Sediment Trap Technology and Sampling in Surface Waters

Edited by Wilford D. Gardner
Department of Oceanography
Texas A&M University
College Station, TX 77843-3146
wgardner@ocean.tamu.edu

Introduction

There is a wide range of possible trap designs and environmental conditions for their use. Protocols for the use of traps have been outlined twice (U.S. GOFS Report No. 10, 1989; IOC Manual and Guides No. 29, 1994), but have not always been followed because of differing opinions and our inability to devise unequivocally accurate calibration schemes that can be routinely followed for surface waters. Thus, we have not progressed to the point of the CO2 measurements in establishing fixed, widely accepted and adhered to protocols.

The objective of this report is to assess the present status of particle trapping in the upper water column and suggest plans on how to:
  1. Estimate the magnitude of errors in trap and trap-related measurements;
  2. Resolve differences in fluxes measured by different methods in the upper 200 m;
  3. Establish which ancillary measurements are needed in future trapping experiments; and
  4. Verify which protocols should be used for traps.

To accomplish this objective, we list and discuss some of the potential causes of biases in using traps in surface waters and estimate the magnitude of their importance. We acknowledge disagreement in this assessment because we have not been able to adequately quantify the errors under all conditions. While many of the issues in this report were discussed in the U.S. GOFS Report No. 10 (1989), half of the references in this report were published after the GOFS report and they demonstrate substantial progress. This report is an update of the major issues with a narrow focus on using and calibrating traps in the upper 200 m of the water column.

As a unifying theme, the discussion centers around the potential impact of each parameter on the carbon imbalance reported by Michaels et al. (1994) at the Bermuda Atlantic Time-Series station (BATS), and to a lesser extent some comparable carbon budgets made at the Hawaii time-series station (HOT). The format of this report is mostly in the less orthodox form of an outline rather than straight prose in an attempt to cut to the heart of each issue.

BATS carbon imbalance: At BATS a one-dimensional (vertical) mass balance was constructed for carbon during the April-December period when sediment traps were deeper than the mixed layer depth. The time-course of total suspended carbon (DIC+DOC+POC) was compared with the balance of all vertical fluxes (measured or estimated). The carbon changes in the water were 3 times greater than the balance of the fluxes. If the discrepancy were entirely due to undertrapping, the traps would have had to collect 6 times as much material. Alternately, advection or vertical migration of zooplankton could account for some of the difference.

HOT carbon balance: At HOT, two different 1-D models predict a carbon export from the upper ocean of approximately the same rate as the measured carbon flux in traps (without accounting for sample dissolution). Not included in this mass balance was an estimate of the carbon flux due to vertical migration of swimmers or the fluxes of DOC. Thus modifications of our sense of trap accuracy to bring the trap estimate more towards addressing the imbalance at BATS might create a disagreement between carbon fluxes and model estimates at HOT.

In constructing a carbon balance, we must remember that carbon can be removed from surface waters by several processes; gravitational settling of particles, vertical migration of zooplankton, vertical mixing of DOC, DIC and POC, advective transport that creates a horizontal gradient, and gas exchange of CO2 with the atmosphere. Traps are intended to collect only the settling particles. Sources of errors in trap measurements include; swimmers, solubilization of carbon within the trap and hydrodynamic effects that include trap geometry, flow, wave-induced trap motion, tilt, and the effects of brine inside the trap.

The central question is whether differences in carbon budgets arise because we don't understand the behavior of the instruments we are using (traps and in-situ filtration systems) or because we don't understand the dynamics of the system (particle fluxes and dynamics in surface waters, radionuclide distributions and interactions with a wide range of particle types, spatial scales etc.) or both?

For each issue there are comments made at the meeting or in response to the draft report, notes on the magnitude of the problem, and recommendations of what to do in the future. Names in parentheses indicate the person responsible for the comment.

I. Possible Trap Biases

A. Swimmers

Comments
  1. Protocols for picking swimmers have been established (IOC Manual and Guides, 1994)
  2. Keep in mind that some swimmers may be part of the signal (Knauer, Jim Murray; Silver et al., 1984)
  3. Swimmers have low Th content (Coale, 1990; Buesseler et al., 1994)
  4. Swimmer-avoidance traps
    1. Coale (1990) - A cylindrical trap with internal funnels and collars designed to separate the active swimmers from the trap sample. It was used in VERTEX as a cod end on Soutar cone traps. It is commercially available, but the only published report of its use appears to be Lee et al., 1988. It removed 25-70% of the swimmers from the sample based on amino acid content, but there were no comparisons made simultaneously with other traps.
    2. Peterson et al. (1993) - A cylinder/cone trap with an indented rotating sphere (IRS) separating upper and lower trap segments. The device is designed to eliminate swimmers from lower regions of the trap and to isolate the sample. On average, the flux of material >850 µm was reduced by 88% relative to identical control traps without the sphere. At the same time, the flux of material < 850 µm decreased to 59-97% (ave 84%) of the flux in identically-shaped control traps without spheres. The C/N ratio of material in the IRS traps was about 10, whereas it was about 8 in the control traps suggesting some differentiation in the type of material passing around the sphere. During the JOGFS EqPac program the IRS traps collected much less material than the cylindrical traps of Murray et al (submitted) when they were surface tethered, but the moored IRS trap fluxes were more similar to the fluxes measured with other moored traps (Cindy Lee).
    3. Hansel and Newton (1994) - A variation of the Coale trap, it relies on random motion of swimmers to trap them away from the sample chamber. It had a 70% exclusion efficiency for copepods in coastal waters, but only 37-72% efficiency for copepods in open ocean waters. Hansell suggests some modification before these traps are used in the ocean. This trap has been used to determine that only 7% of the trapped organic carbon was remineralized to DOC in Monterrey Bay in 1.7 days.
  5. In the Mediterranean, pteropods are regularly and abundantly found in the traps (Miquel et al., 1992, 1994, 1995). In principle, shells with the animal inside do not belong to the trap flux whereas empty shells do contribute to the passive vertical flux. (See Harbison and Gilmer, 1986). Most of the shells are "full" at upper depths (200 m) but deeper (1000 m), a good fraction (sometimes up to 1/2) are empty. On the other hand, if the empty shells are left in the flux, their contribution to the carbonate flux will mask any other qualitative and quantitative change on carbon fluxes with seasons or years. So far, we have considered the pteropods as swimmers and we have a numerical estimate of their importance in the traps, but we are looking for better approaches if they exist (Juan Carlos Miquel).
Magnitude of problem
  1. Decreases with depth (Lee et al., 1988).
  2. Even after manual removal of zooplankton under a strong dissecting microscope, much of the remaining carbon in very shallow traps is still swimmer or swimmer derived particles. This cryptic swimmer problem is described in Michaels et al. (1990); it averaged about 50% of the collected carbon, with a maximum value of 96% of the carbon in a shallow, picked trap, but the problem decreases rapidly below 200 m.
    • Are these data available from the VERTEX project? (Richard Lampitt).
        Some are available (Michaels).
  3. Karl and Knauer (1989) attempted to circumvent the swimmer problem by using combinations of screened and unscreened traps to calculate a swimmer-free flux.
  4. See US GOFS Report (1989).
  5. A preliminary comparison of picking methods between HOT and BATS found a 50-100% difference in total carbon. The screening technique used at HOT was employed on replicates of the BATS traps and the resulting carbon estimate with the HOT technique was higher than the traditional BATS technique.
Effect on BATS/HOT carbon imbalance/balance
  • Swimmer errors in poisoned traps usually bias traps towards higher fluxes. If more swimmers were picked, trap fluxes would be even lower and the BATS imbalance larger. Could some of the organisms classified as swimmers actually be part of the true flux? Could swimmers in traps be compensating for traps missing the vertical migration flux (Michaels).
  • Removing swimmers also removes attached non-swimmer mass which tends to underestimate true fluxes. There is a cross-over point somewhere: no swimmer removal overestimates flux, complete swimmer removal and the attached particles underestimates flux. The "null" point will vary with size spectrum and, perhaps, species present. A most difficult problem indeed. (Dave Karl)
  • If HOT traps were processed like BATS traps, the flux estimate would likely be somewhat lower, creating a difference with the modelled fluxes (Michaels)
  • If some of the residual carbon in traps at either location is due to swimmers, the fluxes are still further from the independent measurements of flux.
  • Could swimmers be eating and leaving?
    • Not likely to be significant if the trap contains poison.
    • Lee et al. (1987) found a 43% loss of carbon in unpoisoned traps at 3 m in a shallow lake and attributed the loss to zooplankton feeding in the traps. The loss was only 1-3% at 8-10 m depth.
Recommendations
  1. All investigators should report how swimmers were removed and quantify their abundance.
  2. Some intercomparisons of swimmer removal techniques is feasible and should be done.

B. Solubilization of particulate matter in the trap

Comments
  1. Organic carbon is lost with time (Lee and Cronin, 1982; Gardner et al., 1983; Knauer et al., 1984; Lee et al., 1992), but decay components can be retained in the brine. Knauer et al. (1984) argued that the quantification of these components (e.g. phosphate) can be used to estimate carbon loss. Dennis Hansell points out, however, that this is true only if the decay products are unique to the sinking particles. If these decay products are also found in herniating swimmers, then the source cannot be identified uniquely. One would also have to know the percentage of DOC release per unit POC from swimmers and sinking particles.
  2. Peterson and Dam (1990) demonstrated that the addition of brine to a trap will cause zooplankton to herniate. Hansell and Newton (1994) found a factor of 10 difference in DOM accumulation between brine and brine-free trap solutions, which they attributed to herniation of swimmers. In deployments of a swimmer-segregating trap (1.7 days), the quantity of DOC released caused only a 7% decrease in the total POC flux compared with standard PITS traps (Hansell and Newton, 1994).
  3. Lee et al. (unpublished data) measured DOC in their EQPAC IRS traps that have reduced swimmer content. It appeared to make up 10-20% of the C flux. This may still a problem, but probably more for those interested in specific organic compounds (Cindy Lee).
  4. Measurements of DOC and DIC in sediment traps obviously have not found a wide acceptance and only very few examples of DOC measurements have been published (Hansell & Newton, 1994). DIC measurements in sediment trap samples have not been carried out or published at all to our knowledge. Besides the additional work burden, theoretical considerations restrict the completeness of any DOC+DIC correction of POC flux measurements. Degradation of POC to DIC in sediment traps would increase the pCO2. This increase would give rise to loss of DIC from the sample prior to the DIC analysis. Excess DIC (above ambient values) would thus be a minimum estimate of carbon lost from the POC sample to the water of the sediment trap cups. The problem of the origin of this excess DIC (sinking particles vs. swimmers) will be similar to that of excess DOC (Section I.B.1). For routine measurements of DOC additional problems arise for those samples poisoned with formaldehyde (the poison most recommended from the JGOFS-protocols; IOC- manual, 1994). The DOC introduced by formaldehyde appears to be a very large background signal, that will not allow detection of excess DOC at a reliable level. (Wolfgang Koeve)
  5. The `Kiel Particle Flux Group' suggests furthermore that not only should the measurement of DOC and DIC in sediment trap cups be optimized, but one should also be aware of possible dissolution of other components. Excess phosphate has been observed frequently (Bodungen et al., 1991) and should be monitored like other constituents of interest to the respective program (e.g. trace metals, amino acids, fatty acids). (As a caveat, see comment 1. above.) If we believe that excess DIC is a significant source of error for our POC flux estimates, dissolved excess Ca also should be monitored to control our estimates of PIC fluxes. It will depend on the precautions carried out to control or even increase the buffer capacity of the seawater in the sediment trap cups (examine effects of different poisons: buffered formaldehyde vs. HgCl2, sodium azide etc., Lee et al., 1992) whether or not dissolution of POC to DIC and the subsequent increase of the pCO2 in the sample give rise to the dissolution of PIC for a given sample. Furthermore, without monitoring excess Ca in the sediment trap cups one would not be able to decide whether any excess DIC should be added to the POC or PIC flux estimate. (Wolfgang Koeve)
  6. Obviously, reliable detection of excess DOC and DIC in sediment trap samples relies on
    1. improving the swimmer avoidance
    2. using other poisons than formaldehyde
    3. control the whole carbonate system in the trap samples
    4. measure and understand the artificial losses of dissolved tracers from sediment trap cups and
    5. model the loss of DOC, DIC and other related dissolved components from sediment trap cups (Wolfgang Koeve)
  7. Experimental lab and field tests carried out showed that losses of dissolved compounds are small for the `Kiel Sediment Traps'. Investigation of losses of the supernatant sodium azide concentration, used as poison during these experiments, was carried out in a series of 11 sediment traps (up to 20 samples each) from moorings deployed over recent years in the North East Atlantic. Deployment period was one year. On average, blank bottles which were not exposed to the open funnel showed losses of 7.5% of the initial value of the poison. Higher loss values (up to 20%) occurred in the sample bottles, which were exposed to the funnel for 8 to 28 days. These higher losses can be explained by means of reaction of the poison with particles, diffusion, swimmer activity and probably turbulent mixing events (Lundgreen et al., in prep). This source of uncertainty needs to be evaluated also for other sediment trap designs (VERTEX, Soutar, Honjo-traps, etc.). Since it will highly depend on the hydrodynamic environment of the sediment trap for a given experiment or field study, regular measurements are recommended. Tracers other than the poison sodium azide need to be discussed, Kremling and Schuessler from Kiel recently used 22Na as a tracer in another study. (Wolfgang Koeve & Ulrich Lundgreen)
  8. Brines also "purge" interstitial fluids of particles that could contain substantial amounts of nutrients (Karl et al. 1984) and biogenic gas (Karl and Tilbrook, 1994).
  9. The VERTEX group found that dense NaCl brines cause CaCO3 to dissolve (Honjo et al., 1992), so they switched to a NaCl, MgCl2, CaCl2, KCl mix in making their brine.
  10. Sample cups in a JGOFS trap mooring with funnel traps lost 20% of the original sodium azide. The loss was believed to have resulted from swimmer activity rather that diffusion or currents (Ulrich Lundgreen). Dissolved components would have been lost also.
  11. Jim Murray didn't see a DOC difference in a two day trap deployment., but Dennis Hansell and Jan Newton report measuring large DOC signals in numerous trap deployments in Monterey Bay. If traps are deployed without brine, DOC could easily diffuse from the trap and would not be detected.
  12. If C, N, P, etc. are lost due to herniation, maybe we should redefine the whole biogeochemical cycle in terms of Si, which has a much smaller magnitude problem with swimmers. If the carbon imbalance problem finally falls on swimmers, perhaps this is a viable solution, or at least a test of the hypothesis (Dave Karl).

Magnitude of problem

  • Depends on length of deployment and what poisons/preservatives are used.
  • Lorenzen et al. (1981) reported a 7% loss per day for 5 days using sediment trap material. Iturriaga (1979) reported an 8% loss per day for zooplankton and 3% lost per day for phytoplankton in 15°C water. Gardner et al. (1983) measured daily losses of 0.1% to 1% per day extended over 106 days in deep traps, and showed that if carbon losses were much greater than 1% per day for long-term, multiple cup trap experiments, seasonal cycles would not be discernable. In the upper water column carbon losses were about 7% in 1.7 days (Hansell and Newton, 1994).
  • In some earlier studies, carbon fluxes were increased by a factor of 2.23 to "correct" for the presumed dissolution in traps. This is no longer done.

Effect on BATS/HOT carbon imbalance/balance

  • The herniating of swimmers would increase carbon in the supernatant even if they are picked out.
  • Hansell and Newton (1994) found only a 7% difference between total organic carbon and particulate organic carbon after a 1.7 day deployment in Monterey Bay.
  • This isn't large enough to solve the carbon imbalance (Michaels), though some of the more extreme correction factors (e.g. the 2.23 fold VERTEX correction) would move the measured flux much closer to the required export.

Recommendation:

  • Several measurements indicate that the carbon loss by solubilization is at most a few percent a day in unpoisoned traps. Unless swimmers are prevented from entering the trap, most of any excess DOC signal in trap supernatant water is derived from herniating swimmers. If trap samples are well-picked for swimmers, the apparent magnitude of the solubilization problem appears to be smaller than previously assumed.

C. Hydrodynamic biases

Comments:
  1. One of the major approaches in evaluating the efficiency of sediment traps has been to calibrate small models of traps in flumes or tanks where the conditions of sedimentation can be controlled and measured (Hargrave and Burns, 1979; Gardner, 1980a; Butman, 1986). The experimentally confirmed assumption has been that in the absence of any current, cylindrical traps accurately intercept the material settling out of the water above the trap. In the presence of velocities up to about 10 cm/sec, cylindrical traps still collect particles at the rate predicted ±30-50%. The predicted rate is determined from the loss of particles above the trap (Hargrave and Burns, 1979; Gardner, 1980a) or the measured settling velocity of the particles used in the calibration (Butman, 1986). In some flume experiments conducted by Gust et al. (in press), VERTEX style cylinders show significant increases in flux between velocities of 5 and 10 cm/s. However, several reviewers versed in fluid dynamics argue that their methods are flawed because they inject particles into the trap through a tube rather than allowing particles to be intercepted by the trap and collected naturally. It is difficult to test traps in flumes at velocities much higher than about 10 cm/sec because settled material is resuspended from the flume bottom and has a second chance to enter the trap.
  2. Moving to the field, one continues to make the reasonable assumption that cylindrical traps still accurately intercept the material settling out of the water above the trap if there is no movement past the trap, though independent verification of this assumption would certainly be desirable. (See section III below on INDEPENDENT MEASURES OF VERTICAL FLUX.) The effects of large-scale turbulence, internal waves, tilt and mooring line motions that are not present in the flume have unknown effects on the comparison of flume and field data. Comparisons have then been made between cylindrical traps and traps of other designs such as funnels that are deployed simultaneously (Honjo et al., 1992). Comparisons of fluxes measured with drifting and moored traps of the same design at the same location can also be used to determine the efficiency of traps under different flow conditions. Baker et al. (1988) have done precisely that experiment.
  3. Baker et al (1988). show that relative trap efficiency in moored, cylindrical traps with a steep inner funnel in the field drops to 20-25% efficiency somewhere in the velocity range of 12-30 cm/sec compared to the same design of trap when it is drifting with the water in the same place over the same time. The corresponding trap Reynolds numbers (Rt = (velocity*trap diameter)/viscosity) for the 20 cm diameter Baker trap are 24,000-60,000. The Rt or range of Rt over which the flux decreases is unknown. The composition of particles was also very different at higher Reynolds numbers. Field experiments by Gardner et al. (submitted) showed no decrease in trapping efficiency up to a Reynolds number of 43,000 in cylindrical traps with an interior funnel moored at 4500 m. If Reynolds number is the controlling parameter in trap efficiency, this suggests there should be no decrease in relative trapping efficiency for the 7.6 cm diameter BATS traps up to a velocity of 32-57 cm/sec, which is faster than any of the velocities reported for those traps in the BATS area by Gust et al. (1992, 1994). However, based on personal observations of flow within model traps at just 20 cm/sec, it is hard to believe that the collection efficiency of traps is not affected at velocities above 20 cm/sec (Gardner). We must remind ourselves that although we can scale the dynamics of flow between models and full-scale traps using dimensional analysis, we may not be able to predict the dynamics of natural marine particles at different scales. Furthermore, in the dynamic region of the upper ocean one must consider whether other factors such as large-scale flow, tilt and mooring dynamics are more important than Reynolds number in applying these results to regions like BATS and HOT (Gardner).
  4. Neutrally buoyant traps have been recommended (U.S. GOFS, 1989) and tested (Honjo, personal communication, 1980; Diercks and Asper, 1994), but have not been widely used due to engineering difficulties. Neutrally buoyant traps are not practical for all environments, but they are well-suited to near-surface deployments in the open ocean. Their use could significantly reduce questions about hydrodynamic effects on trapping efficiency. One must also have information about how tightly coupled a neutrally-buoyant trap is to the surrounding water. Jim Price and Jim Valdes (WHOI) are making a neutrally buoyant trap based on the RAFOS drifting float design and hope to test it at BATS by the end of 1995. This should minimize the hydrodynamic questions, but leaves the swimmer, solubilization, and migration transport questions to be quantified simultaneously.
  5. For further discussion of trap dynamics see Hargrave and Burns (1979), Gardner (1980a,b; 1985), Blomqvist and Kofoed (1981), Butman (1986), Butman et al. (1986) and Hawley (1988).
  6. Gust et al. (1992) show a factor of two increase in flux with a doubling of velocity (in the 10-30 cm/sec range) past large funnel traps during one-day deployments at the BATS site.
    • Is this simply a velocity effect?
    • Could this be a tilt effect? -Tilted cylinder traps collect more than upright traps - 25% at 5 degrees, up to 200-250% at 30 degrees (Gardner, 1985).
    • Tilts measured by Gust et al. (1992) were less than 5°, but tilt effects on funnel efficiencies have not been tested.
  7. Michaels et al. (unpublished data) have measured velocities at the 150 m trap for nearly 4 years of the BATS program and for many of those samples looked in detail at the particle composition of the traps. When all the data are considered together, there is no significant trend with velocity. However, if high (>350 mgC/m2/day) and low (<350 mgC/m2/day) productivity stations are considered separately, there is an apparent pattern in collection of carbon with velocity. The collection differences are 2-3 fold higher with increased velocity over the range of 4-14 cm/sec approach velocities. In examining the major components of the carbon flux, there is no velocity pattern with the dominant particle type, marine snow (assuming this can be adequately identified in the trap). There is a strong velocity dependence for fecal pellets (10-fold differences in collection over the velocity range). Thus, the hydrodynamic effects on aggregates may be very different than more solid particles. Alternately, fecal pellets in traps could be due to swimmers and swimmer collection might well be a function of approach velocity (and hence the amount of water that passes through the trap mouth).
  8. One area that has received little to no attention is the behavior and integrity of aggregates in the eddies and flow generated in and around traps. It is also necessary to study how aggregates cross a dense brine interface inside traps. One of the difficulties is in being able to identify aggregates once they are collected in a trap. Jannasch et al. (1980) attempted to preserve aggregates by adding a polyacrylamide gel in the sample collection area so that they could be thin-sectioned and studied, but the viscosity of the gel was so large that aggregates rolled up like dust balls before they sank into the gel. (Jannasch, personal communication and observation, 1979).
  9. To minimize vertical motion use a spar for the first or only surface float. (Tony Michaels)
  10. A spar buoy is effective, but the VERTEX solution of using a string of small floats achieves the same purpose and is easier to handle. The intent of a spar buoy is to minimize water displacement per vertical displacement or to minimize the waterline area. The string of small floats has a small waterline area and functions like a "flexible spar buoy." (Vernon Asper)
  11. Gust et al., (1994) have measured significant vertical motion and wave energy on floating arrays of the VERTEX design that had a stretchable member.
  12. The VERTEX surface-tethered trap array has a stretchable member to dampen out waves. It is important to be sure that there is enough sub-surface flotation to prevent the stretchable member from ever becoming fully extended or the surface-wave energy and motion will be transmitted to the traps and may affect their efficiency. (Vernon Asper)
  13. The installation of a horizontal drag plate below the trap or at the bottom of the array can act as an effective sea anchor to reduce upward motion of floating traps (Gardner et al., 1985).
  14. Most of the drag on a trap array is from the mooring line itself. This can be minimized using very thin wire (Tony Michaels),
  15. To minimize flow past the trap, position the floats below the Ekman layer to maximize the drag in the vicinity of the traps. (Susanne Neuer)
  16. Traps at multiple depths make it difficult to minimize the flow past traps.
  17. Maybe we're too greedy. Perhaps we should use just a single trap per mooring. (Tony Knap)
  18. One trap per floating array is the approach used by scientists from the Kiel group in recent years. (Wolfgang Koeve)
  19. We keep making recommendations but is anyone listening? How can these be widely accepted and practiced? What is the penalty for non-compliance? (Dave Karl)

    Magnitude of problem

    • Up to a factor of 4-5 based on Baker et al. (1988) data, but potentially only at very high velocities (Rt). Potentially 2-3 fold biases at lower velocities based on field data (Michaels).

    Effect on BATS carbon imbalance/balance

    • This could account for the entire effect, but there is still significant uncertainty about flow effects where floating traps are deployed. If the Reynolds number arguments are relevant in surface waters, then flow effects should be small except at high velocities. The unpublished Michaels et al. data show a difference of a factor of two-three that may be attributed to velocity effects at lower velocities.
    • Approach velocities are likely to be in the same approximate range at HOT as at BATS based on similar drift patterns, array configurations and horizontal velocities. Thus, any ad hoc explanation for the BATS imbalance that involves a flow-based explanation should have the opposite impact on the comparison of the HOT data with the models at that site.

    Recommendations

    • Develop a neutrally buoyant drifting sediment trap (U.S. GOFS Report No. 10, 1989; Diercks and Asper, 1994).
    • More experiments of the type performed by Baker et al. (1988) should be made. The velocity bins need to be more narrowly limited, especially between 12 and 30 cm/sec (Rt = 24,000-60,000) since that is where the trapping efficiency decreased markedly in their study. Use Reynolds numbers when planning experiments and interpreting their results. One should also test funnel traps in this mode since they are one of the most widely used designs because they offer the advantage of large collection area and sample concentration.
    • Field experiments at the JGOFS sites should be done to compare arrays drogued to have different approach velocities to see if there are collection differences as traps are actually used in the field. Although this will only establish a relative accuracy pattern, it will allow a determination of flow impacts in the regime in which these experiments are conducted.

    • Minimize the flow past the traps.
      • Measure flux at a single depth per array and measure the flow.
      • Use thin lines to minimize drag.
      • Put the maximum drag near the trap depth.
      • Continue to decouple the trap from surface wave motion.
        • Continue to use a stretchable segment of mooring line above the trap and be sure it has a rapid response and is never fully extended.
        • Use a spar buoy as the surface float as this removes a lot of the short period wave motion from the array.
        • Use a horizontal drag plate below the trap to reduce upward motion (Gardner et al., 1985).
    • Measure the velocity past traps on all deployments.

     

    D. Effect of adding brine to traps

    Comments
    1. A 50 psu excess brine creates a much larger density difference than the 10-4 - 10-5 density units difference from seawater that exists for aggregates in the ocean (Alldredge and Gotschalk, 1988). Macintyre et al. (1995) have observed aggregates accumulating at density gradients in the water column and report that it can take from hundreds of seconds up to 3 hours for the pore water to exchange in the aggregates.
    2. Flume experiments show that the addition of brine to traps decreases the collection rate (Gardner and Zhang, in press).
      • Brine was 5 psu above ambient compared with a 50 psu excess brine used in BATS and HOT traps.
      • Undertrapping decreased as velocity increased. Efficiency was 54% at 5 cm/sec and 75% at 15 cm/sec.
      • Length of deployment and the sequence of velocities to which traps are exposed may be important in determining the magnitude of the effect. A high velocity at the beginning of the trap deployment could wash out the brine early on so it doesn't inhibit the collection rate as much.
    3. Three BATS field experiments showed 0, 25 and 60% higher carbon fluxes in traps without brine, but showed both increases and decreases in the flux of specific components of the flux. (Michaels).
    4. Scott Nodder (New Zealand) tested cylindrical traps (ID = 9 cm, aspect ratio = 10.6) on frames 3 m above the harbor floor for 24 hours filled with a 50 psu excess brine and found that they collected 2-3 times less material compared with traps that were partially filled with the same brine (equivalent to 1 or 3 trap diameters of brine).
    5. A change in any of the protocols at the JGOFS time-series station needs substantial justification that the new measurement will lead to an increase in the absolute accuracy and it would require an extensive period of simultaneous measurements using both the old and new techniques (e.g. simultaneous brine/no-brine experiments overlapping for a year to test for seasonal differences in hydrography and particle types). Thus, operators of the time-series stations are reluctant to modify one facet of the method (brine) before there is an independent method to determine that the new collection techniques actually result in a flux estimate that is accurate on some absolute standard. It is not considered worth the risk to go through the extensive extra effort of switching one facet to find that the new method without brine is still very inaccurate but for a very different reason than the brine effect (e.g. flow, large-scale turbulence or something else that is not adequately considered now) (Michaels).
    6. Two groups have discussed this issue (US GOFS, 1989, and IOC Manual and Guides, 1994) and recommended that traps be deployed with no more than a 5 psu excess brine only in the sample collection region of a trap (equivalent to one trap diameter in cylinders). Trapping programs should adopt this and the other JGOFS recommendations.

      Magnitude of the problem

      • Flume experiments with a 5 psu brine show up to a factor of 2 loss in the flux measured.
      • Field tests with and without a 50 psu brine in cylindrical traps have shown decreases in flux anywhere from 0% to a factor of 2-3. The increases in flux with floating traps in surface waters without brine have been 0-60% higher than with brine-filled traps.

      Effect on BATS/HOT carbon imbalance/balance

      • A brine effect could increase trap fluxes at BATS but given the brine experiments at the BATS site, the effect is large enough to account for about 10% of the carbon imbalance, or 20% of the projected undertrapping. However, experiments in flumes and field experiments at other locations find brine effects of as much as a factor of 2-3. If these experiments are relevant to BATS, they could account for about half of the carbon imbalance, or up to all of the projected undertrapping. The magnitude of the brine effect decreases with increasing velocity and with increased exposure time. The HOT traps use the same protocol and they report no carbon imbalance in their measurements. Any effect of brine on the carbon budget at one site (BATS) must be applied to other sites (e.g. HOT) assuming hydrodynamic conditions are similar.

      Recommendation

      • Follow the published protocols which call for a 5 psu brine only in the bottom one diameter equivalent of a cylindrical trap.
      • Make necessary tests to convert the BATS/HOT trap protocols to brine only in the bottom. These comparisons need to be done in the context of experiments to determine the absolute accuracy of sediment traps if they are going to be useful for causing a change in protocols at a time-series station (Michaels).

       

      III. System Dynamics Questions

      A. Vertical flux by zooplankton migration

      Comments
      1. The role of particle transport by vertically migrating organisms and respiration has been discussed for many years with little quantification because it is a difficult task Angel (1989).
      2. Traps are not designed to measure the effect of migrant transport. Other methods must be employed to quantify this process.
      3. The question of migrant transport does not directly affect the efficiency of traps, but it is an essential component when constructing mass balances in the upper water column for assessing the accuracy of floating traps. One must understand the dynamics of particles within the upper water column as well as the dynamics of sediment traps.

      Magnitude of the problem

      • Migrant transport measured in the North Atlantic by Longhurst et al. (1988) was 8-28% of the flux measured by BATS traps.
      • Migrant transport measured at the BATS site by Dam et al. (1995) was 18-70% of the carbon flux measured concurrently by floating traps.
      • Walsh et al. (1988) noted a recurring deep (>1000 m) particle flux maximum in MANOP annual sediment trap profiles. They concluded that as much as 50% of the flux measured in traps at 1500-1900 m bypassed or was produced below their shallower traps 500-1000 m. This is well below the zone of interest in this discussion, but suggests migrant transport is not isolated to surface waters. One would expect migrant transport to be largest in surface waters, where diel migration is well-documented.

      Effect on BATS/HOT carbon imbalance/balance

      • The migrant flux was included in the carbon balance at BATS. If the true migrant flux is much higher than that estimate, this could explain some of the imbalance. The larger the magnitude of migrant transport, the smaller the carbon imbalance at BATS.
      • Recent measurements suggest transport by migrating zooplankton is more important than previously measured.
      • Carbon imbalance was not implied by the comparison between the carbon budget and the particle fluxes at HOT, so external transport mechanisms like migrant transport are not needed to balance carbon budgets. If vertical migrant fluxes occur at HOT, then that system is again out of balance with the carbon budget. Longhurst and Harrison estimated that migrant fluxes in the oligotrophic Pacific were 0.8-2.5 mg N/m2/d, comparable to an annual carbon flux of 0.2-0.6 moles C/m2/y. This would increase the HOT annual flux by 22-66%. Is migrant transport less important at HOT than BATS, or are the measurements of migrant transport more likely to be on the low end of the range measured by Dam et al. (1995)? How does migrant transport vary seasonally?

       

      B. Mixed-layer (ML) depth or mixed-layer pumping

      Comments
      1. Turbulent mixing in the mixed layer keeps particles in suspension so fluxes cannot be accurately measured with traps until you are below that depth (Gardner and Richardson, 1992).
      2. Nocturnal increases in the mixed-layer depth can quickly move particles downward where they are isolated and allowed to settle in non-turbulent flow when the mixed layer thins during the day. (Woods and Onken, 1982; Gardner et al., 1995). Conversely, nutrients, pCO2 or any component whose concentration increases with depth will be mixed upward.

      Magnitude of problem

      • Shouldn't be an issue as long as you deploy traps below the ML.

      Effect on BATS/HOT carbon imbalance/balance

      • Trap fluxes in BATS were examined only when ML depth was < trap depth, so there should be no direct influence.
      • At HOT the trap depths were always greater than the mixed layer depths.
      • Total carbon increases with depth, so ML pumping would increase carbon in the surface, not decrease it.

      Recommendation

      • Traps should be deployed below the maximum mixed-layer depth during the time of deployment.
      • The dynamics of 234Th and exchange out of the mixed layer needs to be examined in the context of ML pumping and the types of particles to which 234Th is attached.

       

    C. Spatial inhomogeneity

    Comments

      1. What is the spatial inhomogeneity of vertical fluxes? i.e. patchiness
        1. Just put out a lot of trap arrays and test for homogeneity! (D. Lal)
        2. This would require extensive planning about where, when, the number of arrays required and space and time scales (Dave Karl).
        3. While this would provide information about spatial homogeneity, it does not answer the question of accuracy.
        4. Time-series traps could average out some of the local inhomogeneity if the time scales were chosen appropriately (Cindy Lee).
      2. As discussed below in independent measures of vertical flux, mass balances based on oxygen production match the trap carbon fluxes at the HOT site, but not at the BATS site. Perhaps this results from differences in the degree of spatial inhomogeneity of the two sites.
      3. In Michaels et al. (1994) they analyzed the effect of stochastic events on the carbon imbalance problem. Because of the large number of measurements, it is statistically very unlikely that rare, missed events could explain the imbalance at BATS. With the frequency of sampling, there is only a small number of events that could have occurred and not been seen in the BATS sampling. However, these events would have to account for the entire discrepancy and would require an unreasonably large flux (more than the total POC every day) to make up for the imbalance. Since the HOT sampling is of similar frequency, the same conclusion can be drawn at that site. The traps at the time-series stations are not grossly inaccurate because they miss rare events.
      4. On the smaller scale of individual experiments or deployments, we do not have enough data to see if a local event causes a discrepancy between a trap measurement and a field experiment.
      5. For a regional study, we may not have enough measurements to determine if the error lies in the traps or is a result of large-scale advection. The advective field is not known. Gradients in 234Th and carbon are small. Do we need to know the 3-D flow field (Michaels)? Are three-D experiments necessary for every place traps are deployed? The question of the role of advection is much larger than the trap accuracy issue and hits to the heart of the interpretation of all JGOFS data (Michaels).
      6. Production variability - over what time scale?
        • Traps from short-term deployments at multiple depths each reflect the surface production history for different time periods (Ian Walsh).
        • Traps at multiple depths may also reflect the flux from different source regions (Siegel et al., 1990).

    Magnitude of problem

      • unknown

    Effect on BATS/HOT carbon imbalance/balance

      • Missing rare events likely cannot explain the annual patterns. Advection may play an important role at each station.

    Recommendation

      • Consider deploying multiple trap arrays to test for homogeneity.
      • Do control-volume experiments in JGOFS.

     

    III. Independent Measures of Vertical Flux

    General Comments

      1. What is the accuracy of the trap measurements? Invoking closure of mass balance can't be just a convenience when testing for trap accuracy (Tony Michaels). Conservation of mass must be maintained. Closure of mass balances provides validity for other types of calibrations, but the time scales of each measurement must be comparable. Closure should be attempted whenever possible and when observations violate this condition, it is a powerful constraint on the interpretation of data.
      2. When we see disagreements between traps and other measurements, other processes are often invoked as an explanation, but if agreement is seen then the same questions aren't raised. (Ken Buesseler) This is true of any endeavor in science. We always seek closure of balance and when it is achieved, we move on.

     

    A. Thorium modeled and measured fluxes

    Methodology
    (Moore et al., 1981; Coale and Bruland, 1985, 1987; Eppley and Peterson, 1979; Buesseler, 1991; Buesseler et al., 1994).

      • Measure 234Th deficiency (relative to 238U) in surface waters.
      • From the 234Th deficiency (half life is 24 days), one can predict the 234Th flux down to the depth of disequilibrium. Given this and a 234Th measurement in a trap, one can learn independently if the trap is collecting 234Th-bearing particles in a predictable (i.e. accurate) fashion (Buesseler et al., 1994).

    Comments

      1. The 234Th deficit is confined to the upper 70-200 m in the open ocean, so the method is restricted to predicting the flux to the depth of the disequilibrium. Other studies show that the flux of material decreases rapidly with depth below the upper 50-150 m (VERTEX) or that at least the material one would expect to make up the vertical flux decreases rapidly with depth (Bishop et al., 1980).
      2. If the trap is not collecting particles bearing 234Th in a predictable way, then there is no confidence that they are collecting other particle types in an accurate fashion (Ken Buesseler). However, surface area per unit mass is greatest on the smallest particles, and the small particles are not dominant in the vertical flux of mass.
      3. There is a 234Th deficit down to 1000 m off Cape Hatteras due to scavenging along the continental margin (Peter Santschi) so this method won't work in those regions.
      4. In a dynamic environment with strong currents, trap flux estimates are very likely to be biased. As part of the French KERFIX program in the Southern Ocean, we collected trap data 60 NM SW of Kerguelen Island. Despite this distance from the island, tidal currents are important not only in upper waters but also at 1000 m depth. Under these conditions, estimates of the surface flux from deeper traps was not possible, nor was it possible to estimate flux based on natural radionuclides measurements. At present, we conclude that the site is not an appropriate one for estimates of the flux from surface waters. (Juan Carlos Miquel)
      5. Murray et al. (submitted, DSR) measured the flux of organic carbon from the central equatorial Pacific during EqPac using floating traps at multiple depths and the 234Th approach described above. In their calculations of 234Th flux they included terms to account for upwelling and meridional advection away from the equator. Zonal gradients in 234Th were found to be small (Buesseler et al., 1995) and were neglected. Comparison of the model 234Th fluxes with the corrected 234Th fluxes shows that the model fluxes shallower than 150 m were much less than the trap fluxes. The fluxes at 150 and 200 m agreed to within a factor of +/- 2 from 12°N to 12° S. Murray et al. (submitted, DSR) argue that all drifting sediment trap studies should be conducted as a function of depth and include 234Th analyses. This conflicts with the earlier recommendation that traps be deployed at a single depth and would thus require a large number of arrays deployed simultaneously (Michaels).
      6. Swimmers have low 234Th (Jim Murray; Buesseler et al., 1994).
      7. Error bars on data need to be included (Tony Michaels).
        • How large are the Th error bars? (Gardner).
        • See Buesseler et al., 1994
      8. What are the appropriate time scales (in terms of minimum and maximum fractions or multiples of half-lives) on which to make Th or other radionuclide measurements?

        If you have only a single radionuclide profile, you need to integrate particle fluxes over a comparable time scale appropriate to the tracer activity. If 234Th activities are decreasing with time (i.e. particle fluxes are increasing) a single 234Th profile would actually underestimate the net particle removal at the time you took your sample. If you take a non-steady state approach (Buesseler et al., 1994) and you measure 234Th in a time-series (or better yet, 4D) manner, then you can predict the 234Th export flux within the measurement period, (i.e. within a 2-5 day period is OK, as long as you have data to examine if the 234Th activity is varying with time during this same period). (Ken Buesseler)

    Magnitude of problem

      • Buesseler (1990) showed that fluxes predicted based on 234Th-deficiency and trap-measured 234Th fluxes often disagree by more than a factor of three, with some experiments showing overtrapping and others showing undertrapping by that amount.
      • Time-series measurements of 234Th deficits were made at BATS during trap deployments (Michaels et al., 1994) and can account for much of the carbon imbalance based on the 234Th calibration of the traps and the measured range of C/Th ratios. Over the course of two years of simultaneous 234Th profiles and 234Th collections in traps, both undertrapping and overtrapping have been observed at the BATS site.
      • A previous study of this type at BATS showed that traps on a single cruise tended to overcollect 234Th (Buesseler et al., 1994). Their explanation of the variable over/undertrapping was that their traps overcollected during times of low flux and undercollected during times of high flux, thus dampening out the seasonal signal of the total flux cycle.

    Effect on BATS/HOT carbon imbalance/balance

      • The difference between the predicted and measured 234Th flux is of the same order as the carbon imbalance at BATS (Michaels et al., 1994)
      • In order for advection to account for this 234Th difference, the magnitude of advection would have to be large based on known gradients of [Th], and there would have to be a seasonal trend to account for the flux imbalance. Why should there be a seasonal trend in [Th] in the western North Atlantic?
      • If the 234Th deficiency is caused by scavenging and if scavenging is related to the production and flux of particles, how could there not be a seasonal trend in [Th]? (Vernon Asper)
      • 234Th measurements have not been made around HOT because the scavenging of 234Th is generally low in oligotrophic regions, making the Th disequilibrium very small. When scavenging is low this means that the errors can be large for this calculation (Ken Bruland).
      • There are times when there is no 234Th deficiency at BATS (Buesseler et al., 1994)

    Recommendation

      • See recommendations in section III. B below.

     

    B. 234Th-derived estimates of particulate organic C flux

    Methodology

      • One can calculate a carbon flux by multiplying the 234Th flux by the C/234Th ratio in sinking (trap- or in-situ pump- collected) particles. This approach has its own uncertainties including the time variability and transport terms (Buesseler et al., 1992, 1994; Wei and Murray, 1992) and the C/234Th ratio in sinking particles (Michaels et al., 1994; Buesseler, 1995).
      • Compare the carbon flux measured in the trap with the carbon flux predicted based on the above calculation.

    Comments

      1. To estimate the accuracy of traps at collecting carbon by comparison with the thorium-derived calibration, however, requires the assumption that the 234Th is distributed similarly to organic C among the different types of sinking particles (based on composition and settling velocity distribution) that are responsible for the vertical flux of carbon in the ocean. 234Th adsorption is a function of surface area, and there is much greater surface area per unit mass for small particles that may not be sinking rapidly. How quickly do these small particles become incorporated into larger particles and sink? C/Th ratios vary by particle type, but by how much? The C/Th ratio varies between trap and large-volume filtration samples. The C/Th ratio varied as a function of time during the North Atlantic Bloom Experiment by a factor of 2 and could vary between seasons (Buesseler). Murray et al. (submitted to DSR) found a very different C/Th ratio during two different cruises across the equator. These changes must be adequately measured and incorporated into the models (Buesseler et al., 1995). 234Th on fragile, porous sinking aggregates may break up when they encounter a sediment trap and broken-up pieces may not be retained in the trap, or the aggregate may not be able to penetrate the large density brine used in some traps, thus decreasing the collection of both carbon and 234Th.
      2. Some have suggested that collections of profiles of thorium deficit coupled to pump data on the C/Th ratio of large particles could be used as an independent measure of carbon flux (Buesseler et al., 1992; 1995). The biggest concern is the assumption that the particles collected with in situ pumps are representative of the local sinking particle pool. If some class of exported particles exists that are not sampled (or are masked by "suspended" material on the filter), and it is a dominant fraction of the export flux, and it has a different POC/Th ratio, then this empirical approach fails. So far, the data look very encouraging (Ken Buesseler).
      3. Loss of POC due to vertical migrators would be accounted for in the upper ocean 234Th balance, but not in traps.
      4. Swimmers have low 234Th relative to POC, hence POC/Th ratios in traps may be elevated if swimmers are not adequately removed.
      5. The same concerns for 234Th alone (see section A above) apply here to POC flux, i.e. if the predicted 234Th flux is incorrect, then the POC fluxes would also be in error (assuming the C/Th ratios have been measured and modeled correctly).

    Effect on BATS/HOT carbon imbalance/balance

      • Using the predicted 234Th fluxes and measured POC/234Th on particles, the 234Th-derived POC flux would account for up to 80% of the apparent carbon imbalance at BATS (Michaels et al., 1994).

    Recommendation

      • 234Th is the best independent particle tracer we have. Any JGOFS trap study should have Th measurements made at the same time. (Jim Murray; Buesseler et al., 1994).
      • Further studies are needed to examine the range of C/Th ratios in sorted sediment trap and size-fractionated filtered particle samples to determine if the 234Th-derived trap calibration can be directly applied to POC.
      • Further study of the U-Th system and its transfer between dissolved, colloidal, particulate, and aggregate states is needed.
      • In order to validate the Th-deficiency method for estimates of carbon fluxes, design and conduct a trap/234Th calibration experiment in the highest productivity, lowest energy regime that is practical (i.e. low advection). Continental margins should also be avoided because of the potential of thorium scavenging in those regions (Wilf Gardner) This experiment must include explicit consideration of non-steady state and 3-D effects on the time-scale of the experiment (Tony Michaels and Ken Buesseler).

     

    C. Oxygen mass balance (Emerson)

    Comments

      1. Bermuda
        • Carbon flux calculated from O2 mass balance model: 3±1 mole C/m2/yr (Spitzer and Jenkins, 1989)
        • Carbon flux from floating traps: 0.8±0.2 mole C/m2/yr (Michaels et al., 1994)
      2. HOT station
        • Carbon flux calculated from O2 mass balance model: 1±0.5 mole C/m2/yr (Emerson et al., 1995)
        • Carbon flux from floating traps: 0.9±0.3 mole C/m2/yr (Karl et al., 1995, DSR, in press)
        • Carbon flux calculated from a mass balance of DIC carbon and 13C-DIC in a 1-D model: 0.9±0.5 mole C/m2/yr (Paul Quay, unpublished data).
        • About 25% of the carbon flux is carried as DOC (Emerson et al. 1995).

    Magnitude of error

      • Errors given above.
      • BATS - Factor of three difference between trap, 234Th and carbon budget methods. Budget includes all vertical processes, but not horizontal advection.
      • HOT - Three methods (trap, oxygen and carbon isotope budgets) agree within the accuracy of the data. They do not include a number of other processes which may export carbon (DOC, migrant fluxes). Does not include horizontal advection.
      • The methods show more agreement near Hawaii than Bermuda. However, the comparisons do not include all of the same processes at each of the two sites. Any changes in the BATS flux as a result of a correction for an inferred source of error tend to have the effect of creating an imbalance at HOT.
      • Some infer that the agreement at HOT means that traps are accurate at that site. Perhaps there are fewer environmental variables to contend with around Hawaii than Bermuda (e.g. fronts passing the area, winter overturn, mode-water formation, proximity to a major current like the Gulf Stream with its attendant rings, general advection), so there is a better chance of reaching closure for budgets of carbon, oxygen, and 234Th for calibration with trap fluxes. However, it is also possible that the agreement between one form of carbon flux (traps) and the overall 1-D organic carbon budget means that the trap is likely overcollecting since the current comparison leaves no room for other processes of transport and error like migrant fluxes, hydrodynamics and solubilization of carbon.

    Recommendations

      • Continue to make mass balances of this sort where possible whether or not other means of calibration are available.
      • If possible use multiple independent strategies for comparison with traps.

     

    D. Comparison with sediment accumulation rates

    Comments

      1. Some trap fluxes have matched well with:
        1. accumulation rate of underlying sediments based on radionuclide dating (Pennington, 1974 ; Soutar et al., 1977; Dymond et al., 1981; Gardner et al., 1985);
        2. accumulation of radionuclides (Moore et al., 1981: Anderson et al., 1983; Bacon et al., 1985; Biscaye et al., 1988; Biscaye and Anderson, 1994; Colley et al., 1995);
        3. accumulation above a known sediment horizon in lakes (Pennington, 1974);
        4. varves (Soutar et al., 1977; Brunskill, 1969 as discussed in Gardner, 1980a; Hay et al., 1990).
      2. It is very difficult to use accumulation rates as a calibration standard of the carbon flux for short-term near-surface traps in the open ocean because so much degradation occurs between the surface and seafloor. Even in shallow lakes the time scales can also be orders of magnitude different between trap deployments and accumulation rate measurements (decades for 210Pb and 100-1000 years for 14C).
      3. The flux of inert components such as Al could be used as a calibration standard with the assumption that trapping efficiency for POC matched aluminosilicates. One must always be aware of the possible "contamination" by lateral advection of material resuspended from boundaries - both in traps in in the sediments.
      4. The accumulation of short-lived radionuclides within the water column (e.g 234Th) can be measured on time scales close to the trap deployment time scales.

     

    E. Seasonality

    Comments

      1. Trap fluxes have been shown to have a seasonal cycle. (Deuser and Ross, 1980; Honjo, 1982; Deuser, 1986, 1987)
      2. A seasonal cycle can exist without knowing the absolute flux because the entire cycle or parts of the cycle could be biased high or low depending on the dominant particle type or sinking speed and hydrodynamic conditions.
      3. Buesseler et al. (1994) concluded that at the BATS site floating traps overcollected during periods of low productivity and undercollected during times of high productivity, thus smoothing out the seasonal cycles, but resulting in an annual average undercollection.

     

    F. Correlation with ocean color

      1. Mitchell et al. (in prep.) have compiled floating trap data from numerous projects (RACER, ProMARE, NABE, HOT, CABS, BATS, EqPac) and plotted the fluxes against an algorithm-derived parameter based on sea surface temperature and a blue to green water leaving radiance ratio. The 48 points have an r-squared fit of 0.71 on a log-log plot. This is comparable to the r-squared fit of 0.76 for the same number of points fitting the data of Chl+Phaeopigments versus the blue to green water leaving radiance ratio for the same sites.
      2. One must recognize that there could be a good correlation between flux and the algorithm and still the traps could all be too high or too low. There is also enough scatter that some of them could be high and some could be low by factors of 2-3 as seen in the data collated by Buesseler (1991). Care must be exercised in selecting data for this comparison (e.g. comparable by depth, etc.). Some of the scatter may result from differences like these.
      3. The important, encouraging point is that there appears to be a real correlation between ocean color and particle flux. Such a relationship may seem intuitively obvious to some, but others have questioned whether such a relationship could be demonstrated. It is crucial to cover the entire dynamic range of oceanic conditions to establish this relationship rather than examining only a small portion of the entire range, in which case the correlation might not be so obvious.

     

    IV. Summary of the magnitude of possible errors

    These tables are optimtized for use with Netscape 1.1 or later.
    Here is a text version of the tables.

    Trap Errors

    Traps are designed to collect only the settling particles. Sources of errors in trap measurements include:

      Error Source Error Magnitude
      Swimmers Up to a factor of 2 depending on techniques
      Solubilization of carbon A few percent per day
      Hydrodynamic effects that include:
      Trap geometry Up to several multiples of change
      Flow Zero to several multiples of change
      Wave induced trap motion Not quantified
      Tilt 25-100%
      Effects of brine in the trap 0-300% (60% is max seen in surface waters)

       

    System Errors

      Error Source Error Magnitude
      Vertical migration of zooplankton 8-70% of trap flux
      Vertical mixing of DOC, DIC, and POC 7-25% in two estimates
      Advective transport Undetermined
      Gas exchange of carbon dioxide with atmosphere 2% in one estimate

    Text Version of the Trap and System Error Tables

    Trap errors

    Traps are designed to collect only the settling particles. Sources of errors in trap measurements include

      • Swimmers
        • Up to a factor of 2 depending on techniques
      • Solubilization of carbon
        • A few percent per day
      • Hydrodynamic effects that include
        • Trap geometry
          • Up to several multiples of change
        • Flow
          • Zero to several multiples of change
        • Wave-induced trap motion
          • Not quantified
        • Tilt
          • 25-100%
        • Effects of brine in the trap.
          • 0-300% (60% is max seen in surface waters)

    System errors

      • Vertical migration of zooplankton
        • 8-70% of trap flux
      • Vertical mixing of DOC, DIC and POC
        • 7- 25% in two estimates
      • Advective transport
        • Undetermined
      • Gas exchange of CO2 with the atmosphere
        • 2% in one estimate

         

    V. Summary and Recommendations

    Sediment traps have opened up a new era of investigation of biogeochemical cycles in the ocean. They provided the first proof that seasonal and episodic variations in surface water productivity could result in variable fluxes at depth in the ocean, thus triggering many new questions to pursue. The collection of samples at various depths has allowed studies of the recycling of oceanic particles and helped to elucidate where many processes are occurring. In turn, this information is extremely important in making comparisons with the small residue of biogenic material which reaches the seafloor, and the even smaller residue that is preserved in the sediments. This is critical to accurately (albeit very imperfectly) interpret the paleo record from sediment cores. It is equally important for understanding correlations on short time scales between remotely sensed ocean color data from satellites and processes that lead to the export of carbon from surface waters.

    Traps have proven to be valuable tools. Particle cycling in the upper ocean is more complex that previously realized. The lack of mass balance in some studies based on trap fluxes is part of what has made us realize that complexity. That does not mean a priori that traps don't (or do) work. We need to make a more concerted effort to fully understand both traps and the particle dynamics in the environments in which they are used. Existing data clearly show there are trap designs and regimes in which the results from traps cannot be used in either quantitative (flux) or qualitative (composition) analyses. There are studies using floating traps where the carbon fluxes match well the macroscale carbon budgets derived from oxygen budgets determined by completely independent measurements (Emerson et al., 1995; Karl et al., 1995). Other studies show mismatches of as much as a factor of three (Michaels et al., 1994) Further studies are needed to understand why this discrepancy exists. There needs to be some standard measurements that can be made to verify or validate trap fluxes. One approach has been to calibrate specific trap designs under known conditions (e.g. velocity, Reynolds number, tilt) and then use those traps in the field and accept only those data that are collected within those acceptable physical parameters. A variation on this approach is to compare fluxes between a trap moving with the water and one that is not moving with the water to extend the limits of acceptable hydrodynamic conditions. Only a limited number of studies of this type have been made and the boundaries of regimes where trap fluxes match still-water fluxes must be better defined. The development of neutrally buoyant traps would improve on this method significantly. A second approach is to measure the loss of a radionuclide that is produced in the water as an in-situ calibration for trap fluxes. A third calibration scheme is to develop carbon budgets independent of trap fluxes as a standard for trap carbon fluxes.

    Ultimately, one of the prime JGOFS goals is to develop budgets of carbon, carbonate, silica etc., not to determine a radionuclide flux or trap accuracy. Each system may have a different portion of material transported by vertical settling versus migrant transport, DOM or other mechanisms (advection). From that viewpoint it is a higher priority to develop a system that can be used to predict the removal of carbon (by all pathways combined) from the surface water than determine the absolute accuracy of traps. If the 234Th method could be shown unequivocally to accurately predict the flux of carbon out of surface waters under all conditions, that is a very important advancement for JGOFS, because traps probably don't collect the material carried by vertical migrators and they certainly can't collect DOC or colloidal material. At the same time, trap samples afford the opportunity to examine the composition of settling material if they are functioning as we would like them to. That is what makes it worthwhile to calibrate traps.

    It is important to remember that comparisons between the trap fluxes and 234Th fluxes apply only to the depth over which particle scavenging creates a 234Th depletion (the upper 70-200 m or so in the open ocean). Carbon remineralization is rapid below the euphotic zone, so carbon fluxes decrease rapidly. Still, a proven calibration scheme for traps in this depth range is important for comparisons with short-term processes in the euphotic zone and with satellite data. With regard to long-term sequestration of carbon, fluxes to the deep ocean and to the seafloor are far more significant than carbon fluxes out of the upper 200 m. Agreement between longer-lived radionuclides and trap fluxes at greater depths has been much better, but that is beyond the scope of this report.

    Why is there an apparent agreement between traps and independent measurements at HOT and an apparent 3-fold disagreement at BATS? One valid suggestion is to conduct a 4D-scale carbon and 234Th calibration to verify the carbon and 234Th budget. If this is done at BATS, is this a calibration of the trap methodology or a calibration of the carbon dynamics around the around BATS region? 234Th measurements have not been made around HOT because the scavenging of 234Th is so low in oligotrophic regions that the errors are large for this calculation. Buesseler generally finds enough scavenging at BATS to make these measurements, though there are times when there is no disequilibrium, which means the predicted flux is zero.

    What, then, is the calibration standard? Is it necessary to measure 234Th depletion every time a trap measurements is made in the upper 200 m to determine if it is accurate? If we conduct a 4D-scale trap/234Th experiment at a simple site, does that guarantee that similar trap measurements conducted elsewhere can use the same correction factor or scheme? Or do we take the hydrodynamics viewpoint and determine the conditions under which traps match the flux determined by an independent means such as 234Th depletion or a carbon balance and then say that traps can be used under those conditions and if the conditions are not met, the data must be discarded?

    We can make recommendations for JGOFS, but what is the penalty for non-compliance? What about all the trap measurements made outside of JGOFS? They probably constitute the majority of trap measurements both now and in the future. What advice do we give them? Most people who use traps don't have the resources to measure either 234Th or currents.

    Before NASA sends an instrument into space (except for the Hubble space telescope) it is tested for responses in all conditions it might experience. Many calibration and comparison experiments have been made with sediment traps in both the laboratory and field, but few calibration measurements have been made in the upper 200 m of the water column. The only calibration technique that has been suggested for calibrating traps on short time scales in this region is 234Th. If a 4D 234Th calibration of traps is done to answer the question of trap efficiency, there should be an oversight committee to ensure that all parameters are sufficiently characterized and measured. Such a study would also have to include measurements of vertical migrant and DOM transport.

    Despite all the concern about fluxes measured with floating sediment traps, Mitchell et al. have shown a correlation between ocean color and particle flux when measured over the global range of ocean color. While there is significant scatter in the data and the precision is not known, it provides encouragement that it is possible to develop even better algorithms to make true Global calculations for the Joint Global Ocean Flux Study using floating sediment traps.

    VI. Major Recommendations

      1. Design and conduct a trap/234Th calibration experiment in the highest productivity, lowest energy regime that is practical. There should be community input to the design of such an experiment even if only one or more groups conduct the work. The C/Th ratio and Th cycling between sinking and non-sinking particle pools is one of the crucial points of such an experiment.
      2. Pending the outcome of the above experiment, measure the 234Th deficiency during JGOFS trap studies.
      3. Develop and test neutrally buoyant traps.
      4. Minimize the flow past traps.
      5. Measure flux at a single depth per array. Measurements at multiple depths are always desirable, but we must consider the importance of one measurement in which we have confidence versus several numbers that might be compromised because of velocity effects.
      6. Decouple the trap from surface wave motion.
      7. Measure the velocity past traps.
      8. Deploy traps with 5 psu excess brine only in the bottom of the trap. Make necessary tests to convert the BATS/HOT trap protocols to brine only in the bottom.
      9. Carefully remove swimmers from samples.
      10. Put the JGOFS protocols on the Web so they can be accessed easily.
      11. Fully report methods and errors in all publications.
      12. Make more experiments of the type of Baker et al. (1988).

     

    References

    1. Alldredge, A. L. and C. Gotschalk, 1988. In situ settling behavior of marine snow. Limnol. Oceanogr. 33: 339-351.
    2. Anderson, R. F., M.P. Bacon, P.G. Brewer, 1983. Removal of Th-230 and Pa-231 from the open ocean. Earth Planet. Sci. Lett.. 62: 7-23.
    3. Angel, M. V., 1989. Does mesopelagic biology affect the vertical flux? In: W. H. Berger, V. S. Smetacek, G. Wefer, eds. Productivity of the Ocean: Present and Past. John Wiley & Sons, 155-173.
    4. Bacon, M. P., C.A. Huh, A.P. Fleer, and W.G. Deuser, 1985. Seasonality in the flux of natural radionuclides and Plutonium in the deep Sargasso Sea. Deep-Sea Research. 32: 273-286.
    5. Baker, E. T., H. B. Milburn, and D. A. Tennant, 1988. Field assessment of sediment trap efficiency under varying flow conditions. J. Mar. Res. 46: 573-592.
    6. Biscaye, P. E. and R. F. Anderson, 1994. Fluxes of particulate matter and on the slope of the southern Middle Atlantic Bight: SEEP-II. Deep-Sea Res. II, 41: 459-509.
    7. Biscaye, P. E., R. F. Anderson, and B. L. Deck, 1988. Fluxes of particles and constituents to the eastern United States continental slope and rise: SEEP-I. Cont. Shelf Res. 8: 855-904.
    8. Bishop, J. K. B., R. W. Collier, D. R. Ketten and J. M. Edmond, 1980. The chemistry, biology, and vertical flux of particulate matter from the upper 1500 m of the Panama Basin in the Equatorial Pacific Ocean. Deep-Sea. Res. 27A: 615-640.
    9. Blomqvist, S. and C. Kofoed, 1981. Sediment trapping - a subaquatic in situ experiment. Limnology and Oceanography. 26: 585-590.
    10. v. Bodungen, B. M. Wunsch, H. FŸrderer, 1991. Sampling and analysis of suspended and sinking particles in the northern North Atlantic. Geophys. Monogr. 63: 47-56.
    11. Brunskill, G. J., 1969. Fayetteville Green Lake, New York, III. Precipitation and sedimentation of calcite in a meromictic lake with laminated sediments. Limnol. Oceanogr.. 14: 830-847.
    12. Buesseler, K. O., 1991. Do upper-ocean sediment traps provide an accurate record of particle flux? Nature. 353: 420-423.
    13. Buesseler, K.O., J. K. Cochran, M. P. Bacon and H. D. Livingston, 1992. Carbon and nitrogen export during the JGOFS North Atlantic Bloom Experiment estimated from 234Th:238U disequilibria. Deep-Sea Res. I. 39: 1115-1137
    14. Buesseler, K. O., J.A. Andrews, M. C. Hartman, R. Belastock, and F. Chai, 1995. Regional estimated of the export flux of particulate organic carbon derived from thorium-234 during the JGOFS EqPac program. Deep-Sea Res., 42:777-804.
    15. Buesseler, K. O., A. F. Michaels, D. A. Siegel, and A. H. Knap, 1994. A three-dimensional time-dependent approach to calibrating sediment trap fluxes. Glob. Biogeochem. Cycles 8: 179-193.
    16. Butman, C. A., 1986. Sediment trap biases in turbulent flows: results from a laboratory flume study. J. Mar. Res. 44: 645-693.
    17. Butman, C. A., W. D. Grant, and K. D. Stolzenbach, 1986 a. Predictions of sediment trap biases in turbulent flows: A theoretical analysis based on observations from the literature. Journal of Marine Research 44: 601-644.
    18. Coale, K. H.,1990. Labyrinth of doom: a device to minimize the "swimmer" component in sediment trap collections. Limnol. Oceanogr. 35: 1376-1380.
    19. Coale, K. H. and K. W. Bruland, 1985. 234Th:238U disequilbria within the California Current. Limnol. Oceanogr. 30: 22-33.
    20. Coale, K. H. and K. W. Bruland, 1987. Oceanic stratified euphotic zone as elucidated by 234Th:238U disequilbria. Limnol. Oceanogr. 32: 189-200.
    21. Colley, S., J. Thomson and P.P. Newton, 1995. Detailed 230Th, 232Th and 210Pb fluxes recorded by the 1989/90 BOFS sediment trap time-series at 48¡N, 20¡W, Deep-Sea Res. 42: 833-848.
    22. Dam, H.G., M.R. Roman and M.J. Youngbluth, 1995. Downward export of respiratory carbon and dissolved inorganic nitrogen by diel-migrant mesozooplankton at the JGOFS Bermuda time-series station. Deep-Sea Res. 42:1187-1197.
    23. Deuser, W. G., 1987. Seasonal variations in isotopic composition and deep-water fluxes of the tests of perennially abundant Planktonic Forminifera of the Sargasso sea: results from sediment-trap collections and their Paleoceanographic significance. J. Foram. Res.. 17: 14-27.
    24. Deuser, W. G. and E.H. Ross, 1980. Seasonal change in the flux of organic carbon to the deep Sargasso Sea. Nature. 283: 364-365.
    25. Diercks, A. and V. Asper, 1994. Neutrally buoyant sediment traps: The first designs. EOS, Transactions Amer. Geophys. Union 75:22
    26. DOE, 1994. Handbook of methods for the analysis of the various parameters of the carbon dioxide system in sea water; version 2. Dickson, A. G.; C. Goyet, Editors. ORNL/CIDIAC-74.
    27. Dymond, J., K. Fischer, M. Clauson, R. Cobler, W. Gardner, M.J. Richardson, W. Berger, A Soutar, and R Dunbar, 1981. A sediment trap intercomparison study in the Santa Barbara Basin. Earth and Planetary Science Letters. 53: 409-418.
    28. Emerson, S., P.D. Quay, C. Stump, D. Wilbur, and R.Schudlich, 1995. Chemical tracers of productivity and respiration in the subtropical Pacific Ocean. J. Geophys. Res.. 100: 15,873-15887.
    29. Eppley, R. W. and B.J. Peterson, 1979. Particulate organic matter flux and planktonic organic matter in the surfae layer of the ocean. Deep-Sea Res., 30 (A): 311-323.
    30. Gardner, W. D., 1980a. Field calibration of sediment traps. Journal of Marine Research 38: 41-52.
    31. Gardner, W.D., 1980b. Sediment trap dynamics and calibration: a laboratory evaluation. Journal of Marine Research 38: 17-39.
    32. Gardner, W. D., 1985. The effect of tilt on sediment trap efficiency. Deep-Sea Research 32: 349-361.
    33. Gardner, W. D., S. P. Chung, M. J. Richardson, and I. D. Walsh, 1995, The oceanic mixed-layer pump. Deep-Sea Res. II 42: 757-775.
    34. Gardner, W. D., P.E. Biscaye and M.J. Richardson, Sediment Traps in the Vema Channel: Collectors of vertical or horizontal particulate flux? Deep Sea Research (submitted)
    35. Gardner, W. D., K. R. Hinga, and J. Marra, 1983. Observations on the degradation of biogenic material in the deep ocean with implications on the accuracy of sediment trap fluxes. J. Mar. Res. 41: 195-214.
    36. Gardner, W. D. and M.J. Richardson, 1992. Particle export and resuspension fluxes in the western North Atlantic. In: G.T. Rowe and V. Pariente, (eds.). Deep-Sea Food Chains and the Global Carbon Cycle. Netherlands: Kluwer Academic Publishers pp. 339-364.
    37. Gardner, W. D., J. B. Southard, and C. D. Hollister, 1985. Sedimentation and resuspension in the western North Atlantic. Mar. Geol. 65: 199-242.
    38. Gardner, W.D. and Y. Zhang, 1996. The effect of brine on the collection efficiency of cylindrical particle traps, Deep Sea Research (in press)
    39. Gust, G., W. Bowles, S. Giordano, and M. Huettel. Particle accumulation processes in upright, flow-exposed cylindrical sediment traps and proposed link to in-situ fluxes. Aquat. Sci.. (in press).
    40. Gust, G., R.H. Byrne, R.E. Bernstein, P.R. Betzer, and W. Bowles, 1992. Particle fluxes and moving fluids: experience from synchronous trap collections in the Sargasso Sea. Deep-Sea Res. 39: 1071-1083.
    41. Gust, G., A. F. Michaels, R. Johnson, W. G. Deuser, and W. Bowles, 1994. Mooring line motions and sediment trap hydromechanics: in situ intercomparison of three common deployment designs. Deep-Sea Res. 41: 831-857.
    42. Harbison, G. R. and R. W. Gilmer, 1986. Effects of animal behavior on sediment trap collections: implications for the calculation of aragonite fluxes. Deep-Sea. Res. 33: 1017-1024.
    43. Hansel, D. A. and J. A. Newton, 1994. Design and evaluation of a "swimmer"-segretating particle interceptor trap. Limnol. Oceanogr. 39: 1487-1495.
    44. Hargrave, B. T. and N. M. Burns, 1979. Assessment of sediment trap collection efficiency. Limnol. Oceanogr. 24: 1124-1136.
    45. Hawley, N., 1988. Flow in Cylindrical sediment traps. Journal of Great Lakes Research 14: 76-88.
    46. Hay, B. J., S. Honjo, S. Kempe, V. A. Ittekkot, E. T. Degens, T. Konuk, and E. Izdar, 1990. Interannual variability in particle flux in the southwestern Black Sea. Deep-Sea Res. 37: 911-928.
    47. Honjo, S., 1982 Seasonality and interaction of biogenic and lithogenic particulate flux at the Panama Basin. Science. 218: 883-884.
    48. Honjo, S., D. W. Spencer, and W. D. Gardner, 1992. A sediment trap intercomparison experiment in the Panama Basin, 1979. Deep-Sea Res. 39: 333-358. IOC Manual and Guides No. 29, 1994. Protocols for the Joint Global Ocean Flux Study (JGOFS) core measurements. UNESCO Scientific Committee on Oceanic Research.
    49. Iturriaga, R., 1979. Bacterial activity related to sedimenting particulate matter. Mar. Biol. 55: 157-169.
    50. Jannasch, H. W., O. C. Zafiriou, and J. W. Farrington, 1980. A sequencing sediment trap for time-series studies of fragile particles. Limnol. Oceanogr. 25: 939-943.
    51. Karl, D. M. and G. A. Knauer, 1989. Swimmers: a recapitulation of the problem and a potential solution . Oceanography. 2: 32-35.
    52. Karl, D. M. and B.D. Tilbrook, 1994. Production and transport of methane in oceanic particulate organic matter. Nature, .368: 732-734.
    53. Knauer, G. A., D. M. Karl, J.H. Martin, and C.N. Hunter, C. N., 1984. In situ effects of selected preservatives on total carbon, nitrogen and metals collected in sediment traps. Journal of Marine Research. 42: 445-462.
    54. Lee, C. and C. Cronin, 1982 The vertical flux of particulate organic nitrogen in the sea: Decomposition of amino acids in the Peru upwelling area and the equatorial Atlantic. Journal of Marine Research 40: 227-251.
    55. Lee, C., J. I. Hedges, S. G. Wakeham, and N. Zhu, 1992. Effectiveness of various treatments in retarding microbial activity in sediment trap material and their effects on the collection of swimmers. Limnol. Oceanogr..37: 117-130.
    56. Lee, C., J.A. McKenzie and M. Sturm, 1987. Carbon isotope fractionation and changes in the flux and composition of particulate matter resulting from biological activity during a sediment trap experiment in Lake Greifen, Switzerland. Limnol. Oceanogr. 32: 83-96.
    57. Lee, C., S. G. Wakeham, and J. I. Hedges, 1988. The measurement of oceanic particles flux - are 'swimmers' a problem? (Review and Comment). Oceanography. 1(2): 34-36.
    58. Longhurst, A. R. and W. G. Harrison, 1988. Vertical nitrogen flux from the oceanic photic zone by diel migrant zooplankton and nekton. Deep-Sea Res. 35: 881-889.
    59. Lorenzen, C. J., F.R. Shuman, and J.T. Bennett, 1981. In situ calibration of a sediment trap. Limnology and Oceanography. 26: 580-585.
    60. Macintyre S., Alldredge AL, Gotschalk CC. 1995. Accumulation of arine snow at density discontinuites in the water column. Limnology & Oceanography 40:449-468
    61. 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.
    62. Michaels, A. F., M.W. Silver, M.M. Gowing, and G.A. Knauer, 1990. Cryptic zooplankton "swimmers" in the upper ocean sediment traps. Deep-Sea Res. 37: 1285-1296.
    63. Miquel, J.C., S.W. Fowler and J. La Rosa, 1992. Vertical particulate carbon fluxes in the Ligurian Sea: a time-series study. Rapport et Prochs Verbaux Commission Internationale pour l'Exploration Scientifique de la Mer Miditerranie, 33: 78.
    64. Miquel, J.C., S.W. Fowler, J. La Rosa and P. Buat-Menard, 1994. Dynamics of the downward flux of particles and carbon in the open northwestern Mediterranean Sea. Deep Sea Research 41, 243-261.
    65. Miquel, J.C., S.W. Fowler, B. Mostajir and J. La Rosa, 1995. Long term study of particulate carbon flux in the open NW Mediterranean Sea. In Tsunogai S., K. Iseki, I. Koike et T. Oba (eds.), Global Fluxes of Carbon and its Related Substances in the Coastal Sea-Ocean-Atmosphere System (Proceedings of the 1994 Sapporo IGBP Symposium, 14-17 November 1994, Hokkaido University, Sapporo, Hokkaido, Japan), M&J International, Yokohama, 353-359.
    66. Moore, W. S., K.W. Bruland, J. and Michel, 1981. Fluxes of uranium and thorium series isotopes in the Santa Barbara Basin. Earth and Planetary Science Letters. 53: 391-399.
    67. Murray, J. W., J. N. Downs, S. Strom, C.-L. Wei, and H. W. Jannasch. 1989. Nutrient assimilation, export production and 234Th scavenging in the eastern equatorial Pacific. Deep-Sea Res. 36: 1471-1489.
    68. Pennington, W. Seston and sediment formation in five Lake District lakes. Jour. Ecol.. 1974; 62: 215-251
    69. Peterson, W. and H. G. Dam, 1990. The influence of copepod "swimmers" on pigment fluxes in brine-filled vs. ambient seawater-filled sediment traps. Limnol. Oceanogr. 35: 448-455.
    70. Peterson, M. L., P. J. Hernes, D. S. Thoreson, J. I. Hedges, C. Lee, and S. G. Wakeham. Field evaluation of a valved sediment trap. Limnol. Oceanogr.. 1993; 38: 1741-1761
    71. Siegel, D. A., T. C. Granata, A. F. Michaels, and T. D. Dickey, 1990. Mesoscale eddy diffusion, particle sinking, and the interpretation of sediment trap data. Jour. Geophys. Res. 95: 5305-5311.
    72. Silver, M. W., M.M. Gowing, D.C. Brownlee, and J.O. Corliss, 1984. Ciliated protozoa associated with oceanic sinking detritus. Nature, 309: 246-248.
    73. Soutar, A., S. A. Kling, P. A. Crill, E. Duffrin, and K. W. Bruland, 1977. Monitoring the marine environment through sedimentation. Nature, Lond. 266: 136-139.
    74. Spitzer, W. S. and W. J. Jenkins, 1989. Rates of vertical mixing, gas exchange and new production: estimates from seasonal gas cycles in the upper ocean near Bermuda. Jour. Mar. Res. 47: 169-196.
    75. U.S. GOFS Report No. 10., 1989. Sediment Trap Technology and Sampling. Available from U.S. JGOFS Planning Office, Woods Hole Oceanographic Institution, Woods Hole, MA, 94 pp.
    76. Walsh, I., K. Fischer, D. Murray, and J. Dymond, 1988. Evidence for resuspension of rebound particles from near-bottom sediment traps. Deep-Sea. Res. 35: 59-70.
    77. Wei, C. -L. and J. W. Murray, 1992. Temporal variations of 234Th activity in the water column of Dabob Bay: particle scavenging. Limnol. Oceanogr. 37(2): 296-314.

Attendees at the Villefranche Trap Meeting

(Meeting was open to all who attended the JOGFS Symposium)

    • Nicholas Bates (nick@sargasso.bbsr.edu)
    • U. Bathmann (ubathmann@awi-bremerhaven.de)
    • Ken Buesseler (kbuesseler@whoi.edu)
    • Craig Carlson (ccarlson@bbsr.edu)
    • Fei Chai (fchai@athena.umeoce.maine.edu)
    • Andrew Dickson (adickson@ucsd.edu)
    • Steve Emerson (emerson@u.washington.edu)
    • Wilford Gardner (wgardner@ocean.tamu.edu)
    • Julie Hall (hall@eco.cri.n2)
    • Nobuhiko Handa (h4496a@nucc.cc.nagoya-u.ac.jp)
    • Roger Hanson (rbhanson@nsf.gov)
    • Dennis A. Hansell (dennis@bbsr.edu)
    • Dale A. Kiefer (dale.kiefer@fao.org)
    • Tony Knap (knap@bbsr.edu)
    • Wolfgang Koeve (wkoeve@ifm.uni-kiel.d400.de)
    • D. Lal (dlal@ucsd.edu)
    • Richard Lampitt (rsl@nwo.ua.ac.uk)
    • K. K. Liu (kkliu@ccms.ntu.edu.tw)
    • Jim McCarthy (james_j_mccarthy@harvard.edu)
    • Nick McCave (mccave@esc.cam.ac.uk)
    • Dennis McGillicuddy (mcgillic@epl.whoi.edu)
    • Tony Michaels (tony@bbsr.edu)
    • J. Carlos Miquel (miquel@unice.fr)
    • Jim Murray (jmurray@u.washington.edu)
    • Wajih Nagvi (naqvi@bcgoa.ernet.in)
    • Susanne Neuer (susanne@zfn.uni-bremen.de)
    • John Parslow (parslow@ml.csiro.au)
    • Don Rice (drice@nsf.gov)
    • Javier Ruiz (javier.ruiz@uca.es)
    • T Saino (i45518a@nucc.cc.nagoya-u.ac.jp)
    • Jan Scholten (js@gpi.uni-keil.de)
    • Ian Walsh (walsh@ocean.tamu.edu) Rapporteur

Attendees without e-mail addresses (or incorrect addresses):

    • Detlef Schulz-Bull IfM Kiel, Germany
    • Ning Xiuren, 2nd Institute of Oceanography, SOA, 310012 Hangzhou, China
    • Jan Duinker jduinker@ifm.uni-kiel.d400.de
    • Ulrich Lundgreen IfM Kiel, Germany
    • Andreas Irmisch PTBEO Meeresforschung FAX +4938151509

Invited participants who were not able to attend:

    • Vernon Asper (vasper@whale.st.usm.edu)
    • Mike Bacon (mbacon@whoi.edu)
    • Bodo Bodungen (bodungen@bio.io-warnemuende.d400.de)
    • Steve Calvert (calvert@unixg.ubc.ca)
    • Serge Heussner(heussner@lisa.univ-perp.fr)
    • Susumu Honjo (shonjo@whoi.edu)
    • Cindy Lee (cindylee@ccmail.sunysb.edu>)
    • Phil Newton (100522.3717@compuserve.com)
    • Paul Wassman (paulw@ottar.uit.no)
    • Graham Shimmield (G.Shimmield@edinburgh.ac.uk)
    • Alexis Khripounoff (akripoun@ifremer.fr)

Contributors of material and comments after the first draft:

    • Tony Michaels
    • Ken Buesseler
    • Dave Karl
    • Wolfgang Koeve
    • Uli Lundgren
    • Cindy Lee
    • Vernon Asper
    • Jim Murray
    • Dennis Hansell
    • Greg Mitchell
    • Scott Nodder
    • Juan Carlos Miquel
    • Steve Emerson
    • Ken Bruland
    • Ken Coale