The concentration of bacteriophages in natural unpolluted waters is in general believed to be low, and they have therefore been considered ecologically unimportant. Using a new method for quantitative enumeration, we have found up to 2.5 x 10(8) virus particles per millilitre in natural waters. These concentrations indicate that virus infection may be an important factor in the ecological control of planktonic micro-organisms, and that viruses might mediate genetic exchange among bacteria in natural aquatic environments.

The possible roles of viruses in phytoplankton dynamics were investigated in seawater mesocosms with natural assemblages of phytoplankton growing under various nutrient regimes. Blooms of the marine coccolithophorid Emiliania huxleyi (Lohmann) Hay & Mohler were in some cases succeeded by an increase in the abundance of a morphologically homogeneous population of viruses. The viruses had an hexagonal outline and were about 180 nm in diameter. Similar viruses were found both intracellularly and associated with apparently lysed cells. Viral lysis could account for 25 to 100 % of the net mortality of E huxleyi when the blooms declined under non-limiting nutrient conditions or when the nitrate concentration was low. Production of viruses was limited when the phosphate concentration was low. Dur~ng a bloom of E. huxleyi in Norwegian coastal waters in 1990 we found that the concentration of free algal viruses was increasing at the end of the bloom, indicating that viruses of E. huxleyi may be important under natural conditions as well. These results suggest that viral mortality of phytoplankton may be an important factor regulating community structure, diversity and biomass production in marine environments.

The content of carbon, nitrogen, oxygen, phosphorus and sulfur was measured in individual cells from 6 native aquatic samples and 4 samples of cultured bacteria by X-ray microanalysis using a transmission electron microscope (TEM). The molar C N:P ratio for the pooled sample was 50:lO:l From length and width measurements of unfixed air-dried cells we estimated cell volumes over a total range of 0.0026 to 15.8 pm3. and mean C:volume ratios of 30 to 162 fg for the samples included. For the marine samples we found mean N:C ratios of 0.25 to 0.28, while cells from fresh or brackish waters had mean N:C ratios of 0.17 to 0.20, indicating differences in nutrient availability. The P:C ratios for the samples analyzed varied from 0.040 to 0.090, with a pooled mean of 0.052, which is approximately twice that of the Redfield ratio for P C . For 0:C ratios we estimated a pooled mean of 0.37 and a range of 0.22 to 0.77 for all samples. We may conclude that slow-growing or non-growing cells have low 0 -C ratios. The mean S:C ratio for all samples was 0.031, with a range of 0 016 to 0.084 for the sample means. A general conclus~on is that single-cell analyses of elemental composition give important information on the physiological conditions of cells and on possible nutrient limitations. The rationale for this is the assumption that changes in macromolecular composition are due to nutrient availability.

The quantitative significance of aquatic viruses in coastal and in lake water was investigated. The number of viruses in marine surface waters was found to change on a diurnal basis along with changes in number of bacteria and bacterial activity. By inhibiting the production of viruses, we were able to measure viral decay rates up to 1.1 h-' in marine systems, and up to 0.6 h-' in a freshwater lake, for the majority of the viral population. A minor fraction (4 to 40 %) of the viral population was found to have decay rates lower than 0.05 h-' The fraction of bacteria containing mature virus particles ranged from 2 to 16 % , and the number of viruses released from these bacteria was on average about 50 [range 10 to 300). From these results we eshmate that phages may lyse 2 to 24 % of the bacterial population per hour. Phages may thus be a major cause of bacterial mortality in aquatic ecosystems and may have a significant impact on the carbon and nutrient flow in aquatic food webs.

We estimated prokaryotic mortality due to viruses and bacterivores through salinity gradients in 2 solar salterns. In each saltern system, successive ponds provided steady state environments with a range of salinities from 37 to 372%. Prokaryotic and viral abundance increased with salinity, reaching about 10' prokaryotic cells ml-l and 10q virus-like particles (VLP) ml-' at salinities higher than 250%0. Prokaryotic doubling times became longer than 2 d above 250% salinity until the end of the gradient. Bacterivory accounted for all the production at lower salin~ties but it was found to be zero at the highest salinities. The percentage of visibly infected cells was not different among the ponds where infected cells could be detected and it was always lower than 4 %. From the percentage of infected cells and using conversion factors from the literature we estimated rates of prokaryotic mortality due to viral lysis: about 0.6 to 2 X 10hprokaryotes ml-l were lysed daily by the viruses in the salterns. This number represented a low percentage of prokaryotic abundance and production compared to the prokaryotlc losses due to bacterivores (0.2 to 4 x 107bacteria ml-' d-l). However, viral production reached values higher than 10' VLP ml-l d-l above 250% salinity, due to the large burst size (200 viruses cell-') found in a part~cular morphotype of prokaryotes, the square archaea. These archaea represented more than 25% of the prokaryotic dssemblage above 250%0 salinity. At this point they became the prokaryotic morphotype with the largest percentage of infected cells (1 to 10% of square archaea with visible phages inside). A lemon-shaped virus (sim~lar to one described for some other groups of archaea) was found infecting squdre archaea, its abundance increased in the saltiest ponds together with that of the square archaea. In this system viruses did not exert a strong control over the prokaryotic abundance and growth rate.