Carbon Cycling in the Glacial Ocean: Constraints on the Ocean’s Role in Global Change

A summary of our knowledge about the recent carbon cycle is presented from the perspective of atmospheric observations. Particular emphasis is given to carbon isotopes as a tool to determine global carbon sources and sinks as well as their recent anthropogenic changes. In the case of atmospheric carbon dioxide, the yearly anthropogenic emissions are small if compared to the gross exchange between the most important reservoirs: biosphere and ocean. In order to derive net fluxes caused by climatic or anthropogenic perturbations, the gross fluxes have to be known as accurately as possible. One isotopic tool is to trace the penetration of bomb ¹⁴C into the ocean surface water so as to derive the atmosphere/ocean gas-exchange rate. Moreover, the distribution of δ¹³C in atmospheric CO₂ can help to constrain the amount of excess CO₂ taken up (or released) by the global biosphere. As for atmospheric methane, anthropogenic emissions today exceed natural release rates by nearly a factor of two. Although the total yearly methane emissions to the atmosphere are believed to be known to within 10–20%, large uncertainties still remain in the estimates for individual sources, in particular for natural ecosystems like wetlands and the oceans. Isotope observations provide strong constraints also for the global atmospheric methane budget because different source types (biogenic vs. thermogenic) produce their own characteristic isotopic fractionations during methane production. These isotopic signatures can be used further to predict the future development of atmospheric methane concentrations and to understand better the drastic changes observed in the past under different conditions.

The Hamburg Oceanic Carbon Cycle Circulation Model as a global tracer model suited for glacial ocean studies is briefly described. Experimental strategies for paleoclimatic model experiments are outlined. In sensitivity experiments the effect of a hydrography shift due to a change in sea level on the marine carbon cycle is investigated.

An eleven-box model of the ocean-atmosphere subsystem of the global carbon cycle is developed to study the potential contribution of continental rock weathering and oceanic sedimentation to variations of atmospheric CO₂ pressure over glacial-interglacial timescales. The model is capable of reproducing the distribution of total dissolved inorganic carbon, total alkalinity, phosphate, δ¹³C, and Δ¹⁴C between the various ocean basins today, as well as the partial pressure of atmospheric CO₂. A simple sedimentation scheme at 20 different depth levels drives carbonate deposition and dissolution as a function of the depths of carbonate and aragonite lysoclines in each ocean basins considered (Atlantic, Antarctic and Indo-Pacific). The coral-reef erosion-deposition cycle is also taken into account. Furthermore, a simple cycle of oceanic strontium isotopes has been added to this model to take advantage of the ⁸⁷Sr/⁸⁶Sr data recently published by Dia et al. (1992) for the last 300,000 years. These data emphasize the importance of weathering of continental silicate rocks at glacial-interglacial timescales. They are used to construct several scenarios of changes of continental weathering over the last glacial cycles. They suggest that the flux of alkalinity delivered to the ocean from continental silicate weathering may have been substantially larger during glacial times than today. We show that such variations of continental weathering may explain at least in part the observed changes of the partial pressure of atmospheric CO₂ between glacial and interglacial periods.

A detailed study of paleoclimatic proxy data (stable isotopes, planktonic foraminiferal census data, carbonate content, and Ice Rafted Detritus (IRD)) in the surface sediments of the Greenland, Iceland and Norwegian Seas (GIN-seas) shows that different proxies are closely related to the surface water masses, to the position of oceanic fronts and to the sea ice extent. Both stable isotopes, foraminifers and sedimentological data differentiate between Polar water with extensive sea ice cover, Arctic water with only seasonal sea ice cover, and warm Atlantic water. The fronts that border these surface water masses are also well defined. Polar water is characterized by lower carbon and oxygen isotope values than Arctic water, and a slightly lower content of Neogloboquadrina pachyderma sinistral in the Polar Front region. Carbonate content is low and IRD input is high in Polar waters. Arctic water has highest carbon and oxygen isotope values, and is completely dominated by N. pachyderma sin. The Arctic Front is reflected by a clear isotopic gradient and by a strong switch from N. pachyderma sin. dominance to Globigerina quinqueloba dominance. Atlantic Water is defined by lower carbon and oxygen isotope values and by dominance of N. pachyderma dextral and increased amounts of Globigerina bulloides. The results have implications for paleoceanographic reconstructions of cold environments and point to the possibility of better defining sea ice margins and convective regions as well as frontal positions in past high latitude oceans. Applying these results to the Last Glacial Maximum and the Younger Dryas indicates more dynamic and less sea ice covered surface conditions in the GIN-seas than in earlier reconstructions.

Trace element and ²³⁰Th_excess profiles of sediment cores from two strategic localities yield further experimental arguments to our Mn-CO₂-hypothesis. At the Galapagos Microplate, which is supplied with hydrothermal Mn, we observe low Mn concentrations (as well as low Mn/Al and low Mn/Fe ratios) during isotopic stages 2 and 3 suggesting a longer residence time of Mn in the water column and enhanced export of Mn from this area to the glacial ocean. In the Angola Basin, west of the Congo Fan, peaks of Mn (and of Mn/Fe) coincide with peaks of ²³⁰Th_excess, which exceed the production flux at Terminations II and I. As ²³⁰Th is immobile in the sediments, the peaks suggest that the chemistry in the water column has changed significantly during the last part of glacial stages 6 and 2. Assuming a build-up of Mn in the water column during these periods, we infer that peaks of Mn in the sediments originate from rapid precipitation of MnO₂ at the beginning of interglacials. We estimate the impact of Mn deposition on the alkalinity in the range from 3 to 10) µequ./l. This could explain a part of the rapid increase of atmospheric CO₂ observed in the record of ice cores at the beginning of interglacial stages.

We use benthic foraminiferal assemblages and benthic δ¹³C to interpret glacial-to-interglacial contrasts in two gravity cores from the lower bathyal Antarctic continental margin at 69°S, and the abyssal Agulhas Basin at 43°S. As a recent analogue, sediment surface samples from an eastern Atlantic Ocean and Weddell Sea transect between 20°–70°S are discussed. In the investigated area, benthic foraminiferal assemblages reflect both ocean circulation and surface productivity. Also at most stations from a belt with seasonally high surface productivity between 48°S–55°S, the δ¹³C values of epibenthic Cibicidoides spp., including F. wuellerstorfi are depleted relative to the bottom water δ¹³C_∑CO2 and hence do not follow the 1:1 relationship established for more northern areas. This bears implications for the interpretation of large glacial-interglacial ⁵¹³C shifts from the Southern Ocean: Significant parts of this shift can be caused by a northward migration of high productivity belts associated with the Polar Front and the winter sea-ice limit rather than indicating nutrient-rich glacial Southern Ocean deep and bottom water.

During interglacial climatic optima, seasonally open surface water accompanied by relatively high opal and very low carbonate accumulation characterises the Antarctic continental margin environment. The recent benthic foraminiferal fauna indicates moderate productivity, but during peak warm periods (oxygen isotope Stages: 11, 9, 7.5, 7.3, 5.5 and 1.1) very low numbers of benthic foraminifera are inferred to represent maximum organic matter fluxes with severe calcite dissolution on the sea floor. Equally high δ¹³C values in surface and bottom water as inferred from planktic and benthic foraminifera, may indicate deep convection and bottom water formation during interglacials. In contrast, during glacials, very low opal accumulation, moderate carbonate accumulation, a benthic fauna that is presently associated with low productivity, as well as different benthic and planktic δ¹³C values are consistent with both a reduced primary productivity and a stratified water column, suggesting suppressed bottom water generation.

In the Agulhas Basin high carbonate and low organic carbon (C_org) accumulation reflect the late Holocene position of the site investigated well north of the present-day Polar Front. Low Holocene δ¹³C values of 0.3‰ and a benthic foraminiferal fauna that indicates a southern bottom water mass which is corrosive to carbonate is in agreement with the injection of North Atlantic Deep Water into Circumpolar Deep Water at intermediate depths, which does not affect bottom waters of this basin. During glacial periods, a specific southern fauna, associated with high productivity today, low carbonate, high sediment and C_org accumulation, and by 1.1‰ lower δ¹³C values indicate a bottom water mass of southern origin, a northward shift of the high productivity belt by 7°, and strongly diminished injection of NADW into the Southern Ocean.

Most planktic foraminifera live within the photic zone and exhibit a life style tied to the lunar cycle. They migrate between the reproductive depth (thermocline and/or the chlorophyll maximum) and the uppermost part of the photic zone. This ontogenetic migration pattern sets the initial δ¹³C of the foraminiferal shell. On top of that, biological fractionation processes (vital effects) modify the signal. These processes include photosynthetic activity of the symbionts and respiration of the host/symbiont complex. Globigerinoides sacculifer (Brady) was chosen to model ontogenetic changes in the δ¹³C of the shell as a function of depth migration with and without vital effects.

Oceanic cadmium, carbon isotopes, phosphorus, and dissolved carbon dioxide are correlated on a global scale because of the dominance of the biological cycling of these elements. Some differences between the distributions of these properties are seen in the modern ocean, although these differences can be rationalized. In the glacial maximum (18K) ocean, the differences between benthic foraminiferal δ¹³C and Cd/Ca are much larger than in the modern ocean, suggesting quite different scenarios for the carbon system chemistry of the Southern Ocean, the northwest Pacific Ocean, and the intermediate waters of the Pacific. A reexamination of the modern distribution of δ¹³C and Cd suggests that one major factor which has not been addressed in detail is the role of gas exchange on δ¹³C, but it is not clear that this factor can reconcile the differences between the tracers during the glacial maximum. The modern oceanic Cd-P relationship is reassessed in light of three new data sets and found to be consistent with previous interpretations of deep ocean data. The issue of foraminiferal accuracy and precision for δ¹³C and Cd/Ca reconstructions is considered: more effort on living benthic foraminifera is clearly needed to resolve the uncertainties concerning foraminiferal response to bottom water properties.

Deviations of δ¹³C:PO₄ and Cd:PO₄ ratios from the global trend are observed within the density range 26.8>σ_Θ>27.7 close to where the isopycnal surfaces outcrop at the sea surface. In these areas tracer concentrations seem to be decoupled from nutrient concentrations presumably because of the effects of gas exchange and varying recycling depths for tracers and nutrients. Temperature exerts strong control on the surface water δ¹³C anomalies in high latitudes but results from an ocean-atmosphere δ¹³C -simulating box model imply that the anomalies still exist if air-sea δ¹³C -fractionation is held constant for the warm and cold surface ocean. The warm and northern Atlantic surface waters acquire a lower δ¹³C than expected from biological fractionation alone, whereas the Antarctic surface δ¹³C is somewhat more positive. A dynamic balance is achieved where CO₂ evading from Antarctic waters to the atmosphere has a lower ¹³C/¹²C ratio than that invading, while the isotope ratio evading from the warm ocean is greater than that of the CO₂ return flux.

Box model simulations of Cd and PO₄ distribution suggest that differential uptake of Cd in Antarctic waters affects the ocean Cd distribution globally because these waters are entrained into intermediate waters which spread about the global ocean. Stripping Cd from Antarctic surface waters produces the greatest intermediate water depletion in the Atlantic, deeper recycling produces the greatest intermediate water Cd depletion in the North Pacific with noticeable lower Cd:PO₄ ratios than those of the deep waters in that ocean. Because of the polar mediterranean nature of the circulation in the northern Atlantic differential uptake of Cd in boreal Atlantic surface waters would only be detected locally. In the rest of the ocean that is capped by other surface waters, a linear correlation of Cd and PO₄ would be observed throughout the water column. Changing parameters related to biological cycling in Antarctic surface waters exerts by far the largest control on the distribution of Cd and PO₄ in the world ocean.

Because δ¹³C and Cd are not uniquely tied to phosphorus a reliable interpretation of foraminiferal δ¹³C and Cd/Ca records is difficult, especially if they come from shallow-ocean core sites at high latitudes or other areas of oceanic upwelling. When estimating nutrient concentrations from benthic δ¹³C and/or Cd/Ca ratios we need to know if we are on the low-nutrient or high-nutrient side of the break in the global regression lines. It may be possible to obtain this information by measuring a number of different proxies with different boundary conditions. Subtle differences between independent proxy records may tell us more about past ocean chemical cycling than coherent paleo-proxy records.

The search for palaeo tracers in the last decade has resulted in a number of potentially useful tracers for past temperature/salinity, productivity, pCO₂(aq), nutrients, etc. Recently, it has become evident that some limitations exist on the application of these tracers. Early diagenetic processes may considerably alter the primary signals. The diagenetic fates of palaeo proxies may vary for different environments, which can lead to preferential preservation, decomposition or relocation of particular tracers. Consequently, one must be very cautious when applying a palaeo-tracer that is sensitive to early diagenetic alteration. In this paper we present data from the Madeira Abyssal Plain, the eastern Mediterranean, and Kau Bay that clearly illustrate some of these alterations:

A variety of biological processes associated with upper ocean phyto- and zooplankton may be important in determining the isotopic composition of organic matter sedimenting to the sea floor. In the modern ocean, the natural abundance of ¹⁵N (δ¹⁵N) varies in a regular way within planktonic ecosystems as a result of biological isotopic fractionation: the δ¹⁵N of the primary producers depends on the isotopic composition of the inorganic forms of nitrogen used as a substrate for growth, while the natural abundance of ¹⁵N in zooplankton increases as a function of trophic position, with an average increment of about 3.5‰ per trophic step. Among the processes that are likely to be of critical importance in determining the δ¹⁵N of sedimenting organic matter are the uptake of inorganic nitrogen by phytoplankton, the remineralization of nutrients within the euphotic zone and the production of rapidly-sinking fecal pellets and detrital particles by zooplankton. In addition, a number of recent studies provide evidence that the isotopic signature of different components of planktonic ecosystems may vary dynamically on time scales of days to months in response to perturbations in the nitrogen cycle such as wind-induced vertical mixing or seasonal variations in production. As a result of such temporal changes and the filtering of the isotopic signal that occurs as nitrogen moves through the planktonic food web, the δ¹⁵N of sedimentary organic material is likely to be most useful in studying processes that affect planktonic ecosystems as a whole. The utility of δ¹⁵N measurements in paleoceanographic studies will be greatly enhanced by additional study of the ways in which the physiological processes of phyto- and zooplankton discriminate between the nitrogen isotopes. In this paper, I review the areas of plankton biology that are most relevant to paleoceanographic applications of ¹⁵N natural abundance measurements, and provide a brief overview of areas of current research interest.

Sedimentary ¹⁵N/¹⁴N ratios can be used as a unique measure of past changes in surface ocean nutrient utilization which in turn is a function of past changes in productivity and nutrient input from deeper waters. Due to isotopic fractionation during nitrate uptake by phytoplankton, partial nutrient utilization produces substantial ¹⁵N enrichment in particulate matter reaching the seafloor. Examples are given from the N. Atlantic, Equatorial Pacific, and Southern Ocean. In the latter two regions large variations in nutrient utilization with latitude are observed which go along with gradients in both near-surface ocean and core top ¹⁵N/¹⁴N. Thus isotopic signals generated in surface waters are transferred and preserved in sediments. First down core results from the Southern Ocean indicate that only modest increases in nutrient utilization and perhaps productivity occured in Subantarctic waters uring the last glacial maximum.

Theory, laboratory studies, and ocean observations indicate that the δ¹³C of marine plankton biomass and sedimentary remains thereof can be used as a proxy for ambient molecular CO₂ concentration, [CO_2(aq)] in ocean surface waters. A compilation of in situ ocean data suggests that about 89% of the global δ¹³C_org variation within bulk plankton or seston can be explained by a simple negative linear response to ambient [CO_2(aq)] with the slope of the best-fit line = -0.6 ‰ µM^-1. With this model the standard error of the estimate of surface ocean [CO_2(aq)] is ±2.0 µM when δ¹³C_org is specified. This residual variability may be largely due to effects on plankton δ¹³C_org imparted by changes in phytoplankton CO₂ demand that are independent of [CO_2(aq)]. Within this variability and within the current range of ocean [CO_2(aq)] there are slight differences between this model and various proposed nonlinear fits to observed global data. While an inverse relationship that can be influenced by both CO₂ demand as well as concentration is theoretically expected, it does not provide an improved fit to observations over the negative linear model. When applied to the sedimentary δ¹³C_org record, the latter model predicts that the approximate 80 µatms increase in atmospheric pCO₂ during the last glacial-interglacial transition (as documented by ice core analyses) should have resulted in a 1–2 ‰ decrease in plankton δ¹³C. Indeed, changes of this direction and magnitude are evident in most low-latitude Pleistocene/Holocene sediment core profiles of δ¹³C_org thus far reported. However, some geographic and temporal differences in past plankton isotopic response are present and expected due to i) regional non-equilibrium between ocean and atmospheric [CO₂], and ii) changes in phytoplankton CO₂ demand.

The principles and practice of reconstruction of concentrations of dissolved CO₂ from carbon isotopic compositions of contemporaneous, coeval marine organic matter and carbonates are summarized here. Techniques for determination of the isotopic fractionation accompanying photosynthetic fixation of carbon are exemplified by three oceanic case histories. The northern Gulf of Mexico during the late Quaternary represents an environment with near-equilibrium distribution of CO₂ between the atmosphere and ocean surface, whereas non-equilibrium conditions prevail in the central equatorial Pacific. The Black Sea, with surface waters buffered strongly by high concentrations of dissolved CO₂ at depth, represents an exotic PCO₂ environment. Consideration of a preliminary water-column model focussed on the loci of production of biogenic phases used for reconstruction of paleoceanic PCO₂ illuminates some of the major issues to be considered as this technique is refined.

The δ¹³C record for organic carbon in Late Quaternary sediments of the Eastern Angola Basin shows a pronounced cyclicity with high values (-18.3 to -20‰) in glacial isotopic stages 2 to 4 and stage 6, and lower values (-20 to -21.2‰) in interglacial stages 1 and 5. Seasurface temperatures, derived from the alkenone U^k ₃₇ index, vary in the opposite sense, ranging from 20–22°C in glacial stages up to 24–26°C in interglacial stages. The inverse relationship between δ¹³C_org and SST values suggests that the isotopic variations are not due to temperature-dependent isotopic fractionation during photosynthesis, in which case δ¹³C_org values would be expected to increase with increasing temperature. Relationships between sedimentary δ¹³C_org and C/N ratios indicate that differing proportions of marine and terrigenous organic matter can also be ruled out as a cause for the δ¹³C_org variability. Instead we conclude that changes in surface water CO₂ concentrations were responsible for the observed glacial to interglacial isotopic variations.

Employing the relationships between planktonic δ¹³C_org and dissolved CO₂(aq) derived by POPP et al. (1989) and RAU et al. (1989) we estimate that the partial pressure of carbon dioxide (PCO₂) in the surface water of the Angola Current (AC) has been consistently higher than values given by the Vostok atmospheric CO₂ record for the last 160,000 years. This indicates that the AC has generally acted as a source region for atmospheric CO₂ throughout this period, which is consistent with the modern situation. However, the difference between surface-water and atmospheric pCO₂ was larger in interglacial (pCO₂ ≈ 100–120 µatm) than in glacial times (mean pCO₂ ≈ 70 µatm). This suggests that the AC was a weaker CO₂ source during glacial periods, presumably due to lower sea-surface temperatures and higher biological productivity.

We reconstructed past variations in CO₂ partial pressure (local PCO₂) in the surface waters of the East Atlantic equatorial upwelling zone over the last 330,000 years, based on the δ¹³C record of the (marine) organic matter in ‘Meteor’ core 16772. To deduce the initial δ¹³C_organic values of plankton and the CO₂ solubility in surface water, the δ¹³C record was adjusted for i) past variations in (winter) sea surface temperature, ii) variations in the δ¹³C composition of inorganic carbon dissolved in the surface waters, using the δ¹³C values of G. ruber (white), and iii) isotopic fractionation during the degradation of settling organic matter in the water column and on the sediment surface.

The calculated paleo-PCO₂ variations in the surface waters show a strong signal at the obliquity frequency and are approximately parallel to the VOSTOK ice-core record of atmospheric PCO₂ over the last 140,000 years. Holocene PCO₂ values varied within the range of modern local PCO₂, which is 350–400 ppmv compared to a pre-industrial atmospheric pCO₂ level of 280 ppmv. This positive anomaly demonstrates the persistent CO₂ release from upwelled subsurface water. The glacial-to-interglacial amplitudes of local PCO₂ (at the core site) exceeded those of atmospheric pCO₂ by 20–60%, with values of less than 250 to 300 ppmv during cold isotopic stages, which indicate a decreased net carbon outgassing from the ocean to the atmosphere. The close correlation between high paleo-PCO₂ and low paleo-nutrient contents and paleoproductivity (r=0.7–0.8) suggests that the local PCO₂ variations resulted mainly from CO₂ transfer by phytoplankton production, especially over the last 170,000 years.

There are three fundamentally different proxies recording ocean productivity in sediments: those representing flux (that is, export production), those representing nutrient concentrations (e.g., nitrate or phosphate), and those representing aspects of the trophic structure of the pelagic environment. Flux proxies disagree due to different power-law relationships to the source term. The particular power-laws (expressed in terms of proxy exponents) change as a function of geography, and may change as a function of time. Modification of the accumulation rate by changing preservation on the sea floor (and within the sediment) is an important aspect of this problem.

Combining the use of flux proxies and nutrient proxies makes it possible to estimate the magnitude of physical processes such as mixing and upwelling, and the positive feedback from nutrient buildup in the thermocline. These effects are subsumed into a parameter [v] in the equation Flux = nutrients x [v]. A study of data from the western equatorial Pacific suggests that mixing, upwelling and subsurface nutrient buildup were greatly increased in glacial Stage 6. Diatom abundance does not go parallel to either the flux or the nutrient proxy. Paradoxically, it shows reverse fluctuations. Silicate starvation during glacials may be indicated.

Particle fluxes using sediment traps have been seriously studied in the past decade for various microfossil groups including biogenic opal. We have accumulated a substantial amount of information pertinent to properly interpret the past environments. The majority of biogenic opal production is lost due to dissolution mainly at the sea-floor and only a minute fraction can be preserved in the fossil record. Numerical correlations between fluxes and the fossil record are needed to establish a linkage between modern and fossil assemblages based on the minute quantity preserved. A prototype numerical paleoecological method has been constructed. This method is used to quantitatively interpret paleofluxes of diatoms. Paleoflux results from two diatom taxa are presented from the northeastern Pacific: Denticulopsis seminae and Coscinodiscus marginatus. Denticulopsis seminae, a pennate diatom taxon widely occurring in the high latitude subarctic Pacific, is characterized as a productivity taxon in the pelagic realm with high fossil abundance. The other taxon important for winter conditions in the present day subarctic Pacific is Coscinodiscus marginatus, one of the best preserved diatoms in the fossil record of the region.

We have generated 350 ka records of diatom accumulation rates (DAR) and export production (P_exp.), based on organic carbon accumulation rates, for Core 16772 in the eastern Equatorial Atlantic (EEA). These records were compared to published records for the coastal upwelling areas off Portugal and NW Africa, in order to obtain a better understanding of NE Atlantic paleoproductivity variations.

Accumulation rates of the various siliceous microfossils correlate well at each location. The DAR shows a good overall correspondence to P_exp. at all locations; however, discrepancies between the DAR and P_exp. records at the beginning of the interglacial stages are noted for all areas. Assuming that both records are basically a measure of the same process, the primary productivity in the surface ocean, the observed discrepancies can be explained by changes in opal preservation and/or in the plankton community that dominated the equatorial waters during this time. The EEA records exhibit marked oscillations, with maxima occurring during glacial stages. This oscillating character is interpreted as deriving from variations in productivity through time, so that increased primary production has characterized every major glacial period for the last 350 ka in the area. However, only glacial Stages 6 and 4 are characterized by high abundance spikes of the large diatom Ethmodiscus rex (Rattray) Hendey.

Comparison of this equatorial upwelling record with the record of the coastal upwelling areas (1326 KS11 off Portugal and 12392–1 off NW Africa) for the last 120 ka, shows that although the maxima are of the same order of magnitude, the times of maximum values are not the same. In the eastern Equatorial Atlantic the most intense DAR and P_exp. peaks are observed during Stage 4 but barely recorded in the coastal areas where the most productive period is Stage 2. This high productivity recorded during Stage 2 on coastal areas is also recorded in the EEA by a P_exp. maximum of the same magnitude, although DAR values are 4xs lower than the observed in the coastal record. The difference in the timing of maximum siliceous production (DAR) between oceanic and coastal regimes which is not observed for the carbon production record (P_exp.), is interpreted as an indication of important changes in the silica concentration of the upwelling source waters (intermediate waters).

We compare paleoproductivity proxy records from a set of gravity cores from the Exmouth Plateau (≈ 19°S, 113°E, 950 to 2250 m) and the Perth Basin (≈ 27°S, 111°E, 2750 m) in the southeastern Indian Ocean. In general, these proxies indicate higher surfaceocean productivity in this region at the Last Glacial Maximum (LGM, isotope Stage 2). LGM sediment accumulation rates and the accumulation rates of biogenic sediment components (CaCO₃ and organic carbon) are a factor of 1.5 to 2 higher than Holocene values. Benthic foraminiferal abundances and accumulation rates are both higher in glacial sediments, as are the concentrations and accumulation rates of authigenic uranium in the sediments. These benthic foraminiferal abundance and authigenic uranium data suggest higher surface ocean productivity during the glacial, but we cannot yet relate them to carbon flux quantitatively. In contrast to these three approaches, a productivity proxy based on paired-species benthic foraminiferal δ¹³C differences shows little glacial-Holocene change. Possible explanations for this disagreement are discussed.

Together, the data suggest that the glacial productivity off Western Australia was elevated relative to Holocene values, and support the hypothesis that a north-flowing west Australian current led to coastal upwelling and enhanced primary productivity off western Australia during the Last Glacial Maximum. However, glacial productivity was high only relative to the low productivity characteristic of this region in the modern ocean. We see no evidence of strong upwelling similar to that observed in the modern ocean off the southwestern coasts of Africa and South America.

The seminal proposition by Arrhenius (1952) that colder periods favor higher productivity oceans than warmer ones, together with the relatively recent hypothesis linking the surface ocean export productivity levels and atmospheric pCO₂ have motivated a renewed interest in the reconstruction of the ocean’s paleoproductivity from deep-sea sediments. In this paper we present a record of the relative magnitude and timing of the paleoexport of organic matter to the sea-floor for the last 250 Kyr for the western equatorial Pacific region. The benthic foraminifera preserved in deep-sea sediments are used to estimate the organic matter flux to the sea floor and indirectly they are also a measure of the organic carbon export of from the photic layer. This paleoexport record shows a remarkable coherence with the ice record, as indicated by the δ¹⁸O of benthic foraminifera. We further estimate the relative importance of changing nutrients and upper ocean mixing rates to explain the observed paleoproductivity fluctuations. The record of changes in nutrient concentrations is based on the interpretation of the Δδ¹⁸O_PB between planktics and benthics as a paleonutrient tracer. To estimate the magnitude of the physical forcing, or the upper ocean mixing rates, we use a simple box model, together with a paleotemperature record based on the interpretation of the Δδ¹⁸O_GsPo between two planktic foraminifera living at different depths within the mixed layer. Paleoexport, the nutrient content, and the ocean mixing rates records are compared to estimate the relative importance of each mechanism in explaining the observed fluctuations of the paleoexport record.

Dinoflagellate cysts are useful marine paleoenvironmental indicators of: sea surface temperatures, salinity, and coastal/oceanic water mass boundaries. Most are coastal/neritic; very few oceanic cysts are produced, and deep sea assemblages probably often include many that are transported long distances from coastal environments. This factor must be considered in attempts to apply standard transfer functions in paleoceanographic interpretations from deep sea sediments. Cysts are potentially useful for productivity studies as indicators of upwelling and eutrophication (especially where mineralized microfossils are absent through dissolution).

Benthic foraminifers are an abundant, nearly cosmopolitan and easily preserved component of the deep-sea meiofauna. The benthic foraminiferal faunas are very strongly structured on regional scales. In the western North and South Atlantic the distribution of benthic assemblages follows the distribution of named “water masses,” leading to the expectation that fossil benthic foraminifers from sediment cores allow a reconstruction of past deep-water circulation patterns. However, in many other areas the faunal composition is strongly correlated with the productivity of the overlying surface waters. Such results demonstrate that the benthic meiofauna responds to the integral of the deep environment and it is the limiting or overwhelming influence that determines the success or failure of individual species and thus the composition of the fauna. Analysis of available data clearly indicates that benthic foraminifers are unequivocal indicators of productivity in areas where productivity is high. In areas of low or very uniform productivity the composition of the benthic fauna clearly carries the imprint of deep water mass structure as the dominant feature. Faunal sequences from late Quaternary cores from the Atlantic and Pacific oceans show very strong differentiations. Not only is the glacial/postglacial transition shown, but also much superimposed fine structure. Faunal composition is as sensitive to deep-water environmental change as are geochemical indicators. The faunal composition allows at least some appreciation of the cause for the major changes, but a quantitative differentiation between productivity or water mass changes is not yet possible. If, however, an independent assessment of productivity change is available, then the benthic foraminiferal faunas are exquisite indicators of changes in deep water circulation.

Over the last decade many regional palaeoceanographic studies have found evidence for enhanced primary productivity during glacial episodes, particularly in the equatorial Pacific and off northwest Africa. These studies have given rise to the “glacial productivity hypothesis” which has suggested that enhanced oceanic biomass during glacial times supported an efficient removal of organic carbon from the euphoric zone, contributing to lower atmospheric CO₂ levels recorded in ice cores. Recently, studies from the Southern Ocean have shown that south of the Antarctic Polar Front glacial palaeoproductivity was lower than during interglacial times. Here we present further evidence for enhanced interglacial productivity in the Southern Ocean using a transect of cores collected from the Scotia Sea and Weddell Sea. Biogenic silica, organic carbon and barium are presented as proxy indicators of past productivity. In order to establish such palaeoproductivity records for this region, where an absence of foraminifera precludes the standard use of δ¹⁸O stratigraphy and carbonate ¹⁴C dating, we have developed a method based on the synchronous removal of barium to the sea floor by scavenging and the formation of barite within the frustules of marine diatoms. The barium record is calibrated to a δ¹⁸O Specmap time scale from a single core at 68° 45′S, 5° 53′W (PS 1506) where planktonic and benthic foraminifera are found. One core from the Weddell Sea has been studied using the ²³⁰Th_xs dating method to confirm the barium stratigraphy. Using the age models developed here, we identify important increases in palaeoproductivity during isotope stage 5e, and during the Holocene (Stage 1). Some evidence for a decrease in productivity during a cooling event between 11,000 and 12,000 years BP is observed. Highest palaeoproductivity, defined by biogenic opal accumulation, occurs in the vicinity of the Scotia Arc, just south of the present-day Antarctic Polar Front. Glacial productivity (18 to 72 ky BP) was weaker and displaced slightly to the north. The extent of glacial sea ice is considered to be of primary importance in governing the budget of biogenic detritus to these Southern Ocean sediments.