Is the distribution of nitrogen-fixing cyanobacteria in the oceans related to temperature?

Authors


*E-mail l.stal@nioo.knaw.nl; Tel. +31 113 577497; Fax +31 113 573616.

Summary

Approximately 50% of the global natural fixation of nitrogen occurs in the oceans supporting a considerable part of the new primary production. Virtually all nitrogen fixation in the ocean occurs in the tropics and subtropics where the surface water temperature is 25°C or higher. It is attributed almost exclusively to cyanobacteria. This is remarkable firstly because diazotrophic cyanobacteria are found in other environments irrespective of temperature and secondly because primary production in temperate and cold oceans is generally limited by nitrogen. Cyanobacteria are oxygenic phototrophic organisms that evolved a variety of strategies protecting nitrogenase from oxygen inactivation. Free-living diazotrophic cyanobacteria in the ocean are of the non-heterocystous type, namely the filamentous Trichodesmium and the unicellular groups A–C. I will argue that warm water is a prerequisite for these diazotrophic organisms because of the low-oxygen solubility and high rates of respiration allowing the organism to maintain anoxic conditions in the nitrogen-fixing cell. Heterocystous cyanobacteria are abundant in freshwater and brackish environments in all climatic zones. The heterocyst cell envelope is a tuneable gas diffusion barrier that optimizes the influx of both oxygen and nitrogen, while maintaining anoxic conditions inside the cell. It is not known why heterocystous cyanobacteria are absent from the temperate and cold oceans and seas.

Introduction

Current estimates of the global primary production on Earth indicate that approximately 50% occurs in the oceans (Whitman et al., 1998). It is therefore of major importance for global biogeochemical cycles and the changes they presently experience. Nitrogen is often pointed out as the factor that limits primary production in the ocean (Vitousek and Howarth, 1991). In the euphotic zone of the water column the bioavailable nitrogen (i.e. nitrate, nitrite, ammonium, urea and amino acids, excluding dinitrogen gas) is quickly depleted as a result of the heavy demand of the phytoplankton, growth of which in part depends on the regeneration of nitrogen (Dugdale and Goering, 1967). In the surface-mixed layers of the oligotrophic oceans the concentrations of nitrate and nitrite are usually below the limit of detection (Cavender-Bares et al., 2001). Part of the nitrogen [estimated 25 Tg N year−1 (Gruber and Sarmiento, 1997)] is lost as particulate organic nitrogen from the euphotic zone by sedimentation and is mineralized at greater depth. The concentrations of nitrate are therefore higher at greater depth below the euphotic zone. In upwelling zones this nitrate is transported into the surface waters where it is used by the phytoplankton. This nitrate is often referred to as ‘new’ nitrogen in order to distinguish it from nitrogen that is regenerated within the phytoplankton community.

Nitrate and nitrite are subject to denitrification converting it to gaseous dinitrogen (N2) and to some extent nitrous oxide (N2O), which escape to the atmosphere. Denitrification results in a loss of about 175 Tg N year−1 as N2 and another 4 Tg N year−1 is lost as N2O through various processes (Gruber and Sarmiento, 1997). Anaerobic ammonium oxidation converts ammonium and nitrite into N2. This process could also contribute considerably to the loss of combined nitrogen from the oceans but its magnitude is uncertain (Ward, 2003).

These losses of nitrogen are compensated in part by riverine discharge and atmospheric deposition that account for 76 and 30 Tg N year−1 respectively (Gruber and Sarmiento, 1997; Karl et al., 2002; Galloway et al., 2004). Biological N2 fixation accounts for an estimated 125 Tg N year−1 which represents some 50% of the total global fixation. This estimate of N2 fixation is not based on actual measurements but relies on a semi-conservative parameter, N*, the concentration of nitrate in excess or deficit of phosphate relative to the Redfield stoichiometry (Gruber and Sarmiento, 1997). Actual measurements of N2 fixation in the oceans have largely been made on the filamentous cyanobacterium Trichodesmium, and extrapolating these based on observations of their global distributions. Rates obtained this way are considerable lower than those obtained by the N* approach. Assuming that the oceans are not losing their nitrogen and that the N* approach gives more reliable estimates of N2 fixation, it must be concluded that the current measurements are missing out a considerable part of the actual fixation that is going on in the oceans. New discoveries support this conclusion (Montoya et al., 2004; Needoba et al., 2007).

The biological fixation of N2 is carried out by Bacteria and Archaea (‘Prokaryotes’) that possess the enzyme complex nitrogenase. Nitrogenase is a two-component enzyme consisting of dinitrogenase reductase (also known as the Fe protein) and dinitrogenase (also known as the Mo-Fe protein). The former reduces the latter enzyme using a low-potential electron donor (usually reduced ferredoxin) at the expense of two molecules of ATP per electron. Dinitrogenase subsequently reduces the triple bond of the two nitrogen atoms producing two molecules of ammonia and one molecule H2, which is an obligatory side product of N2 fixation. In many aerobic N2-fixing organisms, H2 is recycled by an uptake hydrogenase and respired. No member of the Eukarya is capable of fixing N2 but various groups have entered symbiotic relationships with diazotrophic Bacteria. Such symbiotic diazotrophic associations between Eukarya and Bacteria are ecologically successful, apparently eliminating the need for the former to evolve their own N2-fixing system.

The fixation of N2 requires a large amount of energy in the form of eight low-potential electrons and 16 ATP. Cyanobacteria are oxygenic photoautotrophic Bacteria that have access to more or less infinite source of energy (sunlight) and electrons (water). Therefore, on the one hand it is not surprising that many of these organisms are potent N2 fixers and that the vast majority of diazotrophs in the oceans are cyanobacteria, but on the other hand it seems paradoxical because nitrogenase is extremely sensitive to oxygen and requires a virtually oxygen-free environment (Gallon, 1992). Hence, diazotrophic cyanobacteria had to invent ways to protect nitrogenase from photosynthetic evolved and environmental oxygen.

It is remarkable that diazotrophic cyanobacteria are only abundantly present in the tropical or subtropical oceans. Hitherto, N2 fixation has not been detected in the pelagic of the temperate or cold oceans. Even when N* values might indicate that N2 fixation could occur in oceanic provinces outside the tropics, measurements to support this are lacking. Moreover, N* values may also be accomplished by transport of water masses and are therefore not necessarily indicative for N2 fixation at the location where positive values of N* are found (Hansell et al., 2004). It is well known that diazotrophic cyanobacteria occur in freshwater lakes and brackish basins and estuaries all over the world, indicating that temperature per se is not a factor that excludes these organisms. This review will summarize what is known about the distribution of diazotrophic cyanobacteria in the sea and attempts to find answers to the question what determines these distributions.

N2-fixing cyanobacteria

Not all cyanobacteria possess nitrogenase and those that do not are therefore incapable of fixing N2. Cyanobacteria that do fix N2 evolved strategies allowing them to provide an anoxic environment for nitrogenase while performing the incompatible oxygenic photosynthesis. Diazotrophic cyanobacteria can be subdivided into three main groups (Stal, 1995). The strategy of the first group can be described by ‘avoidance’ (of O2). These filamentous or unicellular cyanobacteria fix N2 only under anaerobic or microaerobic conditions while oxygenic photosynthesis is (largely) inhibited. For instance, sulfide-rich environments are anoxic and the sulfide inhibits photosystem II (PS-II) or may support anoxygenic photosynthesis in some cyanobacteria (Howsley and Pearson, 1979; Cohen et al., 1986). Such cyanobacteria may also migrate between anoxic (and anoxygenic) conditions and aerobic (and oxygenic) photosynthetic growth, while N2 fixation is confined to the former and thus spatially and temporally separated from O2 evolving photosynthesis (Stal, 1995).

The second group of diazotrophic cyanobacteria comprises filamentous forms that confine N2 fixation to special differentiated cells, the heterocysts (Adams, 2000). The heterocyst is devoid of PS-II and is therefore unable to perform oxygenic photosynthesis, although it can harvest light and generate biochemical energy through PS-I-mediated cyclic phosphorylation. The heterocyst is unable to fix CO2 and depends on the neighbouring vegetative cells for reducing equivalents, provided as sucrose (Curatti et al., 2002), and which in return receive the fixed nitrogen. The heterocyst is enveloped by complex glycolipid layers that act as a gas diffusion barrier, limiting the entrance of O2 (Walsby, 1985). Any O2 entering the heterocyst is scavenged by an efficient and high-affinity respiratory system, rendering the heterocyst interior virtually anoxic. This strategy can be described as ‘spatial separation’ (of oxygenic photosynthesis and CO2 fixation in the vegetative cells and N2 fixation in the heterocyst). This is a very efficient way of combining the incompatible processes of oxygenic photoautotrophic and diazotrophic growth. While heterocystous cyanobacteria fix N2 in the light, concomitantly with oxygenic photosynthesis, many continue the fixation of N2 during the dark, supported by respiration and the import of reducing equivalents from the vegetative cells. However, the reports relating to this matter are not consistent. While some heterocystous cyanobacteria seem to be quite efficient in dark fixation of N2, others seem to shut down this process during the night. Hence, there may be species-specific differences or their behaviour might differ with the circumstances they are exposed to. The differentiation of the heterocyst is a complex process and is still not completely understood. It has been suggested that it evolved as a result of the oxygenation of the atmosphere some 2.500 million years ago (Tomitani et al., 2006).

The third group of diazotrophic cyanobacteria fixes N2 aerobically even although they are non-heterocystous. This group comprises filamentous as well as unicellular species. The strategy by which these cyanobacteria are able to combine photoautotrophic and diazotrophic growth has been described as ‘temporal separation’ or a combination of spatial and temporal separation of N2 fixation and oxygenic photosynthesis but the strategies within this group may be several and are still incompletely understood. The best studied species in this group of aerobic non-heterocystous N2-fixing cyanobacteria are the filamentous Trichodesmium and Lyngbya and the unicellular Gloeothece and Cyanothece (Bergman et al., 1997).

Trichodesmium forms extensive surface blooms in the tropical and subtropical oceans and is considered to be the most important marine diazotroph (Capone, 2001). This organism has been known for a long time but only because Trichodesmium was successfully grown in culture (Ohki et al., 1986; Prufert-Bebout et al., 1993) and with the application of molecular biological techniques, our knowledge about this remarkable diazotroph has rapidly increased. The fixation of N2 in Trichodesmium depends strongly on light and the daily pattern is very similar to that of heterocystous cyanobacteria with the important difference that no fixation occurs during the night (Capone et al., 1990; Staal et al., 2007a). In fact, nitrogenase is inactivated and subsequently turned over at night and synthesized de novo each day. This process has been shown to be under the control of a circadian clock (Chen et al., 1998). The various reports disagree about the question whether all cells in a trichome of Trichodesmium are (capable of) fixing N2 or that this is reserved to a subset of special cells. By using immunolocalization of nitrogenase Fredriksson and Bergman (1997) discovered that the enzyme was present only in a subset of adjacent cells in the trichome of Trichodesmium. They supposed that these were differentiated cells analogous to heterocysts and termed them ‘diazocytes’. However, using an isolate of Trichodesmium, Ohki (2008) did not find evidence of diazocytes and observed nitrogenase in virtually all cells. The immunological detection of nitrogenase does not proof whether it is active and fixes N2 or even whether the antigen is in an active state. There is little doubt that nitrogenase can only be active under virtually anoxic conditions and this does not seem to be compatible with PS-II activity. It is therefore possible that Trichodesmium cells switch between N2 fixation and oxygenic photosynthesis in a fashion that can be described as a combination of ‘temporal and spatial separation’ (Berman-Frank et al., 2001). From single cell fluorescence measurements Küpper and colleagues (2004) concluded that certain cells could rapidly switch between N2 fixation and PS-II activity. This model is not necessarily in contradiction with the immunological observations but makes the concept of irreversible unidirectional cell differentiation such as is the case with heterocysts unlikely. In addition to the cessation of PS-II activity, the diazotrophic cell needs to scavenge any O2 that diffuses into it. Whether this is governed by respiration or by the Mehler reaction, or other O2-scavenging mechanisms, or a combination is currently not precisely known (Staal et al., 2007a).

The unicellular Cyanothece belongs to the ‘Group C’ of the marine unicellular diazotrophic cyanobacteria (Foster et al., 2007; Needoba et al., 2007). Although it is frequently encountered in the tropical oceans, in most cases it does not appear to be abundant. It has also been found as a cyanobiont in a eukaryotic alga, as epiphyte and in benthic mats and biofilms. Cyanothece is a typical example of a diazotrophic cyanobacterium that separates N2 fixation temporally from photosynthesis by confining the former to the dark period (Sherman et al., 1998). The day–night pattern of N2 fixation in this organism is under the control of a circadian rhythm. When transferred to continuous light Cyanothece maintains the cyclic behaviour of nitrogenase activity and gene transcription (Colón-López et al., 1997).

Not much is known about the fixation of N2 by the ‘Group B’ representative of a marine unicellular diazotrophic cyanobacterium, Crocosphaera. Similar as Cyanothece, Crocosphaera also fixes N2 during the dark. Natural communities of Group B nifH phylotypes express this gene during the night (Church et al., 2005). Group B nifH phylotypes are common but rarely abundant. Contrary to Cyanothece, Group B cyanobacteria seem to be typical free-living planktonic organisms. ‘Group A’ organisms do not have cultivated representatives and are only known from their nifH sequences. On the basis of nifH phylogeny Group A is suspected to be a unicellular cyanobacterium. Based on size fractionation and fluorescent in situ hybridization, Group A may belong to the picoplanktonic fraction but they may also occur in aggregates (Biegala and Raimbault, 2008). Zehr and colleagues (2008) carried out metagenome sequencing on cells that were sorted by flow cytometry and thereby were enriched in ‘Group A’ phylotypes. Surprisingly, these organisms lacked the genes coding for PS-II, but possessed those for PS-I, suggesting a photoheterotrophic mode of life of ‘Group A’ organisms. This sheds new light on the observation that, in contrast to the Group B and C cyanobacteria, the highest expression of Group A nifH occurs during the day (Church et al., 2005) and also the actual fixation of N2 may occur during daytime (Montoya et al., 2004). After Trichodesmium, organisms with the Group A nifH phylotype are the most abundant diazotrophs in the tropical oceans. Until cultivated representatives become available it will be difficult to understand the strategy by which these organisms fix N2.

Gloeothece, another unicellular cyanobacterium from freshwater and terrestrial environments, is known to fix N2 since 1969 when Wyatt and Silvey (1969) published their paper in Science, one year after Fay and colleagues (1968) showed that the heterocyst was the site of N2 fixation. Most of our knowledge on this organism is based on the work of J.R. Gallon (Stal, 2003; Stephens et al., 2003). When grown under alternating light–dark cycles, Gloeothece fixes N2 during the dark. However, this organism grows also diazotrophically in continuous light, but still shows a pattern of temporal separation from oxygenic photosynthesis. N2 fixation in this organism does not seem to be under the control of a circadian clock. Interestingly, when grown in continuous culture under alternating light and dark cycles, Gloeothece confined most of its N2 fixation to the light, hence revealing a pattern very much like the Group A organisms (Ortega-Calvo and Stal, 1991). It is still not understood how Gloeothece is capable of apparently concomitantly fixing N2 and growing by oxygenic photosynthesis. Important differences of Gloeothece with Group A are its much larger size and the fact that the cells occur in aggregates or colonies enveloped by a common sheath.

Lyngbya aestuarii is a cosmopolitan marine benthic mat-forming diazotrophic cyanobacterium. It also follows the typical temporal separation of N2 fixation and oxygenic photosynthesis (Stal and Krumbein, 1985; 1987). The diel cycle of N2 fixation in Lyngbya is probably also under control of a circadian clock and its behaviour is very much similar to that of Cyanothece. In continuous light, photosynthesis switches off and respiration is greatly stimulated when nitrogenase is active. Nitrogenase is synthesized de novo for each dark period and inactivated during the day, even when nitrogenase antigen remains present during the full day–night cycle. It is apparently not quickly turned over as is the case in Trichodesmium.

In conclusion, cyanobacteria have evolved a variety of different strategies that aim at providing an anaerobic environment for nitrogenase. These strategies are a trade-off of the morphological and physiological properties of the cyanobacterium and the environment in which it thrives.

Distribution of diazotrophic cyanobacteria in the oceans

N2 fixation in the oceans is restricted to areas with seawater temperature in the range of 20–30°C (Capone et al., 1997), i.e. in the tropical and subtropical oceans. For instance, Mazard and colleagues (2004) detected putative unicellular diazotrophic cyanobacteria only in the area of 0–7°N along the 67°E meridian in the Arabian Sea where the water temperature was > 29°C as well as in one other station with similar conditions and also with a water temperature > 29°C. Similarly, Lugomela and colleagues (2002) found a possible temperature dependence of the occurrence of Trichodesmium in the Indian Ocean near Zanzibar with highest occurrences and nitrogenase activities around 28°C. These results hint to temperature as a key factor for the distribution of marine unicellular diazotrophic cyanobacteria. However, temperature itself is unlikely a factor restricting N2 fixation because it occurs in environments from close to freezing such as in the Antarctic (Shukla et al., 1997) as well as in those as hot as 60°C in hot springs (Miyamoto et al., 1979; Steunou et al., 2008). The reason for this temperature restriction of diazotrophic cyanobacteria and N2 fixation in the oceans is unclear and is discussed below.

The tropical oceans reveal interesting spatial variations of diazotrophic organisms and N2 fixation rates. Montoya and colleagues (2007) compared the relevant data obtained during a number of cruises in the Atlantic Ocean in different years and found no significant differences in the rates of N2 fixation as a function of longitude. However, Trichodesmium was by far more important in the western part, while unicellular diazotrophs appeared to be more important east of 40°W. An explanation for this peculiar distribution of diazotrophs in the tropical Atlantic was not offered, but agreed with observations of Staal and colleagues (2007b) who followed a north–south transect in the eastern Atlantic Ocean. Most of the stations positive for N2 fixation along this transect were characteristic for phycoerythrin-containing unicellular cyanobacteria and only in the area between 11°N and 4°N nitrogenase activity was more characteristic for Trichodesmium. Particularly south of the equator nitrogenase activity possessed typical features of unicellular diazotrophic cyanobacteria. Staal and colleagues (2007b) hypothesized that, due to their small size, unicellular diazotrophic cyanobacteria respond faster to nitrogen deficiency than the large Trichodesmium cells but that the latter eventually becomes dominant because it fixes N2 more efficiently due to its combination of spatial and temporal separation from oxygenic photosynthesis (Berman-Frank et al., 2001) and the possession of gas vesicles that allow buoyancy regulation and, hence, superior competition for light. In the south, the Benguela Current passes through an upwelling area off the coast of Namibia that feeds cold and nutrient-rich water. When nutrients become depleted in the South Equatorial Current, unicellular diazotrophic cyanobacteria inoculated from the warm equatorial waters are the first to profit. Subsequently, the easterly Equatorial Counter Current brings in water masses from the western part of the Atlantic Ocean containing Trichodesmium. This model agrees also with the observations of Fong and colleagues (2008) who reported that an anticyclonic eddy in the North Pacific Ocean supported particularly Trichodesmium as was seen from the transcription of nifH, although gene copies of at least seven other phylotypes were present. These authors explained this observation by the increased nutrient availability in the surface waters induced by the eddy.

Staal and colleagues (2007b) measured N2 fixation along a North–South transect in the Eastern Atlantic Ocean. They observed N2 fixation only between the 14°N and 13°S latitudes. The phytoplankton community in this part of the Atlantic was dominated by unicellular cyanobacteria and the properties of nitrogenase activity (activity mostly at night) hinted to Group B–C type of diazotrophs. When they pooled the stations that were positive for N2 fixation and those that were negative a distinct difference in surface water temperature was noted. The stations positive for N2 fixation were characterized by a water temperature of 28°C while the negative stations were 3°C cooler. This notable difference in temperature was accompanied by a number of other features. For instance, the chlorophyll a content (i.e. the phytoplankton biomass) in the area positive for N2 fixation was twice the amount present in the area without N2 fixation. The higher biomass had also drawn down the phosphate concentration to one-fourth of the amount in the area without N2 fixation and because the concentrations of DIN were not different between the two areas and invariably low, the N : P ratios in the area without N2 fixation were very low (below 2). Such low N : P ratios are a signature for nitrogen limitation and are expected to be selective for diazotrophic phytoplankton. However, apparently such organisms could not develop in this area. Instead, diatoms were an important component of the phytoplankton as was also evidenced from the much lower silicate concentration in the area without N2 fixation. Similar observations were made in an anticyclonic eddy in the North Pacific Ocean (Fong et al., 2008). Hence, it seems that a small difference in water temperature can make the difference whether a diazotrophic community can proliferate or not.

Similarly as for the Atlantic Ocean, the results from four cruises covering the latitude range of 12°S and 54°N in the Pacific Ocean revealed that diazotrophic organisms as identified by their nifH copies were restricted between the 14°N and 32°N latitudes (Church et al., 2008). The nifH phylotypes were most abundant in the oligotrophic regions where the surface temperatures ranged from 24°C to 27°C, although the authors observed low numbers from temperatures as low as 17°C and as high as 32°C. It was unclear whether such populations in these temperature extremes were in fact actively fixing N2 and whether they might have been drifted from other regions. The authors recognized that temperature was an important parameter that determined the distribution of diazotrophic organisms in the Pacific Ocean but other factors could also have played a role. For instance, in the warm equatorial extremely oligotrophic waters diazotrophs were virtually absent and it was hypothesized that the unavailability of iron may have excluded diazotrophs. It seemed that Trichodesmium appeared to have a much more limited temperature range as compared with unicellular diazotrophic cyanobacteria or even chemotrophic phylotypes. Particularly the ‘Group A’nifH phylotype showed a broader temperature range.

Largely similar observations were made for the North Atlantic Ocean (Langlois et al., 2008). These authors measured the abundance of nifH copies by using quantitative TaqMan PCR assay and were able to quantify seven different nifH phylotypes. Cyanobacteria nifH phylotypes were the most abundant (94%) indicating the relative insignificant contribution of other bacteria. Trichodesmium and Group A nifH phylotypes dominated with 51% and 33% respectively. These two groups had narrow temperature limits and were clearly different in their temperature preferences. The highest number of Group A nifH copies were found at considerably lower temperatures (20–23°C) than those of Trichodesmium (28–30°C). Similar as was observed for the Pacific Ocean, nifH copies were very low or even undetectable at the equator and, hence, other factors may here play a role in the distribution of diazotrophs in the North Atlantic Ocean.

The lower-temperature preference of the Group A nifH phylotypes was also observed in a station in the Pacific Ocean just two degrees north of the subtropics (34°N), at the border of the area where the summer SST still reaches 19–25°C (Needoba et al., 2007). The rates of nitrogenase activity these authors measured in October when the water temperature was 19°C were low compared with the tropics and subtropics but could support 10% of the new production in that area. The dominant nifH genotypes in that station belonged to ‘Group A’ and the number of transcripts was low but in accordance with the extrapolated measured rates of N2 fixation. Group A transcripts were found also in the deeper waters where the temperature was as low as 14°C. They did not find Trichodesmium. Neither their trichomes or colonies, nor their nifH genes or gene products were encountered.

Trichodesmium and temperature

The diazotrophic growth of Trichodesmium IMS101 is possible within the limits of 20–34°C, with an optimum of 27°C (Breitbarth et al., 2007). This organism was able to survive for weeks at temperatures as low as 17°C, but was unable to grow at that temperature. At 36°C Trichodesmium IMS101 died within 2 days. The C : N ratio varied a little with temperature and was approximately ‘Redfield’ (6.6) at the optimum growth temperature. Interestingly, at 17°C the C : N ratio was very high (> 9). The authors considered this as a possible artefact or they speculated that both carbon and nitrogen fixation were energy limited at this low temperature forcing the C : N ratio to that of pure protein. This explanation seems unlikely. At low temperature the metabolic rates will be accordingly low, while the light harvested is probably less affected. More likely is that N2 fixation is impossible because Trichodesmium IMS101 will be unable to maintain the diazotrophic cell anoxic as explained in Staal and colleagues (2003) and further elaborated below. As carbon fixation is not subject to the same limitations, the cell will be enriched with carbon relatively to nitrogen. Hence, this observation of Breitbarth and colleagues (2007) supports the conclusions of Staal and colleagues (2003).

Breitbarth and colleagues (2007) concluded that the upper limit of the occurrence of Trichodesmium is set by the sea surface temperature, rather than by the physiology of this cyanobacterium. Occasionally, Trichodesmium is found at temperatures below 20°C. It is generally assumed that this is caused by drifting and that these populations neither are photosynthetic active nor fix N2. These authors considered the consequences of a 3°C rise in the sea surface temperature and calculated that it would lead to an 11% areal increase of Trichodesmium, but this would be counteracted by a 16% decrease in area for optimum growth.

For unicellular diazotrophic cyanobacteria less is known with respect to the effect of temperature on growth and N2 fixation. Falcón and colleagues (2005) investigated three isolates from the Atlantic and Pacific Oceans in continuous cultures and found that optimum growth and N2 fixation occurred at 26–30°C. They also noted that the N : P ratios in all three isolates were low at lower temperature and increased approaching the Redfield value of 16 when grown at their optimum temperature. While these isolates were well adapted to low-phosphate concentrations, the low N : P ratio must have resulted from inefficient N2 fixation. The results showed also that N2 fixation was more affected by low temperature than growth rate. The identity of the isolates was not reported, but it is likely that they belonged to Group B or C.

Other factors that may determine distribution of diazotrophic cyanobacteria

The recent literature provides ample evidence for an important but unexplained role of temperature for the distribution of diazotrophic cyanobacteria in the pelagic ocean, i.e. its restriction to higher temperature (> 25°C). However, a high surface water temperature alone is not sufficient to allow the proliferation of diazotrophic cyanobacteria in the oceans, which is also clear from several of the literature reports. For instance, the equator has often been mentioned as an area where diazotrophs are absent and there are other such oceanic provinces. Hence, other factors are obviously involved determining the proliferation of diazotrophic cyanobacteria in the oceans. At the equator cold nutrient-rich upwelling waters could eliminate the selective advantage of diazotrophs. High concentrations of combined nitrogen would give other phytoplankton a selective advantage over diazotrophic organisms. For instance, Langlois and colleagues (2008) found that nifH copies were only abundant in waters with low-nitrate concentrations (< 0.5 µM). Phosphate and iron have often been identified as possible limiting factors (Sañudo-Wilhelmy et al., 2001; Voss et al., 2004). Tyrrell (1999) stated that the fixation of N2 would eventually lead to the limitation of primary productivity in the oceans by phosphorus. In the extremely oligotrophic oceans phosphate and iron may be too low to support much primary productivity. Diazotrophic organisms have a particular high demand of these nutrients. Severe phosphate limitation affects the energy household of organisms and this will specifically affect the energy-demanding N2 fixation process. Nitrogenase and its electron donor ferredoxin are rich in iron and therefore diazotrophic organisms have an elevated demand of it (Kustka et al., 2003). Langlois and colleagues (2008) found nifH most abundant in areas where the dust deposition was 2–5 g m−2 year−1. Higher or lower deposition rates were associated with lower abundances of nifH. Dust is an important source of iron. High dust deposition rates would also increase the input of other macronutrients including phosphate and combined nitrogen.

Nitrogen fixation by heterocystous cyanobacteria in the brackish Baltic Sea

Diazotrophic cyanobacteria are an important component of the summer phytoplankton blooms in the Baltic Sea. These cyanobacteria are filamentous, heterocystous types and their community consists roughly of three species: Nodularia spumigena, Aphanizomenon flos-aquae and Anabaena sp. (Laamanen and Kuosa, 2005). They all possess gas vesicles providing them with buoyancy. At calm sea they float to the surface where they form conspicuous surface blooms. They may also be transported to the deep by wind mixing of the water column, but even with a stratified and stable water column only a small portion of the total community is present at the surface (Walsby et al., 1995). Nodularia spumigena and A. flos-aquae form aggregates, allowing them to float at higher speeds. Nodularia spumigena is a toxic cyanobacterium and is often the dominant species, while Anabaena is usually a minor component which occurs as single trichomes. Anabaena sp. has been reported particularly from the northern parts of the Baltic Sea and occurs early in the summer (Laamanen, 1996).

It is unclear whether non-heterocystous diazotrophic cyanobacteria contribute to N2 fixation in the Baltic Sea. One report claimed that unicellular cyanobacteria in the size fraction < 10 µm contributed significantly to N2 fixation (Wasmund et al., 2001). These authors used the incorporation of the stable isotope 15N and it is possible that nitrogen fixed by heterocystous cyanobacteria had been transferred to the small size fraction (Ohlendieck et al., 2000). Natural communities of heterocystous cyanobacteria in the may loose 10–90% of the nitrogen they have fixed depending on the conditions (Stal and Walsby, 2000; Larsson et al., 2001; Gallon et al., 2002). Nevertheless, it cannot be totally excluded that non-heterocystous cyanobacteria contribute to N2 fixation in the Baltic Sea. Although none of the isolates of the picocyanobacterial Cyanobium were shown to fix N2, some of the isolates of the tiny filamentous Pseudanabaena possess nifH and are therefore potential diazotrophic (Acinas et al., 2009). Preliminary analyses indicate that this nitrogenase phylotype is expressed in samples taken from the Baltic Sea. If at all diazotrophic, Pseudanabaena belongs to the group of anaerobic N2-fixing cyanobacteria and it might have its niche in anaerobic aggregates. It is unlikely that it competes as a diazotroph with the heterocystous cyanobacteria in the water column.

Cyanobacterial blooms in the Baltic Sea develop after the spring bloom of diatoms that deplete the surface water from combined nitrogen (i.e. nitrate, NO3-) while sinking organic matter depletes the bottom water from oxygen. This causes the liberation of phosphate bound to the bottom sediments into the water column (Kahru et al., 2000). This results in low N : P ratios, which, when below the ‘Redfield ratio’ of 16, is assumed to provide diazotrophic cyanobacteria with a selective advantage (Arrigo, 2005). Diazotrophic blooms in the Baltic Sea developed at N : P ratios below 10 (Laamanen and Kuosa, 2005). Temperature is another important factor that has been brought in connection with the development of the blooms in the Baltic Sea, reasoning that cyanobacteria require warmer conditions for growth although also the stabilization of the water column and the connected decrease in mixing depth has been identified as important (Laamanen and Kuosa, 2005).

While it is a fact that cyanobacterial blooms in the Baltic Sea are only observed in summer and occur invariably at elevated water temperatures (> 10–13°C) (Laamanen and Kuosa, 2005), Stal and Walsby (2000) suggested that the effect of temperature must be indirect, i.e. not attributed to the physiology of the (diazotrophic) cyanobacteria. These authors showed that the compensation point of the daily water column integrated rate of photosynthesis was 22.7 mol photons m−2 of incident light and that even in the middle of summer there were days when this value was not reached, meaning that the blooms would have a negative net production. Moreover, they calculated the critical depth for every day of the year. The critical depth is defined as the depth below which the total water column photosynthetic production is negative. Only when the critical depth exceeds the mixing depth, whole water column production will be positive and this was only the case during the summer months (Stal and Walsby, 2000). Furthermore, they showed that neither the compensation depth (when net photosynthetic production is zero) nor the critical depth was affected by temperature. This means that although the photosynthetic production rate may be lower at lower temperature, this will also be the case with dark respiration, resulting in a lower but still net production.

Another interesting observation was that, in sharp contrast with photosynthesis, the daily and water column integrated rate of N2 fixation was only slightly affected by the incident daily light impinging at the surface (Stal and Walsby, 2000). This is counterintuitive considering the high energy demand for N2 fixation and the photosynthetic nature of the diazotrophic organisms. Moreover, measurements of natural samples and cultures of cyanobacteria isolated from these blooms showed a clear light response. Modelling N2 fixation from such light response curves (Staal et al., 2002) using the daily incident light and the depth distribution of the diazotrophic biomass allowed the investigation of this phenomenon by asking how the organism could increase its daily and water column integral of N2 fixation. It was reasoned (M. Staal and L.J. Stal, unpubl. results) that the organism might optimize its affinity towards light (α), its light dependent part of N2 fixation (Nm) or its light independent part (dark N2 fixation) (Nd) (see for the definition of these model parameters: Staal et al., 2002). The results of this exercise were that a doubling of α increased the daily and depth integral of N2 fixation by 6%, that a doubling of Nm would increase N2 fixation by 36% and that a doubling of Nd resulted in an increase of 51%! Hence, the organism would benefit most from increasing its light independent part of N2 fixation. The reason for this is that this would have an effect both during the day and the night and also contributes more at greater depth where light levels are low or in darkness, while the other two parameters would have an effect only the light.

The next question is which possibilities are at the disposal of the N2-fixing organism in order to increase Nd. The dark fixation of N2 is supported by respiration of intracellular reserve compounds, i.e. glycogen. It was demonstrated that in the dark and at non-saturating light intensities nitrogenase is normally energy limited. Therefore, it is eventually the flux of O2 into the diazotrophic cell that determines dark N2 fixation. While N2 fixation (nitrogenase activity) can be determined it is thus an excellent measure for the flux of O2. A lower flux of O2 would give a lower rate of respiration and as a result less energy generation and less N2 fixation and vice versa. When the flux of O2 into the diazotrophic cell would exceed the capacity of respiration or when e.g. insufficient reducing equivalents (i.e. storage carbohydrate) are available, anoxic conditions will not be maintained and nitrogenase will be inactivated. Only when O2 is scavenged uncoupled from energy generating respiration, there would be no intimate relation between respiration, nitrogenase activity and the O2 flux. This can be envisioned when nitrogenase is not limited by energy. Hence, the only way the N2-fixing organism can increase Nd is to increase the influx of O2 into the N2-fixing cell.

Dynamic gas exchange in the heterocyst

Heterocysts still possess PS-I and this will allow photosynthetic ATP generation through cyclic photophosphorylation in the light. In the dark respiration is the only source of energy. In order to optimize energy generation in the dark, the amount of O2 that is allowed to enter the heterocyst should be as large as possible, i.e. not exceeding the capacity of respiration thereby maintaining anoxic conditions in the heterocyst. This emphasizes that the gas diffusion properties of the glycolipid cell envelope should be carefully tuned to the requirements of optimum N2 fixation. Also, it should be noted that with increasing limitation of the gas diffusion properties of the glycolipid cell envelope of the heterocyst not only the entrance of O2 will decrease but also the entrance of N2. Obviously, the less efficient the gas diffusion properties of the glycolipid envelope are, the more O2 (and N2) can enter the heterocyst and the higher rate of N2 fixation may be achieved.

Cyanobacteria occur in dynamic environments where dissolved O2, temperature, salinity and light fluctuate rapidly. In order to perform optimally, diazotrophic cyanobacteria are expected to adapt to these fluctuations. There is evidence that the external O2 concentration influences the thickness and diffusion properties of the glycolipid envelope (Kangatharalingam et al., 1992). It is, however, difficult to imagine how the glycolipid cell envelope may dynamically change its diffusion characteristics at the time scale where O2 concentrations change. Walsby (2007) proposed therefore that not the glycolipid envelope is instrumental in the dynamic gas exchange between the heterocyst and its environment but the pores that connect the heterocyst with the neighbouring vegetative cells. This is an attractive and plausible possibility. It can be conceived that the pores open or close in a dynamic manner letting in more or less O2. The glycolipid envelope will serve as a more or less permanent gas diffusion barrier. The efficiency of it (i.e. thickness) could be genetically determined or a function of the environment in which the organism lives. Certainly, the synthesis of a thick glycolipid envelope comes with a cost for the organism. When not required, an organism that produces a thinner envelope may have a competitive advantage. Ultimately this would lead to the absence of the glycolipid cell envelope and the redundancy of the heterocyst.

The non-heterocystous diazotrophic cell (diazocyte)

The cell envelope of the N2-fixing cells of non-heterocystous cyanobacteria does not possess the glycolipid envelope that is typical for the heterocyst and presumably does not differ from that of the non-N2-fixing cells. For unicellular and filamentous non-heterocystous diazotrophic cyanobacteria that fix N2 during the dark, it is important that O2 diffuses into the cell in order to fuel respiration and therewith the fixation of N2. Although nitrogenase activity in Trichodesmium is strongly light-dependent it still has a light-independent component. O2 diffuses into the N2-fixing cell and will be scavenged in order to maintain that cell anoxic (Staal et al., 2007a). When the O2 is respired it will contribute to the energy demand of nitrogenase, although other O2-scavenging processes not linked to energy generation may also act in Trichodesmium (Kana, 1993; Milligan et al., 2007).

Gas diffusion into the diazotrophic cell

The flux (J) of O2 into the diazotrophic cell is a function of the diffusion coefficient (D) and the concentration (C) in the surrounding medium and the efficiency (ε) of the gas diffusion barrier (Staal et al., 2003). The efficiency of gas diffusion is a property of the glycolipid cell envelope and is 0> <1. The non-heterocystous diazotrophic cell is lacking this glycolipid envelope meaning that ε = 1. The interior of the diazotrophic cell should be virtually anoxic otherwise no active nitrogenase is expected. The solubility and diffusion coefficient of O2 in water depend both on salinity and temperature (http://www.unisense.com/Default.aspx?ID=117). The solubility of O2 in water decreases with both temperature and salinity. The diffusion coefficient increases with temperature and decreases with salinity. The flux of O2 into the diazotrophic cell would be 32% larger in freshwater than in seawater (salinity 35) at 20°C. In full salinity seawater at temperatures exceeding 20°C, Trichodesmium does not require a glycolipid diffusion barrier because respiration would be sufficiently fast to scavenge the O2 that enters the N2-fixing cell (Staal et al., 2003). Moreover, and perhaps even more important, a highly efficient gas diffusion barrier will limit not only the flux of O2 but also the flux of N2. Both would have serious drawbacks: the former would decrease the energy supply to nitrogenase or the latter simply might under saturate nitrogenase with its substrate N2. When the N2 concentration in the diazotrophic cell is limiting N2 fixation, nitrogenase will divert electrons to the evolution of H2, causing a futile loss of reducing power and energy, eventually leading to death.

Temperature relationships of metabolism and gas diffusion

The temperature coefficient (Q10) is a measure of the rate (R) of change of a process as the result of a change of temperature (T) of 10°C.

image

For many metabolic processes, within the limits of temperature given for an organism, the Q10 is often ∼2. This means that for a temperature increase or decrease of 10°C, the metabolic rate will double or halve respectively. The Q10 for the flux of O2 into a diazotrophic cell in seawater (salinity 35‰) in the temperature range 10–35°C is 1.10 (calculated from the relationship of temperature and the product of O2 solubility and diffusion coefficient: http://www.unisense.com/Default.aspx?ID=117). In freshwater the Q10 is slightly lower: 1.09. If respiration and N2 fixation would have Q10 of ∼2, it would mean that with increasing temperature the O2 flux would severely lag behind the potential rates of respiration and N2 fixation.

Most of the Q10 values of the light-independent (dark) nitrogenase activity, Nd, of the heterocystous Baltic Sea cyanobacteria N. spumigena and Anabaena sp., and the non-heterocystous oceanic Trichodesmium sp. are close to the 1.1 calculated for the O2 flux (Staal et al., 2003) (Table 1). This means that the O2 flux into the diazotrophic cell controls nitrogenase activity (transport control) and any O2 that enters the cell is respired, resulting in virtual anoxic conditions. Exceptions are N. spumigena and Trichodesmium sp. in the lower-temperature regions. At these low temperatures respiration becomes too slow to consume the O2 that diffuses into the diazotrophic cell. Respiration is at its maximum rate (at that temperature) and controls nitrogenase activity (reaction control). It should be noted that the diazotrophic cell cannot be maintained anoxic under these conditions and nitrogenase will be rapidly inactivated. These measurements of Nd were only possible during very short incubations, using an online and real-time set-up for nitrogenase activity monitoring (Staal et al., 2001). The sensitivity of nitrogenase for O2 in addition to the suboptimal temperatures under which these measurements were done explains the exceptional high values of Q10. The values of Q10 of Ntot (the sum of the dark and light-dependent rates of nitrogenase activity) were all close to the expected value of ∼2 for metabolic processes, with the exceptions of the incubations under suboptimal temperatures (Table 1).

Table 1. Q10 values for nitrogenase activity (acetylene reduction assay) in the dark and at saturating irradiance.
OrganismTemperature rangeQ10 (Nd)Q10 (Ntot)
  1. Q10 values for Nd and Ntot were measured for Nodularia spumigena (strain CCY9414) cultured at 20°C and 25°C, Anabaena sp. (strain CCY9901) cultured at 20°C, and Trichodesmium sp. (strain IMS101) cultured at 25°C (Staal et al., 2003).

N. spumigena(20°C)10–15°C24.5338.86
15–30°C1.292.26
N. spumigena(25°C)15–35°C1.331.81
Anabaena sp.5–30°C1.122.34
Trichodesmium sp.15–20°C4.313.36
20–35°C1.121.84

Temperature limits of diazotrophic growth

The intercept of the temperature relationships of Nd and Ntot gives the temperature at which the light-dependent rate of N2 fixation, Nm, equals 0 and Ntot/Nd equals 1 and marks the transition from transport to reaction control. Below this temperature N2 fixation will be impossible because the diazotrophic cell cannot be maintained anoxic (Table 2). The temperatures estimated this way reflect the environment from which the organism originated. According to these estimates Trichodesmium cannot thrive below 16.9°C under diazotrophic conditions. This might seem a bit too low, considering the fact that Trichodesmium is encountered usually at 25°C and higher. However, it should be emphasized that this temperature is the absolute theoretical limit. Below 25°C the growth rate of Trichodesmium drops and the organism may loose the competition in its natural environment. The lower-temperature limit of Trichodesmium estimated here is remarkably similar as the 17°C determined by Breitbarth and colleagues (2007). The transition temperature estimated for Nodularia is 12.3°C, which reflects its occurrence in the Baltic Sea in summer. The transition temperature for Anabaena is estimated at 3.8°C, reflecting its occurrence early in the year and particularly in the northern reaches of the Baltic Sea where the water temperature is still low. These temperatures are also lower than those under which the organisms thrive in the Baltic Sea. Similar as for Trichodesmium these temperatures should be considered as the absolute (theoretical) limit. In order to proliferate in nature, higher growth rates are required to be able to compete with other organisms in the community.

Table 2.  Temperature where the control of N2 fixation changes from enzyme level (reaction control) to gas (O2 or N2) diffusion (transport control).
OrganismO2 (Nd)
  1. Temperatures were obtained from the intersections of the linear fits of Nd and Ntot (y = ax + b) (Staal et al., 2003).

N. spumigena(20°C)12.3
N. spumigena(25°C)9.8
Anabaena sp.3.8
Trichodesmium sp.16.9

In Fig. 1 the relationship between reaction control and transport control is depicted for Trichodesmium grown at 25°C as an example. It shows the lines for the theoretical Q10 = 2 for enzyme activity Q10 = 1.1 for O2 flux which intercept at 16.9°C. The black symbols indicate the theoretical Nd below and above the critical temperature. The open symbols represent the measured values for Nd (circles) and Ntot (squares). Nd follows the theoretical predicted values (calculated Q10 = 1.12; Table 1) although it is lower than predicted below the intercept. Ntot follows the Q10 = 2 (calculated Q10 = 1.84; Table 1). Below the transition temperature respiration (and thus dark N2 fixation) is controlled by nitrogenase activity (reaction control) and above by the flux of O2 into the diazotrophic cell (transport control). In the latter case the cell is anaerobic.

Figure 1.

Temperature relationships of Ntot (squares) and Nd (open circles) of Trichodesmium and the theoretical lines for Q10 = 2 (theoretical enzyme activity) (solid line) and Q10 = 1.1 (gas flux) (broken line) and the theoretical values for Nd (filled symbols). At the intercept of Q10 = 2 and Q10 = 1.1 the theoretical Ntot/Nd = 1. The intercept is at 16.9°C (Table 2). Below this temperature nitrogenase activity follows the Q10 for nitrogenase activity (1.84, Table 1) (‘reaction control’) and the diazotrophic cell will become aerobic. Above 16.9°C Nd will follow Q10 for gas flux (1.12, Table 1) and the diazotrophic cell will be virtually anaerobic.

A similar exercise can be done for the diffusion of N2 (Fig. 2). The Q10 of the N2 flux (solubility multiplied by the diffusion coefficient) is not much different from O2 (1.13 and 1.12 at salinity of 35 ppt and 0 ppt respectively). While Ntot is reaction (nitrogenase) controlled throughout most of its normal temperature range, at higher temperatures it may change to transport (N2) control. This is the case when due to high-temperature nitrogenase activity exceeds the flux of N2 into the diazotrophic cell and the latter becomes the limiting step. In order to determine the temperature where the transition occurs, it would be necessary to measure the N2 flux into the diazotrophic cell or have a measure for it. Because nitrogenase activity was measured using the acetylene reduction assay this cannot be taken as measure for the flux of N2 and it is therefore currently not possible to estimate the transition temperature. In the case of Trichodesmium, the optimum temperature for growth as well as N2 fixation is 27°C (Breitbarth et al., 2007). In Fig. 2 this temperature is presumed to be the transition from ‘reaction control’ to ‘transport control’ of nitrogenase activity. At temperatures below this upper transition point the diazotrophic cell is saturated with N2, a situation that can be described as ‘azotic’ (analogous to ‘aerobic’), while at higher temperatures a low concentration of N2 is expected (depending on the affinity of nitrogenase) and this situation is therefore described as ‘anazotic’ (analogous to ‘anaerobic’ or ‘anoxic’). The actual concentration of N2 in the diazotrophic cell is difficult to determine. It depends also on the affinity of nitrogenase for N2, which is low (0.08 atm) (Christiansen et al., 2000).

Figure 2.

Temperature relationship of Ntot (squares) of Trichodesmium and the theoretical lines for Q10 = 2 (theoretical enzyme activity) (solid line) and Q10 = 1.1 (gas flux) (broken line) and the theoretical values for Ntot (filled symbols). The theoretical intercept of Q10 = 2 and Q10 = 1.1 is set at 27°C [the optimum growth temperature of Trichodesmium (Breitbarth et al., 2007)] at which Ntot/Nd = 2.4. Below this temperature nitrogenase activity follows the Q10 for nitrogenase activity (1.84, Table 1) (‘reaction control’) and the diazotrophic cell is saturated with N2 (‘azotic’). Above 27°C it is expected that the diffusion of N2 in the diazotrophic will control the rate of N2 fixation (‘transport control’) and Ntot will follow Q10 for gas flux (1.12, Table 1) and the diazotrophic cell will be ‘anazotic’ (depending on the affinity of nitrogenase for N2).

Although cyanobacteria would still be capable of fixing N2 when nitrogenase is controlled by the flux of N2 (ignoring the fact that the high temperature may be suboptimal for other processes involved in growth and metabolism of the organism), it will be inefficient. When nitrogenase is under saturated with its substrate N2 it will perform as a hydrogenase and reduce electrons in stead, producing H2 (in addition to the obligatory produced H2). This H2 may be partly recycled through the uptake hydrogenase but because of the unavailability of O2 in the diazotrophic cell much is expected to be lost. Eventually the organism will be unable to maintain itself due to the futile loss of reducing equivalents and energy.

Adaptation of the ratio Ntot/Nd

The Ntot/Nd, i.e. the ratio of the potential nitrogenase activity, the sum of the activity at saturating light and in the dark (Nm + Nd), and the dark activity equals or is larger than 1 and is a measure of the ATP limitation of nitrogenase in the dark. When the ratio equals 1, respiration supplies all ATP necessary for nitrogenase to achieve its maximum activity. Higher ratios indicate a higher energy demand that is supplied by light. Ratios of 1 are not usually observed in cyanobacteria. At the intercept of ‘transport control’ and ‘reaction control’, the Ntot/Nd is greater than 1. This means that at the intercept temperature the maximum rate of respiration is slower than the potential maximum rate of nitrogenase.

Staal and colleagues (2003) measured Ntot/Nd ratios in a large number of heterocystous cyanobacteria grown at 20°C and found that the ratio varied from 2.5 to 4.0 with an overall average of 3.2. For instance, in N. spumigena the average ratio was 2.6 and did not differ much at different growth temperatures. However, when this organism was grown at 20°C and tested at 25°C the ratio was much higher (3.8). Anabaena has an unusual high Ntot/Nd ratio of 9, three times higher than what is found in other heterocystous cyanobacteria. This means that respiration contributes very little to N2 fixation in this organism, obviously because of the limited flux of O2 into the heterocyst. Heterocystous cyanobacteria seem to be able to adjust the ratio to an optimal value but they cannot adjust quickly to changes in temperature. The ratio Ntot/Nd increases because a substantial higher potential nitrogenase activity is achieved (Q10∼2), while Nd changes little because O2 diffusion increases only marginally (Q10∼1.1).

The diazotrophic cell cannot distinguish between O2 and N2 and a gas diffusion barrier will affect the entrance of any gas. The ratio Ntot/Nd is a trade-off between the requirement of limiting the entrance of O2 and allowing sufficient N2 in the diazotrophic cell. The solubility of N2 (Hamme and Emerson, 2004) is 1.9 times that of O2 and the diffusion coefficient of N2 is 0.83 times that of O2 (http://www.unisense.com/Default.aspx?ID=117). The diffusion of N2 (solubility multiplied by the diffusion coefficient) will therefore be 1.6 times that of O2. When the diffusion of O2 and N2 are in balance it would yield an Ntot/Nd of 2.6, as was indeed observed. This assumes that as is the case with O2 the concentration of N2 in the diazotrophic cell is close to 0, which is obviously not the case. Therefore, Ntot/Nd may be slightly lower.

When no other O2-utilizing processes than respiration are going on, Nd can be taken as a measure of the O2 influx into the diazotrophic cell (i.e. the O2 influx limits Nd) and it does therefore not relate to Nm, the maximum light-dependent part of nitrogenase activity. Light (i.e. ATP) will not limit Nm (because it is by definition the activity at light saturation). Factors that could potentially limit Nm include (i) reducing equivalents, (ii) the amount of nitrogenase, or (iii) the flux of N2 (or acetylene in the case of the acetylene reduction assay). An increase in temperature will increase Ntot/Nd, because Nm increases more (Q10∼2) than Nd (Q10∼1.1).

The narrow range in which Ntot/Nd is adjusted can be attributed to the delicate trade-off between the diffusion of O2 and N2 into the diazotrophic cell. Any change in the external concentration of these gases or in the temperature will require acclimation. The cell envelope of the heterocyst consists of several layers of glycolipid and polysaccharide that form the main gas diffusion barrier. The efficiency of this gas diffusion barrier (i.e. thickness) is genetically determined and it is difficult to image how this may allow a dynamic acclimation to changing environmental conditions. Walsby (2007) proposed that the pores that connect the heterocyst with the neighbouring vegetative cell serve as the main ports of gas exchange. It is conceivable that these pores may dynamically open and close, just like stomata in plants. Walsby suggested that the pores (stomata) of heterocysts might close during the night when nitrogenase activity is low and a low demand of N2 exist while at the same time the supply of reducing equivalents is low that are required for respiration. Thus when the closed pores also limit the influx of O2 the heterocyst can be maintained anoxic with a limited supply of reducing equivalents, thereby preventing the inactivation of nitrogenase. While this model may be valid in certain species, other heterocystous cyanobacteria fix considerable N2 at night. As described above, this may increase the ecological success of the organism enormously and make it much less dependent on the actual availability of light. In either case, the fine-tuning of the O2 influx with the rate of respiration (and hence the supply of reducing equivalents) in the dark as well as in the light will optimize the organism's performance with respect to the fixation of N2.

Conclusions

Diazotrophic cyanobacteria in the oceans occur only in the tropical and subtropical regions where the surface water temperature is above 25°C. With the exception of some symbiotic cyanobacteria all of the diazotrophic species in the oceans are of the non-heterocystous type. The temperate and cold oceans and seas seem to be devoid of diazotrophic cyanobacteria but freshwater and brackish environments are commonly occupied by heterocystous N2-fixing cyanobacteria and are found worldwide in all climatic regions. Obviously, the possession of a heterocyst is not a prerequisite for diazotrophic growth of cyanobacteria in the warm waters of the tropical oceans. It is argued that the relative low-oxygen concentration in warm seawater compared with colder or freshwater and the higher rates of respiration at elevated temperature are sufficient to keep the nitrogen-fixing cell anoxic. The possession of a heterocyst under such conditions would become disadvantageous because of the metabolic cost and because of the limited flux of dinitrogen gas. However, there are other factors that determine the proliferation of diazotrophic cyanobacteria in the tropical ocean. Extremely oligotrophic conditions with particularly low iron or phosphate concentrations may prohibit the growth of diazotrophic cyanobacteria, whereas elevated nitrate concentrations provide competitive advantage to other phytoplankton. In pelagic environments with lower temperatures non-heterocystous diazotrophic cyanobacteria will be excluded because they will be unable to maintain anoxic conditions in their diazotrophic cells. Heterocystous cyanobacteria are abundant in freshwater and brackish environments in all climatic zones. It appears that the gas diffusion characteristics of the heterocyst glycolipid cell wall is tuned to optimize between the requirement of maximizing the diffusion of gas (i.e. oxygen and nitrogen) to allow the highest possible respiratory energy generation and nitrogen fixation, while keeping the cell anoxic. The heterocyst cell envelope may be genetically determined and tuned to the environment in which the organism occurs or may adapt to changing conditions. Short-term tuning of the oxygen and nitrogen flux may be accomplished through opening and closing the pores connecting the heterocyst to the neighbouring vegetative cells. It is not known what excludes heterocystous cyanobacteria from the temperate and cold seas.

Acknowledgements

I would like to acknowledge my former student M. Staal who originally developed some of the ideas reviewed in this paper. This is publication nr. 4475 of NIOO-KNAW.

Ancillary