FIGURE 1. The two perspectives of copepod mechanosensing as predator and prey, respectively. A few mechanosensory hairs are sketched in on one antenna and the opposite caudal furca. The parcel of water within which the copepod is embedded is indicated by the dashed ellipse. A. Detection of small prey is via local perturbation of the velocity field. B. Detection of a predator by deformation of the water parcel in which the copepod is embedded. Qualitative features of the deformation are shown in the ellipse containing arrows. Based on analyses in Kiørboe and Visser (1999).

tematic decomposition of fluid dynamic phenomena into their constituent motions.

Kiørboe and Visser (1999) have idealized the motion of both prey and predator as flow around a sphere and have provided clear intuition for the mechanical perspective of a copepod in detecting and distinguishing moving prey and predators (Figure 1). This intuition comes not only from the calculated decomposition, but also from clever experiments that expose copepods to simplified and easily quantified fluid dynamic components of this decomposition. Arrays of mechanoreceptors on the antennae detect prey as local velocity variations. Independently performed experiments and numerical simulations (Bundy et al., 1998) show that the flow field generated by a swimming copepod also can contribute to detection of nonmoving, particulate prey.

Predators of copepods, on the other hand, are large in comparison with the copepod and are detected as larger-scale flow-field deformations that influence the whole space monitored by the copepod; the copepod knows that a predator is near when the stimulus affects the full array of sensors but deflects them with spatially varying velocities. Successful fish predators detect the copepod at a distance visually and decelerate as they approach, dropping the deformation produced by their bow wave below the threshold intensity that elicits the copepod's "jump" escape behavior. This threshold sits orders of magnitude above the neurophysiological detection limits of the mechanoreceptors and just above the level of deformation caused routinely by ambient turbulence. Kiørboe and Visser's (1999) analysis brings intuitive understanding of the process.

The qualitative gain in intuition for mechanoreception by copepods itself is compelling, but the gain is even more impressive when stimuli are quantified. Clearance rates by the ubiquitous omnivore Oithona similis are predictable over three orders of magnitude and on radically differing food particles (swimming protists and settling fecal pellets). Moreover, detection of settling particles by coprophagous copepods is found in calculations and experiments to be highly nonlinearly but predictably dependent upon particle size and settling velocity, to the point where detection of the most rapidly settling pellets at natural copepod abundances is virtually certain. Since McCave's (1975) seminal analysis, large particles or aggregates have been thought to account disproportionately for the flux of material to the sea-floor. Kiørboe and Visser's (1999) analysis suggests instead that particles of intermediate settling velocity that provide less mechanical stimulus for detection may be more successful in running the suspension-feeder gantlet. Particle-type-and settling-velocity-dependent degradation rates are among the most poorly constrained parameters in global carbon budgets, and this new analysis of mechanosensory abilities provides substantial help in the form of new perspectives and predictions from an unexpected direction. Suddenly complicating this range of issues in vertical transport of carbon still further is the documentation of spontaneous assembly of gels (Chin et al., 1998).

Another important part of the carbon cycle is uptake of dissolved organic carbon by heterotrophic bacteria. Escherichia coli, a resident of the human large intestine and colon, has provided the universally used model of chemotaxis. Digestion in humans can be expected to yield large-scale