Nitrogen fixation and transfer in open ocean diatom-cyanobacterial symbioses.

Many diatoms that inhabit low-nutrient waters of the open ocean live in close association with cyanobacteria. Some of these associations are believed to be mutualistic, where N(2)-fixing cyanobacterial symbionts provide N for the diatoms. Rates of N(2) fixation by symbiotic cyanobacteria and the N transfer to their diatom partners were measured using a high-resolution nanometer scale secondary ion mass spectrometry approach in natural populations. Cell-specific rates of N(2) fixation (1.15-71.5 fmol N per cell h(-1)) were similar amongst the symbioses and rapid transfer (within 30 min) of fixed N was also measured. Similar growth rates for the diatoms and their symbionts were determined and the symbiotic growth rates were higher than those estimated for free-living cells. The N(2) fixation rates estimated for Richelia and Calothrix symbionts were 171-420 times higher when the cells were symbiotic compared with the rates estimated for the cells living freely. When combined, the latter two results suggest that the diatom partners influence the growth and metabolism of their cyanobacterial symbionts. We estimated that Richelia fix 81-744% more N than needed for their own growth and up to 97.3% of the fixed N is transferred to the diatom partners. This study provides new information on the mechanisms controlling N input into the open ocean by symbiotic microorganisms, which are widespread and important for oceanic primary production. Further, this is the first demonstration of N transfer from an N(2) fixer to a unicellular partner. These symbioses are important models for molecular regulation and nutrient exchange in symbiotic systems.

Introduction

It is well established that oceanic N2 fixation has a pivotal role in providing ‘fixed' n class="Chemical">nitrogen (N; new production) to surface water communities (Karl , 2002). However, there is a continuing controversy regarding the apparent imbalance in the sources and sinks of N in the global N budget (Codispoti, 1995; Michaels ; Gruber, 2005). Many studies over the last decade have shown the presence and activity of diverse N2-fixing microorganisms (Zehr , 2000; Montoya ), and so it has also been argued that N input from these other N2-fixing microorganisms (that is, unicellular groups) have been underestimated. Furthermore, there is new evidence strongly suggesting a methodological underestimation of oceanic N2 fixation rates by standard isotope tracer experiments (Mohr ). A unique group of open ocean diazotrophs, which have been understudied, including their N2-fixing activities, are the heterocystous cyanobacteria that live symbiotically with other phytoplankton, primarily diatoms.

Richelia intracellularis and n class="Species">Calothrix rhizosoleniae are filamentous heterocystous cyanobacteria that live in presumed symbioses with several diatom genera, including Hemiaulus, Rhizosolenia and Chaetoceros (Figures 1a–c; see also illustrations of diatom ultrastructure in Tomas, 1997). Heterocysts are specialized cells in which N2 fixation is localized (Stewart, 1973), and therefore it is assumed that the cyanobacterial symbionts provide fixed N to the diatom partners. Richelia spp. and Calothrix spp. have morphologically similar filaments, called trichomes, comprised of vegetative cells and a terminal heterocyst. The length, location and number of Richelia and Calothrix trichomes per diatom partner, and phylogeny of the symbionts differ in each symbiosis (Janson ; Foster and Zehr, 2006). There is another less studied symbiosis between the chain-forming pennate diatom, Climacodium frauenfeldianum and unicellular cyanobacteria similar in morphology to the free-living diazotroph, Crocosphaera watsonii (Figure 1d). A recent re-evaluation of the partial 16S rRNA sequence derived from a symbiotic C. frauenfeldianum reported by Carpenter and Janson (2000), found 100% sequence identity to C. watsonii 8501 (personal communication, Shellie Bench). The benefit (or costs) of the symbiotic relationship for either partner has not been clearly characterized in any of the diatom–cyanobacterial symbioses, but a primary interaction has been assumed to be provision of fixed N to the diatom by the potentially N2-fixing symbiont.

Figure 1

Blue light excitation (450–490 nm) images of field collected diatom–cyanobacteria symbioses. The diatom frustules are not easily seen under epi-fluorescence microscopy (except in d), however, the excitation patterns of the cyanobacterial symbionts are clearly different and yellow/orange from their diatom partners (red). (a) Two Hemiaulus membranaceus diatoms with two Richelia intracellularis associated to each diatom. The chlorophyll a within the chloroplast (c) of the diatom fluoresces red, whereas the pigments in the vegetative (v) cells and the terminal heterocyst (h) of Richelia fluoresce yellow–orange. (b) The apical end of a Rhizosolenia clevei diatom with two associated trichomes of R. intracellularis (c) A chain of Chaetoceros spp. diatoms with Calothrix rhizosoleniae attached to the spines (not visible). (d) A chain of Climacodium frauenfeldianum diatoms associated with yellow-fluorescing unicellular cyanobacteria (cyanobionts).

The symbiotic diatoms have been observed in all the major ocean basins (see review, Foster and O'Mullan 2008), where bloom cell densities have been reported (Villareal, 1994; Carpenter ), however, there are few reported N2 (and n class="Chemical">carbon (C)) fixation rate measurements for the symbiotic populations. Of the few reported rate measures, estimates were determined from bulk water assays or cell concentrates (plankton slurries) and therefore, the rate estimates include N2 fixation by other co-existing populations, such as Trichodesmium and unicellular cyanobacteria. Recent methodological advancements have made it possible to directly measure and visualize stable isotopes in individual cells (Römer ; Clode ; Lechene ; Ropa ; Musat ; Finzi-Hart ; Halm ; Ploug ). Using a high-resolution nanometer scale secondary ion mass spectrometry (NanoSIMS; Cameca, Gennevilliers Cedex, France) approach we provide the first experimental evidence of N2 fixation and transfer of fixed N products in field populations of symbiotic diatoms.

Materials and methods

15N2 incubations

Two sets of long-term incubation (0, 24, 48 and 76 h) experiments were run in July 2008 at two locations in the Gulf of California (21 23 °N, 107.05 °W and 24 49 °n class="Chemical">N, 109 00 °W) and one location in the subtropical North Pacific (24 4 °N, 158 27.96 °W). In July 2009, one short-term (0, 30 min, 1, 3 and 12 h) set of experiments was performed at the same location of the subtropical North Pacific. Bulk seawater (SW) was collected from 0 to 10 m using a conductivity temperature depth (CTD) rosette connected to a hydro wire. The 0–10 m depth was selected, as this was the depth where the most symbiotic diatoms were observed with microscopy.

In July 2008, two experimental designs were used, in which in one design, 4.4 l bottles were filled with bulk SW from the CTD, capped and sealed without air bubbles, and subsequently amended with 3 ml of 15N2 (98%+n class="Chemical">15N2, Cambridge isotopes). In the second design, 5 l of bulk SW was gravity filtered, and the >5 μm cell diameter concentrates were incubated with 200 ml filtered SW, capped, sealed as described above, and amended with 150 μl of 15N2. This latter design was used to decrease the time of filtration at the subsequent time points. In July 2009, only 4.4 l bottles were used and treated as described above. All incubations were held at ambient sea surface temperatures by continuously flowing SW and 50% incident surface irradiance.

At the time of injection (time 0), and at the subsequent time points, the entire contents of incubation bottles were filtered onto a membrane filter using a peristaltic pump and fixed overnight at 4 °C in (w/v) 100 μl of 4% paraformaldehyde. The July 2008 long-term experiments used 3 or 10 μm pore size filters and the July 2009 short-term experiments used 3.0 μm pore size filters, which were pre-sputtered with gold (n class="Chemical">Au) and palladium (Pd). Filters were washed three times in 0.1 M phosphate-buffered saline and stored dry at −20 °C.

NanoSIMS analyses

Cells from the July 2008 experiments were re-suspended from the 3.0 μm pore size membrane filters by gentle pippeting, then filtered onto a 3.0 μm pore size Au–n class="Chemical">Pd sputtered membrane filter and washed with 15 ml of Milli-Q water. This step was omitted for the July 2009 filters, as these were filtered directly onto pre-sputtered filters. A 5 mm diameter circle was excised in the area on the filter comprised of a few (that is, >3 symbioses) symbiotic diatoms and/or free-living cells. Epi-fluorescent images of each free-living cell and diatom-containing symbionts were taken before the nanoSIMS analyses. Once dried, the filters were scanned for diatoms containing Richelia/Calothrix and Climacodium cells with Crocosphaera symbionts, and in addition, free-living cells of Richelia, Calothrix and Crocosphaera were also identified using a Zeiss Axioplan (Zeiss, Jena, Germany) epi-fluorescent microscope fitted with blue (450–490 nm) and green (510–560 nm) excitation filters. All cells were identified by shape, cell diameter and excitation patterns. In addition, the taper of the trichome was also used to distinguish the Richelia and Calothrix symbionts. Given the 100% 16S rRNA sequence identity between C. watsonii 8501 and the partial sequences reported previously from symbiotic Climacodium diatoms (Carpenter and Janson, 2000), we assumed that the unicellular symbionts were Crocosphaera.

For each time point of the long-term incubations (0–76 h), two to three different Hemiaulus–n class="Species">Richelia symbioses were analyzed. For the Climacodium and Chaetoceros diatoms containing symbionts, cells were only re-identified at the initial time (time 0) and at 76 h time points. We were not able to find these latter symbioses in the 24 and 48 h time point samples in the long-term experiments of July 2008. Only Hemiaulus–Richelia symbioses, and free-living Calothrix and Richelia (from time points 1 and 12 h, respectively) cells were analyzed in the short-term experiments of July 2009 as these were consistently observed with microscopy. In addition, a large group (19 cells) of Crocosphaera cells were also analyzed from the 12 h time point sample to provide a comparison measure for the Crocosphaera cells living symbiotically with Climacodium. The number of cell replicates for analysis at each time point was limited by the low cell abundances of the symbioses and free-living cells (Richelia, Calothrix and Crocosphaera) at the time of collection. Multiple planes (>50) were recorded to assure a considerable robustness of the single-cell rate measurements.

NanoSIMS analysis was performed using a Cameca n class="Chemical">NanoSIMS 50 l instrument following previously described methods (Musat ). Briefly, samples were rastered with 16 keV Cesium (Cs+) primary ions with current between 1–3 pA. Primary ions were focused into nominal ∼120 nm spot diameter. Mass resolving power in all measurements was >6000. The primary ion beam was used to raster the analyzed area with 256 × 256 pixels over the chosen raster size with dwelling time 1 or 2 ms per pixel. Negative secondary ions were collected simultaneously in electron multiplier detectors.

All scans were corrected for drift of the beam and sample stage after acquisition. Isotope ratio image was created as ratio of a sum of total counts for each pixel over all recorded planes of the investigated isotope and the main isotope. Regions of interest (ROI) around cell structures were defined using the parallel epi-fluorescent image taken before nanoSIMS analyses together with secondary ion images. For each ROI, the ratios 15N/n class="Chemical">14N (for example, inferred from 12C15N/12C14N) were calculated.

Calculations: biovolume, 15N assimilation rates, %N transfer, growth rate, excess N assimilation, global N contribution

Epi-fluorescent images taken before nanoSIMS analyses were used with Axioscope software (Zeiss Axiovision Rel 4.7.2, Zeiss) to estimate heterocyst and unicellular cyanobiont cell diameters, and the apical and trans-apical dimensions of the Hemiaulus diatoms. As the cells were fixed and dried to the filters and the depth (Z plane) was not visible under the microscope, we assumed that the height of the Hemin class="Chemical">aulus, Chaetoceros and Climacodium cells was the same as the diameter of their respective symbiont's heterocyst or unicellular cell diameter.

Measurements of cell diameter, apical and trans-apical axes for the Hemiaulus and Chaetoceros cells were used in a biovolume estimate for an elliptic prism and four cones (23-SL) and a prism on an elliptic base (29-H) as described by Sun and Liu (2003). Similar measures for the Climacodium cells were made in a biovolume estimate for a box plus a prism on a triangle base (30-H; Sun and Liu, 2003). The volume for the Crocosphaera cells (free-living and symbiotic) and the vegetative and heterocyst cells of n class="Species">Richelia and Calothrix was estimated by using the equation for a sphere (Sun and Liu, 2003). C content was calculated for the diatoms and symbionts, using the Strathmann (1967) equations, where the biovolume (V) was used in place of plasma volume. The C content was then used to estimate nitrogen (N) content by assuming a Redfield ratio of 6.6 (Redfield, 1934). As none of the symbiotic diatoms have been brought into pure culture, and at the time of our incubations cell densities were too low for standard elemental analysis, we considered our estimate of the initial N content based on biovolume and Redfield ratio as conservative and a reasonable alternative.

Nitrogen assimilation for individual symbiotic cells (fmol n class="Chemical">N per cell) were estimated by: assimilation=[15Nex × Ncon]/Nsr. The 15Nex is the mean of the 15N/14N ratios of the individual ROIs from the diatoms corrected for by the mean value of the 15N/14N ratios in diatoms of the time zero samples and divided by 100; the Ncon is the N content estimated by V and the Strathmann (1967) equations as described above, and Nsr is the calculated atom percent (AT%) of 15N in the experimental bottle according to Zehr and Montoya (2007). The assimilated N was divided by incubation time to determine cell-specific N2 fixation rates (fmol N per cell h−1).

The percentage of fixed N transferred to the diatom partners was determined by dividing the n class="Chemical">N assimilated as calculated above by the sum of N assimilated into the diatom, vegetative cells and heterocysts and multiplying by 100. Growth rate for the symbionts and diatom partners were estimated by the following, V=1/t (R(F)−R(I)/R(S)−R(I)), where t is time, the R(F) is estimated from mean value of the ROIs (for diatom and symbiont separately) at specific time points, the R(I) is the mean AT% of the ROIs for the diatom or symbiont in the time zero samples, and the R(S) is the calculated AT% of 15N in the experimental bottle according to Zehr and Montoya (2007).

To determine the percent of N fixed by the n class="Species">Richelia that was in excess of their own growth requirement and the percent of the diatom partner's growth supported by Richelia, we assumed growth rates for Hemiaulus and Richelia to be equivalent to that reported by Villareal (1989, 1990). These were 0.77 and 0.67 division per day, respectively, and we converted these to specific growth rate, K′, by K′=division per day × Ln2. These assumed values were also within the range we estimated for growth of both the Richelia and the Hemaiulus, which is described above (Table 1). We estimated the percent of the diatom's growth requirement supported by the Richelia by the following, (assimilation/(Ncon × K′)) × 100, where assimilation is estimated as described above, Ncon is N content estimated from V and Strathmann (1967) equations, and K′ is the assumed growth rate. The percentage of excess N fixed by the Richelia was calculated by the total N assimilated (host, vegetative cells and heterocysts) normalized to incubation time divided by the Ncon of the symbiont multiplied by assumed growth rate (0.67 division per day).

Table 1

Summary of nanoSIMS analyses, cell dimensions and estimates of N2 fixation rates and growth rates

Diatom symbiont	Time (h)	ROIs (n)	AT % (15N/14N)	Biovolume (μm3)	N2 fix rate (fmol N per cell h−1)a	Growth rate (division per day)
Hemiaulus		3	0.3789	4.22 × 103		0.22
Heterocyst 1		1	0.3726	164	21.0
Vegetative	0.5	1	0.3790	8.72		0.22
Heterocyst 2		1	0.3812	37.4
Vegetative		1	0.3806	16.4		0.23
Hemiaulus		3	0.3651	3.49 × 103		0.06
Heterocyst	1	1	0.3761	102	5.88
Vegetative		2	0.3607	8.32		0.05
Hemiaulus		4	0.3759	1.87 × 103		0.10
Heterocyst	1	1	0.3764	85.1	8.94
Vegetative		2	0.3737	15.5		0.09
Hemiaulus			0.3595	1.91 × 103		0.03
Heterocyst	1	1	0.3623	34.3	2.90
Vegetative		2	0.3657	8.62		0.06
Hemiaulus		6	0.4378	4.63 × 103		0.11
Heterocyst		1	0.4238	50.0	9.45
Vegetative	3	3	0.4362	6.0
Heterocyst		1	0.4508	45.6		0.10
Hemiaulus		8	0.4576	4.23 × 103		0.13
Heterocyst 1	3	1	0.4104	72.3	10.6
Vegetative		2	0.4792	29.6		0.16
Heterocyst 2		1	0.4706	44.9
Hemiaulus		6	0.4523	4.32 × 103		0.12
Heterocyst 1		1	0.4118	41.1
Vegetative	3	1	0.4951	11.7	10.3	0.10
Heterocyst 2		1	0.4596	55.0
Vegetative		1	0.4485	13.4		0.11
Hemiaulus	12	5	0.4739	1.56 × 103		0.04
Heterocyst		1	0.4572	13.1	1.41
Hemiaulus		5	0.4519	1.49 × 103		0.04
Heterocyst	12	1	0.4826	10.4	1.15
Vegetative		2	0.4748	5.77		0.04
FL Calothrix
Heterocyst	1	1	0.3570	35.8	0.17	0.06
Vgetative		6	0.3679	28.5
FL Richelia
Heterocyst	12	1	0.4575	44.5	0.12
Vegetative		6	0.7415	45.1		0.11
FL Crocosphaera	12	19	0.3561–0.4519	65–2.66 × 102	0.02–2.39	0.001–0.15
Hemiaulus		3	0.4032	2.39 × 104
Heterocyst 1	0	1	0.3632	9.10 × 102
Heterocyst 2		1	0.3951	3.17 × 102
Hemiaulus		3	0.3494	2.84 × 104
Heterocyst 1		1	0.3689	6.84 × 102
Heterocyst 2	0	1	0.3743	4.37 × 102
vegetative		1	0.3617	2.34 × 102
Hemiaulus		4	4.4440	4.41 × 103		0.43
Heterocyst	24	1	1.7900	75.3	47.8
Vegetative		2	4.4600	18.9		0.43
Hemiaulus		3	4.6600	3.12 × 103		0.45
Heterocyst	24	1	2.6810	71.0	38.7
Vegetative		3	4.7430	27.4		0.46
Hemiaulus		6	4.5960	4.31 × 103		0.45
Heterocyst 1		1	4.7068	38.7	48.7
Vegetative	24	1	4.7815	7.37		0.47
Heterocyst 2		1	4.9327	48.6
Vegetative		2	4.9693	28.8		0.49
Hemiaulus		4	3.1628	2.12 × 103		0.15
heterocyst 1		1	3.1141	39.1	9.43
vegetative	48	2	3.8720	41.6		0.18
heterocyst 2		1	3.9909	35.8
Hemiaulus		3	3.0170	4.30 × 103		0.14
Heterocyst	48	1	3.4009	41.9	15.3
Vegetative		4	3.5284	23.4		0.17
Hemiaulus		3	3.0506	4.68 × 103		0.14
Heterocyst	48	1	3.6647	12.1 × 101	16.5
Vegetative		3	3.5281	16.6		0.33
Hemiaulus		4	9.7529	4.40 × 103		0.48
Heterocyst 1		1	9.8114	74.0	50.4
Vegetative	76	1	10.655	14.0		0.48
Heterocyst 2		1	11.652	68.9
Vegetative		1	11.891	33.2		0.59
Hemiaulus		3	9.6706	4.33 × 103		0.48
Heterocyst	76	1	3.1086	94.8	49.2
Vegetative		1	9.0384	41.6		0.44
Chaetoceros		12	0.3593	nm
Heterocyst 1		1	0.3056	11.9
Vegetative	0	13	0.3339	29.6
Heterocyst 2		1	0.3417	15.6
Vegetative		5	0.3398	44.0
Chaetoceros		3	7.6500	2.17 × 104		0.38
Heterocyst 1		1	4.3552	91.3	71.5
Vegetative	76	3	7.0634	67.0		0.35
Heterocyst 2		1	6.5394	70.9
Vegetative		10	6.9387	66.7		0.32
Climacodium	0	2	0.3797	nm
Crocosphaera		4	0.3710	55.9
Climacodium	76	11	3.1657	1.56 × 105	6.03	0.37
Crocosphaera		28	5.3341	65.4	1.62	0.26

Abbreviations: AT%, atom percent; FL, free living (non-symbiotic); nanoSIMS, nanometer scale secondary ion mass spectrometry; nm, not measured; ROIs, regions of interests.

N2 fixation rates were normalized by number of N2-fixing cell.

To estimate the contribution of the various symbiotic diatoms to basin scale N2 assimilation, we used non-bloom cell abundances reported for the Atlantic (39 cells l−1; Carpenter ) and the Pacific (80 cells l−1, Venrick 1974; Mague ), the range in n class="Chemical">N2 assimilation determined from our nanoSIMS analyses of all the symbiotic cells, a domain of 17.8 and 27.8 km2 × 106 for the Atlantic and Pacific Oceans, respectively (Gruber and Sarmiento, 1997; Mahaffey ; Carpenter and Capone, 2008), and 365 days of activity (12 h per day) to a depth of 10 m.

Results and discussion

Bulk SW samples from two regions of the Pacific Ocean were incubated with 15N2 to trace and quantify the n class="Chemical">N2 fixation by cyanobacterial symbionts of various diatom genera at the cellular level. As diatoms are not able to acquire N from N2, they rely on extracellular dissolved fixed inorganic nitrogen pools (nitrate and ammonium), which in the open ocean surface waters are present at extremely low concentrations. Thus, the diatoms housing symbiotic diazotrophs have a distinct advantage if they can acquire the N2 fixed by their cyanobacterial partners. The nanoSIMS approach made it possible to visualize and quantify the 15N fixed in the symbionts and transferred to the diatom cells (Figures 2 and 3). In addition, the cyanobacterial partners were fully supporting the diatom N requirements for growth, as we always observed equal or higher enrichment in the diatoms than in the vegetative cells.

Figure 2

The images of 15N/14N ratios are shown for symbiont-containing Hemiaulus. The 15N/14N ratio is shown for Hemiaulus–Richelia symbioses at time 0 (a), 30 min (b) and 48 h (c). Inset panels a–c are the epi-fluorescent images taken before the nanoSIMS analyses. The numbers and markings in the figure define regions of interest, which were used for calculating 15N/14N ratios. Scale bars are 5 μm.

Figure 3

The images of 15N/14N ratios are shown for Chaetoceros-Calothrix and the Climacodium-cyanobiont symbioses. The 15N/14N ratio is shown in A and C for Chaetoceros–Calothrix and the Climacodium–Crocosphaera symbioses, respectively. Note the ‘hotspots' (white arrows) of enrichment within the individual cyanobiont (Crocosphaera) cells of Climacodium (c). The corresponding total ion content images images for the same symbioses in a and c are shown in b and d, respectively. The numbers and markings in the Figure define regions of interest, which were used for calculating 15N/14N ratios. Scale bars are 5 μm.

In the initial set of experiments, the 15N enrichment pattern within a chain of Hemin class="Chemical">aulus cells clearly mirrored the location of two associated Richelia trichomes imaged before the analyses with epifluorescence microscopy (Figure 2), showing that the cyanobacteria fixed N. In addition, in areas identified as the diatom chloroplasts (that is, excited red in the epi-fluorescence image) were also enriched, suggesting that N was transferred from symbiont to diatom partner. Similarly, substantial 15N labeling was observed in the Calothrix heterocyst and vegetative cells and also in the Chaetoceros diatoms, suggesting that N was transferred along the trichome of the symbiont, and in addition, across the cell membrane of the diatom partner (Figure 3a). The 15N enrichment was most obvious in <1 μm diameter ‘hotspots' inside the Crocosphaera cells found associated with the Climacodium diatoms (Figure 3c). In the Chaetoceros and Climacodium symbioses N2 fixation had not been previously demonstrated. In general, the transfer of fixed N to the diatom partners in all of the diatom–cyanobacterial symbioses had previously been assumed and had never been demonstrated. The nanoSIMS results directly prove that the N2-fixing cyanobionts provide fixed N to their diatom partners.

The transfer and incorporation of N from the cyanobacteria to the diatom cells was faster than the length of the sampling period in the initial set of experiments (24–76 h). Diatoms, and other cells in oligotrophic oceans are often assumed to grow slowly due to low ambient nutrient concentrations, and thus we anticipated that the symbiont would only transfer a fraction of the n class="Chemical">N2 fixed to the diatom partner within the incubation. However, in our long-term experiments, the enrichment was high after 24 h (Figure 4a) and we were not able to estimate the N transfer over the shorter time interval. In our subsequent short-term experiments, we observed elevated 15N enrichment in Hemiaulus cells incubated for as short as 30 min and as the range of enrichment was similar in the cells measured after 3 and 12 h incubations, it appears that N transfer saturated by 3 h (Figure 4b). The near saturation within 3 h was much faster than anticipated, and means that measured fixation rates may be underestimated. We expected that the transfer would be slow, as the symbionts reside external to the diatom cell membrane. The Richelia symbionts reside between the frustule and the cell membrane (plasmalemma) of the Hemiaulus diatom cell wall (Janson ), whereas the Calothrix symbionts are extracellular on the Chaetoceros diatom (Norris, 1961), and the cyanobiont (Crocosphaera) location is unknown in the Climacodium symbioses. The N transfer we observed along the trichomes of symbiotic and free-living Richelia and Calothrix was not unexpected, as rapid transfer has already been reported for lab cultures of free-living filamentous heterocystous cyanobacteria (Wolk ; Ropa ; Ploug ). Unique to the results presented here, is that the N transferred along the symbiotic trichome (heterocyst to vegetative cells) is also transferred across the diatom cell membrane.

Figure 4

Summary of results from nanoSIMS analysis. (a) The atom % of the 15N/14N ratios for ROIs of individual Hemiaulus–Richelia are shown as a function of incubation time in both long and short-term experiments. (b) The atom % of the 15N/14N ratios for ROIs of individual Hemiaulus–Richelia symbioses from the short-term experiments and represents the values within the red box shown in a. Note that the Richelia enrichment values are given as ratios estimated in the heterocyst and the vegetative cells. The dashed red line indicates the value for a co-occurring non-symbiotic diatom.

In terrestrial systems, rapid nutrient transfer is facilitated by the intracellular location of the symbionts within specialized tissues or organs of multicellular hosts, which are connected to the host tissues via vesicles (Rai ). Nitrogen should be easily transferred from intracellular symbionts, but it is unknown how n class="Chemical">N transfer is facilitated in these planktonic diatom symbioses. The orientation of the Calothrix heterocyst attached to the intercalary spaces of their Chaetocoeros partners may facilitate the transfer of N to the diatom while still exposing the trichome to the surrounding environment (for light and to acquire other nutrients). The efficiencies of nutrient exchange are poorly resolved in most symbioses (Rai ), and the mechanisms in these simple unicellular symbioses may be very different from symbiotic systems with multicellular hosts.

As the 15N labeling could be measured on entire individual cells (symbioses), cell-specific rates of n class="Chemical">N2 fixation could be calculated for the natural populations (Table 1). Using the ratios obtained from the nanoSIMS analyses, we obtained remarkably similar rates of N2 assimilation (range, 1.15–71.5 fmol N per cell h−1) amongst all the symbioses, including the unicellular Crocosphaera symbionts of Climacodium (1.62 fmol N per cell h−1). Our rates were also comparable with the cellular rates previously estimated from bulk N2-fixation rates for a large expansive mono-specific bloom of Richelia associated with Hemiaulus hauckii in the North Atlantic Ocean (50 fmol N per cell h−1; Carpenter ). This result is surprising given that the symbioses differ in diatom association (that is, Hemiaulus, Chaetoceros and Climacodium), diatom cell size (1.49 × 103 to 1.56 × 105 μm3) collection (different ocean basins and during different years), incubation time (30 min–76 h) and cell densities (bloom versus background abundances).

The 15N/n class="Chemical">14N ratios determined on the individual symbiotic cells by nanoSIMS were also used to estimate growth of the symbionts and the diatoms (see Materials and methods). We determined remarkably similar growth rates for all three symbiotic cell types and their respective diatom partners. The estimates of growth rate for the field collected Hemiaulus (0.04–0.48 division per day) were similar to those (0.77 division per day) reported for laboratory cultures of another symbiotic diatom, Rhizosolenia clevi (Villareal, 1990), which associates with a closely related Richelia strain of Hemiaulus (Janson ; Foster and Zehr, 2006). It should be noted that R. clevi are much larger in cell length than the Hemiaulus reported here. The other two symbionts, including the unicellular Crocosphaera cells associated with Climacodium, were also similar (0.38–0.51 per division) to the growth rates of Richelia, which was equally surprising given the same reasons aforementioned and additionally, differing cell types of the symbionts (that is, unicellular and heterocystous).

In several of the symbioses (14 of 19) analyzed with nanoSIMS, the estimated growth rates for symbiont and diatom partners were so similar that a synchronous division would be expected (Table 1). In addition, our results indicate that the co-occurring free-living Richelia and n class="Species">Calothrix cells had substantially reduced growth rate compared with the same symbiont cell types living in association with the diatoms (Table 1). The average rate of N2 fixation (20.4 and 71.5 fmol N per cell h−1) for the symbiotic heterocystous cells (Richelia and Calothrix, respectively) is 170–420 times higher than the N2 fixation rate (0.12 and 0.17 fmol N per cell h−1, respectively) estimated for free-living Richelia and Calothrix cells (Table 1). In terrestrial symbiotic systems, cyanobionts undergo several structural–functional changes, including modifications to their growth rates and metabolism, which is often coordinated by the host to maximize nutrient transfer and balanced growth amongst the partners (Rai ). A similar scenario may exist in these marine symbioses as a means to provide sufficient N for both partners.

As we demonstrated that N was transferred, and growth and n class="Chemical">N2 fixation rates were apparently accelerated under symbiotic conditions, we were curious whether the symbionts were fixing more N than required for their growth (see Materials and methods). Assuming similar growth rates for the Hemiaulus and the Richelia, as was observed in our experiments, we estimate that Richelia symbionts fix between 71 and 651% more N than required for their own growth (see Materials and methods). Other cyanobacteria, that is, free-living Trichodesmium (Mulholland ), also fix more N than required for their own growth, but unique to the symbioses studied here, is that up to 97.3% of the total fixed N is transferred to their diatom partners and not assimilated by the symbionts themselves nor simply released to the environment. Given that N2 fixation is energetically expensive and highly regulated (Postgate, 1972), we suspect that the diatoms may influence the N metabolism of the symbionts.

Determining the contribution of these fragile associations to global N (and C) has been difficult to predict. We used the cell-specific rates of n class="Chemical">N2 fixation determined by nanoSIMS analysis to estimate basin scale N2 assimilation attributed to the symbioses (see details in Materials and methods). We estimated that the symbiotic diatom populations (0.19 and 0.62 Tmol N per year in Atlantic and Pacific Oceans, respectively) could be an equally important source of new N as the free-living colonial diazotroph, Trichodesmium (0.36–0.71 Tmol N per year; Capone et al., 2008), which is usually considered largely responsible for N2 fixation in the open ocean. The densities of the symbioses are not easily detected and most data on the symbioses are reported during bloom densities, and as we are unable to measure the N release from the symbioses, our estimate is likely an underestimate. Moreover, recent evidence demonstrates that N2 fixation will be underestimated when the 15N2 tracer is introduced as a bubble (Mohr ). Although conservative, our calculation reveals the importance of an often un-estimated pool of N, and indicates that diatom symbioses should be included in global N models.