The physiological cost of diazotrophy for Trichodesmium erythraeum IMS101

Trichodesmium plays a significant role in the oligotrophic oceans, fixing nitrogen in an area corresponding to half of the Earth’s surface, representing up to 50% of new production in some oligotrophic tropical and subtropical oceans. Whilst Trichodesmium blooms at the surface exhibit a strong dependence on diazotrophy, colonies at depth or at the surface after a mixing event could be utilising additional N-sources. We conducted experiments to establish how acclimation to varying N-sources affects the growth, elemental composition, light absorption coefficient, N2 fixation, PSII electron transport rate and the relationship between net and gross photosynthetic O2 exchange in T. erythraeum IMS101. To do this, cultures were acclimated to growth medium containing NH4+ and NO3- (replete concentrations) or N2 only (diazotrophic control). The light dependencies of O2 evolution and O2 uptake were measured using membrane inlet mass spectrometry (MIMS), while PSII electron transport rates were measured from fluorescence light curves (FLCs). We found that at a saturating light intensity, Trichodesmium growth was ~ 10% and 13% lower when grown on N2 than with NH4+ and NO3-, respectively. Oxygen uptake increased linearly with net photosynthesis across all light intensities ranging from darkness to 1100 μmol photons m-2 s-1. The maximum rates and initial slopes of light response curves for C-specific gross and net photosynthesis and the slope of the relationship between gross and net photosynthesis increased significantly under non-diazotrophic conditions. We attribute these observations to a reduced expenditure of reductant and ATP for nitrogenase activity under non-diazotrophic conditions which allows NADPH and ATP to be re-directed to CO2 fixation and/or biosynthesis. The energy and reductant conserved through utilising additional N-sources could enhance Trichodesmium’s productivity and growth and have major implications for its role in ocean C and N cycles.


Introduction
In marine ecosystems, phytoplankton primary production is often limited by the bioavailability of fixed N [1][2][3], where N-sources (e.g.NO 3 -, NO 2 NO 3 -concentrations higher at greater depth [5][6][7].Whilst areas of upwelling transport NO 3 into the euphotic zone, there are vast regions of the oligotrophic open oceans that are dependent on the input of new N from N 2 -fixing cyanobacteria.Among the most important marine diazotrophs are Trichodesmium sp., which can form extensive surface blooms in the tropical and subtropical oceans [8][9][10][11][12].Previous studies have highlighted Trichodesmium's capacity to assimilate various forms of combined N-sources [13][14][15][16][17].It is commonly assumed that Trichodesmium obtains most of its nitrogen quota from N 2 fixation, however field-based measurements of N 2 fixation show wide temporal and spatial variability [18].The causes of this variability remain unclear, but environmental factors such as the availability of combined nitrogen may be a contributing factor.

Diazotrophy
Diazotrophic cyanobacteria are able to meet their daily nitrogen quota by fixing dinitrogen (N 2 ).
While N 2 fixation is an extremely energy demanding process, Trichodesmium incurs additional costs related to the protection of nitrogenase from the irreversible inhibition of photosynthetically evolved O 2 [9,19,20].The separation of O 2 evolution and N 2 fixation is regulated over a diurnal cycle of N 2 fixation and photosynthesis [21], involving daily synthesis and degradation of nitrogenase [22,23] and alternation of photosynthetic activity states [24].Temporal separation occurs over short timescales, where peak rates of photosynthesis (~10 am) and N 2 (~12 pm) fixation vary over a diel period.Spatial separation occurs via diazocytes, which are reversibly specialised cells for nitrogen fixation [25,26].Diazocytes contain the necessary proteins to perform photosynthetic CO 2 fixation and N 2 fixation.However, it has been suggested that when fixing N 2 , cells increase cyclic electron transport around PSI to enhance ATP synthesis [21,24], thus allowing the cells to meet the energetic demands of N 2 fixation (Eq 1).

Uptake of additional N-sources
Like other facultative diazotrophic cyanobacteria spp., Trichodesmium can exploit other forms of nitrogen including NH 4 + , NO 3 -, urea and amino acids [16,27].These N compounds are transported into the cell via permeases, metabolised to NH 4 + and then incorporated into carbon skeletons through the glutamine synthetase (GS) and glutamine 2-oxoglutarate aminotransferase (GOGAT) pathways.This process is mediated by nitrate reductase (Eq 2) and nitrite reductase (Eq 3).
For cyanobacteria, nitrate reductase is located in the cytosol and uses NADPH to catalyse the transfer of two electrons.The NO 2 -formed by nitrate reductase is further reduced to NH 4 + via the transfer of six electrons.Thus, the reduction of NO 3 -to NH 4 + can be expressed as; Amino acids are synthesised from ammonia (NH 3 ) via the GS-GOGAT pathway.The initial GS pathway requires ATP and glutamate as a substrate; where glutamine is subsequently transformed to 2-oxoglutarate and reduced using NADPH, forming two moles of glutamate.
Thus, for every mole of glutamate produced, one mole each of NH 3 , NADPH, ATP and 2-oxoglutarate are required.Additionally, ATP is required for the active transport of inorganic NH 4 + or NO 3 -into the cell [28].Different N-sources require different amounts of energy and reductant and as such can be ordered into a hierarchy of energy requirements; where diazotrophy requires the highest investment of electrons and ATP, followed by NO 3 -, NO 2 -and then NH 3 .

Utilising additional N-sources
Global warming is increasing sea surface temperatures (SSTs) which is enhancing water stratification and decreasing vertical mixing [29], potentially increasing the area of N-limited oceans.Whilst detrimental to many phytoplankton, a reduced flux of NO 3 -into the upper mixed layer will increase the competitive advantage of diazotrophs for other limiting nutrients (i.e.Fe or P).Trichodesmium colonies have been observed migrating to the nutricline [30,31] to facilitate the luxury uptake of polyphosphates before returning to the surface.Whilst at these depths, cells are exposed to NO 3 -concentrations greater than those at the surface.As such, Trichodesmium colonies may be assimilating and storing (i.e.cyanophycin granules) more combined N than the blooms frequently measured on the surface [32].This could have major implications for growth rates, primary productivity and biogeochemical cycles [33].
Our approach comprises a systematic experiment where T. erythraeum IMS101 was grown over long durations, at three N-source treatments, with controlled and well-defined growth conditions, ensuring fully acclimated, balanced growth had been achieved.Our aims were to assess the response of T. erythraeum IMS101 growth, light dependency of gross and net O 2 photosynthesis, PSII electron transport rates and elemental composition to different Nsources; investigating the physiological cost of performing diazotrophy.

Experimental setup
Cultures were acclimated to the CO 2 and light intensity for ~4 months (~60 generations) under diazotrophic conditions before the addition of NH 4 + or NO 3 -.Cultures were gradually enriched over a 2/3-week period by increasing the dilution ratio of YBCII media containing NH 4 + or NO 3 -(100 μM).
T. erythraeum IMS101 was grown using YBCII medium [34] at 1.5 L volumes in 2 L pyrex bottles that were acid-washed and autoclaved prior to culturing.Daily growth rates were quantified from changes in baseline fluorescence (F o ) measured between 09:00 to 10:30 on dark-adapted cultures (20 minutes) using a FRRfII FastAct Fluorometer System (Chelsea Technologies Group Ltd, UK).Cultures were regarded as fully acclimated and in balanced growth when both the slope of the linear regression of ln F o versus time and the ratio of live cell to acetone extracted (method detailed below) baseline fluorescence (F o ) were constant following every dilution with fresh YBCII medium.Cultures were kept at the upper section of the exponential growth phase through periodic dilution with new growth media at 3-5 day intervals.Illumination was provided side-on by fluorescent tubes (Sylvania Luxline Plus FHQ49/T5/840).Cultures were constantly mixed using magnetic PTFE stirrer bars and aerated with a filtered (0.2μm pore) air mixture at a rate of ~200 mL s -1 .The CO 2 concentration was regulated (± 2 μatm) by mass flow controllers (Bronkhorst, Newmarket, UK).CO 2 -free air was supplied by an oil free compressor (Bambi Air, UK) via a soda lime gas-tight column which was mixed with a 10% CO 2 in-air mixture from a gas cylinder (BOC Industrial Gases, UK).The CO 2 concentration was continuously monitored and recorded by an infra-red gas analyser (Li-Cor Li-820, Nebraska USA), calibrated weekly by a standard gas (BOC Industrial Gases).
Throughout all culturing, the inorganic carbon chemistry (S1 File) and dissolved inorganic NH 4 + and NO 3 -concentrations (S2 File) were determined prior to diluting with fresh media.Samples for elemental composition, photosynthesis-light response curves, fluorescence light curves (FLC), in vivo light absorption and acetylene reduction assays were collected at the same time of day, approximately 4 and 6 hours into the photo-phase of the L:D cycle.

Measuring O 2 exchange by membrane inlet mass spectrometry (MIMS)
Light dependent rates of O 2 production and consumption were measured with a membrane inlet mass spectrometer (MIMS), using an 18 O 2 technique modified from McKew et al. [35] (S3 File).MIMS measurements consisted of three biological replicates per treatment (S4 File).
Chlorophyll a concentrations at the point of sampling ranged from 80 to 245 μg Chla L -1 .Changes in 16 O 2 and 18 O 2 and thus O 2 consumption (U o ) and O 2 evolution (E o ) were calculated using the following equations [36]; where U o is the rate of O 2 consumption calculated from the decrease of 18  Photosynthesis-light (P-E) curves for gross (E 0 Chl(C) ) and net photosynthesis (P net Chl(C) = E 0 Chl(C) -U 0 Chl(C) ) were fitted to the equations from Platt and Jassby [37];

Measuring nitrogenase activity by acetylene reduction
Acetylene reduction rates were measured using gas chromatography (ATI Unicam 610 series).Gaseous samples were injected into the GC column head (60 ˚C), carried via N 2 gas through a Porapak N column (100 ˚C) to a flame ionising detector (100 ˚C).Peak areas of acetylene and ethylene were quantified by an integrated chromatograph data acquisition unit (Shimadzu C-R8A Integrator) and were converted into concentrations via an acetylene and ethylene standard curve performed with standard gases (Scientific and Technical Gases Ltd., UK).Triplicate 6 mL samples of each biological replicate culture were placed into 12 mL exetainer, screw capped glass vials (Labco Ltd, UK).Exactly 1.2 mL of the headspace was removed and replaced with a 1.2 mL sample of acetylene (BOC Industrial Gases, UK) (headspace = 20% acetylene).The vials were gently inverted for 1 minute before 250 μL of headspace was injected into the GC column for an initial measurement of acetylene and ethylene concentrations (T 0 ).Vials were incubated at 26 ˚C and 400 μmol photons m -2 s -1 in an aluminium temperature block and were gently inverted every 10 minutes to prevent trichomes from settling on the bottom or aggregating at the meniscus.After 1 hour, a second 250 μL gaseous headspace was injected into the GC column for the post-incubation measurement (T 1 ).Temperature and pressure was measured during each set of measurements and accounted for in the calculations.The rate of ethylene production was calculated with the assumption that the concentrations of acetylene and ethylene within the media were always in equilibrium to those in the headspace; where (ΔC 2 H 2 ) is the ethylene production rate (μmol C 2 H 4 h -1 ), C 2 H 2(T0) and C 2 H 2 (T1) are the ethylene concentrations in the headspace at the start (T 0 ) and end (T 1 ) of the incubation, V (I) is the volume of gaseous sample injected into the GC column (L -1 ) and t is the incubation time (min).N 2 fixation rates were calculated to a Chla (μmol N 2 (mg Chla) -1 h -1 ) and total carbon (μmol N 2 (mg C) -1 h -1 ) basis; where ΔC 2 H 2 (μmol h -1 ) is divided by the Chla or total carbon concentration (mg) and multiplied by 0.25 under the assumption that reduction of four moles of acetylene is equivalent to reduction of one mole of dinitrogen.

Fluorescence light curves (FLCs)
A 2 mL sample of each replicate culture was used to measure a fluorescence light curve (FLC) [38].The FLCs were measured with a FRRfII FastAct Fluorometer System, using a white LED actinic light source (Chelsea Technologies Group Ltd, UK).Each FLC lasted 1 hour; comprising 12 light steps which ranged from 10 to 1600 μmol photon m -2 s -1 , each lasting 5 minutes in duration.The FLCs provided measurements of the light absorption cross-section of PSII photochemistry (σ PII ´), the average time constant for the re-opening of a closed PSII reaction centre (τ f ´) and the operating efficiency of PSII photochemistry (F q ´/F m ´); where F m ´is the maximum fluorescence in the light-adapted state and F´is the steady-state fluorescence at any point.
Photosystem II (PSII) electron transport rates were normalised to a Chla (mol e -(g Chla) -1 h -1 ) and total carbon (mol e -(g C) -1 h -1 ) basis; where F q ´/F m ´is the operating efficiency of PSII photochemistry; E is the light intensity (mol photons m -2 s -1 ), a Chl(C) is the Chla-specific (C-specific) effective light absorption (m 2 g -1 Chla and m 2 g -1 C, respectively), FAQ PII is the fraction of absorbed photons directed to PSII, which was set to 0.5 [39], with the assumption that the quantum yield of electron transport of one trapped photon within a reaction centre is equal to 1 [40]; 3600 converts seconds to hours and SCF is a spectral correction factor of 1.194, which converts electron transport rates to the culturing LED spectrum (S1 Fig).
The realised maximum PSII electron transport rate in the presence of photoinhibition (ETR m ), light intensity at which ETR is maximal (E opt ), the light-saturation parameter (E k ) and the light inhibition parameter (E p ) were calculated from the fitted parameters as follows: The ratio of PSII electron transport to gross O 2 evolution (E 0 ) under light-limitation (F eα ) and light-saturation (F em ) were calculated as follow;

Cellular elemental composition and light absorption
Samples for determining particulate organic carbon (POC), nitrogen (PN) and phosphorus (PP) (S5 File), chlorophyll a (S6 File) and in vivo light absorption (S7 File) were collected with each MIMS measurement, with each sample being a biological replicate.
The Chla-specific light absorption coefficient was modelled as the sum of the contribution of all pigments; a Chl mod ðlÞ ¼ where a Chl mod is the modelled in vivo light absorption at a specific wavelength (λ = 400-700 nm); β i is the contribution of each pigment to a Chl mod and a i is the pigment-specific spectral absorption coefficient of pigment i, in m 2 (g pigment i) -1 .
The modelled in vivo light absorption spectra (a Chl mod (λ)) was optimised to the measured spectra between 400 and 700 nm using a reduced sum of squares method (Sigmaplot 11.0).If a zero value was returned for a β i parameter, that pigment was removed from the model and the curve fit reapplied.

Inorganic C-chemistry, growth rate and cell composition
Balanced growth of T. erythraeum IMS101 was 0.34 d -1 when grown on N 2 , increasing by 10% and 13% when grown in the presence of NH 4 + and NO 3 -, respectively (Table 1).Particulate C: N, C:P and N:P ratios were all influenced by the presence of additional N-sources.When compared to the N 2 treatment, C:N decreased by 36% and 43% for the NH 4 + and NO 3 -treatments, respectively.Ratios of C:P and N:P were comparable between NH 4 + and NO 3 -treatments, but were significantly lower (~60% and 35%, respectively) compared to the N 2 treatment (Table 1).Ratios of Chla:C were 80% and 67% higher for the NH 4 + and NO 3 -treatments than for the N 2 treatment, while Chla:N was not significantly different between treatments (Table 1).Carbon and Chla-specific N 2 fixation rates were highest for the N 2 treatment, decreasing significantly by 84% and 80% (Chla-specific) and 73% and 68% (C-specific) for the NH 4 + and NO 3 -treatments, respectively (Table 1).
The inorganic carbon concentration, pH and alkalinity (A T ) did not vary significantly amongst N-source treatments.Overall, CO 2 drawdown ranged between 78 to 92 μatm from the target concentration (i.e.380 μatm) for all N-source treatments (Table 2) and exhibited little variability over a diurnal cycle (S3 Fig) .Inorganic N concentrations were > 1 μM for the N 2 treatment and were ~8 μM for the NH 4 + and NO 3 -treatments at the point of dilution (Table 2).

Light absorption
The effective light absorption coefficients were not significantly different between N-source treatments, nor were the modelled absorption coefficients significantly different to the measured coefficients; with modelled coefficients being only 1 to 3% higher across all N-source treatments (Table 3).

Light-dependence of O 2 exchange
The C-specific maximum rate (E 0m

C
) and initial slope (α g C ) of light-dependent gross photosynthesis increased with additional N-sources (i.e.NH 4 + and NO 3 - ) and was highest for the NH 4 + treatment relative to the N 2 treatment (Table 4).There were also significant effects of additional N-sources on the Chla-specific maximum rate (E 0m

Chl
) and initial slope of lightdependent gross photosynthesis (α g Chl ) (S1 ) increasing significantly by 86% and 100% respectively, relative to the N 2 treatment (Table 4).The light saturation parameter (E k = P netm C /α g C ) for net O 2 evolution did not vary significantly between N-source treatments (Table 4).
The relationship between net and gross O 2 evolution was linear (Fig 2D -2F), with the slope increasing by approximately 40% when cultured in the presence of an additional N-source (Table 4).This linear relationship suggests that light-dependent O 2 consumption (U 0 C ) was a ) and was independent of light intensity for all N-source treatments.Subtracting the slope from unity gave the ratio of light-driven U 0 C to E 0 C , which was significantly lower for the N 2 treatment.
The ratio of gross photosynthesis (E 0 ) to N 2 fixation increased 9-fold and 6-fold for the NH 4 + and NO 3 -treatments relative to the N 2 treatment.In addition, the ratio of net photosynthesis (P net ) to N 2 fixation was 12-fold and 7-fold higher for the NH 4 + and NO 3 -treatments relative to the N 2 treatment (Table 4).

Light-dependence of PSII electron transport
The operating efficiency of PSII photochemistry (F q '/F m ´) increased at low light intensities, reaching a maximum at ~110 to 130 μmol photons m -2 s -1 , before decreasing significantly with increasing light intensity (Fig 3).The light saturation parameter (E k ) and the light at which ETR was maximal (E opt ) were significantly higher for the N 2 treatment than the NH 4 + treatment.Conversely, the light inhibition parameter (E p ), absorption cross-section of PSII  Light absorption coefficients were spectrally corrected to the culture LEDs and were normalised to a chlorophyll a (m 2 g Chla -1 ) and carbon (m 2 g C -1 ) basis.5).
The light intensity at which ETR was maximal (E opt ) was significantly lower (by ~120 μmol photons m -2 s -1 ) for the NH 4 + treatment relative to the N 2 treatment (Fig 4).The Chla and C-specific maximum electron transport rate and initial slope (α ETR ) of the ETR-light curves were not significantly different between N-source treatments (Table 5, S2 Table ).In contrast, the light-saturated photoinhibition slopes (β ETR ) were significantly different, with β increasing by 5% and 10% for the NO 3 -and NH 4 + treatments, relative to the N 2 treatment (Table 5).
The ratio of PSII electron transport to gross O 2 evolution under light-limitation (F eα ) was ~4 and did not vary significantly between N-source treatments.Light saturated ratios (F em ) increased relative to F eα for all N-source treatments, with the N 2 treatment being 46% and 35% higher than the NH 4 + and NO 3 -treatments, respectively (Table 5).

Effect of acclimation to variation of N-sources on growth rates and elemental stoichiometry
Growth rates achieved under diazotrophic conditions were similar to most previous studies [23,[43][44][45], as was the increase in growth rate observed under non-diazotrophic conditions [23,43], which we attribute to the lowered demand of NADPH and ATP for nitrogenase activity, where NADPH and ATP could be re-directed to CO 2 fixation and/or biosynthesis.Our data shows that at saturating light intensity, the energetic cost of diazotrophy constrains Trichodesmium growth by ~13%.However, in a natural system, potential changes to inorganic carbon chemistry (influencing the activity of the carbon concentrating mechanism (CCM)), temperature (influencing enzyme activity), or other key nutrients (i.e.Fe, P), all of which were controlled in our experiments, will almost certainly influence this estimate.

Effect of acclimation to different N-sources on gross photosynthesis
We show an effect of N-source on C-specific light saturated gross O 2 evolution rates.The more than two-fold increase in the maximum O 2 evolution rate and initial slope when T. erythraeum IMS101 was grown on NH 4 + or NO 3 -than when growing diazotrophically was largely due to differences in the ratio of Chla:C as chlorophyll a-specific photosynthetic parameters varied by only 36% between N 2 and NH 4 + treatments.The increase of C-specific gross O 2 evolution rates when Trichodesmium is supplied with NH 4 + or NO 3 -may be due to an increase in the maximum rate of CO 2 fixation and/or to an increase in PSII concentration.Previous studies report high PSI:PSII ratios under diazotrophic conditions (ranging between 1.3 to 4) [47][48][49][50][51], which would allow cyclic photophosphorylation in diazocytes to provide most of the ATP required for N 2 fixation, with glycolysis and the Kreb's cycle providing the required reducing equivalent.It may be that under non- E k μmol photons m -2 s -1 465 (8) [B]  421 (3) [A]  447 (20) α ETR C mmol e -(g C) -1 h -1 (μmol photons m -2 s -1 ) -1 0.133 (0.033) 0.167 (0.033) 0.200 (0.003) C mmol e -(g C) -1 h -1 (μmol photons m -2 s -1 ) -1 5081 (55) [A]  5577 (55) [C]  5332 ( 14) [B]   E opt μmol photons m -2 s -1 1263 ( 22) [B]  1144 (3) [A]  1216 (54) E p μmol photons m -2 s -1 0.99 (0.11) 0.67 (0.08) 0.83 (0.09) , the C-specific light saturated slope of the electron transport rate light response curve; E k , the light saturation parameter; E opt , the light at which ETR is maximal; E p , the light inhibition parameter; F v /F m , the maximum photochemical efficiency of PSII in the dark-adapted state; σ PII , the absorption cross-section of PSII photochemistry in the dark-adapted state; τ f , the average time constant for the re-opening of a closed PSII reaction centre in the dark-adapted state; F em , the light saturated ratio of PSII electron transport to gross O 2 evolution; F eα , the light limited ratio of PSII electron transport to gross O 2 evolution.The r 2 values of all curve fits were > 0.

Effect of acclimation to different N-sources on N 2 fixation
Nitrogenase activity declined significantly by 81-84% when Trichodesmium was cultured in the presence of an additional N-source.Despite being cultured under N-replete concentrations, Trichodesmium cells in the NH 4 + and NO 3 -treatments exhibited a baseline rate of N 2 fixation.Similarly, Milligan et al. [52] reported a ~85% decrease when Trichodesmium was cultured in 100 μM of NO 3 -for 2 weeks and Holl and Montoya [44] reported a 66% decrease when cultured in 20 μM of NO 3 -, accrediting 8% of total N assimilation to diazotrophy despite the presence of additional N-sources.Maintaining the capability to perform N 2 fixation under non-diazotrophic conditions, albeit at a reduced rate, could reflect Trichodesmium's natural environment and act a potential safeguard mechanism to variable light and nutrient regimes.
Noting that 16 moles of ATP are consumed per mole of N 2 fixed (Eq 1) and that 2.56 moles of ATP can be produced per mole of O 2 evolved by photophosphorylation linked LPET [53], we calculated that T. erythraeum IMS101 may use 20% of the ATP that could be generated from gross O 2 evolution to support the observed N 2 fixation rate during diazotrophic growth: This proportion decreases to 2% and 4% for the NO 3 -and NH 4 + treatments, respectively, where the ratio of E 0 :N fix increases to 289 for the NO 3 -treatment and 185 in the NH 4 + treatment, versus 31 in the N 2 treatment (Table 4).
Studies on natural populations of Trichodesmium spp.have shown that the addition of NO 3 -(100 μM) in the morning can cause a gradual decrease of N 2 fixation over the photic period [22].Further studies have also shown that addition of glutamine (10 μM) immediately decreases N 2 fixation rates, indicating a direct effect on enzyme activity as opposed to enzyme synthesis [54].These observations have been accredited to accumulation of N-containing metabolites acting as potential inhibitors to the specific activity rather than abundance of nitrogenase [22,54].
It is well known that intracellular nitrogen pools have a role in regulating nitrogenase activity in diazotrophs [55,56].Dinitrogenase reductase catalyses the reduction of N 2 to NH 4 + , which is assimilated into glutamine (gln) and then into glutamate (glu) via the glutamine synthetase (GS, EC 6.3.1.2)/glutamatesynthase (GOGAT) pathway [54].The intracellular pools of NH 4 + , glu and gln have been identified as important feedback regulators of N uptake and metabolism, with GS activity in Trichodesmium being sensitive to both intra-and extracellular N concentrations [55].It could be hypothesised that the activity of nitrogenase is influenced by internally recycled N (e.g.NH 4 + and gln), while the synthesis of nitrogenase is influenced by newly assimilated N (e.g.NO 3 -).

Effect of acclimation to different N-sources on light-stimulated O 2 consumption and the relationship between net and gross O 2 evolution
Net photosynthesis was significantly lower for the N 2 treatment than for the NH  [43,52].
Several processes demand ATP in excess of the ATP:NADPH produced through linear photophosphorylation; two most notably being N 2 fixation and the operation of the CCM [57].In this study, the carbon chemistry of all cultures was closely regulated to ensure that variation in O 2 consumption and net photosynthesis was due to the N-source treatments only.Linearity between gross O 2 evolution (E 0 ) and O 2 consumption was observed across all Nsource treatments, suggesting that light-dependent O 2 consumption is linked to balancing ATP to NADPH production, as opposed to serving as a mechanism to dissipate excitation energy.
Diazotrophic cells consume more O 2 per evolved O 2 across the entire range of actinic light intensities than the NH 4 + and NO 3 -treatments.This suggests a higher rate of water-water cycling due to either Mehler activity or operation of plastoquinone terminal oxidase when N 2 is being fixed.To maintain a sufficient supply of ATP relative to NADPH, Trichodesmium may utilise pseudocyclic photophosphorylation linked to the Mehler reaction to augment the ATP generated by linear electron transfer from water to NADP + in addition to ATP produced by cyclic electron flow around PSI. Measurements of O 2 evolution, ETR and N 2 fixation were all made at one time of day (4 to 6 hours into the photo-phase of a 12:12 L:D cycle) and as such cannot be extrapolated to a diel response given the reports of temporal separation of photosynthesis and N 2 fixation in Trichodesmium [21].

Effect of acclimation to different N-sources on electron transport rates and photophysiology
Like Eichner et al. [43], we observed a negligible effect of N-source on many photo-physiological parameters, including F q ´/F m ´, σ PII and τ f .Trichodesmium exhibited a light response typical for most cyanobacteria, where the dark-adapted photochemical yield is significantly affected by respiratory electron flow [58].This results from a proportion of PSII reaction centres remaining in a closed state despite being in the dark and is imposed by a reduction in the plastoquinone (PQ) pool, which prevents the oxidation of Q A -. Moving from darkness to a low light intensity increases the electron flux through PSI, alleviates the bottleneck of electron transport through the Cyt b6f complex, thereby increasing F q ´/F m ´and decreasing the re-oxidation time of Q A -. Addition factors such as higher downregulation under dark-adapted conditions may also contribute to the increase in F q ´/F m ´under low light intensities.

Ratio of electron transport to gross O 2 evolution
Electrons are transferred from PSII (where O 2 is evolved) to an intermediate plastoquinone pool and eventually to ferredoxin to produce NADPH [59].A minimum of four moles of electrons are transported through PSII for each mole of O 2 evolved at PSII.Most higher plants exhibit a linear correlation between gross O 2 evolution and electron transport rate [60].In microalgae, this relationship is often ambiguous, especially at high light intensities where the relationship can become non-linear [61,62].
Here we show that at low light intensities, the ratio of PSII electron transport to gross O 2 evolution (F eα ) is close to a 4:1 ratio for all N-sources treatments.However, when light intensities exceed 150 μmol photons m -2 s -1 , F e declines as ETR saturates at a higher light intensity (~900 μmol photons m -2 s -1 ) than E 0 (~400 μmol photons m -2 s -1 ).Similar responses have been reported for diatoms [63], microalgae [64] and the Baltic cyanobacteria, Nostoc [65].Few studies have measured O 2 production rates in Trichodesmium [47,66] and to our knowledge none have reported concurrent PSII electron transport rates.Interestingly, we calculated a higher F e for the N 2 cultures than for the NH 4 + and NO 3 -cultures, irrespective of using the light-limited or -saturated rates.This may be due to overestimating the proportion of light absorbed by PSII in the non-diazotrophic growth conditions (i.e.NH 4 + and NO 3 -) relative to the diazotrophic condition.Here we assumed that 50% of absorbed light was directed to PSII reaction centres and 50% to PSI reaction centres (i.e.FAQ-PII of 0.5).It's likely that FAQ PII was overestimated for diazotrophic treatment (i.e.N 2 ) which may have had a higher ratio of PSI:PSII to support significant rates of cyclic photophosphorylation.In addition, non-diazotrophic cells may undergo more pronounced state transitions with phycobilin proteins being redistributed between PSII and PSI.Finally, a F e > 4 could be accredited to cyclic electron flow around PSII, which may act a mechanism to dissipate excess excitation energy under high light [67].

Implications for future oligotrophic oceans
In N-limited regions of the oligotrophic open ocean, diazotrophy provides a competitive advantage by allowing cells to access N 2 as an N-source against faster growing phytoplankton that rely on fixed N. Current ocean models predict a poleward shift in the 20 ˚C isotherm which could extend Trichodesmium's niche into higher latitudes.On a global scale, this niche expansion is driven by increased SSTs; however, on regional scales persistence in an area may be dictated by Trichodesmium's response to fluctuating nutrient regimes.
At the surface in oligotrophic waters, Trichodesmium is unlikely to encounter NO 2 -, NO 3 or NH 4 + concentrations in excess of 0.1 μM [68], except during mixing events.While Trichodesmium is commonly observed in the upper meters of the water column [69], observations have been recorded down to 200 m depth [70].Thus, Trichodesmium colonies and free trichomes are able to migrate to the nutricline [30,31].Such vertical migration has been suggested to allow luxury uptake of phosphates before colonies return to the surface.In addition to encountering phosphates, Trichodesmium will also encounter high concentrations of NO 3 in the nutricline.As such, NO 3 -uptake is likely at these greater depths or at the surface after a mixing event.
Mulholland et al. [17] reported significant NO 3 -uptake rates with the addition of 1 μM NO 3 -to the growth media.Furthermore, Karl et al. [30] showed that concentrations of dissolved NH 4 + reached 1.5 μM L -1 and dissolved organic N (DON) reaching 13 μM L -1 during a natural bloom of Trichodesmium spp. in the North Pacific gyre.These concentrations are far greater than typical oceanic N pools and could therefore be high enough to inhibit N 2 fixation rates [44].It's therefore possible that Trichodesmium colonies at depth may be utilising more combined N-sources than the blooms frequently measured on the surface.The energy and reductant conserved through utilising additional N-sources could significantly enhance Trichodesmium's productivity and growth which could have major implications for biogeochemical cycles.
Our results indicate the need to seek more information on the potential for natural populations of Trichodesmium to uptake fixed N-sources (e.g.NO 3 -, NH 4 + , labile dissolved organic nitrogen (DON)) at concentrations that migrating colonies or trichomes experience in the nutricline or that are encountered transiently after deep mixing events.The potential significance of Trichodesmium assimilating fixed N is indicated by a modelling study by McGillicuddy [33] which concluded that to obtain realistic simulations of biomass and export production Trichodesmium populations in the North Atlantic must utilise fixed N. Specifically, this study indicated that 15-20% of the N quota of Trichodesmium could be due to uptake of NO 3 -and NH 4 + .Furthermore, although uptake of NO 3 -, NH 4 + or DON will decrease N 2 fixation rates in the short-term, as these N-sources are depleted over longer time periods, the increase in Trichodesmium biomass may lead to increased N 2 fixation and greater competition for other nutrients including Fe and P.

Table 3 .
The mean (± S.E.) measured and modelled effective light absorption coefficients and the relative contribution of each photosynthetic pigment to the total light absorption under the culturing LEDs within T. erythraeum IMS101, when acclimated to three N-sources (

Table 4 .
The parameters (± S.E.) of the C-specific light-response curves for gross and net photosynthetic O 2 evolution of T. erythraeum IMS101 (n

Table 1 . The median (± S.E.) balanced growth rates and mean elemental stoichiometry and N 2 fixation rates for T. erythraeum IMS101 when acclimated to three N- source conditions (N 2 , NH 4 + and NO 3 - ), at a target CO 2 concentration (380 μatm), saturating light intensity (400 μmol photons m -2 s -1 ) and optimal temperature (26 ˚C). Variables Units N 2 NH 4 + NO 3 -
Table), however the effects were more pronounced when expressed as a C-specific rate, where E 0m C increased by 143% from the N 2 to the NH 4

Table 2 . The growth conditions (± S.E.) for T. erythraeum IMS101 when cultured under three N-source conditions (N 2 , NH 4 + and NO 3 - ), at a target CO 2 concentration (380 μatm), saturating light intensity (400 μmol photons m -2 s -1 ) and optimal temperature (26 ˚C).
+ concentration, allowing a mean pH value to be calculated.https://doi.org/10.1371/journal.pone.0195638.t002constant proportion of gross O 2 evolution (E 0 C -specific initial slopes the light response curve for net and gross photosynthesis; ɸ mgross and ɸ mnet are the maximum quantum efficiencies of gross and net N fix and P net :N fix , the ratio of gross and net photosynthesis to N 2 fixation, where rates of E 0 and P net were calculated at 400 μmol photons m -2 s -1 , matching to light intensity of the N 2 fixation incubations; slope, the gradient of the regression between P net C and E 0 C .The r 2 values of all curve fits were > 0.982.Letters in parenthesis indicate significant differences between CO 2 treatments (One Way ANOVA, Tukey post hoc test; P < .05);where [B] is significantly greater than [A] and [C] is significantly greater than [B] and [A].https://doi.org/10.1371/journal.pone.0195638.t004