Picophytoplankton biomass distribution in the global ocean

The smallest marine phytoplankton, collectively termed picophytoplankton, have been routinely enumerated by flow cytometry since the late 1980s during cruises throughout most of the world ocean. We compiled a database of 40 946 data points, with separate abundance entries for Prochlorococcus , Synechococcusand picoeukaryotes. We use average conversion factors for each of the three groups to convert the abundance data to carbon biomass. After gridding with 1 ◦ spacing, the database covers 2.4 % of the ocean surface area, with the best data coverage in the North Atlantic, the South Pacific and North Indian basins, and at least some data in all other basins. The average picophytoplankton biomass is 12 ±22μg C l−1 or 1.9 g C m−2. We estimate a total global picophytoplankton biomass of 0.53–1.32 Pg C (17–39 % Prochlorococcus , 12–15 % Synechococcus and 49–69 % picoeukaryotes), with an intermediate /best estimate of 0.74 Pg C. Future e fforts in this area of research should focus on reporting calibrated cell size and collecting data in undersampled regions. http://doi.pangaea.de /10.1594/PANGAEA.777385 Published by Copernicus Publications. 38 E. T. Buitenhuis et al.: Picophytoplankton biomass distribution in the global ocean


Introduction
Picophytoplankton are usually defined as phytoplankton less than 2 or 3 µm diameter (e.g.Sieburth et al., 1978;Takahashi and Hori, 1984;Vaulot et al., 2008).They are the smallest class of phytoplankton and are composed of both prokaryotes and eukaryotes.The eukaryotes (0.8-3 µm) are a taxonomically diverse group that includes representatives from four algal phyla: the Chlorophyta, Haptophyta, Cryptophyta and Heterokontophyta (Vaulot et al., 2008).The prokaryotes belong to the phylum cyanobacteria and are subdivided into the genera Prochlorococcus (∼ 0.6 µm) and Synechococcus (∼ 1 µm), with each group having many ecotypes that dominate in different ocean regions (Johnson et al., 2006).
Picophytoplankton tend to dominate the phytoplankton biomass under oligotrophic conditions such as in the subtropical gyres (Alvain et al., 2005), where their high surfaceto-volume ratio makes them the best competitors for low nutrient concentrations (Raven, 1998).The abundance of the prokaryotes is often inversely related with the eukaryotes, which are favoured by more physically active mixed layers (e.g.Boumann et al., 2011).Furthermore, with warming of the temperate to subpolar North Atlantic and the Canadian high Arctic, picophytoplankton (specifically picoeukaryotes) have been found to become an increasingly large fraction of the total chlorophyll (Li et al., 2009;Moran et al., 2010).
As part of the marine ecology data synthesis effort (MAREDAT, this special issue), we compiled a database on picophytoplankton in the global ocean.MAREDAT is a community effort to synthesise abundance and carbon biomass data for the major lower trophic level taxonomic groups in the marine ecosystem.It addresses both autotrophs and heterotrophs and covers the size range from bacteria to macrozooplankton.

Data
We compiled data for picophytoplankton abundance in three taxonomic groups: Prochlorococcus, Synechococcus, and picoeukaryotes (Table 1).We used the size range of picoeukaryotes as defined by the contributing researchers.The size range has a large impact on the resulting biomass (see Discussion).All of the data were obtained by flow cytometry.Both the raw data and the gridded data are available from PANGAEA (http://doi.pangaea.de/10.1594/PANGAEA.777385) and the MAREDAT webpage (http:// maremip.uea.ac.uk/.maredat.html).

Conversion factors
Conversion factors from cell abundance to carbon biomass for the three picophytoplankton groups were compiled from the literature (Table 2).Conversion factors were either measured directly on unialgal cultures in the laboratory or derived from indirect methods on in situ samples.Most of the indirect measures were derived from cell sizes that were estimated from average forward-angle light scatter (FALS) multiplied by a carbon content per biovolume.The conversion factors of Veldhuis et al. (1997) were based on nitrate uptake in incubated in situ samples and assuming a C : N ratio of 6.Since the biggest source of variability in the other indirect measures is the carbon content per biovolume, which was measured on laboratory cultures, the advantage of using in situ biovolume to determine conversion factors does not seem to improve the local applicability of these data, and we therefore used the directly measured conversion factors as the standard.

Quality control
Contributed data were assumed to have undergone the contributing researchers' own internal quality control procedures.As a statistical filter for outliers, we applied the Chauvenet criterion (Buitenhuis et al., 2012b) to the total carbon data.The data were not normally distributed, so we logtransformed them, excluding zero values.No high outliers were found by this criterion.The highest picophytoplankton biomass in the database is 575 µg C l −1 , measured near the coast of Oman (Indian Ocean).

Results
The database contains 40 946 data points (Fig. 1).Data are included from a number of stations that have been sampled repeatedly over many years or programs where measurements have been made on a fine-resolution grid.Therefore, after gridding, we obtained 10 747 data points on the World Ocean Atlas grid (1 • × 1 • × 33 vertical layers × 12 months), representing a coverage of vertically integrated and annually averaged biomass for 2.4 % of the ocean surface.For further details on the gridding, see Buitenhuis et al. (2012b)  the overrepresentation of well sampled locations, we present results of the gridded data.Only 15 % of the data are from the Southern Hemisphere (Fig. 2a), 33 % are from the tropics (43 % of the ocean surface), while 13 % are from the polar oceans (5 % of the ocean surface).Observations in the upper 112.5 m make up 81 % of the data (Fig. 2b), but the number of observations decreases more slowly than biomass (Fig. 3), and there are still 480 observations at 200 m depth (Fig. 2b), thus defining the vertical biomass profile fairly well.Zero values make up 1.6 % of the data, and 95 % of those are from below 62.5 m depth.There is some sampling bias towards the growing season, with 67 % of the data sampled in the spring and summer months (Fig. 2c).The average picophytoplankton biomass is 12 ± 22 µg C l −1 (Fig. 4) or 1.9 g C m −2 .Of the vertically integrated biomass, 54 % occurs in the upper 40 m and 93 % in the upper 112.5 m (Fig. 2).Synechococcus is found at the most shallow depths (97 % above 112.5 m, Fig. 5), followed by picoeukaryotes (92 % above 112.5 m), while Prochlorococcus biomass decreases more slowly with depth (87 % above 112.5 m).
The average biomass is slightly higher in the tropics and considerably lower in the Arctic (Figs. 4, 6), but the standard deviation within latitudinal bands is high, so that none of the differences are significant.Antarctica: 11 ± 8 µg C l −1 or 1.2 g C m −2 , south temperate zone (67-23 • S): 13 ± 23 µg C l −1 or 2.2 g C m −2 , tropics: 15 ± 24 µg C l −1 or 2.2 g C m −2 , north temperate zone: 12 ± 22 µg C l −1 or 1.9 g C m −2 , and Arctic: 6 ± 8 µg C l −1 or 0.6 g C m −2 .We calculate the global picophytoplankton biomass from the zonal and time-averaged concentration filled by interpolation across up to 22 • latitude (Fig. 6) multiplied by the volume at each latitude and depth, integrating to the bottom and counting missing values as 0. We thus estimate a total global picophytoplankton biomass of 0.74 Pg C (17 % Prochlorococcus, 15 % Synechococcus and 69 % picoeukaryotes).Interpolation across up to 10 • latitude only leaves a few missing values and estimates 0.73 Pg C. If we use the indirect in situ conversion factors for each of the three groups (Table 2), the total biomass (with up to 22 • interpolation) is 0.53 Pg C (39 % Prochlorococcus, 12 % Synechococcus, 49 % picoeukaryotes).Picoeukaryotes tend to dominate by > 75 % poleward of 40 • , and dominate below 62.5 m depth in the tropics and below 225 m everywhere (Fig. 7).Prochlorococcus tends to dominate above 225 m between 20-40 • N and shares dominance with picoeukaryotes between 10-30 • S and at the surface in the tropics.Synechococcus only dominates around 50 • S and is relatively abundant above 62.5 m between 10-40 • N.This is consistent with the community structure of picophytoplankton that has been analysed by Bouman et al. (2011).

Discussion
Although data coverage, at 2.4 % of the ocean surface, is by no means complete, if we randomly select half of the depth profiles in 10 random samples, the average integrated biomass varies between 96 and 104 % of the value for the whole dataset, while the averages from the Southern and Northern Hemispheres are 119 % and 96 %, respectively.On the other hand, the average using the indirect in situ conversion factors is 72 % of the value estimated using the direct conversion factors.Thus, the main uncertainty in determining the global picophytoplankton biomass in this analysis is the conversion from cell abundance to carbon biomass.There is a fairly tight relationship between forward-angle light scatter (FALS; Cavender- Bares et al., 2001;DuRand et al., 2002) or right-angle light scatter (RALS; Simon et al., 1994;Worden et al., 2004), as measured by flow cytometry, and cell size, which is probably the main source of uncertainty in the conversion factor.Only about a third of our data came with FALS or RALS data, and even in those cases these were in arbitrary units.We recommend the routine measurement of calibrated size as the additional measurement that would do most to improve our knowledge of global picophytoplankton biomass distribution.
www.earth-syst-sci-data.net/4/37/2012/ Earth Syst.Sci.Data, 4, 37-46, 2012  In addition to the uncertainty in the carbon conversion factor, there is uncertainty about the abundance of Prochlorococcus in near-surface oligotrophic waters, where the cellular chlorophyll content, and thus the ability to detect them as algae from their red fluorescence, is at its minimum and near the detection limit of standard flow cytometers (Dusenberry and Frankel, 1994).
It has been repeatedly shown that Prochlorococcus and Synechococcus increase in cell size with depth up to ∼ 150 m.In contrast, previously published results for picoeukaryotes showed little variation in size as a function of depth (Li et al., 1993;DuRand et al., 2001;Grob et al., 2007).We compared the increase in size for the three groups at two locations.At BATS (Bermuda Atlantic Timeseries Station; which includes the data of DuRand et al., 2001), we also find an increase in cell size of Prochlorococcus and Synechococcus but not picoeukaryotes (Fig. 8a).However, in the Western Mediterranean (Almo-1 and -2, Jacquet et al., 2010), we find a similar increase in cell size of Prochlorococcus and Synechococcus, but a much larger increase of picoeukaryotes (Fig. 8b).The difference this could make to the global picophytoplankton biomass is large.If we use the standard conversion factors in the surface and increase these linearly up to a factor 3 below 150 m depth (blue lines in Fig. 8), then the global biomass becomes 1.32 Pg C (+78 %), or if we only apply this increasing conversion factor to Prochlorococcus and Synechococcus, we estimate a global biomass of 0.93 Pg C (+25 %).Our standard conversion factors are taken from laboratory studies.Conversion factors for heterotrophic bacteria from laboratory studies tend to be higher than from in situ measurements (Buitenhuis et al., 2012a).Indeed, even if we do not account for an increase of cell size with depth, the laboratory conversion factors lead to a higher biomass estimate than the indirect conversion factors.Other sources of variability are seasonal variations of cell size (DuRand et al., 2001) of all picophytoplankton and increasing cell size of Prochlorococcus with latitude towards the equator (Viviani et al., 2011).Thus, it is clear that there is considerable uncertainty in the conversion factors, but in the absence of general trends for the cell size variability of each group under all conditions, our estimate of 0.74 Pg C represents our best estimate of the global picophytoplankton biomass.
Le Quéré et al. (2005) estimated that the global picophytoplankton biomass, including nitrogen fixers, is 0.28 Pg C. Our estimate, excluding nitrogen fixers, is considerably higher at 0.74 Pg C, and even our estimate using the indirect conversion factors is still almost double at 0.53 Pg C. Le Quéré et al. (2005) suggested that a third of global phytoplankton biomass is in the pico size class.Therefore, a 2-3fold difference in the estimated picophytoplankton biomass would not only be important for calculating the relative contribution that picophytoplankton make to the phytoplankton but also for calculating the total biomass of phytoplankton as the base of the ocean ecosystem.
For picoeukaryotes, the definition of the size range to be included is a major source of ambiguity.Whether phytoplankton between 2 and 3 µm diameter are included as picophytoplankton not only affects the abundance of the picoeukaryotes, but also which conversion factor is applicable.Here, we have included measurements of cells up to 3 µm diameter in the carbon conversion factor (Table 2).As a consequence, our conclusion that picoeukaryotes constitute 69 % of global picophytoplankton biomass critically depends on the definition of the size cut-off.
In summary, thanks to the routine use of flow cytometry for measurement of picophytoplankton abundance, we obtained a global dataset with reasonable coverage.The two main issues that deserve future attention are better resolution of cell sizes and better sampling coverage in the Southern Hemisphere.

Figure 1 .
Figure 1.Horizontal distribution of the number of observations.Data points have been enlarged to 5 • × 5 • .

Figure 2 .
Figure 2. Number of grid points with data, as a function of (A) latitude.(B) Depth.Observations below 300 m are not shown (1.4 % of the data).The deepest observation is at 3000 m and the deepest nonzero observation at 1100 m. (C) Time.Red: Southern Hemisphere, black: total.

Figure 8 .
Figure 8. Cell size as a function of depth, normalised to cell size at the surface, (black) Prochlorococcus, (red) Synechococcus, (green) picoeukaryotes, (blue) exploratory conversion factor that increases up to a factor 3 below 150 m depth.(A) At BATS, (B) in the Western Mediterranean (Almo-1 and -2).

Table 1 .
. To limit Data sources.

Table 2 .
Cell abundance to carbon biomass conversion factors [fg C cell −1 ].