appearing (e.g. Hernes 2001). Finally, for deep samples (>2 km), the
swimmers are fewer and assumed to be essentially eliminated by coarse (1
mm) screening of samples before analysis; marine snow aggregates
primarily break up and pass the screen (e.g. Honjo et al. 2008).
2 Traps are hydrodynamic barriers that force flow around them into eddies
over the top of a trap. This has the effect of a snow fence on the prairie –
greatly enhanced deposition compared to the actual flux in the region. This
has been modeled in flow flumes and on computers. Two-fold differences
have been found between traps hung at the same depth in pairs, presumably
those leading and trailing in the flow. The quantitative importance of snow-
fence effects remains uncertain but large. At considerable expense in
complexity, mounting traps on floats neutrally buoyant at a depth of interest
may or may not solve this problem (Stanley et al. 2004).
3 In long deployments (and many of the data are from long deployments)
rotting and dissolution of organic matter sedimented into the traps can be
extensive. This causes enthusiasm for solutions of azide (until ∼1985),
mercuric chloride or formaldehyde in the trap bottoms, usually held down
by mixing with dense brine. Unfortunately, dense brine causes osmotic
rupture of trapped cells and animals. The IRS trap mostly solves this, too,
retaining preservative without need for ballasting it with salt (which can
cause osmotic rupture of animals, Peterson & Dam 1990).(^) Indented rotating sphere traps have been modified to determine the distribution of
particle sinking speeds (Peterson et al. 2005). Since the rotating valve drops
accumulated particles intermittently, a stepping motor at the bottom of a fairly long
tube below the valve changes the collecting cups at intervals shorter than the
rotations, dividing the flux according to sinking speeds (despite some viscous “wall
effects”). Trull et al. (2008) applied this system between 200 and 300 m at both
productive and oligotrophic sites in the North Pacific and in the Mediterranean. At all
sites, they found a geometric distribution of settling speeds, with ∼50% of flux
sinking between 1000 and 100 m d−1, the remainder descending progressively slower
to 1 m d−1. Contradicting a general understanding (to be recited below anyway), the
relationship to particle size was not particularly strong, and, despite the general
consensus that mineral ballasting is critical, the amounts of inorganic calcium
carbonate and opal were not greater in the faster fractions.
(^) One extensive class of studies employs very large, conical traps with designs similar
to the PARFLUX series (Honjo et al. 1982): cones, apices down, with sides 14° from
vertical and a 1.54 m^2 opening with a honeycomb baffle in the mouth. They are
suspended above mooring anchors, buoyed up by large Pyrex™ spheres clustered
about the sides. Sedimenting material arriving at the apex cups is generally preserved
by a fixative solution. There are results from both single-cup versions and time-series