Factors controlling bacterial abundance, biomass and growth at the Bermuda Atlantic time-series site
The Bermuda Atlantic Time-series Study (BATS) site represents a unique hydrographical oceanic environment to study the factors controlling bacterial growth dynamics. The region was sampled at monthly intervals from 1991 to 1996. A typical annual cycle was defined by a deep winter mixing, followed by an increasing stratification of the mixed layer through summer and fall. There were striking seasonal patterns in phytoplankton productivity with a strong maximum immediately following the deep winter mixing and the intrusion of inorganic nutrients. Bacterial growth rates showed a similar pattern but had a secondary peak in late summer/fall of the same magnitude as the spring bloom. Bacterial abundance showed only slightly elevated concentrations in spring. A number of time course storage experiments showed that bacterial abundance decreased by 24-50 % within 7-29 days in samples preserved with 2.5 % glutaraldehyde. By adding a protease inhibitor prior to the addition of glutaraldehyde, the loss of bacterial cells was reduced to 17-18 % over a 30 day period. These findings lead to the recommendation that samples for bacterial abundance should be processed immediately for epifluorescence enumeration. An average of 47 % of all bacterial cells passed the pore size of a Whatman GF/F filter and these viable cells should be included in biomass estimates. An annual average of 26 % of estimated C settled below the spigot of a Niskin water sampler. Consequently, C measurements made on GF/F filters must account for the particles settling below the spigot of a water sampler as well as the number of bacteria lost during the process of filtration. This study is the first to present single cell elemental C, N and P measurements from natural bacteria in the Sargasso Sea. A wide range in elemental content was found between single cells and this could be expressed as a function of the cell size. By applying an average cell volume, an annual average of 10 fg C, 1.9 fg N and 0.28 fg P was calculated per bacterial cell. The average percentage integrated stocks of C in the upper 250 m of the water column, was 20 (phytoplankton), 18 (microheterotrophs) and 62 (other non-living detrital matter). Bacterial biomass was higher than phytoplankton outside the spring bloom period, but non-living carbon showed an overall dominance through out the year. Phytoplankton generation time was relatively constant over the season. Bacterial generation time was ten times longer and showed a greater seasonal variation, but largely followed the changes in primary production. Assuming that 50-70 % of the bacterial cells were non-living, the mean bacterial generation time was estimated to be 7 times (0-60 m) and 1.4 times (80-140 m) longer than phytoplankton generation time. During the spring bloom event, an average of 85 % of the bacterial growth rate was removed by grazing and viral lysis. This was the only noted decoupling between growth and removal of bacteria at BATS. During the remainder of the year bacterial growth was balanced by the loss rate, due to grazing and viral lysis. A linear relationship was found between net DOC accumulation and primary production in natural surface waters at BATS. Phytoplankton net DOC excretion constituted 42 % of the primary production rate, while Trichodesmium colonies only excreted 12 % (puffs) and 23 % (tufts). By using a conservative estimate of the bacterial growth efficiency (14 %) and the net DOC accumulation rate from this study, gross DOC excretion was equivalent to the rate of primary particulate C production. Results from this study suggests that bacteria at the BATS site are using the majority of the DOC generated by primary production. Prior to the spring bloom and the associated increase in DOC excretion, bacteria appeared to be C limited in the surface waters at BATS. Following the peak in primary production and coinciding with the depletion of inorganic nutrients in the euphotic zone, the bacterial cells became less C starved, but never reached a true N or P limitation. Regenerated nutrients from grazing and viral lysis of bacteria and new production by diazotrophic Trichodesmium colonies and trichomes, may support the bacteria with N and P in the euphotic zone in summer and fall. The substrate dependent growth and increase in biomass of bacteria exhibited at BATS in spring, is indicative of a bottom-up controlled system, whereas the bacteria appeared to be topdown controlled by grazing and viral lysis for the remainder of the year.