PLEISTOCENE MILLENNIAL- TO ORBITAL-SCALE PROCESSES IN THE CALIFORNIA MARGIN

One primary objective of Leg 167 was to gain a better understanding of millennial-scale variability known to occur along the California margin. The drilling of the Santa Barbara Basin on Leg 146 (Site 893, Kennett, Baldauf, and Lyle, 1995) documented that significant millennial-scale variability exists along the Pacific coast of North America and that this variability is tied to the North Atlantic Dansgaard/Oeschger events, the weak and abrupt interstadials within marine isotope Stages (MIS) 2-5 (Behl and Kennett, 1996; Hendy and Kennett, 1999). The Younger Dryas interval at the MIS 2/1 boundary, a return to nearly full glacial conditions, is also prominent at Site 893. The last glacial stage and the Younger Dryas interval are marked by higher oxygenation of the basin and disappearance of laminated sediments, possibly because of strengthened production of North Pacific Intermediate Water in the glacial northwest Pacific.

One of the major questions that remained from the Site 893 studies in Santa Barbara Basin was the extent to which the Santa Barbara Basin is representative of regional conditions. The 120 m drop in sea level at the last glacial maximum significantly restricted the access of open ocean waters to the Santa Barbara Basin (Fig. 18). It is possible that some of the environmental changes recorded at Site 893 were caused by changes in the basin configuration because of changes in water depth. One of the goals of the Leg 167 drilling was to collect paleoceanographic records in the vicinity of the Santa Barbara Basin to compare to the Site 893 record. A second goal was to extend the record beyond that of Santa Barbara Basin and study more of the Pleistocene.

Sites 1014 (Tanner Basin) and 1017 (Santa Lucia Slope) are both in the vicinity of Santa Barbara Basin (Fig. 18; Table 1) but are bathed by somewhat deeper waters. Bottom waters in Santa Barbara Basin originate from intermediate waters at a depth of 475 m below sea level (mbsl) because of the sill into the basin. Site 1017 was drilled at 956 mbsl, whereas Site 1014 was drilled at 1166 mbsl. Both drill sites are bathed by waters of the open eastern Pacific Ocean and so have minimal artifacts that could be caused by basin restriction.

We found that Dansgaard/Oeschger (D/O, Dansgaard et al., 1982; Oeschger et al., 1985; Dansgaard et al., 1993; McManus et al., 1994) interstadial events are indeed present at both sites, showing up most strongly in the planktonic oxygen isotope signal (Fig. 19; Kennett et al., Chap. 21, this volume; Hendy and Kennett, Chap. 7, this volume). They appear similar to events in the Santa Barbara Basin (Hendy and Kennett, 1999). Oxygen isotopes have been measured on Globigerina bulloides in all three sites, and the D/O events are strongest in Santa Barbara Basin (up to 2, Site 893, Hendy and Kennett, 1999), moderate near Point Conception (up to 1, Site 1017, Kennett et al, Chap. 21, this volume), and more poorly expressed in Tanner Basin (about 0.5, Site 1014, Hendy and Kennett, Chap. 7, this volume). The difference seems largely because of a sedimentary recording effect rather than a regionality in response: Site 893 has a sedimentation rate of about 1200 m/m.y. above the MIS 5/6 boundary, Site 1017 has a rate of about 185 m/m.y., and Site 1014 has a rate of around 112 m/m.y. Bioturbation should smooth the signal at Site 1014 much more completely than at Site 893, and aliasing of the signal by relatively coarse sampling becomes more important.

The D/O events as expressed by stable isotopes are primarily a surface-water phenomenon, in contrast to the strong response of the benthic community to anoxia during the events (Behl and Kennett, 1996; Cannariato et al., 1999). The strongest signals in oxygen and carbon isotopes are in planktonic foraminifers, and the plankton community has major compositional changes during these events. As Hendy and Kennett (1999) have pointed out, the strong surface-water expression of D/O events in the Santa Barbara Basin suggests an atmospheric connection to the source of the D/O oscillations. There are events in the subsurface ocean (e.g., Lund and Mix, 1998) but they are subdued relative to the surface signal. In addition, benthic and planktonic foraminiferal isotope signals do not strongly track each other, at least in the California Borderland. The lack of a strong intermediate D/O signal in carbon isotopes is troubling, if the signal is driven by intermediate-water source changes.

When working with millennial-scale events, assigning ages is a significant problem. Compare, for example, age assignments for Sites 1017 (Kennett et al., Chap. 21, this volume) and 1014 (Hendy and Kennett, Chap. 7, this volume) in Figure 19. It is apparent that the preliminary age models used for these cores need revision assuming that the D/O events are chronostratigraphic between the GISP-2 ice core from Greenland and the California Borderlands drill sites. Because the D/O link is probably atmospheric, we expect little or no phase lag to the signal, which should make it possible to use the D/O events stratigraphically. Development of a D/O "tuned" time scale for the California margin will be highly important for the next level of study of millennial-scale processes and global climate. It will provide a means to study decadal to centennial oceaographic variability. Development of tuned time scales for the California Borderland drill sites must be done in a systematic manner, however, and more confirmation that the D/O events on the California margin are truly in phase with the Greenland ice cores would increase the confidence in this time scale.

During Leg 167, high-resolution records were recovered from central California (Site 1018, Fig. 1), the northern California Eel River Basin (Site 1019, Fig. 20), and the eastern Gorda Ridge (Site 1020, Fig. 20) to compare with the records from the California Borderlands. Millennial-scale events possibly linked to D/O cycles appear in these records (Lyle et al., Chap. 11, this volume; Fig. 21). In Figure 21 we compare the stacked MAR records from Site 1018, Site 1020, and two piston cores (EW9504-17 andW8709-13), normalized by standard deviations from the median. Construction of the stack is discussed in Lyle et al. (Chap. 11, this volume). Stacking smooths out events but also improves signal-to-noise characteristics of the record. The individual records of the stack can be studied in Lyle et al. (Chap. 11, this volume).

Detailed planktonic foraminiferal oxygen isotope records have yet to be constructed for any of the northern drill sites, so we do not know if there is a high-amplitude isotopic response in surface plankton similar to the response recorded at the California Borderlands. However, sedimentary time series of both C_org and CaCO₃ have millennial-scale variability. Lyle et al. (Chap. 11, this volume) argue that this variability represents variability in the production and rain of C_org and CaCO₃ from surface waters, not from rapid changes in dissolution. Lyle et al. (Chap. 11, this volume) reached this conclusion because there is no depth-dependent change in the events, because most of the events exceed the CaCO₃ burial expected from saturation of North Pacific deep waters, and because some of the events even exceed the modern CaCO₃ rain from the surface. Changes in production and/or the ratio of CaCO₃ to C_org in the particulate rain best explain the observations.

Because of the high precision of radiocarbon dating we have the best age control in the interval from about 25 ka to the present. Within this time window lies the last glacial maximum (LGM), a series of millennial-scale warmings and coolings during the deglaciation (including the return to near-glacial conditions in the Younger Dryas interval) and the Holocene. This is also the best known interval along the California margin, having been sampled by numerous studies with long gravity cores and piston cores (Moore, 1973; Gardner et al., 1988; Lyle et al., 1992; Sancetta et al., 1992; Sabin, 1994; Prahl et al., 1995; Thunell and Mortyn, 1995; Dean et al., 1997; Doose et al., 1997; Gardner et al., 1997; Ortiz et al., 1997; Dean and Gardner, 1998).

From the earlier studies it is generally accepted that productivity was lowest in MIS 2 and highest in the Holocene, with MIS 3 being intermediate (Lyle et al., 1992; Dean and Gardner, 1998). There appears to be weaker coastal upwelling in the glacial half of this period, with a higher fraction of productivity derived during the spring bloom off southern Oregon (Sancetta et al., 1992). During the glacial period, nutrient injection into the surface waters may be dominated by diffuse offshore upwelling driven by the curl of the wind stress (Ortiz et al., 1997). Sea-surface temperatures (SST) were significantly colder during the last glacial maximum. Prahl et al. (1995) estimated 5°-6°C temperature difference between the last glacial maximum and modern SST off southern Oregon by both alkenone- and radiolarian-based regressions. Ortiz et al. (1997) estimated the LGM temperature by foraminiferans to be 3.3° ± 1.5°C colder along the same transect. The gradient in SST from north to south appears to have been highest at the LGM (Moore, 1973; Herbert et al., 1995; Prahl et al., 1995; Doose et al., 1997). This remains a controversy because other workers have used abundance of left-coiling Neogloboquadrina pachyderma as an indicator that the SST in the California Borderlands may have been about 8°C colder than modern at the LGM (a glacial SST of about 7°-8°C, about the same SST as southern Oregon; Thunell and Mortyn, 1995; Kennett and Venz, 1995), rather than the 2°-3°C difference measured from alkenone biomarkers (Herbert et al, 1995; Yamamoto et al., Chap. 12, this volume; Ostertag-Henning and Stax, Chap. 26, this volume; Manglesdorf et al., in press). It should be a priority to explore and settle this controversy. It is important to point out that Ortiz and Mix (1997) find 100% left-coiling N. pachyderma from core tops in waters as warm as 12°C. The late Holocene sediments at Site 1019 are also nearly 100% left-coiling N. pachyderma (Mix et al., 1999). Subpolar water is not required for a switch from right- coiling to left-coiling N. pachyderma.

Northern California

ODP drilling is responsible for the highest resolution records from the LGM to the present along the California margin: Site 893 (LGM located about 30 mbsf; Kennett et al., 1995), Site 1017 (LGM at 4.5 mbsf, Kennett et al., Chap. 21, this volume), and Site 1019 (LGM at 11 mbsf, Lyle et al., Chap. 11, this volume; Mix et al., 1999). Site 1019 has the fastest sediment accumulation and highest time resolution of the northern sites drilled on Leg 167 (Fig. 20). This drill site is located just shoreward of the slope break in the Eel River Basin and is in a region of active deformation. A hill 220 m high lies just east of Site 1019 and is formed of late Pleistocene sediments. The major uplift of the hill occurred after 400 ka and perhaps as late as 160 ka (Gallaway, 1997). Sedimentation rates at the site are extremely high during the MIS 6/5 and MIS 2/1 deglaciations, averaging 550 m/m.y. during the MIS 2/1 deglaciation and 360 m/m.y. during the MIS 6/5 deglaciation (Mix et al., 1999; Lyle et al., Chap. 11, this volume).

Lyle et al. (Chap. 11, this volume) have analyzed late Pleistocene and Holocene sediments from Site 1019 for C_org and CaCO₃ at roughly 100 yr intervals (Fig. 22) and find that the deglacial sequence is similar to the North Atlantic: records of both C_org and CaCO₃ mass accumulation rates (MAR) mimic the oxygen isotope record from the GISP-2 Greenland ice core until about 10 ka. The deglaciation has slightly higher C_org MAR than the Holocene, which is somewhat different than records offshore. This may in part be a preservation effect because of the high sedimentation rates. CaCO₃ MAR responds dramatically through the glaciation, marked by highs in late Stage 2 and the Bølling-Allerød (D/O Event 1), and immediately younger than the Younger Dryas interval. CaCO₃ becomes much more rare in the sediments of Site 1019 (as well as elsewhere in the northern California and southern Oregon region) at 8 ka.

At Site 1019, the Younger Dryas interval is more easily recognized in the C_org percentage data (Fig. 23). The Younger Dryas is marked by about a 0.5% C_org difference between it and the warm intervals on either side. There is also a major shift in coiling ratio of N. pachyderma to essentially 100% left-coiling specimens (Mix et al., 1999). The flanking C_org highs have three laminated intervals within them: 19.0-19.13 ka (9.52-9.61 mcd), 14.66-14.96 ka (7.04-7.25 mcd), and 10.9-11.65 ka (4.82-5.26 mcd; Pike et al., 1998). These laminated intervals appear to represent enhanced productivity, because they have high numbers of upwelling diatom flora and extensive benthic foraminiferal colonization, which would rule out anoxic conditions (Pike et al., 1998). Carbon isotope contents of the benthic foraminiferan Uvigerina peregrina are actually enriched in ¹³C during the laminated intervals (Mix et al., 1999) arguing against a strongly oxygen-depleted water source at this time.

Southern California

It is clear that productivity plays a role in the appearance of laminations in the warmer deglacial stages in the Eel River Basin (Site 1019) in Northern California. This is a substantially different interpretation than that for the California Borderlands (Site 893, Santa Barbara Basin), where researchers have argued for changes in strength of the oxygen minimum as the primary driving force shaping the deglacial sedimentary record (Kennett and Ingram, 1995; Behl and Kennett, 1996; Cannariato et al., 1999). Leg 167 postcruise scientific research has concentrated heavily on the late glacial and Holocene record in Site 1017 to investigate the causes of glacial interglacial variability (Behl et al., Chap. 22, this volume; Irino and Pedersen, Chap. 23, this volume; Ishiwatari et al., Chap. 24, this volume; Kennett et al., Chap. 21, this volume; Ostertag-Henning and Stax, Chap. 26, this volume; Tada et al., Chap. 25, this volume). It appears that current activity and/or offshore transport of clastics is an important factor shaping the Site 1017 record.

Strong glacial/interglacial changes occur at Site 1017 but are complicated by active slope depositional processes. The sedimentation processes are an important component of the preserved sediment record here. Because one of the typical environments for high-resolution paleoceanographic records is the upper continental slope and because sedimentation here can be much more variable than in the deep sea, it is critical to understand how sedimentation can affect the paleoceanographic record to properly interpret paleoceanographic change.

Site 1017 is marked by a series of small turbidites (1-3 cm sand layers) in the glacial part of the record, including the Younger Dryas (Irino and Pedersen, Chap. 23, this volume). The likely cause of the turbidites is the lowered sea level and more proximal position of the drill site to the coast during glacial intervals. The provenance of the finer sediment fraction changed during the deglaciation from a more mafic source in MIS 2 and 3 to a more felsic source in the Holocene. Irino and Pedersen (Chap. 23, this volume) argue for a stronger glacial littoral and offshore transport from Franciscan sources to the north of Cape Conception in MIS 2 and 3 to be the cause of the change in provenance.

The glacial part of the record also has larger modal grain size in the nonturbidite intervals (Tada et al., Chap. 25, this volume) and occasional faint laminations interpreted as contourites. Tada et al. (Chap. 25, this volume) argue that the northward-flowing California Undercurrent may have been stronger in MIS 2 and 3, or at least the energy of intermediate waters impinging upon the upper slope was stronger in the vicinity of Site 1017.

Tada et al. (Chap. 25, this volume) also show that the organic content of the sediment is primarily in the finer fraction and argue that any change in current strength will change the deposition of this fraction and the C_org profile. They also argue that C_org variation uncorrelated with the grain-size effect seems to relate to higher production. Sorting out the depositional control from the production control here is obviously of high importance to our ultimate understanding of paleoproductivity and upwelling along the California margin.

Comparison of High-Resolution Time Series

It will become increasingly important to compare high-resolution time series through the deglacial interval from a variety of sites along the North American margin to understand the dynamics of Pleistocene oceanographic change. In Figure 24 we make an initial comparison of the high-resolution CaCO₃ weight percentage time series from Site 893 (Santa Barbara Basin; data from Gardner and Dartnell, 1995), Site 1017 (Santa Lucia Slope; data from Ostertag-Henning and Stax, Chap. 26, this volume), and Site 1019 (Eel River Basin; data from Lyle et al., Chap. 11, this volume).

There is a fundamental difference in the records from near the California Borderlands and the northern California site: the Holocene has higher CaCO₃ overall than the Pleistocene at Sites 893 and 1017, whereas the inverse is true at Site 1019. Nevertheless, one of the prominent features in all the records is the CaCO₃ peak immediately after the Younger Dryas interval in the vicinity of 10-11 ka. This peak has roughly double the CaCO₃ percentage of the Younger Dryas at all sites. At Sites 1019 and 1017 on the open ocean margin, the Younger Dryas interval is bracketed by another CaCO₃ high during the early deglaciation. The early Holocene CaCO₃ peak at Site 1017 is largely composed of coccoliths and is primarily associated with deposition of small Gephyrocapsa species, an upwelling indicator (Tanaka and Tada, Chap. 27, this volume).

There is an apparent offset of about 1000 yr between the equivalent early Holocene CaCO₃ peak at Sites 1019 and 1017. It is very possible that the offset may be a dating artifact, and we encourage systematic detailed age models to be developed and intercalibrated at both sites. If the age offset is real, it may represent the propagation of a CaCO₃ production event from north to south along the California margin.

Benthic Foraminiferal Oxygen Isotopes

A major postcruise research effort for Leg 167 consisted of measuring benthic foraminiferal oxygen isotopes for high-resolution Pleistocene stratigraphy and deep-water properties (Fig. 25; Andreasen et al., Chap. 8, this volume; deMenocal and Baker, Chap. 9, this volume; Hendy and Kennett, Chap. 7, this volume; Kennett et al., Chap. 21, this volume; Lyle et al., Chap. 11, this volume). As a result of these studies we now have benthic stable isotope data from seven Leg 167 drill sites to at least MIS 6 (>130 ka; Fig. 25). These data and that from Santa Barbara Basin (Site 893) not only allow us to develop age models for each of the drill sites but also to investigate the structure of the intermediate- and deep-water column (480 to >3000 mbsl).

Figure 26 shows a comparison of average benthic isotopic composition for the Holocene (MIS 1) plotted against measured modern bottom-water temperatures at each drill site. The bottom-water temperatures were measured while taking the sedimentary thermal gradient (Lyle, Koizumi, Richter, et al., 1997). The Santa Barbara Basin bottom-water temperature is from Emery (1960; Fig. 6). The modern profile of temperature vs. benthic foraminiferal oxygen isotope composition provides a simple internal calibration between temperature and isotopes for further isotopic temperature estimates. The change in isotope composition with depth is essentially a temperature function, because the modern salinity change is small in the intermediate and deep waters along the California margin (0.39 practical salinity units [PSU; 1 PSU approximately equals 1 change] between 500 and 6000 mbsl; Talley and Roemmich, 1991). The salinity gradient should cause minimal differences in oxygen isotope composition, because the oxygen isotope-salinity slope in the North Pacific is 0.322 per PSU (Lynch-Stieglitz et al., 1999).

Figure 26 also shows the MIS 2 average benthic foraminiferal isotopes plotted against modern bottom-water temperature. One would expect, if there were a major change in density gradient, to find the drill sites plot with a different slope or with a significant kink in the profile when compared to the modern profile. Instead, the sites plot along the same slope but with an offset of 1.55 from the Holocene profile. This is slightly less than the reported Pacific deep benthic oxygen isotope shift of 1.70 (Norton et al., 1997). The 0.15 difference, if real, is the equivalent to a 0.6°C average temperature increase relative to deep water, or a 0.5 PSU average decrease in salinity relative to deep water. Because surface temperatures at the last glacial maximum were 2°-5°C colder along the California margin (Doose et al., 1997), it is more likely that the average intermediate-water salinity dropped. Whether or not temperature or salinity was the more important factor, the intermediate-water density gradient appears to have been little affected by the glacial-interglacial climate changes.

Because Leg 167 investigators have thus far concentrated their efforts on individual drill sites, the ability to profile the intermediate-water column has not yet been strongly exploited. However, we expect that follow-up studies will more strongly focus upon this potentially powerful paleoceanographic tool.

SST Records

One of the other major data sets collected during the initial postcruise studies on Leg 167 is alkenone-derived SST records (Kreitz et al., Chap. 10, this volume; Ostertag-Henning and Stax, Chap. 26, this volume; Yamamoto et al., Chap. 12, this volume; Mangelsdorf et al., in press). The California margin is particularly well suited for these studies because of the significant number of biomarker calibration studies in the modern California Current system and by comparisons between alkenone SST, faunal SST, and oxygen isotope SST estimates (Prahl et al., 1993; Prahl et al., 1995; Doose et al., 1997; Herbert et al., 1998). The results of these calibration studies indicate that haptophyte production in slope and shelf locations (water depths <2 km) occurs <30 m deep and has no bias in SST from the mean annual SST, whereas farther from shore the haptophyte production occurs slightly deeper in the water column and there is a slight bias toward colder temperature estimates (Herbert et al., 1998). Relatively high abundance of the coccolithophorid Emiliani huxleyi, a major producer of the alkenones, occurs throughout at least the last 25 k.y. (Tanaka and Tada, Chap. 27, this volume). It is likely that modern SST calibrations are valid for at least the late Pleistocene. Comparisons between radiolarian and alkenone estimates of SST show that the two different estimates are essentially the same (Prahl et al., 1995).

All along the California margin the warmest SST coincides with maximum interglacials but the coldest SST occurs before maximum glaciation (Kreitz et al., Chap. 10, this volume, Herbert et al., 1995). This is true at all Leg 167 drill sites studied (Site 1017, Ostertag- Henning and Stax, Chap. 26, this volume; Site 1016, Yamamoto et al., Chap. 12, this volume; Site 1020, Kreitz et al., Chap. 10, this volume; Sites 1017, 1018, and 1019, Mangelsdorf et al., in press) and Site 893 (Herbert et al., 1995). The SST record for the last 150 k.y. from a northern California drill site (Site 1020; Kreitz et al., Chap. 10, this volume) and a southern California drill site (Site 1016; Yamamoto et al., Chap. 12, this volume) are shown in Figure 27. Because both sites are away from the coast and well out into the California Current (Fig. 1), similar SST time series are found at both sites. The SST record of Site 1020 is offset from that of Site 1016 by ~3°C throughout the last 160 k.y. Site 1018, which is approximately midway between Sites 1020 and 1016 (Fig. 1), has an SST profile that lies midway between them (Mangelsdorf et al., in press). The 3°C temperature difference between Sites 1016 and 1020 is slightly larger than the modern annual gradient of about 1.5°C but is the same as the spring seasonal temperature gradient (Levitus and Boyer, 1994).The temperature gradient should become smaller if the California Current gains strength.

The glacial maxima are marked by the largest alkenone SST differences between the north and south (about 5°C), even though the minimum alkenone SST does not coincide with this time. Doose et al. (1997) have interpreted the alkenone SST difference between north and south to be a measure of California Current strength, where maximum difference in SST equates to minimum flow in the current. Using this model, California Current flow at the glacial maxima (MIS 2 and 6) is around 60% of the flow at other times. If other data sets confirm a weak California Current at glacial maxima, this represents a major decrease in heat exchange between the subarctic and subtropical Pacific. Ultimately this may help to end glaciation: low import of heat into the North Pacific lowers evaporative water loss from the North Pacific, perhaps curtailing water supply to North American glaciers.

Longer time series that cover more than the last glacial-interglacial cycle are still uncommon for the Leg 167 drill sites because of the high sedimentation rates and the thickness of sedimentary sections that must be analyzed. We now have the most information about multiple glacial cycles from Site 1020 from Northern California (Fig. 21 and Fig. 28; SST from Kreitz et al., Chap. 10, this volume; pollen data from Heusser et al., Chap. 17, this volume; oxygen isotopes and carbon measurements from Lyle et al., Chap. 11, this volume; biogenic silica from Kuroda et al., Chap. 14, this volume). The longer records illustrate some of the possible directions for future research.

MIS 12/11 Transition

One of the important features in the Site 1020 time series is a transition at roughly the MIS 12-11 boundary (~423 ka on the Imbrie et al. [1984] time scale) apparent in SST, CaCO₃, and biogenic silica (Fig. 28). The glacial stages prior to MIS 12 are marked by relatively small glacial/interglacial changes in alkenone SST—around 3°-4°C for the maximum difference between glacials and interglacials. Beginning with the MIS12-11 glacial-interglacial pair, the alkenone SST difference between glacials and interglacials doubled and the SST variance strongly increased at higher frequencies. The MIS12/11 boundary roughly marks the end of a relatively high CaCO₃ interval and a high biogenic silica interval. Part of the decrease may result from higher terrigenous input in the younger section (Kuroda et al., Chap. 14, this volume). Lower terrigenous dilution in the section older than MIS 11 may explain part but not all of the elevated biogenic contents of the pre-MIS 11 interval. Sedimentation rates were 20% lower in the section older than MIS 11 based upon the stage boundaries in Figure 28, but biogenic silica and CaCO₃ contents average a factor of 2 higher. The older sediments have significantly higher biogenic MAR's as well as somewhat lower terrigenous dilution than post-MIS 11 sediments.

The biogenic silica record (Kuroda et al., Chap. 14, this volume) indicates that the period of high biogenic deposition may have lasted only for about 200 k.y. and began at about the MIS 15/16 boundary. We have only coarse-scale shipboard measurements of CaCO₃ and C_org along with reflectometry records through the interval from MIS 11 to 16 to estimate the length of the high CaCO₃ burial interval (Lyle, Koizumi, Richter, et al., 1997). Both would suggest that high CaCO₃ burial began nearly coincident with the beginning of the high biogenic silica burial, near the MIS 15/16 boundary. Lyle et al. (Chap. 11, this volume) have noted a similar CaCO₃ high prior to MIS 10 at Site 1021, but no detailed records exist to the south. The estimated CaCO₃ percent from reflectometry for the California Borderlands sites (Sites 1011, 1012, 1013, and 1014) would suggest that this event is much less prominent in the southern California margin. Detailed longer records from the California Borderlands need to be analyzed and compared to the records from the north.

Alkenone SST records would suggest that the period of relatively warm glacial intervals in the North Pacific may have lasted from the Brunhes/Matuyama boundary to MIS 12. We do not yet understand why the temperate North Pacific Ocean remained relatively warm during this period; nor do we understand why there was high biogenic deposition in the period from MIS 15 to 11, although it may be related to the 400-k.y. orbital forcing. Obviously we need to analyze more long time series to better understand the evolution of North Pacific oceanographic conditions during the Pleistocene. Most of the Leg 167 drill sites have the potential for time series of 1000 k.y. or longer.

Tantalizing comparisons can be made with time series from other regions in the Pacific. In the eastern equatorial Pacific at Site 847, Murray et al. (1995) noted peak CaCO₃ percent and MAR for the last 1 m.y. at about 400 ka. The interval between 0.35 and about 1.1 Ma has higher than average CaCO₃ percentages and CaCO₃ MAR. In the Peru Basin in the southeastern Pacific, high CaCO₃ percentages and CaCO₃ MAR are also found in the interval 0.4-1.1 Ma (Weber et al., 1995). These intervals are somewhat longer than the Site 1020 record. It is not yet clear if there is any significance to the coincidence of high CaCO₃ around 400-600 ka in the eastern Pacific, but better regional comparisons are clearly needed.

CaCO₃ and C_org Records

High-resolution records of carbon burial for the last 500 k.y. are being generated for Sites 1020, 1018, and 1012 (Lyle et al., Chap. 11, this volume; M. Lyle, unpubl. data). Figure 28 shows the record from Site 1020. Whereas the general Pacific pattern of high CaCO₃ in glacial intervals typically occurs, the Site 1020 time series is much more spiky than a typical orbitally forced record—it has high variance in frequencies higher than the orbitally forced ones. The CaCO₃ events can be correlated to other northern California drill sites (Sites 1019 and 1018; Lyle et al., Chap. 11, this volume) and show the potential for millennial-scale correlation between sedimentary records within the region throughout the late Pleistocene.

C_org has a tendency to be higher in the interglacial periods, in contrast to CaCO₃. The level of variability of C_org is significantly smaller than CaCO₃, however, and has a two-times range vs. a ~10-times range for CaCO₃. Surprisingly, the CaCO₃ spikes tend to also be high C_org events. C_org has a slight tendency to lead a CaCO₃ burial spike, however.

Lyle et al. (Chap. 11, this volume) have suggested that the orbital-scale glacial-interglacial variation in CaCO₃ may be driven in part by changes in the saturation state of [CO₃]⁼ in Pacific deep water, but that the majority of the variance in the record is driven by changes in the production of CaCO₃ in surface waters combined with change in CaCO₃/C_org in the particulate rain to the seafloor. This hypothesis implies that the surface of the North Pacific Ocean is highly sensitive to millennial-scale climate processes.

One of the lines of evidence that the CaCO₃ record is driven by the surface ocean depends upon the SST record. Highest CaCO₃ deposition is not inversely correlated with SST but instead occurs when SST is in the range of about 7.5°-9°C. Even the prominent CaCO₃ spike in the middle of MIS 8 occurs after the minimum SST and in the shoulder of moderate SST immediately prior to the first MIS 7 interglacial event. The 7.5°-9°C temperature range marks one in which sediment trap experiments record high CaCO₃ rain with high CaCO₃/C_org leaving modern North Pacific surface waters (Lyle et al., Chap. 11, this volume). High CaCO₃ rain and high CaCO₃/C_org maximizes the delivery of CaCO₃ to the ocean floor and minimizes the C_org-mediated dissolution at the bottom.

Further work linking the biogenic sedimentary records (including biogenic silica) to specific high-resolution micropaleontological and stable isotope records will be crucial to establish the change in the surface North Pacific Ocean through the Pleistocene and will provide a better understanding of the implications of these changes for climate and biogeochemical cycles.

The Terrestrial-Marine Connection

One of the other priorities from Leg 167 is to better understand the linkages between the oceanographic conditions along the California margin and terrestrial climate. Some researchers have been studying the provenance and mass accumulation rates of terrestrial aluminosilicate detritus in the Leg 167 cores (Irino and Pedersen, Chap. 23, this volume; Hovan et al., Chap. 18, this volume) to understand changes in continental erosion and transport. In northern and central California (Sites 1018 and 1020), Hovan et al. (Chap. 18, this volume) note a link between terrigenous MAR and orbital-scale climate variability. Highest terrigenous MARs are associated with glacial periods, indicating increased supply from source regions, because of exposure of continental shelves and/or increased fluvial sediment discharge. Provenance of the material suggests a primary source region from the Klamath Mountains or Cape Mendocino region to Site 1020 during glacial intervals. In the south, Tada et al. (Chap. 25, this volume) and Irino and Pedersen (Chap. 23, this volume) also show a glacial-interglacial change, suggesting higher off-shelf transport or fluvial discharge for the period MIS 3-2.

Pollen from coastal plant communities is also buried in sediments along the California margin and provides a way to link changes in the terrestrial climate to changes in paleoceanographic conditions offshore (Heusser, 1998; Heusser et al., Chap. 17, this volume). In Northern California, peak interglacials are marked by a distinctive floral succession. On deglaciations, alder (Alnus) pioneers reforestation by the interglacial community, followed by lowland oak (Quercus), and finally by redwood (Sequoia). Redwood is typically found in the late stages of strong interglacial periods. We only have information about the last 160 k.y. from southern California (Site 893, Heusser, 1995), but see a strong glacial-interglacial signal alternating between oak and nonarboreal pollen in the interglacials and juniperus or pine in the glacials. Pine most strongly appears at the glacial maximum in MIS 2.

Figure 28 allows the comparison of the Sequoia record with that of the benthic foraminiferal oxygen isotopes and the alkenone SST record at Site 1020. Sequoia pollen is only present in significant quantities when the SST goes above 10°C. For greater than 90% of the last 500 k.y. in Northern California, Sequoia has been a rare genus and not the dominant tree type. If one looks closely at the record it becomes apparent that the peak Sequoia pollen percentages lag peak SST. It also is clear that the highest abundance of Sequoia pollen is found during intervals when SST remained for the longest time above 10°C (e.g., MIS1 and 11). Short intense SST warmings (e.g. MIS 5e) don't seem to allow time for the Sequoia forest to reach its maximum extent. The very short SST warmings in MIS 7 lead to a small response by Sequoia.

The response in the plant community to glacial-interglacial cycles is complex, being a response to temperature, insolation, precipitation, seasonality, and competition. Nevertheless, there is a typical response to the glacial cycle that should eventually provide an important means to understand how insolation changes affect these factors on the adjacent continent. There is not only a typical response to each glacial-interglacial change but also a unique response specific to individual glacial-interglacial pairs (Heusser et al., Chap. 17, this volume) that implies an evolution to the system through time or some inherent instabilities that we do not yet understand.

By lumping the pollen data from genera that are common in the Holocene coastal region from Northern California to Canada into an 'interglacial lowland forest' community (sitka spruce, western red cedar, western hemlock, redwood, oak, ferns; Fig. 29) and comparing the aggregate to SST we find a more linear response from the group than from any of the individual genera. It appears that changes in SST cause a relatively consistent expansion or contraction of the community as a whole, even though there are also important shifts in importance of species within the community.

MST Records and Color Reflectometry

It is now typical within ODP to measure wet bulk density, magnetic susceptibility, natural gamma-ray activity, and P-wave velocity at high resolution using the shipboard multisensor track (MST). On Leg 167, we also measured color reflectometry by two systems: the ODP digital color imaging system (Nederbragt et al., Chap. 29, this volume) and the Oregon State University (OSU) SCAT spectral reflectometer (see "Explanatory Notes" chapter in Lyle, Koizumi, Richter, et al., 1997; Harris et al., 1997). Nederbragt et al. (Chap. 29, this volume) took combined data from three-channel (RGB) video images to make 1-cm color averages downcore, using the L*a*b* system (lightness, green-red, and blue-yellow). The OSU system, in contrast, measures 1000 channels of color information from the near-infrared to near-ultraviolet wavelengths at discrete 2 cm spots downcore.

The most useful records for the high-resolution stratigraphic correlation proved to be the color data, because sediments with fast sedimentation rates tend to be rich in C_org and gassy. Gaps and expanded sections caused significant artifacts in GRAPE (gamma-ray attenuation porosity evaluator) bulk density data at many sites, whereas diagenesis highly attenuated the magnetic susceptibility signal after ~20 mbsf, except when diagenetic sulfides were formed in high quantities. The GRAPE density and magnetic susceptibility records are extremely good offshore at sites with more moderate sedimentation rates (e.g., Sites 1010, 1016, and 1021), as we will show when discussing the Neogene sediments.

Color changes in sediments are caused by a change in the relative abundance of different sedimentary phases. The best known example is CaCO₃, where changes in its abundance change the lightness of the sediment color. Color changes can also occur when other minerals change abundance. When hematite concentration changes in equatorial Atlantic sediments, for example, the amount of red color in the sediment changes (Balsam et al., 1995).

The resolution of video images and the rapidity at which they can be collected makes this an attractive tool to study sedimentary core material. Significant problems are encountered when using video images quantitatively, however, because of nonuniform illumination and the changing color of the light source over time. These require postcruise processing before use (Nederbragt et al., Chap. 29, this volume). Figure 30 shows the processed profiles for the four drill sites in the California Borderlands. Also included are benthic oxygen isotope time series (where available) for independent correlation between sites (Andreasen et al., Chap. 8, this volume; Hendy and Kennett, Chap. 7, this volume). The L* curves have extremely high resolution and provide a simple way to correlate between drill sites or to other cores from the California Borderlands region. They also may provide a high-resolution analytical tool if we can understand the cause of the signal observed.

We have CaCO₃ and C_org data from Site 1011 at 10-cm resolution (Lyle et al., Chap. 11, this volume) to compare to the L* curve and test if the change in sediment lightness is essentially a measure of CaCO₃ content (Fig. 31). A quick comparison shows that the L* and CaCO₃ time series have significant differences. The intervals that are most anomalous in the L*-CaCO₃ comparison also happen to be extremes in the C_org record. High C_org is a major factor causing a darker average sediment color. A simple linear model can be created to estimate L* by subtracting 85% of the C_org variation (expressed as percentage of C_org range in the intervals) from CaCO₃ in percentage of its range. Because L* is a mixed signal nearly equally weighted to both CaCO₃ and C_org variability, one cannot use the L* time series as a proxy for either CaCO₃ or C_org in the California Borderlands, even though it is highly useful for correlations. Further work must be done to use the other color information recorded by the digital imaging system to distinguish between C_org and CaCO₃.

The OSU SCAT spectral reflectometer, which takes 1024 channels of reflection data, has a stronger ability to distinguish between sedimentary components by their color spectra. To get the spectral resolution, one trades off depth resolution. Maximum resolution on the SCAT is 20 mm vs. ~0.25 mm for the digital imaging technology. Typical spacing of SCAT measurements on Leg 167 were 4-6 cm. The use of this data is still in its infancy because of the lack of good calibration data sets. Independent high-quality data sets of sedimentary components must be measured on the core material to determine the spectral response of each component.

Some calibration has been done using shipboard data, with impressive results. Figure 32 shows a comparison between an estimate of C_org by the SCAT reflectometer on Hole 1019C (Eel River Basin) and the splice C_org data from Lyle et al. (Chap. 11, this volume), which is a combination of data from Holes 1019C and 1019E. None of the data used for the reflectometry calibration is from this data set. The estimated and measured C_org profiles overlay each other except for the interval between about 21 and 30 mcd. The discrepant interval is one with essentially no CaCO₃ in the section, as compared to an average of about 5% elsewhere, so it is abnormally darker than the average sediment. For this reason, the C_org was overestimated. The C_org estimate can probably be improved by a recalibration with sediments expressing a full range of sediment variability.