The Southern Ocean Biological Pump

Contemporary scientific issues related to the atmospheric accumulation of greenhouse gases and their effects on global environmental change demand a comprehensive understanding of both the carbon cycle and global ecology. Consequently, detailed in situ investigations of terrestrial and marine microbial, plant and animal populations are necessary prerequisites for developing a predictive capability for global environmental change. To be useful, these investigations need to address broad questions regarding the distributions, abundances, diversity and controls of microbial, plant and animal populations, their effects on the chemical composition of the surrounding habitats and their response to physical perturbations. Ideally, these field studies should be conducted at strategic sites that are representative of larger regions. Furthermore, these ecological studies should be conducted for at least a decade, and preferably longer, in order to distinguish natural variability from that induced by human activities.

The global inventory of carbon is approximately 4-5 x 10¹⁹ g (40-50 x 10³ GT), distributed unequally among four active reservoirs: atmosphere (1.7%), terrestrial (4.5%), geological (8.9%) and oceanic (85%). The small size of the atmospheric reservoir and the large size of the oceanic reservoir are both of great importance in issues related to global climate and environmental variability. Since the beginning of the industrial revolution, humankind has been returning to the atmosphere carbon that accumulated over millions of years in fossil fuels and carbonate minerals, and is conducting an uncontrolled, global scale biogeochemical experiment. Since 1800, the atmospheric inventory of CO₂ has increased by about 30%, and will continue to rise as world population increases. This perturbation in the global carbon cycle is likely to result in a redistribution of mass among the active reservoirs and attendant alterations in exchange rates and, perhaps, exchange processes. However at the present time we have only a rudimentary understanding of the rates of transfer and the mechanisms that comprise the global carbon cycle.

The large and dynamic oceanic reservoir of carbon, approximately 4 x 10¹⁹ g distributed unequally among dissolved and particulate constituents with various redox states, plays an important role in global biogeochemical cycles. The two largest pools are dissolved inorganic carbon (DIC = [H₂CO₃] + [HCO₃^-] + [CO₃⁼]) and the less oxidized pool of mostly uncharacterized dissolved organic carbon (DOC). A chemical disequilibrium between DIC and organic matter is produced and maintained by numerous biological processes. The reversible, usually biologically-mediated interconversions between dissolved and particulate carbon pools in the sea collectively define the oceanic carbon cycle.

Primary conversion of oxidized DIC to reduced organic matter (dissolved and particulate pools) is generally restricted to the euphotic zone of the world ocean through the process of photosynthesis. The supply of reduced carbon and energy required to support subeuphotic zone metabolic processes is ultimately derived from the upper ocean and is transported down by advection and diffusion of dissolved organic matter, gravitational settling of particulate matter and by the vertical migrations of pelagic animals and phytoplankton. Each of these individual processes, collectively termed the "biological pump" is controlled by a distinct set of environmental factors and, therefore, the relative contribution of each process may be expected to vary with changes in habitat or with water depth for a given habitat. Three components of the biological pump, each representing a separate set of ecological processes have recently been defined. The rotary pump circulates materials through the microbial food web, the Archimedian pump defines the gravitational flux of fecal pellets and aggregated materials and the reciprocating pump represents the daily bi-directional migration of animals in response to light. For open ocean ecosystems, the relative contributions of these processes are poorly known. Although the Archimedian pump is generally assumed to dominate total euphotic zone export, the role of yet another component, the diffusion pump, may also be important in selected habitats. The rates at which the individual components of the biological pump operate are under the control of both physical (light, temperature, turbulence) and biological (species composition, growth rate, food web structure) controls.

Each year, the biological pump removes an estimated 7 GT C (1 GT = 10¹⁵ g) from the surface waters of the world ocean, a value that is equivalent to ~ 15% of the annual global ocean primary production. Microbial transformation of sinking particles in the thermocline that gives rise to increased C:N and C:P ratios with depth can potentially drive a net atmosphere-ocean flux of CO₂. Episodic flux "events" carry to the deep sea large amounts of "fresh" organic matter.

The role of the ocean as a net sink in the global carbon cycle is dependent largely upon the balance between the export flux of planktonic primary production and the rate of dissolved inorganic resupply by upward eddy-diffusion processes. When particulate export is expressed as a percentage of contemporaneous primary production, this value is termed the export ratio. Results from broad-scale, cross-ecosystem analyses suggest that the export ratio (generally measured/reported only as the Archimedian component of total export) in oceanic habitats is a positive, non-linear function of total integrated primary production, with values ranging from less than 10% in oligotrophic waters to greater than 50% in productive coastal regions. It should be emphasized, however, that the field data from which the existing export production models were derived are extremely limited and that open ocean and high latitude habitats, in particular, are underrepresented. Because most global ocean primary and export production occurs in oceanic habitats, it is important to understand the mechanisms that control the biological pump so that we can make accurate and meaningful predictions of the response of the oceanic carbon cycle to global environmental change.

Marine microbes have a major impact on both local and global environments. In addition to their primary metabolic activities, many species produce, or consume, growth-stimulating and growth-suppressing organic compounds (e.g. vitamins, organic toxins), while others emit "greenhouse" or ozone-destroying gases (e.g. methane and nitrous oxide) as normal by- products of cellular metabolism. Complex interactions and nutritional interdependencies of organisms occur in the marine environment.

Superimposed on these integrated metabolic processes are numerous physical and chemical interactions that define the complex and temporally variable marine habitat. Some of the higher frequency temporal variability is predictable (e.g., diurnal, tidal, seasonal) but other processes (e.g., storms, volcanic eruptions) are stochastic. On longer time scales (e.g., interannual to centuries), coupled ocean-atmosphere interactions and other global processes can cause habitat variability and provide a mechanism for biodiversity or evolutionary change. Such is the complex nature of microbial life in the sea.

Of all the marine habitats investigated to date, those in the Southern Ocean are among the least well understood. Although the antarctic marine environment (including all ocean, sea ice and island components south of the Antarctic Convergence zone) is one of the largest ecosystems on Earth (36 x 10⁶ km²), it is undersampled relative to other more accessible locations. Furthermore, the extreme variability in climate, solar radiation and sea-ice extent yields a physically- and biologically-variable habitat that is difficult to fully appreciate from a single expedition. Repeated observations over several years and during all seasons, and comprehensive synoptic assessments of ocean circulation, chemistry and biology are required to elucidate ecosystem structure and function.

The Palmer Long-Term Ecological Research (Pal-LTER) program was established in 1990 at Palmer Station, Antarctica as an interdisciplinary study to seek a general understanding of ecosystem processes and to model the interactions between key groups of organisms and the physical environment. For the past several years, one component of the Pal-LTER study ("Microbial Dynamics and Carbon Flux," D. Karl, P.I.) has focused on microbiological oceanography in the coastal shelf waters west of the Antarctic Peninsula and within the seasonal ice pack. The central hypothesis of the Pal-LTER program is that the annual advance and retreat of sea ice is a major physical determinant of spatial and temporal changes in the structure of antarctic ecosystems. It is well known that sea ice can influence the timing and magnitude of primary production, krill recruitment and reproductive success of apex predators. Sea ice also provides one of the major habitats for microorganisms and certain food webs are entirely ice-associated.

Since 1992 we have had a year-round sediment trap mooring deployed at a representative site in the Pal-LTER study area (64°30'S, 66°W). This autonomous experiment allows us to measure seasonal and interannual variations in particle export and to ascertain controls on the Southern Ocean biological pump. We fully anticipated a strong seasonal phasing to particulate matter export in this Antarctic coastal environment. We also expected a large dynamic range in export from the large, phytoplankton bloom-supported peak export processes in summer to the low wintertime fluxes. We did not realize at the start of this project that both these values would represent global maxima and minima, respectively.

Two of the more interesting and unexpected results of this initial data set are: (1) the extremely low implied values for annual primary production based on the time-integrated particulate matter export and (2) the unusual, non-Redfield molar C:N:P stoichiometry (high C relative to N and P) of the summertime particulate matter export pulse. It is important to distinguish between gross particulate carbon export (measured by sediment traps) and net carbon export (the difference between gross export and upward advection/diffusion of dissolved inorganic carbon, DIC). A finite rate of export production is a necessary, but not sufficient, condition for a net flux of carbon out of the surface ocean. If the elemental ratios (C:N:P) are the same in the upward inorganic flux and in the organic downward flux, there is no net export of carbon, and no potential for sequestration of carbon dioxide. The distinct stoichiometric uncoupling of the particulate matter export each summer with much greater than Redfield (C:N and C:P) export is consistent with a net carbon export from the euphotic zone.

Figures to accompany summary

Fig. 1: Outline of presentation

Fig. 2: Self-explanatory justifications for establishing a long-term project in Antarctica

Fig. 3: Two examples of how global climate change may affect Southern Ocean ecosystems

Fig. 4: Oceanic food web showing the traditional nano- and microplankton (algae) based food chain on the left and the microbial-based food web processes, supported by the large pool of dissolved organic matter (DOM), on the right. Both components contribute to the export of organic matter as particulate matter flux from the euphotic zone.

Fig. 5: Inventory of specific carbon pools and carbon fluxes currently measured in the Palmer LTER program.

Fig. 6: Schematic representation of ice-associated biological carbon pumps based on unique biological [top] and physical-chemical [bottom] characteristics of these regions.

Fig. 7: Pal-LTER study area along the western Antarctic Peninsula region. The solid circles are the standard hydrostations where samples are routinely collected. The contours are water depth, in meters.

Fig. 8: Chronology of near surface water [symbols] and atmospheric [line] partial pressure of carbon dioxide (pCO₂) along a cruise track from Punta Arenas to the LTER study area and back again for 1995. The surface water supersaturations (symbol > line) indicate regions where deep waters are upwelled and the surface water undersaturations (symbol < line) indicate regions where the biological carbon pump has removed carbon dioxide faster than gas exchange processes can resupply it. Most of the Pal-LTER study area has lower than expected carbon dioxide values in the surface ocean, supporting the claims of high rates of biological activity during austral summer periods.

Fig. 9: A comparison of the partial pressure of carbon dioxide [center] and dissolved oxygen [bottom] concentrations versus date of sampling (1997) along the cruise track shown in the upper panel. The numbers in the upper panel indicate the Julian day when those data were collected. The strong negative relationship between carbon dioxide and oxygen is consistent with biological activity (i.e., net primary production).

Fig. 10: Sea ice coverage in the LTER study area as a function of season [top] and year [bottom]. From: Stammerjohn and Smith (1996).

Fig. 11: Satellite-derived percentage ice cover for the region of the Palmer LTER study area corresponding to the location of the sediment trap mooring. Both the extent (total coverage) and timing (rates of accretion and ablation) of ice cover vary during this period of observation, with 1994 corresponding to the most ice-extensive winter and the most extended period of ice cover.

Fig. 12: Particulate carbon flux [top] and log particulate carbon flux [bottom] versus time at the Palmer LTER sediment trap station. The annual flux patterns, though similar with maxima in spring-summer and minima in winter, also display substantial differences, including: (1) interannual variations in the timing of the peak export (early Nov for the 1992-93 austral summer versus late Jan for the 1993-94 period), (2) interannual variations in peak export (largest export following the ice-extensive 1994 winter) and (3) order of magnitude interannual variations in winter export with lowest fluxes during the period of most extensive ice cover (e.g., 1994 winter; see bottom figure). These export fluxes are also characterized by very rapid order of magnitude changes in particulate carbon flux between consecutive collection periods, especially at the beginning and at the end of the peak export periods.

Fig. 13: C:N [top], N:P [center] and C:P [bottom] elemental ratios (mol/mol basis) for sinking particulate matter collected in bottom-moored sediment traps at the Palmer LTER sediment trap station. The solid line indicates the Redfield Ratio stoichiometry of C₁₀₆:N₁₆:P₁. For all three reference determinations, there is a significant decoupling with elevated C:N, N:P and C:P ratios each summer period. The non-Redfield C:N / C:P export of particulate matter has implications for the net export and sequestration of carbon dioxide.

Fig. 14: Summary of primary production and particulate carbon flux estimates for samples collected in the Palmer LTER study region west of the Antarctic Peninsula (shown here as summer Nov-Feb vs. winter Jun-Aug extrema and annual average values). The two independent estimates of primary production, based on export production models and chl a distributions, are more than an order of magnitude different for the LTER region. The coastal Antarctic data sets are compared here to similar data collected by us at the Hawaii Ocean Time-series (HOT) North Pacific ocean subtropical gyre Station ALOHA. By comparison to the HOT annual average, the Palmer LTER data set suggests lower primary production and lower export and may qualify this coastal Antarctic shelf environment as the global minimum for both of these important ecosystem rates.