HISTORY OF SEDIMENTATION UNDER THE DEEP WESTERN BOUNDARY CURRENT

The history of the DWBC since the early Miocene is contained in abyssal sediment drifts located east of New Zealand (Carter, L., and McCave, 1994). The four key drift drill sites are Site 1121 (winnowed veneer accompanied by long erosion), Site 1122 (contourite followed by fan deposition), Site 1123 (continuous 20.5- to 0-Ma record of drift deposition), and Site 1124 (drift deposits since 27 Ma punctuated by erosional paraconformities). Fluctuations in DWBC flow are reflected in variations of the terrigenous grain size (Hall et al., 2001; cf. McCave et al., 1995), variations in linear sedimentation or mass accumulation rates (Handwerger and Jarrard, in press), and variations in magnetic fabric (Joseph et al., 1998). The mean grain diameter for sortable silt grain sizes is expected to rise with increasing flow until, at speeds of more than ~0.15 m/s, erosion outweighs deposition. In general, the magnitude and frequency of individual erosion and deposition events cannot be resolved in the record. In the late Quaternary age control points are several thousand years apart, whereas in the Oligocene and Miocene they are several million years apart, so sedimentation rates and hiatuses can only be tightly resolved for intervals where an orbitally tuned timescale has been achieved.

Graham et al. (in press) showed from ¹⁰Be dating that the skin drift at this site has an age of ~18 Ma (early Miocene) at ~7-8 m depth (Fig. F7). The low average sedimentation rates this implies (<0.5 m/m.y.) indicate that throughout the Neogene the area behaved more as a sediment source than a sink. At Site 1123, over the same period ~535 m of sediment accumulated. Particularly low sedimentation rates occurred between 15 and 12, 10 and 8, and 1.5 and 0 Ma. As seen in other records (e.g., Flower and Kennett, 1995; Zachos et al., 1992), the period around 15-13.50 Ma is associated with growth of the East Antarctic Ice Sheet. Zhou and Kyte (1992) described the geochemistry of abyssal clays from DSDP Leg 91 Site 596 on the central Pacific plain and inferred (1) a period of development of Mn crusts and vigorous bottom currents between ~14.0 and 8.4 Ma and (2) an erosional interval at ~8.4 Ma, which is also the time of onset of a strong cooling trend identified from isotope data by Shackleton and Kennett (1975; cf. Zachos et al., 2001). Lastly, 1.5-0 Ma spans the development of the intense cooling of the Quaternary glacial period.

It is a significant result that the same timing applies to inferred periods of bottom erosion—and therefore to increased cold-water flow and enhanced Fe-Mn deposition—in both the southwest and west Central Pacific Ocean. If the Site 1121 record is representative of the erosional remnants of the whole Campbell sediment apron, then the apron represents a very large sediment source with an area >5 x 10⁴ km² (Carter, L., and McCave, 1997) that supplied material to regions farther north for >20 m.y. through the Neogene. The record of rapidly accumulated radiolarian-rich Cretaceous-Paleocene sedimentation at Site 1121 predates 55 Ma, and former overlying sediments up to early Oligocene in age are inferred to have been removed by DWBC/ACC erosion after the ~33.7-Ma opening of the Tasmanian Gateway to deep flow (cf. Carter, R., McCave, Richter, Carter, L., et al., 1999).

The Bounty Fan developed across the path of the DWBC at the mouth of the Bounty Trough. Site 1122 was cored into the left bank levee succession of the fan, which was deposited under the influence of east-traveling, north-overspilling turbidity currents from the Bounty Channel and the north-traveling DWBC (Carter, R., and Carter, L., 1996).

Using the updated age model of Wilson (2000a), a ~5-m.y. hiatus separates early Pliocene from middle Miocene drift sediments at 485 mbsf. At ~2.2 Ma (~440 mbsf), injection of sediment from the Bounty Trough, and perhaps also from erosion near Site 1121 farther south, resulted in a change in sedimentation style beneath the DWBC, increased sediment accumulation rates (to ~10 cm/k.y.), and inception of the abyssal Bounty Fan. Despite this increase in sedimentation rate, discrete turbidites do not appear in the Site 1122 record until ~1.7 Ma (~380 mbsf). The first turbidites are interbedded with cross-laminated sediments that record a continuing influence of DWBC/ACC flow. The increase in sediment injection from the Bounty Trough at ~2.2 Ma occurs also, though a little earlier at ~2.5 Ma, at DSDP Site 594 (Nelson et al., 1985). These changes may result from late Pliocene climatic deterioration or, alternatively, may be related to increased uplift along the Southern Alps plate boundary in the west. At ~0.7 Ma, turbidite frequency and thickness increases, as turbidity currents from the Bounty Trough cause average sedimentation rates to increase to >40 cm/k.y., probably in response to further uplift along the Southern Alps.

Handwerger and Jarrard (in press) used downhole logging data for the record younger than 25 Ma at Sites 1123 and 1124 to estimate the relative proportions of carbonate and terrigenous sediment using principal component analysis. The logging data analyzed were bulk density, P-wave velocity, total gamma ray, spectral gamma ray (K, U, and Th), resistivity, photoelectric factor, and depth. The Site 1123 records show relatively high carbonate percentages (~70 wt%) and MARs (~2.5 g/cm²/k.y.) during the periods 0.6-13.5 and 19-20.5 Ma and lower carbonate percentages (~40 wt%) and MARs (1.5 g/cm²/k.y.) during the intervening period (14.5-17.5 Ma) (Fig. F20). Milankovitch-scale 41-k.y. cyclicity occurs superimposed on these background averages, within which particularly strong depletions in carbonate content occur at ~17.5, 16, 15, 11.5, 10, 4.5, and 1.2 Ma. Assuming that carbonate accumulation at this drift site is controlled primarily by DWBC transport vigor, this suggests the occurrence of generally slower flow late in the early Miocene between 14.5 and 17.5 Ma.

Three other postcruise studies of the Site 1123 core used grain size records as a proxy for changing flow speeds under the DWBC (cf. McCave et al., 1995), one for the middle Miocene (Hall et al., 2003), one for the last 3 m.y. (Hall et al., 2002), and one focusing at high resolution on the last 1.2 m.y. of the Quaternary (Hall et al., 2001) (Fig. F21).

Between ~15.5 and 12.5 Ma (middle Miocene), fluctuating sortable silt mean grain sizes indicate current variations at the period of 41-k.y. orbital obliquity forcing, with faster flow speeds during intervals of colder climate (Hall et al., 2003) (Fig. F21A). In addition to the Milankovitch-scale forcing, longer-term changes of current speed occur, with increasing DWBC strength between ~14.8 and 14.3, ~14.3 and 13.8, and 13.8 and 13.15 Ma, and a decreasing trend between 13.2 and 12.5 Ma. The trends of increased DWBC speed that culminate at ~14.3 and ~13.8 Ma correspond to the known middle Miocene cooling phases Mi3b (= CM4; Woodruff and Savin, 1989) and Mi3a (Flower and Kennett, 1995), respectively.

Orbital-scale flow changes in the DWBC also occurred in the Pliocene-Pleistocene (Hall et al., 2001, 2002). Sortable silt grain sizes clearly indicate that faster flow occurred in glacial compared to interglacial periods. Spectral analysis of the sortable silt record shows significant spectral peaks, coherent with both planktonic and benthic oxygen isotope records, at each orbital frequency. There is no phase lag between sortable silt and benthic ¹⁸O at the 100-k.y. period. Although the DWBC has a very large flux, data from current meters, nepheloid layers, and bottom photographs show that it is presently sluggish around the New Zealand margin. In contrast, very large current scours occur at the seabed around volcanic pinnacles (Carter, L., and McCave, 1994; McCave and Carter, L., 1997). The Leg 181 results suggest that these scours most probably formed under stronger bottom flows during glaciations, and farther downstream these enhanced flows may also have driven the glacial increase in sediment focusing that has been recorded in the central equatorial Pacific over the past 300 k.y. (Marcantonio et al., 2001).

Collectively, these variations indicate that a strong coupling exists between changes in the speed of the deep Pacific inflow and high-latitude climatic forcing (Fig. F21B). We conclude that this coupling has probably been a persistent feature of the global thermohaline circulation system for at least the past 15 m.y. Furthermore, longer-term changes in the mean flow speed suggest that intensification of the DWBC occurred in parallel with increases in the production rate of the intermediate-depth SCW, as indicated by isotopic data at DSDP sites north of New Zealand (Flower and Kennett, 1995). Site 1123 results, therefore, provide evidence that the middle Miocene growth of ice on Antarctica caused a significant increase in the production of deep cold water, thus in turn increasing the vigor of the DWBC and perhaps global thermohaline circulation and marking an important step in the development of Neogene icehouse climate. A particularly elegant detail of the Site 1123 record is the increase in grain size recorded at 14.8-14.6 Ma, followed by a short hiatus at 14.6-14.3 Ma, which is precisely the pattern that would be expected from increasing current flow that eventually exceeded the erosion threshold of seafloor sediment. The hiatus is then followed by sedimentation of finer grain sizes, indicating the resumption of accumulation once the flow had slowed.

Using the Pleistocene oxygen isotope stratigraphy as a stratigraphic control, Hall et al. (2001) also recorded three long-term periods of differing mean flow speeds of the DWBC at Site 1123 (Fig. F21C). First, a period of moderately high flow occurred prior to the termination of glacial MIS 22 (which is the first large 100-k.y. cycle in the record); second, a transitional period of lower-speed flow occurred during 0.87-0.45 Ma; and third, another phase of higher-speed mean flow occurred from 0.45 Ma to the present. The mean current speeds for these inferred flow periods are significantly different (P < 0.001). The middle flow phase corresponds to the MPT, marking the change in response of the Earth's climate from orbital obliquity (41 k.y.) to eccentricity (100 k.y.) forcing; at Site 1123, this change extended over several hundred thousand years rather than comprising a sharp event. This gradual change in a proxy record indicative of the strength of the DWBC demonstrates that a changing global thermohaline circulation accompanied the MPT. Over the most recent period, the strength of DWBC flow in the interglacials has clearly increased as part of a long-term trend that has typified the last 0.9 m.y.

The benthic ¹³C record at Site 1123 is in phase with ¹⁸O at major orbital frequencies, with light ¹³C associated with heavy ¹⁸O. Nutrient-enriched carbon values of -0.6 are present in glacial sediment and depleted values of +0.8 in interglacial sediment. Deepwater ¹³C spatial gradients can be used to shed light on water mass aging trends, and Site 1123 is ideally located for the assessment of Pacific ¹³C gradients because it represents the entrance of CDW. As CDW travels across the Pacific, it mixes with the overlying waters and progressively accumulates nutrients through the remineralization of sinking organic material—thus depleting the ¹³C signal. The gradient between Site 1123 and eastern equatorial Leg 138 Site 849, ¹³C_(1123-849), varies by >0.5 and displays a larger gradient during interglacial periods. The spectrum of ¹³C_(1123-849) shows strong obliquity power and is coherent with ¹⁸O, with the records in phase. A clear relationship is also seen between ¹³C_(1123-849) and the sortable silt index, with periods of reduced ventilation (high ¹³C_(1123-849)) associated with reduced DWBC flow speeds. An important related feature is a very large glacial-interglacial variation of ~1.4 in the ¹³C of the water emanating from the ACC and flowing into the Pacific, part of which is attributable to changes in the global carbon reservoir.

Overall, the carbon isotope and sortable silt data demonstrate strong glacial-interglacial variability in the strength of flow and water mass properties of the southwest Pacific DWBC sector of the global thermohaline system. The data are consistent with greater DWBC flow, and therefore enhanced formation of AABW and greater Pacific ventilation, as a persistent feature of glacial periods over the past 1.2 m.y. This Pleistocene record provides a solid basis for studies going back further in time if reliable isotopic data can be assembled.

Using the same techniques as for Site 1123, Handwerger and Jarrard (in press) also analyzed sediment accumulation rate proxies at Site 1124. The Site 1124 record (Fig. F20) displays a long-term decline in average carbonate percentage from ~85 to 15 wt% (25-10 Ma), a late Miocene period of low (10-20 wt%) carbonate content (10-5 Ma), a rise from ~15 to 55 wt% (5-3.2 Ma), and a final period during which carbonate values oscillate repeatedly and rapidly between ~10 and 55 wt% (3.2-0 Ma). The long-term decline in carbonate during the Miocene continues across the core gap at 22.5-17.6 Ma and the hiatuses at 16.5-15 and 14-11 Ma. High carbonate peaks occur superimposed on the long-term trends at ~20, 10.5, and 3.2 Ma, and notable lows occur at ~23.6, 11, 2.4, and 1.2 Ma. Mass accumulation rates of both carbonate and terrigenous material show a sharp reduction at ~23.7 Ma, and terrigenous MAR shows highs at 14.8-14.2 and 10.9-9.7 Ma and a sharp increase at ~1.1 Ma.

The Site 1124 core was also studied by Joseph et al. (in press), who used grain size and magnetic fabric measurements as proxies for DWBC intensity. These data (Fig. F22) clearly show sharp increases in terrigenous:carbonate ratio starting at ~23.6 and 10.4 Ma, a longer-term but markedly fluctuating decrease in the same ratio between ~5 and 1.5 Ma, and finally another significant increase in the ratio over the last 1.5 m.y. Higher-frequency fluctuations are also apparent throughout the Joseph et al. data set and represent an orbital signal, though the sampling interval was not quite close enough to capture the full 41-k.y. Milankovitch cyclicity.

The changes in sediment parameters at Site 1124 documented by Jarrard and Handwerger (in press) and Joseph et al. (in press) are consistent with our other conclusions regarding the ~23.7-Ma settling of the DWBC along pathways at speeds similar to those of today as the strong predecessor current through the Tasmanian Gateway waned with the the widening of the Southern Ocean and perhaps the opening of Drake Passage (Fulthorpe et al., 1996). The grain size and fabric measurements seem to be particularly sensitive to fluctuations of the DWBC between 23 and 9 Ma, in accord with major Antarctic episodes of glacial development and associated circulation changes (e.g., Miller et al., 1991). The hiatus between 22.5 and 17.6 Ma represents an increase in ACC circulation around Antarctica (perhaps caused by the opening of deep circum-Antarctic gateways) (cf. Rack, 1993; Handwerger and Jarrard, in press), whereas the hiatuses at 16.5-15 and 14-11 Ma encompass the late middle Miocene cooling events described by several earlier writers (Woodruff and Savin, 1991; Wright and Miller, 1992; Flower and Kennett, 1993) and represent glacial deepenings in Antarctica. The time period between 11 and 9 Ma is characterized by the signature of formation and stabilization of the West Antarctic Ice Sheet. A decrease in carbonate and terrigenous MAR at this time is most marked at the deeper Site 1124, an effect that may indicate the presence of a stronger and more corrosive DWBC there. Again at Site 1124, a Pleistocene surge in accumulation rate occurs after 1.5 Ma, which probably marks the diversion of the Hikurangi Channel onto the abyssal Pacific floor close to Rekohu Drift (cf. Lewis et al., 1998; Hall et al., 2002).

Ice-Rafted Debris on the Campbell Plateau

Records of ice-rafted debris (IRD) in seven site survey cores and at Site 1120 reveal a pattern of periodic iceberg incursion into the southwest Pacific over the last 200 k.y. (Carter, L., et al., 2002b). Modern icebergs originate in Antarctica and are moved east and north to the margins of the Campbell Plateau by the ACC. Once they are in the southwest Pacific, local currents and winds disperse icebergs as far north as Chatham Rise at 43°S (Cullen, 1965; Brodie and Dawson, 1971). The shallowness of the rise crest, the rise-parallel circulation, and the strong thermal gradient associated with the nearby STF inhibit further northward transport. Although IRD concentrations are very low in the studied cores (926 grains/g), a ¹⁸O chronology reveals distinct IRD peaks (1) at the transition from MIS 7 to 6, (2) during late MIS 5, and (3) during the LGM (MIS 2). Smaller peaks occur in MIS 4 and 3. A similar pattern is also known off Antarctica, suggesting that the periodic destabilization of ice shelves was the main driving force behind Campbell Plateau IRD events. Differences between the Antarctic and New Zealand IRD records probably reflect paleoceanographic influences on iceberg dispersal. For instance, the MIS 5 event is more strongly represented near Antarctica, as would be expected. Conversely, the MIS 2 event is relatively better shown on Campbell Plateau, consistent with a more vigorous oceanic circulation then causing more icebergs to reach the distant plateau. Comparison of IRD records from the Campbell Plateau with those from the southeast Atlantic and southwest Indian Oceans suggests that IRD events can be correlated across the Southern Ocean over at least the last 70 k.y.

Changes in the Position of the STF near Site 1119

Site 1119 lies just seaward of the modern STF, east of South Island. The upper 86.19 mcd of Site 1119 was deposited in the last 0.252 m.y., during MIS 1-8 (Fig. F7). The underlying sediments to 100 mcd, beneath a ~25-k.y.-long unconformity, represent MIS 8.5-11 (0.278-0.370 Ma). Interglacial MIS 5, 7, and 9 are represented by silty clay, which encompasses small groups of 5- to 65-cm-thick, sharp-based, Chondrites-burrowed, olive-gray, graded fine sands-muds. The sands are shelly (especially Tawera) and conspicuously rich in foraminifers, sometimes including temperate-water forms (Orbulina universa, Globorotalia inflata, and Globorotalia truncatulinoides). The intervening micaceous glacial muds may be bedded on a centimeter scale but are more usually massive and bioturbated; they contain the cold-water scallop Zygochlamys delicatula and an enhanced siliceous and impoverished calcareous microfauna.

During interglacials, the water at Site 1119 was deeper, the shoreline was distant from the site, and the broad shelf was ventilated by the subtropical Southland Current. At these times, transgressive shelly sand followed by highstand mud accumulated at low rates of 5-32 cm/k.y. along the upper slope, which was bathed in colder SAW. As climatic cooling progressed, the falling sea level was accompanied by a narrowing of the shelf; sedimentation rates rose successively through the MIS 10, 6, and 2 glaciations to 45, 69, and 140 cm/k.y., respectively, as a result of the delivery of river-borne sediment directly to lowstand shorelines.

Against this lithologic background, the upper 100 m of the Site 1119 core records the seaward movement of the STF during glacial periods, accompanied by the incursion of warmer STW above the site and landward movement during interglacials, resulting in a dominant influence then of colder SAW. Counterintuitively and forced by the bathymetric control of a laterally moving shoreline during glacial-interglacial and interglacial-glacial transitions, the Site 1119 core records a southerly (seaward) movement of the STF during glacial periods accompanied by the incursion of subtropical water (STW) above the site and northerly (landward) movement during interglacials, resulting in a dominant influence then of subantarctic surface water (SAW). These different water masses are clearly delineated by their characteristic ¹³C values (Carter, R., et al., in press). Intervals of thin, sharp-based, graded sands-muds occur within the cold periods MIS 2-3, 6.2, and 7.4. During these glaciations, an increased flow of ACC cold water circulated clockwise in the head of the Bounty Trough (Carter, L., et al., 2000; Neil et al., submitted [N1]) and the glacial STF east of South Island was marked by a zone of intense oceanographic gradients (Weaver et al., 1998). The currents that developed along the glacial STF transported sand beds with grains up to 150 µm in diameter and are therefore inferred to have reached speeds of at least 40 cm/s in waters to ~250 m deep. The beds of very fine sand that occupy the cold-climate intervals at Site 1119 are marked also by conspicuous gamma ray lows, color reflectance (carbonate content) highs, and stable isotope signatures. In common with other Southern Hemisphere records, the cold period that represents the LGM commenced at ~22.4 ka at Site 1119, at which time the STF and SAF may have coalesced into a single zone of enhanced oceanographic gradients around the head of the adjacent Bounty Trough.

The deeper parts of Site 1119 comprise a succession of silty clay punctuated by thin sand intervals, similar to the lithologies described above for the uppermost 100 m. Shipboard observations indicate that the 495-m base of the core terminates in the late Pliocene at ~3.9 Ma (Carter, R., McCave, Richter, Carter, L., et al., 1999), and research is continuing on the interpretation of the earlier parts of this important climatic record.

Glacial-Interglacial Changes in Salt Sources for Deep Water

Pore water samples from the upper 100 m of Site 1123 were measured for chloride concentration and oxygen isotopic composition by Adkins et al. (2002), in order to reconstruct the salinity and temperature of the CDW during the LGM. By comparing the results from Site 1123 with similar measurements made at sites elsewhere in the Southern and Atlantic Oceans, these authors showed that LGM deep ocean temperatures were homogeneous everywhere and within error of the freezing point of seawater at the sea surface. In contrast, glacial salinity values varied, with the saltiest deep water present in the Southern Ocean rather than the North Atlantic (NADW). These results have been termed "a landmark change in our understanding of deep-ocean circulation" (Boyle, 2002) because they imply that during Termination I (~20-10 ka) the salt source for the deep sea switched from the Antarctic to the North Atlantic, with concomitant circulation changes.