SYNOPSIS OF THE U.S. JGOFS SMP WORKSHOP ON MIDWATER PROCESSES Shining a Light on the Twilight Zone March 25 - 27, 2002 San Antonio, TX Adrian Burd, George Jackson, Richard Lampitt and Mick Follows

Midwater Workshop Home

Recent years have seen considerable progress in understanding the export of organic material from the surface ocean. However, the fate of material that sinks below the euphotic zone is not so well understood, and yet remains important for understanding how much surface production gets sequestered in the oceans. The main aim of this workshop was to assess our current understanding of processes affecting particle flux between depths of 100-1000 m where there is insufficient light for photosynthesis but enough to see (the “Twilight Zone”). Our hope was to be able to arrive at a consensus of our current state of knowledge, the uncertainties and future directions.

The meeting was loosely organized around four themes: particles, microbes, zooplankton and modeling. A general review was given for each theme followed by shorter, more specific talks and a general discussion. On the final day, four questions were asked of each theme: What are the biggest uncertainties? What is crucial? What would you like from other groups? What technologies would you like to see? The rationale for this was the view that understanding particle remineralization will require in interdisciplinary approach.

Traditionally, the change of particle flux with depth has been depicted using power-law relationships F(z)=az^-b, where the constants a and b are determined empirically. The most well known of these is that developed from the VERTEX data by Martin et al. (1987) who estimated a value of b=0.858. Such formulations have been used in global biogechemical models to predict the remineralization of particle flux within the twilight zone. However, it is difficult to incorporate spatial and temporal variability using such empirically-based descriptions. There is a growing need to understand the biological and physical factors affecting particle flux, with the hope of developing process-based models that can be used to predict spatial and temporal variations in particle remineralization.

The consensus view.

Fortunately, all participants at the meeting agreed that particle flux within the oceans does, for the most part, decrease with depth. Generally, only 10% of the material sinking below the mixed layer reaches a depth of 1000 m. Although there are large spatial and temporal variations in fluxes between 100 and 1000 m, there is remarkable uniformity in flux at 3000 m (Antia, Berelson) suggesting that the twilight zone communities can process large and varying amounts of material, providing something akin to a buffering capacity.

There was broad acceptance that although problems exist with sediment trap data, traps provide a good means of obtaining flux data. This is particularly the case when trap data are combined with other techniques such as the use of Thorium radioisotopes (Antia, Bacon). From such data the spatial distribution of annual particle flux is more similar to that of benthic fluxes than primary production. This is seen in assembled data sets (Jahnke), for example for the North Atlantic (Antia) and in the results of inverse models (Schlitzer).

A single relationship cannot accurately describe changes in particle flux with depth globally. Spatial and temporal variations in mesopelagic communities are in part responsible for changes in the depth of remineralization and changes in the size of the carbon sink in the oceans. Inverse modeling of global nutrient fields (Schlitzer) indicates considerable variation in these parameters, with a global average value of b=0.973 which is closer to the estimate of Suess (1980), but larger than that estimated by Martin et al. (1987). Interestingly, the inverse model predicts a value of b for the VERTEX region that is close to the Martin et al. value.

Particulate material is processed by both microbial and zooplankton communities (Steinberg, Carlson, Rivkin). Biomass and metabolic rates of both zooplankton and bacteria decrease with depth. Zooplankton in the twilight zone are known to consume particulate material (Steinberg, Wishner, Lampitt) and numbers of zooplankton per aggregate increase with aggregate size. Gut content analysis reveals the presence of olive green debris and diatoms in the guts of non-migrating zooplankton, indicating that these animals have been feeding on particulate material that originated in surface waters. However, zooplankton do not feed exclusively on particles, but can also act as carnivores.

Zooplankton that undertake diel vertical migrations tend to have gut residence times that are longer than weak or seasonal migrators. However, these are only sufficient to transport material 200-300 meters. Even so, this would potentially take that material out of the euphotic zone and prevent further processing of it in the upper water column (Steinberg). Vertically migrating zooplankton however, may provide a food source for deep living zooplankton.

Organisms appear not to be homogenously distributed in the twilight zone. Some zooplankton, for example, have vertical distributions that show concentrations coinciding with strong chemical gradients (such as the base of oxygen minimum zone Wishner). However, in the Arabian Sea, there appears to be no systematic horizontal spatial relationship between bacterial production, bacterial biomass and POC flux. Both long and short timescales were recognized as being important. Annual timescales are important for processes relevant to the global climate cycle and climate change. Shorter timescales are important for understanding how changes in biological communities affect the particle flux.

Vertical transport of organic material is not only in the particulate form, but also dissolved. Approximately 20% of the total export globally is in the form of DOC and "DOC export occurs when exportable DOC is present at the time of overturning" (Hansell). DOC export can also result from excretion by vertically migrating zooplankton (Steinberg).

The complexity of the midwater food web is a major issue. Zooplankton in the Twilight Zone are omnivores, feeding carnivorously and on detrital particles (Steinberg, Wishner). Zooplankton can alter their food preferences and life cycle according to conditions. For example, Lucicutia grandis in the Arabian Sea has a reproductive cycle that correlates with the timing of the monsoons and resulting increases in particle flux. This introduces significant complications in developing a food web model to predict changes in particle flux. Simple predator-prey models which include a particle feeder and a predator reveal that the dynamics of zooplankton in the twilight zone can have a dramatic effect on the particle flux reaching the deep ocean (Burd, Jackson). However, models incorporating particle coagulation, disaggregation, zooplankton feeding and bacterial degradation are consistent with observed changes in particle flux with depth (Stemmann).

Bacteria also affect particle flux. Bacteria attached to sinking particles aid in particle degradation by remineralizing organic material. But attached bacteria make up only about 10% of the bacterial biomass. Both bacterial biomass and metabolic rates decrease with depth in the oceans. In addition, bacterial community structure changes below the euphotic zone where archaea play a more important role (Carlson).

Uncertainties, Contradictions and Ignorance.

One of the most significant areas of uncertainty was the measurements of particle flux itself (everyone). Sediment traps have long been known to have problems, but to date they provide the instrument of choice for wide coverage in flux measurements. Problems relating to sediment traps include identification of the material (Antia) that is collected and this makes estimating what material is being processed very difficult. Optical measurements of particle concentration combined with satellite measurements of primary productivity allow measurements of particle flux at depth to be related to surface processes (Gardner). Many measurements in the twilight zone are defined operationally, which raises the question of what it is that these measurements are measuring (Wakeham)? A related problem is determining the accuracy of measurements in the midwater. For example, zooplankton that can swim into a trap pose a problem for flux measurements since they can consume the material already in the trap. This problem is usually dealt with by poisoning the trap, however, these poisons can leach dissolved material which may bias other measurements (Fisher).

Interactions between the various fractions (e.g., dissolved, particulate) remain uncertain (Antia, Jackson). Processes in the twilight zone may mask signals in the particle flux. For example, measurements using ²³⁰Th have indicated large (±50%) seasonal swings in particle flux which are too large to be accounted for by variations in surface production (Bacon). This indicates that significant processing and repackaging of particles can occur within the twilight zone. Whether these changes result from purely biological interactions (e.g., repackaging material into fecal pellets) or a combination of biological and physical (coagulation and particle break up) interactions is unclear.

Although both zooplankton and bacteria affect particle flux, the relative contribution of these groups of organisms to remineralization remains uncertain and contradictory. Part of this uncertainty arises from estimates being made in different parts of the water column. Estimates of the proportion of particle flux consumed or remineralized by zooplankton in the upper part of the twilight zone vary from 2% to 100%. Similar estimates for bacteria can exceed 100% (Jackson, Steinberg, Carlson). However, such conclusions often rely on estimates of bacterial production efficiencies which are not well known (Carlson, Rivkin) and can be lower than 20% for bacteria in the mesopelagic.

Our knowledge of processes in the twilight zone is hampered by our lack of knowledge as to who is there, where they are and what they are doing (Steinberg, Wishner). Satellites give us a great deal of spatial and temporal information about primary production, but similar coverage is not available for the twilight zone. So although we have lists of species, we have little information on their spatial distributions (both vertically and horizontally), patchiness and how these distributions change with time. Certain species are known to concentrate around chemical and physical gradients, e.g. Lucicutia grandis in the Arabian Sea and Pacific which may also have a reproductive cycle that is connected to seasonal changes in particle flux (Wishner).

Little is known about the structure, composition and role of the biological communities within the twilight zone. In many cases there is insufficient coverage of taxonomic or functional groups to enable food webs to be constructed (Ianson). One group that is particularly poorly represented is micro-zooplankton. This lack of detail presents a problem since without it we cannot know about the various trophic pathways that carbon takes once it is in the twilight zone (Steinberg, Wishner). One further complicating factor is that mid-water zooplankton can feed on both detritus as well as other zooplankton, thereby making it harder to determine the partitioning of carbon (Steinberg, Wishner).

Free-living bacteria dominate the twilight zone bacterial populations. However, the DOC consumed by them has to come from somewhere. So although attached bacteria comprise only a small fraction of the total, they may in fact be supporting a much larger fraction of the bacterial population. It is therefore important to know whether particles are being respired or solubilized. Regional differences in bacterial populations could also have a strong impact on changes in particle flux, as would any differences between the activities of archaea and bacteria.

The structure of surface communities play a significant role in determining particle flux in the twilight zone. For example, blooms of salps in surface waters can produce copious quantities of large, rapidly settling fecal pellets which can traverse the twilight zone relatively unscathed (Steinberg). Similarly, particles arising from diatoms can have different settling speeds and rates of remineralization than those arising from other phytoplankton (Berelson). This may be in part due to differences in composition (e.g., ballast, and hence particle density) or particle structure.

The role of DOC in the twilight zone remains unclear (Carlson, Hansell, Rivkin, Steinberg). Measurements show that mid-water DOC concentrations change throughout the ocean, generally decreasing from the North Atlantic, through the Indian and Pacific Oceans. The lability of this material is crucial to knowing if it will be taken up by bacteria. DOC can either be formed in situ (e.g., by solubilizing particles or through zooplankton excretion) or mixed into the twilight zone through deep convective mixing and the relative contributions of these mechanisms remains unclear.

Future Directions.

Understanding the processes in the twilight zone involves a problem of choosing the correct temporal and spatial scales. Large spatial scales yield insights into processes affecting global changes. However, these maybe related to the smaller scale of interactions between different groups of organisms (e.g., bacterial hotspots). Similarly, the longer timescales are important for carbon sequestration but these depend on the shorter timescales of animal interactions. This implies that an array of different techniques should be used.

Addressing the uncertainties and gaps in our knowledge will require combined modeling and field programs where primary production, particle flux and properties, zooplankton feeding and bacterial activity are all measured in concert (Steinberg). This will provide a view of the different roles that different communities play, in particular the relative roles of zooplankton and bacterial communities as well as POC and DOC, in particle remineralization. In addition, these studies should be made at different locations and at different times. Combining field programs with simultaneous modeling efforts will help synthesize data and improve our understanding of the role mid-water processes play in the global carbon cycle.

The development of inverse models has already been fruitful in providing regional predictions of the parameters a and b (Schlitzer). However, improved predictive power will come from the development of mechanistic models which can focus on the interactions between different material fractions. One possible direction is a coupling of forward and inverse models. A standardized framework for model comparison should be developed, and much can be learnt from existing SMP efforts along these lines. Consequently, there is a need for models that incorporate valid, tractable formulations for the biological and physical processes occurring in the mid-water.

There should be a effort made to improve our understanding of particle properties and how they affect particle flux and remineralization. Better understanding of the changes in particle composition and size distributions will help determine processes affecting particle settling speed and how it changes in the water column. In particular it will be important to understand how the average particle settling speed (which is what is measured) is related to the properties of the individual particles. At the present we cannot accurately relate particle properties to the flux or vice versa. Such an ability is important since we if we can deduce particle properties we can say something about particle flux. Similarly, we need to understand not only particle transformation within the twilight zone but also particle production within the twilight zone.

Spatial and temporal coverage of mid-water organisms and particle flux is currently patchy. However, there is a significant amount of information available in the Russian scientific literature which is now available (Banse). This may assist in filling some gaps in spatial and temporal distributions of organisms.

Mid-water biological communities respond to changes in external forcing. This can be seen in the life-cycles and distributions of zooplankton in the Arabian Sea for example (Wishner). We need a greater understanding of how zooplankton and bacteria respond to changes in particle flux (both composition and magnitude) and DOC export.

New technologies open up the promise of new understanding. In particular, the holocam (Lampitt) presents the possibility of studying three dimensional small scale spatial distributions of organisms and particles in situ. This will greatly increase our understanding of the interactions between particles and zooplankton, and also between zooplankton themselves. The large volume and spatial nature of the images captured by the class=SpellE>holocam should also provide reliable estimates of particle concentrations and particle size distributions. If combined with other optical techniques (Gardner) and traditional measurement it should make for a good comparison.

Recreating the history of a particle retrieved from a sediment trap is something of a forensic science (or art). There is a strong need for a suite of biomarkers that can be used to discover what processes have acted on particles as they have fallen through the twilight zone - in other words, who has done what to whom? Along a similar vein, there needs to be developed the capability of performing in situ physiological measurements on zooplankton (e.g., respiration, feeding responses etc.)

References

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

Suess, E. (1980). Particulate organic carbon flux in the oceans, surface productivity and oxygen utilization. Nature 288:260-263.

Some remarks

“I want one!” - referring to the holocam.

“Lability… is that a word?”

“Wow!!!” (response of a non-biologist to seeing neat movies of critters doing their stuff)

Participants