Bacterial biomass distribution in the global ocean

E. T. Buitenhuis, W. K. W. Li, M. W. Lomas, D. M. Karl, M. R. Landry, and S. Jacquet School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK Fisheries and Oceans Canada, Bedford Institute of Oceanography, Dartmouth, Nova Scotia, Canada Bermuda Institute of Ocean Sciences, St. George’s GE01, Bermuda, USA Department of Oceanography, University of Hawaii, Honolulu, HI 96822, USA Scripps Institution of Oceanography, University of California San Diego, La Jolla, California, USA INRA, UMR CARRTEL, 75 Avenue de Corzent, 74200 Thonon-les-Bains, France


Introduction
Heterotrophic bacteria are the main degraders of detritus in the ocean (Azam and Malfatti, 2007). Most bacteria (>95 %, Cho and Azam, 1988;Turley and Stutt, 2000) occur 15 as detached bacteria, living mostly on dissolved organic matter (DOM, with minor contributions from other energy sources such as reduced nitrogen). Attached bacteria living in and on particulate detritus, although less abundant, have a higher specific activity (up to 12 % of bacterial production, Turley and Stutt, 2000). Bacteria that spend part of their time attached to particles both attach and detach from particles on a timescale 20 of hours (Kiørboe et al., 2002). They also produce ectoenzymes that solubilize POC to DOC that can be subsequently used by detached bacteria (Thor et al., 2003;Azam and Malfatti, 2007). Thus, the relative importance of attached bacteria may be higher still than their contribution to bacterial production suggests.
Bacteria have a higher biomass than the metabolic theory of ecology would predict 25 based on their small size (Brown et al., 2004). This may be due in part to the fact 302 ESSDD 5,[301][302][303][304][305][306][307][308][309][310][311][312][313][314][315]2012 Bacterial biomass distribution in the global ocean E. T. Buitenhuis et al. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | that they respire organic matter that is formed as losses at all trophic levels, i.e. that their trophic status is unrelated to their size. Furthermore, not all bacteria show the same activity, ranging from ghost cells with a cell membrane but no internal structures, dead cells containing nucleic acids but with a compromised cell membrane, low nucleic acid cells with a lower specific activity and high nucleic acid cells (Gasol et al., 1999;5 Longnecker et al., 2006;Ortega-Retuerta et al., 2008;Morán et al., 2011). These dead or less active bacteria would contribute to a higher bacterial biomass than the metabolic theory would predict.
Here, we present a database of bacterial abundance and biomass in the global ocean. This is a contribution towards a world ocean atlas of plankton functional types 10 (MAREDAT, this special issue), which we hope will help resolve some of the important issues on ecosystem functioning and its representation by models. Table 1 summarises the data that were compiled for this synthesis. Most of the data were obtained by flowcytometry. Cells were stained with nucleic acid stains, and there-15 fore include (presumably recently) dead cells with compromised cell membranes, but not ghost cells. The data at BATS were stained with DAPI and counted microscopically, and could therefore include ghost cells. We treat Bacteria and Archaea as one group. In some cases, cyanobacteria will also have been included, especially Prochlorococcus near the surface, which have low red fluorescence and are therefore 20 difficult to distinguish from heterotrophic bacteria. The data are available from PAN-GAEA (http://doi.pangaea.de/10.1594/PANGAEA.779142) and the MAREDAT webpage (http://lgmacweb.env.uea.ac.uk/maremip/data/essd.shtml). 5,[301][302][303][304][305][306][307][308][309][310][311][312][313][314][315]2012 Bacterial biomass distribution in the global ocean   Table 2), but not enough data are available to define the controlling factors for this increase or how it graduates to the open ocean value with distance from the coast. We are also unaware of measurements showing how the carbon content of bacteria varies with growth conditions. We therefore use a single conversion factor of 9.1 fg cell −1 for the whole database.

Quality control
As a statistical filter for outliers, we applied the Chauvenet criterion (Buitenhuis et al., 2012) to the total carbon data. The data were not normally distributed, so we log transformed them, excluding 51 zero values. No high outliers were found by this criterion. The highest bacterial biomass in the database is 74 µg C l −1 , measured near the coast 15 of Oman.

Results
The database contains 39 766 data points. After gridding, we obtained 9272 points on the World Ocean Atlas grid (1 • × 1 • × 33 vertical layers × 12 months), i.e. we obtain a coverage of vertically integrated and annually averaged biomass for 1.3 % of the 20 ocean surface. Only 6 % of the data are from the Southern Hemisphere (58 % of the ocean surface; Fig. 1a), 24 % are from the tropics (43 % of the ocean surface), while 15 % are from the polar oceans (5 % of the ocean surface). Observations in the upper 112.5 m make up 57 % of the data (Fig. 1b) Although there are some zero values in the raw database, presumably because of a detection limit in small samples, there are no zero values in the gridded dataset. There is some sampling bias towards the growing season, with 72 % of the data sampled during the spring and summer months (Fig. 1c). The average abundance is 4.3 × 10 8 ± 3.9 × 10 8 bacteria l −1 with a median of 5 3.1 × 10 8 bacteria l −1 . The average biomass is 3.9 ± 3.6 µg l −1 (Fig. 2) with a median of 2.8 µg l −1 . The biomass decreases with depth, from 7.3 ± 4.3 µg l −1 at the surface to 0.36 ± 0.19 µg l −1 at 2750-4750 m depth (Fig. 3). The average biomass in the top 225 m is slightly higher in the Northern temperate region (23-67 • N, 5.5 ± 3.7 µg l −1 ) and tropics (5.5 ± 3.6 µg l −1 ) than in Antarctica (3.2 ± 1.9 µg l −1 ), the Arctic (2.4 ± 2.1 µg l −1 ) and 10 Southern temperate region (3.1 ± 1.9 µg l −1 ). The differences between most of these regions are significant (one-way ANOVA with violated homogeneity of variances, Games Howell post-hoc test, p < 0.001), except for Antarctica, for which there are only 23 measurements in the upper 225 m, and which was only significantly different from the tropics (p = 0.014).

15
If we calculate a total ocean bacterial biomass based on the average profile with depth ( Fig. 3) and multiplying by the volume of ocean water at each depth we calculate an inventory of 1.1 Pg C, of which 0.28 Pg C is found in the upper 225 m, and 0.51 Pg C below 950 m. If we calculate the inventory separately in the top 225 m for the 5 regions mentioned above, he inventory is higher at 0.35 Pg C due to the larger ocean volume 20 at low latitudes. Since we do not have enough data to calculate regional differences in the deep sea, this would bring the total ocean bacterial inventory up to 1.2 Pg C.

Discussion
We could find only few measurements of carbon content of bacteria that were measured directly after collection, i.e. without incubation, from open ocean waters (Table 2). 25 The range in these measurements is considerable, from 5.5 to 23.5 fg C cell −1 . Thus, there is a corresponding uncertainty in our conversion from cell abundance to carbon biomass. In addition, a higher conversion factor has been found in coastal waters (Fukuda et al., 1998). However, it has not been established how far this higher conversion factor extends between the coastal bay waters and the open ocean. If we assume the higher 5 conversion factor is valid up to a water depth to the bottom of 225 m (i.e. the continental shelf), then, based on the average profile of bacterial biomass (Fig. 3), increasing the conversion factor from 9.1 to 30.  Table 1 were measured on samples from the upper 250 m, so whether the conversion factor changes with depth is yet to be resolved. Whitman et al. (1998) estimate the global ocean bacterial inventory at 2.0 Pg C. This higher estimate is entirely due to their use of a higher conversion factor of 20 fg C cell −1 .