DISCUSSION

We can assess the validity of our results from Site 1020 in several ways. The -SST estimates from our Holocene samples at Site 1020 most closely resemble modern winter SST in the area (Levitus, 1994; Fig. 4). This means that any cooling in our record cannot result from a shift toward a colder season of productivity. Temperatures warmer than Holocene could reflect a bias associated with changes in the seasonal patterns of production toward spring or summer months; however, the Holocene marine isotope Stage (MIS) 5.5 temperature increase at Site 1020 is identical to that at Site 893, which shows no seasonal bias occurring today (Herbert et al., 1995).

The magnitude of glacial-interglacial SST variability holds remarkably constant over the last 450 k.y. with glacial temperatures around 5°C cooler and interglacial SSTs 2°-3°C warmer than today. The LGM-Holocene warming of ~4°C that we infer from Site 1020 also agrees with data from other locations along the California margin. To the south, -SST records at Site 893 (Santa Barbara Basin) and Core EW9504-03PC (32°04´, 117°35´) show ~2°-3°C warming from LGM-Holocene (Fig. 7; Herbert et al., 1995). In all three cases, MIS 5.5 is warmer than modern. To the north, -SST records at Multitracers Site Midway (W8709-8PC and TC) indicate roughly a 5°C warming since the LGM, but unfortunately, this record does not extend to MIS 5.5.

The consistency in glacial-interglacial warming strengthens our confidence in the data and strongly suggests a margin-wide response as climate changed from full glacial to interglacial conditions. Other marine isotopic and faunal studies (Sabin and Pisias, 1996; Ortiz et al., 1997) confirm the alkenone SST warming we see along the California margin, but these records extend only to the LGM.

At Site 1020, profiles of C_org and total alkenone concentrations (Lyle et al., Chap. 11, this volume) are greatest around warmest alkenone SSTs from the LGM-Holocene. This inferred increase in sedimentary C_org and alkenone concentrations is also seen farther south in alkenone data from Site 893 (Herbert et al., 1995) and in Core EW9504-03PC (T.D. Herbert et al., unpubl. data). To the north, Lyle et al. (1992) saw increases in C_org mass accumulation rates since the LGM in the Multitracers Transect. Ortiz et al. (1997) inferred increases in productivity since LGM at 42°N from isotopic gradients in Gephyrocapsa bulloides and from planktonic foraminiferal shell accumulation rates. Based on our data from Site 1020 and the data from these earlier studies, we infer that productivity may have been reduced during the LGM and has slowly increased in strength as modern conditions evolved. Terrestrial proxies such as coastal redwood pollen also suggest that upwelling ceased or diminished along the California margin (Heusser, Chap. 20, this volume). Sancetta et al. (1992) analyzed Core W8709-3 from 42°N and found less redwood pollen during the LGM, matching more recent data by Heusser et al. (Chap. 17, this volume). They concluded that weakened summer upwelling was responsible for the decreases in pollen. At Site 1020, correlations between redwood pollen and alkenone SST also suggest that upwelling was strongest during interglacial periods and weaker or nonexistent during glacial periods (Heusser et al., Chap. 17, this volume).

In addition to observed LGM-Holocene SST warming and Haptophyte productivity increases, alkenone SST records along the California margin also show similar relationships to ice volume. The coldest alkenone SSTs at Sites 1020 and 893 and Core EW9504-03PC all lead ¹⁸O maximum ice volume by at least 5 k.y. (Fig. 7; Herbert et al., 1995) in the south, and we speculate by a similar amount at Site 1020 in the north. The geographic consistency of this relationship suggests that there is an early response of SST warming with respect to ice volume along the California margin. The length of our record at Site 1020, coupled with the Site 893 record (Herbert et al., 1995), allows us for the first time to infer the behavior of sea-surface temperatures within the CCS beyond the last glacial period. We see that this early deglacial warming occurs consistently over ~350 k.y.; however, we are less confident in the preliminary age model much beyond this point.

A margin-wide SST response along the California margin would require a large-scale forcing during a glacial-interglacial transition. Two proposed mechanisms for cooler SSTs involve contributions from atmospheric effects resulting from the presence of glacial ice sheets and/or changes in thermohaline transport within the CCS.

An expansion of the Alaskan Gyre southward, perhaps because of changes in either oceanic or atmospheric circulation patterns, would aid in cooling the sea surface of the northeast Pacific Ocean. This encroachment of cool, subarctic water in the glacial northeast Pacific is similar to the scenario proposed by CLIMAP Project Members (1981) in the LGM North Atlantic, which shows the deflection of the Gulf Stream by the advancing polar front in the glacial North Atlantic. The expansion may have deflected the North Pacific Current/Kuroshio Current Extension and the west wind drift to the south, allowing cooler temperatures to exist for a longer duration in the northern part of the CCS.

A map of the southernmost extent of ice-rafted debris (IRD) in the northeast Pacific (Fig. 8) provides evidence independent of alkenone estimates that a cooler northeast Pacific existed during glacial times. ODP Site 887 (Gulf of Alaska) shows a large range in IRD abundance (Krissek, 1995) preserved there, and references to IRD as far south as off the coast of Oregon are made in Griggs and Kulm (1969). These deposits suggest that major ice sources must have been located proximal to the Gulf of Alaska and offshore of the northwest coast of North America during glacial times. Thus, the presence of IRD in the northeast Pacific supports the notion that the icebergs must have originated there. The petrology of the Site 887 IRD suggests that the provenance was the Alaskan coastal region and the Pacific Northwest of North America (Krissek, 1995). If SSTs in present areas of sea ice are analogs of the past, water temperatures proximal to sea ice would have been on the order of 2°-4°C (CLIMAP, 1981) for the LGM northern northeast Pacific. Therefore, the glacial temperatures as cold as 4°-6°C at Site 1020 are plausible.

The southward expansion of the northeast Pacific IRD belt is not necessarily the ultimate mechanism for glacial cooling of the sea surface. What could cause this belt to expand? For the IRD belt to propagate southward, the sea surface must be cool enough to allow the southerly migration of ice. COHMAP Members (1988) have modeled the atmospheric surface winds for the Holocene and LGM, and their results suggest an increase in the influence of the Aleutian Low, typically seen in modern Northern Hemisphere winter (Fig. 2; Huyer, 1983). We believe that the persistence of cyclonic wind fields over the northeast Pacific Ocean may have played a part in increasing the size of the Alaskan Gyre, causing its sphere of influence to be felt farther south than today. The idea of the southerly migration of the Alaskan Gyre is not new. Moore (1973), who documented subarctic radiolarian fauna along the northern California/southern Oregon border, was first to propose that Alaskan fauna existed this far south during the LGM. Lyle et al. (1992) acknowledged the possibility of the Alaskan Gyre planktonic community thriving off the coast of Oregon based upon productivity and organic carbon fluxes along the Multitracers Transect at 42°N. However, Lyle et al. (1992) suggested reduced coastal upwelling or subsurface replacement of nutrient-rich water as a better explanation of their data. Ortiz et al. (1997), using transfer functions from a 10-core transect at 42°N, saw that the strongest modern analogs of glacial fauna currently live in the Alaskan Gyre.

The northward winds modeled by COHMAP Members (1988) for the LGM along the California margin would have reduced upwelling at Site 1020. We see this reduction evidenced by lower C_org, terrestrial pollen, and SC₃₇ concentrations. One factor that may have contributed to the reduction of upwelling is an increase in northward, coastal flow of the geostrophic Davidson Current (Hickey, 1979), which would be possible under the atmospheric scenario proposed by COHMAP Members (1988).

We must also consider the possibility that a disturbance in the heat budget of the tropics and higher latitudes existed during the LGM, as suggested by Sr/Ca measurements of corals (Beck et al., 1992). If the tropical LGM anomaly was as large as this method predicts, this disturbance would have serious consequences regarding thermohaline reorganization along the California margin.

The modern relationship of colder SST and enhanced upwelling/organic preservation (La Niña) and anomalously warm SST with reductions in upwelling (ENSO) along the California margin does not persist on an orbital time scale. The increases in the records of sedimentary alkenone concentration, C_org, and terrestrial pollen assemblages all center around warmer SSTs at Site 1020. From this, we have inferred maximum productivity or preservation of sediment alkenone concentrations during warmer periods of SST, although periods that are more productive have existed during the past (for example, OIS 13).