Small Phytoplankton and Carbon Export from the Surface Ocean

See allHide authors and affiliations

Science  09 Feb 2007:
Vol. 315, Issue 5813, pp. 838-840
DOI: 10.1126/science.1133471


Autotrophic picoplankton dominate primary production over large oceanic regions but are believed to contribute relatively little to carbon export from surface layers. Using analyses of data from the equatorial Pacific Ocean and Arabian Sea, we show that the relative direct and indirect contribution of picoplankton to export is proportional to their total net primary production, despite their small size. We suggest that all primary producers, not just the large cells, can contribute to export from the surface layer of the ocean at rates proportional to their production rates.

Carbon export from the oceanic surface layer is controlled by biological transformations that occur within the pelagic food web (1, 2). Autotrophic picoplankton (<2 μm in diameter) often dominate primary production in these regions but are believed to contribute relatively little to carbon export from surface layers because of their small sizes, slow sinking rates, and rapid utilization in the microbial loop. Large, rapidly sinking phytoplankton, such as diatoms, are believed to control carbon flux from upper ocean layers, and their contributions to export are believed to be disproportionately larger than their contributions to total primary production (1). Here we ask whether the contributions of picoplankton to carbon fluxes from the surface ocean are, in fact, disproportionately low.

Using results from inverse and network analyses of U.S. Joint Global Ocean Flux Study (JGOFS) data from the equatorial Pacific (EqPac) and Arabian Sea, we show that the relative contributions of various phytoplankton size classes to carbon export are proportional to their contributions to total net primary production (NPP) (3) (Fig. 1). Potential export pathways include direct routes, such as aggregation and incorporation into settling detritus, and indirect export through the consumption of picoplankton aggregates by organisms at higher trophic levels. The network analysis used in these studies (4) estimates the total particulate organic carbon (POC) flux that originated from each input source, here the NPP of each phytoplankton size class. In the EqPac, for example, picoplankton contributed 70% or more of the total NPP measured during JGOFS EqPac time series cruises and were responsible for 87% of POC export via detritus and 76% of carbon exported through the mesozooplankton (5). Mesozooplankton consumed both picoplankton-derived detritus and picoplankton-fed micrograzers. Similarly, at a northern station in the Arabian Sea during the Northeast Monsoon, picoplankton NPP was 86% of the total and was the source for 97 and 75% of the total carbon exported from the euphotic zone via the POC and mesozooplankton pathways, respectively (6). The relative contributions of the picoplankton to carbon export decreased where they produced relatively less of the total NPP but were nonetheless substantial (Fig. 1 and Table 1).

Fig. 1.

Proportional contributions of varying phytoplankton groups or size classes (picoplankton, diatoms, pelagophytes, and prymnesiophytes for the EqPac study and pico-, nano-, and microphytoplankton for the Arabian Sea) to NPP versus their proportional contributions to export as detritus (A) or through consumption of mesozooplankton (B). Proportional contributions were calculated as NPP or export due to the size class/total NPP or export flux.

Table 1.

Direct and indirect contributions of picoplankton to carbon export from the euphotic zone for two cruises in the EqPac and at three stations (N7, S2, and S11) during three seasons in the Arabian Sea. Abbreviations are as follows: TS1, time series 1 (March to April 1992); TS2 Early, time series 2 cruise (early October 1992); TS2 Mid, midway through TS2 (mid-October 1992); TS2 Late, late October 1992; NEM, Northeast Monsoon; SIM, Spring Intermonsoon; and SWM, Southwest Monsoon.

ModelExport pathwayCarbon export due to picoplankton (mmol of C m-2 day-1)% Contribution of picoplankton to total carbon export flux
TS1 POC 1.7 87
Mesozooplankton 10.4 76
TS2 Early POC 0.7 57
Mesozooplankton 11.5 60
TS2 Mid POC 1.5 43
Mesozooplankton 11.0 50
TS2 Late POC 1.7 63
Mesozooplankton 11.2 63
Arabian Sea
NEM N7 POC 1.0 97
Mesozooplankton 11.5 75
NEM S2 POC 3.8 56
Mesozooplankton 5.0 33
NEM S11 POC 4.3 99
Mesozooplankton 11.1 83
SIM N7 POC 3.7 73
Mesozooplankton 3.4 24
SIM S2 POC 5.9 92
Mesozooplankton 2.4 89
SIM S11 POC 2.8 98
Mesozooplankton 11.0 80
SWM N7 POC 2.3 91
Mesozooplankton 1.7 68
SWM S2 POC 13.8 53
Mesozooplankton 1.4 34
SWM S11 POC 10.5 44
Mesozooplankton 0.8 1

We believe that the contributions of picoplankton to carbon flux from oceanic surface layers have been overlooked because of assumptions that are made about cell size, sinking dynamics, and the trophic pathways of picoplankton through ecosystems. The seminal paper by Smayda (7) documented the tendency of larger cells to sink faster. However, there are at least three potential mechanisms for increasing the effective size of picoplankton and their removal rates from the euphotic zone. First, the aggregation of picoplankton cells into larger detrital particles (8, 9) enhances their vertical settling velocities and resulting export fluxes (1012). A simple model of aggregation and cell sinking for phytoplankton of varying diameters (3) (Fig. 2) shows that, at peak velocities, the average settling rate for aggregated cells from 1to30 μm in diameter is similar, although aggregation does not substantially increase the settling rates of the largest cells modeled (100 μm). We conclude that an individual alga does not need to be large to sediment out, even though settling rates are generally faster for single cells that are larger. Aggregation may be enhanced if cells are nutrient-depleted (13), and settling may be enhanced by the incorporation of mineral matter (1416).

Fig. 2.

Average sinking velocity versus time (in days) for phytoplankton cells with diameters of 1, 3, 10, 30, and 100 μm. A cell density was assigned that is consistent with the velocity versus cell diameter graph of Smayda (7) for 3-, 10-, 30-, and 100-μm cells. Because 1-μm cells are outside the range analyzed by Smayda, a density difference of 0.0725 g cm–3, characteristic of Synechococcus lividus, was used (3). Cells grow exponentially in a mixed layer of constant thickness, and there was no light or nutrient limitation. Algal cells collide to form aggregates at rates that depend on their abundances as well as sizes. Large cells settle faster than small ones. With time, cell concentrations increase, causing an increase in the fraction of material in aggregates. The resulting increase in average particle size leads to an increase in average settling speed. Peaks in total cell concentration occur when the enhanced losses due to settling balance the gains due to cell division. The maximum average settling rate for particles formed from the 1-μm cells is not substantially different from that for particles formed from the 30-μm cells.

The inclusion of small phytoplankton in marine snow aggregates is well documented (2, 1719). Aggregates of intact Emiliania huxleyi were first found in sediment traps deployed for short time periods (24 to 48 hours) in the North Sea (20). In the northeast Atlantic, water samples captured with bottles showed concentrations of Synechococcus-like cyanobacteria on aggregates that were 3000 to 12,000 times higher in concentration than that seen in freeliving forms (18). Picoplankton aggregates have also been observed in sediment traps equipped with acrylamide gels during short-term (48 hours) deployments in high-nutrient, low-chlorophyll waters off New Zealand (19). It is noteworthy that observations of aggregates of small phytoplankton in sediment traps have been in traps deployed for short time periods, which will minimize loss of material and maximize potential identification of constituent organisms as compared to longer-term deployments (21).

Although we did not explicitly include the process of aggregation in our EqPac and Arabian Sea analyses, we did allow ungrazed picoplankton production to flow directly to the detritus pool, where it could settle out as detritus or be consumed by larger zooplankton. Models of food webs and biogeochemical cycling (2225) generally assume that all picoplankton production is cycled through the microbial loop and that none sinks from the euphotic zone directly. This follows the classic analysis of pelagic community structure and nitrogen fluxes by Michaels and Silver (1), who assumed that “only large particles, with consequently high potential sinking rates, can exit the euphotic zone” and that “large colonies of smaller algae” decompose (and are presumably recycled) within the euphotic zone. Implicit in these food web structures is the assumption that growth and grazing in picoplankton-dominated open oceans are in balance over long (annual) time scales (26). However, our studies (5, 6) and those of others (27) have shown that microzooplankton grazing does not always balance picoplankton growth on shorter time scales (weeks to months). Food web models that force all picoplankton production through the microbial loop and do not allow direct picoplankton export may, therefore, be misleading. The conclusion of Michaels and Silver (1) that “picoplankton, the dominant producers in the model, contribute little to the sinking material” was based on a model (theirs) that was structured in such a way that no other conclusion could be reached.

A second pathway for the accelerated sinking of picoplankton-derived material is through mesozooplankton grazing. Picoplankton are usually considered to be too small to be captured effectively by larger grazers such as copepods (28), but the increase in effective size through aggregation makes them available and thus can enhance their export from surface layers through their incorporation into fast-sinking fecal pellets (9). When mesozooplankton consume picoplankton-containing aggregates, the picoplankton carbon short-circuits the microbial loop and results in higher than expected efficiency of carbon transfer from the euphotic zone (18, 29, 30). For animals such as salps that are capable of feeding on particles as small as 1 μm, aggregation is unnecessary to produce fecal pellets. The aggregates described in (19) were thought to originate in salp or other tunicate fecal material. The incorporation of picoplankton-sized particles into the rapidly sinking mucous nets and feces of filter feeders such as salps or pteropods provides an additional route for picoplankton removal (2, 31).

We propose, therefore, that the conventional view of carbon cycling, in which picoplankton carbon is recycled within the microbial loop and only larger phytoplankton carbon is exported, should be revised to include the alternative pathways for picoplankton carbon cycling discussed above. This alternative view relies on aggregation as a mechanism for both direct sinking (the export of picoplankton as POC) and mesozooplankton- or large filter feeder–mediated sinking of picoplankton-based production, but we believe that the evidence for both these processes under an array of environmental conditions is well established. Despite their small size, picoplankton may contribute more to oceanic carbon export than is currently recognized, and these contributions should be considered in current models of trophic dynamics in the ocean.

Supporting Online Material

Materials and Methods


References and Notes

View Abstract

Navigate This Article