High-resolution climatic evolution of coastal northern California during the past 16,000 years

Authors


Abstract

[1] Holocene and latest Pleistocene oceanographic conditions and the coastal climate of northern California have varied greatly, based upon high-resolution studies (ca. every 100 years) of diatoms, alkenones, pollen, CaCO3%, and total organic carbon at Ocean Drilling Program (ODP) Site 1019 (41.682°N, 124.930°W, 980 m water depth). Marine climate proxies (alkenone sea surface temperatures [SSTs] and CaCO3%) behaved remarkably like the Greenland Ice Sheet Project (GISP)-2 oxygen isotope record during the Bølling-Allerod, Younger Dryas (YD), and early part of the Holocene. During the YD, alkenone SSTs decreased by >3°C below mean Bølling-Allerod and Holocene SSTs. The early Holocene (ca. 11.6 to 8.2 ka) was a time of generally warm conditions and moderate CaCO3 content (generally >4%). The middle part of the Holocene (ca. 8.2 to 3.2 ka) was marked by alkenone SSTs that were consistently 1–2°C cooler than either the earlier or later parts of the Holocene, by greatly reduced numbers of the gyre-diatom Pseudoeunotia doliolus (<10%), and by a permanent drop in CaCO3% to <3%. Starting at ca. 5.2 ka, coastal redwood and alder began a steady rise, arguing for increasing effective moisture and the development of the north coast temperate rain forest. At ca. 3.2 ka, a permanent ca. 1°C increase in alkenone SST and a threefold increase in P. doliolus signaled a warming of fall and winter SSTs. Intensified (higher amplitude and more frequent) cycles of pine pollen alternating with increased alder and redwood pollen are evidence that rapid changes in effective moisture and seasonal temperature (enhanced El Niño–Southern Oscillation [ENSO] cycles) have characterized the Site 1019 record since about 3.5 ka.

1. Introduction

[2] Recent studies [Sabin and Pisias, 1996; Mix et al., 1999; Pisias et al., 2001] demonstrate that sediments deposited beneath northeast Pacific waters off the coasts of northern California and southern Oregon are an excellent monitor of latest Quaternary climatic change. These studies and that of Kienast and McKay [2001] provide convincing evidence that during the past 16,000 years, climate changes in this middle-latitude region of the northeast Pacific are strongly coupled to the Greenland Ice Sheet Project (GISP)-2 δ18O and to North Atlantic sediment records. Kienast and McKay [2001] cite the atmospheric modeling study of Peteet et al. [1997], who argue that Pacific water vapor provides this link by affecting the growth of the Laurentide ice sheet and thereby influencing thermohaline circulation in the North Atlantic. Detailed comparisons of marine microfossil and pollen assemblages in cores taken off northern California and Oregon also demonstrate a strong link between oceanic and continental climate change for most of California and Oregon [Pisias et al., 2001].

1.1. Setting

[3] Waters off the coasts of northern California and southern Oregon lie near the modern-day boundary between the subarctic and subtropical gyres of the North Pacific, where they are influenced by the strength and character of the California Current [Huyer, 1983]. The California Current begins at the divergence of the North Pacific Drift, which lies between about 42° and 50°N along the western margin of North America (Figure 1). During much of the spring and summer, juxtaposition of the North Pacific High and the North American Low results in strong, persistent northwesterly winds which induce coastal upwelling and lead to high biologic productivity [Hood et al., 1990]. Winters are influenced by a weakened North American Low, the migration of the North Pacific High to south of 35°N, and the migration of the jet stream and associated Aleutian low-pressure cells to an average position of 38°N. Winters are typically mild, wet, and stormy, with southwesterly winds and a noticable lack of upwelling [Huyer, 1983].

Figure 1.

Map of the middle-latitude northeast Pacific off the west coast of the United States showing ODP Site 1019 and other core sites discussed, as well as the California Current.

[4] Sediments in this region are characterized by low CaCO3 (typically <5%), but they are relatively rich in organic carbon (Corg) content (>1%). Due to an active tectonic setting and yearlong terrigenous input from rivers, Holocene sedimentation rates off northern California and southern Oregon commonly exceed 20 cm/kyr [Lyle et al., 2000], allowing the possibility of high-resolution studies. Lyle et al. [2000] demonstrate that millennial-scale events in %CaCO3 and %Corg can be regionally correlated in Pleistocene and Holocene sediments from offshore California and Oregon. They note that peaks of weight %CaCO3 and Corg are not coincident, and they are not correlative to glacial and interglacial stages in a straightforward manner. However, in support of the studies of Dean et al. [1997] and Mix et al. [1999], Lyle et al. [2000] observe that higher values of Corg tend to occur in the Holocene and in earlier interglacials, presumably reflecting increased biologic productivity associated with enhanced coastal upwelling. Lyle et al. [2000] present evidence that CaCO3 events are controlled by-production, not dissolution, and they postulate that higher CaCO3 values occur when coastal upwelling is weakened, causing increased blooms of coccolithophorids.

1.2. Previous Paleoclimate Studies

[5] A number of recent studies off the coasts of California and Oregon have estimated the amount of sea surface temperature (SST) change between the last glacial maximum (LGM) and the present. Estimates of deglacial SST change based on oxygen isotopes and/or planktic foraminifers have generally been higher in the south than in the north. Off southern California planktic foraminifers estimate SST change between the LGM and present above 5°C [Mortyn et al., 1996, 6–10°C; Kennett and Ingram, 1995, ca. 7°C; Hendy and Kennett, 2000, ca. 7°C). Off northern California and southern Oregon, on the other hand, Ortiz et al. [1997] propose a deglacial SST change of 3.3° ± 1.5°C SST, similar to the 2–3°C change suggested by Mix et al. [1999].

[6] Doose et al. [1997] give alkenone-based estimates of deglacial SST change of ≤1–2°C south of 36°N and 3–5°C north of 36°N based on a study of 17 cores between 33° and 42°N. Their northern estimates are supported by the studies of Prahl et al. [1995] (ca. 4°C off southernmost Oregon) and Kreitz et al. [2000] (ca. 5°C at 41°N). The studies of Herbert et al. [1995], who report a 2–3°C warming from the LGM to the Holocene in the Santa Barbara Basin, support Doose et al's [1997] estimates for the south; however, Ostertag-Henning and Stax's [2000] alkenone results at Ocean Drilling Program (ODP) Site 1017 (34°32′N; Figure 1) suggest a somewhat larger deglacial SST change of ca. 4°C.

[7] Sabin and Pisias [1996] presented radiolarian-based sea surface temperature (SST) reconstructions for the middle-latitude coastal region of the northeast Pacific for the past 150,000 years in 12 well-dated deep-sea cores ranging in latitude from 33.60°N to 54.42°N. Their factor analysis of radiolarian assemblages suggested that the regional pattern of oceanic circulation reached its present configuration at 13.0 ka. They argued that prior to 13.0 ka, the North Pacific Drift and the Transition Zone lay further to the south in response to a more southerly position of the North Pacific high pressure cell. The reconstructions of Sabin and Pisias [1996] revealed regional differences in both deglacial SST changes and SST variations within the Holocene. For example, they suggested that deglacial SST change between 20.0 and 10.0 ka was about +2–3°C between 33° and 36°N, but rose to +4°C between 37° and 43°N. This observation is closer to the conclusions of Doose et al. [1997] based on alkenones. Within the Holocene, Sabin and Pisias [1996] predicted SST variations of 1–2°C in the region between 33° and 36°N and typically 2°C within the region between 37° and 43°N, again supporting previous alkenone studies such as Prahl et al. [1995] and Doose et al. [1997]. Sabin and Pisias [1996] report that SSTs were at a maximum at about 10.0 ka within both the northern and southern regions; however, their reconstructions suggest that the middle part of the Holocene (ca. 8.0 to 5.0 ka) stands out as the coolest period (by 1–2°C) of the Holocene in the northern region (37° to 43°N).

[8] Studies of modern conditions, such as those by Huyer [1983] and Strub et al. [1987], emphasize the major regional differences in seasonal cycles of currents, SST, winds and sea level along the North American coast between 33°N and 48°N. Strub et al. [1987] point out that during the fall and winter, monthly mean winds are northward for 3–6 months in the region north of 35°N (Point Conception), whereas south of Point Conception these winds are near zero or weakly southward. They stress that the magnitudes of the seasonal cycles of all variables are at a maximum between about 38°N and 43°N (northern California to southern Oregon), implying a strong sensitivity to climatic cycles such as El Niño–Southern Oscillation (ENSO).

[9] Within these climatically sensitive waters off northern California and southern Oregon, two recent high-resolution studies of the last deglaciation stand out. Mix et al. [1999] measured the δ18O of the planktic foraminifer Neogloboquadrina pachyderma at ODP Site 1019 (41.682°N, 124.930°W, 980 m water depth) (Figure 1). Their results showed relatively large climatic oscillations during the last deglaciation, including Younger Dryas (YD) values that were about 0.5 to 0.8‰ greater (or 2–3°C cooler if they are entirely due to temperature) than the intervals immediately proceeding and following. Mix et al. [1999] remarked that δ18O trends through the Bølling-Allerod (decreasing), Younger Dryas (increasing), and earliest Holocene (an initial decrease centered on 10.0 ka, followed by an increase toward 8.0 ka) were mirrored by percentages of left-coiling N. pachyderma, suggesting that these trends reflect sea surface temperature at Site 1019. They argued that surface ocean changes differ little from those of the Bølling-Allerod and Younger Dryas observed in Greenland and the North Atlantic. Mix et al. [1999] went on to suggest that the dominance of left-coiling forms of N. pachyderma after 8.0 ka suggested that during the middle part of the Holocene, surface water conditions above Site 1019 were similar to those of the glacial maximum. Mix et al. [1999] observed a decrease in values of δ18O for N. pachyderma after about 5.5 ka that might imply warming of sea surface temperatures; however, because they did not observe change in N. pachyderma coiling ratios, they chose to interpret this isotopic change to be a reflection of a reduction in salinity in the upper ocean.

[10] Pisias et al. [2001] compared detailed radiolarian and pollen records from two piston cores from offshore southern Oregon (W8709A-13PC and EW9504-17PC), from ODP Site 1019 off northern California and from piston core EW9504-13PC off central California (Figure 1), for the past 60 kyr with the ice core δ18O record. They argued that at wavelengths >3000 years, warm events in Greenland correlate to increased coastal upwelling off Oregon, a decline in very cold North Pacific radiolarian assemblages, and increases in pollen associated with wetter coastal environments. Pisias et al. [2001] observed that warming in coastal regions is due to reduced advection of the California Current, even though it is moderated by an increase in coastal upwelling. They inferred that SST variability in this region of the northeast Pacific during the past 150 kyr was about 2°C.

[11] In a study of the climate record of the past 500 kyr in cores recovered by ODP Leg 167 off California, Lyle et al. [2001] concluded that terrestrial vegetation responded primarily to regional SST. They also noted that coastal ocean productivity appeared highest when SST was moderately high, not during peak interglacial conditions nor during insolation maxima.

[12] There have been no previous attempts to interpret SSTs using diatoms in the Holocene of offshore California. However, various researchers, such as Barron in the studies by Gardner et al. [1988], Sancetta et al. [1992], and Hemphill-Haley and Gardner [1994], have recorded Holocene fluctuations of the warm-water diatom, Pseudoeunotia doliolus, which may be an indication of fluctuating SSTs.

[13] Mix et al. [1999] and Pisias et al. [2001] emphasize the importance of the latest Pleistocene and Holocene record at ODP Site 1019, as a well-dated record, in a key climatic region, that has relatively high sediment accumulation rates (ca. 40 cm/kyr). High-resolution studies using phytoplankton-based proxies (i.e., diatom assemblages and alkenones from coccolithophorids) have not been completed at Site 1019. With this in mind, a high-resolution study of the past 15,000 years at Site 1019 was initiated, combining study of diatom and pollen assemblages, alkenone production, and sediment properties on paired samples.

2. Materials and Methods

[14] Samples were chosen at 5-cm intervals from the upper 7.10 m composite depth (mcd) of ODP Site 1019 after referring to Mix et al.'s [1999] age model, which suggests that the last 15,000 years were sampled. In the case of sediment and pollen studies, these samples were supplemented by samples previously studied by Lyle et al. [2000] and Heusser et al. [2000] in order to bring the record back to 16,000 years B.P.

2.1. Age Model

[15] Mix et al. [1999] obtained paired AMS dates on mixed planktic and benthic foraminifers from seven samples taken between 4.17 and 8.21 mcd of Hole 1019A. Their only date above 4 mcd came from a piece of bark at 2.84 mcd. They used the CALIB 4.1 program [after Stuiver and Reimer, 1993] to calculate calendar ages for these radiocarbon dates, following Southon et al. [1990] and Toggweiler et al. [1989] respectively, in choosing reservoir ages of 800 years for planktic foraminifers and 1750 years for benthic foraminifers (Table 1). Kienast and McKay [2001] accepted the 800-year-old reservoir age for planktic foraminifers for radiocarbon ages younger than 12,000 years; however, they used a 1100-year-old reservoir age for planktic foraminifers older than 12,100 radiocarbon years after Kovanen and Easterbrook [2002]. With this in mind, the calendar ages of the planktic foraminiferal AMS dates from 7.11 and 8.21 mcd have, therefore, been recalculated on Table 1 using a reservoir age of 1100 years. All ages discussed in this manuscript are calendar years B.P. unless indicated otherwise.

Table 1. Radiocarbon Age Data From the Upper 8.21 mcda of ODP Site 1019 and Calculated Calendar Agesb
Hole-Core-Section, Interval, cmMaterial (Foraminifers)Depth, mcd14C Age, years B.P.Reservoir Age, yearReservoir Corrected Age, years B.P.Calendar Age, years B.P.
  • a

    Meters composite depth.

  • b

    The top five dates are from this paper, the others are from Mix et al. [1999]. The calendar ages for the lower two planktic ages are recalculated using a reservoir age of 1100 years.

1019D-1H-1, 14–20Globobulimina spp.0.162040 ± 401750290 ± 40310 ± 80
1019D-1H-1, 14–20mixed benthic forams0.162130 ± 401750380 ± 40430 ± 90
1019D-1H-1, 54–60Globobulimina spp.0.573040 ± 4017501290 ± 401250 ± 60
1019D-1H-1, 114–120mixed benthic forams1.165100 ± 70017503350 ± 7003670 ± 910
1019D-1H-2, 53–60mixed benthic forams2.076400 ± 4017504650 ± 405430 ± 90
 
Data fromMix et al. [1999]
1019A-1H-1, 40–46bark2.846030 ± 1506030 ± 1506820 ± 240
1019A-1H-2, 22–28mixed planktic4.179950 ± 1108009150 ± 11010,290 ± 370
1019A-1H-2, 22–28mixed benthic4.1710,810 ± 12017509060 ± 12010,270 ± 160
1019A-1H-2, 97–103mixed planktic4.9210,210 ± 1208009410 ± 12010,450 ± 410
1019A-1H-2, 97–103mixed benthic4.9211,130 ± 8017509380 ± 8010,480 ± 410
1019A-1H-2, 122–128mixed planktic5.1611,410 ± 17080010,610 ± 17012,470 ± 570
1019A-1H-2, 122–128mixed benthic5.1611,540 ± 9017509790 ± 9011,120 ± 390
1019A-1H-3, 38–44mixed planktic5.8111,580 ± 14080010,780 ± 14012,700 ± 320
1019A-1H-3, 38–44mixed benthic5.8113,290 ± 220175011,540 ± 22013,430 ± 380
1019A-1H-3, 78–84mixed planktic6.2111,950 ± 11080011,150 ± 11013,020 ± 150
1019A-1H-3, 78–84mixed benthic6.2112,830 ± 90175011,080 ± 9012,990 ± 240
1019A-1H-4, 24–30mixed planktic7.1113,350 ± 120110012,250 ± 12014,120 ± 630
1019A-1H-4, 24–30mixed benthic7.1114,260 ± 140175012,510 ± 14014,340 ± 630
1019A-1H-4, 128–134mixed planktic8.2115,080 ± 120110013,980 ± 12016,650 ± 360
1019A-1H-4, 128–134mixed benthic8.2116,210 ± 190175014,460 ± 19017,200 ± 400

[16] To improve the age model for the younger part of Site 1019, four samples (each 40 cc in size) were obtained for AMS dating from the less heavily sampled Hole 1019D. Correlation of physical property measurements [Shipboard Scientific Party, 1997; Lyle et al., 2000] between Holes 1019A, 1019C, and 1019D allows integration of ages at Site 1019. The samples were sieved at 63 μm and picked for benthic foraminifers that were carbonized at the USGS laboratory of Jack McGeehin, and five AMS radiocarbon dates were obtained for these samples at the Lawrence Livermore AMS facility (Table 1). These dates have been corrected to calendar years using the CALIB 4.3 program [Stuiver et al., 1998], assuming a benthic foraminifer reservoir age of 1750 years [Mix et al., 1999]. The large error associated with the sample from 2.07 mcd is due to the relatively small number of benthic foraminifers that were obtained from that sample.

[17] The new calendar ages and those of Mix et al. [1999] are plotted on Figure 2a. Stuiver et al. [1998] note that the CALIB 4.3 program differs from the CALIB 4.1 program used by Mix et al. [1999] only by the use of the calibration data set that is selected for each sample rather than for a batch of samples. Application of the CALIB 4.3 program to Mix et al.'s [1999] AMS dates yields the same calendar ages as previously reported by him. Following Mix et al. [1999], calendar ages from planktic foraminifers are preferred over those from benthic foraminifers. An exception, however, is the sample from 5.16 mcd, where a planktic date of 12.47 ± 0.57 ka. seems anomalously old compared to the benthic date of 11.12 ka, which was used here. In proposing an age model for the Holocene of Site 1019, Mix et al. [1999] used their planktic foraminiferal calendar age of 10.29 ± 0.37 ka at 4.17 mcd and extrapolated a line through the assumed age of 0 ka for the uppermost sediments recovered at Site 1019, implying a sedimentation accumulation rate of ca. 41 cm/kyr. Figure 2a shows that the five new calendar AMS ages on benthic foraminifers above 2.1 mcd give a better fit to a sediment accumulation rate of 38 cm/kyr Below 2.1 mcd, however, adoption of an age model for Site 1019 is not so straightforward (Figure 2a) [Mix et al., 1999]. A simple approach would be to use a line corresponding to a sediment accumulation rate of 55 cm/kyr, which passes through the error bars of all of Mix et al.'s [1999] preferred calendar dates (Figure 2a), with the exception of the planktic date from 4.17 mcd (10.29 ± 0.37 ka). Acceptance of this date would necessitate using a sedimentation rate of >450 cm/kyr between 4.17 and 4.92 mcd, a 75-cm-long interval that does not stand out as being any different from surrounding intervals on the lithologic logs and core photographs [Shipboard Scientific Party, 1997], or in the detailed sediment property data of Lyle et al. [2000]. Use of an averaged (55 cm/kyr) sediment accumulation rate between 8.21 and 2.1 mcd seems more appropriate until further age dating has been completed.

Figure 2.

A) Preliminary age model for ODP Site 1019 showing calendar ages calculated from radiocarbon ages (Table 1) and their age constraints. B) Adjusted age model for Site 1019 incorporating correlation tie points “A”, “B”, “C”, and “D” between the alkenone SST curve and the GISP-2 oxygen isotope record of Grootes and Stuiver [1997] (Figure 3a).

2.2. Alkenone Methods

[18] Sediment samples were treated and analyzed for alkenones according to the methods described in Herbert et al. [1998]. As determined by replicate analyses of selected sediment samples, precision of better than 0.005 units (∼0.15°C) is routinely obtained. Alkenone concentrations were calculated (ΣC37; nmol/g dry sediment) as the sum of the C37:2, C37:3, and C37:4 molecules, and it is assumed that ΣC37 represents a proxy for haptophyte algal productivity [Villanueva et al., 1997; Rosell-Melé et al., 1997]. These concentrations reflect the productivity and preservation of one important component of marine phytoplankton, but are subject to dilution from other sources.

[19] Comparison of core top and LGM-SSTs from Prahl et al. [1995], Doose et al. [1997], and Herbert et al. [1998] show excellent agreement between laboratories Sea surface temperatures were estimated on Site 1019 samples using the Prahl et al. [1988] calibration, which is indistinguishable from a global sediment core top temperature relation subsequently published by Muller et al. [1998].

[20] Alkenone temperature estimates reflect a weighted average of production and preservation from different depths and times of the year. In gyre settings, alkenone production is greatest in a subsurface chlorophyll maximum zone, and during winter months [Prahl et al., 1993]. The U37k′ in sediments underlying such regions reproduces this bias from mean annual temperature (by ∼1°C) [Kreitz et al., 2000]. Closer to the continental margin, alkenone production apparently occurs very near the surface, and with very little seasonal bias [Herbert et al., 1998]. Our core top temperature estimate of 11.7°C agrees well with the mean annual SST of 11.9°C taken from the World Ocean Atlas [Levitus, 1994], and coastal SST recorded at Trinidad Beach (41.05°N: 11.6°C) and Crescent City (41.7°N: 12.0°C) [data obtained from http://nemo.ucsd.edu/]. We therefore assume that down core variations in alkenone temperature estimates at Site 1019 reflect changes in mean annual SST, although such changes might have been driven by changes in seasonal temperature patterns.

2.3. Sediment Methods

[21] The CaCO3 and Corg data of Lyle et al. [2000] have been supplemented by additional carbon analyses also run at Boise State University. Bulk sediment was combusted at 1000°C in a combustion furnace and the resulting CO2 determined by coulometry. Calcium carbonate was then removed by acidification and another combustion was done to give the organic carbon fraction. Calcium carbonate was determined by subtracting the organic carbon fraction from the total carbon fraction (from the first combustion) and dividing by 0.12, the fraction of C in CaCO3. The data are available in the work of Lyle et al. [2000] as well as on request to the authors.

[22] Lyle et al. [2000] estimated the accuracy and precision of their data by including two standards in each sample run and by repeat analyses of the unknown samples. Their results on total carbon and organic carbon measurements reveal that the average difference between the repeated samples is <0.01% carbon.

2.4. Diatom Methods

[23] Samples were disaggregated in distilled water and then processed by boiling them in 30% hydrogen peroxide and 37% hydrochloric acid. The acid was then removed through several washings in distilled water separated by at least 4 hr of settling and decanting away of the liquid. The final sample was stored in a vial containing at least 7–10 times as much distilled water as sample. To prepare slides, the vial was shaken and a drop of the suspension was taken after 5–10 s of settling from near the top of the vial, transferred to a 22 × 20 mm cover slip and allowed to dry on a warming tray overnight. Slides were then mounted in Hyrax (index of diffraction = 1.71). At least 300 individual diatoms were counted using the counting techniques of Schrader and Gersonde [1978] by making random traverses of the slide under the light microscope at 1250×.

[24] Resting spores of Chaetoceros dominate the Holocene diatom assemblage of Site 1019, so following the recommendations of Sancetta [1992], other diatom taxa have been tabulated on a Chaetoceros-free basis based on counts of at least 200 other individuals. Chaetoceros spores are easily transported downslope and their inclusion in assemblage data can obscure environmental interpretations suggested by other diatom taxa.

[25] Diatom abundances were estimated by recording the number of diatom valves encountered while making vertical traverses of the slide (length of traverse = 22 mm) at 1250× (total area covered per traverse = 4.114 mm2). Random traverses were made until >300 diatom valves were counted.

2.5. Pollen Methods

[26] Standard processing procedures (KOH and HF digestion and acetolysis) were preceded and succeeded by sieving through 7-μm nylon screening [Heusser and Stock, 1984]. An exotic tracer (Lycopodium) was used to determine pollen concentration. Taxonomic identification of pollen was based on comparison with modern pollen reference collections from western North America. At least 300 pollen grains were identified from each sample.

3. Results

3.1. Alkenone Results

[27] Alkenone SST estimates for the past ca. 15,000 years at Site 1019 are plotted versus age on Figure 3a according to the age model of Figure 2. Additional studies are underway by Herbert to extend this record further back into the last glacial. Similar to the alkenone SST record of Kienast and McKay [2001] in piston core JT96-09PC off the coast of British Columbia (48°54.7′N, 126°53.4′N, water depth 920 m), the Site 1019 alkenone SST record bears remarkable resemblance to the GISP-2 oxygen isotope record of the past 15,000 years [Grootes and Stuiver, 1997], faithfully recording the Bølling-Allerod, Younger Dryas, and the earliest part of the Holocene. Peteet et al. [1997] argue that North Pacific SST changes documented in the late glacial and Younger Dryas provide a rapid mechanism for widespread hemispheric cooling due to the loss of water vapor as a greenhouse gas. Using a Goddard Institute for Space Studies (GISS) general circulation model (GCM), they show that colder North Pacific SSTs result in a cooler, drier air blowing over North America leading to increased sensible heat fluxes there and lower temperatures over land. As a mechanism for Laurentide Ice Sheet growth, Peteet et al. [1997] cite tracer modeling of precipitation sources in the GISS GCM by Koster et al. [1986] which indicates that the North Pacific is a major moisture source for northern Canada in fall and winter which are the significant seasons for snow accumulation. They further note that unlike their previous sensitivity experiments using a colder North Atlantic, their North Pacific SST sensitivity run produced a response in the entire northern hemisphere.

Figure 3.

A) The alkenone SST record for ODP Site 1019 plotted for the past 16,000 years (bold line) using the preliminary age model of Figure 2a compared with the GISP-2 oxygen isotope record (normal wt. line). Suggested correlation tie points “A”, “B”, “C”, and “D”. Modern annual SST at Site 1019 is shown by a dashed line. (ka* refers to 1000 calendar years before present.) B) The alkenone SST record for Site 1019 plotted for the past 16,000 years using the adjusted age model of Figure 2b compared with the GISP-2 oxygen isotope record. These data can be accessed at ftp://ftp.ngdc.noaa.gov/paleo/contributions_by_author/barron2002/.

[28] With Peteet et al.'s [1997] arguments in mind, it seems reasonable to attempt to correlate the early part of the Site 1019 alkenone record with the GISP-2 oxygen isotope record at least within the constraints of the Site 1019 age model. The Bølling-Allerod (14,600–12,900 cal years B.P) stands out clearly by alkenone SSTs that are >10°C, but rise to peaks of ca. 13 and 12°C. Correlation of prominent events “D” and “C” in two records suggest that the age model used for Site 1019 (Figure 2a) may yield ages which are about 200 years older than possible coeval events in the GISP-2 record. A distinctive interval of lower alkenone SSTs (<8°C, or 3–4°C cooler than surrounding intervals) correlates remarkably well with the Younger Dryas (12.9 to 11.6 ka), although it is possible that the age of its termination (event “B”) may be estimated to be about 200 years too young by the Site 1019 age model of Figure 2a.

[29] During the Holocene, alkenone SSTs fluctuated between about 10 and 13°C, with highest values (typically >12°C) in the early part of the Holocene (11.6–8 ka), generally reduced (averaging about 10.5°C) values in the middle part of the Holocene (7.6–3.2 ka), and moderate (ca. 11.4–12.2°C) values in the later part of the Holocene (<3.2 ka). These fluctuations in Holocene SST at Site 1019 seem more extreme than the GISP-2 oxygen isotope record, so correlation is not so apparent. A possible exception, however, is a sharp decline in SST to ca. 10°C (event “A”) that is estimated by the age model of Figure 2 to be 8.0 ka but may coincide with the distinctive 8.2 ka cold event of the GISP-2 oxygen isotope record. Recent studies such as that of de Menocal et al. [2000] indicate that the 8.2 ka cold event is recognizable even in low-latitude SST records.

[30] Thus, correlations of the Site 1019 alkenone SST record with the GISP-2 oxygen isotope record suggest that the age model of Figure 2 may yield ages which are about 200 years too young in the intervals of events “A” and “B”, and about 200 years too old in the intervals of events “C” and “D”, if one were to accept synchronicity between the two climate curves. On Figure 2b the stratigraphic positions of these four events are plotted on the Site 1019 age model. An alternate age model, which includes these 4 events and also passes through the error bar of Mix et al.'s [1999] planktic foraminiferal AMS date (10.29 ± 0.37 ka) from 4.17 mcd, is shown on Figure 2b. This new age model is preferred because it falls within the constraints of the accepted AMS dates and improves the correlation of events “A” through “D” between the alkenone SST record and the GISP-2 oxygen isotope record. The alkenone SST record from Site 1019 is plotted versus age using this new age model on Figure 3b.

3.2. Sediment Composition Results

[31] Weight % CaCO3 and Corg data from Site 1019 are plotted versus age on Figure 4 following the adjusted age model of Figure 2b. Opal percent was estimated using the shipboard Oregon State University color reflectance data (A. C. Mix and D. C. Lund, unpublished data, 1997) and multiple linear regression equations generated from the Leg 167 site-survey color cores calibrated to measured opal data [Shipboard Scientific Party, 1997]. The estimated opal % is plotted on the bottom of Figure 4. The levels of opal seem low and may be underestimated.

Figure 4.

The record of sediment properties—weight percent and mass accumulation rate of calcium carbonate (A), total organic carbon and its mass accumulation rate (B), and opal content estimated using a multiple linear regression equation generated from the Leg 167 site-survey color reflectance and opal data [Shipboard Scientific Party, 1997] (C) at ODP Site 1019 for the last 16,000 years compared with changes in sediment accumulation rates (C). (ka* refers to 1000 calendar years before present.) These data can be accessed at ftp://ftp.ngdc.noaa.gov/paleo/contributions_by_author/barron2002/.

[32] These sediment proxies define six distinctive intervals within the sediment record of the last 16,000 years at Site 1019 (Figure 4). The late glacial (16.0–14.7 ka) is marked by relatively moderate values of weight % CaCO3, low values of Corg and even lower estimated opal content, suggesting reduced biologic productivity [Lyle et al., 2000]. The Bølling-Allerod is characterized by relatively high values of weight % CaCO3 and Corg and higher estimated opal content, supporting the findings of Lyle et al. [2000] and Pisias et al. [2001] who both report that increased biogenic productivity (carbonate and opal) occurred during the Bølling-Allerod off northern California and southern Oregon. The Younger Dryas is distinguished by reduced weight % CaCO3 and Corg and low estimated opal content, marking a period of reduced biologic productivity, again supporting the findings of Lyle et al. [2000] and Pisias et al. [2001]. A turbidite at 6.47 mcd (visible in the core photographs at 1019C-1H-4, 46–47 cm), stands out at ca. 13.3 ka by its greatly reduced weight %CaCO3 and Corg values and by decreased estimated opal content.

[33] The early Holocene between ca. 11.4 and 8.2 ka is characterized by relatively high (>4%) but fluctuating wt. % CaCO3 compared to the rest of the Holocene, a feature that is typical for the California margin [Lyle et al., 2000]. This early Holocene interval is marked by generally higher wt. % Corg and moderately higher estimated opal content than surrounding intervals, possibly suggesting increased biologic productivity and diatom content. A peak in estimated opal content between ca. 11.6 and 11.1 ka is present in the earliest Holocene, where values (ca. 4%) are comparable to those of the Bølling-Allerod.

[34] One might argue that the wt. % CaCO3 may be controlled by terrigenous dilution. As shown by Figure 4, however, CaCO3 MAR effectively mirrors wt. % CaCO3 at Site 1019 for much of the record of the past 16 kyr, with offsets corresponding to changes in the sediment accumulation rate, the most extreme of which occurring between ca. 11 and 10 ka. This relationship argues that carbonate productivity is being recorded and not varying effects of terrigenous dilution [see Lyle et al., 2000]. Multitracer sediment trap studies by Lyle et al. [1992] show that surface water biologic productivity events, as measured by CaCO3 and Corg, are faithfully recorded in the hemipelagic sediments off southern Oregon. Furthermore, Lyle et al. [2000] demonstrate that fluctuations on the order of ca. 10–2 wt. % CaCO3 and ca. 1.5–0.5 wt. % Corg can be accurately correlated for at least the past 500 kyr in the hemipelagic sediment record off coastal California, implying regional paleoproductivity patterns.

[35] Decreasing wt.% CaCO3, relatively low, but increasing wt.% Corg, and moderate to low estimated opal content typify the middle part of the Holocene between ca. 8.2 and 3.3 ka. An abrupt increase in wt.% Corg and abrupt increase in estimated opal content at ca. 3.4–3.3 ka marks the onset of a late Holocene interval distinguished by increased diatom content (increased biologic productivity?). It should be pointed out, however, that the mass accumulation rate (MAR) of Corg is a better proxy for biologic productivity proxy than wt.% Corg [Gardner et al., 1997], because it accounts for effects of dilution. Consequently, Corg MAR is plotted on Figure 4 (bold line) for comparison.

[36] Corg MAR shows the effects of sediment accumulation rate changes (Figure 4), displaying a major peaks in the early Holocene between about 11.0 and 10.0 ka, where the age model (Figure 2b) calls for an abrupt increase in the sediment accumulation rate (to ca. 80 cm/kyr) that is not supported by gross lithologic changes [Shipboard Scientific Party, 1997] but does coincide in part with a 44-cm-thick interval of laminated sediment [Pike et al., 1998; Lyle et al., 2001]. According to Pike et al. [1998], this laminated interval (and two older ones) appears to represent enhanced productivity, because it is characterized by high numbers of upwelling diatom flora and extensive benthic foraminiferal colonization, which would rule out anoxic conditions. It is possible that further refinement of the Site 1019 age model would reveal that this early Holocene Corg MAR is not so pronounced.

[37] At ca. 5.4 ka the sediment accumulation rate declines from ca. 47 to 38 cm/kyr (Figure 2), causing a decrease in both CaCO3 and C-org MAR (Figure 4). A similar late Holocene trend of decreasing Corg MAR is shown for piston core W8709-13PC (42.12°N) off the southern Oregon Coast by Dean and Gardner [1998], suggesting a regional productivity pattern.

3.3. Diatom Results

[38] The relative abundance of selected diatom taxa during the past 15,000 years at Site 1019 is shown on Figure 5 using the age model of Figure 2b. The very poor preservation of diatoms in the late glacial interval precludes their study. Chaetoceros spores are tabulated as a relative percent of the entire diatom assemblage, whereas other taxa are tabulated on a Chaetoceros-free basis. An estimate of relative diatom abundance (number of diatom frustules encountered per vertical traverse (length 22 mm of a slide at ×1250) is included for comparison, but it is recognized that this estimate may be biased by minor variations in the preparation of the diatom strewn slides.

Figure 5.

Relative percent of selected diatom taxa (B) and the number of whole diatom valves counted on a vertical transect (at X1250) of a microscope slide (width 22 mm) (C) during the past 15,000 years at ODP Site 1019 compared with the alkenone SST record (A). (* = counts of these diatom taxa are on a Chaetoceros-free basis; ka* refers to 1000 calendar years before present.) These data can be accessed at ftp://ftp.ngdc.noaa.gov/paleo/contributions_by_author/barron2002/.

[39] Sediment trap studies at nearshore Multitracers site W8709A-13PC (see Figure 1) display a maximum flux of Chaetoceros spores from June to August 1988, with a secondary maximum from September to November [Sancetta, 1992], agreeing with observations by others that the taxon is a characteristic component of spring-summer upwelling (see references in the work of Sancetta et al. [1992]). Yet, Sancetta [1992] points out that surface sediments in core W8709A-13PC are most similar in composition to trap samples from the less productive fall and early winter. She suggests that advection of displaced material off the shelf may be responsible for these differences and cautions that robust diatoms, such as Chaetoceros spores, may be preferentially preserved in sediments. Thus, the relative percentage of Chaetoceros spores in sediments may not be a direct indicator of the strength of summer upwelling and diatom production.

[40] Chaetoceros spores account for between 70 and 85% of the total Holocene diatom assemblage of Site 1019 with little apparent change throughout most of the past 15,000 years (Figure 5). Virtually no vegetative valves of Chaetoceros were recorded. The dominance of Chaetoceros resting spores, however, appears to drop by about 10% for the past 1000 years, although this may be a result of better diatom preservation in these younger samples resulting in increased numbers of delicate diatoms that are not preserved in older samples.

[41] Thalassionema nitzschioides is a temperate to subtropical taxon that appears to represent spring-season production within a broad region extending seaward from the coastal zone [Sancetta, 1992]. Thalassionema nitzschioides is most abundant in the Midway Mulitracers sediment trap during the early spring (March to April) according to Sancetta [1992], and it may be indicative of the oceanic upwelling that occurs at that time. Thalassionema nitzschioides comprises between 40 and 70% of the Chaetoceros-free Holocene diatom assemblage, with abundances of 20% or more characterizing the early and late Holocene, and relatively high (ca. 65–70%) and less variable (5–10%), values during the middle part of the Holocene (Figure 5).

[42] Pseudoeunotia doliolus is a diatom associated with the warm-water Central Gyre that enters coastal waters off northern California and southern Oregon in late August to October, when the California Current relaxes [Sancetta, 1992]. In the Multitracer sediment trap series, it is most common at the Gyre site (17%) and decreases shoreward (7% at Midway and 3% at Nearshore) [Sancetta, 1992]. Pseudoeuntia doliolus appears to represent warm, highly stratified waters with low nutrient availability. Pseudoeunotia doliolus makes up >12% of the Chaetoceros-free diatom assemblage between ca. 10.2 and 7.6 ka, with increased values corresponding to higher alkenone SSTs within this interval of the Holocene. Reduced P. doliolus (generally <10%) characterizes the middle Holocene interval between about 7.6 and 3.3 ka, an interval that is marked by lower alkenone SSTs. Diatoms appear to have been more common during selected intervals of the middle Holocene than they were during the early Holocene; however, the estimated opal content data (Figure 4) argue that diatom productivity was slightly reduced compared to that of the early Holocene. During the middle part of the Holocene, P. doliolus was largely replaced by an increasing contribution of Thalassionema nitzschioides, suggesting enhanced oceanic upwelling and a reduced influence of the central gyre [Sancetta, 1992]. Increased oceanic upwelling during the middle part of the Holocene is also supported by the alkenone SST data, if one accepts the arguments of Popp and Prahl [2001] that alkenone SSTs reflect temperatures at the depth of the thermocline (ca. 50 m).

[43] At ca. 3.2 ka, a three-fold increase in the relative percent of P. doliolus coincides with a permanent warming of alkenone SSTs by ca. 1°C. In the modern ocean, P. doliolus enters waters in the region of Site 1019 during the late summer and early fall, when the California Current relaxes and waters of the Central Gyre move shoreward. The 1°C increase in alkenone SST therefore may reflect warming of winter SSTs off northern California caused by increasing wintertime insolation in the late Holocene.

[44] Neodenticula seminae is a subarctic diatom which is persistent, but rare (typically <10%) throughout most of the Holocene at Site 1019. On average, N. seminae is more common (4–8%) during the late early to middle part of the Holocene (ca, 9.0–3.5 ka) than it is during the later part of the Holocene (typically <3%), supporting the suggestion from the alkenones of a warming of wintertime SSTs during the late Holocene. Alternatively, northward migration and stabilization of the position of the North Pacific subarctic front may have occurred during the late Holocene as suggested by West [1990].

3.4. Pollen Results

[45] The record of selected pollen taxa from Site 1019 versus time is presented in Figure 6 using the age model of Figure 2b. During the Bølling-Allerod, the appearance of alder (Alnus) in the sage brush (Artemesia) and pine (Pinus) pollen assemblages that characterize the late glacial marks the beginning of the replacement of glacial open pine-woodland by Holocene oak and coastal redwood assemblages. The rapid climatic shifts seen in SSTs offshore are similar to the high-amplitude variations in pine and alder, suggesting alternating cool, dry and warm, wet environments. Alder, a colonizer of disturbed environments, grows in moist sites, along streams and in marshy places in coastal redwood and mixed evergreen forests, and in moist places in the north Coastal Range [Barbour and Major, 1977; Barbour et al., 1980]. Increased alder in sediments deposited on the California margin may thus reflect expansion and/or recolonization of alder habitats related to increased precipitation and/or increased runoff from snowmelt during deglaciation [Heusser and Shackleton, 1979]. A brief interval of cooler and/or drier conditions suggested by the pine peak, alder minimum, and decrease in oak (Quercus) at the beginning of the Younger Dryas is also inferred from a slight reexpansion of pine, spruce and western hemlock ca. 13.0 ka in nearby northeast Pacific sediment cores [Heusser and Barron, 2002]. The subsequent rapid rise in the successional alder to its highest values at the end of the Younger Dryas (ca. 11.55 ka) implies an increase in effective moisture and warming (flooding?, melting of snowpacks?) as nonglacial vegetation replaced glacial vegetation [Heusser and Barron, 2002]. The lack of a strong expression of Younger Dryas cooling in pollen assemblages from Site 1019, from other cores from offshore Oregon and northern California [Pisias et al., 2001] and from onshore Oregon [Grigg and Whitlock, 1998] contrasts with the sustained low SSTs recorded in Site 1019. The pollen data imply a rapid recovery from a brief cooling early in the Younger Dryas, whereas in adjacent coastal waters warming and increased biologic productivity (increased carbonate and Corg MAR) appear to have occurred nearly 600 years later at the end of the Younger Dryas (Figures 35).

Figure 6.

Relative percent of selected pollen taxa during the past 16,000 years at ODP Site 1019 compared with the alkenone SST record. (ka* refers to 1000 calendar years before present.) These data can be accessed at ftp://ftp.ngdc.noaa.gov/paleo/contributions_by_author/barron2002/.

[46] The ca. 5°C increase in alkenone SST during the earliest Holocene coincides with a doubling of alder percentages to the peak values that characterize the late glacial/Holocene boundary in numerous Pacific Northwest pollen diagrams [Heusser, 1985]. The apparent lag in alkenone SST may be an artifact of differing sampling intervals. After ca. 11.0 ka, trends in pollen at Site 1019 (a rapid decrease in alder, relatively high oscillating values of pine, and a gradual increase in oak) appear to reflect the onset of warm dry conditions that are observed throughout the Pacific Northwest during the early Holocene [Heusser, 1985; Heusser et al., 1985]. Between ca. 9.0 and 5.5 ka, oak and redwood displayed modest increases, while decreasing alder values and generally high pine values point to a persistence of a warming trend into the middle part of the Holocene., as seen elsewhere in the Pacific Northwest [Heusser et al., 1985; Rypins et al., 1989; Koehler and Anderson, 1994; Leonard and Reasoner, 1999].

[47] A rapid and steady increase of redwood, beginning ca. 5.5 ka, marks the development of the “modern” north coast temperate rain forest [Worona and Whitlock, 1995]. The increased cover of conifers along with a modest rise in alder and modest decline in oak suggests cooling and/or an increase in effective moisture. and decreased seasonal contrast. In general, winters became milder and the summers possibly became cooler and characterized by the presence of coastal fog [Heusser, 1998]. Presumably, this trend reflects the onset and intensification of modern maritime conditions along the adjacent coast of northernmost California.

4. Chronologic Summary

[48] Prominent events in the evolution of the marine and coastal continental climate systems of northern California during the past 16,000 years are summarized on Figure 7.

Figure 7.

Evolution of continental and marine climate conditions in the northern California region of ODP Site 1019 during the past 16,000 yr. (ka* refers to 1000 calendar years before present.)

4.1. Late Glacial

[49] Only sediment and pollen data adequately cover the late glacial interval between ca. 16.0 and 14.7 ka. The late glacial (16.0–14.7 ka) is marked by relatively moderate values of weight % CaCO3, low values of Corg and lower estimated opal content, suggesting reduced biologic productivity [Lyle et al., 2000]. The domination of pine and prominence of sage brush, a remnant of glacial vegetation, suggests a dry, cool environment. The end of the interval is marked by low numbers of alder, oak, and coastal redwood, pollen that are indicative of the warmer, wetter conditions of the Holocene [Heusser et al., 2000; Pisias et al., 2001].

4.2. Bølling-Allerod

[50] During the Bølling-Allerod (ca. 14.6 to 12.9 ka), alkenone SSTs rose to average about 10–11°C and reached a maximum of ca. 13°C at about 13.6 ka. The Bølling-Allerod interval is characterized by higher wt.% CaCO3 and higher organic carbon MAR's than surrounding intervals. Increased estimated opal content and increased numbers of diatoms indicate that biogenic productivity and upwelling were relatively high during the Bølling-Allerod, supporting the conclusions of Mix et al. [1999] and Pisias et al. [2001]. Modest increases in alder, coupled with modest decreases in pine after ca. 14.8 ka may indicate a slightly warmer, wetter climate, but conditions appear to have been highly variable.

4.3. Younger Dryas

[51] At the beginning of the Younger Dryas (ca. 12.9 to 11.6 ka), alkenone SSTs decreased to <8°C, while both carbonate and diatom productivity declined sharply. The Younger Dryas is marked by a distinctive 0.5% decline in organic carbon content from the warm intervals on either side. Increased pine pollen during the early part of the Younger Dryas suggests cooler, drier climates, whereas increasing alder during its later half signals a trend toward warmer, wetter climates that began nearly 600 years earlier than the warming of SSTs offshore. Although Pisias et al. [2001] report that the Younger Dryas was characterized by reduced coastal upwelling in cores off Oregon and northern California, they do not observe a significant cooling of radiolarian SSTs. This contrasts with the conclusions of Mix et al. [1999] who suggest a 2–3°C cooling during the Younger Dryas at Site 1019 based on oxygen isotopic and planktic foraminiferal studies. Similarly, Kienast and McKay [2001] observed a 3 to 4°C cooling in alkenone SSTs during the Younger Dryas off British Columbia. These values are comparable to the 4–8°C cooling suggested by Hendy and Kennett [2000] in the Santa Barbara Basin based on isotopes and planktic foraminiferal assemblages.

4.4. Early Holocene

[52] The early Holocene (ca. 11.6 to 8.2 ka) was a time of generally warm conditions and moderate carbonate values (generally >4%). Alkenone SSTs were typically >11°C or more than 3°C warmer than the Younger Dryas and 1–2°C warmer than the middle part of the Holocene. Two prominent peaks where alkenone SSTs reached 13°C occurred between ca. 11.3 and 11.1 ka and again between ca. 10.2 and 9.9 ka. The warm-water diatom, Pseudoeuntia doliolus, was virtually absent in the Site 1019 record prior to 10.2 ka, supporting previous studies by Gardner et al. [1988], Sancetta et al. [1992], and Hemphill-Haley and Gardner [1994] off northern California and Oregon. The appearance of P. doliolus off northern California at ca. 10.2 ka may reflect the onset of a late summer-early fall migration of gyre waters into coastal waters that were favorable for diatom production. It is likely that the early Holocene increase in alkenone SSTs, warm-water planktic foraminifers [Mix et al., 1999], and P. doliolus at Site 1019 prior to ca. 8.0 ka was due to a generally weaker flow of the California Current, as the relative accumulation of CaCO3 (Figure 4) was rather high compared to that of diatoms (Figure 5).

[53] The mass accumulation rate (MAR) of organic carbon peaked to values that are more than twice those of the proceeding and following periods between ca. 10.8 and 10.0 ka, a possible deglacial productivity spike [Lyle et al., 2000] that coincides with a brief interval of apparent accelerated accumulations rates (Figure 2b). Organic carbon MAR's remained rather low (0.6–0.7 gm/cm2/kyr) for the remainder of the Holocene, supporting the conclusions of Gardner et al. [1997] that biologic productivity declined after the early Holocene in waters north of ca. 38°N off the California-Oregon.

4.5. Middle Holocene

[54] With the exception of a brief moderately warm event at ca. 7.9 ka, the middle part of the Holocene between 8.2 and 3.3 ka was characterized by alkenone SSTs (<11°C) that were typically 1°C cooler than the proceeding and following periods. Carbonate values continued to drop during the middle part of the Holocene at Site 1019, until they consistently remained below 2% after about 4.5 ka [see Lyle et al., 2000]. The mass accumulation rate of organic carbon at Hole 1019A decreased by ∼30% at ca. 5.2 ka (Figure 4), suggesting that biogenic productivity also declined late in the middle part of the Holocene.

[55] Reduced relative abundances of the gyre-diatom Pseudoeunotia doliolus (<10%,) between ca. 7.6 and 3.3 ka are indicative of a decreased influence of the Central Gyre at the location of Site 1019. An increased contribution of Thalassionema nitzschioides suggests enhanced oceanic upwelling [Sancetta, 1992] and a fairly broad California Current system. The persistent presence of Neodenticula seminae in the middle Holocene diatom assemblage indicates that subarctic waters exerted some influence over Site 1019.

[56] Relatively low abundances of the gyre-diatom P. doliolus during the middle part of the Holocene is commonplace in cores collected off northern and central California. Hemphill-Haley and Gardner [1994] show that P. doliolus displays a bimodal distribution similar to that shown at Site 1019 in three Holocene cores recovered off north central California (38.5°N) with early and late Holocene peaks of 10–20% (relative % of the total assemblage) separated by an interval of low (5% or less) relative percent centered in the middle part of the Holocene at about 5 ka. Sancetta et al. [1992] report a similar trend for P. doliolus during the early and middle parts of the Holocene of piston core W8709-13PC (42.12°N) off the southern Oregon Coast, including a two- to four-fold decline in its abundance at about 5.0 ka. Unfortunately, the Holocene record in W8709-13PC does not extend above 5.0 ka. Similarly, at Site 893 in the Santa Barbara Basin (34.17°N) Hemphill-Haley and Fourtanier [1995] report a reduced relative abundance of P. doliolus in the middle part of the Holocene between 6.3 and 5.4 ka.

[57] Cooler SSTs during the middle part of the Holocene after about 8.0 ka are also recorded at Site 1019 by Mix et al. [1999] and elsewhere along the central and northern California margin by Sabin and Pisias [1996] and Ostertag-Henning and Stax [2000]. Similarly, Prahl et al. [1995] report alkenone-based SSTs of ca. 13°C at ca. 9.0 and 2.3 ka, bracketing a middle Holocene that was 1–2°C cooler in cores off the coast of southernmost Oregon. Pisias et al. [2001], however, note that radiolarians record little or no cooling of SSTs during the middle part of the Holocene at either SITE 1019 or EW9504-13PC, although an SST decrease of ca. 2°C is recorded by them between about 6.0 and 4.0 ka in core EW9504-17PC off southern Oregon (Figure 1). In all three cores, Pisias et al. [2001] show a decline in the numbers of subtropical radiolarians during the middle part of the Holocene, which one might interpret to reflect a reduced influence of the Central Gyre. Middle Holocene cooling is not apparent in cores from the southern California continental borderland according to the alkenone studies of Herbert et al. [2001], nor is it recorded by Kienast and McKay [2001] off British Columbia, suggesting that it was limited in extent.

[58] Are these cooler middle Holocene SSTs indicative of increased coastal upwelling off much of California and southern Oregon? Van Geen et al. [1992] infer that a 30% reduction in Cd/Ca in the shells of foraminifers from a sediment core taken near the mouth of the San Francisco Bay reflects a reduction in coastal upwelling over the past 4,000 years. Similarly, Gardner et al. [1997], who synthesized evidence for biogenic productivity during the past 60 kyr in a suite of 17 cores along the California coast, report that both biogenic opal and organic carbon mass accumulation rates were at a maximum during the middle part of the Holocene (ca. 8.0 and 4.0 ka) between 36 and 39°N, implying that biologic productivity and presumably coastal upwelling were also high compared to modern values.

[59] On the other hand, an increasing contribution of redwood pollen after 5.2 ka (Figure 7) might be taken as evidence of increasing coastal upwelling at Site 1019 during the later part of the Holocene [Heusser, 1998; Heusser et al., 2000]. Forests of coastal redwood are best developed where fog associated with cold upwelling waters moderates summer temperatures (mean July temperature of 17°C) and ameliorates drought. Similarly, an increase in the relative number of diatoms per microscope slide traverse (Figure 5) and an increase in the estimated opal content of the sediment (Figure 4) after ca. 3.4–3.2 ka would seem to imply that diatoms became more common in the late Holocene samples studied; however, a late Holocene decline in sediment accumulation rates at Site 1019 (Figures 2b and 4) offsets these trends.

[60] Reduced alkenone SSTs after ca. 8.2 ka, a permanent decline in CaCO3% at ca. 8.2 ka, and a decline in the gyre diatom-Pseudoeunotia doliolus at ca. 7.6 ka are evidence that the California Current increased in width (and possibly in strength) during the middle part of the Holocene. Atmospheric modeling studies such as Bush [1999] and Clement et al. [2000], the latter of which argues that El Niño events were suppressed during the middle part of the Holocene, suggest that middle Holocene sediment records off California would be less likely to sample climatically warm years and therefore should appear relatively cooler than modern sediment records. A long-term trend of decreasing alder and pine pollen and increase in coastal redwood pollen between ca. 11.0 and 5.2 ka indicate a decline in seasonality (possibly warmer winters after the middle Holocene xerothermic interval). This would be expected as the Suptropical High weakened [Bartlein et al., 1998] and the California Current intensified in strength.

[61] The mid Holocene interval of reduced alkenone SSTs (ca. 8.2 to 3.2 ka) coincides with a period of widespread aridity in the western interior of North America [Fritz et al., 2001]. Dean et al. [1986] invoke an enhancement of westerly winds as a possible cause for this mid Holocene aridity. Bartlein et al.'s [1998] paleoclimate reconstructions for January of 6 ka using the NCAR Community Climate Model Version 1 both show a strengthening of upper level (500 Mb) westerly winds over the western US and increasing northerly surface winds off the California coast, which should be expected with decreased winter SSTs.

[62] This same ca. 8 to 4 ka mid Holocene interval coincides with reduced methane concentrations in ice core records from both Greenland and western Antarctica [Brook et al., 1999]. A reduction in the extent of tropical wetlands is commonly cited as the cause for a reduction of atmospheric methane concentrations during the mid Holocene [Brook et al., 1999]; however, a number of low latitude climatic records such as those of Haug et al. [2001] for Venezuela and Gasse [2000] for North Africa point to increased lake levels and/or precipitation in the tropics during the mid Holocene.

4.6. Late Holocene

[63] Between ca. 5.2 and 3.5 ka, a major increase in coastal redwood pollen and modest increase in alder pollen, coupled with a major decline in pine pollen, signal the establishment of modern mesic coastal communities and the transition to a more maritime climate, characterized by mild winters and cool summers dominated by fog, on the coast adjacent to Site 1019. This trend of increasing redwood pollen during the latter part of the Holocene at Site 1019 (Figure 6) is also seen in the records of other cores off California and Oregon. In piston core EW9504-17PC off southern Oregon (Figure 1) redwood pollen begins a steady increase at ca. 9.0 ka (from ca. 3% of the pollen assemblage to >10% after 7.0 ka), followed by a secondary increase (to >20%) at ca. 3.5 ka [Heusser et al., 2000]. Off the southern Oregon coast (W8709-13PC, 42°07.01′N, 125°45.00′N, water depth 2712 m; Figure 1), a steady increase in redwood pollen (from ca. 5% to >15%) occurs between ca. 10.0 and 5.0 ka. The last 5,000 years are missing in that core, but in piston core W87-0A-8 (42°15.74′N, 127°40.68′W, 3111 m water depth), which lies further seaward (Figure 1), a second, major increase in redwood pollen from ca. 10% to ca. 30–40% occurred between 3.0 and 2.0 ka. Further south, in piston core V1-80-P3 off the northern California coast (38°25.51′N, 123°47.77′W, water depth 1,600 m), an abrupt increase in Sequoia pollen (from ca. 10% to >20%) occurred at about 10.0 ka followed by a secondary increase (from ca. 20% to ca. 30%) at about 2.5 ka [Heusser, 1998].

[64] Intensified (higher amplitude and more frequent) cycles of pine (both in percentage and concentration), that are mirrored in alder and coastal redwood, suggest that rapid changes in effective moisture and seasonal temperature are likely present in the Site 1019 record after about 3.5 ka. The presence of strong millennial-scale oscillations in neoglacial vegetation on the Pacific Northwest coast is also documented in the well-dated, detailed (ca. 40 years average sampling interval) pollen records from Kasten core W77-10A-26K (44°50.2′N, 125°8.5′W, water depth 1822 m) [Heusser and Barron, 2002].

[65] These continental climatic events proceed a ca. 3.2 ka shift to warmer alkenone SSTs, a threefold increase in the relative percentage contribution of the gyre-diatom Pseudoeunotia doliolus (Figure 5), and increased diatom preservation (Figure 5) that together signal the onset of modern oceanographic conditions at Site 1019. In the modern ocean, P. doliolus enters waters in the region of Site 1019 during the late summer and early fall, when the California Current relaxes and waters of the Central Gyre move shoreward, resulting in a relatively steep east-west SST gradient. A late Holocene development of this sharp offshore SST gradient and an intensification of ENSO cycles [Clement et al., 2000] documented in coastal Peru after 3.2–2.8 ka [Sandweiss et al., 2001] may be responsible for the intensification of terrestrial and marine cycles seen in the Site 1019 record. At the same time, late Holocene warming of winter SSTs off northern California, as suggested by the alkenone data, would be favored by increasing wintertime insolation. The relative increase in coastal redwood pollen (Figure 7) after ca. 5.2 ka is possible evidence of warmer winters in adjacent coastal regions.

5. Conclusions

[66] Trends in marine climate proxies (alkenone SSTs and CaCO3%) from Site 1019 are remarkably similar to the GISP-2 oxygen isotope record during the Bølling-Allerod, Younger Dryas, and earliest part of the Holocene (ca. 15 to 8.2 ka) (Figures 3 and 4), suggesting a strong teleconnection, possibly through transport of Pacific water vapor [Peteet et al., 1997; Kienast and McKay, 2001]. Climatically warmer intervals of the Bølling-Allerod and early Holocene are characterized by higher CaCO3% values, implying a strong influence of oligotrophic, carbonate-rich waters [Lyle et al., 2000], such as can be found in modern California offshore waters south of about 36°N [Gardner et al., 1997], where the California Current is relatively weak and diatom production is relatively low.

[67] During the Bølling-Allerod (ca. 14.6 to 12.9 ka), alkenone SSTs rose to an average of about 10–11°C, reaching a maximum of ca. 13°C at about 13.6 ka. Variation in the relative abundances of alder versus pine pollen suggest cold/dry versus warm/wet cycles.

[68] At the beginning of the Younger Dryas (ca. 12.9 to 11.6 ka), alkenone SSTs decreased to <8°C, while both carbonate and diatom productivity declined sharply. Increased pine pollen during the early part of the Younger Dryas suggests cooler, drier climates, whereas increasing alder during its later half signals the start of a trend toward warmer, wetter climates.

[69] The early Holocene (ca. 11.6 to 8.2 ka) was a time of generally warm conditions and moderate carbonate values (generally ca. 9–10%). Alkenone SSTs were typically 12 to 13°C, or 4 to 5°C warmer than during the Younger Dryas. Warming and increased precipitation are inferred from the abrupt increase in the successional alder to 50% of the total pollen. After ca. 11.0 ka, modest rises in pine and oak pollen suggest warmer, drier climates in the Pacific northwest. The gyre-diatom Pseudoeunotia doliolus rose to prominence at 10.2 ka, when surface waters became more conducive to diatom production.

[70] The middle part of the Holocene (ca. 8.2 to 3.2 ka) was characterized by lower alkenone SSTs (<11°C). Lower numbers of P. doliolus (<10%) between 7.6 and 3.2 ka are evidence of a reduced influence of the Central Gyre at Site 1019, suggesting a relatively broad California Current. Calcium carbonate values dropped below 3% after ca. 8.2 ka and remained low for the rest of the Holocene. Increasingly warm and dry continental conditions are suggested for the early part of the middle Holocene by a steady decline in alder pollen and modest increase in oak; but starting at ca. 5.2 ka, coastal redwood (Sequoia sempervirens) and alder begin a steady rise, arguing for increasing effective moisture and the development of the north coast temperate rain forest.

[71] At ca. 3.2 ka a sustained ca. 1°C increase in alkenone SST and threefold increase in P. doliolus signaled a warming of fall and winter SSTs. Intensified (higher amplitude and more frequent) cycles of pine pollen (both in percentage and concentration) alternating with increased alder and redwood pollen suggest that rapid changes in effective moisture and seasonal temperature have characterized the Site 1019 record since about 3.5 ka. The modern maritime climate of the northern California coastal region with cool, coastal upwelling-dominated summers and relatively warm, wet winters was established during the late Holocene between approximately 5.2 and 3.2 ka.

Acknowledgments

[72] We are grateful to Nick Pisias for sending us a preprint of his paper and to Constance Sancetta and Pat Bartlein for discussion. Thanks are due to Jack McGeehin of the USGS for 14C dating of foraminifer samples. Scott Starratt of the USGS provided valuable research and editorial assistance. We are grateful to the Ocean Drilling Program for providing the samples. Walt Dean and Mary McGann of the USGS and two anonymous reviewers provided very helpful reviews of the manuscript. The thorough editorial review of Larry Peterson was also very helpful. Lyle was supported by NSF grants OCE-9811272 and EPS-0132626.

Ancillary