gradients relative to the body size of a bacterium (Dade et al., 1990), and the tumble-and-run approach makes good sense in the episodically stirred gut environment. E. coli runs in a straight line, then tumbles and goes off at a random heading from the original. Run duration increases with nutrient concentration, biasing the random walk and moving the bacterium up-gradient (Berg, 1993).
Most suggested sources of dissolved organic matter for bacteria are small in length and hence subject to rapid diffusion. In the case of leaking or exuding phytoplankton, the source may be quasi-steady, but in the cases of sloppy feeding, fecal pellet ejection, or autolysis, the source will be not only small but also short lived. An important question is whether marine bacteria show the same chemotactic behaviors as E. coli. Small bacteria apparently cannot find their way up gradient, and taxis is not practiced by bacteria smaller than about 0.6 gin in diameter (Dusenbery, 1998a). Below this size, the course of the bacterium would be changed too rapidly by Brownian rotation to allow directed movement. Conversely, the elongate shapes of chemotactic bacteria generally observed are particularly resistant to Brownian rotation and thereby increase the time over which the organism practically can integrate stimulus strength, and hence increase greatly its sensitivity to chemical gradients (Dusenbery, 1998b).
Recently, swimming paths of large marine bacteria have been recorded optically by taking advantage of light scatter under dark-field illumination (Blackburn et al., 1998). Unlike E. coli, they do not turn in a random direction at the end of a run but instead double back at close to 180° from the original heading, with some deviation due to Brownian rotation. This visualization was done in still water, but simulations in shear fields expected from decaying turbulence suggest that doubling back is far more effective than the tumble-and-run behavior in staying near a small, spherical source (Luchsinger et al., in review). There clearly is diversity in chemotactic strategies and patterns among marine bacteria (e.g., Barbara and Mitchell, 1996); E. coli is no longer an appropriate universal model. Models also suggest yet undocumented tactics. Quite contrary to intuition, bacteria can in principle use spatial sensing (difference in concentration on two parts of the cell) rather than temporal sensing (difference in concentration over time) to detect stimulus gradients (Dusenbery, 1998c).
Letting released enzymes do the searching for particulate material appears to be another useful bacterial strategy in either large aggregates of particles or in sediments (Vetter et al., 1998; Vetter and Deming, 1999) and explains the paradox of oversolubilization of aggregates by bacteria (more soluble products made than used; Smith et al., 1992). Further challenges to biological-physical modeling, to measurement of both bacterial tactics and carbon dynamics and even to the discrimination of dissolved from particulate carbon are the rapid self assembly and state changes of biogenic polymer gels (Chin et al., 1998).
The conceptual simplifications provided by Kiørboe and Visser (1999), steadily increasing abilities to, visualize flows and organisms both optically and acoustically, and increasing computational capabilities poise the study of fluid dynamic and chemical interactions with and among organisms for rapid advance. These advances promise in particular to help understand the vast and beautiful morphological diversity of protists and phytoplankton (though some modem classifications include the phytoplankton and even macroalgae with protists) living at low Reynolds numbers. Easier numerical modeling than at high Reynolds numbers is partial compensation in these regimes for lack of intuition about life in a fluid environment filled with dynamical chemical and physical signals.
Over what spatial scales are marine populations connected via dispersal of early life stages?
What are the dynamics of marine food webs, and how will they respond to environmental perturbations?
As models and synoptic data now are used to forecast the weather, can one forecast changes in physical-chemical-biological interactions in the sea that affect fisheries yields, food web dynamics, and ecosystem services that the sea provides?
This section covers population and community ecology but in the specific context of the ocean. Important facets of this context are the unquantified connectedness of subpopulations and the pervasiveness of fluid transport of propagules. This unquantified connectedness remains the greatest obstacle to rational establishment of marine preserves and management policy in general.
The OEUVRE workshop, to no one's surprise, cited identification of strong indirect effects through experimental manipulation as one of the great successes of marine ecology over the last 30 years and prediction of which interactions would be strong ones among the greate it challenges for the future. For few communities are the majority of interaction strengths known, but it is clear that in communities of even modest diversity the majority are weak (Paine, 1992). and identifying the important ones by manipulating all the species individually is a daunting empirical task. I left the workshop seeing no clear route to progress through predictive theory, either.
Theory suddenly has jumped to the rescue and shifted perspective by 180° by putting focus not on the strong interactors but on the weak ones. McCann et al. (1998) departed from classic food web models in two ways. They modeled functional responses as saturating rather than linear in prey concentration, and they allowed population abundances to be away from equilibrium values. With these additions to realism, food web models better reflect the added