By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
There is approximately 50 times more inorganic carbon in the global ocean than in the atmosphere. On time scales of decades to millions of years, the interaction between these two geophysical fluids determines atmospheric CO2 levels. During glacial periods, for example, the ocean serves as the major sink for atmospheric CO2, while during glacial–interglacial transitions, it is a source of CO2 to the atmosphere. The mechanisms responsible for determining the sign of the net exchange of CO2 between the ocean and the atmosphere remain unresolved. There is evidence that during glacial periods, phytoplankton primary productivity increased, leading to an enhanced sedimentation of particulate organic carbon into the ocean interior. The stimulation of primary production in glacial episodes can be correlated with increased inputs of nutrients limiting productivity, especially aeolian iron. Iron directly enhances primary production in high nutrient (nitrate and phosphate) regions of the ocean, of which the Southern Ocean is the most important. This trace element can also enhance nitrogen fixation, and thereby indirectly stimulate primary production throughout the low nutrient regions of the central ocean basins. While the export flux of organic carbon to the ocean interior was enhanced during glacial periods, this process does not fully account for the sequestration of atmospheric CO2. Heterotrophic oxidation of the newly formed organic carbon, forming weak acids, would have hydrolyzed CaCO3 in the sediments, increasing thereby oceanic alkalinity which, in turn, would have promoted the drawdown of atmospheric CO2. This latter mechanism is consistent with the stable carbon isotope pattern derived from air trapped in ice cores. The oceans have also played a major role as a sink for up to 30% of the anthropogenic CO2 produced during the industrial revolution. In large part this is due to CO2 solution in the surface ocean; however, some, poorly quantified fraction is a result of increased new production due to anthropogenic inputs of combined N, P and Fe. Based on ‘circulation as usual’, models predict that future anthropogenic CO2 inputs to the atmosphere will, in part, continue to be sequestered in the ocean. Human intervention (large-scale Fe fertilization; direct CO2 burial in the deep ocean) could increase carbon sequestration in the oceans, but could also result in unpredicted environmental perturbations. Changes in the oceanic thermohaline circulation as a result of global climate change would greatly alter the predictions of C sequestration that are possible on a ‘circulation as usual’ basis.
The oceans cover 70% of the Earth's surface and contain about 50 times more soluble inorganic carbon than the atmosphere. The fluxes of CO2 between these two geophysical fluids are enormous, averaging approximately 100 Pg C per annum. Since the Industrial Revolution, there has been a net flux of CO2 from the atmosphere to the oceans, which presently amounts to about 2 Pg per annum. Up to an additional 1·5 Pg C may be sequestered by terrestrial ecosystems. On time scales of millions of years, the net flux of CO2 between the oceans and atmospheres is driven by very small changes in both inorganic and biological processes. On time scales of years to decades, the net flux between the oceans and atmosphere is primarily driven by physico-chemical processes related to solubilization of CO2 in the oceans, while on time scales of centuries to millennia biological responses play a key role (see review by Sarmiento & Bender 1994). Geochemical models clearly suggest that removal of atmospheric CO2 by a terrestrial sink (e.g. enhanced net primary production) will result in outgassing of CO2 from the oceans. We strive here to explain the critical factors that determine the net fluxes of CO2 between the oceans and atmosphere, with a goal of explaining key feed-backs in the global carbon cycle. We begin by considering the global carbon cycle in terms of the chemical components and the biogeochemical mechanisms involved, the pool sizes of the components and the magnitude of the fluxes among them.
THE ROLE OF THE OCEANS IN THE NATURAL GLOBAL CARBON CYCLE
On geological time scales (greater than hundreds or thousands of years), inorganic carbon is sequestered in the oceans as either Ca or Mg mineral salts. The geochemical reaction schemes responsible for the formation of the Ca and Mg mineral salts were first described by Harold Urey (Urey 1952; see also Berner, Lasaga & Garrels 1983; Berner & Berner 1996). These so-called ‘Urey reactions’ are summarized thus:
CO2 + CaSiO3→ CaCO3 + SiO2 (1)
CO2 + MgSiO3→ MgCO3 + SiO2 (2)
It should be noted that CaSiO3 and MgSiO3 represent any generic Ca or Mg silicate produced at high temperatures in the Earth's crust, that SiO2 represents any sedimented silicate which has not been subjected to high temperatures, and that MgCO3 is the Mg component of the mixed Mg and Ca carbonate, dolomite (Berner et al. 1983). In Eqns 1 and 2 ‘weathering’ occurs on the land surface (noting that continental crust has existed for as long as we have fossil evidence of cyanobacteria, i.e. 3·45 billion years: Raven 1998a) and yields bicarbonates rather than carbonates. The precipitation of CaCO3 and MgCO3 (and SiO2) occurs only in the oceans. Thus, the terrestrial part of the weathering reactions in Eqns 1 and 2 may be represented as:
2 CO2 + H2O + CaSiO3→ Ca(HCO3)2 + SiO2 (3)
2 CO2 + H2O + MgSiO3→ Mg(HCO3)2 + SiO2 (4)
While these reactions are inorganic, they are influenced by biological processes. With the emergence of terrestrial vegetation as cyanobacteria and microalgae 1·2 billion years ago, and especially as higher plants (embryophytes) 420 million years ago (Raven 1998), the weathering reactions were accelerated. Terrestrial plants fix atmospheric CO2, and transfer a fraction of the fixed organic carbon to the soil surface, where it is respired. This terrestrial CO2 pump enriches groundwater, and hence rivers, with inorganic carbon, with little possibility of diffusive efflux to the atmosphere (Berner et al. 1983; Berner 1992, 1994, 1997, Retallack 1997; Algeo & Schleckler 1998; Berner 1998; Elick, Driese & Mora 1998). The flux of inorganic carbon and its associated divalent cations enriches the oceans with these inorganic molecules on time scales of millennia.
Following the transport of Ca(HCO3)2, Mg(HCO3)2 and SiO2 (as (Si(OH)4), from fluvial sources to the ocean, precipitation reactions become dominant. The marine precipitation processes are represented as:
Ca(HCO3)2→ CaCO3 + CO2 + H2O (5)
Mg(HCO3)2→ MgCO3 + CO2 + H2O (6)
The sum of Eqns 3 and 5 is the weathering component of Eqn 1, while the sum of Eqns 4 and 6 is the weathering component of Eqn 2. In the contemporary oceans there is very little precipitation of dolomite; and essentially all of the precipitation of CaCO3 (as calcite) and SiO2 is biologically mediated (Berner et al. 1983; Berner & Berner 1996).
Another component of Urey reactions is the sequestration of CO2 on land by the dissolution of carbonates that had precipitated previously in the ocean according to Eqns 5 and 6, but escaped metamorphism and subsequently were uplifted and exposed. This carbonate dissolution pathway effectively is the reverse of Eqns 5 and 6, thus:
The estimated fluxes of CO2 associated with Urey reactions were calculated by Berner et al. (1983). The uptake of CO2 from the atmosphere related to carbonate dissolution on land (Eqns 7 and 8) is 0·14 Pg C per year, and that associated with silicate weathering (Eqns 3 and 4) is 0·14 Pg C per year, yielding a total of 0·28 Pg C per year. Assuming a steady-state, CO2 input resulting from calcite precipitation in the ocean (Eqns 5 and 6) amounts to 0·209 Pg C per year. Vulcanism and other metamorphic fluxes (the ‘metamorphic’ direction of Eqns 1 and 2) produce 0·071 Pg C per year. Hence, the total of 0·28 Pg C per year.
These global fluxes, driven by geochemical processes are relatively small compared with biological fluxes of carbon. Terrestrial net primary production (NPP) is presently estimated at approximately 56 Pg C per year, while marine NPP is approximately 45 Pg C per year (Field et al. 1998). However, small changes in the Urey reactions are very significant as long-term (millions to billions of years) determinants of atmospheric CO2 (Berner et al. 1983; Berner 1994, 1997; Retallack 1997). Among the more important processes altering the rates of CO2 input to the atmosphere by metamorphism (Eqns 1 and 2) is the extent of tectonic plate subduction. More subduction leads to a greater magmatic production of CO2 by metamorphism (Berner et al. 1983). CO2 removal by the weathering processes described in Eqns 3, 4, 7, and 8 is also influenced by plate tectonics in that tectonic processes determine the continental area exposed to weathering (Berner et al. 1983).
Much of the rest of this paper is concerned with the factors controlling primary productivity in the long term sequestration of carbon in the oceans. Our approach uses palaeo-ecological data on the role of oceans and their biota in previous episodes of variation in atmospheric CO2, especially the last glaciation, as well as mechanistic models of what may happen as CO2 increases. We also consider interventionist procedures such as iron fertilization and direct transfer of CO2 to the deep ocean.
CONSTRAINTS ON PRIMARY PRODUCTIVITY AND ITS RECYCLING TO CO2 IN TODAY's OCEAN AND ITS IMPLICATIONS FOR ATMOSPHERIC CO2
Supply of inorganic and organic C from terrestrial biota
The ocean is a net recipient of both organic and inorganic carbon from terrestrial systems. Respiration in soils inevitably leads to the production of CO2, which equilibrates between the solution and gas phases. While diffusion of some CO2 gas to the atmosphere occurs, some of the dissolved CO2 becomes hydrated and promotes the weathering reactions depicted in Eqns 3, 4, 7 and 8. The hydrated CO2 (i.e. H2CO3 + HCO3–+ CO32–), together with some of the free dissolved CO2 is ultimately transported to bodies of freshwater where biologically mediated oxidation of allochthonously produced (i.e. imported) organic C generates more CO2. In open freshwater aquatic ecosystems, CO2 can become supersaturated and, despite photosynthetic consumption, can outgas to the atmosphere (Cole et al. 1994). The net effect of these processes is a flux of CO2 from freshwaters to the atmosphere amounting to approximately 0·14 Pg C (Cole et al. 1994).
Essentially all of the HCO3– (0·28 Pg C) generated in the reactions defined in Eqns 3, 4, 7 and 8 is transported to the ocean in rivers. The flux of organic carbon amounts to approximately 0·4 Pg C (see Meybeck 1993; Watson & Liss 1998). The terrestrially derived organic carbon is often identified by the presence of lignin (which is not significantly produced by marine photoautotrophs). In the oceans, lignin, which is relatively resistant to decomposition, is found only in coastal environments, especially in sediments of continental margins (Premuzic et al. 1982). Hence, the coastal oceans are the primary sink for terrestrially derived organic matter exported from the continents. The total flux of inorganic plus organic carbon from terrestrial to marine ecosystems is approximately 0·7 Pg C per annum. The inorganic component helps to subsidize marine primary productivity, while the organic carbon component is of a similar magnitude to the long-term storage of organic carbon in oceans (Beran 1995).
Ocean circulation and distribution of CO2 and other solutes
On time scales of centuries to millennia, oceanic inorganic C controls the atmospheric level of CO2, not vice versa. The equilibrium distribution of CO2 between the ocean and the atmosphere is critically dependent upon the temperature, alkalinity and salinity of surface waters (Broecker 1985; Takahashi 1989; Bigg 1996; Sarmiento & Quéré 1996; Broecker 1997). Mean oceanic salinity and alkalinity vary slightly with freshwater sequestration in ice sheets over millennial time-scales, while mean sea surface temperature varies over a shorter time scale. Higher temperatures and salinity mean lower CO2 solubility.
There are variations in temperature, alkalinity and salinity driven by solar radiation via the hydrological cycle and the warming of low-latitude oceans. In the absence of any biological activity in the oceans, water which cools and then downwells at high latitudes has high CO2 concentrations (not necessarily in full equilibrium with the atmosphere: Smith 1985; Watson, Upstill-Goddard & Liss 1991). This situation occurs only in the North Atlantic, off the coast of Greenland and Labrador, and in the Southern Ocean, but not in the North Pacific. Upon upwelling at low latitudes the cold, CO2-enriched, water from the ocean interior warms and becomes supersaturated with respect to CO2. Prior to the Industrial Revolution, the net flux of carbon from high latitudes to low latitudes in the ocean amounted to some 1 Pg C per year (see Watson & Liss 1998). This flux was balanced by the transport of 1 Pg C from low to high latitudes in the atmosphere (Fig. 1). The oceanic absorption of CO2 at high latitudes and subsequent subduction is called the ‘solubility pump’, and is primarily maintained by cold temperatures in the ocean interior. Consequently, the concentration of inorganic carbon below the upper 500 m of the ocean is significantly higher than the air–sea equilibrium values (Fig. 2).
Oceanic photosynthesis further modifies this situation. A fraction of the carbon fixed by phytoplankton in the upper ocean is exported (sinks) to the ocean interior, where it is oxidized by microbes, regenerating CO2 in the deep ocean. This ‘biological pump’ (Volk & Hoffert 1985) produces a vertical-inorganic carbon profile that is further enriched at depth (Fig. 2). A similar profile is found for other biologically active solutes in oceans (e.g. NO3–, NH4+, HPO42–, Si(OH)4, and Fe). Unlike other nutrients, however, the fractional difference in total inorganic carbon concentrations between surface and deeper water is much smaller than for NO3–, NH4+ and HPO42–, providing prima facie evidence that inorganic carbon does not limit primary productivity in the oceans.
Physical processes that influence vertical fluxes of nutrients are the critical determinants of primary production in the ocean. Solar heating of surface ocean waters isolates the upper mixed layer from the cold, deep ocean interior. The thermal gradient constrains phytoplankton in a well-illuminated (euphotic) zone, but retards mixing with deeper, nutrient-rich waters. This situation permanently characterizes tropical and subtropical seas. Seasonal heating and cooling in temperate and boreal regions, coastal upwelling driven by winds, and the dissipation of kinetic energy by turbulent mixing (e.g. storms, eddies, and high frequency internal waves), facilitates the injection of nutrients into the euphotic zone from deeper waters. The importation of ‘new’ nutrients into the euphotic zone permits transient blooms of phytoplankton. In the steady-state these fluxes are balanced by the vertical ‘export’ flux of organic material to the ocean interior. This export flux fuels biological processes in the deep ocean. Alternatively, phytoplankton can be consumed within the euphotic zone, leading to a ‘recycling’ of nutrients. About 70% of global marine primary productivity is ‘recycled’ and some 30% is ‘new’ (Falkowski & Raven 1997; Field et al. 1998). While distribution of new productivity is highly variable spatially and temporally, it is important to emphasize that neither ‘new’ nor recycled production affects the global net exchange of CO2 between the atmosphere and ocean.
The average elemental ratio of inorganic C : N : P regenerated by the oxidation of organic matter in the ocean is highly constrained. This so-called ‘Redfield ratio’ of 106C : 16N : 1P (by atoms), is unique to marine ecosystems and reflects the annually and spatially averaged, proximate, highly conserved, chemical composition of phytoplankton (Redfield 1958; Copin-Montegut & Copin-Montegut 1983). The Redfield ratio is useful in calculating the organic carbon potentially produced for a given amount of fixed inorganic nitrogen and phosphate, regardless of whether the production is regenerated or ‘new’.
Throughout most of the central oceans, the ratio of fixed inorganic nitrogen (in the form of NO3–) to P (as HPO42–) averages 14·7 rather than 16 (Falkowski 1997a; Gruber & Sarmiento 1997). This small difference is a consequence of losses of combined nitrogen due to denitrification relative to nitrogen inputs via N2 fixation or riverine fluxes (Codispoti 1995). Carbon and phosphorus, and to a much lesser extent, nitrogen, can be lost by sedimentation (Berner 1980). Prior to the Industrial Revolution, the quantity of organic carbon buried in ocean sediments amounted to some 0·2 Pg C per year (Fig. 1 of Watson & Liss 1998; cf. Chin, Orelland & Verdugo 1998; Wells 1998). This mass balance estimate implies that less organic carbon is sedimented in the central oceans basins than is delivered by rivers. Virtually all of the oceanic sedimentation of organic carbon occurs in coastal waters (see Hartnett et al. 1998; Reimers 1998), and a significant fraction of this organic carbon is derived from terrestrial primary productivity. The accumulation of organic carbon in ocean margin sediments represents very small imbalances between organic carbon production and respiration. Almost no terrestrial or aquatic NPP escapes oxidation to CO2 either biologically through heterotrophic respiration or, for some terrestrial habitats, by combustion from fires. The average turnover time of carbon in terrestrial ecosystems is on the order of two decades, whilst in aquatic ecosystems the average turnover time is on the order of a week (Beran 1995; Falkowski, Barber & Smetacek 1998).
To summarize, it is generally assumed that prior to the Industrial Revolution, the global carbon cycle was ‘steady state’ during the relatively brief (on geological time scales) period from 10 000 to 250 years before the present. There was a net flux of CO2 from the oceans to the atmosphere amounting to some 0·5 Pg C per year during this period. This flux was balanced by an equal flux of CO2 into organic matter in terrestrial biota, and thence back to the oceans in rivers. Approximately 1 Pg C (as inorganic carbon) flowed from high latitudes to low latitudes in the ocean. An equal flux of CO2 was conveyed through the atmosphere from warm, low-latitude regions to colder high-latitudes (Fig. 1). Approximately 0·2 Pg C as CO2 supplied to the atmosphere from vulcanism was incorporated into terrestrial biota and thence, via rivers, to marine sediments (and thus back to the C source for CO2 emitted in vulcanism): Watson & Liss (1998) (Table 1). The geochemical assumption of a steady-state carbon cycle for at least 10 000 years prior to the Industrial Revolution is predicated on the principle that a deviation from steady-state requires that at least one of three conditions must be met: (1) nutrients limiting primary production must be added to the ocean from external sources; (2) limiting nutrients in the euphotic zone that are unused on an annual basis must be consumed, and/or (3) the chemical composition of phytoplankton must change.
Table 1. . Carbon fluxes in the pre-industrial (up to approximately 1750 AD) part of the present interglacial. Adapted from Fig. 1 of Watson & Liss (1998). Note that, to ensure closure of cycles, the values differ slightly from those in the text
‘Bottom-up’ versus ‘top-down’ control of marine primary productivity
The standing stock of photosynthetic biomass in the world oceans is extremely small relative to terrestrial ecosystems. The estimated oceanic photosynthetic biomass of approximately 1 Pg C amounts to approximately 0·2% of the total photosynthetic biomass on Earth (Falkowski & Raven 1997; Field et al. 1998). The rapid turnover of this biomass is related to both the rate of production and the rate of removal. The balance between these two processes as a function of time can be described by the simple differential equation:
dN/dt = N(μ–m)
where N is the ensemble of photosynthetic biomass, μ is the specific growth rate, and m is the specific mortality. Both μ and m have units of (time)–1.
Any factor that limits the intrinsic physiological rate of growth, expressed as μ, is considered ‘bottom-up’ control. Such a factor may be a limiting resource (light, C, N, P, Fe, Si, vitamins, etc.) or a physical condition (thermal energy, osmolality). Such restrictions on the rate of biomass increase are termed ‘stress’ by Grime (1979). The history of bottom-up control can be traced to von Liebig (1840), with what is now termed ‘extent’ limitation (as in a batch culture). The concept of ‘rate’ limitation (as in a chemostat or turbidostat) is generally attributed to Blackman (1905) and Nathansohn (1908) (de Baar 1994; Falkowski 1994).
Primary production in the oceans is generally limited both in extent and rate by nutrient fluxes (i.e. ‘bottom-up’ control). Throughout most of the upper ocean, the concentrations of total fixed inorganic nitrogen and phosphorus are extremely low (< 1 mmol m–3). Vertical profiles of these two nutrients clearly reveal that fixed inorganic nitrogen is almost always depleted more rapidly than phosphate in the euphotic zone (Fanning 1992), and indeed, to a first approximation, fixed nitrogen limits phytoplankton biomass and growth throughout most of the world oceans (Falkowski et al. 1998; Downing 1997; notwithstanding).
The paucity of nitrogen is curious. In most temperate and tropical lakes, when fixed inorganic nitrogen concentrations become low, nitrogen fixation is stimulated. Consequently, phosphorus, which, unlike nitrogen, has no atmospheric source, almost always limits primary production in lakes (Hutchinson 1957). In the ocean, however, diazotrophic organisms are relatively sparse and confined to very few taxa. The lack of both numbers and diversity of nitrogen fixers in marine ecosystems has led to the notion that nitrogen limitation is itself limited by some other factor, the most probable being iron (Falkowski 1997a).
A major source of iron for the oceans is aeolian dust (Duce 1986). Ferric iron is virtually insoluble in seawater, and precipitates as hydrated oxides or phosphates. The concentration of iron throughout most of the euphotic zone in the world's oceans is approximately 1 mmol m–3 or less. Terrestrially derived iron, primarily from deserts, is delivered by prevailing winds, and provides the essential transition metal for phytoplankton growth (Blank, Leinen & Prospero 1985; Prospero & Nees 1985). Nitrogen fixers have an extremely high iron requirement, and in oceanic areas far removed from an aeolian source, iron fluxes cannot sustain high rates of fixation. The iron limitation of nitrogen fixation is particularly acute in the southern hemisphere, especially in the Pacific.
In three regions of the world oceans, namely the eastern equatorial Pacific, the subarctic Pacific, and the Southern (Antarctic) Oceans, iron limitation is so severe that photosynthetic electron transport is limited by the availability of the metal (Kolber et al. 1994; Behrenfield et al. 1996; P. Boyd, personal communication). In these three regions, iron limitation actually prevents the utilization of fixed nitrogen and phosphorus in the euphotic zone. Hence, the concentration of NO3– and HPO2–4 in these regions can be relatively high, approaching 30 to 50 mmol m–3 in some areas. Stimulation of primary production by the addition of iron to these so-called ‘high nutrient (N, P)-low chlorophyll’ regions of the ocean can significantly influence atmospheric CO2 influx (Martin, Gordon & Fitzwater 1990;Falkowski 1994; de Baar et al. 1995; Raven 1995; Coale et al. 1996; Cooper, Watson & Nightingale 1996; Pakulski et al. 1996; Falkowski 1997a; cf. Hart 1934; Harvey 1937).
Specific taxa are sometimes limited by specific elements. Diatoms are perhaps the most important taxon mediating the export flux of carbon in the oceans. This group of organisms uses silica to make opaline cell walls. Silica is supplied to the oceans from fluvial sources and the element often limits diatom production in coastal waters (Dugdale & Wilkerson 1998; Smetacek 1998). Vitamins have been inferred to limit dinoflagellates and other groups (Provasoli & Carlucci 1974). While these compounds and other specific elements, such as Cu, Co, Mo and Zn, are sometimes implicated as factors selecting specific phytoplankton taxa, they do not limit overall NPP in the oceans (Morel et al. 1994).
The mortality of photoautotrophs involves grazers and parasites, namely processes dependent on other trophic levels. Regulation of biomass via such processes is termed ‘top-down’ control. Mortality is essential to the recycling of nutrients, which permits the growth of the remaining photoautotrophic biomass; however, it cannot alter the maximum biomass supported by nutrient supply. Hence, while ‘top-down’ control plays an extremely important role in determining the structure of marine food webs and nutrient recycling efficiency (Banse 1992; Azam 1998), from a geochemical perspective, it plays a relatively unimportant role in regulating the net exchange of CO2 between the oceans and atmosphere.
Influence of increased atmospheric CO2 on marine primary productivity via increases in surface ocean inorganic carbon
With an atmospheric CO2 concentration of 365 μmol mol–1 (a typical value over the last decade), the equilibrium value for total inorganic carbon at the ocean surface is approximately 2 mol m–3 at 18 °C (the mean global sea surface temperature). Approximately 95% of the inorganic carbon in the ocean is in the form of bicarbonate anion; the equilibrium concentration of CO2 is only approximately 10 mmol m–3. This concentration of CO2 is about an order of magnitude lower than that required to saturate RuBisCO, and CO2 is frequently below air-equilibrium in regions with high levels of other (noncarbon) nutrients (Codispoti et al. 1982; Codispoti, Friedrich & Hood 1986; Watson et al. 1991a,b). Hence, in the absence of either an alternative carbon fixing pathway or a CO2 concentrating mechanism, marine phytoplankton (and macrophyte) photosynthesis at light saturation in air-equilibrated seawater would be limited by inorganic carbon (Raven 1970).
A large number of experiments have been conducted to examine potential limitation of photosynthesis by inorganic carbon in marine photoautotrophs. Most of the currently available data concern short-term (minutes – tens of minutes) measurements of photosynthetic rates as a function of inorganic carbon concentration for organisms grown at or near air-equilibrium CO2 levels in seawater enriched with other nutrients. Many of these data may not reflect organisms grown at air-equilibrium, since growth procedures frequently allow CO2 (and other components of the inorganic carbon system) to be depleted. Despite the difficulty in constraining all potential experimental variables, the outcome of these measurements is that the light-saturated or light-limited rate of photosynthesis is not limited by inorganic carbon in equilibrium with the atmosphere for all marine cyanobacteria, most eukaryotic microalgae and many eukaryotic macroalgae (Muñoz & Merrett 1989; Riebesell et al. 1993; Kübler & Raven 1994; Beer & Koch 1996; Kübler & Raven 1996; Hein & Sand-Jensen 1997; Raven 1997; Tortell, Reinfelder & Morel 1997; Beardall et al. 1998b). Virtually all marine embryophytes (i.e. seagrasses) are limited by inorganic carbon (Beer & Koch 1996; Raven 1997; Zimmerman et al. 1997; Beardall et al. 1998b). These results strongly suggest that changes in atmospheric CO2 will not (and historically, have not) directly affect overall photosynthetic rates in marine planktonic ecosystems.
The inorganic carbon system in the oceans is the primary pH buffer. This buffer is dependent upon the availability of alkaline earth cations, especially Ca2+. The precipitation of carbonate is an ancient pathway that has provided most of the 60 000 000 Pg of C as CaCO3 in the lithosphere. In the open ocean, primary producers, such as coccolithophorids and symbiotic foraminifera, are a significant source of CaCO3 in the form of calcite. In coastal waters, CaCO3-precipitating primary producers include symbiotic corals and foramenifera, as well as red, green and brown macroalgae, which primarily form aragonite. Aragonite is relatively easily dissolved in sediments, while calcite is generally well preserved.
Two key issues emerge regarding the biological flux of CaCO3 in relation to the carbon cycle. First, it is important to remember (Eqn 5) that the precipitation of CaCO3 increases the partial pressure of CO2 and, furthermore, the CO2 produced per unit CaCO3 precipitated increases as ambient CO2 increases (Frankignoule, Canon & Gattuso 1994). This is an example of a positive feedback of increasing CO2 level on CO2 production during calcification. However, as CO2 rises, pH declines. Calcification is highly pH sensitive, and essentially does not occur below c. pH 7·6. Hence, while calcification can theoretically produce high concentrations of CO2, it is self regulated via pH; this is an example of a negative feedback in a biogeochemical cycle. The factors controlling calcification in the ocean are very poorly understood.
THE OCEANS AS CO2 SINKS IN THE PRE-INDUSTRIAL PLEISTOCENE
Measurements of air trapped in ice-cores show that for at least two thousand years prior to the Industrial Revolution, atmospheric CO2 concentrations were approximately 280 μmol mol–1. This value appears to be typical of interglacial conditions over the last four glacial cycles. During glacial periods, atmospheric CO2 concentrations reached minima of approximately 180 μmol mol–1 (Jouzel et al. 1993; Raynaud et al. 1993). The causes of the 100 μmol mol–1 variations in atmospheric CO2 are unclear. While there is a consensus that the oceans sequester CO2 during glacial periods and are a source of CO2 during glacial–interglacial transitions, the mechanism(s) for the exchange are contentious, especially regarding the role of primary producers.
Other things being equal, enhanced sedimentation of organic carbon during glacial periods reflects an increase in ‘new’ primary productivity. The proxy data suggest that the upper ocean concentrations of NO3– and HPO42– during glacial episodes were elevated. Furthermore, there is clear evidence for increased aeolian inputs of iron (Martin et al. 1990; Falkowski & Raven 1997). The elevated aeolian iron fluxes suggest that primary productivity in the ‘high nutrient (i.e. NO3–, HPO42–)–low chlorophyll’ parts of the ocean would have been higher (Martin 1990; Falkowski 1997a; Falkowski & Raven 1997; Sunda & Huntsman 1997). The increased iron flux would have also stimulated biological N2 fixation (Raven 1988, 1990; Falkowski 1997a), thereby potentially enhancing the fixed nitrogen inventory of the oceans.
The effects of these changes in NPP can be quantitatively assessed using a box modelling approach (Broecker, Peng & Engh 1980). If the aeolian fluxes of iron were sufficient to top up the inventory of fixed inorganic N such that the N : P ratio conformed to the Redfield value, atmospheric CO2 would have decreased from 280 to approximately 245 μmol mol–1 (Figs 3a & b). Utilization of 30% of the nutrients in the Southern Ocean would have led to a further drawdown in CO2 to 190 μmol mol–1, the glacial minimum (see legend to Fig. 3). While these calculations do not prove that changes in oceanic NPP were responsible for the changes in atmospheric CO2, they clearly demonstrate their sensitivity to oceanic biological processes.
One line of evidence related to the role of primary producers in influencing glacial/interglacial CO2 concentrations in the atmosphere can be inferred from the isotopic 13C/12C ratio (δ13C). If the lowering of atmospheric CO2 at the onset of a glacial episode were a result of increased ‘new’ primary productivity, atmospheric 13CO2/12CO2 would be expected to increase. This prediction is based on the fact that marine photoautotrophs would preferentially assimilate 12C, thereby enriching the remaining inorganic carbon with the heavier isotope. Simple diffusive equilibration of the surface ocean with the atmosphere would lead to enrichment of the latter in 13C. Two independent lines of evidence indicate, however, that atmospheric CO2 was enriched with 12C during the last glacial episode. Leuenberger, Siegenthaler & Langway (1992) showed that samples of atmospheric CO2 derived from Antarctic ice cores had lower13C/12C ratios during the last glacial interval. Marino et al. (1992) used the 13C/12C of the C4 terrestrial plant Atriplex confertifolia taken from pack-rat middens of known age as a proxy for the 13C/12C of atmospheric CO2, and also inferred a lower13C/12C in atmospheric CO2 during the last glacial episode. These two data sets have been invoked to suggest that the oceanic ‘biological pump’ was less active in the glacial episodes, and that the drawdown of atmospheric CO2 was a result of changes in oceanic physics and/or physical chemistry (Keir 1992; Kerr 1992; Leuenberger et al. 1992; Marino et al. 1992; Raven 1992; François et al. 1997; Raven 1999).
This apparent paradox may be reconciled, albeit not yet quantitatively, by consideration of the solubilization of CaCO3. An enhanced flux of organic carbon to the ocean interior would stimulate respiration. A by-product of respiration is acidification, which promotes solubilization of CaCO3 in sedimentary particles (Archer & Maier-Reimer 1994). The liberated Ca2+ acts as a trapping agent for HCO3–, via inorganic carbon equilibria within the ocean and between ocean and atmosphere, account for the observed decrease in atmospheric CO2 levels. This sequence of reactions is described schematically by:
CO2 + 2H2O → (CH2O) + O2 + H2O
(primary production) (increases pH)
CH2O + O2 + H2O → CO2 + 2H2O
(respiration) (decreases pH)
CO2 + H2O + CaCO3→ 2HCO3– + Ca2+
(calcium carbonate dissolution)
(CaCO3 + H2O + CO2→ Ca2+ + 2HCO3– )
(net reaction of above)
This carbonate intermediate pathway is quantitatively consistent with the observed decrease in calcite preservation in marine sediments during glacial periods as well as consistent with evidence of increased total, and exported marine primary productivity. Such evidence includes an increased benthic-to-planktonic gradient in δ13C, and higher sedimentary organic carbon concentrations and burial rates.
Further evidence which is generally consistent with this hypothesis relating the decreased CO2 concentration in the ocean surface-waters (and hence atmosphere) to an increased alkalinity and pH of the ocean comes from the δ11B work of Sanyal et al. (1995). These workers studied the natural abundance 11B/10B ratio of B(OH)4– in fossil foramenifera skeletons as a measure of palaeo-pH of seawater. In seawater, boron exists as boric acid (B(OH)3) and borate ion (B(OH)4–). The δ11B ratio is 19‰ higher in B(OH)4– than in B(OH)3 as a consequence of a thermodynamic isotope effect. The ratio of B(OH)4– : B(OH)3 increases with increasing pH. Hence a higher δ11B indicates a higher pH. The δ11B data indicated a seawater pH that was 0·3 ± 0·1 (deep water) and 0·2 ± 0·1 (surface) units higher during the last glaciation than it is today, which could account for the observed difference in atmospheric CO2 levels, albeit with some ‘implications that are difficult to accept’ (Sanyal et al. 1995). Despite these problematic implications, this model does predict the direction (lower, i.e. more negative), if not necessarily the extent, of the change in the δ13C of atmospheric CO2 in the last glacial episode relative to today.
OCEAN INORGANIC CARBON IN RELATION TO THE INDUSTRIAL REVOLUTION AND GLOBAL ENVIRONMENTAL CHANGE
Clearly, in the contemporary ocean, the assumption of steady-state does not apply (Falkowski et al. 1998). Increasing CO2 inputs into the atmosphere from fossil fuel burning, deforestation and cement manufacture currently amount to some 7 Pg C per year (Table 2). The pre-industrial 0·5 Pg C per year CO2 net flux from ocean to atmosphere has been replaced by an atmosphere to ocean net CO2 flux of up to 2 Pg C per year. The pre-industrial 0·5 Pg C per year net CO2 flux from atmosphere to terrestrial biota is increased to a net flux of approximately 1·5 Pg C per year (Houghton et al. 1996; Bigg 1996). Thus, approximately 30% of the anthropogenic CO2 (i.e. some 100 Pg C) has been taken up by the oceans between 1850 and 1996.
Most of the CO2 absorbed by the oceans is a consequence of direct solubilization due to increased partial pressure of the gas in the atmosphere. This flux can be estimated from tracing radiocarbon released to the atmosphere as a consequence of thermonuclear explosions and ultimately equilibrating with the ocean (Broecker & Peng 1982), as well as very high precision analyses of the concentration of total inorganic carbon in seawater over the past several decades (Sabine, Wallace & Millero 1997).
The interaction between biological and physical processes in the ocean produces a significant ocean–surface–atmosphere disequilibrium with respect to CO2. The net invasion of CO2 results in the transfer of approximately 1 Pg C per year from low-latitude to high-latitude surface ocean waters (Watson & Liss 1998; Table 2; Fig. 2). Note that this gradient is the opposite of that prior to the Industrial Revolution. Thus, CO2 uptake as a result of the CO2 solubility effect is a decrease in net CO2 evolution from the low-latitude ocean and an increased net CO2 uptake by the high latitude ocean. The rate of physico-chemical uptake of CO2 by the ocean is reduced as ocean temperature increases. Such warming has apparently occurred over the 1850–1996 time interval (Houghton et al. 1996). The solubilization effects are complicated by any influence that anthropogenic climate change has had on oceanic circulation, and thus on the atmospheric CO2 transfer described (Watson & Liss 1998). While major effects on global ocean circulation do not seem to have occurred between the beginning of the Industrial Revolution and today (Broecker 1997), they almost certainly will occur in the coming centuries (Sarmiento et al. 1998).
Human activities since 1750 have increased not only atmospheric CO2, but also the riverine and atmospheric transfer of HPO42–, NOx, NHy, organic N and transition metals (Cornell, Rendell & Jickells 1995; Berner & Berner 1996; Falkowski & Raven 1997; Raven & Yin 1998). Much of this additional nutrient input, with a possible increase in global primary productivity of up to 0·1 Gt C per annum, involves coastal waters (Walsh 1988), and so would not have been evident in the work of Falkowski & Wilson (1992, 1993) comparing oceanic phytoplankton biomass and primary production over the last century. At all events there is evidence of an increased input of potentially limiting resources to the ocean which could account for increased organic carbon sedimentation. A less significant effect, discussed earlier, is the variable stimulation of photosynthesis and of primary productivity in the ocean as a result of increased CO2 levels over the last 250 years.
WHAT WILL HAPPEN WITH EXPECTED FURTHER INCREASES IN ATMOSPHERIC CO2 AND SURFACE OCEAN TEMPERATURE?
Given ‘business as usual scenarios’ used by the Intergovernmental Panel on Climate Change (IPCC) to examine climate forcing (Houghton et al. 1996) atmospheric CO2 will have doubled from the pre-Industrial Revolution concentration by the middle of the 21st century. The solubility of CO2 in the surface ocean via the solubility pump could be partly offset by the predicted higher temperatures. Moreover, the sequestration of dissolved inorganic carbon in the ocean interior will probably be reduced as a consequence of increased stratification of the upper ocean (Sarmiento et al. 1998). Further increases in nutrient inputs from the land to seas could increase new production and thus, potentially, organic carbon incorporation into sediments, leading to drawdown of atmospheric CO2 via the biological pump; however, there are caveats regarding the efficiency and magnitude with which increased nutrient fluxes impact on the carbon cycle. Primary amongst these is that, despite coastal eutrophication, denitrification on continental margins often keeps apace of nitrogen inputs, such that the net change of fixed nitrogen in the oceans from anthropogenic sources is extremely small, accounting for approximately 0·2 Pg C per annum (Walsh 1991). Changes in that flux are difficult to predict, but are not likely to lead to a large sink for anthropogenic CO2 in the coming century. Even greater problems ensue if atmospheric changes alter climate sufficiently to cause very markedly ‘non-linear’ responses, for example, to ocean circulation (Broecker 1997; Falkowski et al. 1998).
WHAT ARE THE PROSPECTS FOR INTERVENTION IN INCREASING THE ROLE OF THE OCEANS IN DISPOSING OF ANTHROPOGENIC CO2?
As anthropogenic emissions of CO2 increase, there will be increased pressure to develop methods of enhancing sinks or reducing the source term. For the former, it has been proposed that primary productivity could be increased by deliberate addition of a limiting nutrient to a part of the ocean (Martin et al. 1994). For the latter, it has been suggested that non-biological, engineering approaches be used for trapping the CO2 from major point sources (e.g. fossil fuel electricity generating stations) and depositing it directly in the deep ocean, where oceanic circulation would prevent its return to the atmosphere for decades or centuries (Haughan & Drange 1992; Orr 1992). These two groups of methods will be considered in turn.
In principle, any of the nutrients (other than CO2!) which limit primary productivity in different areas of the ocean could be deliberately added to the appropriate location. This would encompass inorganic nitrogen (NH4+ or NO3–), HPO42–, Fe, possibly Zn and (for diatoms) Si. In practice, the method is most conveniently applied to nutrients required in small quantities relative to carbon (i.e. trace elements). This approach was suggested by the late John Martin (Martin, Gordon & Fitzwater 1990; cf. Hart 1934; Harvey 1937) for Fe, which is a limiting resource for phytoplankton growth in two areas of the Pacific Ocean and the Southern Ocean. There are reservations about the large-scale use of this method (Fuhrman & Capone 1991). The large increase in organic carbon export, which would follow large-scale fertilization, could lead to anoxia in the deeper ocean. Anoxia could, in turn, enhance both methanogenesis and denitrification. The latter would lead to the production of N2O. Subsequent outgassing of both methane and N2O would increase the greenhouse effect to a greater extent than the decrease produced by the drawdown of CO2. At present atmospheric levels, a given absolute change in CH4 and N2O alters the greenhouse effect orders of magnitude more per molecule than CO2.
With such cautionary observations in mind, small-scale experiments have been conducted in the Eastern Tropical Pacific, namely, IRONEX I and IRONEX II. Here FeSO4, together with SF6 as an inert tracer, was added to 65 km2 areas of ocean to yield final concentrations of iron of approximately 2 to 4 μmol m–3. Lessons from the first experiment were used in designing and executing the second experiment, which yielded a very significant increase in phytoplankton primary productivity and decrease in ocean surface dissolved CO2 concentration (Kerr 1994; Coale et al. 1996; Cooper, Watson & Nightingale 1996). However, as was anticipated, the iron effect only lasted a few weeks as a result of the oxidation and precipitation of the iron, and its physical advection and diffusion in the ocean. Consequently, large-scale CO2 drawdown would need continuous Fe addition. Presumably this problem could be overcome with engineering and technical resources; however, estimates of increased carbon sequestration resulting from Fe fertilization need to be better constrained.
The rationale of non-biological intervention is to trap the CO2 from major point sources, such as electricity-generating stations which are major consumers of fossil fuels, and transfer the CO2 to the deep ocean (see Haughan & Drange 1992; Orr 1992). The original suggestions involved the deep injection of CO2 as a gas, a liquid, or a highly concentrated aqueous solution. If an appropriate site was chosen, then the CO2 should not come back to the surface for a century or more. However, most such sites are remote from power stations, so that the mechanics of such CO2 disposal would be very complex (Orr 1992).
A subsequent suggestion (Haughan & Drange 1992) is much less logistically complex. Haughan & Drange (1992) argue that injection at approximately 150 m of CO2 gas under pressure would convert 99% or more of the injected CO2 into a solution of CO2 in seawater. The solution is denser than the surrounding normal seawater and, under appropriate conditions, could sink and thus mimic the results of the much more complex deep injection. The prerequisites for the shallow injection procedure are that the water movement regime is adequate to cause solution of all of the added CO2, but not so great as to prevent sinking of the CO2-containing water. Furthermore, even the shallow injection process has a significant energy cost; Orr (1992) suggests that even the removal of CO2 from the power station exhaust and its compression would derate the station by 35% and double its construction costs, without allowing for the costs of movement to an appropriate shoreline.
Since injection of such quantities of CO2 would lower the pH of seawater locally to 4·0, either deep or shallow CO2 injection could have deleterious effects on biota near the site of injection (Haughan & Drange 1992; Orr 1992). In this respect, the effect would mimic the T–K boundary bolide impact (O’Keefe & Ahrens 1989). The shallow injection option would have more obvious impacts on benthic biota. It should be noted, however, that the CO2-enriched seawater would interact with CaCO3 in sediments with which it came into contact, thus increasing the alkalinity of the water (Orr 1992). In the extreme case, this might even make the originally CO2-enriched deep water act as neither a sink nor a source for atmospheric CO2 rather than a source when it is eventually upwelled, provided all of the injected CO2 has been titrated by dissolution of CaCO3 according to Eqn 7.
1. Oceans dominate the global C cycle over 101–106 and more years.
2. Glacial drawdown of atmospheric CO2 in the Pleistocene mainly involved the ocean. This was due at least in part to increased phytoplankton primary productivity and organic C sedimentation related to (for example) increased aeolian Fe input to the Southern Ocean. Decreased ocean CaCO3 precipitation, with increased soluble alkalinity, could have had a role.
3. Oceans have sequestered up to 30% of the additional CO2 emitted to the atmosphere since the start of the Industrial Revolution. This is due in part to the solubility effect from increased atmospheric CO2, and in part to increased primary productivity and organic C sedimentation.
4. Responses to further anthropogenic CO2 inputs will follow those outlined in (3), provided that global environmental change does not massively alter the oceanic thermohaline circulation. Human intervention by Fe fertilization of phytoplankton Fe-deficient areas of the ocean, and direct CO2 burial in the deep ocean, could increase the C sequestration in the ocean but with an environmental cost.
Work in J.A.R.'s laboratory on inorganic carbon acquisition by marine algae and its interaction with other environmental factors is supported by the Natural Environment Research Council (UK) and the Scottish Office Agriculture, Environment and Fisheries Department. P.G.F. is supported by the National Aeronautics and Space Administration, the US Department of Energy and the Office of Naval Research.