Primary production fueled by new nutrient inputs is balanced over large time and space scales by export of organic matter (both particulate and dissolved; see below) from the upper productive layer into the deep sea. The export flux has been studied principally by sediment traps moored in the deep ocean [ Honjo et al., 1992; Jasper and Deuser, 1993; Milliman, 1993], and also by increasingly numerous deployments of freely drifting traps in the upper 1000 m [ Silver and Gowing, 1991]. Time series observations show that variations in surface productivity are mirrored by fluxes into shallow drifting and deeper moored traps [ Altabet et al., 1991; Karl et al., 1991; Asper et al., 1992], suggesting that the traps provide an accurate reflection of the timing of the export process. Comparison of trap collections with camera profiles of large aggregates indicate that fluxes are proportional to large particle concentrations in the water column [ Walsh and Gardner, 1992].
The accuracy of sediment traps has been questioned, particularly in the upper ocean where biological activity and variability are greatest. At shallower depths in warmer water, bacterial activity must be controlled using poisons to prevent particle decomposition [ Lee et al., 1992; Hedges et al., 1993] but poisoning results in accumulation of actively swimming zooplankters (`swimmers') in the traps, leading to overestimates of fluxes. New trap designs appear to segregate the larger swimmers, but small swimmers are not efficiently removed from the passively sinking material [ Peterson, 1993; Hansell and Newton, 1994]. Even if the multitude of biological effects can be corrected or prevented, the behavior of traps in moving fluid still needs to be addressed. Traps appear to collect particles in proportion to the particle approach velocities [ Gust et al., 1992]. The bias will be largest in the upper ocean where velocity shear is greatest.
[ Buesseler et al., 1992a] appears to be a powerful tracer of
particle flux in the upper ocean. Comparison of observed 
fluxes
into shallow sediment traps with modeled 
removal rates on sinking
particles suggests that shallow drifting traps may undercollect or
overcollect the particulate flux by factors of 3-10 [ Buesseler, 1991].
-based estimates of particulate carbon export from 
profiles in JGOFS NABE ranged from 5-42% of the primary production
[ Buesseler et al., 1 992b], agreeing with net observed changes in
particulate carbon in the upper water column [ Bender et al., 1992], but
were up to 3 times greater than estimates derived from carbon flux into
sediment traps [ Martin et al., 1993]. The 
and 
Th
observations in NABE support the idea that microbial activity is responsible
for particle aggregation/disaggregation [ Cochran et al., 1993].
Transparent exopolymer particles (TEP) of O(100 mm) derived from the extracellular products of bacteria and diatoms harbor bacteria and might constitute sites for particle aggregation and enhanced microbial hydrolytic enzyme activity [ Smith et al., 1992; Alldredge et al., 1993]. In the subarctic north Pacific (SUPER Program) bacterial production averaged 13% of primary production in the euphotic zone, and bracketed estimates of the particulate organic carbon (POC) flux in the mesopelagic region (100-1000 m depths; [ Simon et al., 1992], suggesting that decomposition of the POC flux supported bacterial metabolism beneath the surface layer. In the Arabian Sea however, even conservative estimates of mesopelagic bacterial production (100--1000 m) were in excess of the POC fluxes through the water column, indicating the possible need for an addition source of carbon for the bacteria [ Ducklow, 1993]. Virus activity might play a role in regulating bacterial activity in sinking particles by attacking the particle-associated bacteria and phytoplankton in the vertical flux [ Proctor and Fuhrman, 1991].