Report 18 -- Appendix 2D

Large Scale Biological Models

Ian J. Totterdell
James Rennell Centre for Ocean Circulation, NERC, Gamma House
Chilworth Research Centre
Chilworth, Southampton, SO16 7NS, United Kingdom

January 1995

INTRODUCTION

In Section 2 of this talk I shall review those biological models that have been developed to be generally applicable over large areas, particularly those that have been, or are in the process of being, incorporated in Ocean General Circulation Models (OGCMs). I am restricting the choice to dynamic models (i.e. not those that examine a steady-state), and to Eulerian models. The principal Lagrangian model (Wolf & Woods 1988; Woods & Barkmann 1994) is too computer-intensive to be included in any large physical model, and is best suited to small scale studies.

The impetus to incorporate biological models in OGCMs has come so far from studies of the global carbon cycle, which are primarily interested in the ecosystems in general terms, such as primary production and export flux, rather than in the details of the ecosystems themselves. However, if the biological interactions within the ecosystems are to be modeled and understood more detailed models will be needed. They will also give more confidence that the current extremely simple, carbon-cycle-oriented models are producing good simulations for the right reasons. The most urgently needed improvements in the models, and the measurements that will be required to provide validations and parameter values, are discussed in Section 3.

CURRENT LARGE-SCALE BIOLOGICAL MODELS

The Fasham-Ducklow-McKelvie (1990) model (hereafter FDM90) has been incorporated in several large-scale models. The ecosystem is based on nitrogen, and features seven compartments: nitrate, ammonia, bacteria, phytoplankton, zooplankton, detritus and dissolved organic nitrogen (DON). As important as the variables in a model however are the biological processes represented. In this case the phytoplankton photosynthesize (taking up nitrate and ammonia, but exuding DON) and suffer natural mortality (becoming detritus); the zooplankton graze phytoplankton, bacteria and detritus (producing more detritus as fecal pellets), excrete ammonia and DON and suffer mortality; the bacteria take in ammonia and DON, and excrete ammonia; and detritus sinks and breaks down to DON. The inhibition by ammonia of the uptake of nitrate by phytoplankton is included.

In the original paper this ecosystem existed in a slab mixed layer, the thickness of which was specified over an annual cycle, and biological quantities could be lost from this layer by turbulent mixing, detrainment (in the spring) and sinking (detritus only), while nitrate was resupplied by entrainment (from the end of summer) and by turbulent mixing throughout the year. The zooplankton variable is not specified to be any particular species, and in fact combines aspects of both micro- and mesozooplankton in that it both grazes bacteria (as well as phytoplankton and detritus) and produces fecal pellets (or detrital particles) which have a significant sinking rate. The zooplankton graze on each of their three types of prey with a preference governed by that prey's availability; hence a 'switching' behavior is produced.

The model was tested initially against data from Bermuda Station 'S', and later (Fasham 1993) against data from OWS 'India' (59°N, 19°W). The comparison of the _-dimensional model at OWS 'India' showed poor agreement unless the function describing zooplankton mortality (including predation by higher organisms) was altered so that there was a lower specific rate in winter. This change enabled more zooplankton to survive the winter and so provided a larger initial population which was able to respond in time to limit the spring bloom of phytoplankton before the nutrients were depleted.

FDM90 has been embedded in the Princeton OGCM and results published for simulations of the North Atlantic (Sarmiento et al. 1993; Fasham et al. 1993). The model output was compared both with satellite CZCS data and with in situ observations at Bermuda Station 'S' and OWS 'India'. In both of the latter comparisons the biological model over-estimates the phytoplankton concentrations in the spring bloom and under-predicts the summer populations. Also the predicted zooplankton biomass is higher than that observed. These problems relate to the balance between the rates of primary production and grazing, and will be considered again later.

A version of FDM90, with carbon flows coupled to the biology (Anderson 1993) is also being run in the Princeton OGCM, with global coverage, but as yet no results are available. Drange (1994) has independently coupled carbon flows to FDM90, and embedded it in an isopycnic-coordinate OGCM of the North Atlantic. Both this implementation and that in the Princeton OGCM restrict the active biology to the upper levels of the water-column, 180 m and 123 m respectively. Another interesting study involved embedding FDM90 in a quasi-geostrophic eddy-resolving model (Burren 1993).

A more complex version of FDM90 has been developed (Ducklow & Fasham 1992). In this model (DF92), the phytoplankton compartment of FDM90 has been split into separate picoplankton and net phytoplankton compartments, while the original zooplankton compartment has become separate protozoa and mesozooplankton compartments. These additions enabled some representation to be made of the microbial loop, and indeed were made so that the role of bacteria in recycling carbon and nutrients could be examined.

Finally, and with particular interest for this meeting, some work was undertaken to embed FDM90 in a model of the Southern Ocean. The model was a version of the Fine Resolution Antarctic Model (FRAM), but with coarser resolution (1x2° rather than _ x _; hence 'CRAM'). The coarser resolution was required to reduce the computer storage and cpu-time to realistic levels. However, problems were encountered with the climatological forcing used for CRAM (and FRAM) which involved annual-mean quantities for many of the variables. In fact, the only seasonality was provided by the variation of the incident solar radiation. Also the lack of a suitable ice-model, the coarse vertical resolution (20 m) in the euphotic zone and the lack of a wind-mixed-layer meant that the project was abandoned after just a few trial runs.

A simpler biological model has been developed by Kurz (1993) for inclusion in the Hamburg HAMMOC3 ocean model. The limiting nutrient is phosphate, and there are also compartments for phytoplankton and zooplankton. The biological processes included are primary production, grazing, natural mortality of phytoplankton and loss from the zooplankton to sinking detritus (due to fecal pellet production and/or higher predation). The biological model is restricted to the mixed-layer, which in the Hamburg model has no seasonal variation. This model has the advantages that it is simple to understand and runs exceedingly fast (the biological time step is one week; the physical time step is one month). It is also an improvement on the previous implicit biology of the Hamburg model. However, particularly in its current implementation, this model is most suitable for long climate runs rather than studying the dynamics of ecosystems.

Taylor, Harbour, Harris, Burkill and Edwards (1993) describe a model (hereafter THHBE93) of the ecosystem in the North Atlantic, based on work by Azam et al. (1983) and Taylor & Joint (1990). This nitrogen-based model includes bacteria, picophytoplankton, heteroflagellates, phytoflagellates, micrograzers (or ciliates), net diatoms and net phytoplankton, as well as silicate, nitrate, ammonium, DOC and detritus. With so many compartments it is obviously impractical to list every biological interaction, but in general the modeled processes are photosynthesis (including uptake of nutrients), grazing (that by mesozooplankton is imposed from measured rates), excretion and remineralization, as well as sinking (diatoms and detritus only). The detailed inclusion of the smaller organisms and their interactions allows a good description of the microbial loop, while having diatoms as a separate variable allows the silicate-limitation of their growth to be included.

The model is compared to the (relatively-detailed) observations made during the North Atlantic Bloom Experiment (NABE) in 1989 and on cruises during the following two years. The area of study covers both areas that exhibit nutrient depletion in the summer (e.g. 47°N, 20°W) and those which remain nutrient-replete (e.g. 60°N, 20°W), and THHBE93 successfully simulates both types of annual cycle. This model has so far only been run in half-dimensional mode (confined to a mixed layer).

The Hadley Centre for Climate Prediction and Research, part of the U.K. Meteorological Office, have developed a coupled ocean-atmosphere GCM, which can also be run (with suitable climatological forcings) in ocean-only or atmosphere-only mode. Work is nearing completion on a model of the global carbon cycle to be embedded in this coupled OAGCM, and ocean-only runs have been performed with a sub-model describing the marine biology. This model is nitrogen-based (with coupled carbon flows) and has nutrient, phytoplankton, zooplankton and detritus as the variables. The phytoplankton photosynthesize (taking up nutrient) and suffer mortality (becoming detritus); the zooplankton graze on both phytoplankton and detritus, produce fecal pellets (detritus) and suffer mortality due to predation (mainly becoming detritus); while the detritus sinks and is remineralized.

In the current implementation the biology is not restricted to the top couple of hundred meters, though the rate of photosynthesis is set to zero below 180 m. A fraction of the detritus can be found sinking at all depths, and zooplankton acting as detritivores can be found to below 400 m in the aftermath of a North Atlantic bloom. This model can, with a single parameter set, simulate the low-chlorophyll, high-nutrient ecosystem observed at OWS 'Papa' and in many parts of the Southern Ocean, and also the nutrient-depleting blooms of the North Atlantic, given suitable climatological forcings. This model is particularly constrained by computer storage and cpu-time requirements, as the Hadley Centre coupled OAGCM cannot be run in an 'off-line' mode. Having only one compartment for nutrients, it cannot distinguish new and regenerated production.

There are other models that deserve mention. Hofmann & Ambler (1988) produced a biological model for the US continental shelf featuring detritus, two size-classes of phytoplankton and five stages of the copepod life-cycle (including the egg and the adult stages). This was then embedded in a 2-D physical model (Hofmann 1988). There are also reports that an ecosystem model developed jointly at the Marine Research Centre in Helsinki, Finland and the Marine Research Institute in Tallinn, Estonia has been experimentally incorporated in the OPYC isopycnic code in Hamburg (OPYC circular no. 46, 22nd Dec. '93).

THE FUTURE DEVELOPMENT OF LARGE-SCALE BIOLOGICAL MODELS.

The three biological models that have been run incorporated in OGCMs (FDM90, Kurz's model and that at the Hadley Centre) have been very successful, in fact surprisingly so. Computing restraints have restricted the number of variables representing actual organisms in these models to just three, two and two respectively, and yet something recognizable as the annual cycles observed in different parts of the ocean still results. However, because of the aggregation of the biology into just two or three compartments, there are many parameters that are not well known or that it is not possible to measure - for example the mortality of 'zooplankton' or even the grazing rate of the same. Undeterminable parameters are able to be adjusted (within reasonable limits) to give a good fit to the available data. The value giving the best fit for today's climatological forcings may not be the best for future forcings, so this is an especially dangerous (but often necessary) course to follow in climate prediction experiments. These parameter adjustments are often needed because the data does not exist, and the ecosystems are not understood well enough to produce a detailed model and so suggest parameter values for the more aggregated model.

Also, while the large-scale models developed to date have mainly been intended - at least as incorporated in the OGCMs - to examine the role played by the ocean biology in the global carbon cycle, that will not always be the case. The marine ecosystems are interesting in themselves, and to simulate them in any detail, and understand the interactions, more complex models with better parameterized processes will have to be developed.

Microzooplankton

Excessive aggregation forces modelers to use one variable to fulfill two (or more) functions. In the models, fecal pellets which sink (as detritus) are produced by the zooplankton compartment, which is also responsible for such grazing-control of the phytoplankton as is observed. In the real ocean, there is little doubt that fecal pellets are produced by mesozooplankton. However, there is increasingly strong evidence (e.g. Burkill et al. 1993) that most of the grazing of primary production is by microzooplankton, which have short doubling times. In theory this should please modelers, because the complicated life-cycles and behavioral responses displayed by the mesozooplankton (e.g. vertical migration on a diel or annual cycle) are very difficult to incorporate in a model, and if the important agents for grazing control of phytoplankton are the microzooplankton then our job becomes easier. However, whereas many studies have been made of copepod feeding rates, data on microzooplankton grazing rates are much more scarce. Measurements were made in the Bellingshausen Sea in the austral summer of 1992-3 that indicated that the microzooplankton play as pivotal a role in the ecosystem around Antarctica as in the Northeast Atlantic (P. Burkill, pers. comm.).

I see it as crucial, therefore, that extensive measurements are made of microzooplankton biomass and grazing rates in any field program in the Southern Ocean. As yet there are no automated techniques, but that should not be allowed to prevent the measurements being made.

Lateral Heterogeneity

The large-scale models usually have very coarse resolution; for example the Hadley Centre OGCM has grid points spaced 2.5° N-S and 3.75° E-W. There is usually no attempt to take account of the patchiness and variability that occurs on scales much less than this. Experimentally determined growth rates for phytoplankton are assumed to be equally valid for the average population in a box several hundred kilometers square. In some cases this may be a good approximation; for example Burren (1993) examined the annual primary production calculated by FDM90 embedded in an eddy-resolving model and found that it differed by just one per cent from that calculated in a coarse-resolution model that did not display any mesoscale variability. However, the time of the predicted bloom was earlier by about a week. More pertinently, the model was run in an area where nutrients were depleted in the summer, and so the primary production should be limited by the available nutrient. In areas that show grazing limitation, the faster growth of the phytoplankton (producing an earlier the earlier bloom) should enable them to "escape" from the grazing control for longer and so show a greater productivity.

The experimental equipment for taking long, continuous and automated readings of chlorophyll in the top 500 m of the water column already exists. Laboratories in the U.K., particularly IOSDL and PML, have made extensive use of the SeaSoar for this purpose, so it should be possible to get detailed information on the spatial variability of phytoplankton.

The lateral heterogeneity of the mesozooplankton populations is probably even more important to measure. This should be especially true in the Southern Ocean where krill swarms can devastate regions of high productivity. Indeed, it may not be possible to model this trophic level accurately by an Eulerian model. Acoustic techniques for measuring large populations of mesozooplankton are now becoming available, and should be used in any Southern Ocean survey.

Lastly, the Bellingshausen Sea surveys along 85°W in the austral summer 1992-3 discovered a 'jet' of extremely high chlorophyll (D. Turner, pers. comm.). It was at about 67°S-69°S, and the chlorophyll values were as high as 5-6 mg-Chl a/m³. Further west (upstream), the values were 7 mg-Chl a/m³, and it was believed that the jet was carrying biological material advected from a zone of very high productivity further west, possibly near Peter the First Island. The ecosystem to the south of this jet was totally dominated by material detrained from it, at least at the time of the survey, which was over a month after the retreat of the ice. Should we aim to be able to model such locally-important features?

Iron Limitation

It is now clear that there are phytoplankton species in the ocean that are limited by the availability of the micro-nutrient iron. However the impact of iron-limitation on a full ecosystem on time-scales of longer than a few weeks remains to be demonstrated. None of the large-scale biological models feature iron limitation, both because of restrictions on the number of compartments and because of a lack of data. Since the Southern Ocean is one of the main areas claimed to be iron-limited, any experimental evidence on the question could be important.

Ice Models

Satellite imagery has shown strong blooms in regions around Antarctica that have been influenced by ice. Lancelot et al. (1991) modeled the development of an ice edge phytoplankton bloom that had been observed on the European Polarstern Study (EPOS) leg 2. It is also reported that blooms can begin to occur within the ice itself if conditions are suitable, and seed such blooms. Given that the productivity can be very high, it is important that the processes are understood. This is particularly the case if the high-chlorophyll water is then advected into the main ocean, as the Bellingshausen Sea cruises may have observed, and strongly influences the ecosystem there. Murphy et al. (1988) have suggested that the scales of interaction of the krill with their prey are much smaller in under-ice systems than in the open ocean.

REFERENCES

Anderson, T.R. 1992. Modeling the influence of food C:N ratio and respiration on growth and nitrogen excretion in marine zooplankton and bacteria. J. Plankton Res., 14, 1645-1671.

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

Burkill, P.H., E.S. Edwards, A.W.G. John & M.A. Sleigh. 1993. Microzooplankton and their herbivorous activity in the northeastern Atlantic Ocean. Deep-Sea Res. II, 40, 479-493.

Burren, C.L. 1993. The effect of mesoscale heterogeneity on oceanic primary productivity. Unpublished PhD. thesis, University of Southampton, U.K.

Drange, H. 1994. An Isopycnic Coordinate Carbon Cycle Model for the North Atlantic; and the Possibility of Disposing of Fossil Fuel CO₂ in the Ocean. Unpublished PhD. thesis, University of Bergen, Norway. 286pp.

Ducklow, H.W. & M.J.R. Fasham. 1992. Bacteria in the greenhouse: modeling the role of oceanic plankton in the global carbon cycle. In: Mitchell, Ralph (Ed.) "Environmental Microbiology", pp 1-31. Wiley-Liss.

Fasham, M.J.R. 1993. Modeling the marine biota. In: Heimann, M. (Ed.) "The Global Carbon Cycle", NATO ASI Series, Vol. I 15, pp457-504. Springer-Verlag.

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

Fasham, M.J.R., J.L. Sarmiento, R.D. Slater, H.W. Ducklow & R. Williams. 1993. Ecosystem behavior at Bermuda Station "S" and Ocean Weather Station "India": a general circulation model and observational analysis. Global Biogeochem. Cycles, 7, 379-415.

Hofmann, E.E. 1988. Plankton dynamics on the outer southeastern U.S. continental shelf. Part III: A coupled physical-biological model. J. Mar. Res., 46, 919-946.

Hofmann, E.E. & J.W. Ambler. 1988. Plankton dynamics on the outer southeastern U.S. continental shelf. Part II: A time-dependent biological model. J. Mar. Res., 46, 883-917.

Holling, C.S. 1965. The functional response of predators to prey density and its role in mimicry and population regulation. Mem. Entomol. Soc. Canada, No. 45. (61pp).

Kurz, K.D. 1993. Zur saisonalen Variation des ozeanischen Kohlendioxidpartialdrucks [The seasonal variation of oceanic pCO₂]. Examensarbeit Nr. 18, Max-Planck-Institut fur Meteorologie, Germany. 107pp.

Lancelot, C., C. Veth & S. Mathot. 1991. Modeling ice-edge phytoplankton bloom in the Scotia-Weddell sea sector of the Southern Ocean during spring 1988. J. Mar. Syst., 2, 333-346.

Murphy, E.J., J. Morris, J.L. Watkins & J. Priddle. 1988. Scales of interaction between Antarctic krill and the environment. In: Sahrhage, D. (Ed.), 'Antarctic Ocean and Resources Variability', pp120-130. Springer-Verlag.

Sarmiento, J.L., R.D. Slater, M.J.R. Fasham, H.W. Ducklow, J.R. Toggweiler & G.T. Evans. 1993. A seasonal three-dimensional ecosystem model of nitrogen cycling in the North Atlantic euphotic zone. Global Biogeochem. Cycles, 7, 417-450.

Taylor, A.H., D.S. Harbour, R.P. Harris, P.H. Burkill & E.S. Edwards. 1993. Seasonal succession in the pelagic ecosystem of the North Atlantic and the utilization of nitrogen. J. Plankton Res., 15, 875-891.

Taylor, A.H. & I.R. Joint. 1990. A steady-state analysis of the "microbial loop" in stratified systems. Mar. Ecol. Prog. Ser., 59, 1-17.

Wolf, K.U. & J.D. Woods. 1988. Lagrangian simulation of primary production in the physical environment - the deep chlorophyll maximum and nutricline. In: Rothschild, B.J. (Ed.), 'Towards a Theory of Biological-Physical Interactions in the World Ocean', p51. Reidel.

Woods, J. & W. Barkmann. 1994. Simulating plankton ecosystems by the Lagrangian ensemble method. Phil. Trans. R. Soc. Lond. B, 343, 27-31.