Combined effects of different CO2 levels and N sources on the diazotrophic cyanobacterium Trichodesmium



To predict effects of climate change and possible feedbacks, it is crucial to understand the mechanisms behind CO2 responses of biogeochemically relevant phytoplankton species. Previous experiments on the abundant N2 fixers Trichodesmium demonstrated strong CO2 responses, which were attributed to an energy reallocation between its carbon (C) and nitrogen (N) acquisition. Pursuing this hypothesis, we manipulated the cellular energy budget by growing Trichodesmium erythraeum IMS101 under different CO2 partial pressure (pCO2) levels (180, 380, 980 and 1400 µatm) and N sources (N2 and NO3). Subsequently, biomass production and the main energy-generating processes (photosynthesis and respiration) and energy-consuming processes (N2 fixation and C acquisition) were measured. While oxygen fluxes and chlorophyll fluorescence indicated that energy generation and its diurnal cycle was neither affected by pCO2 nor N source, cells differed in production rates and composition. Elevated pCO2 increased N2 fixation and organic C and N contents. The degree of stimulation was higher for nitrogenase activity than for cell contents, indicating a pCO2 effect on the transfer efficiency from N2 to biomass. pCO2-dependent changes in the diurnal cycle of N2 fixation correlated well with C affinities, confirming the interactions between N and C acquisition. Regarding effects of the N source, production rates were enhanced in NO3 grown cells, which we attribute to the higher N retention and lower ATP demand compared with N2 fixation. pCO2 effects on C affinity were less pronounced in NO3 users than N2 fixers. Our study illustrates the necessity to understand energy budgets and fluxes under different environmental conditions for explaining indirect effects of rising pCO2.


acetylene reduction assay


carbonic anhydrase


carbon concentrating mechanism

chl a

chlorophyll a


inorganic carbon


dissolved inorganic carbon


PSII photochemical quantum yield measured in dark-adapted state


PSII photochemical quantum yield measured in light-adapted state


4-(2-hydroxylethyl)-1-piperazine-ethanesulfonic acid


honest significant difference


half-saturation concentration


CO2 partial pressure


particulate organic carbon


particulate organic nitrogen




ribulose-1,5-bisphosphate carboxylase/oxygenase


total alkalinity


maximum rate


PSII functional absorption cross section


QA re-oxidation time


The release of anthropogenic carbon (C) has caused atmospheric CO2 partial pressure (pCO2) to increase from 280 to 390 µatm since pre-industrial times and pCO2 levels are expected to rise further to 750 µatm or even beyond 1000 µatm by the end of this century (IPCC 2007, Raupach et al. 2007). As CO2 is taken up by the ocean, seawater CO2 concentrations increase and pH levels decrease, a phenomenon termed ocean acidification (Caldeira and Wickett 2003). These changes in carbonate chemistry are expected to have diverse effects on marine phytoplankton (Rost et al. 2008, Riebesell and Tortell 2011). By fixing CO2 into organic matter, phytoplankton acts as a C sink and plays a potential role as a negative feedback mechanism to atmospheric pCO2 increase (Raven and Falkowski 1999, De La Rocha and Passow 2007).

In marine ecosystems, phytoplankton productivity is often limited by availability of nitrogen (N). Fixation of atmospheric N2 by diazotrophic cyanobacteria thus plays a crucial role for primary productivity, particularly in oligotrophic regions of the world ocean. With global change, the marine N cycle is subject to an array of perturbations. On the one hand, increasing deposition of anthropogenic N leads to eutrophication in coastal regions (Duce et al. 2008). On the other hand, the expansion of oxygen minimum zones favors N loss processes such as denitrification and anammox (Lam and Kuypers 2011). Additionally, ocean acidification is expected to decrease marine nitrification rates (Beman et al. 2011), and global warming intensifies stratification and therewith lowers nutrient input into the upper mixed layer (Doney 2006). As the latter processes are likely to decrease the overall NO3 availability in the surface ocean, marine N2 fixation may become more important, helping to restore the global N budget.

The cyanobacterium Trichodesmium is considered one of the most important marine N2 fixers with an estimated contribution of up to 50% to global marine N2 fixation (Mahaffey et al. 2005). Previous studies found this diazotroph to be exceptionally sensitive to rising pCO2. Laboratory experiments exposing cultures to pCO2 levels projected for the end of this century showed significant increases in the production of particulate organic C and particulate organic nitrogen (POC and PON) as well as N2 fixation rates (Barcelos é Ramos et al. 2007, Hutchins et al. 2007, 2013, Kranz et al. 2009, Levitan et al. 2007); the magnitude of these effects yet differed strongly between investigations. In several follow-up studies, CO2 effects on Trichodesmium were found to be strongly modulated by other environmental factors such as iron (Shi et al. 2012) and light (Kranz et al. 2010, Levitan et al. 2010, Garcia et al. 2011), the latter highlighting the importance of energy in the modulation of CO2 effects.

Cyanobacteria have to invest a considerable share of energy into the accumulation of inorganic carbon (Ci) by carbon concentrating mechanisms (CCMs) owing to a competing reaction with O2 and a particularly low CO2 affinity of their ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO) (Badger et al. 1998). The CCM of Trichodesmium involves a distinct assembly of RubisCO and carbonic anhydrase (CA) within carboxysomes, as well as two Ci acquisition systems (Badger et al. 2006, Price et al. 2008). HCO3 is taken up via a Na+-dependent HCO3 transporter (BicA) whereas diffusive uptake of CO2 is facilitated by the so-called NDH-14 complex, converting CO2 to HCO3. Next to Ci acquisition, another important energy sink in Trichodesmium is N2 fixation (Kranz et al. 2011). As CCM activity was found to be downregulated at high pCO2 levels, while N2 fixation rates were simultaneously increased in this species, a reallocation of energy between C and N2 fixing pathways has been suggested to fuel the increase in production at high pCO2 (Kranz et al. 2010).

Similarly to RubisCO, nitrogenase is characterized by a high sensitivity toward O2 (Falkowski 1997). In consequence, while the fixation of N2 is an extremely energy demanding reaction in itself (Eqn (1)), diazotrophs face additional costs, which are related to the protection of nitrogenase from photosynthetically evolved O2 (Großkopf and LaRoche 2012). To separate O2 evolution from N2 fixation, Trichodesmium has a tightly regulated diurnal cycle of N2 fixation and photosynthesis (Berman-Frank et al. 2001), involving daily synthesis and degradation of nitrogenase (Capone et al. 1990, Sandh et al. 2009) and alternation of photosynthetic activity states (Küpper et al. 2004). Moreover, nitrogenase is expressed only in subsets of cells within filaments, the diazocytes (Lin et al. 1998, Berman-Frank et al. 2001). As no trans-cellular transport mechanisms for N compounds have been found in Trichodesmium, diazocytes have to release N for use by their neighboring cells (Mulholland and Capone 2000). Uptake mechanisms for N sources other than N2 are thus indispensable for this species.

Laboratory studies have shown that Trichodesmium can use NO3 and NH4+ as well as organic N compounds (glutamine, glutamate or urea; Mulholland et al. 1999), all of them requiring different amounts and types of energy equivalents. NO3 is taken up in cyanobacteria by high-affinity ATP-dependent transporters and subsequently reduced to NH4+ in a two-step ferredoxin-dependent reaction catalyzed by nitrate reductase and nitrite reductase (Flores et al. 2005, Wang et al. 2000) (Eqn 2).

display math(1)
display math(2)

In Trichodesmium, N2 fixation was shown to be inhibited in cultures grown in NO3-containing media (Ohki et al. 1991, Fu and Bell 2003, Holl and Montoya 2005, Sandh et al. 2011). As the uptake and reduction of NO3 requires little ATP (Eqn 2), it can be expected that NO3 addition to culture media will alter the energy budget of the cells in comparison to N2 fixing conditions.

In this study, Trichodesmium erythraeum IMS101 was acclimated in a matrix of four different pCO2 levels (ranging from 180 to 1400 µatm) and two different N sources (N2 and NO3). In addition to acclimation effects on the level of growth and composition (C, N and pigments), physiological key processes (N2 fixation, O2 fluxes and electron transport) were analyzed to improve our understanding of the plasticity in energy and resource allocation under the different energetic requirements imposed by changing environmental conditions.

Materials and methods

Culture conditions

Trichodesmium erythraeum IMS101 was grown in semi-continuous batch cultures at 25°C and 150 µmol photons m−2 s−1 with a 12:12 h light:dark cycle. Light was provided by white fluorescent bulbs (BIOLUX, Osram, München, Germany). Cultures were grown in 0.2-µm-filtered artificial seawater (YBCII medium; Chen et al. 1996) and kept in exponential growth phase by regular dilution with culture medium. Cultures consisted of single trichomes and cell densities ranged between approximately 6000 and 180 000 cells ml−1. Cells were acclimated in 1 l culture flasks, which were continuously bubbled with 0.2-µm-filtered air with pCO2 levels of 180, 380, 980 and 1400 µatm. Gas mixtures were generated with a custom-made gas flow controller. Prior to experiments, cells were allowed to acclimate to the respective pCO2 for at least 2 weeks. Cultures in which pH had drifted by >0.09 compared with cell-free reference media were excluded from further analysis. In treatments with NO3 as the N source, 0.2-µm-filtered NaNO3 was added to achieve mean concentrations of 97 ± 2 µM in the experiments, never falling below 65 µM. Concentrations were monitored photometrically according to Collos et al. (1999). Consumption of NO3 by cellular uptake was compensated for by regular additions of NaNO3. Cultures were acclimated to NO3 for at least 1 week before measurements.

Carbonate chemistry

To compensate for an increase in total alkalinity (TA) due to NO3 uptake (Wolf-Gladrow et al. 2007), appropriate quantities of HCl were added according to the daily changes in NO3 concentration. TA was determined by potentiometric titration with a TitroLine alpha plus titrator (Schott Instruments, Mainz, Germany) and calculation from linear Gran plots (Gran 1952). Average precision was ±5 µmol kg−1. Samples for dissolved inorganic carbon (DIC) analysis were filtered through 0.2 µm cellulose acetate filters and measured colorimetrically (TRAACS CS800 autoanalyzer, Seal, Norderstedt, Germany). Average precision was ±5 µmol kg−1. Certified reference materials supplied by A. Dickson (Scripps Institution of Oceanography) were used to correct for inaccuracies of TA and DIC measurements. pH values of the acclimation media were measured potentiometrically on the NBS scale [pH meter pH3110, Wissenschaftlich-Technische Werkstätten (WTW) GmbH, Weilheim, Germany]. Carbonate chemistry of the different pCO2 and N treatments is shown in Table 1.

Table 1. Carbonate chemistry for each pCO2 and N treatment acquired in daily measurements during the experiment. Attained pCO2 of the media was calculated from pH, TA, [PO43], temperature and salinity using the CO2sys program (Pierrot et al. 2006) with equilibrium constants K1 and K2 given by Mehrbach et al. (1973), refit by Dickson and Millero (1987). Errors denote 1 sd (n ≥ 6).
pCO2 treatment (µatm)NO3pH (NBS)TA (µmol kg−1)DIC (µmol kg−1)pCO2 attained (µatm)
1808.48 ± 0.042415 ± 111847 ± 10 173 ± 12 
+8.49 ± 0.042438 ± 371868 ± 22168 ± 4 
3808.26 ± 0.032408 ± 152014 ± 6 329 ± 22
+8.26 ± 0.042389 ± 7 2003 ± 12325 ± 10
9807.88 ± 0.022377 ± 152142 ± 17918 ± 12
+7.89 ± 0.032392 ± 192166 ± 26912 ± 35
14007.74 ± 0.022399 ± 442231 ± 781354 ± 30 
+7.76 ± 0.052413 ± 372237 ± 201298 ± 74 

Growth and elemental composition

Samples for determination of growth and elemental composition of cells were generally taken between 1 and 2.5 h after beginning of the photoperiod to account for changes due to the diurnal rhythm of Trichodesmium. Duplicate samples for chlorophyll a (chl a) determination were extracted in acetone for >12 h and determined fluorometrically (TD-700 Fluorometer, Turner Designs, Sunnyvale, CA; Holm-Hansen and Riemann 1978). Specific growth rates (µ) were estimated by exponential regression through chl a concentrations measured daily over at least 4 days. Duplicate samples for analysis of POC and PON were filtered onto pre-combusted GF/F filters and stored at −20°C. Prior to analysis, filters were acidified with 200 µl HCl (0.2 M) to remove all inorganic C. POC and PON contents as well as PON isotopic composition (δ15N) were measured with an EA mass spectrometer (ANCA SL 20-20, Sercon Ltd, Crewe, UK). Daily production rates of POC and PON were obtained by multiplication of the respective elemental contents and growth rates.

N2 fixation

N2 fixation rates were determined using the acetylene reduction assay (ARA) (Capone 1993). Samples were spiked with acetylene (20% of head space volume) in crimp vials followed by incubation for 1 h at acclimation light and temperature with continuous agitation to avoid aggregation of cells. The amount of acetylene reduced to ethylene was then measured by gas chromatography (Trace GC, Thermo Finnigan, Bremen, Germany). Solubility of acetylene in the aqueous phase was taken into account by applying the Bunsen coefficient (0.088 at 25°C and salinity 32; Breitbarth et al. 2004). A conversion factor of 4:1 (Capone and Montoya 2001) was used to convert acetylene reduction rates to N2 fixation rates.

O2 fluxes

Cellular O2 fluxes were measured by means of membrane inlet mass spectrometry (MIMS) as described by Rost et al. (2007). Assays were performed in YBCII medium buffered with 4-(2-hydroxylethyl)-1-piperazine-ethanesulfonic acid (HEPES, 50 mM, pH 8.0) at 25°C and acclimation light intensity unless otherwise specified. To account for the diurnal cycle of O2 fluxes in Trichodesmium, measurements were performed three times over the day, during time intervals from 0 to 1.5, 5.5 to 7 and 9 to 10.5 h after beginning of the photoperiod. For normalization of the O2 traces, duplicate samples for chl a analysis were taken after each measurement.

In the first set of measurements, net O2 evolution was determined as a function of DIC using the disequilibrium method described by Badger et al. (1994). O2 fluxes were monitored during consecutive dark and light phases typically lasting 4 min, starting with DIC concentrations close to zero (media bubbled with CO2-free air), which were subsequently increased by step-wise addition of NaHCO3 up to a maximum of approximately 4500 µM DIC. DIC-saturated rates of photosynthesis [Vmax (DIC)] and half-saturation concentrations [K1/2 (DIC)] were obtained by fitting a Michaelis–Menten function to the data.

In a second approach, O2 evolution and uptake were assessed as a function of light intensity according to Fock and Sültemeyer (1989). Prior to measurements, HEPES-buffered YBCII media were bubbled with N2 to remove 16O2, then spiked with 18O2 gas (Chemotrade, Düsseldorf, Germany) and allowed to equilibrate for >30 min, reaching O2 concentrations of approximately 21%. DIC concentration was adjusted to approximately 2000 µM by addition of 1 M NaHCO3 solution prior to measurements. Fluxes of 16O2 and 18O2 were monitored during consecutive 4 min dark and light intervals, applying a range of light intensities from 8 to 2000 µmol photons m−2 s−1. The light intensity at which photosynthesis starts to enter saturation (Ik) was obtained by a curve fit as specified by Rokitta and Rost (2012).

Fluorescence measurements

Chl a fluorescence was measured using the Fluorescence Induction and Relaxation (FIRe) method with a FIRe Fluorometer System (Satlantic, Halifax, Canada) and the associated actinic light source. Measurements were performed in parallel to the 18O2 assays, following the same protocol of dark and light intervals as well as light intensities. PSII photochemical quantum yield (Fv/Fm: measured in dark-adapted state; Fv′/Fm′: measured in light-adapted state) and functional absorption cross section of PSII (σ) as well as QA re-oxidation time (τ) were assessed by analysis of the single turnover flash response using the Fireworx matlab code (, written by Audrey Barnett).

Statistical analysis

Data were analyzed using r for significance of differences by two-way analysis of variance (anova) tests, followed by Tukey's test for Honest Significant Differences (TukeyHSD) for specification of differences between means where appropriate. A significance level of P ≤ 0.05 was applied.


Growth and composition

Cellular chl a contents stayed relatively constant over the range of pCO2 levels and N sources tested, with a mean of 1.04 ± 0.28 pg chl a cell−1, and were therefore used for normalization. Growth decreased slightly with increasing pCO2 in both N treatments when all data were included (Fig. 1, anova, P < 0.005). However, when testing CO2 levels individually, the differences between 980 and 380 µatm as well as 180 µatm depended on the N source: with N2 there was a significant pCO2 effect (TukeyHSD, adjP < 0.05) while with NO3 there was no effect (TukeyHSD, adjP > 0.05). In contrast, contents of POC and PON significantly increased by approximately 33% from 180 to 1400 µatm pCO2 (anova, P < 0.0001). As a result of these opposing trends, production rates of POC and PON did not change over the range of pCO2 levels tested (Table 2, anova, P > 0.05). There was no clear trend in the pCO2 effect on the POC:PON ratio (Table 2). Regarding effects of the N source, cells grown on NO3 had slightly higher growth rates than N2 fixing cells (anova, P < 0.05). Consequently, also production rates of POC and PON were higher in NO3 grown cultures (anova, P < 0.0001), even though contents of POC and PON were not significantly affected (anova, P > 0.05). POC:PON ratios were significantly lower in NO3 users than in N2 fixers (Table 2, anova, P < 0.0001).

Figure 1.

Acclimation responses of Trichodesmium grown under different pCO2 levels and N sources (N2 and NO3). (A) Growth rates (n ≥ 4). (B) Ratios of POC to chl a (n ≥ 3). (C) Ratios of PON to chl a (n ≥ 3). Error bars denote 1 sd.

Table 2. POC and PON production and ratios of Trichodesmium grown under different pCO2 levels and N sources (N2 and NO3). Errors denote 1 sd (n ≥ 3).
pCO2 treatment (µatm)NO3POC production (µmol (µg chl a)−1 day−1)PON production (µmol (µg chl a)−1 day−1)POC:PON (mol:mol)
1801.75 ± 0.100.36 ± 0.024.83 ± 0.10
+1.89 ± 0.240.39 ± 0.054.82 ± 0.06
3801.60 ± 0.220.33 ± 0.044.81 ± 0.08
+1.82 ± 0.140.39 ± 0.034.68 ± 0.07
9801.26 ± 0.420.26 ± 0.094.89 ± 0.09
+2.15 ± 0.190.45 ± 0.054.82 ± 0.12
14001.62 ± 0.200.33 ± 0.034.93 ± 0.21
+2.02 ± 0.150.44 ± 0.034.56 ± 0.07

N2 fixation

N2 fixation was inhibited by the addition of NO3, with N2 fixation rates being close to detection limit (data not shown). Moreover, the δ15N of PON clearly differed between treatments (anova, P < 0.0001), with −1.3 ± 1.0‰ in N2 fixing cells and +4.8 ± 1.3‰ in NO3 grown cells. N2 fixation rates displayed a typical diurnal cycle with high rates during midday (Fig. 2). At elevated pCO2, there was a change in the diurnal pattern toward N2 fixation rates remaining high until the end of the photoperiod. Consequently, integrated rates of N2 fixation over the photoperiod increased by approximately 60% from 27 to 43 nmol N2 (µg chl a)−1 day−1 at 380 and 1400 µatm pCO2, respectively. At 180 and 980 µatm pCO2, integrated rates were 25 and 39 nmol N2 (µg chl a)−1 day−1, respectively.

Figure 2.

Diurnal cycle of N2 fixation in Trichodesmium grown at 380 µatm (circles) and 1400 µatm (triangles) pCO2 in a 12:12 h light:dark cycle. Open and closed symbols represent biological duplicates.

O2 fluxes

To characterize the energy generating processes, O2 evolution was firstly assessed as a function of DIC concentrations. In all treatments, maximal net O2 evolution [Vmax (DIC)] followed a typical diurnal cycle with lowest rates during midday (Fig. 3, anova, P < 0.0001). Values of Vmax (DIC) were not affected by the N source or pCO2 (anova, P > 0.5). Half saturation concentration [K1/2 (DIC)] followed a distinct diurnal pattern with significantly lower values in the morning than for the rest of the day (Fig. 3, anova, P < 0.0001). Effects of pCO2 on K1/2 (DIC) were modulated by the N source and the time of day. While under NO3 grown conditions, there was no significant pCO2 effect (TukeyHSD, adjP > 0.5), K1/2 (DIC) was significantly lower at 380 µatm than at 1400 µatm pCO2 under N2 fixing conditions (TukeyHSD, adjP < 0.005). The difference between pCO2 treatments in N2 fixers was especially pronounced toward the evening (TukeyHSD, adjP < 0.0001), with K1/2 (DIC) decreasing in cells grown at 380 µatm but remaining high in cells grown at 1400 µatm pCO2.

Figure 3.

Typical diurnal cycle of O2 evolution as a function of DIC [(A) measured in Trichodesmium grown at 1400 µatm −NO3]; Diurnal cycle of maximal net O2 evolution (B) and half saturation DIC concentrations (C) in Trichodesmium grown under different pCO2 levels and N sources (N2 and NO3). Error bars denote 1 sd (n ≥ 3; except for 1400 µatm −NO3 morning and midday and 380 µatm +NO3 morning with n = 2).

In a second approach, evolution and uptake of O2 were assessed over a range of light intensities. Net O2 evolution typically reached light compensation between 10 and 60 µmol photons m−2 s−1 and started to enter saturation (Ik) at approximately 280 µmol photons m−2 s−1 (Fig. 4). Light-dependent O2 uptake, i.e. an excess of O2 uptake in the light over O2 uptake in the dark, was detected at irradiances >60 µmol photons m−2 s−1 (Fig. 4). Gross O2 evolution typically saturated at higher light intensities than net photosynthesis, which is consistent with the increase in light-dependent O2 uptake with increasing irradiance. At acclimation light intensity, diel mean values for dark respiration and light-dependent O2 uptake amounted to 20 and 13% of gross O2 evolution, respectively. With irradiances increasing beyond acclimation levels, light-dependent O2 uptake increased further and equaled about 19% of gross O2 evolution at 1300 µmol photons m−2 s−1. Regarding effects of pCO2 and N source on gross O2 evolution and light-dependent O2 uptake at acclimation light, no clear trend was found and rates followed a diurnal pattern with highest values in the morning (data not shown). Also dark respiration was not affected by pCO2 or N source, yet rates were lowest in the morning (data not shown).

Figure 4.

Typical example of light dependence of O2 fluxes and chlorophyll fluorescence parameters (measured in Trichodesmium grown at 380 µatm pCO2 with NO3). Total O2 uptake consists of dark respiration and light-dependent O2 uptake. Fv/Fm, PSII photochemical quantum yield; σ, PSII functional absorption cross section; τ, QA re-oxidation time.

Chlorophyll fluorescence

PSII photochemical quantum yield (Fv/Fm and Fv′/Fm′) was higher at acclimation light than in the dark and decreased with increasing light intensities beyond acclimation level, covering a range between approximately 0.5 and 0.1 (Fig. 4). At low irradiances, the functional absorption cross section of PSII (σ) increased with light (by approximately 30% from approximately 8 to 60 µmol photons m−2 s−1) while at higher light intensities, it slightly decreased (by approximately 10% from approximately 60 to 2000 µmol photons m−2 s−1). QA re-oxidation time (τ) was longest in the dark and decreased by approximately 40% from approximately 60 to 2000 µmol photons m−2 s−1.

Regarding changes in the diurnal cycle, light-adapted Fv′/Fm′ values were highest in the morning (0.48 ± 0.04 at acclimation light) and lowest at midday (0.40 ± 0.05 at acclimation light, anova, P < 0.0001, Fig. 5). Variability over the course of the day was even more pronounced regarding dark-adapted Fv/Fm (anova, P < 0.0001), with values ranging from 0.35 ± 0.07 (morning) to 0.17 ± 0.01 (midday). Likewise, functional absorption cross section of PSII (σ) was always highest in the morning (anova, P < 0.0001), irrespective of light conditions. QA re-oxidation time, which was measured in the dark (τdark), was significantly lower in the morning than during the rest of the day (anova, P < 0.0001), while τlight decreased slightly by approximately 10% over the course of the day (anova, P < 0.05).

Figure 5.

Diurnal cycle of chlorophyll fluorescence of Trichodesmium cultures grown under different pCO2 levels and N sources (N2 and NO3), measured in the dark (A, C and E) and at acclimation light intensity (B, D and F). Error bars denote 1 sd (n ≥ 3, except for τ of 380 µatm +NO3 midday with n = 1). Fv/Fm, PSII photochemical quantum yield; σ, PSII functional absorption cross section; τ, QA re-oxidation time.

Concerning responses to pCO2 and the N source, Fv/Fm and Fv′/Fm′ were not significantly affected by either of the two parameters (Fig. 5, anova, P > 0.05). Irrespective of light conditions, functional absorption cross section of PSII (σ) was not affected by the N source (anova, P > 0.05), while it was slightly higher at 1400 µatm than 380 µatm pCO2, however, only by approximately 10% (anova, P < 0.0001). Neither τdark nor τlight were significantly affected by pCO2 or N source (anova, P > 0.5).


To investigate CO2 effects on Trichodesmium under altered energy requirements, cultures were grown over a range of different pCO2 levels under N2 fixing conditions as well as with NO3, the latter providing a N source with a significantly lower demand in ATP but higher electron requirements (Eqns (1) and (2)). We also tested NH4+ as an alternative N source, which would have altered the energy requirements most strongly, lowering the ATP as well as the electron demand compared with N2 fixation. However, measurements revealed NH4+ to be toxic to Trichodesmium in concentrations as low as 10 µM (data not shown), which equaled the average daily N consumption in our cultures, and therefore argued against the applicability in dilute batch incubations. Additionally, concentrations could not be kept stable because of pH-dependent out-gassing of NH3 (data not shown), rendering it impossible to perform pCO2 manipulations without simultaneously affecting the N availability. The addition of NO3, on the other hand, had no negative effects on Trichodesmium and was not influenced by pH. As a consequence, we chose NO3 to impose a change in the energy status of cells. The change in N usage upon NO3 addition was demonstrated by direct measurements of nitrogenase activity as well as by the change in 15N composition of PON. NO3 assimilation resulted in corresponding changes in TA, which were compensated by additions of HCl in equimolar amounts to keep the carbonate system comparable between N2 fixing and NO3 using cultures. Growth rates and Fv/Fm indicate that cells were not stressed in any of the treatments. Please note that the light level applied in the acclimations (150 µmol photons m−2 s−1) was below saturation (Fig. 4), imposing a general energy constraint in the cell.

Acclimation effects of different pCO2 levels and N sources

The increase of POC as well as PON with pCO2 (Fig. 1) is in accordance with previous results (Kranz et al. 2009, 2010). Respective production rates, however, stayed relatively constant due to the concomitant decrease in growth (Table 2, Fig. 1). In other words, cells contained less biomass and divided more quickly at low and medium pCO2, while at high pCO2, cell quotas were higher and cells divided more slowly. Among the previous studies on T. erythraeum IMS101 testing the effect of pCO2 levels up to 750 or 1000 µatm, some showed an increase in growth rate with pCO2 (Barcelos é Ramos et al. 2007, Levitan et al. 2007, Kranz et al. 2010, Garcia et al. 2011), while others did not find significant differences (Hutchins et al. 2007, Kranz et al. 2009). Only one study has tested CO2 levels comparable to our highest CO2 treatment, finding that positive effects on growth leveled off between 760 and 1500 µatm in T. erythraeum GBRTRL101 (Hutchins et al. 2007). In a recent study investigating pCO2 effects under low iron availability representative for oligotrophic oceans, growth rates of Trichodesmium were shown to decrease with pCO2 (380 vs 750 µatm; Shi et al. 2012).

The stimulation in PON production in NO3 grown cells may be directly attributed to the lower energy requirement for N assimilation (Table 2, Eqns (1) and (2)) as well as the fact that filaments are not subject to N loss during transfer from diazocytes to non-diazotrophic cells when grown on NO3. The effect on POC production, however, cannot be directly linked to N assimilation and suggests a more global effect of NO3 on the cells' metabolism such as reallocation of energy from N to C assimilation. The cost reduction associated with the switch from N2 to NO3 assimilation is also reflected in the lower POC:PON ratios under these conditions (Table 2). Changes in POC:PON ratios have been found in response to nutrient limitation in different phytoplankton (Sterner and Elser 2002). Even though none of our treatments was N limited, the observed changes in POC:PON ratios may simply reflect the higher N assimilation costs in N2 fixers.

O2 fluxes and electron transport

To better understand how the observed effects of pCO2 and N source on POC and PON were fuelled, we investigated photosynthesis as a measure of energy generation. Concerning treatment effects, neither pCO2 nor the N source had a significant effect on net O2 evolution (Fig. 3). However, net O2 evolution can be uncoupled from energy generation by high rates of O2 uptake or cyclic electron transport (Heber 2002). Thus, gross and net O2 fluxes as well as chlorophyll fluorescence need to be considered to obtain a more complete picture of energy generating processes. In all cultures, irrespective of pCO2 or N source, about one third of the gross O2 evolved was consumed by dark respiration and light-dependent O2 uptake (Fig. 4), the latter being indicative for either the classical Mehler reaction (Mehler 1951) or the equivalent reduction of O2 by flavoproteins (Helman et al. 2003). High rates of O2 uptake by dark respiration and Mehler reaction have been suggested to protect nitrogenase from O2 inhibition in Trichodesmium (Kana 1993, Carpenter and Roenneberg 1995, Berman-Frank et al. 2001, Milligan et al. 2007). Rates of Mehler reaction equaled only about 10% of gross O2 evolution at acclimation light intensity, yet rates increased when light intensities exceeded acclimation levels (Fig. 4). This light dependency could either indicate a role for Mehler reaction in photoprotection and/or reflect the enhanced need for nitrogenase protection at high gross O2 evolution rates. Moreover, the fact that light-dependent O2 uptake was not significantly affected by the N source seems surprising, considering the proposed role of Mehler reaction in nitrogenase protection in Trichodesmium (Milligan et al. 2007).

Chlorophyll fluorescence showed a light response typical for cyanobacteria, with dark-adapted fluorescence being controlled by respiratory electron flow that introduces electrons into the plastoquinone (PQ) pool (reviewed by Campbell et al. 1998). At low light intensities, electron flux through PSI is induced, oxidizing the PQ pool and thereby increasing Fv′/Fm′ and decreasing QA re-oxidation time (τ) (Fig. 4). When light intensities increase beyond acclimation light, input of excitation energy can become higher than the cells' capacity of ferredoxin re-oxidation, making cells vulnerable to photodamage. However, being adapted to high and variable light regimes, Trichodesmium employs effective photoprotective mechanisms (Breitbarth et al. 2008, Andresen et al. 2009). First of all, state transitions lead to a re-arrangement of phycobilisomes toward PSI, decreasing the PSII functional absorption cross section (σ) and therewith Fv′/Fm′ (Fig. 4). Second, the enhanced rates of Mehler reaction dissipate electrons at high light (Fig. 4). The effectiveness of these photoprotective mechanisms is reflected in a decreasing QA re-oxidation time (τ) whilst gross O2 evolution increases with light (Fig. 4).

To cover the high ATP demand of N2 fixation (Table 3; Eqn (1)), Trichodesmium depends on high rates of cyclic electron transport and Mehler reaction, increasing the ATP:NADPH ratio beyond that provided by linear photosynthetic electron transport. High rates of cyclic electron transport have been proposed to result in chemical reduction of the PQ pool, increasing re-oxidation time of QA (Berman-Frank et al. 2001). Assuming that cells adjust their energy generation closely to their needs, we expected the treatment-dependent differences in energy demand to be reflected in chlorophyll fluorescence. Contrary to our assumption, none of the fluorescence parameters measured was affected by pCO2 or N source with the exception of a small pCO2 effect on functional absorption cross section of PSII (σ, Fig. 5).

Table 3. Theoretical ATP and electron (e) costs of cellular processes and costs calculated for the observed POC and PON production rates under two different N sources (N2 and NO3). Theoretical demands of C fixation, CCM and POC production were normalized to mol N using the average POC:PON ratio measured in the experiment. CCM costs are based on 80% HCO3 use and a transport cost of 0.5 mol ATP per mol HCO3, assuming 50% CO2 leakage. Costs of NO3 assimilation include 1 mol ATP for uptake. Loss of fixed nitrogen (e.g. NH4+) is not accounted for. Please note that numbers given do not include costs for synthesis of enzymes and transporters, which would significantly increase the estimates for fixation of carbon as well as nitrogen (Brown et al. 2008). POM (particulate organic matter) is the sum of POC and PON.
ProcessUnitATPeATP:NADPH + H+Reference
C fixationmol (mol N)−114 191.5

Allen 2002

CCMmol (mol N)−14 0 

Hopkinson et al. 2011

POC productionmol (mol N)−118 191.9 
N2 assimilation to NH4+mol (mol N)−18 44.0

Flores and Herrero 1994

NO3 assimilation to NH4+mol (mol N)−11 80.3

Flores et al. 2005

NH4+ assimilation to glutamatemol (mol N)−11 21.0

Flores et al. 2005

PON production N2 fixermol (mol N)−19 63.0 
PON production NO3 usermol (mol N)−12 100.4 
POC production measured in N2 fixerµmol (µg chl a)−1 day− 
POC production measured in NO3 userµmol (µg chl a)−1 day− 
POC production difference NO3 vs N2µmol (µg chl a)−1 day−11.61.7  
PON production measured in N2 fixerµmol (µg chl a)−1 day− 
PON production measured in NO3 userµmol (µg chl a)−1 day− 
PON production difference NO3 vs N2µmol (µg chl a)−1 day−1−2.0 2.3  
Total POM production in N2 fixerµmol (µg chl a)−1 day− 
Total POM production in NO3 userµmol (µg chl a)−1 day−18.312.11.4 
POM production difference NO3 vs N2µmol (µg chl a)−1 day−1−0.5 3.9  

Regarding the diurnal cycle, there was a characteristic downregulation of maximal net photosynthesis as well as Fv/Fm during midday, which has been shown previously in Trichodesmium as part of the cells' mechanisms to reduce O2 concentrations during the phase of highest N2 fixation (Berman-Frank et al. 2001). In the morning, highly efficient electron transport was indicated by high Fv/Fm and a large PSII functional absorption cross section (Fig. 5), which is in line with the high gross O2 evolution (data not shown). Dark respiration, as indicated by τdark and O2 flux measurements, was lowest in the morning, while rates of Mehler reaction were at their maximum. Later during the day, rates of photosynthetic electron transport decreased, reflected by lower Fv/Fm, functional absorption cross section of PSII (σ), O2 evolution as well as Mehler reaction, while dark respiration increased. Interestingly, the diurnal cycle of O2 evolution and uptake as well as electron transport was maintained also in NO3 grown cultures. Studies on the diurnal cycle of nitrogenase protein abundance in Trichodesmium showed that nitrogenase is synthesized de novo every day (Zehr et al. 1996), resulting in a significant energy demand for protein synthesis (Brown et al. 2008). Nitrogenase was found to be synthesized, yet not activated by post-translational modification, in cells grown even at high levels of NO3 (Ohki et al. 1991). These findings suggest that although nitrogenase was not active, NO3 grown cells in our study may still have invested a considerable amount of energy for synthesis of nitrogenase. This would cause similar energy requirements as well as protection of nitrogenase from O2 also in NO3 grown cells (i.e. O2 consumption by dark respiration and Mehler reaction as well as downregulation of photosynthesis during midday), explaining the lack of N effects on chlorophyll fluorescence and O2 fluxes observed in our study. There is, however, also data suggesting significantly lower expression levels of nitrogenase subunits NifK and NifH in NO3 grown cells (Sandh et al. 2011).

In summary, the lack of a clear pCO2 or N effect on photosynthesis, dark respiration or Mehler reaction confirms that there was no difference in energy generation (ATP and reducing equivalents). The observed treatment effects on contents and production of POC and PON can thus not be explained by differences in the overall energy availability, indicating potential changes down-stream of the electron transport chain. To identify alterations in the energy consuming processes we therefore measured rates of N2 fixation and C acquisition.

N2 fixation

In agreement with previous results (Kranz et al. 2010), a characteristic change in the diurnal pattern of N2 fixation was observed at elevated pCO2, with the phase of high nitrogenase activity being prolonged toward the evening (Fig. 2). Although integrated daily N2 fixation rates increased by as much as 60% between 380 and 1400 µatm pCO2, PON production was not significantly affected by the different pCO2 levels. The ARA used for estimating N2 fixation rates gives a measure of the maximal nitrogenase enzyme activity under the respective assay conditions (approximating gross N2 fixation) while PON production rates reflect how much N is ultimately incorporated into the cells (approximating net N2 fixation). While there are indications that a considerable share of fixed N is lost before incorporation into PON (Mulholland and Capone 2000, Mulholland 2007), significant uncertainties remain with respect to the absolute values due to methodological issues (Mulholland and Capone 2001 and references therein). It also has to be noted that, in contrast to acetylene reduction during ARA, actual N2 fixation is dependent on ammonium consumption by downstream metabolism (e.g. Herrero et al. 2001). However, interpretation of trends within results of each of the methods should be valid. In accordance with our findings on CO2 sensitivity, previous studies found ARA-based estimates of N2 fixation to increase more strongly with pCO2 than estimates of PON production based on cell quotas or 15N fixation (Kranz et al. 2010, Garcia et al. 2011). In the natural environment, N release by Trichodesmium has been suggested to provide an important N source for a range of associated organisms (Mulholland and Capone 2000, Mulholland 2007), which may be enhanced under elevated pCO2 according to our data. The high assimilation costs and unavoidable N loss in N2 fixers impose higher energy requirements compared with NO3 users, especially under elevated pCO2. As all treatments, however, showed the same energy generation, we expect changes in other energy sinks.

Inorganic C acquisition

Acquisition of inorganic C constitutes a major energy sink in Trichodesmium due to the high CCM activities required to compensate for the poor CO2 affinity of its RubisCO (Kranz et al. 2009). Similarly to O2 and electron fluxes as well as N2 fixation, also the affinity for inorganic C was subject to a strong diurnal cycle (Fig. 3), which was previously described by Kranz et al. (2009). The high affinity for inorganic C in the mornings observed in all treatments is in line with the high rates of photosynthesis discussed above. The overall lower affinities at high pCO2, especially during the second half of the day, suggest significantly lower operational costs for the CCM which, in turn, allow for the enhanced N2 fixation observed (Figs 2 and 3). These CO2-dependent changes in affinities and the anti-correlation with N2 fixation are in agreement with previous results (Kranz et al. 2010). The fact that pCO2 effects are larger in N2 fixers than in NO3 using cells can be attributed to the higher overall energy requirements of N2 fixation as well as differences in the stoichiometry of ATP and electron demand (Fig. 6): Provided that the downregulation of CCM activity mainly saves ATP, this surplus energy can be readily used in N2 fixers to cover the high ATP demand of nitrogenase. In contrast, NO3 usage requires only little ATP (for uptake) and is, instead, likely to be limited by the supply of reducing equivalents. Consequently, a downregulation of the CCM in NO3 users would not have the same stimulatory effect on PON production as in N2 fixers.

Figure 6.

Schematic diagram of the distribution of energy equivalents for PON production under different N sources (N2 and NO3). Due to the different requirements of N2 and NO3 assimilation with respect to ATP and electron (e) stoichiometry, N2 fixation is prone to limitation by ATP while NO3 assimilation tends to be limited by e supply. Please note that the ultimate outcome in terms of PON production in the different N treatments is strongly dependent on the ratio of ATP per e available, which is, in turn, controlled by the ratio of (pseudo-) cyclic to linear e transport and the use of energy equivalents by other cellular processes. NaR, nitrate reductase; NiR, nitrite reductase.

Energy requirements of the CCM are generally dependent on the C sources and uptake mechanisms. CCM operation in Trichodesmium is considered to predominantly consume ATP, as the main C source for this species is HCO3 (approximately 80%; Kranz et al. 2009, 2010), which is taken up via a transporter fuelled indirectly by ATP (BicA; Price et al. 2008). Such HCO3 transporters are dependent on a Na+ gradient across the plasma membrane and presumably consume 0.5 mol ATP per mol HCO3 (Espie and Kandasamy 1994, Hopkinson et al. 2011). Furthermore, the so-called NDH-14 complex converts CO2 to HCO3, thereby driving uptake of CO2 as well as an internal recycling to prevent CO2 leakage (Price et al. 2002, 2008). The reaction is involved in the electron transport chain, receiving electrons from NADPH or ferredoxin that are subsequently transferred to PQ. Intriguingly, NDH-14 activity leads to a release of protons into the thylakoid lumen, which in turn increases the pH gradient used for ATP synthesis. The observation that this complex seems to be especially active at high pCO2 (Kranz et al. 2010) is in line with the increased ATP demand by enhanced N2 fixation under these conditions (Fig. 2). It has to be noted that the operational costs for BicA and NDH-14 are still under debate. Provided that the two CCM components have opposing effects on cellular ATP levels, it is crucial to investigate their differential regulation in response to different environmental conditions.


Despite the change in energy demand imposed by the different pCO2 levels and N sources, Trichodesmium showed no alteration in energy producing pathways. Yet, elevated pCO2 increased cellular POC and PON contents in both N treatments. In N2 fixers, also nitrogenase activity was strongly enhanced with pCO2. Concurrently, CCM activity was downregulated, reducing the use of ATP in active HCO3 uptake and allowing its allocation to N2 fixation. The increase in N2 fixation was, however, not reflected in PON production, possibly due to an increase in N loss with increasing pCO2. In NO3 users, the lower N-normalized ATP demand for PON production (Table 3) and the better N retention allowed for higher production rates of POC as well as PON compared with N2 fixers. A calculation of the theoretical energy demands of the measured POC and PON production rates (Table 3) revealed that most of the ATP saved from the switch to NO3 use (approximately 80%) was invested into increasing the production rates of POC and PON, resulting in almost unaltered ATP demand in our cultures (0.5 ATP residue, Table 3). The concomitant increase in the demand of reducing equivalents may have prevented a full implementation of ATP savings into the production of particulate organic matter (POM). The effects of pCO2 on CCM activity were smaller in NO3 users than in N2 fixers, highlighting the dependence of energy reallocation on the stoichiometric demands in energy equivalents: As NO3 assimilation requires only little ATP and is limited by electrons (Fig. 6), any spare ATP arising from downregulation of the CCM would not have the same stimulatory effect as in N2 fixers. Interestingly, the diurnal pattern in O2 fluxes usually attributed to protection of nitrogenase was maintained also in NO3 grown cells. Further studies are necessary to unravel the effects of different environmental conditions on cellular energy budgets, focusing on energization of the CCM as well as the intricate effects of the NDH-14 complex on C use efficiency and energy balance.

Author contributions

M. E., S. A. K. and B. R. conceived and designed the experiment. M. E. and S. A. K. performed the experiments. M. E. analyzed the data; M. E., S. A. K. and B. R. wrote the paper.


We thank Ulrike Richter, Jana Hölscher and Klaus-Uwe Richter for laboratory assistance and technical support. Grant support was provided by the European Research Council under the European Community's Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement (205150).