Variable export fluxes and efficiencies for calcite, opal, and organic carbon in the Atlantic Ocean: A ballast effect in action?

Authors


Abstract

[1] Latitudinal variability in export fluxes and efficiencies (ThE) of calcite, opal, and particulate organic carbon (POC) were examined during a basin-scale Atlantic Ocean cruise. A clear relationship between integrated euphotic zone POC and calcite export combined with similarities in average ThE for calcite (0.26), opal (0.31), and POC (0.29) implies a potential association between biomineral and POC export. However, such similarity conceals substantial uncorrelated variability when ThE values are compared on regional scales, with ThE of POC often being much higher than that of calcite or opal. High-euphotic zone ThE for POC (0.3–0.4) relative to that found in deep sea sediment traps (<0.05) suggests that considerable remineralization occurs below the euphotic zone. We suggest (1) that regional variability in the mechanisms by which biominerals and POC become associated are more important in determining the efficient export of organic carbon than that of ballast materials; and (2) that, because of the preferential remineralization of POC relative to calcite/opal dissolution during subeuphotic processes, the potential for effective ballasting increases with depth in the water column.

1. Introduction

[2] The capacity of the biological carbon pump to transfer atmospheric CO2 into the deep ocean is dependent upon the efficiency of particulate organic carbon export and the degree of remineralization as particles sink [Sarmiento et al., 2004]. The fate of particulate matter produced within the euphotic zone is governed by competition between dissolution/remineralization and export, with a return to the solution phase being enhanced by increasing residence times in the upper ocean. Strong correlations observed in the deep ocean between the vertical fluxes of particulate organic carbon (POC) and of inorganic material (calcite, opal, clay) have been used to suggest that mineral phases may enhance the export and survival of organic matter as it sinks into the deep ocean (the “ballast effect”), by increasing the density and sinking speeds of particle aggregates [Klaas and Archer, 2002], and by providing some protection from remineralization [Armstrong et al., 2002]. However, the underlying processes are not well understood and the contrary has been suggested whereby organic aggregates scavenge nonsinking mineral material, so that the flux of POC determines the flux of minerals to the deep sea and not the reverse [Passow, 2004; Passow and De La Rocha, 2006]. Rapid sinking of material will prevent significant dissolution of opal and calcite and remineralization of organic carbon [Ragueneau et al., 2000]. However, biomineral dissolution in the upper ocean may preclude the efficient export of inorganic, and any associated organic material [Nelson et al., 1995; Milliman et al., 1999; Feely et al., 2002; Brzezinski et al., 2003].

[3] Realistic modelling of the biological pump and the oceanic pathways for atmospheric CO2 sequestration [Sarmiento et al., 2004] depends on developing an improved understanding of the mechanisms of particle formation and rates of remineralization/dissolution, and their variability with respect to depth and regional hydrography [Buesseler et al., 2001; Cochran et al., 2000, Ragueneau et al., 2006]. Although there are relatively few measurements of biomineral formation, the accepted paradigm is that a substantial proportion (∼50–80%) of the calcite and opal produced in the euphotic zone dissolves within the upper (<1 km) ocean [Nelson et al., 1995; Milliman et al., 1999; Feely et al., 2002; Brzezinski et al., 2003]. Opal dissolution is mediated by low, undersaturated concentrations of silicate relative to particulate concentrations [Nelson et al., 1995; Brzezinski et al., 2003] and bacterial action [Passow, 2004; Bidle and Azam, 1999], while calcite dissolution may be mediated by localized acidic conditions, such as within (micro-/macro-) zooplankton guts [Harris, 1994; Pond et al., 1995; Hansen et al., 1996; Jansen and Wolf-Gladrow, 2001; Langer et al., 2007] and in microenvironments during aggregate (faecal pellets, marine snow) formation [Milliman et al., 1999].

[4] A comparison of measurements of primary production and biomineralization in the (sub)tropical Atlantic with published sediment trap data [Poulton et al., 2006a] indicated that the proportion of organic carbon associated with mineralizing phytoplankton production is greater than that exported to the deep sea, that calcite is the major biomineral and has a turnover time within the euphotic zone comparable to that of the phytoplankton (∼3 d), and that ∼70% of the calcite being formed is dissolved in the upper 2–3 km of the ocean. Clearly, the nature of the association between particulate organic carbon and mineral material both within and below the euphotic zone is fundamental to understanding how organic carbon export is controlled. Within the euphotic zone, ballasting may result from physical associations during biomineral and organic carbon formation (cellular), packaging (grazing) or aggregation [Passow, 2004].

[5] In this study, we compare the production and export of biominerals and organic carbon from the euphotic zone in the (sub)tropical Atlantic Ocean in order to examine (1) whether the strong correlations observed between particulate and biomineral fluxes in deep (>2 km) sediment trap data [e.g., Klaas and Archer, 2002] are evident at shallower water depths (i.e., within the euphotic zone); and (2) if the export efficiencies (ThE = export/surface production) of the different particle types are related. Export efficiencies of POC, calcite, and opal are determined by comparing surface production rates of organic carbon (i.e., photosynthetic rates) and biomineral phases (calcification, silicification) [from Poulton et al., 2006a], to shallow (<100 m) particulate export fluxes of organic carbon, calcite and opal. High organic carbon export can result from low-productivity systems with efficient export (little remineralization) or from high-productivity systems with inefficient export (extensive remineralization) [see Francois et al., 2002]. Thus in order to examine the relationships between biomineral and organic carbon export, it is necessary to consider both the magnitude of export and the amount of export relative to surface production as expressed by the export efficiency (ThE).

2. Sampling and Methods

[6] Measurements of particulate production and export were made during a cruise of the Atlantic Meridional Transect (AMT-14) programme between the Falkland Islands (50°S) and the UK (50°N) (see Figure 1) [Robinson et al., 2006]. Samples of large particulate (>50 μm) material were collected from predawn (0000–0700 LT) deployments of Stand Alone Pumps (SAPS; 1 depth) filtering ∼1500 L, while water samples (∼20 L) were collected from Niskin bottles attached to a CTD rosette sampler (10 depths). Simulated in situ phytoplankton organic carbon and biomineral production measurements were made on deck for 5 “light” depths (97, 55, 33, 14, and 1% incident irradiance) as determined from the in situ PAR sensor on the CTD frame. In this study, the euphotic zone was defined as the layer between the surface and the depth of 1% incident irradiance.

Figure 1.

Cruise track and positions of export sampling stations for the June 2004 Atlantic Meridional Transect (AMT-14). Cruise track is overlaid on a monthly sea surface chlorophyll-a (mg m−3) composite (April–June 2004). Note station positions relative to surface chlorophyll-a concentrations: (1) relatively high chlorophyll-a temperate waters in both hemispheres (48.6°N, 41.6°S); (2) relatively high chlorophyll-a waters of both the northern (38.4°N) and southern (32.6°S) subtropical convergences; (3) intermediate chlorophyll-a concentrations for equatorial stations (0.1°S, 11.2°N); and (4) low chlorophyll-a concentrations in the northern (22.3°N, 29.3°N) and southern (12.2°S, 24.1°S) subtropical gyres. Image courtesy of ORBIMAGE.

[7] The methods for determining calcite, opal, and POC concentration (precision <15%) and their rates of production can be found in the work of Poulton et al. [2006a]. The mean relative standard deviation (RSD) for calcite, opal, and POC production rates, calculated as the average of all the relative standard deviations for production rates derived from triplicate measurements, were 31%, 28% and 14% respectively [Poulton et al., 2006a]. The small volumes (<0.3 L) used for production measurements undersample large and rare mineralizing plankton (foraminifera, pteropods, radiolarians), whereas SAPS (∼1500 L) should better sample such organisms. This difference in sampling volumes implies that estimates of export efficiency for calcite and opal may be based on underestimates of total calcification and silicification. However, large calcifiers and silicifiers are relatively rare in (sub)tropical waters. For example, in the upper ocean coccolithophores and diatoms are present at cell densities of 104–107 cells m−3, whereas forams, pteropods, and radiolarians are present at cell densities of 102–104 cells m−3 [Baumann et al., 2004]. Thus any overestimation of mineral export efficiencies is likely to be small. In interpreting the data set we present here, it is important to note that the biomineralizing component of the planktonic community was dominated by calcite rather than opal producing organisms [Poulton et al., 2006a].

[8] Export of POC, calcite, and opal were calculated from water column 234Th/238U disequilibria [e.g., Buesseler, 1998; Thomalla et al., 2006]. Estimates of the total methodological error in export fluxes were based on propagated methodological uncertainties (full details in the work of Thomalla et al. [2006]) for the 234Th deficit calculations and an average relative standard deviation for particulate measurements of 15%. Single activity profiles as obtained during this study lack information on spatial-temporal variability and necessitate the assumption of steady state. The data therefore do not allow estimation of horizontal and vertical advective contributions to thorium sinking fluxes, but they are assumed insignificant compared to scavenging rates and export on sinking particles. This assumption provides adequate resolution of 234Th fluxes in many steady state settings [e.g., Tanaka et al., 1983; Wei and Murray, 1991; Moran and Buesseler, 1993; Buesseler et al., 2001]. Non-steady-state effects are however important during periods of significant 234Th drawdown, such as during phytoplankton blooms and postbloom conditions [Buesseler et al., 1992, 1998, 2001; Cochran et al., 1997], as well as in regions of high upwelling velocity [Buesseler et al., 1995; Bacon et al., 1996; Dunne and Murray, 1999]. As we have ignored the upwelling term, it is likely that our 234Th and subsequent POC, calcite, and opal flux estimates in equatorial upwelling waters may represent lower limits. At two stations where Poulton et al. [2006a] report biomineral production data the depletion of 234Th in the upper ocean was not statistically significant, therefore estimates of export from these stations are given in Table 1 as zero.

Table 1. Compilation of Station Positions, Euphotic Zone Depths (Zeup), Estimates of Particulate Organic Carbon (POC), Calcite and Opal Export Fluxes, and Export Efficienciesa
LatitudeZeup, mExport, mmol m−2 d−1Export efficiencies, ThE
POCCalciteOpalPOCCalciteOpal
  • a

    Calculated errors for export fluxes are based on accounting for methodological uncertainties [Thomalla et al., 2006; see section 2]. ND indicates not determined. Square brackets contain values calculated from mean values of production and export. Parentheses indicate standard deviations.

  • b

    Means calculated from average values of production [Poulton et al., 2006a] and export.

41.6°S655.36 ± 1.860.39 ± 0.130.02 ± 0.010.140.090.04
32.6°S1033.69 ± 2.870.11 ± 0.080.04 ± 0.030.280.040.13
24.1°S1222.53 ± 3.260.10 ± 0.130.02 ± 0.020.20ND0.12
12.2°S1303.29 ± 3.250.05 ± 0.050.01 ± 0.010.220.020.05
0.1°S596.33 ± 3.580.34 ± 0.190.01 ± 0.010.260.140.01
11.2°N816.26 ± 6.810.41 ± 0.450.08 ± 0.090.210.790.14
22.3°N1150.000.000.00NDNDND
29.3°N1320.000.000.00NDNDND
38.4°N5011.76 ± 5.150.61 ± 0.270.24 ± 0.110.710.701.51
48.6°N282.84 ± 2.330.03 ± 0.030.08 ± 0.060.110.030.48
Mean 5.260.260.060.27 [0.25]b0.26 [0.12]b0.31 [0.19]b
(S.D.) (3.03)(0.21)(0.08)(0.19)(0.34)(0.51)

[9] Ratios of particulate 234Th to POC (range 6–15, mean 8.6 ± 2.9), to calcite (0.07–0.82, 0.42 ± 0.25) and to opal (0.01–0.25, 0.10 ± 0.07) were used to calculate the export of the various phases from the 234Th deficit measurements. These ratios were determined from samples obtained from 50 μm filters fitted on SAPS deployed at, or close to, the base of the euphotic layer [see Thomalla et al., 2006, Table 3]. To calculate export efficiencies (the numerical ratio of export/production; ThE after Buesseler et al. [1998]) for the different particle types, we have used the depth of the euphotic zone as the depth of integration for the steady state 234Th disequilibria model.

3. Results and Discussion

3.1. Export Fluxes and Efficiencies for Organic Carbon, Calcite, and Opal

[10] The magnitude of (molar) export fluxes for the different particulate materials (Figure 2) were higher for organic carbon (range 2.53–11.76 mmol C m−2 d−1) than for either calcite (0.03–0.61 mmol C m−2 d−1) or opal (0.01–0.24 mmol Si m−2 d−1) (Table 1). Calcite export fluxes were generally an order of magnitude higher than opal fluxes (Table 1). The station at 38.4°N with the highest organic carbon flux (11.76 mmol C m−2 d−1) was also the station with the highest biomineral fluxes (0.61 mmol C m−2 d−1 for calcite and 0.24 mmol Si m−2 d−1) (Table 1).

Figure 2.

Euphotic zone integrated (molar) rates of production (bottom white bars) and export (top gray bars) for (a) calcite (mmol C m−2 d−1), (b) opal (mmol Si m−2 d−1), and (c) organic carbon (mmol C m−2 d−1). Ratio of export to production (export efficiency) for each particle type is provided. ND indicates not determined. Note that export fluxes from stations in the northern subtropical gyre (22.3°N, 29.3°N) were below the detection limits of the 234Th technique.

[11] The export efficiencies (ThE) of all three particulate phases were highly variable (Figure 2) and ranged from 0.1 to 0.7 for POC, from 0.05 to 1.51 for opal and from 0.02 to 0.79 for calcite (Table 1). The mean values of ThE for the three phases (Table 1) were however similar (0.26 for calcite, 0.31 for opal, and 0.29 for organic carbon), suggesting that similar processes determine the efficiencies of mineral and organic carbon export. However, when ThE values for the three phases are compared on a station by station basis (Figures 3c and 3d) they appear unrelated (over the euphotic zone) in a way that is not predictable from surface rates of primary or biomineral productivity (Figure 2 and Table 1) and it becomes clear that such averages mask considerable variability (spatial and/or temporal) in the processes that determine ThE.

Figure 3.

Comparison of export fluxes (mmol m−2 d−1) and export efficiencies (export/production) for calcite, opal, and particulate organic carbon: (a) calcite and particulate organic carbon fluxes, (b) opal and particulate organic carbon fluxes, (c) calcite and particulate organic carbon export efficiencies (ThE), and (d) opal and particulate organic carbon export efficiencies (ThE). Dashed lines (c, d) represent a 1:1 line.

[12] As is found in deep sediment trap data [e.g., Klaas and Archer, 2002], there do appear to be linear relationships between organic carbon, calcite (Figure 3a) and opal fluxes (Figure 3b). Regression analysis on the statistical relationships between standing stocks, production rates, export fluxes and ThE for all three phases (Table 2) were carried out to examine whether processes leading to the efficient export of the various phases were correlated and could potentially have similar causal mechanisms. No significant relationships were found between the production rate of any of the phases and the export or ThE of any of the phases (Table 2). Clearly, high production rates do not automatically lead to high rates of export. For the full data set, POC ThE was found to be significantly correlated with POC export flux, opal export flux, and opal ThE (Table 2). Hence the processes leading to POC and opal export could be similar, whereas the absence of significant correlations between POC ThE and calcite fluxes and calcite ThE may indicate the lack of similar mechanisms. However, if the high export station (38.4°N) is ignored in the analysis, these relationships break down (Table 2) and thus the link between POC and opal export appears to be driven by high export fluxes.

Table 2. Regression Analysis for the Relationships Between Standing Stocks, Production Rates, Export Fluxes, and Export Efficiencies (ThE) for the Different Particle Types: Particulate Organic Carbon (POC), Calcite (PIC), and Opal (BSi)a
 Standing StocksProduction RatesExport FluxesExport Efficiencies
POCPICBSiPOCPICBSiPOCPICBSiPOCPICBSi
  • a

    Significant correlation coefficients are shown along with the level of significance as (a) p < 0.05, (b) p < 0.01, and (c) p < 0.001. NS indicates nonsignificant. Values in parentheses refer to correlation coefficients for relationships where the high export station (38.4°N) has been omitted. Values of organic carbon and biomineral standing stocks are taken from Poulton et al. [2006a].

Standing stocksPOC------------
 PICNS-----------
 BSiNSNSNS---------
Production ratesPOCNSNSNS---------
 PIC0.85b (0.83a)NSNSNS--------
 BSiNSNSNSNSNS-------
Export fluxesPOCNSNSNSNSNSNS------
 PICNSNSNSNSNSNS0.94c (0.90b)-----
 BSiNSNSNSNSNSNS0.82a (NS)NS ---
Export efficienciesPOCNSNSNSNSNSNS0.87b (NS)NS0.85b (NS)---
 PICNSNS−0.76a (0.96b)NSNSNSNS0.77b (NS)NSNS--
 BSiNSNSNSNSNSNS0.77a (NS)NS0.97c (NS)0.87b (NS)NS-

[13] There are no relationships between POC standing stocks and biomineral ThE (Table 2), as would be expected if POC scavenging were an important process in the (sub)tropical Atlantic Ocean (Table 2). Furthermore, POC fluxes and standing stocks are relatively large compared to biomineral fluxes and standing stocks [Poulton et al., 2006a, Table 1], suggesting that biomineral protection for organic matter [Armstrong et al., 2002] is likely to be of secondary importance in regulating organic carbon export in the upper ocean of the (sub)tropical Atlantic Ocean. Finally, there is no relationship between PIC or opal standing stocks and organic carbon flux or ThE, implying that a directly mediated ballast effect (i.e., only organic carbon directly associated with coccolithophore and diatom cells is exported) is unlikely to be occurring within our data set.

[14] The general pattern displayed at nearly all stations is for ThE of all three phases to be relatively low and for ThE of organic carbon to be higher than ThE for either biomineral (Figures 3c, 3d, and Table 1). However, at 38.4°N (G in Figures 3c and 3d; also see Table 1) ThE of all three phases were large and the biomineral export efficiencies were larger than or comparable to that of organic carbon. A comparison of biomineral to organic carbon ratios for material synthesised in the euphotic zone relative to the exported material provides a proxy for the relative changes in the characteristics of exported particles (Figure 4) [see Brown et al., 2006]. At most stations, these ratios show that if the ecosystem is operating at steady state, there is relatively more dissolution of biominerals in the euphotic zone than remineralization of organic carbon, although there is considerable variability between stations (Figure 4).

Figure 4.

Relative changes in the molar ratios of particulate material composition between surface production and export: (a) calcite to particulate organic carbon (calcite: organic carbon) and (b) opal to particulate organic carbon (opal: organic carbon). Values above the 1:1 (dashed lines) line indicate relatively high mineral dissolution from relatively high organic carbon remineralization.

[15] Although the reasons for generally higher ThE for POC than biominerals are unclear, it is an extremely important point to note since below the euphotic zone the two biomineral phases are better preserved than organic carbon [Poulton et al., 2006a]. Hence there must be a depth horizon within the twilight zone (below the euphotic zone) at which the relative labilities of the biomineral and organic carbon phases become reversed. It is also noticeable that at the one station where ThE were high (38.4°N), those for biominerals were comparable to or higher than those for organic carbon. Thus the mean conditions whereby organic matter is exported more efficiently from the euphotic zone is compensated for by phytoplankton bloom conditions with relatively more efficient export of biominerals. It may be the latter process that accounts for better preservation of biominerals than organic carbon over the full depth of the water column.

[16] Comparison of both the magnitude of export fluxes and the export relative to surface production (ThE) for organic carbon, calcite, and opal in the (sub)tropical Atlantic Ocean show that (1) although linear relationships appear to characterise the relationship between absolute fluxes of organic carbon, calcite and opal, there is significant regional variability and high biomineral fluxes are not always related to high organic carbon fluxes particularly with respect to opal (Figure 3b and Table 2); and (2) although average ThE were similar for the three particulate types, there is considerable station to station differences suggesting that the efficient export of biominerals does not necessarily enhance ThE of organic carbon (Table 2). Such regional disparity in export fluxes and ThE are likely to be related to regional patterns in hydrography and planktonic ecosystem structure in the (sub)tropical Atlantic Ocean and are examined in the context of these factors in section 3.2.

3.2. Regional Variability in Export Flux and Export Efficiency

[17] A variety of hydrographic provinces and ecosystems are sampled along the AMT transect, including the permanently stratified oligotrophic subtropical gyres, tropical equatorial upwelling waters, seasonally variable subtropical convergences and the seasonally mixed temperate waters at either end of the transect [Robinson et al., 2006]. Surface and upper ocean (<50 m) chlorophyll-a concentrations are low (<0.10 mg m−3) within both subtropical gyres of the Atlantic Ocean, with elevated chlorophyll-a concentrations (>0.3 mg m−3) associated with the equatorial upwelling (10°S–15°N), and seasonally in the subtropical convergences and temperate waters of both hemispheres [Robinson et al., 2006, Figure 1].

[18] Generally, small picophytoplankton (<2 μm) dominate both biomass and organic carbon production throughout the AMT transect [Zubkov et al., 1998; Maranon et al., 2000; Poulton et al., 2006b], although there are slight increases in the biomass and productivity of larger phytoplankton cells in equatorial upwelling waters [Perez et al., 2005] and in temperate waters during (northern) spring [Tarran et al., 2006]. The grazer community is more variable over the AMT transect, with nanoflagellate (2–20 μm) and microzooplankton (20–200 μm) grazers dominant in the subtropical gyres, while mesozooplankton (>200 μm) biomass and grazing pressure are relatively more important in equatorial and temperate waters [Huskin et al., 2001; Isla et al., 2004].

[19] Low rates of calcification and especially silicification characterised the central subtropical gyres (Figure 2) [Poulton et al., 2006a], highlighting the low biomass of coccolithophores and diatoms in the subtropical and tropical ocean [Nelson et al., 1995; Haider and Thierstein, 2001]. In the southern subtropical gyre (12.2°S, 24.1°S, 32.6°S), low ThE for calcite (0.02–0.04), opal (0.05–0.13) and organic carbon (0.20–0.28) also suggest that a large proportion of surface production was retained in the upper ocean. A comparison of biomineral to organic carbon ratios for surface production and in exported material for stations in the southern subtropical gyre (Figure 4) both indicate preferential dissolution of mineral material (opal, calcite) relative to organic carbon.

[20] Efficient nanoflagellate and microzooplankton grazing, as well as bacterial activity, are likely to enhance silica dissolution in subtropical surface waters by exposing individual diatom frustules to warm, undersaturated surface water silica concentrations [Nelson et al., 1995; Passow et al., 2003; Ragueneau et al., 2006] and by facilitating intracellular or aggregate associated calcite dissolution [Hansen et al., 1996; Milliman et al., 1999]. Dominance of the community by small cells, with slow sinking rates, may also promote the dissolution of calcite and opal by increasing their residence times in surface waters. The biological pump in the subtropical gyres, characterised by an active microbial loop, is likely to lead to efficient recycling of biomineral and organic carbon production [Brzezinski et al., 2003].

[21] By contrast, in equatorial waters of the Atlantic Ocean (0.1°S, 11.2°N), elevated nutrient concentrations due to localized upwelling [Longhurst, 1993; Perez et al., 2005] cause increases in chlorophyll-a concentrations (Figure 1) [Robinson et al., 2006] and increased rates of organic production relative to the subtropical gyres (Figure 2c) [Maranon et al., 2000; Poulton et al., 2006b]. Because of the strong grazing pressure by meso- and microzooplankton, there is little change in the size structure of the phytoplankton community [Perez et al., 2005]. Nevertheless, increases in the integrated rates of calcification and silicification (Figures 2b and 2c) [Poulton et al., 2006a] indicate that larger phytoplankton cells are present and efficiently grazed (therefore exported) by the high mesozooplankton biomass found in equatorial waters [Huskin et al., 2001; Isla et al., 2004]. Export efficiencies were high for calcite (0.14–0.79) and organic carbon (0.21–0.26) but low for opal (0.01–0.14) in equatorial waters (Figure 2 and Table 1). A comparison of biomineral to organic carbon ratios of surface production and export implies preferential organic carbon remineralization relative to calcite dissolution and preferential opal dissolution relative to carbon remineralization (Figure 4).

[22] Significant dissolution of opal in equatorial waters (Figure 3b) is likely to be mediated by enhanced mesozooplankton grazing [Huskin et al., 2001; Isla et al., 2004], which exposes individual diatom frustules to relatively low ambient silicate concentrations [Nelson et al., 1995], and also increases the number of broken diatom frustules [Roman and Rublee, 1980] and their subsequent colonization by bacteria [Ragueneau et al., 2006]. High-calcite ThE (0.79) in the northern equatorial station (11.5°N) was associated with relatively low ThE for organic carbon (0.21), which implies that there was no enhancement of organic carbon ThE associated with relatively high calcite ThE. Rather, in equatorial waters there was preferential export of calcite relative to organic carbon. Several mechanisms could be responsible for this; for example, feeding by equatorial mesozooplankton on coccolithophores, as opposed to the dominant picophytoplankton community, would produce calcite rich faecal pellets and promote calcite export. Another mechanism may be spatial-temporal uncoupling in production and export caused by mesoscale eddies, which are characteristic of low-latitude current systems [Longhurst, 1993; Perez et al., 2005]. Our observations of high-calcite ThE in the equatorial Atlantic also explain observations of high equatorial calcification rates coupled with low standing stocks of calcite [Poulton et al., 2006a].

[23] The subtropical convergences of the Atlantic Ocean (41.6°S, 38.4°N) are sites of seasonally enhanced biomass (Figure 1), production (Figure 2), and seasonal differences in the structure of the plankton community [Maranon et al., 2000; Poulton et al., 2006b; Robinson et al., 2006]. During AMT-14 (May 2004), early autumn conditions of enhanced mixing, low irradiance and moderate nutrient concentrations characterised the southern subtropical convergence while the northern subtropical convergence was experiencing late spring conditions of reduced mixing, high irradiance but decreasing nutrient concentrations [Maranon et al., 2000; Poulton et al., 2006a, 2006b; Robinson et al., 2006].

[24] In the southern subtropical convergence (32.6°S), rates of calcite, opal, and organic carbon production were all high, whereas the corresponding ThE values were all very low (0.09, 0.04, and 0.14, respectively; Figure 2 and Table 1). However, in the northern subtropical convergence (38.4°N) the opposite trends were found with low calcite, opal, and organic carbon production, whereas the ThE values were among the highest values found in this study (0.70, 1.51, and 0.71, respectively; Figure 2 and Table 1). A comparison of biomineral to organic carbon ratios for surface production and export also showed different patterns in the two subtropical convergences (Figure 4). In the southern subtropical convergence (32.6°S) there was preferential dissolution of both calcite and opal relative to organic carbon remineralization, whereas in the northern subtropical convergence (38.4°N) there was preferential organic carbon remineralization relative to opal dissolution, and to a lesser extent, calcite dissolution (Figure 4). As such differences are based on rather few data points, we cannot yet extrapolate to characterise seasonal or regional patterns in the relationships between POC and biominerals.

[25] Nevertheless, differences in ThE and relative biomineral dissolution/organic carbon remineralization patterns between the two subtropical convergences may be caused by seasonal differences in the temporal coupling of production and export. In the northern subtropical convergence, material being exported may represent the remnants of the spring bloom community settling out of the water column as low-nutrient conditions develop. Thorium-based measurements of export integrate over relatively long timescales (∼31 d; Buesseler [1998]) compared to relatively instantaneous measurements of primary production (<1 d). Elevated ThE may therefore result from significant temporal decoupling between export and primary production. Indeed, averaged SeaWiFS chlorophyll-a estimates obtained for the 34°N station clearly reveal elevated chlorophyll-a concentrations in the region ∼2 weeks prior to our in situ sampling and suggest that high-234Th flux rates measured here could represent export from a previous episode of elevated productivity [Thomalla et al., 2006].

[26] In contrast, consistently low ThE in the southern subtropical convergence suggest that the autumnal bloom may not have progressed to the export phase because of continuing high-nutrient conditions at the time of sampling [see Poulton et al., 2006a, Figure 2]. Spatial decoupling of production and export may also be important in the patterns observed in the northern subtropical convergence, as both convergences are strongly influenced by mesoscale physical instabilities [Garcon et al., 2001; Mouriño et al., 2003], with stations close to 38.4°S (35.5°N, 41.6°N) having high diatom and coccolithophore cell densities (A. Poulton, unpublished results, 2004, and T. Adey, personal communication, 2005) and high rates of calcification and silicification [Poulton et al., 2006a].

[27] It is likely that seasonal export from the northern subtropical convergence is due to large aggregates which sink much faster (∼100 m d−1) than the timescale for significant biotic dissolution of calcite [Jansen et al., 2002], or opal [Hill, 1992; Alldredge and Jackson, 1995], although there is evidence of preferential organic carbon remineralization relative to opal dissolution (Figure 4b). Under these conditions, it is possible for sinking material to scavenge other particulate material present in the water column (i.e., calcite, opal, and other phytoplankton cells) [Passow and De La Rocha, 2006] to further enhance ThE for all components.

[28] High opal ThE (0.48) relative to calcite (0.03) and organic carbon (0.11) ThE values was observed for the northern temperate waters (48.6°N) indicating that the large amount of material exported at this time contained relatively little calcite or organic carbon, although some export of each was measured (Table 1). Comparison of the ratio of opal to organic carbon in material produced in the surface waters relative to that exported (Figure 4b) showed relatively more organic carbon remineralization than opal dissolution. These differences may result from either (1) a phytoplankton community that is dominated by opal export originating from heavily silicified nutrient-stressed diatoms [Hutchins and Bruland, 1998; Timmermans et al., 2004], (2) the significant remineralization of organic carbon as material is repackaged and/or sinks, and/or (3) the export of predominantly aggregated diatoms which would reduce opal dissolution [Moriceau et al., 2007], while organic carbon continues to be utilised by bacteria within the aggregates [Smith et al., 1992].

4. Conclusions

[29] The positive correlation observed between euphotic zone calcite and organic carbon export fluxes (Figure 3a and Table 2) and the similarities in average ThE for the different particle types (Table 1) suggests a mechanistic relationship between efficient organic carbon and biomineral export in the (sub)tropical Atlantic Ocean. However, when the data are viewed on a regional basis, these relationships break down (Figure 3; see also Figure 2 and Tables 1 and 2). Similarly, a comparison of the biomineral to organic carbon ratio of surface production and exported material generally suggests preferential calcite or opal dissolution relative to organic carbon remineralization (Figure 4). We suggest that regional patterns of export and ThE for calcite, opal, and organic carbon (Figure 2) and relative differences in calcite/opal dissolution versus organic carbon remineralization (Figure 4) result from variability in the mechanisms controlling export in the (sub)tropical Atlantic Ocean. These may include latitudinal and seasonal differences in planktonic ecosystem structure, physical forcing and the degree of spatial-temporal coupling of surface production and export.

[30] As calcite dominates ballasting material in the (sub) tropical Atlantic ocean [Francois et al., 2002; Poulton et al., 2006a], a good relationship exists between calcite and organic carbon fluxes (Figure 3a and Table 2) rather than between opal and organic carbon fluxes (Figure 3b and Table 2). However, the lack of a distinct relationship between ThE for calcite, opal, and organic carbon (Figures 3c, 3d, and Table 2) implies that efficient organic carbon export from the productive euphotic zone may not always be enhanced by ThE of biominerals. Instead, efficient particulate export from the euphotic zone may be more dependent on the characteristics of the planktonic community composition (e.g., size spectra, taxa) and ecology (e.g., physiology, grazing pressure). Nevertheless, at one station in the northern subtropical convergence (38.4°N), the highest organic carbon fluxes were associated with high biomineral export and ThE was associated with high biomineral ThE (Tables 1 and 2). Therefore important exceptions to the pattern of decoupling between organic carbon and biomineral fluxes exist and may significantly influence annual export.

[31] At some stations (41.6°S, 32.6°S, 24.1°S, 12.2°S, 0.1°S), ThE for organic carbon (0.14–0.28) exceeded that for calcite (0.02–0.14) and opal (0.01–0.13) (Table 1); which is the reverse of ThE patterns observed in deep sediment traps, where only ∼1–2% of surface organic carbon production reaches the deep sea [Sarmiento et al., 2004] relative to ∼30–50% for biominerals [see Poulton et al., 2006a, Table 3]. These patterns imply significant remineralization of organic carbon relative to biominerals below the euphotic zone as particles sink and/or scavenge biomineral material as they settle [Passow and De La Rocha, 2006], such that opal and calcite are “chemically” decoupled from organic carbon during sinking leading to an increase in the biomineral to organic carbon ratio with depth. The potential for effective ballasting of the remaining organic carbon therefore increases with depth, which may account for the observation of a “ballast effect” in sediment trap material [Klaas and Archer, 2002].

[32] Overall, biomineral production at the stations sampled in the (sub)tropical Atlantic Ocean was dominated by calcite producing organisms rather than those producing opaline frustules [Poulton et al., 2006a]. Thus the results presented here cannot be considered representative of those from highly productive systems such as the North Atlantic spring diatom bloom or coastal upwelling systems. Clearly, making parallel measurements of biomineral and organic carbon production in such regions is a high priority for the global community.

Acknowledgments

[33] We thank the officers and crew of the RRS James Clark Ross and the technical support of the UKORS staff. We are grateful for assistance from M. Stinchcombe, T. Adey, L. Brown, P. Warwick, K. Chamberlain, M. Woodward, W. Balch, R. Head, D. Green, and R. Pearce. Support for this study came from the UK NERC through a small grant awarded to R. Sanders, the Atlantic Meridional Transect consortium (NER/O/S/2001/00680), and the George Deacon Division. S. Thomalla gratefully acknowledges support from a commonwealth split-site Ph.D. scholarship. This is contribution 153 of the AMT programme.

Ancillary