We have three basic lines of evidence to show that CaCO3 production and, probably, an increase in CaCO3 production relative to total productivity is responsible for the CaCO3 k.y.-scale events.
There is a large body of literature on the geochemical processes that control dissolution, and we will not review it here. CaCO3 burial is basically a competition between the processes that bring CaCO3 to the seafloor and those that destroy it at the bottom. A simple-minded mass balance can be stated as follows:
where CC stands for CaCO3 flux, rain refers to the fall of particulate matter to the bottom from the surface ocean, f(Corg rain) is the function that describes dissolution of CaCO3 within sediments by organic matter degradation, and CCinorganic refers to the dissolution flux by inorganic processes related to changes in calcite saturation of deep ocean waters. Archer (1991b) estimates that 30%-50% of the total CaCO3 that reaches the seafloor dissolves because of oxic degradation of organic matter, so it is clear that productivity can have both a positive effect (higher CaCO3 rain to the bottom) and a negative one (higher CaCO3 dissolution because of Corg degradation). To achieve a high burial rate of CaCO3, high productivity is needed but with a high CCO3/Corg in the particulate rain from the euphotic zone.
The California margin is highly susceptible to CaCO3 dissolution by Corg degradation, because it is a region of high Corg rain relative to CaCO3 (CCO3/Corg is ~0.5, Dymond and Lyle, 1994) as compared to the tropical Pacific (CCO3/Corg is ~2.0). A change in CaCO3 burial in the California margin could either reflect a change in rain of CaCO3 relative to Corg or, conversely, small-scale temporal changes in inorganic dissolution.
The geochemical processes controlling inorganic dissolution in the oceans and their glacial-interglacial changes have been relatively well investigated. However, almost nothing is known about the processes that cause high CaCO3 production relative to Corg, and how those processes might change through time. The basic implication of a change in CaCO3 production without an equivalent change in Corg production is that there must be a reorganization of the plankton ecosystem with a shift towards conditions that favor coccolithophorids and foraminifers in contrast to primary producers and zooplankton that do not produce carbonate, such as diatoms, radiolarians, or other "bare" algae.
The first line of evidence we present for a production control of the CaCO3 time series is a depth transect near Cape Mendocino (Fig. 15: Site 1019, 988 m; Site 1018, 2476 m; Site 1020, 3038 m; and Site 1021, 4212 m). We observe no trend with depth in terms of CaCO3 MAR for the k.y.-scale events, even though depth-dependent dissolution shapes the overall record of the deeper drill sites. Site 1021, the deepest of the northern California drill sites, best illustrates the depth effect. It has a much lower average CaCO3 MAR than any of the other sites in the depth transect and is very near the local carbonate compensation depth (Karlin et al., 1992).
When the k.y.-scale events are examined, we find that CaCO3 MAR peaks for which we have highest confidence (e.g., CC 2-1,~16 ka, with radiocarbon age control) is a factor of three higher than the average MIS 2 or early MIS 3 CaCO3 MAR no matter what water depth the drill site is located in. There is no obvious change in preservation with depth, as would be expected if the CaCO3 MAR were controlled by inorganic dissolution.
The major flaw with this argument is that the depth of calcite saturation in the North Pacific is relatively shallow, on the order of 700-800 m near the California margin based upon [CO3]= data presented in Feely et al. (1984) and a [CO3]=sat for calcite at ~1000 m (100 bar pressure) and 5°C of 63 mmol/kg, calculated from equations in Pytkowicz (1969). All sites in the depth transect are beneath ocean waters undersaturated with respect to calcite, although Site 1019 should be minimally affected.
The second line of evidence for production control of the CaCO3 MAR time series is based upon a combination of sediment trap data near Site 1020 (Lyle et al., 1992; Dymond and Lyle, 1994) and chemical arguments. We show that many of the peaks in the Site 1020 CaCO3 MAR record are larger than can be supported by changes in dissolution alone acting upon the modern CaCO3 particulate rain to the seafloor.
Archer (1991a) summarized arguments of change in deep-water [CO3]= on glacial-interglacial time scales and came to the conclusion that glacial Pacific deep water was enriched in [CO3]= by about 7-11 µmol/kg. At the depth of Site 1020, [CO3]= is about 8 µmol/kg less than calcite saturation (43 vs. 51 µmol/kg at saturation; calculated from data in Pytkowicz, 1969). The glacial/interglacial changes in ocean chemistry should have driven the 3-km horizon of the North Pacific essentially to saturation with calcite. If inorganic dissolution rather than production is the primary control of the CaCO3 record, one should have a record marked by events reaching saturation, or roughly reaching the same CaCO3 MAR.
We can also calculate the approximate size of a saturation CaCO3 MAR event based upon nearby sediment trap data and rough estimates of the effect of Corg degradation on CaCO3 preservation (Fig. 16). Two multiyear sediment-trap deployments were taken as part of the Multitracers project about 100 km north of Site 1020 (Lyle et al., 1992; Dymond and Lyle, 1994). The average CaCO3 rain (Dymond and Lyle, 1994) was 1.5 g/ cm2/k.y. at the Nearshore site (125°45´ W) and 0.9 g/cm2/k.y. at the Midway site (127°35´ W). Site 1020 should have a CaCO3 rain rate that is the average of these two, because it is in between them, or about 1.2 g/cm2/k.y. Of this fraction, essentially none (only about 0.1 g/cm2/k.y.) gets buried in the sediments today. Approximately 1.1 g/cm2/k.y. of the CaCO3 rain dissolves at the seafloor, but only a fraction of this is dissolved by inorganic processes. Archer (1991b) estimated that 30%-50% of the total CaCO3 rain falling to the seafloor is dissolved by CO2 released by the oxic degradation of Corg rain, the magnitude depending upon the CCO3/Corg ratio. The ratio we have observed at both Nearshore and Midway is larger than Archer modeled, so we expect that the dissolution by Corg rain is roughly half of the total CaCO3 rain. This leaves only about 0.5 g/cm2/k.y. to be dissolved by inorganic processes. By this back of the envelope calculation, all CaCO3 MAR peaks >0.6 g/cm2/k.y. are too large to have been caused by changes in calcite saturation alone.
The "saturation" CaCO3 MAR of 0.6 g/cm2/k.y. is near the baseline CaCO3 MAR at Site 1020 for MIS 2, 3, and 4 (Fig. 16) in accord with a scenario where glacial Pacific deep water neared calcite saturation. Most of the CaCO3 MAR peaks in the last 140 k.y., however, are larger than expected if CaCO3 MAR were responding to saturation of deep water alone. Two peaks in particular, CC 2-1 and CC 5-2, are significantly larger than the modern CaCO3 particulate rain. They can only have been caused by a combination of increased CaCO3 production and reductions of both inorganic and organically mediated dissolution. Other peaks above the saturation line must have been driven either by changes in the magnitude of CaCO3 particulate rain or by strong relative reductions of Corg particulate rain to reduce the organically mediated dissolution in the sediments.
The k.y.-scale CaCO3 MAR events were probably caused by a change in plankton community and a resultant increase in CaCO3 production and/or increase in CCO3/Corg in the particulate rain rather than because of an overall increase in primary productivity. There is definite structure to the MAR profiles that suggest an evolution of the surface conditions rather than diagenesis within the sediments. The CaCO3 MAR peaks typically do not coincide with Corg MAR peaks although there is a relationship between the two. A cross spectral analysis of the stacked Corg and CaCO3 MAR time series indicates that typical Milankovitch periods have significant coherence (Fig. 14), and periods at ~11, ~5, ~4, and ~ 1.6 k.y. However, the CaCO3 MAR record leads the Corg MAR record by about 2 k.y. at all significant periods except the 1.6 k.y. period. Clearly, k.y.-scale CaCO3 events are not high Corg events.
Ortiz et al.(1997) have already proposed a mechanism for higher CaCO3 production relative to Corg in the glacial northeastern Pacific based upon shifts from coastal upwelling to more diffuse curl-of-windstress upwelling offshore. Ravelo et al. (1997) have also suggested that anomalous late Pliocene CaCO3 MAR may result from a similar change in ecosystem dynamics. We will amplify upon this hypothesis based upon our examination of modern sediment trap data.
It is possible to treat the k.y.-scale changes in surface oceanographic conditions to be equivalent to movements of modern north Pacific plankton communities over the positions of the Leg 167 drill sites. If this is so, one would expect that the Ortiz et al. (1997) model of increased curl-of-windstress upwelling at the last glacial maximum to be represented by an increase in the equivalent modern community now found offshore at the drill sites being studied. One would also predict that the modern offshore community should produce more CaCO3 relative to Corg.
Subarctic radiolarian fauna were also an important part of the glacial plankton community, suggesting significantly colder water temperatures and subarctic oceanographic conditions (Moore, 1973; Prahl et al., 1995) so CaCO3 and Corg production in the modern subarctic Pacific may be a useful indicator of production under glacial conditions along the northern California margin. One of the surprises of the modern sediment trap data set is the relative strength of CaCO3 production in the northern Pacific.
A series of sediment trap experiments in the northeastern Pacific (Table 14; Martin and Knauer, 1983; Roth and Dymond, 1989; Lyle et al., 1992; Dymond and Lyle, 1994; Honjo et al., 1995) can be used to assess the modern CaCO3 and Corg production. We have collated data from deep sediment traps in Table 14, to examine only the particulate rain and to avoid flux changes associated with the dynamics of the euphotic zone. Note that we list the mass flux of C associated with CaCO3 in the table, not the CaCO3 flux, for better comparison to the Corg flux. To obtain the CaCO3 flux, multiply the CCO3 flux by 8.33. Figure 17 shows the location of sediment traps on a map of the COADS February SST climatology and shows how the CCO3/Corg changes throughout the region. The February SST was chosen because this is usually reasonably close to the alkenone temperature recorded at the top of the nutricline where coccolithophorids are most abundant (Prahl et al., 1993; Doose et al., 1997).
CCO3/Corg is highest in the modern particulate rain away from the coastal upwelling region. The highest CCO3/Corg is in the north, with a peak around 48°N, not where waters are warmer. However, highest standing stocks of coccolithophorids in the north central Pacific are also found north of 45°N (Okada and Honjo, 1973), primarily consisting of the cosmopolitan species Emiliana huxleyi. Only one sediment trap experiment (NS) had higher CaCO3 rain than the subarctic trap experiments JDF and PC85, and the NS mooring had a factor of 4 to 5 higher Corg rain than the subarctic traps. As far as CaCO3 burial is concerned, a switch anywhere along the California margin to the offshore environment around 48°N would both increase absolute CaCO3 rain and significantly reduce Corg-mediated losses from the sediments. It would be ideal to cause the k.y.-scale CaCO3 events.
There is some slight evidence that peak CaCO3 production relative to Corg may peak in the temperature range of 7°-9°C, the SST associated with the JDF and PC85 sediment trap moorings, because the ratio dips in the one sediment trap mooring to the north at 50°N (P82-84). This may be an artifact, however, because variability between years in this 3-yr P82-84 experiment was extremely high. In support of the idea that there is a moderate SST associated with peak CaCO3 rain, we point out that we have noticed a correlation between the rate of C37 alkenone burial (a biomarker for coccolithophorids, Kreitz et al., Chap. 10, this volume) and CaCO3 burial in preliminary data at Site 1020, with highest burial for both being associated with an alkenone-estimated SST between 8° and 10°C.