Introduction

Nitrogen is recognized as the proximate limiting nutrient for primary production in the oceans1. The oceanic nitrogen reservoir is controlled by a balance between fixed nitrogen gains (via dinitrogen -N2- fixation) and losses (denitrification)2. While global nitrogen budget estimations determine that denitrification exceeds N2 fixation considerably3, recent improvements in the 15N2 isotope tracer method used to measure biological N2 fixation have evidenced that formerly published rates could be underestimated by a factor of ~2 to 64,5,6,7,8, and thus could be high enough to balance denitrification on a global basis. However, the variability among N2 fixation rates obtained when using the two different methods (adding 15N2 as a bubble or pre-dissolved in seawater)4,9 can be high7 and at times not significant10,11,12. A mechanistic understanding of which factors determine the degree of discrepancy between the two 15N2 methods is currently lacking. Moreover, marine pelagic N2 fixation had been long attributed to the tropical and subtropical latitudinal bands of the ocean, e.g.13, while other ecological niches such as high latitude waters, oxygen minimum zones and the vast dark realm of the ocean (below the euphotic zone) are now recognized as active N2 fixation sites14,15,16. It is likely that the inclusion of these previously unaccounted for active N2 fixation sites will be more important in equilibrating denitrification and N2 fixation rates than the underestimation of rates due to discrepancies between isotopic tracer methods.

In chronically stratified open ocean regions such as the vast subtropical gyres, primary production depends largely on external fixed nitrogen inputs provided by N2 fixation performed by prokaryotes termed ‘diazotrophs’. Diazotrophic cyanobacteria are photosynthetic prokaryotes (with the exception of the photoheterotrophic Candidatus Atelocyanobacterium thalassa which cannot photosynthesize)17 that thrive in oligotrophic tropical and subtropical waters of the oceans where they provide an important source of fixed nitrogen for other phytoplankton13. Despite being classically regarded as photoautotrophs, some unicellular diazotrophic cyanobacteria like Cyanothece are able to take up dissolved organic matter (DOM) molecules photoheterotrophically18. As well, various filamentous diazotrophic cyanobacteria such as Anabaena bear genes for amino acids transport, which may be used to incorporate amino acids from the in situ DOM pool, or to assimilate amino acids self-produced during diazotrophic growth19.

The filamentous diazotrophic cyanobacterium Trichodesmium is ubiquitous in the tropical and subtropical oceans where it is estimated to contribute 60–80% of global N2 fixation inputs20. Trichodesmium is limited by iron and/or phosphate availability21 which are often scarce in oligotrophic subtropical gyres. The concomitant accumulation of DOM in these oligotrophic gyres22 where Trichodesmium thrives, suggests it could benefit from organic compounds. While most of the marine DOM is composed of refractory molecules that persist in seawater for millennia, labile DOM (degraded within hours or days) accumulates preferentially at the surface ocean as a result of photosynthesis products23. Trichodesmium has been shown to assimilate organic phosphorus compounds such as phosphomonoesters and phosphonates when phosphate is scarce24,25 in an equally efficient manner to phosphate consumption26, although a minimum availability of inorganic nutrients may be needed before Trichodesmium can cleave the carbon-phosphorus bond of phosphonates27. On the other hand, while the uptake of carbon or nitrogen-rich DOM compounds has been studied in cultures of Trichodesmium (e.g. refs 28 and 29), such activity has not been revisited for almost two decades. The extent to which mixotrophic nutrition facilitates the growth and/or N2 fixation in Trichodesmium remains poorly known, particularly for natural colonies. Here we quantify the uptake of carbohydrates and amino acids and their effect on N2 fixation by natural Trichodesmium colonies using nano-scale secondary ion mass spectrometry (nanoSIMS).

Results

Station LDA presented relatively oligotrophic conditions at the surface with inorganic nutrient concentrations below the detection limit (0.02 μM for both nitrate -NO3−- and phosphate -PO43−-; Table S1), but high dissolved organic carbon (DOC; 95.34 ± 2.81 μM) and relatively high chlorophyll a concentrations (0.36 ± 0.05 μg L−1; Table S1) when compared to typical open ocean regional values30. Station LDB was sampled in an elevated chlorophyll a patch (0.83 ± 0.07 μg L−1) and exhibited lower DOC concentrations (70.65 ± 0.09 μM). Bacterial abundance was > 3-fold higher at LDB than at LDA (Table S1).

Bulk N2 fixation rates at LDA were 2.23, 4.61 and 4.10 nmol N L−1 d−1 for the control, carbohydrate and amino acid treatments, respectively. At station LDB, bulk N2 fixation rates were ~9-, 5- and 2-fold higher than at LDA (21.28, 23.44 and 10.79 nmol N L−1 d−1, respectively; Fig. 1a,b). At LDA, the addition of both carbohydrates and amino acids increased bulk N2 fixation but the variability among replicates was high, resulting in non-significant differences (p > 0.05) as observed in previous similar experiments15. No significant enhancement of bulk N2 fixation rates were observed at station LDB for either treatment.

Figure 1
figure 1

Bulk seawater N uptake rates at (a) station LDA, and (b) station LDB (note the different scale ranges on the y-axis). Trichodesmium (c,d) 13C -filled circles- and 15N -open circles- atom % enrichment values, and (e,f) 13C and 15N uptake rates at stations LDA and LDB, respectively. Error bars for C or N uptake rates (a,b,e,f) represent the standard error, while error bars for atom % graphs (c,d) represent the standard deviation of the mean. Straight and dotted lines in (c,d) indicate the average 13C and 15N atom % enrichment, respectively, of time zero samples (average natural abundance). Asterisks indicate statistically significant differences (Mann Whitney test, p < 0.05).

Full size image

At both stations LDA and LDB, nanoSIMS analyses of individual trichomes (Table S2) revealed significant 13C-enrichments (p < 0.0001) by ~1.7-fold relative to the control upon both carbohydrate and amino acid additions (Fig. 1c,d). These additions also significantly enhanced the 15N-enrichment of Trichodesmium by ~1.2-fold at station LDA (both p < 0.0001), but not at station LDB, where a high degree of variability was observed between filaments (Fig. 1c,d). When comparing both stations, we observed that per-trichome carbon and nitrogen uptake rates were ~2- and 5-fold higher at LDB than at LDA (Fig. 1e,f). NanoSIMS example images of 13C and 15N enriched trichomes in control, carbohydrate and amino acid treatments are shown in Fig. 2a–f, respectively. The addition of both carbohydrate or amino acids enhanced per-trichome nitrogen uptake rates at each station, although increases were only statistically significant for amino acid additions (Fig. 1e,f).

Figure 2
figure 2

NanoSIMS trichome 13C/12C ratio images of (a) control, (b) carbohydrate-amended and (c) amino acid-amended samples. The corresponding 15N/14N ratio images are displayed below (d–f).

Full size image

Discussion

We present evidence of carbohydrate and amino acid uptake by natural Trichodesmium colonies in conditions usually regarded as optimal for their autotrophic growth (Table S1). The addition of either carbohydrates or amino acids increased per-trichome N2 fixation rates compared to the control at both LDA and LDB, but only amino acid additions induced statistically significant per-trichome N2 fixation enhancements (Fig. 1e,f). At LDB, the enhancement of per-trichome N2 fixation rates with respect to the control upon the addition of either carbohydrates or amino acids was ~5-fold higher than at LDA, suggesting different DOM degradation patterns at LDB. LDB was located inside a massive chlorophyll patch, which had been drifting eastwards for several months (see chlorophyll a satellite images time lapse, where station LDB is represented by a pink cross; de Verneil, 2015). The persistence of the patch could be maintained by a regular input of inorganic nutrients via wet deposition, which often enhances primary production due to the high nutrient and trace metal content of volcanic ashes in this highly seismic active area of the South Pacific Ocean31. The wet deposition of inorganic nutrients could maintain both photosynthetic and N2 fixation activities while promoting a dynamic production and consumption of DOM. The higher DOM uptake by Trichodesmium at LDB (Fig. 1d,f) suggests that higher in situ DOM availability at LDB compared to LDA, promoted the heterotrophic nutrition of Trichodesmium over growth on inorganic nutrients. DOC standing stocks were lower at LDB than at LDA likely due to a tight coupling between its production and consumption, as suggested by the higher bacterial abundance (Table S1) and production at LDB than at LDA (F. Van Wambeke, personal communication). Alternatively, differences in Trichodesmium DOM uptake between stations LDA and LDB could be influenced by different DOM uptake genetic ability in different Trichodesmium strains32.

Given the high energetic cost of CO2 and N2 fixation in cyanobacteria33,34, the alternative nutrition on DOM is thought to alleviate energy shortages. For example, unicellular cyanobacteria use glycerol as an alternative to CO218, but the use of organic carbon substrates such as carbohydrates by Trichodesmium has seldom been observed and at low rates35. Although the external input of combined nitrogen is thought to preclude N2 fixation in Trichodesmium29,36,37, our results show a significant enhancement of per-trichome N2 fixation rates upon the addition of amino acids, as observed in other sites where heterotrophic diazotrophs predominate, like in mesopelagic waters15,38,39. Amino acids may provide a more readily accessible source of organic carbon than carbohydrates, resulting in a greater enhancement of N2 fixation rates.

DOM utilization likely confers nutritional plasticity to Trichodesmium in oligotrophic environments, reinforcing the obsoleteness of the categorical division of marine microbes into autotrophs or heterotrophs. Although we did not conduct 13C-labeled bicarbonate uptake experiments during this cruise, previous experiments performed in the Southwest Pacific Ocean and in cultures of Trichodesmium IMS101 have shown per trichome bicarbonate uptake rates of ~2–3 × 106 fmol C trichome−1 h−1 40,41, which are in the same order of magnitude as the per trichome carbohydrate uptake rates measured here (~1–9 × 106 fmol C trichome−1 h−1). This suggests that under certain environmental conditions, Trichodesmium may be able to exploit carbon comparably from inorganic and organic carbon sources.

Our results cannot however confirm whether DOM molecules were directly taken up by Trichodesmium, or if they were primarily reduced by epibiont bacteria and then transferred to the trichomes. For example, heterotrophic bacterial epibionts are known to facilitate dissolved organic phosphorus acquisition in Trichodesmium colonies42. Thus, the degree and/or functional diversity of epibiont bacteria colonization among sampling stations could have also influenced DOM uptake rates in our Trichodesmium samples43. We observed bacteria appearing to be attached to trichomes in our samples (Fig. S1), and thus cannot rule out this possibility. Different incubation time span experiments are needed to discern whether DOM passes through bacteria before being taken up by Trichodesmium, or if Trichodesmium assimilates DOM directly. However, such short-term experiments would require a high isotopic enrichment of the source DOM pool, which would likely bias the measured uptake rates.

We present evidence of carbohydrate and amino acid uptake by natural Trichodesmium colonies. Climate change scenarios predict inorganic nutrient limitation and increased DOM retention within the photic zone44, which will likely promote mixotrophy in Trichodesmium. Further studies on Trichodesmium organic versus inorganic nutrient acquisition are thus needed to predict how this important diazotroph will respond to climate alterations.

Methods

We sampled seawater at two stations in the Southwest Pacific (LDA: 19.21°S-164.68°E, LDB: 18.24°S-170.80°W, on 26 February and 15 March 2015, respectively) at depths receiving 50% of surface photosynthetically active radiation (corresponding to 7 and 9 m depth, respectively). The samples were incubated under in situ simulated conditions for 36 h with equimolar quantities of 13C-labeled carbohydrates (sodium pyruvate, sodium acetate and glucose) or amino acids (alanine, leucine and glutamic acid; Sigma-Aldrich, Munich, Germany), added at concentrations of 4 μM C (final concentration for the mix of all three carbohydrates or all three amino acids38). While the real marine DOM pool is molecularly highly complex and mostly refractory, these commercially available compounds were chosen as representative of carbohydrate and small organic acids typically found in marine labile DOM15,38,45. Seawater was distributed into sixteen 4.3 L transparent polycarbonate bottles (Nalgene, Rochester, NY, USA). Four bottles were filtered immediately upon collection (T0), four were amended with the carbohydrate mix, and another four with the amino acids mix. The last four bottles were used as a control without amendments. All bottles were labeled with 6 mL 98.9 atom% 15N2 gas (Cambridge Isotope Laboratories, Tewksbury, MA, USA) to assay N2 fixation simultaneously. Of each quadruplicate set, three bottles were used to estimate bulk N2 fixation rates (expressed as ‘N uptake’) and one bottle was used for nanoSIMS analyses (see Supplementary Information). Mann-Whitney statistical tests were used to test the significance of our results.

Additional Information

How to cite this article: Benavides, M. et al. Dissolved organic matter uptake by Trichodesmium in the Southwest Pacific. Sci. Rep. 7, 41315; doi: 10.1038/srep41315 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.