Dissolved inorganic carbon dynamics and air-sea carbon dioxide fluxes during coccolithophore blooms in the northwest European continental margin (northern Bay of Biscay)

Authors


Abstract

[1] We report a data set of dissolved inorganic carbon (DIC) obtained during three cruises in the northern Bay of Biscay carried out in June 2006, May 2007, and May 2008. During these cruises, blooms of the coccolithophore Emiliania huxleyi occurred, as indicated by patches of high reflectance on remote sensing images, phytoplankton pigment signatures, and microscopic examinations. Total alkalinity showed a nonconservative behavior as a function of salinity due to the cumulative effect of net community calcification (NCC) on seawater carbonate chemistry during bloom development. The cumulative effect of NCC and net community production (NCP) on DIC and the partial pressure of CO2 (pCO2) were evaluated. The decrease of DIC (and increase of pCO2) due to NCC was overwhelmingly lower than the decrease of DIC (and decrease of pCO2) due to NCP (NCC:NCP ≪ 1). During the cruises, the northern Bay of Biscay acted as a sink of atmospheric CO2 (on average ∼−9.7 mmol C m−2 d−1 for the three cruises). The overall effect of NCC in decreasing the CO2 sink during the cruises was low (on average ∼12% of total air-sea CO2 flux). If this is a general feature in naturally occurring phytoplankton blooms in the North Atlantic Ocean (where blooms of coccolithophores are the most intense and recurrent), and in the global ocean, then the potential feedback on increasing atmospheric CO2 of the projected decrease of pelagic calcification due to thermodynamic CO2 “production” from calcification is probably minor compared to potential feedbacks related to changes of NCP.

1. Introduction

[2] Balch et al. [2007] evaluated from remote sensing data the contemporary global calcification related to coccolithophores to 1.6 ± 0.3 Pg PIC yr−1 (1 Pg = 1015 g; PIC, particulate inorganic carbon). Other estimates of contemporary global pelagic calcification range between 0.7 Pg PIC yr−1, based on accumulation rates and sediment trap data [Milliman et al., 1999], and 1.4 Pg PIC yr−1, based on the seasonal cycle of total alkalinity (TA) in the euphotic zone [Lee, 2001]. Each of these estimates of global pelagic calcification suffers from specific shortcomings. For instance, the estimate of Milliman et al. [1999] might be underestimated due to supralysoclinal dissolution of calcium carbonate (CaCO3) [e.g., Wollast and Chou, 1998; Beaufort et al., 2007; Berelson et al., 2007], and the TA analysis of Lee [2001] could be affected by other processes than calcification such as nutrient uptake and release or mixing of water masses. The estimate of Balch et al. [2007] is probably less biased by the sparseness of data coverage compared to the two other approaches, although strongly dependent on the predictive capability of the algorithm used to process the remote sensing data. Three-dimensional circulation models mostly tuned to reproduce the observed TA distributions also provide a wide range of global pelagic calcification values from 0.6 to 1.8 Pg PIC yr−1 [Bacastow and Maier-Reimer, 1990; Yamanaka and Tajika, 1996; Archer et al., 1998; Murnane et al., 1999; Heinze et al., 2003; Moore et al., 2004; Jin et al., 2006; Gehlen et al., 2007; Hofmann and Schellnhuber, 2009]. The fact that the estimate of Balch et al. [2007] of contemporary global pelagic calcification related to coccolithophores is comparable to the other estimates would imply that coccolithophores are the most important pelagic calcifiers in the oceans.

[3] The development of coccolithophore blooms affects the seawater carbonate chemistry, and air-sea CO2 fluxes, through the organic carbon pump and the carbonate counterpump. The organic carbon pump relies on organic carbon production by photosynthesis and leads to an uptake of CO2 from surface waters, according to:

equation image

[4] The carbonate counterpump relies on the production of CaCO3, leading to a thermodynamic shift of HCO3 to CO2, hence, a release of CO2 to surrounding surface waters, according to:

equation image

[5] The ratio between calcification (carbonate counterpump), and organic carbon production (organic carbon pump), the C:P ratio, depends on the life cycle (bloom development) and growth conditions of coccolithophores [Fernández et al., 1993; Paasche and Brubak, 1994; Paasche, 2002; Delille et al., 2005]. At the onset of the coccolithophore bloom, when nutrients are available for growth, organic carbon production dominates over calcification (C:P ≪ 1, the so-called organic phase). At the end of the bloom, in nutrient-depleted conditions and high irradiances (due to stronger stratification), organic carbon production decreases and calcification increases (C:P ≤ 1, the so-called inorganic phase).

[6] The accumulation of anthropogenic CO2 in the oceans [e.g., Sabine et al., 2004] has altered carbonate chemistry in surface waters (ocean acidification) since preindustrial times, and is expected to continue to do so in the coming centuries [e.g., Caldeira and Wickett, 2003; Orr et al., 2005; Cao et al., 2007; McNeil and Matear, 2007]. Changes of the carbonate chemistry of surface waters related to ocean acidification can alter the rates and fates of primary production and calcification of numerous marine organisms and communities [as reviewed by Raven et al., 2005; Kleypas et al., 2006; Fabry et al., 2008; Doney et al., 2009]. Such changes can provide either positive or negative feedbacks on increasing atmospheric CO2 by modifying the flux of CO2 between the ocean and the atmosphere.

[7] Several manipulative experiments to test the effect of ocean acidification on coccolithophores have shown that while calcification would decrease, the export of organic carbon would increase mainly through increasing production of transparent exopolymer particles (TEP) [Riebesell et al., 2000; Engel et al., 2004a, 2004b; Delille et al., 2005; Riebesell et al., 2007]. On the other hand, the reduction of pelagic calcification due to ocean acidification could also lead to a reduction of carbon export due to the decrease of the ballast effect of CaCO3 on marine particles [e.g., Armstrong et al., 2002; Klaas and Archer, 2002; Barker et al., 2003; Hofmann and Schellnhuber, 2009]. The modelling study of Hofmann and Schellnhuber [2009] suggested that the positive feedback on increasing atmospheric CO2 related to the decrease of carbon export from the reduction of ballast effect of CaCO3 on marine particles would represent ∼40% of the negative feedback related to the decrease of the CO2 emission to the atmosphere due to the reduction of pelagic calcification. For a credible implementation in mathematical models of such feedback mechanisms to allow the projection of a future evolution of marine carbon biogeochemistry under global change, it is required to understand present day biogeochemistry and ecology of naturally occurring pelagic calcifying communities. In particular, the overall effect of phytoplankton communities on the C:P ratio, carbonate chemistry, and air-sea CO2 fluxes.

[8] In the northwest European continental margin, blooms of the coccolithophore Emiliania huxleyi have been frequently reported [Holligan et al., 1983; Garcia-Soto et al., 1995; Wollast and Chou, 1998, 2001; Godoi et al., 2009; Harlay et al., 2009, 2010]. Here, we report a data set of carbonate chemistry in surface waters obtained during three cruises in the northern Bay of Biscay (Figure 1). We evaluate the relative effect of calcification and organic carbon production on seawater carbonate chemistry and air-sea CO2 fluxes.

Figure 1.

Map of the study site showing the sampling stations in June 2006 (grey circles), May 2007 (open circles), and May 2008 (black circles), the 200 m, 1000 m, 2000 m, and 4000 m isobaths (solid lines), and the general residual circulation (dotted arrows) based on Pingree and Le Cann [1989], Pingree [1993], Pingree et al. [1999], and Huthnance et al. [2001].

2. Material and Methods

2.1. Cruises and Sampling

[9] Three cruises were carried out in the northern Bay of Biscay from 31 May to 9 June 2006 (BG06/11 cruise), 10 May to 24 May 2007 (BG07/12 cruise), and 7 May to 23 May 2008 (BG08/12 cruise). Sampling of pH, TA, oxygen (O2), and phosphate (PO43−) was carried out with a rosette of 12 Niskin bottles (12 L) coupled to a conductivity-temperature-depth probe (Seabird SBE21). Depths of sampling covered surface waters, thermocline, and bottom waters down to the seafloor over the continental shelf and down to maximum 1400 m over the continental slope. Because of shorter ship-time, sampling during the June 2006 cruise was limited to the area around the La Chapelle Bank, while during the other two cruises sampling was also carried out further north along the continental shelf break (Figure 1).

2.2. Analytical Methods

[10] Underway measurements of the partial pressure of CO2 (pCO2) were carried out in surface waters (2 m depth) using an equilibrator [Frankignoulle et al., 2001], and a nondispersive infrared CO2 analyzer (Li-Cor 6262) calibrated with pure nitrogen (Air Liquide Belgium) and two gas mixtures with a CO2 molar fraction of 366 and 810 ppm (Air Liquide Belgium), that were calibrated against standards with a CO2 molar fraction of 361 and 774 ppm acquired from National Oceanic and Atmospheric Administration (NOAA, Global Monitoring Division, Carbon Cycle Greenhouse Gases Group). Temperature at the outlet of the equilibrator was measured with a Metrohm Pt-100 temperature sensor. The correction of pCO2 for the difference between equilibrator and in situ temperature was carried out with the algorithm given by Takahashi et al. [1993]. Sea surface temperature (SST) and salinity were measured underway (2 m depth) with a Seabird thermosalinograph (SBE 21).

[11] Measurements of pH were carried out with a combined electrode (Metrohm 6.0232.100), calibrated on the total hydrogen ion concentration scale, using TRIS (2-amino-2-hydroxymethyl-1,3-propanediol) and AMP (2-aminopyridine) buffers prepared at a salinity of 35 according to Dickson [1993]. Measurements of TA were carried out by open-cell titration with HCl 0.1 M according to Gran [1952] on 100 mL filtered (GF/F Whatman) seawater samples, and data were quality checked with certified reference material acquired from Andrew Dickson (Scripps Institution of Oceanography, University of California, San Diego). Dissolved inorganic carbon (DIC) and the saturation state of calcite (ΩCAL) were computed from pH and TA using the carbonic acid dissociation constants of Mehrbach et al. [1973] refitted by Dickson and Millero [1987], and the calcite solubility of Mucci [1983].

[12] Concentrations of dissolved O2 were measured by Winkler titration with a potentiometric end point determination. Reagents and standardizations were similar to those described by Knap et al. [1996]. The O2 saturation level (%O2) was calculated from the measured O2 concentration and the O2 concentration at saturation computed with the algorithm given by Benson and Krause [1984]. PO43− was measured colorimetrically with the molybdate and ascorbic acid method described by Grasshoff et al. [1983].

2.3. Evaluation of the Effect of Biological Activity on Carbonate Chemistry of Surface Waters and Air-Sea CO2 Fluxes

[13] The effect on DIC, pCO2, and ΩCAL in surface waters of net community production (NCP) and net community calcification (NCC) was estimated on the basis of the changes between the surface layer and the deep layer of nutrients and of TA, respectively. Changes in nutrients and TA were computed as the difference of the average concentration in the surface layer (top 20 m, 〈X〉0–20m) and the average concentration of the deep layer (80 m to seafloor, 〈X〉80m–bottom). The lower limit of the surface layer (20 m) was chosen to include data within the euphotic layer (euphotic depth typically of ∼30 m during the cruises, not shown), while the upper limit of the deep layer (80 m) was chosen to avoid the base of the thermocline (at some stations down to 70 m, not shown). For the deeper stations on the continental slope, the averaging of the deep layer data was made down to 150 m.

[14] TA and DIC were normalized to a constant salinity of 35.5 (TA35.5 and DIC35.5, respectively) according to:

equation image

where TA and DIC are the observed values at the observed salinity (S).

[15] Hereafter, pCO2@SST refers to pCO2 at SST and pCO2@13°C refers to the pCO2 normalized to a temperature of 13°C applying the algorithm of Takahashi et al. [1993]. The change of pCO2 due to temperature (T) changes (ΔpCO2SST) was computed with the algorithm of Takahashi et al. [1993] and the change of TT):

equation image

[16] For the anomaly of TA (TAanomaly) to only represent CaCO3 production and dissolution and to exclude the effect of organic matter production and degradation, TA values have to be corrected for PO43− and nitrate (NO3) assimilation and release [Brewer and Goldman, 1976]. PO43− assimilation is given by the term 〈PO43−80m–bottom − 〈PO43−0–20m. In absence of NO3 data during the June 2006 cruise, NO3 assimilation was computed for the three cruises from PO43− assimilation, using the Redfield ratio (16:1). During the May 2007 and May 2008 cruises, the measured NO3:PO43− ratios were 16.7 and 15.7, respectively (not shown), hence close to the Redfield ratio. The TAanomaly was computed according to:

equation image

[17] The change of DIC in surface waters due to net organic carbon production (ΔDICorg) was computed from the change of PO43− converted to carbon using the Redfield ratio (106:1), according to:

equation image

[18] The change of DIC in surface waters due to net CaCO3 production (ΔDICinorg) was computed using the TA anomaly technique [e.g., Smith and Key, 1975] according to:

equation image

[19] To estimate the change of pCO2 due to ΔDICorg and ΔDICinorg (ΔpCO2org and ΔpCO2inorg, respectively), we first computed DIC at atmospheric CO2 equilibrium (DICeq) from an average TA of 2340 μmol kg−1, a pCO2 of 380 ppm, an average temperature of 13°C, and an average salinity of 35.5.

[20] The value ΔpCO2org was computed as the difference between atmospheric equilibrium (380 ppm) and pCO2 computed from DICorg, a TA of 2340 μmol kg−1, a temperature of 13°C and a salinity of 35.5, where DICorg is given by:

equation image

ΔpCO2inorg was computed as the difference between atmospheric equilibrium (380 ppm) and pCO2 computed from DICinorg, TAinorg, a temperature of 13°C, and a salinity of 35.5, where DICinorg and TAinorg are given by:

equation image

[21] The change of ΩCAL due to net organic carbon production (ΔΩCALorg) was calculated as the difference between ΩCALeq and ΩCALorg, where ΩCALeq was computed from DICeq, a TA of 2340 μmol kg−1, a temperature of 13°C, and a salinity of 35.5, and ΩCALorg was computed from DICorg, a TA of 2340 μmol kg−1, a temperature of 13°C and a salinity of 35.5. The change of ΩCAL due to net CaCO3 production (ΔΩCALinorg) was calculated as the difference between ΩCALeq and ΩCALinorg,, where ΩCALinorg was computed from DICinorg, TAinorg, a temperature of 13°C, and a salinity of 35.5. The change of ΩCAL due to T change (ΔΩCALSST) was calculated as the difference between ΩCALeq and ΩCALSST, where ΩCALSST was computed from DICeq, a TA of 2340 μmol kg−1, a temperature = 13+ΔT, and a salinity of 35.5. The net effects of NCP and NCC on DIC, pCO2, and ΩCAL were calculated according to:

equation image

The observed changes of DIC, pCO2 and ΩCAL were calculated according to:

equation image

[22] Air-sea CO2 fluxes (F) were computed according to:

equation image

where ΔpCO2air-sea is the air-sea gradient of pCO2, α is the solubility coefficient of CO2 computed using the algorithm given by Weiss [1974], and k is the gas transfer velocity computed from wind speed using the k-wind parameterization given by Ho et al. [2006].

[23] Wind speed data were obtained from the National Centers for Environmental Prediction reanalysis daily averages surface flux (http://www.cdc.noaa.gov/) at five grid points covering the sampled region (50.475°N, 7.080°W; 50.475°N, 7.120°W; 48.571°N, 7.150°W; 48.571°N, 7.650°W; 48.571°N, 8.030°W). F was computed using daily wind speed values (average of the five grid points) for a time interval of 30 days centered on the date of the middle of the cruises. ΔpCO2air-sea was computed from measured pCO2 in seawater and the monthly atmospheric pCO2 at Mace Head (53.33°N, 9.00°W, Southern Ireland) obtained from the NOAA Climate Monitoring and Diagnostics Laboratory air sampling network (http://www.cmdl.noaa.gov/). Atmospheric pCO2 was converted to wet air using the water vapor algorithm given by Weiss and Price [1980].

3. Results and Discussion

3.1. General Setting of the Cruises

[24] Time series of remotely sensed surface chlorophyll-a (Chl-a) in the study area indicate that seasonal cycles of phytoplankton biomass were remarkably similar during the 3 years at the La Chapelle Bank and Goban Spur regions (Figure 2). The main spring bloom associated to diatoms peaked in mid-April, followed by a strong increase in normalized water leaving radiance at 550 nm (Lwn[555]) indicative of the occurrence of coccolithophore blooms from early May to late June. Overall, higher Chl-a as well as Lwn(555) values were observed at Goban Spur than at La Chapelle Bank during the 3 years. Some interannual variability in remotely sensed Chl-a was observed with highest concentrations in 2008 (up to 5 μg L−1 at Goban Spur). The three cruises were carried out after the main spring bloom, during the period of peak to declining Lwn(555).

Figure 2.

Time-series in 2006 (grey circles), 2007 (open circles), and 2008 (black circles) of weekly averages of Chl-a and Lwn(555) at La Chapelle Bank [47.0°N; 49.0°N; 9.0°W; 6.0°W] and Goban Spur [49.0°N; 51.5°N; 11.0°W; 9.0°W] (Sea-viewing Wide Field-of-view Sensor (SeaWiFS), retrieved from http://reason.gsfc.nasa.gov/Giovanni/). Start and end of cruises are indicated by vertical lines (solid grey for June 2006; dotted black for May 2007; solid black for May 2008). Normalized water leaving radiance at 550 nm data from January to March were removed, since cloud coverage leads to biased values during wintertime (S. Groom, personal communication, 2009).

[25] During the three cruises, remote sensing images revealed several patches of cold water (SST < 14°C) along the shelf break in the whole study area (Figure 3) corresponding to the signature of enhanced vertical mixing due to turbulent dissipation related to the generation of internal tides [Pingree and New, 1995; Wollast and Chou, 2001]. The highest Chl-a concentrations (>0.8 μg L−1) at the continental margin were observed inshore of the 200 m isobath. The upwelled nutrient rich cold water at the shelf break, characterized by lower Chl-a values (∼0.4 μg L−1), warmed and stratified as it propagated from the shelf break both off- and onshore, leading to enhanced biological activity [Wollast and Chou, 2001; Harlay et al., 2010]. As these water masses propagated further on the shelf, phytoplankton development caused nutrient depletion, hence, the highest Chl-a waters were confined close to the shelf break. The high Chl-a waters were also associated to high reflectance patches indicating the presence of coccolithophores at the end of the inorganic phase, since coccoliths that are shed from coccolithophores increase reflectance. High performance liquid chromatography (HPLC) pigment measurements (N. Van Oostende and K. Sabbe, personal communication, not shown) indicate that prymnesiophytes (i.e., coccolithophores) accounted, depending on the station, for 10%–71%, 0%–59%, and 2%–49% of the total Chl-a during the June 2006, May 2007, and May 2008 cruises, respectively. The other most abundant phytoplankton community in terms of total Chl-a was usually composed of diatoms, and at some rare occasions of dinoflagellates or chlorophytes. Scanning electron microscopy (SEM) of coccolithophores in samples obtained during the June 2007 and June 2008 cruises identified Emiliania huxleyi as the dominant species (N. Van Oostende and K. Sabbe, personal communication; A. Engel, personal communication).

Figure 3.

Remote sensing images of SST (Advanced Very High Resolution Radiometer), Chl-a (SeaWiFS), and reflectance (unitless, false-color [443, 490, and 555 nm bands], SeaWiFS) contemporary to the May 2006, June 2007, and 2008 cruises in the Bay of Biscay (6 May 2006; composite 20–22 May 2007; composite 18 May 2008) (courtesy of S. Groom, Remote Sensing Group of the Plymouth Marine Laboratory), and bathymetry (200 m and 2000 m isobaths).

3.2. Distribution of pCO2 and TA in Surface Waters

[26] The distribution of pCO2 during the three cruises was patchy and strong horizontal gradients of pCO2 were observed (Figure 4). At the continental margin, the range of pCO2 variations in surface waters was 248–270 ppm, 288–342 ppm, and 250–269 ppm, during the June 2006, May 2007, and May 2008 cruises, respectively. The pCO2 values were systematically below atmospheric equilibrium (378 ppm, 382 ppm, and 385 ppm in June 2006, May 2007, and May 2008, respectively), hence, the area acted as a sink for atmospheric CO2. Average air-sea CO2 fluxes were −9.8, −11.9, and −7.4 mmol C m−2 d−1 in June 2006, May 2007, and May 2008, respectively (Table 1), in agreement with air-sea CO2 fluxes reported previously in the area at this time of the year [Frankignoulle and Borges, 2001; Borges et al., 2006; Padin et al., 2009; de la Paz et al., 2010; Harlay et al., 2010].

Figure 4.

Distribution of pCO2@SST in surface waters in the northern Bay of Biscay in June 2006, May 2007, and May 2008, and bathymetry (250 m, 1000 m, 2000 m, and 4000 m isobaths).

Table 1. Date, Position, ΔpCO2air-sea, F, and Change of F Due to NCC in the Northern Bay of Biscay in June 2006, May 2007, and May 2008
DateLongitude (°W)Latitude (°N)ΔpCO2air-sea (ppm)F (mmol C m−2 d−1)Change of F Due to NCC (%)
31 May 20067.00047.749−109−13.15
1 June 20067.16647.533−100−12.14
2 June 20067.90247.901−59−7.13
2 June 20067.50248.100−85−10.216
6 June 20068.90148.500−55−6.623
7 June 20068.09948.400−67−8.114
8 June 20067.50048.100−73−8.721
9 June 20067.00147.750−107−12.95
Average  −82−9.811
10 May 20076.89847.800−34−5.572
12 May 20077.62248.202−86−13.912
13 May 20078.49948.498−73−11.88
14 May 20079.49549.204−71−11.423
15 May 200710.50949.500−85−13.86
16 May 200710.50051.344−94−15.37
21 May 20078.49248.505−84−13.64
22 May 20077.57748.228−88−14.211
23 May 20077.26747.417−60−9.622
23 May 20078.19847.681−93−15.07
24 May 20077.24147.774−39−6.436
Average  −73−11.919
7 May 20086.00148.499−72−6.40
7 May 20087.16547.533−43−3.821
8 May 20086.89847.801−92−8.11
9 May 20087.90947.898−36−3.22
10 May 20087.59448.204−89−7.93
11 May 20088.49948.501−93−8.26
12 May 20089.50249.199−116−10.31
13 May 200810.50549.498−103−9.18
14 May 200810.50250.502−101−8.95
19 May 20089.99751.003−92−8.19
20 May 200810.34150.002−96−8.514
21 May 20089.49649.202−88−7.822
22 May 20087.60048.200−76−6.73
23 May 20087.26747.418−69−6.09
Average  −83−7.48

[27] During the three cruises, TA in surface waters showed a strong nonconservative behavior with respect to salinity compared to the climatological TA-S relationship for the North Atlantic Ocean surface waters reported by Millero et al. [1998] (Figure 5). Some of the TA data points (in particular during the June 2008 cruise) were above the climatological TA-S relationship of Millero et al. [1998]. While these TA data points remained within the standard deviation of the linear fit of the climatological TA-S relationship of Millero et al. [1998] (±9 μmol kg−1), this could be due to local influence in the northern Bay of Biscay of river inputs [Kelly-Gerreyn et al., 2006] characterized by high TA values, such as the Loire river [TA ∼2800 μmol kg−1 in the freshwater end member, Abril et al., 2003] and the Gironde river [TA ∼2400 μmol kg−1 in the freshwater end member, Abril et al., 1999]. Yet, the nonconservative behavior of TA as a function of salinity indicated the drawdown of TA by calcification. This is consistent with the presence of coccolithophores in the area during the three cruises as indicated by the high reflectance patches in remote sensing images (Figure 3), HPLC and SEM measurements.

Figure 5.

Variation of TA versus salinity (top 20 m) in the northern Bay of Biscay in June 2006 (grey circles), May 2007 (open circles), and May 2008 (black circles). The dotted line corresponds to the climatological TA-S relationship given by Millero et al. [1998] for the surface waters of the North Atlantic Ocean.

3.3. Distribution of Variables as Function of the Degree of Stratification

[28] Stratification is one of the most important variables controlling the intensity of primary production, and the succession of phytoplankton communities [e.g., Margalef, 1997]. Increasing stratification enhances light availability but leads to the decrease of vertical input of inorganic nutrients, imposing a succession of phytoplankton communities with variable light and inorganic nutrient requirements. Coccolithophores (and especially Emiliania huxleyi) have a high tolerance to elevated irradiances [Nanninga and Tyrrell, 1996] as well as a high affinity for inorganic [Holligan et al., 1993] and organic nutrients [Riegman et al., 2000]. Emiliania huxleyi has the ability to use organic sources for nitrogen and phosphorous when inorganic nutrients are limiting for the development of other phytoplanktonic species [Palenik and Henson, 1997; Riegman et al., 2000]. Hence, Emiliania huxleyi usually blooms in stratified and inorganic nutrient-depleted conditions typically after the blooms of diatoms [e.g., Margalef, 1997].

[29] The degree of stratification was computed as the difference between seawater density at 100 m depth and the seawater density at 10 m depth. These depths were chosen to make sure that the upper value was within the mixed layer (where density was homogeneous), and that the bottom value was below the base of the thermocline (down to 70 m at some stations, not shown). Variables from different stations were plotted as a function of the degree of stratification as a way of reconstructing the effect of the bloom development on seawater carbonate chemistry and other biogeochemical variables. Such an approach is useful and appropriate in the study area, where there is a more or less continuous generation of low stratified and nutrient rich waters at the continental shelf break that propagate on shelf, stratify, and host high phytoplankton biomass at the period of the year when the cruises were carried out (section 3.1). It should be noted that the changes of the biogeochemical variables as a function of the stratification degree correspond to a cumulative signal of biological activity, and these patterns do not necessarily indicate an increase of biological rates with stratification. Prior to the May 2007 cruise, a major storm occurred in the area that induced enhanced vertical mixing compared to the June 2006 and May 2008 cruises, and this might have added additional variability in the 2007 data set as a function of the degree of stratification. Hence, linear regressions of variables as a function of the degree of stratification were computed either on the basis of the 2006 and 2008 data sets or on the 2006, 2007, and 2008 data sets.

[30] The patterns of the variables in surface waters as a function of the stratification degree were remarkably consistent considering that data from three cruises carried out in different years were merged together (Figure 6). The increase of %O2 and decrease of PO43− were consistent with organic carbon production during the bloom development (increasing stratification). The pattern in TAanomaly was consistent with CaCO3 production during the bloom development (increasing stratification) of mixed phytoplanktonic communities. The strongest TAanomaly (−32 μmol kg−1) was comparable to the strongest values of −28 μmol kg−1 reported by Holligan et al. [1993] in the North Atlantic in June 1991, and of −35 μmol kg−1 reported by Bates et al. [1996] in the Sargasso Sea in February 1992, but weaker than the value of −82 μmol kg−1 reported by Murata and Takizawa [2002] in the Bering Sea in October 2000. These TAanomaly values observed in naturally occurring blooms remain well below the TA drawdown observed in confined environments with coccolithophores of ∼300 μmol kg1 in mesocosms [e.g., Delille et al., 2005] and of ∼1000 μmol kg−1 in batch cultures [e.g., De Bodt et al., 2010]. pCO2@13°C showed a decreasing pattern with stratification, indicative that the net effect on pCO2 of organic carbon production dominated over the net effect of calcification. The pattern with stratification of pCO2@SST has lower statistical significance than that of pCO2@13°C due to the increase of SST with stratification and subsequent effect on the CO2 solubility and pCO2@SST. DIC decreased during the bloom development (increasing stratification) due to the combined effect of NCC and NCP, as discussed hereafter. Although the linear regression has a low statistical significance, the general increasing pattern of ΩCAL during the bloom development (increasing stratification) suggests that the combined (increasing) effect of NCP and SST change dominated over the (decreasing) effect of NCC, as discussed hereafter. Also, the values of ΩCAL were higher than the wintertime value of ∼3.7 (computed from a TA of 2340 μmol kg−1, a salinity of 35.5, a SST of 12°C and pCO2 at atmospheric equilibrium, based on wintertime data in the area reported by Frankignoulle and Borges [2001]). This shows that NCP increased ΩCAL values with regards to wintertime values. A temperature increase from the wintertime value of 12°C to 13°C (average SST during the cruises) can only account for an increase of ΩCAL of 3%, while the increase between observed ΩCAL values and the wintertime ΩCAL value ranged between 9% and 34%.

Figure 6.

Average values in surface waters (top 20 m) of %O2, TAanomaly, DIC, ΩCAL, and PO43−, and underway pCO2@13°C and pCO2@SST, and SST (2 m) as a function of the degree of stratification, during the June 2006 (grey circles), May 2007 (open circles), and May 2008 (black circles) cruises in the northern Bay of Biscay. The solid line corresponds to the linear regression based on the 2006 and 2008 data sets (corresponding r2 is not italicized), the dotted line corresponds to the linear regression based on the 2006, 2007, and 2008 data sets (corresponding r2 is italicized).

[31] The low values of pCO2 and DIC in surface waters observed in the area during the cruises could be in part due to NCP during the diatom blooms that occurred prior to the cruises and peaked in mid-April (Figure 2). However, mixing at the shelf break (section 3.1) also induced the increase of pCO2 and DIC in surface waters as indicated by the high values of these quantities for the lowest degree of stratification (Figure 6). The mixing at the shelf break and injection of nutrients to surface waters are assumed to be the factors that trigger the coccolithophore blooms in the area at the period of the year the cruises were carried out [Harlay et al., 2010]. Since the stations sampled over the continental shelf and slope were close to the shelf break, we assume that the sampled water masses were derived from the cold and nutrient rich source waters at the shelf break. Hence, we also assume that the patterns of biogeochemical variables as a function of stratification in Figure 6, and the effects of NCP and NCC on seawater carbonate chemistry evaluated hereafter are mainly related to the activity of the recently bloomed mixed phytoplanktonic assemblages dominated by coccolithophores.

3.4. Evaluation of the Effect of NCP, NCC, and SST on DIC, pCO2, and ΩCAL

[32] The cumulative effects of NCP, NCC, and SST change on DIC, pCO2 and ΩCAL were evaluated as described in section 2.3. ΔDICorg and ΔpCO2org decreased, and ΔΩCALorg increased with increasing stratification, consistent with a cumulative effect of NCP as the water mass stratified and the bloom developed (Figure 7). ΔDICinorg and ΔΩCALinorg decreased, and ΔpCO2inorg increased, with increasing stratification, consistent with a cumulative effect of NCC as the water mass stratified and the bloom developed. ΔpCO2SST and ΔΩCALSST increased with increasing stratification, consistent with the warming of surface waters as the water mass stratified. The cumulative effect of NCP on DIC as well as on pCO2 was stronger (ΔDICorg down to −60 μmol kg−1 and ΔpCO2org down to −100 ppm) than the effect of NCC (ΔDICinorg down to −20 μmol kg−1 and ΔpCO2inorg up to +20 ppm). The cumulative effect of NCC on pCO2 was also weaker than the effect of SST change (ΔpCO2SST up to +60 ppm). The highest ΔpCO2inorg values we report were close to maximal values reported by Robertson et al. [1994] in northeast North Atlantic Ocean in June 1991 (∼15 ppm), and those given by the model of Buitenhuis et al. [2001] (∼25 ppm) calibrated with data from the northern North Sea in June 1993 [Buitenhuis et al., 1996]. However, the ΔpCO2inorg values we report were lower than the highest value reported by Murata and Takizawa [2002] in the Bering Sea in October 2000 (∼100 ppm). This is mainly due to very high TA drawdown (−82 μmol kg−1) [Murata and Takizawa, 2002], and to a lesser extent due to the lower seawater buffering capacity in the Bering Sea (salinity ∼32). The cumulative effect of NCP as well as SST change on ΔΩCAL was stronger (ΔΩCALorg up to +0.9, ΔΩCALSST up to +0.4) than the effect of NCC (ΔΩCALinorg down to −0.3).

Figure 7.

The ΔDICorg, ΔDICinorg, ΔpCO2org, ΔpCO2inorg, ΔpCO2SST, ΔΩCALorg, ΔΩCALinorg, and ΔΩCALSST as a function of the degree of stratification, during the June 2006 (grey symbols), May 2007 (open symbols), and May 2008 (black symbols) cruises in the northern Bay of Biscay. The solid line corresponds to the linear regression based on the 2006 and 2008 data sets (corresponding r2 is not italicized), the dotted line corresponds to the linear regression based on the 2006, 2007, and 2008 data sets (corresponding r2 is italicized).

[33] The ΔDICinorg:ΔDICorg ratio increased with stratification (Figure 8), in agreement with the life cycle of coccolithophores. As the water mass stratifies and nutrients become depleted in surface waters, coccolithophores shift from the organic carbon production and growth phase to the stationary and calcification phase [Fernández et al., 1993; Paasche and Brubak, 1994; Paasche, 2002; Delille et al., 2005]. The ΔDICinorg:ΔDICorg values (0.00 to 0.42, on average 0.13 for the three cruises) were in relatively good agreement with C:P values based on calcification and primary production rates derived from 14C incubations reported during several studies in the North Atlantic Ocean: 0.14–0.19 in the northeast North Atlantic Ocean in June 1991 [Fernández et al., 1993], 0.03–0.18 in the northern North Sea in July 1993 [van der Wal et al., 1995], 0.14–0.16 in the northern North Sea in July 1994 [Marañon and González, 1997], 0.03–0.18 in the northern North Sea in June 1999 [Rees et al., 2002], 0.02–0.31 in the northern Bay of Biscay in June 2004 [Harlay et al., 2010].

Figure 8.

The ΔDICinorg:ΔDICorg as a function of the degree of stratification, during the June 2006 (grey circles), May 2007 (open circles), and May 2008 (black circles) cruises in the northern Bay of Biscay. The solid line corresponds to the linear regression based on the 2006 and 2008 data sets (corresponding r2 is not italicized), the dotted line corresponds to the linear regression based on the 2006, 2007, and 2008 data sets (corresponding r2 is italicized). The data point in brackets was assumed to be an outlier and was excluded from the linear regression analysis.

[34] The much larger effect of NCP than NCC on seawater carbonate chemistry is confirmed by the value of the Revelle factor of 8.4 in the top 30 m of the water column (for the whole data set). This value is closer to the theoretical value of 10.0 if only DIC uptake and release by net organic carbon production and degradation occurred in the water column, than the theoretical value of −7.5 if only DIC uptake and release by CaCO3 precipitation and dissolution occurred in the water column (computed for a temperature of 13°C, salinity of 35.5, pCO2 of 300 ppm, and TA of 2340 μmol kg−1) [Frankignoulle, 1994]. Based on the Revelle factor computed from measured profiles in the top 30 m of the water column from the whole data set (8.4), we evaluated theoretically the C:P ratio yielding a value of 0.12, in close agreement with the average ΔDICinorg:ΔDICorg ratio (0.13) derived independently.

[35] The computed changes of DIC, pCO2, and ΩCAL (ΔDICcomputed, ΔpCO2computed, and ΔΩCALcomputed, respectively) were remarkably consistent with the observed changes of DIC, pCO2, and ΩCAL (ΔDICobserved, ΔpCO2observed, and ΔΩCALobserved, respectively), considering that these quantities were evaluated independently (Figure 9). Yet, the values of ΔDICobserved and ΔpCO2observed were almost always systematically lower than the values of ΔDICcomputed, ΔpCO2computed, and the values of ΔΩCALobserved were almost always systematically higher than the values of ΩCALcomputed. This could indicate that additional processes affecting seawater carbonate chemistry dynamics in surface waters were not taken into account in the computations. First, phosphorous was assumed to be assimilated by phytoplankton exclusively as PO43−, however, coccolithophores can rely on dissolved organic phosphorous (DOP) to meet part of their phosphorous requirements [Egge and Heimdal, 1994; Paasche, 2002; Lessard et al., 2005]. Secondly, in low or depleted nutrient conditions, a significant amount of carbon fixed by photosynthesis is released as dissolved organic carbon (documented for coccolithophores by Fernández et al. [1996] and Engel et al. [2004a, 2004b]), which cannot be accounted for in our computations since it is independent of nutrient assimilation (carbon overconsumption) [e.g., Toggweiller, 1993; Anderson and Sarmiento, 1994; Banse, 1994]. A significant amount of dissolved primary production in the study area is consistent with the high TEP concentrations reported during the cruises [Harlay et al., 2009; J. Harlay, C. De Bodt, and L. Chou, unpublished data, 2009]. Both these explanations are consistent with the fact that the differences between observed and computed ΔDIC, ΔpCO2, and ΔΩCAL increased toward the strongest changes (Figure 9) which correspond to the most stratified and nutrient-depleted conditions (Figure 8).

Figure 9.

The ΔDICobserved versus ΔDICcomputed, ΔpCO2observed versus ΔpCO2computed, and ΔΩCALobserved versus ΔΩCALcomputed during the June 2006 (grey circles), May 2007 (open circles), and May 2008 (black circles) cruises in the northern Bay of Biscay. The solid line corresponds to the linear regression (forced through 0) based on the 2006 and 2008 data sets (corresponding r2 is not italicized), the dotted line corresponds to the linear regression (forced through 0) based on the 2006, 2007, and 2008 data sets (corresponding r2 is italicized). The 1:1 line is in bold.

3.5. Effect of Calcification on Air-Sea CO2 Fluxes

[36] Air-sea CO2 fluxes were computed at each sampled station (Table 1), and the cumulative effect of NCC on air-sea CO2 fluxes was evaluated by recomputing the air-sea CO2 fluxes with ΔpCO2air-sea from which ΔpCO2inorg was removed. The cumulative effect of NCC in decreasing the CO2 sink in the study area ranged between 0% and 72% depending on the station, but on average it was small, ∼12%. If this finding is confirmed in other oceanic regions, it would imply that the potential feedback on increasing atmospheric CO2 of the projected decrease of pelagic calcification [e.g., Gehlen et al., 2007; Ridgwell et al., 2007; Hofmann and Schellnhuber, 2009] due to thermodynamic CO2 “production” from calcification is probably minor, compared to the potential feedback related to the increase of NCP and carbon export [e.g., Riebesell et al., 2007].

4. Conclusions

[37] We report a data set of seawater carbonate chemistry obtained during blooms of the coccolithophore Emiliania huxleyi at the continental margin of the northern Bay of Biscay. Physical settings were the major factors controlling phytoplankton dynamics under the springtime high irradiance conditions. The cold, nutrient rich upwelled water at the shelf break moved over the shelf, warmed and stratified and then became depleted in nutrients, constraining the bloom close to the shelf break. Coccolithophores were present during these cruises as testified by high reflectance patches in remote sensed images, HPCL and SEM measurements. On the basis of HLPC measurements, coccolithophores accounted up to 70% of total Chl-a. Calcification by coccolithophores led to a marked drawdown of TA. Yet, the decrease of DIC (and increase of pCO2) due to NCC was overwhelmingly lower than the decrease of DIC and pCO2 due to NCP (NCC:NCP [ΔDICinorg:ΔDICorg] ratios, on average 0.13 for the three cruises). The NCC:NCP (ΔDICinorg:ΔDICorg) ratios could in fact be lower, since the computed effect of NCP on DIC was probably underestimated due DOP assimilation and/or dissolved primary production (carbon overconsumption).

[38] The overall effect of NCC in decreasing the CO2 sink in the area during the cruises was low (on average ∼12%). The primary production and calcification rates based on 14C incubations measured in the study area [Harlay et al., 2010; J. Harlay, unpublished data, 2009] are consistent with those typically reported in literature in the North Atlantic Ocean during blooms of coccolithophores [Fernández et al., 1993; van der Wal et al., 1995; Marañon and González, 1997; Rees et al., 2002]. The NCC:NCP (ΔDICinorg:ΔDICorg) ratios are also consistent with those reported in literature for the North Atlantic Ocean. The North Atlantic Ocean is one of the main regions of the global ocean where coccolithophore blooms are the most intense and recurrent [Brown and Yoder, 1994; Balch et al., 2005, 2007]. Also, Emiliania huxleyi is the most abundant and ubiquitous coccolithophore in the modern ocean [Paasche, 2002]. Hence, if one assumes that the overall low effect of NCC in decreasing the CO2 sink is a general feature in naturally occurring blooms in the global ocean, then the potential feedback on increasing atmospheric CO2 of the projected decrease of pelagic calcification [e.g., Gehlen et al., 2007; Ridgwell et al., 2007; Hofmann and Schellnhuber, 2009] due to thermodynamic CO2 “production” from calcification is probably minor. Furthermore, in naturally occurring blooms, the cumulative effect of NCP leads to an increase of ΩCAL above wintertime values (at or close to CO2 atmospheric equilibrium). Hence, a future increase of NCP [e.g., Riebesell et al., 2007] could to some extent counteract the effect of ocean acidification on ΩCAL, in addition to the predicted effect of SST increase on ΩCAL [Cao et al., 2007; McNeil and Matear, 2007].

Acknowledgments

[39] We are grateful to the officers and crewmembers of the R.V. Belgica and to J. Backers, J.-P. De Blauw and G. Deschepper (Unit of the North Sea Mathematical Models) for assistance during the cruises, to S. Groom (Remote Sensing Group of the Plymouth Marine Laboratory) for providing remote sensing images, to N. Van Oostende and K. Sabbe (Ghent University) for providing the HPLC and SEM data, to A. Engel (Alfred Wegener Institute) for additional SEM data, to M.-V. Commarieu for analytical assistance, and to P. Boyd (Associate Editor) and two anonymous reviewers for encouraging comments on the previous version of the ms. This work was carried out in the frame of Belgian Science Policy project “Role of pelagic calcification and export of carbonate production in climate change” (PEACE, SD/CS/03), and contributes to the European Integrated Project “Toward an integrated marine carbon sources and sinks assessment” (CARBOOCEAN, 511176). AVB and BD are research associates at the Fonds National de la Recherche Scientifique. The first and last authors equally contributed to data analysis and manuscript drafting.

Ancillary