The biological pump
In 1934, Redfield noticed that the growth, sinking and remineralization of phytoplankton generate a difference of carbon concentrations between surface and deep waters. This difference was named “soft-tissue carbon pump” by Volk and Hoffert (1985) better known nowadays as the biological carbon pump (BCP). The biological pump is a suite of biologically mediated processes (left part of In surface waters, phytoplankton assemblages take up nutrients and DIC during photosynthesis and convert it into particulate organic matter (POM) and biomineral compounds (calcium carbonate, also named calcite – CaCO3, for coccolithophores or biogenic silica, also named opal – BSi, for diatoms). Photosynthesis is a reductive chemical reaction transforming CO2 and H2O into organic molecules (e.g. sugars; Equation 1.3).
This carbon fixation, also defined as primary production (PP) in surface waters, produces particulate organic carbon (POC) which can then be exported to the deep ocean through the sinking of organic particles.
This organic matter can be consumed by zooplankton through grazing, which then excretes fecal pellets and DOC. POC can also be remineralized from the particulate to the dissolved phase, and oxidized from the organic to the inorganic form (DIC) through bacterial activity or zooplankton respiration. If this remineralization occurs in the surface ocean, the released CO2 can be then transferred back to the atmosphere.
As a result, only a small fraction of POC (< 0.1 Pg C.year-1) reaches the seabed, trapping the carbon for thousands to hundreds of thousand years (Sabine and Feely, 2007).
The carbonate pump
Another important part of the biological pump involves the formation of PIC via the precipitation of calcium carbonate by specific plankton species such as coccolithophores, foraminifera and pteropods (e.g. Emilio et al., 1993). These calcifying species directly use DIC to synthesize their carbonate shell, thereby releasing CO2 (Equation 1.4). This process is known as the carbonate counter pump (left part of Figure 1.2).
The precipitation of carbonates results in an increase of the surface ocean pCO2 (Frankignoulle et al., 1993) on timescales of 100-1000 years (Zeebe, 2012). However, an opposite effect of the carbonate counter pump is the increased organic carbon export flux to the deep ocean (0.1 Pg C.year-1; Sabine and Feely, 2007), ballasted by the calcium carbonate (Francois et al., 2002).
The microbial pump
The majority of the organic matter is remineralized by respiration, releasing CO2 back to the atmosphere. However, a large reservoir of dissolved organic carbon (DOC) exists, in which 95% is recalcitrant DOC (RDOC). Being resistant to biological decomposition, the RDOC can persist in the ocean for at least 100 years and represents thus a reservoir for carbon. The origin of this RDOC is still misunderstood but some authors have suggested that RDOC may be formed by the microbes ‘pumping’ the bioavailable DOC into this RDOC (Jiao et al., 2010; Legendre et al., 2015).
Physical transfers of carbon via the solubility pump are one order larger in magnitude than the carbon export via the biological pump (265 Pg C.yr-1 compared to 11 Pg C.yr-1) but in obduction regions, the carbon related to the solubility pump is released back to the atmosphere (276 Pg C.yr-1; Levy et al., 2013).
Sarmiento and Gruber (2006) have compared the impact of the oceanic pumps on the surface-to-deep gradient of DIC concentrations (ΔsDIC; Figure 1.5). The authors have estimated that the biological carbon pump (ΔCsoft) is responsible for 70% of the DIC increase in the deep-waters whereas the solubility pump (ΔCgas-ex) and the carbonate counter pump (ΔCcarb) account for 10 and 20% respectively of the observed gradient. This finding clearly highlights that the biological carbon pump is the most important process controlling the distribution of DIC in ocean and therefore the atmospheric CO2 concentrations.
Figure 1.5: Contribution of the different carbon pump components (the solubility pump ΔC gas-ex; the carbonate counter pump ΔC carb; and the biological carbon pump ΔC soft) to DIC increase in the deep ocean (Gruber and Sarmiento, 2006).
In the following section, the biological pump will be at the core of the different chapters.
The biological carbon pump
The first step of the BCP occurs in the surface ocean, where phytoplankton species, which are photoautotrophic organisms, use energy from sunlight to synthesize complex organic compounds such as carbohydrates, lipids and proteins. This is known as primary production. Phytoplankton can be eukaryotes (the cell contains a nucleus and other organelles enclosed within membranes) such as the opal-secreting diatoms, prymnesiophytes (including the CaCO3-secreting coccolithophores) and dinoflagellates; or prokaryotes (the cell lacks a distinct nucleus and membrane bound organelles) such as cyanobacteria (Sigman and Hain, 2012). In addition to carbon, phytoplankton require a wide range of nutrients, many of which become limiting in certain regions of the ocean, exerting a significant control on surface ocean production and ultimately the efficiency of the BCP (Moore et al., 2013). These elements are presented in section 220.127.116.11. Other controls of the primary production in surface waters include the dissolved CO2 concentrations, temperature, stratification of the water column, and grazing.
The second step of the BCP is the export of organic matter from the sunlit surface ocean to depth through different types of vertical transfer: 1) the passive sinking of particles and aggregates, 2) the physical advection and diffusion between the surface and the deep ocean and 3) the daily vertical migration of zooplankton, grazing phytoplankton, ingesting thus the associated-chemical elements in the surface ocean and then excreting fecal pellets after their descent to mesopelagic depths (100-1000 m). Sinking particles transport carbon but also all the other elements composing POM, previously incorporated as nutrients by phytoplankton. The export occurs when there is an excess density of the POM, i.e. when the density of the POM is greater than the density of seawater. This generally occurs at the death of organisms but most phytoplankton cells are too small to sink individually, so sinking occurs once they aggregate into larger particles or are repackaged into fecal pellets by zooplankton. The biogenic minerals (BSi and CaCO3), as well as the lithogenic minerals (dust, clay particles), can be incorporated into aggregates, thereby increasing the excess density of settling particles and inducing an increase of the settling velocity (De La Rocha and Passow, 2007; Honjo, 1996, see in section 18.104.22.168).
Once formed, settling particles also face the possibility of decomposition through consumption and respiration by herbivorous zooplankton and degradation by bacteria: it is the third component of the BCP. Most of these recycling processes take place within the upper 500m of the water column (Guidi et al., 2009). The remineralization rates of the different elements composing the particles can be influenced by a wide range of processes (e.g. Boyd et al., 2010; Twining et al., 2014). For example, carbon (C), nitrogen (N) and phosphorous (P) are remineralized faster and hence at shallower depths than the biomineral silicon (Si) (Boyd et al., 2010; Boyd and Trull, 2007; Twining et al., 2014). This can be explained by the presence of Si in opal frustule of diatoms that are protected of the degradation by an organic coating (Bidle and Azam, 1999) whereas P, N or C are associated to the organic matter. The different remineralization length scales are thus regulated by the lability of the material, which likely depends on the chemical composition of the sinking particles, with lithogenic, opal or calcareous materials more resistant to the degradation (Boyd et al., 2010). Moreover, sinking particles can be progressively enriched in trace elements such as iron (Fe) relative to C, N or P with depth (Boyd et al. 2010; Twinning et al. 2014) due to scavenging of any freshly released Fe back onto particles (Boyd et al. 2010). Such mechanisms induce different remineralization rates that can have a strong feedback on surface processes, in particular influencing the concentrations of micro- and macronutrients, locally or, through interactions with the large scale oceanic circulation.
The net result of these three components of the BCP is that a relatively small fraction (1%, Martin et al., 1987, Figures 1.3 and 1.6) of the primary production is exported to deep waters or even to sediments. As shown on Figure 1.6, the POC flux generally decreases exponentially with depth as more than 50% of the primary production in the euphotic zone is grazed or degraded by microbes and bacteria.
Therefore, it is important to determine POC fluxes from the surface to the deep sea in order to assess the efficiency of the BCP. This efficiency will depend on the amount of primary production in surface waters, itself depending on nutrient and light availability; the rate at which the particles sink, depending on their size and density; and the degradation rate of organic C and other elements composing the particles, following grazing and remineralisation.
– Macro- and micro-nutrients availability
To grow, phytoplankton requires chemical elements, which are typically identified as nutrients (Sigman and Hain, 2012). Nutrients can be divided in two categories: macro- and micro-nutrients, this distinction being made by their concentration in seawater (µmol.L-1 and nmol.L-1 to pmol.L-1, respectively). In the following, the cycle of the elements of interest are briefly described.
Redfield et al. (1963) discovered that relatively invariant elemental ratios of C, N, and P found in marine organisms are intimately entwined with the co-variation (106C:16N:1P) of these elements in the ocean.
Dissolved phosphorous is present in seawater under the phosphate form (PO43-; < 0.1 – 3.0 µmol.L-1). Important inputs of PO43- are delivered through riverine runoff, atmospheric deposition and submarine groundwater discharge (e.g. Paytan and McLaughlin, 2007). In polar areas, ice sheets could also bring large quantities of P in solution (Hawkings et al., 2016). These PO43- sources fuel phytoplankton blooms in surface waters but in some regions, like in the subtropical Sargasso Sea, P has been reported to limit the phytoplankton developments (Wu et al., 2000). Phosphorus is easily remineralized in the water column (Clark and Ingall, 1998), and its preferential remineralization compared to C is explained by the more labile nature of organic compounds containing P (Berner, 1980).
Inorganic nitrogen is mainly present in the marine system as nitrate (NO3-) but also under other forms such as ammonium (NH4+), nitrite (NO2-), nitrous oxide (N2O), nitric oxide (NO) and dinitrogen (N2) and is brought through the rivers, the atmospheric depositions or the N2 fixation (Capone et al., 2005; Gruber, 2008). Ammonium (< 0.3 – 1.2 µmol.L-1) is the preferred source of nitrogen for phytoplankton because its assimilation requires less energy (Zehr and Ward, 2002). Nitrate uptake requires more energy but since nitrate concentrations are more abundant in the ocean (< 1 – 35 µmol.L-1), most phytoplankton have the enzymes to initiate the reduction of NO3- in order to grow (nitrate reductase), with the exception of prokaryote species such as Synechococcus (Moore et al., 2002). Phytoplankton can also use NO2- as a nitrogen source but its concentration in the ocean is usually low (< 0.2 – 1.5 µmol.L-1). Another important source of N, when the other forms are exhausted or when the N/P ratio is not “Redfieldian”, (with P being in excess compared to N in Redfield stoichometry) is via the biological N2 fixation which transforms N2 into PON. Organisms able to fix dissolved N2, called diazotrophs, are found among free and/or symbiotic cyanobacteria some filamentous such as bloom forming Trichodesmium spp. (Capone et al., 1997; Carpenter et al., 1999), Richelia (Foster et al., 2007; Foster and Zehr, 2006) and others unicellullar diazotrophic cyanobacteria (Zehr et al., 2001) as well as heterotrophic bacteria (Zehr et al., 1998). Once phytoplankton has satisfied its nitrogen demand for growth, the different dissolved nitrogen forms are transformed in PON. Most of the PON is returned back to dissolved inorganic nitrogen by remineralization processes generated by bacteria: ammonification transforming PON to NH4+, nitrification transforming NH4+ to NO2- (ammonium oxidation) and then NO2- to NO3- (nitrite oxidation or true nitrification). In the ocean, the fluxes of PON to the seabed are small. The major sink for nitrogen is the denitrification transforming the NO3- into N2 in low oxygen environment (oxygen minimum zones, shelf or margin sediments), leading to a nitrogen loss through the atmosphere (Galloway et al., 2004; Gruber, 2004). In addition, nitrogen is also lost as N2O produced during the nitrification as well as during denitrification.
Silicic acid (H4SiO4 or dSi) is also considered as a macronutrient and is particularly essential for some plankton taxa known as silicifiers such as diatoms (autotrophic phytoplankton), silicoflagellates and radiolarians (heterotrophic zooplankton), as they use it to build their shells which is then referred as biogenic silica (BSi or bSiO2). Silicic acid is brought to the ocean essentially via the rivers (about 66% of the total net inputs, Tréguer and De La Rocha, 2013) and its concentrations vary from < 0.2 to 170 µmol.L-1. In the world ocean, 56% of BSi is recycled in the upper 100m and the remaining fraction is transported to the deep ocean where BSi can also undergo dissolution. Only 3% of BSi is estimated to reach the seafloor (Tréguer and De La Rocha, 2013).
Other elements, such as the following metals: Fe, Zn, Mn, Ni, Cu, Co, Cd, Mo that are present at subnanomolar concentrations in seawater exert a key role in many metabolic processes (Table 1.1) and are referred as key micronutrients. As a result, the oceanic trace element biogeochemical cycles have been intensely studied over the past decade (e.g. see review in SCOR working group, 2007). Their sources are very diverse being atmospheric (e.g. Jickells et al., 2005; Shelley et al., 2015), riverine (e.g. de Baar and Jong, 2001), sedimentary (e.g. Bruland and Lohan, 2008), hydrothermal (e.g. Tagliabue et al., 2010), cold seep vents (e.g. Lemaitre et al., 2014), ice melting (e.g. Lannuzel et al., 2011), anthropogenic (e.g. Gao et al., 2014), groundwater discharge (e.g. Trezzi et al., 2016), extraterrestrial (e.g. Johnson, 2001) or volcanic activity (e.g. Achterberg et al., 2013), shelves (Elrod et al., 2004). However, only a small fraction of trace metals originating from these sources is soluble in seawater and accessible to the phytoplankton. In the surface, a significant fraction of trace metals is bound to strong organic complexes, that increase their solubility and can favor their bioavailability (e.g. Gledhill and Buck, 2012; Rue and Bruland, 1995). Some trace metals are easily scavenged onto particles and are subsequently removed from the surface waters when particles sink. In the mesopelagic layer, trace metals are also affected by bacterial activity, at different remineralization rates. For example, Twining et al. (2014) have shown that Ni or Zn were remineralized faster than Fe.
It is typically the case for the Southern Ocean which is isolated from any major iron sources, inducing a limited primary production (Martin, 1990). However, in this HNLC Southern Ocean, diatoms (the major phytoplankton group in this region) can develop important adaptive strategies to these particular conditions (Boyd et al., 2007; de Baar, 2005), such as a heavily silicified and thick shell, a slower growth rate and a high iron storage capacity (Boyd, 2013).
A nutrient limitation, as seen above, drives the succession of the phytoplankton communities. In the North Atlantic, for example, a strong diatom bloom usually thrives during spring, consuming the available nutrients. When silicic acid levels are depleted, there is then a transition from diatoms to coccolithophores (Henson et al., 2006; Sanders et al., 2014). Moreover, at the end of the productive period, a Fe limitation or co-limitation with silicic acid has also been reported (Nielsdottir et al., 2009) inducing seasonal HNLC conditions in this area.
These different limitations have broad implications for the oceanic ‘biological pump’ that links nutrient and carbon cycling.
– Suspended “ballast” mineral
Ballast minerals, including lithogenic minerals (e.g. dust, clays) and biogenic minerals (BSi and CaCO3), are important for the settling of POM to depth. Indeed, the density of the organic matter (≈ 1.05 g.cm-3) is almost the same as seawater (≈ 1.03 g.cm-3) and thus particles require a “ballast” effect to sink. Biogenic mineral ballast are mainly calcium carbonate (CaCO3) and biogenic silica (bSiO2) that form the shells of coccolithophores, pteropods, foraminifers for the former, and of diatoms and radiolarians respectively (Lam and Bishop, 2007). Calcite density is 2.71 g.cm-3, opal density is 2.1 g.cm-3 and lithogenic material density can reach 2.65 g.cm-3 (e.g. quartz; Klaas and Archer, 2002). Some authors have proposed a relationship between the POC flux and the flux of ballast minerals in deep sediment traps (Armstrong et al., 2002; Francois et al., 2002; Klaas and Archer, 2002; Figure 1.8) as well as in the upper water column (Sanders et al., 2010; Thomalla et al., 2008).
Figure 1.8: Correlations between POC and mineral fluxes collected in 52 sediment traps below 1000 m and around the world ocean (Klaas and Archer, 2002).
The biogenic and lithogenic minerals have the potential to control the fraction of primary production that reaches the seabed by protecting the organic matter from oxidation, by initiating a more rapid aggregation or by increasing density and in fine, sinking velocity (Armstrong et al., 2002; Le Moigne et al., 2013b). The transfer efficiency of POC to the deep ocean has been strongly related to the calcite flux (François et al., 2002) because calcite is denser, more abundant than opal and clays (Klass and Archer, 2002) and is susceptible to dissolve less than opal in the upper water column (Nelson and Brzezinski, 1997). A recent study has also highlighted the importance of calcium carbonate to export POC in the Southern Ocean (Salter et al., 2014), where the carbon export has usually been attributed to large and heavily silicified diatoms. Nonetheless, diatoms represent an important vector of carbon export as they carry out half of the primary production in the world ocean (Buesseler, 1998; De La Rocha and Passow, 2007; Tréguer et al., 1995).
Finally, recent studies have demonstrated that the effect of ballast minerals on the POC flux varies regionally (Frédéric A C Le Moigne et al., 2014; Le Moigne et al., 2012). In the high-latitude North Atlantic, mineral ballasting seems to be important with 60% of the POC flux associated with minerals (on average, mostly CaCO3 and BSi) whereas this fraction is only about 40% in Southern Ocean (on average, mostly BSi). The remainder of the export flux is unballasted, suggesting that the association between POC and minerals is not always necessary to promote the export of organic matter. In this case, low export efficiencies due to the easier degradation of the unballasted POC have been reported (e.g. Le Moigne et al., 2012). However, some discrepancies have been observed, indicating that other parameters such as low microbial activity or large zooplankton migration can impact the magnitude of surface export (Le Moigne et al., 2016).
Spatial variations in phytoplankton size structure are known to exert a control on the magnitude of the POC export flux (Boyd and Newton, 1999) and high POC exports are usually related to a greater size of the sinking phytoplankton cells (Alldredge and Silver, 1988; Guidi et al., 2009). In addition to the size structure, the composition of the phytoplankton community has been shown to influence the magnitude of the biogeochemical export fluxes. As seen in the previous section, diatoms and calcifying species are important vectors of carbon export, as their ‘hard parts’ (BSi and CaCO3, respectively) can be incorporated into aggregates and increase the excess density of suspended particles provoking an increase of the sinking velocity (Honjo, 1996). The life stage of the certain phytoplankton communities can also influence the magnitude of the export fluxes. For example, the formation of diatom resting spores has been shown to strongly enhance the deep carbon export (Rembauville et al., 2016b, 2015a; Rynearson et al., 2013; Salter et al., 2012). The succession of the different phytoplankton communities as well as the succession of the different life stages can be triggered by a nutrient limitation (see section 22.214.171.124; Sugie and Kuma, 2008) but also by turbulence (Margalef, 1978) or light changes (Lasbleiz et al., 2016).
The magnitude and the fate of the primary production exported from surface waters to deeper depths are also related to remineralization processes that are led by zooplankton and bacteria activities in the euphotic zone but also in the mesopelagic zone to support their metabolic demands (Figure 1.9). Modest variations of the remineralization depth have been shown to impact strongly the air-sea carbon balance. For example, a remineralization depth increasing by only 24 m would induce a decrease in atmospheric CO2 concentrations from 10 to 30 ppm, i.e., ~3 to 9% of the present atmospheric pCO2 (Kwon et al., 2009).
Table of contents :
Chapter 1: General Introduction
1.1. The Carbon cycle
1.2. The biological carbon pump
1.3. The different approaches to study the biological carbon pump
1.4. Study areas, objectives and thesis outline
Chapter 2: Material and Methods
2.1. Sample collection, chemical processing and analyses
2.2. 234Th Export determination
2.3. Elemental export fluxes
2.4. Carbon remineralization fluxes determined from Baxs distributions
Chapter 3: High variability of particulate organic carbon export fluxes in the North Atlantic Ocean
Chapter 4: Particulate barium tracing significant mesopelagic carbon remineralization in the North Atlantic Ocean
Chapter 5: Biogenic trace element export fluxes in the North Atlantic Ocean
Chapter 6: Impact of the natural Fe-fertilization on the magnitude, stoichiometry and efficiency of particulate biogenic silica, nitrogen and iron export fluxes
Chapter 7: General conclusions and future directions
7.1. Major findings
7.2. Comparison of the two study areas
7.4. The fate of the BCP with global warming