area of research. The Rrst computation of globaloceanic primary production using the remote-sensing approach appeared in the literature in 1995(Figure 1). Other, similar computations havesince appeared in the literature. It is a method thatwill continue to improve, with improvements insatellite technology as well as in the techniques forextrapolation of local biological measurements tolarge scales.

See also

Microbial Loops. Network Analysis of Food Webs.Ocean Gyre Ecosystems. Pelagic Biogeography.Primary Production Processes. Primary Produc-tion Methods.

Further ReadingChisholm SW and Morel FMM (eds) (1991) What Con-

trols Phytoplankton Production in Nutrient-Rich Areasof the Open Sea? vol. 36. Lawrence, KS: AmericanSociety of Limnology and Oceanography.

Falkowski PG and Woodhead AD (eds) (1992) PrimaryProductivity and Biogeochemical Cycles in the Sea.New York: Plenum Press.

Geider RJ and Osborne BA (1992). Algal Photosynthesis.New York: Chapman & Hall.

Li WKW and Maestrini SY (eds) (1993) Measurement ofPrimary Production from the Molecular to the GlobalScale, ICES Marine Science Symposia, vol. 197.Copenhagen: International Council for the Explorationof the Sea.

Longhurst A (1998) Ecological Geography of the Sea. SanDiego: Academic Press.

Longhurst A, Sathyendranath S, Platt T and CaverhillC (1995) An estimate of global primary production inthe ocean from satellite radiometer data. Journal ofPlankton Research 17: 1245}1271.

Mann KH and Lazier JRN (1991) Dynamics of MarineEcosystems. Biological}Physical Interactions in theOceans. Cambridge, USA: Blackwell Science.

Platt T and Sathyendranath S (1993) Estimators of pri-mary production for interpretation of remotely senseddata on ocean color. Journal of Geophysical Research98: 14561}14576.

Platt T, Harrison WG, Lewis MR et al. (1989) Biologicalproduction of the oceans: the case for a consensus.Marine Ecology Progress Series 52: 77}88.

PRIMARY PRODUCTION METHODS

J. J. Cullen, Department of Oceanography,Halifax, Canada

Copyright ^ 2001 Academic Press

doi:10.1006/rwos.2001.0203

Introduction

Primary production is the synthesis of organic ma-terial from inorganic compounds, such as CO2 andwater. The synthesis of organic carbon from CO2 iscommonly called carbon Rxation: CO2 is Rxed byboth photosynthesis and chemosynthesis. By far,photosynthesis by phytoplankton accounts for mostmarine primary production. Carbon Rxation bymacroalgae, microphytobenthos, chemosyntheticmicrobes, and symbiotic associations can be locallyimportant.

Only the measurement of marine planktonicprimary production will be discussed here. Thesemeasurements have been made for many decadesusing a variety of approaches. It has long beenrecognized that different methods yield differentresults, yet it is equally clear that the variability ofprimary productivity, with depth, time of day,season, and region, has been well described by most

measurement programs. However, details of thesepatterns can depend on methodology, so it isimportant to appreciate the uncertainties and built-in biases associated with different methods formeasuring primary production.

De\nitions

Primary production is centrally important to eco-logical processes and biogeochemical cycling inmarine systems. It is thus surprising, if not discon-certing, that (as discussed by Williams in 1993),there is no consensus on a deRnition of planktonicprimary productivity, or its major components, netand gross primary production. One major reasonfor the problem is that descriptions of ecosystemsrequire clear conceptual deRnitions for processes(e.g., net daily production of organic material byphytoplankton), whereas the interpretation ofmeasurements requires precise operational deRni-tions, for example, net accumulation of radiolabeledCO2 in particulate matter during a 24 h incubation.Conceptual and operational deRnitions can be rec-onciled for particular approaches, but no one set ofdeRnitions is sufRciently general, yet detailed, toserve as a framework both for measuring planktonic

PRIMARY PRODUCTION METHODS 2277

primary production with a broad variety of methodsand for interpreting the measurements in a range ofscientiRc contexts. It is nonetheless useful to deRnethree components of primary production that can beestimated from measurements in closed systems:

f Gross primary production (Pg) is the rate ofphotosynthesis, not reduced for losses to excre-tion or to respiration in its various forms

f Net primary production (Pn) is gross primaryproduction less losses to respiration by phyto-plankton

f Net community production (Pnc) is net primaryproduction less losses to respiration by hetero-trophic microorganisms and metazoans.

Other components of primary production, such asnew production, regenerated production, and exportproduction, must be characterized to describe food-web dynamics and biogeochemical cycling. Aspointed out by Platt and Sathyendranath in 1993, inany such analysis, great care must be taken toreconcile the temporal and spatial scales of boththe measurements and the processes they describe.

Marine primary production is commonly ex-pressed as grams or moles of carbon Rxed per unitvolume, or pet unit area, of sea water per unit time.The timescale of interest is generally 1 day or1 year. Rates are characterized for the euphoticzone, commonly deRned as extending to the depthof 1% of the surface level of photosyntheticallyactive radiation (PAR: 400}700nm). This conve-nient deRnition of euphotic depth (sometimes sim-pliRed further to three times the depth at whicha Secchi disk disappears) is a crude and often inac-curate approximation of where gross primary pro-duction over 24 h matches losses to respiration andexcretion by phytoplankton. Regardless, rates ofphotosynthesis are generally insigniRcant below thedepth of 0.1% surface PAR.

Photosynthesis and Growth ofPhytoplankton

Primary production is generally measured by quan-tifying light-dependent synthesis of organic carbonfrom CO2 or evolution of O2 consistent with thesimpliRed description of photosynthesis as thereaction:

CO2#2H2O&8hl&P (CH2O)#H2O#O2 [1]

Absorbed photons are signiRed by hl and the carbo-hydrates generated by photosynthesis are represent-

ed as CH2O. Carbon dioxide in sea water is foundin several chemical forms which exchange quicklyenough to be considered in aggregate as total CO2

(TCO2). In principle, photosynthesis can be quanti-Red by measuring any of three light-dependent pro-cesses: (1) the increase in organic carbon; (2) thedecrease of TCO2; or (3) the increase of O2. How-ever, growth of phytoplankton is not so simple:since phytoplankton are composed of proteins,lipids, nucleic acids, and other compoundsbesides carbohydrate, both photosynthesis andthe assimilation of nutrients are required. Conse-quently, many chemical transformations are asso-ciated with primary production, and eqn [1] doesnot accurately describe the process of light-depen-dent growth.

It is therefore useful to describe the growth ofphytoplankton (i.e., net primary production) witha more general reaction that describes how trans-formations of carbon and oxygen depend on thesource of nutrients (particularly nitrogen) and onthe chemical composition of phytoplankton. Forgrowth on nitrate:

1.0 NO~3 #5.7CO2#5.4H2O

P(C5.7H9.8O2.3N)#8.25O2#1.0OH~ [2]

The idealized organic product, C5.7H9.8O2.3N, rep-resents the elemental composition of phytoplankton.Ammonium is more reduced than nitrate, so lesswater is required to satisfy the demand for reduc-tant:

1.0NH4̀ #5.7CO2#3.4H2O

P(C5.7H9.8O2.3N)#6.25O2#1.0 H` [3]

The photosynthetic quotient (PQ; mol mol~1) is theratio of O2 evolved to inorganic C assimilated. Itmust be speciRed to convert increases of oxygen tothe synthesis of organic carbon. For growth on ni-trate as described by eqn [2], PQ is 1.45 mol mol~1;with ammonium as the source of N, PQ is 1.10. Thephotosynthetic quotient also reSects the end prod-ucts of photosynthesis, the mixture of which variesaccording to environmental conditions and the spe-cies composition of phytoplankton. For example, ifthe synthesis of carbohydrate is favored, as canoccur in high light or low nutrient conditions, PQ islower because the reaction described in eqn [1] be-comes more important. Uncertainty in PQ is oftenignored. This can be justiRed when the synthesis oforganic carbon is measured directly, but large errorscan be introduced when attempts are made toinfer carbon Rxation from the dynamics of oxygen.

2278 PRIMARY PRODUCTION METHODS

Table 1 Measurements that can be related to primary production

Measurement Advantages Disadvantages Comments

Change in TCO2 Direct measure of netinorganic C fixation

Relatively insensitive: smallchange relative to largebackground

Not generally practical foropen-ocean work

Change in oxygenconcentration (highprecision titration)

Direct measures ofO2 dynamics can yieldestimates of net and grossproduction

Small change relative to largebackground

Interpretation of light-darkincubations is not simple

Very useful if applied withgreat care.

Requires knowledge of PQto convert to C-fixation

Incorporation of14C-bicarbonate into organicmaterial (radioactiveisotope)

Very sensitive and relativelyeasy.

Small volumes can be usedand many samples can beprocessed

Tracer dynamics complicateinterpretations

Radioactive I requires specialprecautions and permission

The most commonly usedmethod in oceanography

Incorporation of13C-bicarbonate into organicmaterial (stable isotope)

No problems with radioactivity Less sensitive and more workthan 14C method

Larger volumes required

A common choice when14C method isimpractical

Measurement of18O2 production from H2

18OMeasures photosynthesis

without interference fromrespiration

Requires special equipment A powerful research tool,not generally used forroutine measurements

Excretion of organic material would have a smallinSuence on PQ and is not considered here.

Approaches

Primary production can be estimated from chloro-phyll (from satellite color or in situ Suorescence) ifcarbon uptake per unit of chlorophyll is known.Therefore, ‘global’ estimates of primary productiondepend on direct measurements by incubation. Thetechnical objectives are to obtain a representativesample of sea water, contain it so that no signiRcantexchange of materials occurs, and to measure light-dependent changes in carbon or oxygen during incu-bations that simulate the natural environment.Methods vary widely, and each approach involvescompromises between needs for logistical conveni-ence, precision, and the simulation of natural condi-tions. Each program of measurement involves manydecisions, each of which has consequences for theresulting measurements. Several options are listed inTables 1 and 2 and discussed below.

Light-dependent Change in Dissolved Oxygen

The light-dark oxygen method is a standard ap-proach for measuring photosynthesis in aquatic sys-tems, and it was the principal method for measuringmarine primary production until it was supplantedby the 14C method, which is described below. Accu-mulation of oxygen in a clear container (light bottle)represents net production by the enclosed commun-ity, and the consumption of oxygen in a dark bottleis a measure of respiration. Gross primary produc-tion is estimated by subtracting the dark bottle re-

sult from that for the light bottle. It is thus assumedthat respiration in the light equals that in the dark.As documented by Geider and Osborne in their1992 monograph, this assumption does not gener-ally hold, so errors in estimation of the respiratorycomponent of Pg must be tolerated unless isotopi-cally labelled oxygen is used (see below).

Methods based on the direct measurement ofoxygen are less sensitive than techniques using theisotopic tracer 14C. However, careful implementa-tion of procedures using automated titration orpulsed oxygen electrodes can yield useful andreliable data, even from oligotrophic waters of theopen ocean. Interpretation of results is complicatedby containment effects common to all methods fordirect measurement of primary production (see be-low). Also, a value for photosynthetic quotient mustbe assumed in order to infer carbon Rxation fromoxygen production. Abiotic consumption of oxygenthrough photochemical reactions with dissolvedorganic matter can also contribute to the measure-ment, primarily near the surface, where the effectiveultraviolet wavelengths penetrate.

Light-dependent Change in Dissolved InorganicCarbon

Changes in TCO2 during incubations of sea watercan be measured by several methods. Uncertaintiesrelated to biological effects on pH-alkalinity-TCO2

relationships are avoided through the use ofcoulometric titration or infrared gas analysis afteracidiRcation. Measurement of gross primary pro-duction and net production of the enclosed com-munity is like that for the light-dark oxygen

PRIMARY PRODUCTION METHODS 2279

method, but there is no need to assume a photo-synthetic quotient. However, precision of theanalyses is not quite as good as for bulk oxygenmethods. Extra procedures, such as Rltration, wouldbe required to assess precipitation of calcium car-bonate (e.g., by coccolithophores) and photochemi-cal production of CO2. These processes causechanges in TCO2 that are not due to primary pro-duction. The TCO2 method is not used routinely formeasurement of primary production in the ocean.

The 14C Method

Marine primary production is most commonly mea-sured by the 14C method, which was introduced bySteemann Nielsen in 1952. Samples are collectedand the dissolved inorganic carbon pool is labeledwith a known amount of radioactive 14C-bicarbon-ate. After incubation in clear containers, carbonRxation is quantiRed by liquid scintillation countingto detect the appearance of 14C in organic form.Generally, organic carbon is collected as particleson a Rlter. Both dissolved and particulate organiccarbon can be quantiRed by analyzing whole waterafter acidiRcation to purge the inorganic carbon. Itis prudent to correct measurements for the amountof label incorporated during incubations in the dark.The 14C method can be very sensitive, and goodprecision can be obtained through replication andadequate time for scintillation counting. Themethod has drawbacks, however. Use of radioiso-topes requires special procedures for handling anddisposal that can greatly complicate or precludesome Reld operations. Also, because 14C is added asdissolved inorganic carbon and gradually enterspools of particulate and dissolved matter, thedynamics of the labeled carbon cannot accuratelyrepresent all relevant transformations betweenorganic and inorganic carbon pools. For example,respiration cannot be quantiRed directly. The inter-pretation of 14C uptake (discussed below) is thusanything but straightforward.

The 13C Method

The 13C method is similar to the 14C method in thata carbon tracer is used. Bicarbonate enriched withthe stable isotope 13C is added to sea water and theincorporation of CO2 into particulate matter is fol-lowed by measuring changes in the 13C : 12C ratio ofparticles relative to that in the TCO2 pool. Isotoperatios are measured by mass spectrometry or emis-sion spectrometry. Problems associated with radio-isotopes are avoided, but the method can be morecumbersome than the 14C method (e.g., largervolumes are generally needed) and some sensitivityis lost.

The 18O Method

Gross photosynthesis can be measured as the pro-duction of 18O-labeled O2 from water labeled withthis heavy isotope of oxygen (see eqn [1]). Detectionis carried out by mass spectrometry. Net primaryproduction of the enclosed community is measuredas the increase of oxygen in the light bottle, andrespiration is estimated by difference. In principle,the difference between gross production measuredwith 18O and gross production from light-darkoxygen changes is due to light-dependent changes inrespiration and photochemical consumption ofoxygen. Respiration can also be measured directlyby tracking the production of H2

18O from 18O2.The 18O method is sufRciently sensitive to yield

useful results even in oligotrophic waters. It is notcommonly used, but when the measurements havebeen made and compared to other measures of pro-ductivity, important insights have been developed.

Methodological Considerations

Many choices are involved in the measurement ofprimary production. Most inSuence the results,some more predictably than others. A brief reviewof methodological choices, with an emphasis on the14C method, reveals that the measurement of pri-mary production is not an exact science.

Sampling

Every effort should be made to avoid contaminationof samples obtained for the measurement of primaryproduction. Concerns about toxic trace elementsare especially important in oceanic waters. Tracemetal-clean procedures, including the use ofspecially cleaned GO-FLO sampling bottles sus-pended from KevlarTM line, prevent the toxiccontamination associated with other samplers, parti-cularly those with neoprene closure mechanisms. Fre-quently, facilitates for trace metal-clean samplingare unavailable. Through careful choice of materialsand procedures, it is possible to minimize toxiccontamination, but enrichment with tracenutrients such as iron is probably unavoidable.Such enrichment could stimulate the photosynthesisof phytoplankton, but only after several hours orlonger.

Exposure of samples to turbulence during samp-ling can damage the phytoplankton and othermicrobes, altering measured rates. Also, signiRcantinhibition of photosynthesis can occur when deepsamples acclimated to low irradiance are exposedto bright light, even for brief periods, duringsampling.

2280 PRIMARY PRODUCTION METHODS

Table 2 Approaches for incubating samples for the measurement of primary production

Incubation system Advantages Disadvantages Comments

Incubation in situ Best simulation of the naturalfield of light and temperature

Limits mobility of the shipVertical mixing is not simulatedArtifacts possible if deployed or

recovered in the light

Not perfect, but a goodstandard method ifa station can beoccupied all day

Simulated in situ Many stations can be surveyedEasy to conduct time-courses

and experimentalmanipulations

Special measures must betaken to stimulate spectralirradiance and temperature

Vertical mixing not simulated

Commonly used whenmany stations must besampled.

Significant errors possibleif incubated samples areexposed to unnaturalirradiance andtemperature

Photosynthesis versusirradiance (P versus E)incubator (14C)

Data can be used to modelphotosynthesis in the watercolumn

With care, vertical mixing canbe addressed

Extra expenses andprecautions are required

Spectral irradiance is notmatched to nature

Results depend on timescale ofmeasurement

Analysis can be tricky

A powerful approach whenapplied with caution

Method of Incubation

Samples of seawater can be incubated in situ, undersimulated in situ (SIS) conditions, or in incubatorsilluminated by lamps. Each method has advantagesand disadvantages (Table 2).

Incubation in situ ensures the best possible simu-lation of natural conditions at the depths of samp-ling. Ideally, samples are collected, prepared, anddeployed before dawn in a drifting array. Samplesare retrieved and processed after dusk or before thenext sunrise. If deployment or retrieval occur duringdaylight, deep samples can be exposed to unnatural-ly high irradiance during transit, which can lead toartifactually high photosynthesis and perhaps tocounteracting inhibitory damage. Incubation ofsamples in situ limits the number of stations thatcan be visited during a survey, because the shipmust stay near the station in order to retrieve thesamples. Specialized systems both capture andinoculate samples in situ, thereby avoiding somelogistical problems.

Ship operations can be much more Sexible if pri-mary productivity is measured using SIS incuba-tions. Water can be collected at any time of day andincubated for 24 h on deck in transparent incubatorsto measure daily rates. The incubators, or bottles inthe incubators, are commonly screened with neutraldensity Rlters (mesh or perforated metal screen) toreproduce Rxed percentages of PAR at the surface.Light penetration at the station must be estimated tochoose the sampling depths corresponding to theselight levels. Cooling comes from surface sea water.

This system has many advantages, including im-proved security of samples compared with in situdeployment, convenient access to incubations fortime-course measurements, and freedom of shipmovement after sampling. Because the spectrallyneutral attenuation of sunlight by screens does notmimic the ocean, signiRcant errors can be intro-duced for samples from the lower photic zone wherethe percentage of surface PAR imposed by a screenwill not match the percentage of photosyntheticallyutilizable radiation (PUR, spectrally weighted forphotosynthetic absorption) at the sampling depth.Incubators can be Rtted with colored Rlters to simu-late subsurface irradiance for particular water types.Also, chillers can be used to match subsurfacetemperatures, avoiding artifactual warming of deepsamples.

ArtiRcial incubators are used to measure photo-synthesis as a function of irradiance (P versus E).Illumination is produced by lamps, and a variety ofmethods are used to provide a range of light levelsto as many as 24 or more subsamples. Temperatureis controlled by a water bath. The duration ofincubation generally ranges from about 20 min toseveral hours, and results are Rtted statistically toa P versus E curve. If P versus E is determined forsamples at two or more depths (to account forphysiological differences), results can be used todescribe photosynthesis in the water column asa function of irradiance. Such a calculation requiresmeasurement of light penetration in the water andconsideration of spectral differences between the in-cubator and natural waters. Because many samples,

PRIMARY PRODUCTION METHODS 2281

Table 3 Containers for incubations

Container Advantages Disadvantages Comments

Polycarbonate bottle Good for minimizing traceelement contamination

Nearly unbreakableAffordable

Excludes UV radiationCompressible, leading to gas

dissolution and filtrationproblems for deep samples

Many advantages forroutine and specializedmeasurements at sea

Laboratory grade borosilicateglass (e.g., PyrexTM)

More transparent to UVIncompressible

More trace elementcontamination

Breakable

A reasonable choice ifcompromises areevaluated

Borosilicate glass scintillationvials

InexpensivePractical choice for P versus

E

Contaminate samples withtrace elements and Si

Exclude UV radiation

Can be used with cautionfor short-term P versusE measurements

Polyethylene bag InexpensiveCompactUV-transparent

More difficult to handleRequires caution with respect

to contamination

Used for special projects,e.g., effect of UV

Quartz, TeflonTM UV-transparentTeflonTM does not

contaminate

Relatively expensive Used for work assessingeffects of UV

Small volume (1}25 ml) Good for P versus ESamples can be processed by

acidification (no filtration)

Cannot sample large, rarephytoplankton evenly

Containment effects morelikely

Used for P versus E withmany replicates

Large volume (1}20 l) Some containment artifactsare minimized

Potential for time-coursemeasurements

More workLonger filtration times with

possible artifacts

Required for some types ofanalysis, e.g., 13C

usually of small volume, must be processed quickly,only the 14C method is appropriate for most Pversus E measurements in the ocean.

Containers

Ideally, containers for the measurement of primaryproduction should be transparent to ultraviolet andvisible solar radiation, completely clean, and inert(Table 3). Years ago, soft glass bottles were used.Now it is recognized that they can contaminatesamples with trace elements and exclude naturallyoccurring ultraviolet radiation. Glass scintillationvials are still used for some P versus E measure-ments of short duration; checks for effects of con-taminants are warranted. Compared with soft glass,laboratory-grade borosilicate glass bottles (e.g.,PyrexTM) have better optical properties, excludingonly UV-B (280d320 nm) radiation. Also, they con-taminate less. Laboratory-grade glass bottles arecommonly used for oxygen measurements. Polycar-bonate bottles are favored in many studies becausethey are relatively inexpensive, unbreakable, andcan be cleaned meticulously. Polycarbonate absorbsUV-B and some UV-A (320d400 nm) radiation, sonear-surface inhibition of photosynthesis can beunderestimated. The error can be signiRcant veryclose to the surface, but not when the entire watercolumn is considered. TeSonTM bottles, more expen-sive than polycarbonate, are noncontaminating and

they transmit both visible and UV (280d400 nm)radiation. When the primary emphasis is an assess-ing effects of UV radiation, incubations are conduc-ted in polyethylene bags or in bottles made ofquartz or TeSonTM.

The size of the container is an important consid-eration. Small containers (450 ml) are needed whenmany samples must be processed (e.g. for P versusE) or when not much water is available. However,small samples cannot represent the planktonic as-semblage accurately when large, rare organisms orcolonies are in the water. Smaller containers havegreater surface-to-volume ratios, and thus smallsamples have greater susceptibility to contamina-tion. If it is practical, larger samples should be usedfor the measurement of primary production. Theproblems with large samples are mostly logistical.More water, time, and materials are needed, moreradioactive waste is generated, and some measure-ments can be compromised if handling times are toolong.

Duration of Incubation

Conditions in containers differ from those in openwater, and the physiological and chemical differ-ences between samples and nature increase as theincubations proceed. Unnatural changes duringincubation include: extra accumulation of phyto-plankton due to exclusion of grazers; enhanced inhi-

2282 PRIMARY PRODUCTION METHODS

Table 4 Incubation times for the measurement of primary production

Incubation time Advantages Disadvantages Comments

Short (41 h) Little time for unnaturalphysiological changes

Usually requires artificial illuminationUncertain extrapolation to daily rates

in nature

Closer to Pg

1}6 h ConvenientAppropriate for some process

studies

Uncertain extrapolation to daily ratesin nature

Used for P versus E, especiallywith larger samples

Dawn}dusk Good for standardization ofmethodology

Limits the number of stations that canbe sampled

Containment effects

A good choice for standardmethod using in situincubation

Vertical mixing is not simulated,leading to artifacts

Closer to Pnc near the surface;close to Pg deep in the photiczone

24 h Good for standardization ofmethodology

Results may vary depending on starttime.

A good standard for SISincubations.

Longer time for containment effects toact

Close to Pnc near the surface;closer to Pg deep in thephotic zone

bition of photosynthesis in samples collected frommixed layers and incubated at near-surface irra-diance; stimulation of growth due to contaminationwith a limiting trace nutrient such as iron; andpoisoning of phytoplankton with a contaminant,such as copper. When photosynthesis is measuredwith a tracer, the distribution of the tracer amongpools changes with time, depending on the rates ofphotosynthesis, respiration, and grazing. All of theseeffects, except possibly toxicity, are minimized byrestricting the time of incubation, so a succession ofshort incubations, or P versus E measurements, canin principle yield more accurate data than a day-long incubation. This requires much effort, how-ever, and extrapolation of results to daily productiv-ity is still uncertain. The routine use of dawn-to-dusk or 24 h incubations may be subject to artifactsof containment, but it has the advantage of beingmuch easier to standardize.

Filtration or Acidi\cation

Generally, an incubation with 14C or 13C is termin-ated by Rltration. Labeled particles are collected ona Rlter for subsequent analysis. Residual dissolvedinorganic carbon can be removed by careful rinsingwith Rltered sea water; exposure of the Rlter to acidpurges both dissolved inorganic carbon and precipi-tated carbonate. The choice of Rlter can inSuencethe result. Whatman GF/F glass-Rber Rlters, withnominal pore size 0.7lm, are commonly used andwidely (although not universally) considered to cap-ture all sizes of phytoplankton quantitatively. Per-forated Rlters with uniform pore sizes ranging from0.2 to 5 lm or more can be used for size-fractiona-tion. Particles larger than the pores can squeeze

through, especially when vacuum is applied. TheRlters are also subject to clogging, leading to reten-tion of small particles.

Labeled dissolved organic carbon, including ex-creted photosynthate and cell contents releasedthrough ‘sloppy feeding’ of grazers, is not collectedon Rlters. These losses are generally several percentof total or less, but under some conditions, excre-tion can be much more. When 14C samples areprocessed with a more cumbersome acidiRcationand bubbling technique, both dissolved and partic-ulate organic carbon is measured.

Interpretation of Carbon Uptake

Because the labeled carbon is initially only in theinorganic pool, short incubations with 14C (41 h)characterize something close to gross production.As incubations proceed, cellular pools of organiccarbon are labeled, and some 14C is respired. Also,some excreted 14C organic carbon is assimilated byheterotrophic microbes, and some of the phyto-plankton are consumed by grazers. So, with time,the measurement comes closer to an estimate of thenet primary production of the enclosed community(Table 4). However, many factors, including theratio of photosynthesis to respiration, inSuence thedegree to which 14C uptake resembles gross versusnet production. Consequently, critical interpretationof 14C primary production measurements requiresreference to models of carbon Sow in the system.

Conclusions

Primary production is not like temperature, salinityor the concentration of nitrate, which can in

PRIMARY PRODUCTION METHODS 2283

principle be measured exactly. It is a biologicalprocess that cannot proceed unaltered when phytop-lankton are removed from their natural surround-ings. Artifacts are unavoidable, but many insults tothe sampled plankton can be minimized through theexercise of caution and skill. Still, the observed rateswill be inSuenced by the methods chosen for mak-ing the measurements. Interpretation is also uncer-tain: the 14C method is the standard operationaltechnique for measuring marine primary produc-tion, yet there are no generally applicable rules forrelating 14C measurements to either gross or netprimary production.

Fortunately, uncertainties in the measurementsand their interpretation, although signiRcant, arenot large enough to mask important patterns ofprimary productivity in nature. Years of data onmarine primary production have yielded informa-tion that has been centrally important to our under-standing of marine ecology and biogeochemicalcycling. Clearly, measurements of marine primaryproduction are useful and important for understand-ing the ocean. It is nonetheless prudent to recognizethat the measurements themselves require circum-spect interpretation.

See also

Carbon Cycle. Fluorometry for Biological Sensing.Ocean Carbon System, Modelling of. NetworkAnalysis of Food Webs. Ocean Color from Satel-

lites. Primary Production Distribution. PrimaryProduction Processes. Tracers of Ocean Produc-tivity.

Further ReadingGeider RJ and Osborne BA (1992) Algal Photosynthesis:

The Measurement of Algal Gas Exchange. New York:Chapman and Hall.

Morris I (1981) Photosynthetic products, physiologicalstate, and phytoplankton growth. In: Platt T (ed.)Physiological Bases of Phytoplankton Ecology. Cana-dian Bulletin of Fisheries and Aquatic Science 210:83d102.

Peterson BJ (1980) Aquatic primary productivity and the14C}CO2 method: a history of the productivity prob-lem. Annual Review of Ecology and Systematics 11:359}385.

Platt T and Sathyendranath S (1993) Fundamental issuesin measurement of primary production. ICES MarineScience Symposium 197: 3}8.

Sakshaug E, Bricaud A, Dandonneau Y et al. (1997)Parameters of photosynthesis: deRnitions, theory andinterpretation of results. Journal of Plankton Research19: 1637}1670.

Steemann Nielsen E (1963) Productivity, deRnition andmeasurement. In: Hill MW (ed.) The Sea, vol. 1, pp.129}164. New York: John Wiley.

Williams PJL (1993a) Chemical and tracer methods ofmeasuring plankton production. ICES Marine ScienceSymposium 197: 20}36.

Williams PJL (1993b) On the deRnition of plankton pro-duction terms. ICES Marine Science Symposium 197:9}19.

PRIMARY PRODUCTION PROCESSES

J. A. Raven, Biological Sciences, University ofDundee, Dundee, UK

Copyright ^ 2001 Academic Press

doi:10.1006/rwos.2001.0202

Introduction

This article summarizes the information availableon the magnitude of and the spatial and temporalvariations in, marine plankton primary productivity.The causes of these variations are discussed interms of the biological processes involved, theorganisms which bring them about, and the rela-tionships to oceanic physics and chemistry. Thediscussion begins with a deRnition of primaryproduction.

Primary producers are organisms that relyon external energy sources such as light energy(photolithotrophs) or inorganic chemical reactions(chemolithotrophs). These organisms are furthercharacterized by obtaining their elemental require-ments from inorganic sources, e.g. carbon frominorganic carbon such as carbon dioxide and bicar-bonate, nitrogen from nitrate and ammonium (and,for some, dinitrogen), and phosphate from inorganicphosphate. These organisms form the basis of foodwebs, supporting all organisms at higher trophiclevels. While chemolithotrophy may well have had avital role in the origin and early evolution of life, therole of chemolithotrophs in the present ocean is mi-nor in energy and carbon terms (Table 1), but is veryimportant in biogeochemical element cycling, forexample in the conversion of ammonium to nitrate.

2284 PRIMARY PRODUCTION PROCESSES