Our conceptual understanding of the role of heterotrophic bacteria in pelaglc ecosystems and in ocean biogeochemical cycles is closely linked to our understanding of how their growth rate, abundance, and diversity is controlled. Here we discuss consequences of the simplifying assumption that there are only 5 potentially important interactions between heterotrophic bacteria and their biological and chemical environment. We consider 3 possible types of growth rate limitation: (1) organic carbon, (2) inorganic phosphate, and (3) organic and inorganic nitrogen; and 2 types of cell losses: (1) predation by heterotrophic flagellates, or (2) lysis by infectious viruses. Incorporating this into sirnple food web structures, we discuss 4 classes of models, 2 based on carbon limitation and 2 based on mineral nutrient limitation of bacterial growth rate. Bacterial abundance is assumed to be controlled by protozoan predation in all cases. For each class, we derive expressions describing bacterial carbon demand, and dscuss the control of bacterial carbon demand, growth rate and diversity. It is shown how models predicting an ecosystem production of dissolved organic carbon (DOC) exceeding bacterial carbon demand may be constructed assuming either a low degradability of the DOC, or mineral nutrient h i t a t i o n of bacterial growth rate. For 2 classes of models, infectious viruses are shown to affect neither growth rate nor abundance of the steady state bacterial community. For all 4 classes of models, viruses are suggested to control diversity of the steady state bacterial community.

Growth and loss of planktonic bacteria are thought to be roughly in balance, but are rarely measured together. The loss rate for a bacterial assemblage in surface waters of Villefranche Bay (NW Mediterranean Sea) was estimated using 2 independent techniques. The disappearance rate of from cold-TCA-insoluble material following a labeling of the natural assemblage with 3H-thymidine gave a turnover of 2.2% h-', while the disappearance of 32P from the bacterial size fraction (0.2 to 1 pm) following an initial uptake period and a subsequent cold chase with orthophosphate gave a bacterial turnover rate of 2.5% h-' The similarity of the 2 estimates suggests that the same loss processes were measured and that processes independent of bacterial population turnover, such as rapid uptake and release of labels, were of minor importance. The mortality estimates were close to thymidine-based production estimates of 2 to 2.3% h-'. Viral abundance (ca 2 X 106 ml-l) was about 3 to 4 times that of bacteria, and relatively constant. Attempts to measure bacterial mortality due to viral infection were complicated by filtration artifacts. Passage of the thymidine-labelled assemblage through a 0.6 pm filter in order to separate bacteria and viruses from larger bacterivorous organisms removed 60% of the bacterial label. Label loss rates were undetectable in the flltered assemblage over 96 h incubations, suggesting that viruses were minor loss agents of this (minority) si.ze fraction of bacteria. In the experiments w~t h 32P, most of the label was transferred directly from the bacterial size fraction to dissolved compounds, with relatively minor amounts (10 to 20%) transferred to larger size fractions.