5.1. Evidence of OMZ Intensification
 The OMZ, defined as that portion of the water column where oxygen is ≤0.5 mL/L, extends from approximately 750 to 1300 m water depth off the west coast of Vancouver Island. During the 1996 research cruise the lowest oxygen concentration (0.3 mL/L) was measured at a water depth of 920 m (i.e., the site where Core JT96-09 was collected). Unlike cores from the California and Mexican continental margins, no laminated sediments have been preserved in Core JT96-09. However, this does not necessarily imply that bottom water oxygen concentrations did not fluctuate because laminated sediments are only preserved when bottom water oxygen levels drop below 0.1 mL/L [Behl and Kennett, 1996]. Significant variations in bottom water oxygen concentration (i.e., OMZ intensity) over the past 16 kyr are inferred from changes in the accumulation of redox-sensitive trace metals and by changes in the assemblage of benthic foraminifera.
 The concentration of certain redox-sensitive trace metals in sediments is directly or indirectly controlled by redox conditions through either a change in redox state (e.g., Re and U) and/or speciation (e.g., Mo) which results in their accumulation or loss. Other redox-sensitive metals (e.g., Cd) have a single redox state, but readily react with the reduced forms of other elements such as sulphur and this results in their accumulation under reducing conditions. The authigenic flux into the sediment (i.e., the diffusion of metals from the overlying water column into the sediment) is primarily controlled by the oxygen concentration in the sediment and overlying bottom water. In general, when suboxic and anoxic redox boundaries (defined here, respectively, as the locations where the oxygen content falls to zero and sulphate reduction commences) are shallow, the flux from the overlying water column into the sediment is enhanced. Such conditions commonly occur when organic matter flux to the seafloor is high and/or oxygen concentration in the bottom water is low.
 The reduction and subsequent precipitation of Re and U begins once pore water O2 is depleted and this leads to their enrichment in both suboxic and anoxic sediments [Ravizza et al., 1991; Colodner et al., 1993; Crusius et al., 1996]. Cadmium has a single redox state, but form insoluble sulphides when trace amounts of H2S are available [Rosenthal et al., 1995]. This leads to minor Cd accumulation in suboxic sediments and large accumulations in anoxic sediments. In comparison, Mo enrichment is only observed in anoxic sediments [Francois, 1988; Emerson and Huested, 1991; Crusius et al., 1996; Morford et al., 2001; Ivanochko and Pedersen, 2004]. In the presence of >11 μM H2S, molybdate (MoO4−2) is converted to thiomolybdate (MoS4−2) which is scavenged by pyrite [Helz et al., 1996; Erickson and Helz, 2000]. The presence of zero-valent sulphur speeds up this process, and also aids in the reduction of Mo and formation of Mo-Fe-S complexes which are also rapidly scavenged by pyrite [Vorlicek et al., 2004].
 Modern sediments deposited within the OMZ off Vancouver Island become suboxic within millimeters of the sediment-water interface and are thus characterized by relatively low Mn/Al ratios (i.e., similar to, or less than, the average shale ratio of 0.010; Figure 3a). However, near-surface sediments never become fully anoxic and thus a large Mo enrichment above the typical lithogenic concentration (i.e., ∼0.6 μg/g [Morford et al., 2001]) is not observed in the upper 35 cm of Core JT96-09 (Figure 3b). In comparison, two periods of high Mo accumulation, relative to modern sediments, are observed in older sediments (zones 1 and 2, Figure 3b). These two zones are also enriched in Cd, Re and U (Figures 3c–3e). Molybdenum enrichment suggests that in the past sediments were more reducing (i.e., anoxic conditions existed in the near-surface sediments), although it does not imply that the overlying bottom water was anoxic. The fact that these sediments are not laminated indicates that the bottom water was somewhat oxygenated and that bioturbation was occurring.
 The first episode of marked trace metal enrichment is observed in the Allerød (zone 1, Figure 3). There is no evidence that this enrichment is result of metal remobilization due to oxygen influx (i.e., burn down) and subsequent reprecipitation in underlying reduced sediments because the distribution of all redox-sensitive metals is similar. Burn down commonly redistributes elements at different depths because of their different chemical behaviors [Colodner et al., 1992; Thomson et al., 1993, 1995; Crusius and Thomson, 2000]. Therefore metal enrichment in zone 1 must reflect the development of anoxic conditions within the sediment shortly after deposition. Such conditions may have developed as a result of (1) increased sedimentation rate and corresponding decreased oxygen influx, (2) decreased ventilation of the bottom water, and/or (3) increased carbon flux to the sediment. Increased sedimentation is ruled out as the principle cause of trace metal enrichment because the sedimentation rate was substantially higher during deposition of the trace metal-poor sediments in the Bølling (average 114 versus 169 cm/kyr; Figure 2). Whether oxygen depletion in near-surface sediments was the result of decreased ventilation of bottom waters and/or increased organic carbon flux to the sediment, which also would have caused water column oxygen content to decline, is discussed in sections 5.2 and 5.3. In either case, these trace metal data imply that oxygen depletion of the OMZ was more pronounced during the Allerød.
 The second interval of trace metal enrichment begins at ∼11.0 kyr (i.e., the end of the Younger Dryas) and continues into the Holocene (zone 2, Figure 3). The increase in Re at ∼4 kyr (i.e., 20 cm below the sediment-water interface) reflects the approximate depth where Re reduction and accumulation are occurring at present [McKay, 2003]. The relatively low Holocene sedimentation rate (∼5 cm/kyr) coupled with sufficient oxidant demand has allowed authigenic metal enrichment to occur well below the sediment-water interface, possibly overprinting the paleosignal. Although this observation constrains detailed interpretation of OMZ history during the early Holocene, these data clearly indicated that near-surface sediments at Site JT96-09 have been continuously reducing since the Younger Dryas, and imply relatively low bottom water oxygen concentrations during the Holocene. This conclusion is consistent with deductions made off California [Anderson et al., 1987; Behl and Kennett, 1996; Cannariato and Kennett, 1999; Ivanochko and Pedersen, 2004] and Mexico [Ganeshram et al., 1995], all of which indicate that a relatively intense OMZ has been a permanent fixture during the Holocene.
 Molybdenum enrichment in the Allerød is contemporaneous with a dramatic increase in the numbers of Bolivina species (predominantly B. argentea, B. pacifica and B. subadvena) and Bulliminella tenuata (Figure 4b). A similar, although more subtle, increase occurs between ∼11.0 and 10.0 kyr (Figure 4b), which is also a period of enhanced Mo accumulation. Bottom water oxygen concentration has a direct influence on the species of benthic foraminifera that occur in sediments [Kaiho, 1994]. Bolivina-dominated assemblages are typical of the most intense portions of the OMZ in the eastern Pacific and thrive at dissolved oxygen levels of <0.3 mL/L [Ingle and Keller, 1980; Mullins et al., 1985; Sen Gupta and Machain-Castillo, 1993]. In contrast, Uvigerina species prefer a more oxygenated environment (0.3 to 1.5 mL/L oxygen [Cannariato and Kennett, 1999]). This relationship has been documented at many locations along the northeastern Pacific margin [e.g., Mullins et al., 1985; Quinterno and Gardner, 1987] and it appears to hold true in the past [Behl and Kennett, 1996; Cannariato et al., 1999; Cannariato and Kennett, 1999]. The presence of Uvigerina species also may be related to a high organic carbon supply to the sediment [Quinterno and Gardner, 1987], but this association is not observed if bottom water oxygen is too low [Kaiho, 1994]. We conclude that the occurrence of a benthic foraminiferal assemblage composed almost exclusively of Bolivina species and Buliminella tenuata, in sediments characterized by relatively high Mo concentrations, is further evidence that the bottom water oxygen concentration was lower than at present during the Allerød (13.5 to 12.6 kyr) and between ∼11.0 and 10.0 kyr. There is however no evidence that bottom water oxygen concentration was low during the Bølling. In fact, it could be argued that bottom water oxygen concentration was relatively high given that the most abundant species of benthic foraminifera in Bølling sediments is Epistominella which prefer an oxic environment (i.e., 0.3 to 1.5 mL/L oxygen [Cannariato and Kennett, 1999]).
 OMZ intensification off Vancouver Island brackets the Younger Dryas, and in this respect the timing is similar to that observed throughout the North Pacific, with one notable difference. Intensification of the OMZ off Vancouver Island began in the Allerød while at many other locations it began much earlier [Cannariato and Kennett, 1999; Zheng et al., 2000; Crusius et al., 2004; Ivanochko and Pedersen, 2004]. On the California margin, for example, OMZ intensification based on high Mo concentrations, commenced at ∼15 kyr while off Vancouver Island it was delayed until ∼13.5 kyr, a lag of 1500 years (Figure 6). There is no prerequisite that the timing of OMZ intensification be the same throughout the North Pacific. In fact, because of regional differences in surface water productivity and proximity to sources of NPIW and SSW variability is to be expected. Even cores that are located relatively close to one and other can exhibit nonsynchronous behavior [van Geen et al., 2003]. It remains to be determined whether the delay observed off Vancouver Island is regional, reflecting some large-scale oceanographic difference, or local.
Figure 6. Down-core profiles of Mo concentration: (a) Core JT96-09 from the Vancouver Island Margin (48°54.76′N, 126°53.44′W), (b) Core 1019A from off northern California (41°40.963′N, 124°55.979′W), and (c) Core 893A from the Santa Barbara Basin, California (34°17.25′N, 120°02.19′W). Molybdenum data for the California cores are from Ivanochko and Pedersen . The age model for Core 1019A is from Mix et al. , and the age model for Core 893A is from Ingram and Kennett . The locations of calibrated 14C dates are indicated by the stars.
Download figure to PowerPoint
5.2. Ventilation Changes?
 At present, ventilation of the OMZ along the eastern margin of the North Pacific reflects a balance between the input of relatively oxygen-rich North Pacific Intermediate Water (NPIW) from the northwestern Pacific and oxygen-poor Subtropical Subsurface Water (SSW) from the eastern tropical Pacific. Any change in the supply and/or oxygen concentration of either the NPIW or SSW would directly impact the OMZ.
 Oxygen-depleted SSW is transported northward along the west coast of North America as far north as Vancouver Island by the California Undercurrent [Reed and Halpern, 1976; Mackas et al., 1987]. Mixing with adjacent water masses modifies the physical and chemical properties of SSW as it moves northward (e.g., decreasing temperature [Halpern et al., 1978] and decreasing δ15N values [Kienast et al., 2002]). Off Vancouver Island the highest percentage of SSW is observed at a depth of 100 to 300 m where the California Undercurrent is strongest [Reed and Halpern, 1976]. Weaker flow below 300 m allows increased mixing of SSW with other water masses; however, SSW is still recognizable as deep as 1300 m [Reed and Halpern, 1976].
 Oxygen-rich NPIW, which forms just east of the Sea of Okhotsk, occurs throughout the North Pacific Subtropical Gyre and extends as far north as the Gulf of Alaska. A number of studies have suggested that enhanced formation of NPIW can explain the weakening of the OMZ during cold climatic periods [Duplessy et al., 1988; Keigwin and Jones, 1990; Kennett and Ingram, 1995; van Geen et al., 1996; Behl and Kennett, 1996; Keigwin, 1998; Zheng et al., 2000]. Better ventilation of NPIW during the last glacial has been inferred from relatively high δ13C values of benthic foraminifera which suggest the presence of a younger, relatively nutrient-poor intermediate water mass in the northwest Pacific [Duplessy et al., 1988; Keigwin, 1998]. Radiocarbon data for coeval benthic and planktonic foraminifera (i.e., benthic-planktonic age differences) for cores from the northwestern Pacific also suggest increased ventilation between ∼17 and 13 kyr and during the Younger Dryas, as well as decreased ventilation during the Bølling-Allerød [Duplessy et al., 1989; Ahagon et al., 2003]. Radiocarbon data from the Santa Barbara Basin (Core 893A, 588 m), on the eastern side of the Pacific, also indicate increased ventilation during the Last Glacial Maximum and Younger Dryas [Ingram and Kennett, 1995; Kennett and Ingram, 1995]. However, radiocarbon data from cores collected on the open California margin are more ambiguous. Core F2-92-P3, taken from 800 m (35°N), yields evidence of decreased ventilation between 11 and 9 kyr [van Geen et al., 1996]. In contrast, 14C measurements made on ODP Core 1019 collected from the continental slope (980 m water depth) off northern California (41°N) suggest increased, not decreased, ventilation during the early Holocene and Bølling-Allerød at the same time as OMZ intensification occurred [Mix et al., 1999].
 The age of intermediate water off Vancouver Island is recorded by benthic foraminifera in Core JT96-09 while the age of surface water is recorded by planktonic foraminifera. The difference in the ages of benthic and planktonic foraminifera obtained from the same sample therefore establishes the age difference between the intermediate and surface waters. If intensification of the OMZ inferred from trace metal and benthic foraminifera data was the result of a substantial decrease in ventilation, the benthic-planktonic age difference should be greater, reflecting the reduced influence of relatively young NPIW and the increased influence of older, oxygen-depleted SSW.
 Direct determination of the modern benthic-planktonic (B-P) age difference is not possible because of the lack of foraminifera in surface sediments. However, using Δ14C water column data of Östlund and Stuiver  for the North Pacific off the coast of California we estimate that benthic foraminifera from a depth of 900 m off Vancouver Island should be ∼1780 years old or perhaps slightly younger given that our study area lies closer to regions of NPIW ventilation. The age of planktonic foraminifera presently growing within surface waters (i.e., reservoir age of surface water) is ∼800 years [Robinson and Thompson, 1981; Southon et al., 1990; Southon and Fedje, 2003; Hutchinson et al., 2004]. Thus the modern B-P age difference should be ∼1000 years.
 The paucity of foraminifera in Core JT96-09 places some limitations on our use of B-P age differences to infer changes in ventilation. Most notably, we can say nothing about changes in ventilation during the Younger Dryas and Holocene. However, we do have data for those times when intensification of the OMZ occurred. With the exception of one data point (sample 2), most of the B-P age differences are similar to, or slightly less than, the estimated modern value of ∼1000 years (Figure 5). There is no evidence of decreased ventilation (i.e., larger B-P age differences) during the first period of OMZ intensification (i.e., Allerød; Figure 5). In fact, the B-P age difference decreases slightly from 860 to 610 years throughout this interval. This decrease implies slightly better ventilation of the OMZ, which is inconsistent with trace metal and foraminiferal data that indicate the exact opposite. The explanation for this apparent contradiction is discussed below. The only substantial decrease in ventilation appears to occur just after the Younger Dryas and is coincident with the second period of OMZ intensification. This interpretation is based on a single data point however and there is evidence that the planktonic age may be anomalously young (Figure 2), leading to the larger B-P age difference.
 When using B-P age differences to infer changes in ventilation we assume that the reservoir age of the surface water, and thus the planktonic age, is constant. However, this is not always true. Fluctuations in atmospheric 14C concentration and atmosphere-ocean exchange, as well as changes in oceanic circulation (e.g., upwelling) can all influence the reservoir age of surface waters. The modern reservoir age in the northeast Pacific is ∼800 years and has not changed significantly since the Younger Dryas [Southon et al., 1990; Southon and Fedje, 2003]. There is however evidence of larger reservoir ages (900 to 1300 years) during the late glacial and early deglacial [Kovanen and Easterbrook, 2002; Hutchinson et al., 2004]. These authors suggest that melting of Cordilleran ice might have supplied relatively old (i.e., 14C depleted) CO2 to surface waters, thereby increasing the reservoir age. If this were the case, B-P age differences off the west coast of Vancouver Island should have decreased when rapid retreat of the Cordilleran ice sheet commenced between 15,000 and 14,000 14C years [Clague and James, 2002]. Since most of the meltwater influx occurred prior to the Allerød it was probably not responsible for the smaller B-P age differences during the Allerød. It is more likely that increased upwelling of 14C-depleted subsurface waters caused the planktonic ages to increase, thus decreasing the B-P age difference. Upwelling was greatly reduced along the northern and central portions of the California Current system during the Last Glacial Maximum, and was reestablished by the Allerød (∼13.0 calendar kyr [Sabin and Pisias, 1996]). The lower B-P age difference observed for the Allerød in Core JT96-09 could reflect the return of upwelling conditions off Vancouver Island. If this hypothesis is correct primary production should have increased at the same time. We present evidence in support of this scenario in section 5.3.
5.3. Changes in Productivity?
 Modern primary production off the west coast of Vancouver Island is influenced by large-scale atmospheric circulation. In late spring to early fall the North Pacific High drives northerly winds, offshore Ekman transport and upwelling of nutrient-rich water [Huyer, 1983]. The strength of these winds and thus upwelling intensity is affected by the strength of the pressure gradient between the North Pacific High and the continental thermal low, such that the larger the gradient the more intense the upwelling [Bakun, 1990]. In winter when the North Pacific High shifts southward from ∼38°N to ∼28°N and is replaced by the Aleutian Low, winds switch direction and upwelling ceases north of ∼40°N [Huyer, 1983].
 Climate modeling and paleoevidence suggest that during the Last Glacial Maximum the North Pacific High was positioned further south in summer [COHMAP Members, 1988; Thunell and Mortyn, 1995; Mortyn et al., 1996; Sabin and Pisias, 1996; Doose et al., 1997], a situation analogous to modern winters. As a result, upwelling, and thus primary production, along the central and northern portions of the California Current system were greatly diminished during the Last Glacial Maximum [Dymond et al., 1992; Lyle et al., 1992; Sancetta et al., 1992; Ortiz et al., 1997; Dean and Gardner, 1998; Mix et al., 1999]. Off Vancouver Island the accumulation rate of marine organic matter was also relatively low during the late glacial (Figure 7) [McKay et al., 2004]. However, during the Bølling-Allerød (14.3 to 12.6 kyr) marine organic carbon accumulation increased substantially (Figure 7). The most dramatic increase (i.e., an apparent sixfold increase relative to late the glacial) occurred in the Allerød, coincident with the first period of OMZ intensification. A small increase in the marine organic carbon accumulation rate is also evident during the second period of OMZ intensification (11.0 to 10.0 kyr; Figure 7). On the basis of the marine organic carbon record and the B-P age differences we conclude that OMZ intensification during the Allerød was the result of increased primary productivity, rather than decreased ventilation. Enhanced productivity was most likely caused by the onset of upwelling off Vancouver Island as atmospheric circulation switched from a glacial mode (i.e., influenced by the Aleutian Low throughout the year) to an interglacial mode (i.e., influenced by the North Pacific High from late spring to early fall). Upwelling of relatively old waters also may explain why B-P age differences decrease slightly during the Allerød.
 Despite apparently high organic carbon mass accumulation rates during the Bølling there is no evidence of OMZ intensification at this time. It is possible that the high organic carbon accumulation rates are an artifact of how mass accumulation rates are calculated. During the Bølling the sedimentation rate was exceptionally high (169 cm/kyr) leading to high organic carbon mass accumulation rates even though the organic carbon concentration did not increase. In comparison, during the Allerød both the concentration and accumulation of marine organic matter increased (Figure 7). It is also possible that the OMZ was slightly better ventilated during the Bølling and that this counterbalanced the increase in organic carbon flux to the sediment. A small change in the ventilation would not be detected when using B-P age differences because of their large associated errors.
 It could be argued that rather than causing intensification of the OMZ, increased organic carbon burial during the Allerød was the result of better preservation of recalcitrant organic matter because of lower oxygen concentrations in the bottom water [e.g., Dean et al., 1994; Zheng et al., 2000]. In the geological record, laminated sediments commonly have high organic carbon contents and this led to the hypothesis that organic matter preservation is enhanced in anoxic environments [Emerson, 1985]. The presumption being that anaerobic bacteria are less efficient at degrading complex organic molecules. However, the sediments in Core JT96-09 are bioturbated, even during periods of inferred OMZ intensification, and thus bottom waters never became anoxic. Therefore enhanced preservation of organic matter resulting from anoxic bottom waters cannot explain increased organic carbon burial during the Allerød. More recently, it has been suggested that high sedimentation rate and low oxygen concentration work together to enhance organic matter preservation by controlling the length of time that refractory organic compounds are exposed to oxygen [Hedges and Keil, 1995; Gélinas et al., 2001]. We have estimated oxygen exposure times (OETs) for the Holocene, Allerød, Bølling and the Late Glacial at Site JT96-09 (Table 2). If the oxygen penetration depth remained constant, then OETs for the Allerød and Bølling are similar (i.e., 1.7 and 1.3 years, respectively). If the oxygen penetration depth decreased in the Allerød, as the trace metal data imply, but remained the same for the Bølling, computed OETs remain similar (i.e., <1 and 1.3 years). To establish a substantial difference in OETs would required a deeper oxygen penetration depth during the Bølling, which is unlikely given the higher sedimentation rate at the time. There is also no geochemical evidence to suggest that oxygen penetrated more deeply during the Bølling. Sedimentary Mn/Al ratios, for example, are relatively low during both periods (Figure 3a) implying that near-surface sediments were continuously suboxic throughout the Bølling-Allerød. We cannot rule out the possibility that a combination of high sedimentation rate and low bottom water oxygen concentration played a role in enhancing organic matter accumulation during the Allerød. However, the large increase in organic matter accumulation during the Allerød relative to the Bølling, given that OETs were probably similar, suggests that high export productivity was the dominant factor leading to high organic carbon accumulation. This conclusion is supported by data for other paleoproductivity proxies (e.g., biogenic barium and opal) [McKay et al., 2004].
Table 2. Oxygen Exposure Times for Sediments in Core JT96-09
|Time Period||Sedimentation Rate, cm/kyr||Oxygen Penetration Depth,a cm||OET,a,b years|
|Allerød||116||0.2 (0.1)||1.7 (<1)|
|Bølling||150||0.2 (1.0)||1.3 (7)|
|Late glacial||48||0.2 (1.0)||4.2 (20)|