Introduction

The global distribution of dinitrogen (N2)-fixing cyanobacteria across the world’s oceans exhibits a striking relationship with temperature. Marine N2-fixing cyanobacteria occur in high numbers only in the tropics and subtropics, at water temperatures above 20 °C, whereas they are virtually absent from temperate and polar regions (Staal et al., 2003; Stal, 2009). This temperature-related geographic pattern seems to hold for different phylogenetic groups of oceanic N2-fixing cyanobacteria, such as the filamentous Trichodesmium (Capone et al., 1997; Lugomela et al., 2002; Chen et al., 2003; Langlois et al., 2005, 2008), unicellular cyanobacteria (UCYN) of the groups B and C (Mazard et al., 2004; Staal et al., 2007; Church et al., 2008; Langlois et al., 2008; Moisander et al., 2010) and the heterocystous symbionts of diatoms (Foster et al., 2007; Dore et al., 2008; Fong et al., 2008). Representatives of the uncultured symbiotic UCYN-A show peak abundances in subtropical waters, but it appears that their temperature range is somewhat broader (Needoba et al., 2007; Church et al., 2008; Langlois et al., 2008; Moisander et al., 2010).

The effect of temperature on the global distribution pattern of N2-fixing cyanobacteria might be direct or indirect. Indirect temperature effects may operate for instance through negative correlations of temperature with the availability of nitrate and ammonium, which may favor the dominance of N2 fixers in warm waters, because nitrogen fixation provides an important additional source of nitrogen (Monteiro et al., 2011). Furthermore, high surface temperature strengthens water column stratification, which may favor the dominance of floating cyanobacteria such as the diazotroph Trichodesmium (Sonntag and Hense, 2011). Temperature can also be directly responsible for the ecological success of N2 fixers in the tropical ocean, for example, if their nitrogenase activity is favored by high temperature. For instance, Breitbarth et al. (2007) showed that the temperature range for growth and N2 fixation of Trichodesmium IMS-101 in laboratory experiments matched the biogeographical distribution of Trichodesmium in the ocean. Similar experiments revealed that N2-fixing UCYN also have their temperature niche above 20 °C (Falcón et al., 2005; Webb et al., 2009).

A possible physiological mechanism explaining the thermal properties of N2-fixing cyanobacteria was proposed by Staal et al. (2003) and Stal (2009). These authors hypothesized that N2-fixing cyanobacteria are restricted to warm waters, because at low temperature the organism is unable to maintain sufficiently low levels of oxygen in the N2-fixing cell to avoid the inactivation of nitrogenase. Here we refine this hypothesis for the UCYN groups B and C, which contribute substantially to the global oceanic nitrogen budget (Zehr, 2011; Großkopf et al., 2012). Representatives of UCYN-B and -C exhibit a strong diurnal pattern in N2 fixation, respiration and photosynthesis (Peschek et al., 1991; Schneegurt et al., 1994b; Colón-López et al., 1997; Červený and Nedbal, 2009). In order to protect nitrogenase from high oxygen concentrations produced during photosynthesis, these organisms fix nitrogen typically during the night. Their nocturnal N2 fixation is fueled by respiration of the glycogen pool that has been built up by photosynthesis during the daytime (Schneegurt et al., 1994a; Dron et al., 2012).

As pointed out by Staal et al. (2003) and Stal (2009), respiration of glycogen requires oxygen, and UCYN therefore face an important trade-off. Oxygen levels in the cell should be sufficiently low to prevent inactivation of nitrogenase, yet the influx of oxygen should be sufficiently high to enable respiration for N2 fixation. Oxygen diffusion into cells shows only a weak temperature dependence (Figure 1a). In contrast, respiration and N2 fixation are enzymatic processes that, in potential, will increase faster with temperature than oxygen diffusion. Hence, at high temperature the actual N2-fixation activity may become limited by the oxygen influx (Figure 1a). Conversely, at low temperature nitrogenase activity will be the rate-limiting step for N2 fixation. However, the respiration rate needs to be high enough to protect nitrogenase against the diffusive influx of oxygen. A low nitrogenase activity in combination with sustained high levels of respiration results in high respiration costs per molecule of N2 fixed at low temperature (Figure 1b). We thus derived two testable hypotheses: (1) N2-fixation rates of UCYN will increase more strongly with temperature at low than at high temperature (Figure 1a) and (2) low temperature enhances the respiratory cost of N2 fixation (Figure 1b).

Figure 1
figure 1

Conceptual visualization of the two hypotheses explaining temperature dependence of N2 fixation in unicellular cyanobacteria. (a) Rates of oxygen diffusion (black dashed line) and potential N2 fixation (black solid line) as function of temperature. The actual N2-fixation rate (thick grey line) equals the potential N2-fixation rate at low temperature, but is limited by the oxygen diffusion rate at high temperature. The respiration rate (grey dashed line) matches the rate of oxygen diffusion into the cells in order to maintain a low-oxygen environment for nitrogenase. (b) Comparison of the respiration rate with the actual N2-fixation rate, in panel a, shows that the respiratory cost of N2 fixation (that is, the oxygen respired per molecule of N2 fixed) will increase strongly at low temperature. Background shading indicates temperature regions where nitrogen fixation is controlled by the reaction rate of nitrogenase (white background) or the rate of oxygen diffusion (dark background).

Full size image

In this study we test these hypotheses by investigating N2 fixation and respiration rates of the unicellular cyanobacterium Cyanothece BG 043511 at different temperatures. This strain assimilates nitrate if sufficient nitrate is available, but switches to N2 fixation if nitrate has been depleted (Agawin et al., 2007). Hence, by manipulating the nitrogen source, we can compare the respiration rates of Cyanothece under diazotrophic versus non-diazotrophic growth conditions to infer the respiratory cost of N2 fixation. In addition, we measured the light response of the N2 fixation and growth rates of Cyanothece at different temperatures. Our results show strong effects of temperature on the respiratory cost as well as the timing of N2 fixation in Cyanothece, which may offer a mechanistic explanation for the temperature dependence of the global biogeographical distribution of many other N2-fixing UCYN as well.

Materials and methods

Experimental design

The strain Cyanothece sp. Miami BG043511, which belongs to the group of UCYN-C, was grown in batch cultures with a volume of 50 ml in 250 ml Erlenmeyer flasks on artificial sea water medium enriched with nutrients (Mitsui and Cao, 1988). The mineral medium was provided either with 1.5 mM nitrate (non-diazotrophic condition) or without combined nitrogen (diazotrophic condition). The cultures were unialgal but not axenic. However, regular microscopic inspection confirmed that abundances of other bacteria remained low (<1% of the total biomass) for the entire duration of the experiments. The cultures were grown under an alternating 12:12 h light:dark cycle. A shaking cryostat bath controlled the temperature and guaranteed continuous mixing of the cultures. We investigated the growth rate and N2-fixation rate of Cyanothece at seven different temperatures from 14 to 38 °C, nine light intensities from 3 to 133 μmol photons m−2 s−1 and with or without nitrate in the mineral medium using a full factorial design. This resulted in 7 × 9 × 2=126 different experimental treatments. Before the experiments, the Cyanothece cultures were acclimated to the applied temperature and nitrogen source for at least 1 week, using an intermediate light intensity of ∼30 μmol photons m−2 s−1.

Growth rates

The specific growth rate of each population was determined during the exponential growth phase. We used one batch culture per experimental treatment. Cell numbers in this batch culture were counted in triplicate on 5 consecutive days using an automated cell counter (Casy Cell Coulter, Schaerfe System GmbH, Reutlingen, Germany). Specific growth rates were calculated as the slope of the regression line of the natural logarithm of population density versus time.

At each experimental temperature, the light response of the specific growth rate, μ(I), was described by the Monod equation (Monod, 1950; Huisman, 1999):

where I is light intensity, μmax is the maximum specific growth rate and α is the initial slope of the μ(I) function. Equation (1) was fitted to the measured growth rates using non-linear regression with minimization of the residual sum of squares (R version 2.13.1).

Subsequently, the temperature dependence of the maximum specific growth rate was described by the Arrhenius equation (Gillooly et al., 2001):

where c is a normalization constant, E is the activation energy, k is Boltzmann’s constant (8.62 × 10−5 eV K−1) and T is the absolute temperature in Kelvin. The activation energy quantifies the temperature dependence, and was calculated as the slope of the linear regression of ln(μmax) plotted against the inverse of temperature, 1/kT. Activation energy was converted to Q10-values according to Vasseur and McCann (2005).

Nitrogen fixation rates

N2-fixation rate was determined during the exponential growth phase using the acetylene reduction assay (Hardy et al., 1968). For each experimental treatment, whole-night N2 fixation was measured in duplicate incubations over a 12-h dark period. In addition, we used six consecutive incubations of 2 h each to follow the temporal pattern of nocturnal N2 fixation. Whole-night incubations were done in 5-ml crimp-top vials (Chrompack, Middelburg, The Netherlands) using 1 ml of Cyanothece culture and a headspace acetylene concentration of 20%. For the 2-h incubations, 3 ml of culture and 11% acetylene was used. After incubation, gas samples of 300 μl were withdrawn from the headspace with a disposable syringe and injected into a gas chromatograph (GC14A, Shimadzu, Kyoto, Japan) to measure the concentrations of ethylene and acetylene. The gas chromatograph was equipped with a flame ionization detector and a 25-m long wide-bore silica-fused Porapak U column (0.53 mm inner diameter; Chrompack). The temperatures of oven, injector and detector were set to 60, 100 and 150 °C, respectively. N2 fixation was expressed as acetylene reduction, and was calculated according to Stal (1988) using acetylene as internal standard.

Analogous to the specific growth rates, the light response of N2 fixation was fitted to the Monod equation (Equation 1) by expressing the whole-night N2 fixation as a function of the light intensity provided during the preceding daytime. Subsequently, the maximum whole-night N2 fixation obtained from the Monod equation (nmax) was fitted to the Arrhenius equation (Equation 2) to quantify its temperature dependence.

Respiration rates

Respiration was monitored during the 12-h dark period using automated recording of oxygen consumption. Cyanothece was grown in Erlenmeyer flasks under high-light conditions (130 μmol photons m−2 s−1) and a 12:12 h light:dark cycle, at five different temperatures (from 18 to 30 °C) under both N2-fixing and non-N2-fixing conditions. The cultures were continuously mixed with magnetic stirrers. Each Erlenmeyer flask was connected to a temperature-controlled flow-through cuvette with a chamber volume of ∼5 ml. Every 30 min a peristaltic pump (Watson Marlow 101U, Falmouth, UK) connected to a computer-controlled USB-time relay (H-TRONIC, Hirschau, Germany) flushed the flow-through cuvette during 7 min with a new air-saturated sample taken from the Erlenmeyer flask. Subsequently, the pump was stopped and the decline in oxygen concentration was measured online during 10 min by using optical oxygen sensors connected to a fiber optic oxygen transmitter (Oxy-4-mini, PreSens, Regensburg, Germany). The respiration rate was calculated as the slope of the regression line of oxygen concentration versus time.

Cellular carbon and nitrogen

At the beginning and end of the 12-h dark period, 5 to 10 ml of the high-light cultures were filtrated on pre-combusted glass fiber filters (GF/F, Watson). The filters were rinsed with demineralized water and stored at −20 °C. Amounts of organic carbon and nitrogen on each filter were determined with an element analyzer (NA-2500, Thermo Scientific, Whaltham, MA, USA) to determine cellular carbon and nitrogen contents.

Cost of nitrogen fixation

We used two measures to quantify the respiratory cost of N2 fixation. First, we calculated the gross cost of N2 fixation (Cgross), which we defined as the whole-night respiration of N2-fixing cultures divided by the whole-night N2 fixation:

where rN2(t) is the respiration rate of the N2-fixing culture as function of time, and Nf(t) is its N2-fixation rate. Accordingly, the gross cost of N2 fixation includes respiration associated directly with N2 fixation as well as respiration required for the maintenance of other cellular processes.

Second, we calculated the net cost of N2 fixation (Cnet), defined as the difference between the whole-night respiration of N2-fixing and non-N2-fixing cultures divided by the whole-night N2 fixation:

where rNO3(t) is the respiration rate of the non-N2-fixing culture as function of time. Hence, the net cost of N2 fixation excludes the maintenance cost associated with other metabolic processes. For both measures, a conversion factor of 4 was used to calculate N2 fixation from acetylene reduction (for example, Stal, 1988).

Results

Growth and nitrogen fixation

The specific growth rate of Cyanothece was an increasing decelerating function of light intensity (Figure 2). The exact response to light, however, depended on temperature and nitrogen source. At 14 °C, Cyanothece showed only marginal growth under non-N2-fixing conditions, while we were unable to grow Cyanothece diazotrophically at this temperature (Figure 2a). At 18 and 22 °C, Cyanothece grew well under both N2-fixing and non-N2-fixing conditions, and the specific growth rate already leveled off at a light intensity of 40–50 μmol photons m−2 s−1 (Figures 2b and c). At 34 and 38 °C, the specific growth rate had not yet leveled off at a light intensity of >150 μmol photons m−2 s−1 (Figures 2f and g).

Figure 2
figure 2

Light-response curves of the specific growth rate at different temperatures. (a) 14 °C, (b) 18 °C, (c) 22 °C, (d) 26 °C, (e) 30 °C, (f) 34 °C, (g) 38 °C. The panels compare the light response of N2-fixing cultures (open circles) versus cultures grown on nitrate (closed circles). Each data point is based on a time series of five measurements. The specific growth rate was calculated as the slope of the linear regression of ln[population density] versus time; error bars represent the s.e. of the slope (N=5). Solid lines show light-response curves fitted by the Monod equation.

Full size image

Whole-night N2 fixation in cultures without nitrate was also an increasing decelerating function of the light intensity received during the preceding daytime, and varied with temperature (Figure 3). N2 fixation (acetylene reduction) was not detectable in the cultures with nitrate (results not shown), which confirmed that Cyanothece grows non-diazotrophically when provided with sufficient nitrate. For the diazotrophic cultures, we compared the amount of nitrogen fixed per cell calculated from acetylene reduction (Nfix) with the accumulation of nitrogen in cells calculated from the cellular nitrogen contents at the beginning and end of the dark period (Nacc). Both values generally agreed quite well, with an average ratio of Nfix/Nacc=0.86 (s.e.=0.22; N=63), which illustrates the consistency of the data.

Figure 3
figure 3

Light-response curves of whole-night N2 fixation at different temperatures. (a) 14 °C, (b) 18 °C, (c) 22 °C, (d) 26 °C, (e) 30 °C, (f) 34 °C, (g) 38 °C. Each data point represents an independent incubation. N2 fixation is expressed as acetylene reduction, integrated over the entire 12-h night. Light intensities were measured during the preceding daytime. Solid lines show light-response curves fitted by the Monod equation.

Full size image

The light responses were fitted to the Monod equation (solid lines in Figures 2 and 3), and the parameter values obtained from these Monod fits are plotted as function of temperature in Figure 4. Up to 30 °C, the maximum specific growth rate of the N2-fixing and non-N2-fixing cultures increased with temperature in a similar manner (Figure 4a), with activation energies of 0.81 eV (95% confidence interval: 0.76–0.85 eV, df=3, P<0.001; Q10=2.89) and 0.90 eV (95% confidence interval: 0.70–1.11 eV, df=3, P=0.0134; Q10=3.28) under N2-fixing and non-N2-fixing conditions, respectively. The maximum specific growth rate continued to increase with temperature up to 34 °C under non-N2-fixing conditions, whereas it leveled off above 30 °C under N2-fixing conditions. Hence, at high temperature, Cyanothece reached much higher specific growth rates under non-N2-fixing than under N2-fixing conditions.

Figure 4
figure 4

Monod parameters for the light response of the specific growth rate and of the whole-night N2 fixation, plotted as function of temperature. (a) Maximum specific growth rate, μmax; (b) maximum whole-night N2-fixation, nmax; (c) initial slope, αμ, of the light-response curves of the specific growth rates, (d) initial slope, αn, of the light-response curves of whole-night N2 fixation. Open circles represent N2-fixing cultures; closed circles represent cultures grown on nitrate. The parameter estimates are based on the Monod fits shown in Figure 2 (N=9) and Figure 3 (N=18). Error bars represent s.e. of the parameter estimates.

Full size image

Maximum whole-night N2 fixation showed a unimodal response to temperature (Figure 4b). At 14 and 18 °C, whole-night N2 fixation did not exceed the detection limit in most cases (see Figure 3b). Maximum whole-night N2 fixation increased strongly with temperature from 22 to 30 °C, with an activation energy of 1.27 eV (95% confidence interval: 0.05–2.5 eV, df=2, P=0.29; Q10=5.2). Above 30 °C, however, maximum N2 fixation decreased with temperature (Figure 4b).

The initial slope of the light-dependent growth rate (αμ) at 18 and 22 °C was higher for non-N2-fixing cultures than for N2-fixing cultures (Figure 4c). Above 22 °C, the initial slope of the light-dependent growth rate seemed largely temperature independent, and was essentially similar under both N2-fixing and non-N2-fixing conditions. The initial slope of the light-dependent whole-night N2 fixation (αn) showed considerable variation, but seemed also largely independent of temperature (Figure 4d). The low value of αn at 18 °C might be an outlier due to the technical difficulty to obtain an accurate light-response curve from the low N2-fixation rates measured at this temperature (cf. Figure 3b).

Nocturnal patterns of nitrogen fixation and respiration

As a next step, we investigated how temperature affects temporal patterns of N2 fixation (acetylene reduction) and respiration (oxygen consumption) during the 12-h dark period. This revealed that N2 fixation and respiration exhibited very similar nocturnal patterns (Figure 5). In most cases, N2 fixation started after the onset of darkness and ceased before daybreak. However, the precise timing of the N2-fixation activity was strongly determined by temperature. The N2-fixation peak shifted from the end of the dark period at 18 °C to the beginning of the dark period at 30 °C (Figures 5a–e). N2 fixation reached the highest rate at intermediate temperatures from 21 to 27 °C (Figures 5b–d). At each temperature, N2-fixation rates in the night increased with the light intensity provided during the preceding daytime, but light intensity did not affect the temporal pattern of N2 fixation (Figures 5a–e).

Figure 5
figure 5

Nocturnal patterns of N2 fixation (acetylene reduction) and respiration (oxygen consumption) at different temperatures. (a–e) The N2-fixation rate was measured during six consecutive intervals of two hours each, covering the entire 12-h dark period. The different light intensities provided during daytime are indicated by different colors. (f–j) Respiration rate of N2-fixing cultures (open circles) and non-N2-fixing cultures (closed circles) during the 12-h dark period, after exposure to a daytime light intensity of 130 μmol photons m−2 s−1. N2 fixation and respiration rates were measured in separate cultures.

Full size image

The temporal pattern of respiration reflected that of N2 fixation (Figures 5f–j). Respiration rates of N2-fixing cultures reached a maximum during the night, and the timing of this maximum shifted from the end of the dark period at low temperature to the beginning of the dark period at high temperature (open circles in Figures 5f–j). The highest respiration rates were found at intermediate temperatures (Figures 5g–i). In contrast, respiration rates of non-N2-fixing cultures remained low and were basically constant during the dark period (closed circles in Figures 5f–j).

Cost of nitrogen fixation

We introduced two measures of the respiratory cost per molecule of N2 fixed. The gross cost of N2 fixation considers the whole-night respiration of the N2-fixing culture (that is, the area underneath the open circles in Figures 5f–j integrated over the 12 h-dark period), whereas the net cost considers the difference in whole-night respiration between the N2-fixing and non-N2-fixing culture (that is, the area between the open and closed circles in Figures 5f–j integrated over the 12-h dark period). Both measures strongly increased with decreasing temperature (Figure 6a). At 18 °C, the gross cost of N2 fixation was about five times higher than at 30 °C, whereas the net cost of N2 fixation was about four times higher. Likewise, the C:N ratio of the cells determined at the end of the 12-h dark period increased at low temperature, particularly for the N2-fixing cultures at 14 and 18 °C (Figure 6b).

Figure 6
figure 6

Respiratory cost of N2 fixation. (a) Gross respiratory cost (closed triangles) and net respiratory cost (open squares) of N2 fixation as function of temperature. Gross respiratory cost represents the ratio of whole-night respiration over whole-night N2 fixation, net respiratory cost represents the same ratio but corrected for respiration not associated with N2 fixation. (b) Cellular C:N ratio of N2-fixing (open circles) and non-N2-fixing cultures (closed circles). The C:N ratios were measured at the end of the 12-h dark period in cultures exposed to 80–130 μmol photons m−2 s−1 during the preceding daytime.

Full size image

Discussion

Mechanisms controlling temperature dependence

The results show that temperature has major effects on growth, N2 fixation and respiration of the unicellular N2-fixing cyanobacterium Cyanothece. In agreement with the first hypothesis, displayed in Figure 1a, the maximum specific growth rate of Cyanothece grown under N2-fixing conditions showed a strong temperature dependence at low temperature but a weak temperature dependence at high temperature (Figure 4a). The switch from strong to weak temperature dependence occurred at 30 °C, suggesting that the diffusive influx of oxygen (and possibly N2) limited a further increase of the diazotrophic growth rate at higher temperatures. In contrast, the specific growth rate under non-N2-fixing conditions still increased steeply with temperature at this temperature range, demonstrating that cells were capable of a much higher growth rate if they were not limited by the N2-fixation process (Figure 4a).

Although the specific growth rate leveled off above 30 °C, the maximum whole-night N2 fixation declined above this temperature. Gallon et al. (1993) observed a similar decline of the N2-fixation rate in the cyanobacteria Anabaena cylindrica and Gloeothece. Their results indicate that the decline is probably not caused by thermal inactivation of nitrogenase, because in vitro studies showed that thermal inactivation of nitrogenase required much higher temperatures. Instead, Gallon et al. (1993) attributed the decreased N2-fixation rate to an enhanced sensitivity of nitrogenase to oxygen inhibition at elevated temperature. Another possible explanation is that the observed decrease of the acetylene reduction rate above 30 °C does not represent a decrease of the actual N2-fixation rate. This latter explanation would be compatible with the observation that the growth rate did not decline (Figure 4a) and cellular C:N ratio remained low (Figure 6b) above 30 °C.

Respiratory cost of nitrogen fixation

Our findings support the hypothesis, in Figure 1b, that the respiratory cost of N2 fixation increases at low temperature (Figure 6a). One might argue that our Cyanothece cultures were not axenic, and that other bacteria may have confounded the respiration measurements. However, our cultures were provided with a mineral medium designed exclusively for photoautotrophic organisms, without suitable substrates to respire by heterotrophic bacteria. Microscopic inspection confirmed only minor contamination by other bacteria, suggesting that their contribution to the total respiration rate was probably small compared with that of Cyanothece. Furthermore, respiration rates under N2-fixing conditions were much higher than in the nitrate-grown cultures, and showed the same temporal pattern as the nitrogenase activity. This clearly indicates a predominant contribution of Cyanothece to the respiration rate, because one would not expect such patterns when the respiration rate would be dominated by contaminating heterotrophic bacteria. Finally, the net respiratory cost of N2 fixation calculated with Equation (4) is based on the difference in respiration rate between N2-fixing and nitrate-grown cultures. Hence, this calculation removes the background respiration not associated with N2 fixation, such as the respiratory contributions of contaminating bacteria. Our results therefore support unequivocally the hypothesis that the respiratory cost of N2 fixation increases with decreasing temperature, as predicted by theory.

It is well known that N2-fixing cyanobacteria often have elevated respiration rates compared with non-N2-fixing organisms (for example, Wastyn et al., 1988; Peschek et al., 1991). This is typically ascribed to the high energy demand of N2 fixation (Bergman et al., 1997). However, the observed increase in respiratory cost at low temperature suggests that part of the respiratory oxygen consumption is independent of the energy demand of N2 fixation, and is used to maintain low intracellular oxygen levels. Großkopf and LaRoche (2012) estimated that the N2-fixing UCYN Crocosphaera watsonii invested only 40% of the energy obtained from respiration directly into N2 fixation and maintenance metabolism, while up to 60% of the respiratory cost represented an indirect cost for removal of intracellular oxygen or reversal of oxidative damage, for example, by de novo nitrogenase synthesis. Our results show that this indirect cost varies with temperature. More specifically, the fourfold rise in net respiratory cost above the baseline value (Figure 6a) suggests that at low temperature, the indirect cost associated with oxygen removal may constitute up to 80% of the total cost of N2 fixation.

We currently do not know which oxygen-scavenging mechanisms are present in Cyanothece sp. BG 043511. One possible strategy is ‘respiratory protection,’ where organisms possess a branched electron transport chain with multiple terminal oxidases, as has been shown for a number of bacteria and the unicellular N2-fixing cyanobacterium Gloeothece (Maryan et al., 1986; Gallon, 1992). An alternative strategy is ’autoprotection,’ where the nitrogenase enzyme protects itself by reducing oxygen to H2O and H2O2 (Gallon, 1992; Bergman et al., 1997).

High respiratory cost of N2 fixation at low temperature have also been found in diazotrophic symbionts of terrestrial plants. Using Frankia-infected root nodules of alder, Winship and Tjepkema (1985) showed that nitrogenase activity decreased faster with temperature than the respiration rate. As a consequence, the respiratory cost per molecule of fixed N2 rose sharply at low temperature, similar to our findings. This indicates that the high respiratory cost of N2 fixation at low temperature, required to maintain low intracellular oxygen concentrations, might be a common challenge for N2-fixing organisms in both aquatic and terrestrial habitats.

Timing of dinitrogen fixation

In addition to the predicted results, we found that temperature has also a remarkable effect on the timing of N2 fixation. The timing of N2 fixation and many other processes involved in the nitrogen and carbon metabolism of the genus Cyanothece is under the tight control of a circadian clock. N2 fixation and respiration show a distinct 24-h periodicity, and photosynthesis peaks about 12 h out of phase (Peschek et al., 1991; Schneegurt et al., 1994b; Colón-López et al., 1997; Červený and Nedbal, 2009). The circadian oscillations of N2-fixation and other processes are expressed independently of external triggers, under continuous light (Colón-López et al., 1997; Toepel et al., 2008; Červený and Nedbal, 2009), under continuous darkness (Schneegurt et al., 1994b) and under various day–night cycles (Červený and Nedbal, 2009; Toepel et al., 2009). Studies on the phosphorylation cycle of the cyanobacterial clock protein KaiC demonstrated that the period of oscillations is independent of nutrient status and temperature (Nakajima et al., 2005; Tomita et al., 2005). Likewise, our study shows that differences in light intensity during the daytime do not affect the timing of N2 fixation during the night.

However, decreasing the temperature strongly delays the onset of N2 fixation (Figure 5). A possible explanation might be that the synthesis of nitrogenase proceeds at a lower rate at low temperature. In N2-fixing UCYN nitrogenase is inactivated by oxygen during the day and has to be resynthesized de novo every night (Mullineaux et al., 1981; Huang et al., 1988; Gallon, 1992; Bergman et al., 1997). Thus, even if the circadian clock is temperature-compensated and always initiates the synthesis of nitrogenase at exactly the same time, there might be a temperature-dependent delay in the delivery of the functional enzyme. This may cause a decrease in N2-fixation capacity if low temperature would delay the synthesis of active nitrogenase to such an extent that the remaining nighttime becomes insufficient to cover the nitrogen demands of the cell. The observed delay in the timing of N2 fixation may thus contribute to the steep decline of the whole-night N2-fixation activity (Figure 4b) and to the high cellular C:N ratio (Figure 6b) at low temperature.

Biogeographical distribution

Although many environmental factors may influence the distribution of N2-fixing cyanobacteria, such as the availability of dissolved nitrogen, iron and phosphorus (for example, Sañudo-Wilhelmy et al., 2001; Mills et al., 2004; Monteiro et al., 2011), there is ample evidence that temperature has also an important role (Stal, 2009). According to our laboratory results, Cyanothece can grow over a temperature range from 18 to at least 38 °C. This is in good agreement with field observations, which indicate that N2-fixing UCYN-B and -C occur only in natural waters above 20 °C (Mazard et al., 2004; Langlois et al., 2005; Staal et al., 2007; Church et al., 2008; Langlois et al., 2008; Moisander et al., 2010). Their global biogeographical distribution might thus reflect the direct effect of temperature on the physiology and growth of these organisms (see also Breitbarth et al., 2007). In nature, competitive interactions with other species may further restrict the temperature range of N2-fixing UCYN, particularly below 21 °C when the cost of N2 fixation becomes high. Our experiments thus suggest that the lower temperature limit of N2-fixing UCYN-B and -C is due to the enhanced respiratory cost and strongly delayed timing of N2 fixation at low temperature.

Together, these results provide novel experimental evidence for the general hypothesis that the temperature sensitivity of N2 fixation is a key determinant of the biogeographical distribution of N2-fixing UCYN across the world’s oceans. An important implication of these findings is that global warming is likely to facilitate the expansion of N2-fixing UCYN to higher latitudes.