHz seismic record of the upper sediments taken on the approach to Site 1019 shows a pinchout in the interval between ~20 and 32 mbsf (~ 20-34 rmcd; Fig. 8) with the strongest pinchout between ~21 and 27 mbsf (~21.3-28.5 rmcd). We have constructed a preliminary age model (Table 9), but expect that the 50- to 250-ka interval will be adjusted as more high-resolution data becomes available.
Because the sedimentation rate at Site 1011 is slow when compared to the northern sites (average of 4.3 cm/k.y. over the last 340 k.y., Table 10) and because there is a likelihood that the carbon records could change as we move south, the primary correlation tool we used is the benthic oxygen isotope record. Once we established a preliminary age model in this manner we adjusted the records to a minor extent to match CaCO3 and Corg peaks.
Y74-2-22PC, near the tip of the Baja California Peninsula (Fig. 1) is our southernmost core and the core with the slowest sedimentation rate, averaging 1.3 cm/k.y. over the last 660 k.y. (Table 11). Although this core is unsuitable for k.y.-scale comparisons, it does allow us to monitor general trends. We also used the benthic oxygen isotope record at this core as the primary correlation tool to construct an age model and made minor adjustments to match carbon peaks.