Influence of the Amazon River on dissolved and intracellular metal concentrations in Trichodesmium colonies along the western boundary of the sub-tropical North Atlantic Ocean

Introduction Conclusions References


Introduction
Marine cyanobacteria in the genus Trichodesmium are important N 2 -fixing organisms in oligotrophic tropical and subtropical oceans, playing an important role in primary production and in the new nitrogen budget (Karl et al., 1997;Capone et al., 1997;LaRoche and Breitbarth, 2005;Westberry and Siegel, 2006).Therefore, important advances have been made in understanding their geographical distribution and abundance (e.g., Carpenter et al., 2004;Westberry and Siegel, 2006), nutrient requirements (Mills et al., 2004;Tuit et al., 2004;Kustka et al., 2002;Sa ñudo-Wilhelmy et al., 2001) and the effect of environmental variables on bloom dynamics (Subramaniam et al., 2008;Tovar-Sanchez et al., 2006).However, some critical issues still remain largely unknown, such as evaluating the impact of the Amazon River on the cycling of bioactive dissolved metals, and the impact of that river on intracellular metal levels in field populations of Trichodesmium.Whereas total metal composition of field-collected Trichodesmium colonies have been reported (Tovar-Sanchez et al., 2006), their internal metal pool and its relation to both the Amazon River plume and bloom dynamics are still unknown.The internal metal pool represents the biological fraction (e.g., cellular quota, Hassler and Schoemann, 2009;Tovar-Sanchez et al., 2003), and therefore quantification of this metal pool is important to understand the biochemical status and/or the nutritional requirements of Trichodemium in the field.Furthermore, simultaneous measurements of bioactive trace metals in the intracellular and in the dissolved water column pools could help us to better understand the processes and mechanisms influencing the bioavailability of trace elements in the field.
In this study, we investigated the geographical distribution of dissolved elements in surface waters both in and out of the influence of the Amazon River plume, and the relationship of the plume with the internal metal composition of

Sampling and analysis
Sampling was carried out during April-May 2003 under high riverine flow conditions (DeMaster and Pope, 1996).A total of 42 locations, distributed along the western tropical North Atlantic Ocean within and outside of the influence of the Amazon River (Fig. 1), were sampled in this study.Dissolved surface seawater samples were collected from a "towed fish" deployed at 2 m below the surface and towed at about 5 knots during sampling.Seawater was pumped through acid-cleaned Teflon tubing coupled to a C-flex tubing (for the Cole-Parmer peristaltic pump head), filtered through an acidcleaned polypropylene cartridge filter (0.22 µm, MSI, Calyx®), and collected in a 1 L LDPE acid-washed bottle.Seawater samples were acidified with sub-boiling quartz distilled HCl (Q-HCl) to pH <1.5 and stored for at least 1 month prior to analysis.Metal concentrations (Ag, Cd, Co, Cu, Fe, Mo, Ni, Pb, and V) were determined by Inductively Coupled Plasma Mass Spectrometry (ICP-MS; ThermoFinigan, Element 2) after pre-concentration with an ammonium1-pyrrolidine-dithiocarbamate/diethylammonium diethyldithiocarbamate (APDC/DDDC) organic extraction (Bruland et al., 1985).Dissolved phosphorous concentrations were determined using the MAGIC method (Karl and Tien, 1992).
Trichodesmium colonies were collected at a depth of ∼5 m using an acid-cleaned all-plastic 100-micron mesh plankton net.Individual colonies were removed from the acid-cleaned polyethylene net collector with a plastic inoculating loop in a class-100 laminar flow hood.Approximately 100 colonies were collected at each location and stored frozen in Teflon vials until acid digestion.Samples were digested in Teflon digestion vials using combined Q-HNO 3 (60%), Q-HCl (30%) and Q-HF (10%), and heated on a hot plate until complete digestion (Sa ñudo-Wilhelmy et al., 2001;2004).described by Tovar-Sanchez et al. (2003).

Dissolved trace metal distributions
Dissolved metal (Ag, Cd, Co, Cu, Fe, Mo, Ni, Pb, V) and P concentrations measured in surface waters during our study are reported in Table S1 in the auxiliary material section.The impact of the Amazon River plume on the chemical composition of our sampling locations is reflected in the salinity gradient (22.4-36.3)as well as in the dissolved trace element distributions (Table S1).and S1a-f in the supplementary information), suggesting that the Amazon River plume was the main source of these elements to the area of study during our sampling.
In contrast, the spatial distribution of dissolved Cd, Mo and V (range from low to high salinity: Cd, 0.31-0.53nM; Mo, 0.08-0.13µm; V, 20.92-35.69nM) varied directly with salinity, suggesting that the Amazon River plume is not the source of these elements, but rather the river input diluted their concentration in the Ocean (Figs. 2b and S1g-h in the supplementary information).
To further establish the impact of the Amazon River plume on trace element distributions in our area of study, we collected water samples at 6 stations located in and out of the plume in a 24 h period (salinity range: 22.4-30.9;stations 30-35 in Table S1).
The positive or negative slopes obtained for different elements in the linear regression analysis with salinity confirms our results regarding the different influence of the river on metal concentrations discussed above (insets in Figs. 2 and S1 in the supplementary information).Although our data is very limited, we used those metal-salinity linear Introduction

Conclusions References
Tables Figures

Back Close
Full regressions to obtain a first-order approximation of the "hypothetical" metal concentrations at the low-salinity end-member of the Amazon River basin that would explain the concentrations measured at the seawater end member assuming conservative estuarine mixing (Table 1).Predicted metal concentrations at salinities 5 and 10 for Cd (0.10 and 0.16 nM), V (11. 5 and 14.3 nM), Co (145.5 and 130.0 pM) and Ni (2.4 and 2.3 nM) were in good agreement with the concentrations measured by Seyler and Boaventura (2003) in the upper Amazon River basin (Table 1).However, calculated concentrations at those salinities for Mo (31.3-44.5 nM) and Cu (16.1 and 13.6 nM) were higher than those reported for the upper river basin (Mo: 0.5 ± 0.3 nM and 0.4 ± 0.2 nM; Cu: 2.9 ± 0.6 nM and 6.7 ± 4.7 nM, at salinities 5 and 10 respectively).These discrepancies are not totally unexpected for some metals, as inputs from shelf sediments could increase metal levels within the high-salinity end-member of the Amazon plume.For example, Breckel et al. (2005), reported that the Amazon shelf sediment is a significant source of Mo to the ocean (with a estimated flux to the dissolved phase of 0.6×10 7 mol/yr), which could cause an overestimation in our hypothetical low-salinity end-member calculations, due to non-conservative excesses relative to simple mixing of freshwater and seawater.Further studies will need to address all of those issues.

Trace element composition in Trichodesmium colonies
In order to investigate the influence of dissolved metal concentrations on the elemental composition in Trichodesmium, we analyzed two different metal pools, intracellular and surface-scavenged, in the colonies collected at the western tropical North Atlantic Ocean (Fig. 1).Total (intracellular + surface-scavenged; Ag, Al, Cd, Co, Cu, Fe, Mn, Mo, Ni, P, Pb, V) and intracellular metal (Cd, Co, Cu, Fe, Mn, Mo, Ni, P, V) concentrations measured in those colonies are reported in Table S1.
Element abundances in the intracellular fraction ranged from 0.49 ± 0.27 mol per colony for P to undetectable levels for Pb (P>Fe>V>Ni>Mo>Zn>Cu>Mn>Co>Cd>Pb) (Table S1).The general concentration pattern seems to be consistent with the metabolic requirements of Trichodesmium.essential nutrients and considered limiting elements for N 2 fixation, oceanic primary production and carbon dioxide sequestration (Sa ñudo- Wilhelmy et al., 2001;Mills et al., 2004).The high internal content of Fe and Mo in the colonies is consistent with the requirements for these elements by the nitrogenase enzyme for nitrogen fixation (Tuit et al., 2004), and for other Fe proteins involved in aerobic nitrogen fixation in Trichodesmium spp.(Zehr et al., 1997).
Explanation for the high internal concentrations of Ni and V are not totally clear.
In the case of Ni, recent research indicates that Ni can be a limiting factor for nitrogen fixation by Trichodesmium in oceanic regions (Ho and Hu, 2010).This could be associated with the synthesis of Ni-superoxide dismutase by Trichodesmium to eliminate oxygen radicals during the Mehler reaction, as previously shown in other marine cyanobacteria (Dupont et al., 2008).Only two biological roles for V have been identified in cyanobacteria -as cofactors in V-nitrogenases and V-haloperoxidases (VHPOs; Rehder 2008).To date, no marine cyanobacteria are known to possess V-nitrogenases (Walmsley and Kennedy, 1991;Stal and Zehr, 2008).Though VHPO activity has so far only been observed in eukaryotes (Winter and Moore 2009), several cyanobacterial genomes (e.g.Synechoccus sp CC931, Acaryochloris marina MBIC11017) possess genes both annotated in the IMG database as likely VHPOs and with high sequence homology to characterized eukaryotic VHPOs (http://img.igi.doe.gov).A number of other cyanobacteria, including the published Trichodesmium genome, contain an uncharacterized enzyme annotated as VHPO /acid phosphatase-related with predicted peroxidase function.It is possible that the high V quotas measured in this study are due to usage as a cofactor in this uncharacterized enzyme.Vanadate and phosphate are also chemical analogs, which suggest that V incorporation could happen as part of phosphate uptake (Ray et al., 1993).
The relative abundances of the extracellular metal pools (i.e.Cd>Al>V>Ag>Pb>Co >Fe>Mo>Mn>P>Ni>Cu>Zn) follows the sequence characteristic of scavenged elements in the ocean, suggesting that extracellular metal composition is mediated by process of surface-adsorption (Clegg and Sarmiento, 1989) In order to provide a comparison of the metal concentrations measured in the colonies independent of cell volume, all of the Trichodesmium metal data were normalized to the amount of phosphorous (mol:mol).Averages of intracellular metal: P were 3.5×10 −5 , 4.7×10 −5 , 7.4×10 −4 , 9.9×10 −4 , 3.2×10 −3 , 4.5×10 −3 and 1.5×10 −2 , for Cd, Co, Mn, Cu, Mo, V, Ni and Fe, respectively (Table S1 and Fig. 3).While these intracellular contents probably denote the biological requirements for these metals, others elements like Mn was mostly found on the cell surface (∼74%; Fig. 3), suggestive of a lower cellular demand or lower bioaccumulation rates (Tang and Morel, 2006;Whitfield, 2001).
The range of intracellular metal concentrations measured in the Trichodesmium colonies were in agreement with the metal:P range reported for other phytoplankton species (i.e.diatoms Thalassiosira wissflogii and Ethmodiscus rex and gazellae), grown under laboratory conditions (Table 2), suggesting similar metal requirements among different marine phytoplankton species.The exceptions were Co and V whose concentrations in Trichodesmium (Co, range: 0.01-0.14mmol/mol; V, range: 5.0-11.4mmol/mol) were one and two order of magnitude higher than in T. wissflogii (Co, range: 0.07-0.08mmol/mol; V, range: 0.018-0.02mmol/mol, respectively; Table 2).As previously suggested, metal content of field populations of phytoplankton seems to be more variable than those reported in laboratory culture studies (e.g., Tovar-Sanchez et al., 2006).
We found significant linear correlations between the intracellular and the total fraction for many elements measured in the colonies along the spatial gradient of dissolved metal concentrations (Figure 4a-g).The best linear correlations were for Co (r 2 = 0.69, p < 0.001), Cu (r 2 = 0.58, p < 0.001), V (r 2 = 0.56, p < 0.001), Cd (r 2 = 0.51, p = 0.001) and Mo (r 2 = 0.37, p = 0.008).Despite the significant correlations between intracellular and total metal pools in the field colonies, prediction of the intracellular pool using the total metal content produced inaccurate results.Those correlations accounted, at the most, for 59% of the measured intracellular Cu (even lower for the rest of the metals; Cd, 17%; Co, 33%; Fe, 21%; Mo, 18%; Ni, 47% and V, 33%).With the exception of Introduction

Conclusions References
Tables Figures

Back Close
Full Cu, for which concentrations in the colonies seems to increase as water column levels increased, in general, trace element concentrations measured in the Trichodesmium did not show a clear trend with the dissolved concentrations.These results suggest that the dissolved metal fraction was not a good predictor of the metal trends observed in the colonies (Fig. 4).This is not totally unexpected, as the total dissolved pool includes biologically unavailable metals.We used a Principal Component Analysis (PCA) to try to indirectly determine whether biological processes such as nitrogen fixation, carbon drawdown and biomass production (measured as chlorophyll a) were influenced by trace elements as well as by other physical variables such as salinity, temperature, and mixed layer depth (Fig. 5).
All of the biological data (i.e., nitrogen fixation and biological drawdown of DIC) as well as mixed-layer depth measured during our sampling campaign were previously reported by Subramaniam et al. (2008).In our PCA, 69% of total variance was explained by the first two principal components.The first PC is defined by the positive relationship of biomass (as chlorophyll a) and the biological drawdown of dissolved inorganic carbon (DIC) with the internal content of Fe, Co, Cd, Cu and Mn, in Trichodesmium.The relationship of DIC with Fe and Mn is in agreement with the use of these elements by different enzymes for carbon fixation (Webb et al., 2001;K üper et al., 2008;Tchernov and Lipschultz, 2008).Copper is required by cytochrome oxidase, which has been suggested to be involved in nitrogenase protection in Trichodesmium spp (Bergman et al., 1993).Cobalt is required for the synthesis of cobalamin (vitamin B 12 ), as diazotrophs seem to be major producers of this organic growth factor in oligotrophic waters (Bonnet et al., 2010).Although Cd does not seem to be involved in any enzymatic process to fix carbon in Trichodesmium, labile forms of Cd in seawater have been found complexed to humic material derived from Trichodesmium, suggesting a previous internalization mechanism (Jones et al., 1986).
The PCA also showed that nitrogen fixation during our sampling seems to be influenced by mixed layer depth, instead of intracellular trace elements (PC2).fixation in the subtropical Atlantic.The MLD can increase or decrease primary production by supplying deep nutrients or decreasing the light penetration (Polovina et al., 1995).Most of the dissolved metals also clustered in the PC2 with MLD and salinity, suggesting the influence of the Amazon River and MLD on the dissolved fraction of trace metals.In our area of study, salinity, together with temperature and density, determine the MLD (de Boyer Mont égut et al., 2004).
A negative trend was found between nitrogen fixation with dissolved P and the internal content of Mo, Ni and V in Trichodesmium, suggesting a biological use of these elements during nitrogen fixation.N 2 -fixing diazotrophs may contribute to the low dissolved P concentrations in surface waters in this area, making P the element controlling nitrogen fixation under high-Fe conditions (Sa ñudo- Wilhelmy et al., 2001).Recently, Ho and Hu (2010) reported that, under low environmental Ni concentrations, growth of Trichodesmium may be strongly limited.The minimum intracellular Ni to achieve maximum growth rate has been estimated at about 5.0 mmol to mol of P (Ho and Hu, 2010), which is within the intracellular range quotes obtained in this study (from 0.7 to 11.2 mmol/mol to P; Table S1).On the other hand, while the use of Mo mediated by the enzyme nitrogenase is well known, the biological role of V with the nitrogen metabolism in Trichodesmium remains obscure.
In summary, our results show that dissolved trace elements in the Western Tropical North Atlantic are strongly influenced by the Amazon River.The concentrations of dissolved Cd, Co, Ni and V appear to have a conservative behavior relative to simple mixing of freshwater and seawater, while Mo and Cu behave non-conservatively.Analysis of internal and extracellular metal composition in Trichodesmium colonies revealed positive correlations between the two pools and differences in abundances.While the extracellular fraction results from a scavenging process, the internal content seems to  1. Estimation of metal concentrations in the upper Amazon River Basin.We used the regression equations obtained from metal concentrations measured along a wide salinity range under the influence of the Amazon plume.Values are compared with concentrations reported in the upper Amazon River Basin by Seyler and Boaventura, 2003.This work Amazon Basin (Seyler and Boaventura, 2003) Salinity Full  Full Metal concentrations (Ag, Cd, Co, Cu, Fe, Mn, Mo, Ni, P and V) were determined by ICP-MS in the acid-digests.Intracellular metal levels were determined after washing a sub-sample of the field-collected Trichodesmium colonies with the oxalate reagent Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | For example, P and Fe are well known 6528 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Sa ñudo-Wilhelmy et al. (2001) also reported that MLD was a major factor controlling nitrogen Discussion Paper | Discussion Paper | Discussion Paper | follow biological requirements of the cyanobacteria.With the exception of Cu, metal content in Trichodesmium does not reflect the water column concentration.The PCA analysis indicates that while chlorophyll a concentration and the biological drawdown of dissolved inorganic carbon (DIC) by Trichodesmium are related to the internal content Discussion Paper | Discussion Paper | Discussion Paper | of Fe, Co, Cd, Cu and Mn, nitrogen fixation is limited by the internal concentration of Mo, Ni and V and by dissolved P levels.Discussion Paper | Discussion Paper | Discussion Paper | Table Discussion Paper | Discussion Paper | Discussion Paper |

Table 2 .
Range values of metal concentrations normalized to P Trichodesmium and Diatoms.Trichodesmium 2 T. Wissflogii 3 Ethmodiscus Introduction 1