Although the principal merit of a rigorous protocol, namely the comparability of data determined by different groups and in different environments, is a highly worthy one, the problems outlined above dictate that such rigour would be presently inappropriate. The protocols presented here aim to (i) constrain procedures to an extent that sediment trap data comparability between different deployments and laboratories is improved, and (ii) provide realistic estimates of downward particulate fluxes. The broad consensus protocol is given, supported by comments summarizing other recommendations, guidelines and considerations (many aspects of which are described in more detail by US GOFS Working Group, 1989). It is intended that the flexibility within these protocols be further reduced, according to the consensus opinion of participating groups, during coordinated studies where comparability of data sets is paramount e.g. region-specific process studies, studies of mesoscale variability.

2.0 Scope and Field of Application
The sediment trap technique may be used for the collection of downward-settling particulate matter in the ocean water column. Analysis of the material thus collected enables (a) the estimation of the downward particulate fluxes of a wide variety of chemical and biological components and (b) the elucidation of the qualitative nature of these components. Traps can be deployed throughout the water column moored either to the seabed or to drifting surface buoys, and samples collected over time periods, and through time-series, of up to the order of one year. Temporal and spatial information is thus accessible.

3.0 Definition
The downward particulate flux of a component is defined as the quantity of that component settling through a given horizontal area in a given time. This flux is expressed in dimensions of quantity per horizontal area per time. A sediment trap provides an estimate of this flux by collecting, during a measured time period, material settling to the bottom of the trap having entered through a trap opening of known area. A number of physical, biological, chemical and hydrodynamic factors, reflecting both natural oceanographic processes and artifacts induced by the trap/mooring and sample treatment/ analysis, can affect the accuracy of this estimate (summarized here, and in detail by US GOFS Working Group, 1989). Flux estimates are therefore dependent to some extent on choices of trap/mooring design and hardware, deployment methods, sample treatments,
analytical procedures and data interpretation.

4.0 Principle of Analysis
Sediment traps can be used to collect settling particles. There is considerable evidence that much of the mass of particulate material transported from surface waters to the deep ocean and ocean floor is in the form of large, fast-settling particles. The mass of material collected should thus be dominated by this particle component. Information derived from the analysis of such samples can be used to help identify and quantify the chemical, physical and biological processes affecting and influenced by downward fluxes.

5.0 Apparatus
Protocol sediment trap designs are to take into account the hydrodynamic effects of trap geometry (aspect ratio) and the configuration of the baffles at the trap opening. Guidelines are not rigid (US GOFS Working Group, 1989). All moorings should be designed to maintain vertical trap orientation and are to be instrumented with pressure and flow sensors at trap depths. Drifting arrays should further be designed (i) to minimize current
flow relative to each trap (so as to reduce hydrodynamic interference), (ii) so that the surface buoy is not wind-driven, and (iii) to achieve effective decoupling of the traps from surface waves.

Comments: In addition to flow and current sensors, tilt meters on each trap are recommended. Mesh screens below trap baffles, proposed for the exclusion of large “swimmers” (section 8.2), are not recommended for routine deployments without further justification (US GOFS Working Group 1989).

6.0 Reagents
No specific recommendations other than described in sections 7.4.1 and 8.6. Reagents used in trap solutions and sample treatments should be of a quality that does not contaminate with intended analytes or components that interfere with their analysis.

7.0 Sampling
7.1 Deployment Environment: Avoid high current environments for fixed moorings and high current shears for drifting arrays; present state-of-the-art mooring and trap technology is inappropriate for such conditions. Minimize flow relative to drifting traps by careful mooring design (especially when shears are high). Acceptable flow limits cannot be generally defined for all trap designs, but relative flows of several tens of
cm/second are not acceptable (Baker et al. 1988). Groups simultaneously deploying drifting arrays as part of site-specific studies should use the same trap depths selected according to oceanographic features. Fixed mooring trap depths in the upper water column should be limited by the flow environment. International reference depths to be at 1000 m below surface (if possible) and deeper 1000 m interval horizons.

7.2 Temporal Resolution: The finest temporal resolution that allows a continuous time-series throughout the study period and provides sufficient material in each sample for the intended analyses is the ideal. Likely diurnal and seasonal variability should be considered when deciding resolution for drifting and fixed traps respectively.

7.3 Trap Solutions: Sample cup solutions are designed to preserve collected material (including swimmers) and to reduce diffusive, advective and resuspension losses of sample cup contents. This is attempted by using seawater dosed with preservative and NaCl (to provide a density discontinuity relative to the ambient seawater). The solution is prepared by dosing seawater (from deployment depth or filtered surface
water) to a final concentration of 2% buffered formaldehyde (5% buffered formalin) and a 5psu excess salinity. Formalin is buffered by saturation with borate. An aliquot of cup solution is retained for blank corrections.

7.4 Comments

7.4.1 A preservative is essential for long-term deployments. Its use is less attractive for short-term (few day) deployments where organic carbon degradation may induce less error than artifacts from preservative use. However, short-term deployments are usually in shallow waters where swimmer contamination (section 8.2) is often a major problem. In such a case the use of a preser-vative/ poison is contentious; large amounts of dead/fragmented swimmers may pose a greater problem than allowing swimmer activity and organic mat-ter
degradation.

7.4.2 Formaldehyde appears to be the most effective and suitable general purpose preservative of those tested (e.g. Knauer et al. 1984), and is recommended until more viable alternatives are proven. The major drawback of formaldehyde is that it precludes the accurate measurement of dissolved natural C lev-els,
and thus prevents the determination of particulate C leached into solution. Although it has previously been held that the non-carbon based alternatives, poisons such as Hg salts and azides, are not such effective inhibitors of the degradation of organic matter, recent evidence (Lee et al. 1992) suggests that mercuric chloride and sodium azide can be as effective as formaldehyde in this respect. These poisons have the advantage of allowing dissolved natural C determinations, but are less effective than formaldehyde at preventing swimmer fragmentation and may cause difficulties with trace metal contamination.

7.4.3 The salinity enhancement is recommended, despite certain potential drawbacks such as particles not settling through the density discontinuity and chemical effects inducing enhanced leaching. The relative importance of such drawbacks compared to the benefits (see above) of a salinity enhancement has not been demonstrated for field deployments. Opinions are divided; in the interests of consistency and comparability, and pending further investigations to resolve this question, the use of a salinity enhancement of 5psu, as
per the previous JGOFS protocols, is recommended at least for long-term moorings aiming to contribute to the global flux database. Excess density solutions should exist only in the sample collection area owing to their effect on trap aspect ratio (section 4; US GOFS Working Group, 1989). For short-term study-specific deployments, individual laboratories and collaborating groups should decide. Non-particle reactive components in trap solutions may provide useful information in quantifying diffusive and advective losses.

8.0 Post-collection Procedures
8.1 Handling and Storage: Samples are to be isolated under non-contaminating conditions  (dependent on intended analytes) immediately following trap recovery. The sample cup solution supernatants are sub-sampled and stored as is appropriate for the analysis of components that may have leached from the collected particulates (section 8.5). Storage prior to separation of particles from solution should be under
refrigeration in the dark.

8.2 Description and Swimmer Picking:

8.2.1 The wet sample is inspected and qualitatively described using an optical dissecting microscope (magnification up to about x50). Swimmers - those organisms deemed to have actively entered the trap - must be removed. Swimmers may be picked out with forceps during microscopic inspection. Alternatively, samples may be first screened to remove large swimmers, followed by microscopic inspection and picking of both fractions to ensure that (a) swimmers smaller than the mesh size are removed, and (b) that non-swimmer
particles retained by the sieve are not removed. Picked swimmer data are to be recorded as organism category, sizes, numbers, and an estimate of total swimmer volume as a fraction of sample volume.

8.2.2 It is recognized that to some extent a swimmer is operationally-defined and dependent in some degree on the sample type. The general recommendation is to remove intact, recognizable zooplankton greater than 330 microns in size. Beyond this, laboratories should consider potential contamination by “cryptic swimmers” (Michaels et al. 1990) - swimmers that are difficult to see or remove, or structures such as feeding webs brought into the trap by a swimmer - and the effects of removing zooplankton that may be part of the passive flux. The identification and removal of small and cryptic swimmers is particularly difficult if samples are filtered or centrifuged before picking; separation of particles from solution prior to picking is thus not recom-mended.
Individual laboratories should take responsibility for swimmer identification strategy based on sample type.

8.3 Subdivision of Samples:

8.3.1 When a trap sample is intended for several analyses, subdivision must produce subsamples that are compositionally representative of the original sample. A rotating, high precision plankton splitter has been shown to be effective for wet samples (Honjo 1978). Other methods can be used if their precision is demonstrated.

8.3.2 The impact of rare large particles on split precision with respect to certain components should be considered.

8.4 Particulate Components to be Analyzed:

8.4.1 Determination of dry mass, total carbon, organic carbon, inorganic carbon, total nitrogen, total phosphorus, total silicon.

8.4.2 Additional analyses might include biogenic and non-biogenic silicon, organic and inorganic nitrogen, aluminium, appropriate radionuclides, stable isotopes, organic biomarkers, lithogenic components, trace elements, other major elements, detailed microscopic examination, etc.

8.5 Dissolved Components to be Analyzed

8.5.1 The analysis of trap solution supernatants (section 8.1) and sample cup solution blanks (section 7.3) for all components analysed in the particulate phase,plus pH. The solution phase should also be sampled when the particles are isolated, and similarly analysed. Organic carbon (and possibly nitrogen) con-tamination
from the formalin preservative precludes the analysis of this (these) component(s).

8.5.2 Different dissolved species (e.g. N and P species) may be differentiated if speciation changes due to sample handling and storage artifacts can be assessed. Given such solution phase analyses, corrections for components leached from collected particulates can be applied if it is assumed that dissolved material is not lost from the high density solution during deployment, and with the caveat that the impact of swimmers on solution phase composition should be considered. The determination of natural dissolved organic carbon is not presently feasible in the presence of formalin. An organic carbon leaching correction cannot therefore be determined. However, estimates of carbon fluxes using preservatives are generally considered to be more reliable than those measured without preservation (section 7.3). pH determination checks effectiveness of preservative buffering and enables assessment of the possibility of significant carbonate dissolution.

8.6 Analytical Methods

8.6.1 Mass determination is by filtration onto preweighed 0.2-0.5 mm filters, removal of sea salt by buffered (pH 7-10) rinsing, drying at < 60°C and reweighing. Alternatively, filtered solids can be resuspended in the rinse solution in a preweighed bottle, freeze-dried and reweighed.

8.6.2 Particulate total carbon and nitrogen are determined by the detection of the gaseous by-products of high-temperature combustion of weighed subsamples of similarly filtered, rinsed and dried samples.

8.6.3 Particulate organic carbon is measured by the same high-temperature combustion method following the removal of the inorganic carbon.

8.6.4 Methods to be used for the determination of other analytes are as given in the JGOFS protocols, where appropriate; otherwise according to the judgement of individual and collaborating laboratories and programs. The community should work towards establishing common analytical protocols as methodologies
improve. The state-of-the-art of particle analysis is described by Hurd and Spencer (1991).

8.7 Comments:

8.7.1 Mass determinations should aim for the rinsing solution to be isotonic with respect to seawater to minimize potential mass losses owing to cell lysis.

8.7.2 Of the several methods for inorganic carbon removal, none reliably remove 100% of the inorganic phase whilst leaving 100% of the organic phase. Whilst it is recommended that the inorganic carbon is removed by direct treatment with, preferably, sulphurous acid (Verardo et al. 1990), owing to the less oxidizing nature of this acid compared to HCl, it is realized that until existing and new methods are further assessed and improved, some laboratories will prefer to continue fuming samples with HCl. Care should be taken
not to lose organic carbon that is solubilized by the direct acid treatments. Methods used should be reported when presenting data.

9.0 Calculation and Expression of Results
Downward particulate flux estimates should be expressed in units of quantity per horizontal area per time. Attempts should be made to correct data for dissolution of collected particles using dissolved phase and cup solution blank concentrations and swimmer data (sections 8.2 and 8.5). All reported sediment trap data should be accompanied by reports of: trap geometry and baffle configuration; current and other mooring sensor data summaries; trap solution preparation methods; swimmer removal criteria, procedures and estimated volume contribution to total sample; sample subdivision method; analytical methods; details of corrections for dissolution of collected particles and possible leaching from swimmers; estimated precision of sample processing and analytical procedures.

10.0 Quality Control/Quality Assessment
In order to improve quality control and quality assessment of sediment trap data, efforts should be focussed on: mooring sensor development and sensor data interpretation (section 5); trap and mooring design and development; replicate traps; horizontal-scale variability of fluxes; the verification and application of trap calibration by radionuclide trapping efficiency integrated over long time-series’; swimmer prevention; swimmer
removal and impact assessment; precision of sample subdivision methods; suitability, accuracy and precision of analytical methods; estimated overall precision of sample processing and analytical procedures. An area of particular concern is the compromising of samples by swimmers. Efforts should be directed towards their exclusion from traps during deployment (e.g. Coale 1990). If swimmers do contaminate a sample, an assessment of their likely impact should be attempted e.g. within-trap solution activity, effects of leaching, effects of activity of swimmers colonizing trap but not trap solution. The difficulty of swimmer impact
assessment necessitates the reporting of swimmer data (sections 8.2, 9).

11.0 Intercomparison
Intercomparisons should be encouraged in the areas of: trap design; mooring design; sample processing; analytical methods. Increased collaboration between groups in planning deployments and discussing methods is desirable. Some such intercomparisons are presently underway, but there is scope for much further work. The offer of S. Honjo (WHOI) to provide a bulk homogenized “standard” of sediment trap material to the
international community for analytical methods intercomparisons and intercalibrations should be taken up; the participation of a large number of laboratories in such a comparison would be invaluable.

12.0 Notes
At all stages of the sediment trapping experiment (trap and mooring design, trap preparation, deployment, sampling, recovery, storage, sample handling and analysis) attention must be paid to avoiding contamination of the samples with intended analytes or components interfering with their analysis.

13.0 References
Baker, E. T., Milburn, H. B. and Tennant, D. A. (1988). Field assessment of sediment trap efficiency under varying flow conditions. J. Mar. Res. 46: 573-592.
Coale, K. H. (1990). Labyrinth of doom: A device to minimize the “swimmer” component in sediment trap collections. Limnol. Oceanogr. 35: 1376-1381.
Honjo, S. (1978). Sedimentation of materials in the Sargasso Sea at 5367m deep station. J. Mar. Res. 36: 469-492.
Hurd, D. C. and Spencer, D. W. (1991). Editors; Marine Particles: Analysis and Characterization. Geophysical Monograph 63. AGU. Washington DC. 472p.
Knauer, G. A., Karl, D. M., Martin, J. H. and Hunter, C. N. (1984). In situ effects of selected preservatives on total carbon, nitrogen and metals collected in sediment traps. J. Mar. Res. 42: 445-462.
Lee, C., Hedges, J. I., Wakeham, S. G. and Zhu, N. (1992). The effectiveness of various treatments in retarding bacterial activity in sediment-trap material and their effects on the collection of swimmers. Limnol. Oceanogr. 37:117-130.
Michaels, A. F., Silver, M. W., Gowing, M. M. and Knauer, G. A. (1990). Cryptic zooplankton “swimmers” in upper ocean sediment traps. Deep Sea Res. 37: 1285-1296.
US GOFS Working Group (1989). Sediment Trap Technology and Sampling. Planning Report No. 10. November 1988. WHOI, USA.JGOFS Protocols—June 1994 201
Verardo, D. J., Froelich, P. N. and MacIntyre, A. (1990). Determination of organic carbon and nitrogen in marine sediments using the Carlo Erba NA-1500 Analyser. Deep Sea Res. 37: 157-165.