It is a common belief among microbial ecologists that planktonic heterotrophic prokaryotes in the upper layers of aquatic ecosystems are strongly dependent on the organic matter synthesized in situ by phytoplankton (Baines & Pace, 1991). This idea has been central to our understanding of the role of microorganisms in biogeochemical cycling (Azam, 1998). Fouilland & Mostajir (2010), by examining concurrently measured rates of particulate primary production and heterotrophic bacterial production (BP), question this view and arrive at a controversial conclusion: that bacteria do not depend on phytoplankton products for growth and metabolism. The fact that bacterial carbon demand (BCD) exceeds gross primary production in oligotrophic waters has fostered a long debate (del Giorgio & Duarte, 2002; Karl et al., 2003). In connection with this, Fouilland and Mostajir claim for a major shift in a well-established aquatic microbiology paradigm.
Two important criticisms arise from their analysis and conclusions. Our first concern is methodological, because the authors made several assumptions that affected the reliability of their estimates. The excess of BCD over total PP was detected mainly for low productive waters (see their Fig. 3), where accurate estimates of dissolved primary production (DPP), BP and bacterial respiration (BR) are difficult to obtain. BP estimates in oligotrophic systems can be strongly influenced by the conversion factors (CFs) used to transform leucine or thymidine uptake rates into biomass production (Gasol et al., 2008). Two thirds of the studies included in the analysis used the thymidine method (Table 1) and applied a single CF (2 × 1018 cells mol−1 thymidine assuming 20 fg C per cell) that yields unrealistically high BP estimates for oceanic samples (Ducklow, 2000). Leucine-based studies also used CFs that are approximately one order of magnitude higher than empirical determinations in oceanic oligotrophic environments (e.g. Alonso-Sáez et al., 2007). Because the total bacterial carbon flux was calculated from BP measurements, considering this initial source of variability that can lead to a large overestimation of BP values becomes essential.
In order to estimate BR, the authors applied Robinson's (2008) empirical equation relating BP and BR, which included data only from marine systems. Besides compromising statistical independence, this is only one of the models available for predicting BR or the related variable bacterial growth efficiency (BGE=BP/BCD) when direct measurements are lacking (del Giorgio & Cole, 1998, 2000). Using the model by Robinson (2008), the lowest range of BP estimates (0.01–0.1 μg C L−1 day−1) yielded a BR range of 0.72–2.75 μg C L−1 day−1 (see their Fig. 3b), equivalent to BGEs of 1.4–3.5%. The methodological approaches to estimate BR and BGE in situ are controversial, including size fractionation and the relatively long incubation times. However, the minimum BGEs resulting from BR vs. BP empirical models are clearly lower than the values directly measured in oceanic environments (usually above 5%, Robinson, 2008). The result of applying a BGE of 1%, 5% or 12% (the open-ocean average in del Giorgio & Cole, 2000) to the lower limit of BP values in Fouilland and Mostajir's analysis (i.e. 0.01 μg C L−1 day−1) would yield a one order of magnitude difference in BR (0.07–0.99 μg C L−1 day−1). The consequences for the scatterplots in their Fig. 3b, c are obvious, with data lying closer to the 1 : 1 relationship with increasing BGE values. This effect would be even more pronounced if BP values were overestimated, as explained above. The pivotal role of BR and BGE in their calculations is worth more detailed arguments, and applying models that use independent variables to estimate these parameters (e.g. López-Urrutia & Morán, 2007) could also help constrain a range of realistic BCD values. Remarkably, in contrast to the results of the study and using the same model, Robinson (2008) found that BCD was generally met by total PP across a latitudinal transect in the Atlantic, rendering unnecessary external inputs of organic carbon.
Our second major objection to the paper is conceptual: if bacteria do not depend on primary production, what do they rely on? Other carbon sources vaguely mentioned in the article are ‘sloppy feeding […], viral lysis and external dissolved organic carbon (DOC) inputs’. Grazing by protozoa and mesozooplankton has been recognized as a major carbon source, making up to 65% of labile DOC rapidly consumed by bacteria (Nagata, 2000). However, despite involving other trophic levels, its ultimate origin is photosynthetic and thus should not be considered a nonphytoplanktonic carbon source. Perhaps it is just a question of where to draw the line between autochthonous and allochthonous carbon. The same would apply to bacterial utilization of carbon produced in previous episodes of phytoplanktonic production. The lack of synchronicity between BP and PP in oceanic systems is argued throughout the paper as a proof of the bacterial independence on phytoplankton. Temporary uncoupling of phytoplankton production and bacterial consumption has been reported in different marine systems (Carlson et al., 1994; Thingstad et al., 1997), partly associated with inorganic nutrient limitation. Yet, the existence of time lags between DOC production and consumption does not prove bacterial independence on the carbon produced by phytoplankton, at least on an annual basis.
There is compelling evidence that bacterial heterotrophs rely on the most labile fraction of DOC, which appears to be strongly linked to primary production (Davis & Benner, 2007). A different story would be to demonstrate that bacteria rely mainly on old, refractory DOC. Although the consumption of old DOC has been reported, it does not seem to be a major carbon source for bacteria (Cherrier et al., 1999). Photochemical and other types of transformations of refractory DOC are common (Moran & Zepp, 1997), but the role of these processes in sustaining BCD is unclear. The global contribution of atmospheric and coastal inputs has also been estimated to be rather low as compared with primary production (del Giorgio & Duarte, 2002). Until it can be proved that a substantial amount of BCD is not met either by freshly plankton-produced substrates or by previously accumulated DOC on a seasonal scale, the independence of bacteria on phytoplankton will have to wait its turn in the list of new paradigms.