U.S. JGOFS Synthesis and Modeling Project

27 July - 2 August 1997

Meeting Overview

Scott C. Doney

Oceanography Section, NCAR, Boulder, CO 80307


Jorge L. Sarmiento

AOS Program, Princeton University, Princeton, NJ 08544-0710

Printed copies of the full report are available from the U.S. JGOFS Planning Office. Contact Mary Zawoysky at mzawoysky@whoi.edu or (508) 289-2834 if you would like a copy.


The U.S. JGOFS summer 1997 Synthesis and Modeling Workshop at Snowbird, Utah highlighted the Hawaii (HOT) and Bermuda (BATS) JGOFS ocean time-series stations and their role, along with the process studies, in generating new paradigms for pelagic marine biogeochemistry. Major topics addressed at the meeting included multi-element (Fe, N, Si, P) controls on phytoplankton production, the magnitude and effect of nitrogen fixation, the links among community structure, export flux and remineralization, and the impact of spatial and temporal variability on ecosystem structure. Two underlying questions were emphasized throughout the meeting to help focus the discussion, namely: what are the potential biogeochemical responses to climate variability and change, and how do we combine observational and modeling efforts to improve our conceptual understanding and predictive capabilities.


The U.S. JGOFS field work began in the Fall of 1988 with the inception of two ocean biogeochemical time-series stations near Hawaii (HOT) and Bermuda (BATS), and the programs are currently nearing their tenth year anniversary. The HOT and BATS stations have played a crucial part in improving our understanding of ocean biogeochemistry over the last decade as well as offering unique opportunities to develop and refine new techniques and technologies to monitor long-term, secular changes in the ocean. The time-series stations will likely continue in operation beyond the end of the JGOFS program, forming one of U.S. JGOFS' enduring legacies. It is fitting then that the first scientific summer workshop of the U.S. JGOFS Synthesis and Modeling Project (SMP) should focus on the time-series aspects of both the HOT and BATS stations, various process studies, primarily EqPAC and Arabian Sea, and other relevant time-series programs such as the Palmer LTER (Smith et al., 1995).

The overall goal of the U.S. JGOFS SMP is to develop models with:

"predictive skill for the partitioning of carbon between the atmosphere-ocean and ocean reservoirs"

on time and space scales, seasonal to centennial and basin to global, that are generally larger than encompassed by traditional process studies. From the beginning of JGOFS, the time-series stations were envisioned as a key component by providing information on the biogeochemical response of the ocean on annual to inter-annual scales. Shorter time-series of one to two weeks duration have also been incorporated into a number of the process studies to study dynamics on the diurnal to event scale.

The time-series workshop was intended to highlight recent scientific advances and to foster improved interactions between observational and modeling efforts. The meeting was organized, therefore, around the following set of questions:

-What processes are important for predicting ocean carbon dynamics?

-How will processes be modified under potential climate change?

-How should those processes be modeled and are proper dynamics included in models?

-What are the major problems with current models?

-How do we make progress from here?

The invited speakers and panel leaders (see agenda, page 83) were each requested to fold these questions into their presentations and/or discussions.

Following a number of introductory presentations, the meeting progressed on the first three and a half days (see agenda, page 83) through a series of invited overview talks and panel discussions on euphotic zone production and remineralization, export production, particle transport and remineralization and finally physical forcing and numerical models. The meeting participants then divided into three working groups to focus in more detail on issues related to nutrient limitation, export and remineralization, and unresolved processes. Meeting participants were invited to bring posters on individual research topics, showcased at an evening poster session. An informal group also met to discuss the technological challenges of mounting southern ocean time series. The three working group reports are included below followed by the individual abstracts from the overview talks, panels, and posters.


By their nature, time-series data sets typically record the temporal variation in a limited number of standing stocks (e.g., chlorophyll; dissolved inorganic carbon) and fluxes (e.g., primary production; sediment trap particle flux) at a single point in space on a range of time and forcing scales from the event or synoptic scale out to the seasonal cycle and inter-annual variability. The underlying dynamics of the system are then often inferred from the data using a combination of empirical and numerical modeling techniques.

Time-series stations have, therefore, been an attractive framework for the development and refinement of numerical models, which are often cast as a set of time-evolving, prognostic (predictive) equations. This approach, however, can be plagued by problems associated with under-resolution of temporal/spatial variability about the time-series station and/or the lack of key dynamical processes in the data and models as discussed below. Because of their supporting infrastructure and long-term data record, time-series stations also provide an excellent site for the testing of new hypotheses (e.g., eddy pumping) and new technologies.

The meeting began with an overview talk by T. Michaels on the time-series programs and some of the scientific highlights to date (e.g., Karl and Michaels, 1996 and references in Deep-Sea Research Time-series volume). The historical antecedents of the time-series stations can be traced to the pioneering work on the seasonal cycles of primary production, nutrients, and mixing at Hydrostation S in the late 1950's/early 1960's (e.g., Menzel and Ryther, 1960) and the sediment trap measurements of the VERTEX program (Martin et al., 1987) among others. The BATS and HOT stations are sampled at approximately monthly intervals using nearly identical techniques (Karl and Lukas, 1996; Michaels and Knap, 1996). Thorough methods and sampling descriptions and data are available on-line via the world wide web:

BATS http://www.bbsr.edu/bats/bats.html; and

HOT http://hahana.soest.hawaii.edu/hot/hot.html.

The observational procedure and measurement suite has evolved slightly with time and is only briefly summarized below. Over a two to three day occupation, the upper water column physical structure is measured by repeat CTD casts and discrete water samples are collected for the analysis of nutrients, oxygen, dissolved inorganic and organic carbon, chlorophyll and pigments, and particulate organic material. Dawn-to-dusk, in-situ C-14 primary productivity incubations are also completed at a set of standard depths, and bacterial production rates are also measured using radio-labeling techniques. The sinking particle flux below the base of the euphotic zone is measured using short (2 to 3 day) deployments of floating sediment traps. Underway systems (e.g., pCO2) collect data on the transits to and from the stations. Both programs have also carried out survey cruises about the time-series station locations to study how representative the stations are of a larger domain and to explore specific hypotheses.

The time-series programs also form the framework for a wide variety of ancillary measurement programs ranging from ocean optics (Siegel et al., 1995; Siegel and Michaels, 1996) to biogenic gases (Schudlich and Emerson, 1996). At Bermuda, related time-series of biogeochemical relevance include Deuser's deep sediment trap array (e.g., Deuser and Ross, 1980; Deuser, 1986), the ROLAID bottom lander data set (Sayles and Martin, 1995) and the AEROCE atmospheric tower. Surface mooring systems have recently been established at both sites including meteorological packages and subsurface instrumentation (Dickey et al., 1997).

To place the JGOFS time-series results in perspective, one must recall the biogeochemical paradigms prevalent in the late-1980's. At the start of JGOFS, nitrogen was thought to be the major nutrient limiting primary production throughout most of the ocean, and most conceptual models of the Nitrogen cycle (Dugdale and Goering, 1967) centered on the input of new nitrogen as NO3 via vertical mixing, recycling and regeneration in the euphotic zone through NH4 and small molecular weight organic nitrogen compounds, and export via sinking particles. The importance of both microzooplankton grazing (Frost, 1991) and the microbial food web (Azam et al., 1983) were recognized but not fully explored, and the general biogeochemical framework at the time can be seen in the seminal modeling work of Fasham et al. (1990). The idea of iron limitation had been advanced (Martin and Fitzwater, 1988) but remained a hypothesis, there was revived interest but few high quality measurements of dissolved organic carbon (DOC) (Sugimura and Suzuki, 1988), and geochemical estimates suggested much higher new production rates for the subtropics than could be supported from most production estimates (e.g., Jenkins and Goldman, 1985).

The subsequent decade has seen rapid progress on a number of biogeochemical fronts from research both within and outside of JGOFS. Advances include the dramatic results of the IronEx experiments (e.g., Coale et al., 1996), the refinement of DOC measurement techniques and discovery of its contribution to the export flux (e.g., Carlson et al., 1994), and the increasing awareness of the role of nitrogen fixation in the subtropical North Pacific (Karl et al., 1997) and North Atlantic (Michaels et al., 1996). Results from the North Atlantic Bloom Experiment reaffirmed the importance of understanding mesoscale variability when interpreting biogeochemical data even in the open ocean (McGillicuddy et al., 1995).

Meeting Overview

The results of the meeting can be broadly grouped into three main themes, which are also reflected in the working group topics, namely:

-Multiple elements (N, P, Si, Fe ...) are involved in limiting autotrophic production, and these controlling factors vary from site to site and with time.

-Community structure has a strong impact on carbon export and remineralization, and the bulk of the particle export is derived from large-cell, diatom based plankton assemblages.

-Physical variability plays a key role in disrupting tightly coupled, steady-state systems through the injection of limiting nutrients; the enhanced new production and modified community structure of the subsequent transient system support large, episodic export events.

Although few of the concepts presented at the meeting are truly novel or revolutionary, the meeting highlighted growing appreciation for the linkages among nutrient limitation, community structure, and physical variability.

Multi-element Controls

The role of iron as a limiting nutrient in the high-nutrient, low chlorophyll regions of the equatorial Pacific has been confirmed through incubation studies and the IRONEX purposeful release experiments, which in IRONEX2 produced a significant phytoplankton bloom and drawdown of surface nutrients and pCO2 (Coale et al., 1996; Landry et al., 1997). P. Falkowski pointed out that the phytoplankton physiological response to iron, primarily an increase in the quantum efficiency, occured rapidly within hours of the iron injection and that the iron response was observed across the phytoplankton size spectrum.

The natural HNLC system is characterized by a close balance between phytoplankton growth and grazing losses, and the addition of iron, either artificially or from physical mixing events, leads to periods of unbalanced growth where the grazers can not keep up with the increasing growth particularly of the larger diatoms (D. Barber). Particle export appears to be driven by the periods of unbalanced growth and a shift towards a diatom dominated ecosystem due to episodic limiting nutrient input, in this case iron. Most of the iron required to fuel export production in the Equatorial Pacific appears to be supplied from below rather than from atmospheric deposition, and a number of Fe based 1-D modeling studies were presented in the poster session. The question of silica regulation of HNLC regimes was also raised following recent work of Dugdale and Wilkerson (1998).

For the subtropical oceans, recent evidence suggests that nitrogen fixation may contribute a substantial fraction of the annual new supply of nitrogen to the oligotrophic gyres (Michaels et al., 1996; Capone et al., 1997), up to half of the particle nitrogen export flux at the HOT site during some years (Karl et al., 1997). Nitrogen fixation is an energy intense process requiring high light (stratified) conditions and iron. At the HOT site, elevated abundance of Trichodesmium and prevalence of the nitrogen fixation biogeochemical signals are related to an extended period of ENSO like conditions from 1992 on with weaker winds and increased stratification (Karl et al., 1997). A corresponding shift from a nitrogen to a phosphorus limited ecosystem is observed at HOT. Active biological transport of nutrients from below the euphotic zone may be a crucial biogeochemical process but is poorly characterized (Villareal et al., 1996).

The apparent large summer drawdown of surface DIC at the BATS site has raised a number of questions regarding whether the decrease is biological or physical in origin and, if it is biological, what supplies the required nutrients since the water column is well stratified with a deep nutricline during that time of year (Michaels et al., 1994). Using a simple carbon budget model for BATS constrained by pCO2, alkalinity, DIC and DIC C-13 measurements, N. Gruber showed that summer DIC loss is governed by a substantial net community carbon uptake, and both T. Michaels and N. Gruber suggest that nitrogen fixation may supply the missing source of nitrogen (Michaels et al., 1996; Gruber and Sarmiento, 1997). This conclusion is based on the elevated N/P ratio in the Sargasso Sea thermocline, resulting from remineralization of nitrogen rich particles produced by fixation near the surface, and geochemical ventilation rate estimates. D. Barber suggested that the Sargasso Sea nitrogen fixation may in fact be a transient event caused by the increased dust transport from the drought-stricken Sahel over the last several decades.

P. Falkowski also discussed nitrogen fixation from an evolutionary and geological perspective, noting it is one of the oldest biochemical processes on earth. He also pointed out that from the overall oceanic N/P ratio, marine production does not appear to have been phosphorus limited, at least on the largest scales. Further, the present marine nitrogen cycle may not be in steady state, with excess denitrification over fixation. During glacial periods in contrast, one would expect decreased denitrification because of reduced continental shelf extent and increased fixation because of higher dust deposition and thus iron input. Much of the debate regarding the role of nitrogen fixation is based on indirect geochemical measures and will not be resolved, likely, without corroborating biological evidence from taxonomic distributions, molecular probe studies and direct measurements of the nitrogen fixation rates.

Community Structure and Export Production

D. Barber and M. Landry outlined a growing consensus that nutrient levels per se do not control community structure; rather, structure is influenced by the degree of disturbance, either by changes in the nutrient inputs or mixing and light environment, which allows for a decoupling of the phytoplankton and zooplankton growth rates. In simplest terms, the hypothesis suggests that steady regimes are marked by essentially a balanced ecosystem where phytoplankton growth approximately equals grazing, chlorophyll concentrations and particle export are low, and the community is dominated by picophytoplankton and microzooplankton. Transient events induce a shift toward a more diatom based community and bloom conditions, which in turn produces a larger particle export flux. From a series of phytoplankton growth and grazing dilution experiments, M. Landry showed that despite large differences in the background nutrient concentrations of the equatorial Pacific and oligotrophic Arabian, broad similarities exist between the community structure of the two regions. P. Falkowski noted, however, that the overall dynamics of community structure are not well known and that we are particularly weak on the theoretical basis for such changes.

Despite a number of methodological issues (e.g., Gardner, 1997), floating sediment traps remain one of the only tools available for estimating the sinking particle export flux. K. Buesseler discussed efforts to calibrate the shallow BATS traps using the upper water column 234Th deficit and the 234Th/C ratio in the trap material, showing a tendency to underpredict during high flux periods and overpredict during low flux periods. More generally, he showed that the correlation of primary production and trap estimated particle export is weak, with the trap fluxes as a fraction of production scattering between 2-10%. Based on 234Th export estimates from a variety of environments, a general theory was outlined where export ratio is low (2-10%) and remineralization shallow over most of the pelagic ocean. The exceptions are bloom events and high latitude systems with a correlation to diatom abundance, where the export ratio can increase to 20 to >50% with a much larger remineralization depth scale that may also be a function of community structure.

The production and export of dissolved organic carbon (DOC) is likely also related to community structure, but an overview by D. Hansell of regional net community production seasonal patterns, which accounts for both DOC and POC export and accumulation, suggests no clear, single relationship. Traditional C-14 primary production incubations may underestimate total carbon fixation because a substantial fraction of the fixed carbon may be released into the filtrate as DOC, which subsequently may be exported by mixing and/or advection. Reevaluations of the DOC export from the equatorial Pacific now show DOC accounts for only about 20% of the overall DIC loss (Hansell, 1997). A significant seasonal DOC build-up and export by deep winter mixing occurs at BATS (Carlson et al., 1994; Michaels et al., 1994), while estimates for the DOC export at the HOT site are comparable to the sinking particulate flux (Emerson et al., 1997). Hansel also presented preliminary data from the Ross Sea spring bloom, where the large net community production in Phaeocystis dominated regions appears to be stored in the suspended particle pool, and not exported or converted to DOC, for the duration of the bloom.


Transient tracer based estimates of the aphotic zone oxygen utilization rates place some of the strongest constraints on shallow remineralization rates (e.g., Jenkins, 1988) and have typically been several times the trap derived export fluxes (W. Jenkins, personal communication). The geochemical approaches by their nature integrate over relatively large time and space scales and thus include transient events which may not be sampled well by routine process or time-series sampling. The regional variation in OUR scale heights may also provide useful information on the character of the underlying biological dynamics, but one caveat to the technique is that the full 3-D oxygen field needs to be determined for accurate results.

The mechanistic controls governing remineralization are much less well known, but the stage appears to be set for considerable progress. Major uncertainties remain regarding the relative roles of zooplankton versus bacterial remineralization, zooplankton migration, and the biologically mediated dynamics of particle aggregation/disaggregation and exchange between the dissolved and suspended pools. A variety of techniques can be applied to the problem including: the bulk composition and organic geochemistry of suspended and sinking particles (M. Conte, personal communication); rate estimates for bacterial respiration and growth and zooplankton metabolism and feeding; and radionuclide distribution and partitioning coupled with particle dynamics models (K. Cochran, personal communication). Of critical importance for JGOFS is to understand how the aphotic ecosystem and remineralization scales may change in response to the overlying productivity and export flux.

Microzooplankton populations and rates are not measured routinely as part of the time-series programs but mesozooplankton biomass is, showing substantial diurnal migration (M. Landry, personal communication). The effective vertical carbon transport due to metabolism, excretion and predation at depth is only about 10-20% of sinking particle flux HOT (M. Landry, personal communication) and about 8% of the annual mean trap flux at 150m for BATS (D. Steinberg, personal communication).

Time-Space Variability

There is a growing realization within the biogeochemical community that sub-seasonal space and time-scale variability is not simply an annoyance to be averaged over but rather may be a crucial trigger controlling community structure and integrated carbon flux. Using a zooplankton data set from the North Sea, J. Steele's talk drew attention to the clear mismatch between the high levels of variability in observations compared with that in simple models. The question arises: how much of the variability is driven by physical forcing, for example synoptic events and mesoscale eddies that are typically missing from such models, versus the inherent variability in the biological dynamics? Does this variability rectify or amplify into the large-scale mean? One approach is to include stochastic elements either in model forcing or dynamics and then to compare ensemble behavior of the model and observations.

Alternatively, we can explicitly treat the underlying physical variability as highlighted at the meeting by D. McGillicuddy's presentation on mesoscale modeling work for the Sargasso Sea. Mesoscale eddies can enhance the effective upward flux of nutrients contributing to a larger new production by lifting nutrient rich isopycnal surfaces into the euphotic zone. Using an idealized 3-D model, McGillicuddy and Robinson (1997) show that the eddy pumping effect for the Sargasso Sea should be pronounced and can maintain on average a substantial new production rate over the well stratified summer period. Preliminary results from the summer 1997 BATS validation cruise also show significant promise for predicting the location of such eddy events using data assimilation models and satellite altimetry.

Geochemical techniques such as the He3--nitrate flux gauge (Jenkins, 1988) integrate over such episodic eddy events and also provide large estimates at BATS of the effective vertical mixing rate (10-3 m2/s) and upward nutrient flux (about 0.75 mol/m2/yr) comparable to the eddy pumping estimates. The He3 data, however, require transport of material all the way into the surface mixed layer, rather than just into the euphotic zone, and the physics of that process are not well described. Further, both the He3 flux gauge and the mesoscale eddy results confound the summer DIC drawdown problem because they both imply an additional large input of DIC. Considerable work will also be required to reconcile the large new production values from all of these approaches with the small sediment trap fluxes over the summer. The importance of spatial and temporal variability also highlights the need to develop new approaches for oceanographic biogeochemical sampling with growing emphasis on moorings and satellite remote sensing, in-situ chemical/biological measurement capabilities, purposeful manipulation experiments, and closer integration of field sampling, models and data assimilation.

Climate Variability

The marine carbon system is but one component of the global biogeochemical carbon cycle but is a potential key reservoir for the uptake and storage of anthropogenic carbon (Schimel et al., 1994). Growing concern over the possible climate impacts of increasing atmospheric CO2 levels has focused efforts on understanding the biological responses and feedbacks to climate change (Denman et al., 1996). As discussed by J. Sarmiento, recent advances in coupled ocean-atmosphere models (e.g., Manabe and Stouffer, 1993; Murphy and Mitchell, 1995) have resulted in a series of reasonably credible physical scenarios for the next century. Although regional features among the models vary widely, some aspects such as increased vertical stratification in both the low latitudes (warming) and high latitudes (freshening) are consistently observed. The projected circulation effects are pronounced at high latitudes, particularly for the southern hemisphere. Other changing environmental factors with potential impacts on marine biogeochemistry include dust (trace metals), SST, cloudiness and carbon chemistry.

Of major concern is whether increased vertical stability and trace metal input could lead to shifts in community structure, export flux and subsequently air-sea carbon exchange. Non-Redfield processes in the euphotic zone are a key area of concern since they could alter the C/N and/or C/P ratios in sinking particulate material and thus modify the coupling of carbon to nutrients. Other hypotheses focus on possible modification of: nitrogen fixation rates (trace metal deposition); community allocation of sinking versus suspended/dissolved material; calcification rates; and remineralization depth scales. The JGOFS time-series data sets provide an unique opportunity to study the ecosystem level marine biogeochemical response to climate variability, serving as a proxy for predicting future climate impacts. The extended ENSO signal of the mid-1990's appears to have caused an enhancement in nitrogen fixation at the HOT site associated with increased water column stability. On longer decadal time-scales, a potential regime shift with a two fold increase in primary production may have occured across the subtropical North Pacific as evidenced by the historical and HOT data.

Models and Data

Many of the key questions confronting the JGOFS Synthesis and Modeling Project involve the role of the ocean in the global carbon cycle and thus require us to extrapolate the results from the time-series and regional process studies to the basin and global scale. An underlying theme through the early JGOFS literature is that the time-series station data should be used to refine process models which can then be incorporated into 3-D general circulation models. Box models and 1-D vertical models are the most commonly applied classes to the time-series data sets (e.g., Doney et al., 1996; Evans and Garcon, 1997; Hurtt and Armstrong, 1996). But even such simple models require the integration of a number of different elements - surface forcing and initial conditions; physics; biological parameters - besides the actual biogeochemical model equations, and it is often difficult to separate errors in one component from another when judging the overall skill of the system. The HOT and BATS data sets have and will continue to prove especially fruitful for developing and improving process models; estimating poorly known or unknown parameters; intercomparing models; and testing new hypotheses.

A range of model complexities are available for both physical and biogeochemical models, and computational resources have placed a natural limit on the type of calculations run to date. For example, generally only very simple ecosystem models have been implemented in full eddy resolving models where quite sophisticated and complex food web models are commonly included in box model calculations. Other limitations to model complexity also arise, however, from our need to understand the model results and dynamics. A clear need expressed at the meeting was for a class of multi-element models (e.g., carbon, nitrogen, phosphorus, iron, silica) where the budgets of all the material fluxes could be solved for, if possible, simultaneously. Other discussions focused on how to represent community structure effects within biogeochemical models because it is recognized that the present practice of lumping all species into a single trophic component, for example "phytoplankton", is clearly inadequate. An intermediate solution to including all relevant species is the multi-functional group where a limited number of groups (e.g., nitrogen fixers; calcifiers; large-diatoms, etc.) are included based on their biogeochemical impact.

Models are simply tools, and more complex models are not always necessarily more useful or better representations of reality depending on the task. That point was nicely made by R. Armstrong's presentation of a very simple diagnostic model for estimating remineralization and export over the North Atlantic. Given satellite ocean color data, the model computes biomass and primary production, and the sum of export and remineralization is calculated by difference. The partitioning between export and remineralization is found using a simple functional relationship between f-ratio and biomass. At least for these measures, the diagnostic model produces more plausible fields than are generated from current 3-D biogeochemical GCMs.

Finally, it was reiterated at the meeting that field work and numerical modeling are but different facets of one science and that some of the main barriers to progress in marine biogeochemistry are not technical but sociological in how the scientists from these sub-disciplines interact. Observational data are not simply an end product that can be ported to models at the end of a field program; the scientists involved in collecting and analyzing the data bring with them the background and deep conceptual understanding of the data which is key to developing and utilizing better models. Conversely, models can provide a scientific context for planning and implementing field efforts. A variety of mechanisms, including SMP workshops, are likely required to improve the exchange of information and feedback between the observational and modeling communities but the ultimate benefits are significant.


Azam, F., T. Fenchel, J. G. Field, J. S. Gray, L. A. Meyer-Reil, and F. Thingstad (1983). The ecological role of water-column microbes in the sea. Mar. Ecol. Prog. Ser., 10: 257-263.

Capone, D. G., J. P. Zehr, H. W. Paerl, B. Bergman, and E. J. Carpenter (1997). Trichodesmium, a globally significant marine cyanobacterium. Science, 276: 1221-1229.

Carlson, C. A., H. W. Ducklow, and A. F. Michaels (1994). Annual flux of dissolved organic carbon from the euphotic zone in the northwest Sargasso Sea. Nature, 371: 405-408.

Coale, K. H., K. S. Johnson, S. E. Fitzwater, R. M. Gordon, S. Tanner, F. P. Chavez, L. Ferioli, C. Sakamoto, P. Rogers, F. Millero, P. Steinberg, P. Nightingale, D. Cooper, W. P. Cochlan, M. R. Landry, J. Constantinou, G. Rollwagen, A. Trasvina, and R. Kudela (1996). A massive phytoplankton bloom induced by an ecosystem-scale iron fertilization experiment in the equatorial Pacific Ocean. Nature, 383: 495-501.

Deuser, W. G. (1986). Seasonal and interannual variations in deep-water particle fluxes in the Sargasso Sea. Deep-Sea Res., 33: 225-247.

Deuser, W. G., and E. H. Ross (1980). Seasonal changes in the flux of organic carbon to the deep Sargasso Sea. Nature, 283: 364-365.

Denman, K., E. Hofmann, and H. Marchant (1996). Marine biotic responses to environmental change and feedbacks to climate. in Climate Change 1995, IPCC, ed. J. T. Houghton, L. G. Meira Filho, B. A. Callander, N. Harris, A. Kattenberg, and K. Maskell, Cambridge University Press, 487-516.

Dickey, T. D., D. Frye, H. Jannisch, E. Boyle, and A. H. Knap (1997). Bermuda Sensor System Testbed. Sea Technology, April 1997: 81-86.

Doney, S. C., D. M. Glover, and R. G. Najjar (1996). A new coupled, one-dimensional biological--physical model for the upper ocean: applications to the JGOFS Bermuda Atlantic Time-series (BATS) site. Deep-Sea Res. II, 43: 591-624.

Ducklow, H. W. (1983). Production and fate of bacteria in the oceans. Bioscience, 33: 494-501.

Dugdale, R. C., and J. J. Goering (1967). Uptake of new and regenerated forms of nitrogen in primary production. Limnol. Oceanogr., 12: 196-206.

Dugdale, R. C., and F. P. Wilkerson (1998). Silicate regulation of new production in the equatorial Pacific upwelling. Nature, 391: 270-273.

Emerson, S., P. Quay, D. Karl, C. Winn, L. Tupas, and M. Landry (1997). Experimental determination of the organic carbon flux from the open-ocean surface waters. Nature, 389: 951-954.

Evans, G. T., and V. C. Garcon, eds. (1997). One-dimensional models of water column biogeochemistry. JGOFS Report 23/97, 85pp., JGOFS, Bergen, Norway.

Fasham, M. J. R., H. W. Ducklow, and S. M. McKelvie (1990). A nitrogen-based model of plankton dynamics in the oceanic mixed layer. J. Mar. Res., 48: 591-639.

Frost, B. W. (1991). The role of grazing in nutrient-rich areas of the open sea. Limnol. Oceanogr., 36: 1616-1630.

Gardner, W. D. (1997). Sediment trap sampling in surface waters, First International JGOFS Symposium, Villefranche sur Mer, May 1995, Cambridge University Press.

Gruber, N., and J. L. Sarmiento (1997). Global patterns of marine nitrogen fixation and denitrification. Global Biogeochem. Cycles, 11: 235-266.

Hurtt, G. C., and R. A. Armstrong (1996). A pelagic ecosystem model calibrated with BATS data. Deep-Sea Res. II, 43: 653-683.

Jenkins, W. J. (1988). Nitrate flux into the euphotic zone near Bermuda. Nature, 331: 521-523.

Jenkins, W. J., and J. C. Goldman (1985). Seasonal oxygen cycling and primary production in the Sargasso Sea. J. Mar. Res., 43: 465-491.

Karl, D., R. Letelier, L. Tupas, J. Dore, J. Christian, and D. Hebel (1997). The role of nitrogen fixation in biogeochemical cycling in the subtropical North Pacific Ocean. Nature, 388: 533-538.

Karl, D. M., and R. Lukas (1996). The Hawaii Ocean Time-series (HOT) program: background, rationale and field implementation. Deep-Sea Res. II, 43: 129-156.

Karl, D. M., and A. F. Michaels (1996). Preface: the Hawaiian Ocean Time-series (HOT) and Bermuda Atlantic Time-series Study (BATS). Deep-Sea Res. II, 43: 127-128.

Landry, M. R., R. T. Barber, R. R. Bidigare, F. Chai, K. H. Coale, H. G. Dam, M. R. Lewis, S. T. Lindley, J. J. McCarthy, M. R. Roman, D. K. Stoecker, P. G. Verity, and J. R. White (1997). Iron and grazing constraints on primary production in the central equatorial Pacific: An EqPac synthesis. Limnol. Oceanogr., 42: 405-418.

Manabe, S., and R. J. Stouffer (1993). Century-scale effects of increased atmospheric CO2 on the ocean-atmosphere system. Nature, 364: 215-218.

Martin, J. H., and S. E. Fitzwater (1988). Iron deficiency limits phytoplankton growth in the north-east Pacific Subarctic. Nature, 331: 341-343.

Martin, J. H., G. A. Knauer, D. M. Karl and W. W. Broenkow (1987). VERTEX: carbon cycling in the northeast Pacific. Deep-Sea Res., 34: 267-285.

McGillicuddy, D. J., and A. R. Robinson (1997). Eddy-induced nutrient supply and new production. Deep-Sea Res. I, 44: 1427-1450.

McGillicuddy, D. J., A. R. Robinson, and J. J. McCarthy (1995). Coupled physical and biological modeling of the spring bloom in the North Atlantic: (ii) three dimensional bloom and post-bloom effects. Deep-Sea Res. I, 42: 1359-1398.

Menzel, D. W., and J. H. Ryther (1960). The annual cycle of primary production in the Sargasso Sea off Bermuda. Deep-Sea Res., 6: 351-367.

Michaels, A. F., N. R. Bates, K. O. Buesseler, C. A. Carlson, and A. H. Knap (1994). Carbon-cycle imbalances in the Sargasso Sea. Nature, 372: 537-540.

Michaels, A. F., and A. H. Knap (1996). Overview of the U.S. JGOFS Bermuda Atlantic Time-series Study and Hydrostation S program. Deep-Sea Res. II, 43: 157-198.

Michaels, A. F., D. Olson, J. L. Sarmiento, J. W. Ammerman, K. Fanning, R. Jahnke, A. H. Knap, F. Lipschultz, and J. M. Prospero (1996). Inputs, losses and transformations of nitrogen and phosphorus in the pelagic North Atlantic Ocean. Biogeochemistry, 35: 181-226.

Murphy, J. M., and J. F. B. Mitchell (1995). Transient response of the Hadley Centre coupled ocean-atmosphere model to increasing carbon dioxide. Part II: spatial and temporal structure of response. J. Climate, 8: 57.

Sayles, F. L., and W. R. Martin (1995). In situ tracer studies of solute transport across the sediment-water interface at the Bermuda Time Series site. Deep-Sea. Res. I, 42: 31-52.

Schudlich, R., and S. Emerson (1996). Gas supersaturation in the surface ocean: The roles of heat flux, gas exchange and bubbles. Deep-Sea Res. II, 43: 569-589.

Schimel, D., I. G. Enting, M. Heimann, T. M. L. Wigley, D. Raynaud, D. Alves, and U. Siegenthaler (1994). CO2 and the carbon cycle, in Climate Change 1994, Intergovernmental Panel on Climate Change, ed. J. T. Houghton, L. G. Meira Filho, J. Bruce, Hoesung Lee, B. A. Callander, E. Haites, N. Harris, and K. Maskell, Cambridge University Press, 39-71.

Siegel, D. A., A. F. Michaels, J. C. Sorensen, M. C. O'Brien and M. A. Hammer (1995). Seasonal variability of light availability and utilization in the Sargasso Sea. J. Geophys. Res., 100: 8695-8713.

Siegel, D. A., and A. F. Michaels (1996). Quantification of non-algal light attenuation in the Sargasso Sea: Implications for biogeochemistry and remote sensing. Deep-Sea Res. II, 43: 321-346.

Smith, R. C., K. S. Baker, W. R. Fraser, E. E. Hofmann, D. M. Karl, J. M. Klinck, L. B. Quetin, B. B. Prezelin, R. M. Ross, W. Z. Trivelpiece, and M. Vernet (1995). The Palmer LTER: A long-term ecological research program at Palmer Station, Antarctica. Oceanography, 8: 77-86.

Sugimura, Y., and Y. Suzuki (1988). A high-temperature catalytic oxidation method for the determination of non-volatile dissolved organic carbon in seawater by direct injection of a liquid sample. Mar. Chem., 24: 105-131.

Villareal, T. A., S. Woods, and K. Culver-Rymsza (1996). Vertical migration of Rhizosolenia mats and their significance to NO3 fluxes in the central North Pacific. J. Plankton Res., 18: 1103.