Convergent estimates of marine nitrogen fixation.

Uncertainty in the global patterns of marine nitrogen fixation limits our understanding of the response of the ocean's nitrogen and carbon cycles to environmental change. The geographical distribution of and ecological controls on nitrogen fixation are difficult to constrain with limited in situ measurements. Here we present convergent estimates of nitrogen fixation from an inverse biogeochemical and a prognostic ocean model. Our results demonstrate strong spatial variability in the nitrogen-to-phosphorus ratio of exported organic matter that greatly increases the global nitrogen-fixation rate (because phytoplankton manage with less phosphorus when it is in short supply). We find that the input of newly fixed nitrogen from microbial fixation and external inputs (atmospheric deposition and river fluxes) accounts for up to 50 per cent of carbon export in subtropical gyres. We also find that nitrogen fixation and denitrification are spatially decoupled but that nevertheless nitrogen sources and sinks appear to be balanced over the past few decades. Moreover, we propose a role for top-down zooplankton grazing control in shaping the global patterns of nitrogen fixation. Our findings suggest that biological carbon export in the ocean is higher than expected and that stabilizing nitrogen-cycle feedbacks are weaker than previously thought.

Bacterial biogeography of the Indian Ocean.

Historically, our understanding of bacterial ecology in the Indian Ocean has been limited to regional studies that place emphasis on community structure and function within oxygen minimum zones. Thus, bacterial community dynamics across the wider Indian Ocean are largely undescribed. As part of Bio-GO-SHIP, we sequenced the 16S rRNA gene from 465 samples collected on sections I07N and I09N. We found that (i) there were 23 distinct bioregions within the Indian Ocean, (ii) the southeastern gyre had the largest gradient in bacterial alpha-diversity, (iii) the Indian Ocean surface microbiome was primarily composed of a core set of taxa, and (iv) bioregions were characterized by transitions in physical and geochemical conditions. Overall, we showed that bacterial community structure spatially delineated the surface Indian Ocean and that these microbially-defined regions were reflective of subtle ocean physical and geochemical gradients. Therefore, incorporating metrics of in-situ microbial communities into marine ecological regions traditionally defined by remote sensing will improve our ability to delineate warm, oligotrophic regions.

Investigating the eco-evolutionary response of microbiomes to environmental change.

Microorganisms are the primary engines of biogeochemical processes and foundational to the provisioning of ecosystem services to human society. Free-living microbial communities (microbiomes) and their functioning are now known to be highly sensitive to environmental change. Given microorganisms' capacity for rapid evolution, evolutionary processes could play a role in this response. Currently, however, few models of biogeochemical processes explicitly consider how microbial evolution will affect biogeochemical responses to environmental change. Here, we propose a conceptual framework for explicitly integrating evolution into microbiome-functioning relationships. We consider how microbiomes respond simultaneously to environmental change via four interrelated processes that affect overall microbiome functioning (physiological acclimation, demography, dispersal and evolution). Recent evidence in both the laboratory and the field suggests that ecological and evolutionary dynamics occur simultaneously within microbiomes; however, the implications for biogeochemistry under environmental change will depend on the timescales over which these processes contribute to a microbiome's response. Over the long term, evolution may play an increasingly important role for microbially driven biogeochemical responses to environmental change, particularly to conditions without recent historical precedent.

Latitudinal gradient in the respiration quotient and the implications for ocean oxygen availability.

Climate-driven depletion of ocean oxygen strongly impacts the global cycles of carbon and nutrients as well as the survival of many animal species. One of the main uncertainties in predicting changes to marine oxygen levels is the regulation of the biological respiration demand associated with the biological pump. Derived from the Redfield ratio, the molar ratio of oxygen to organic carbon consumed during respiration (i.e., the respiration quotient, [Formula: see text]) is consistently assumed constant but rarely, if ever, measured. Using a prognostic Earth system model, we show that a 0.1 increase in the respiration quotient from 1.0 leads to a 2.3% decline in global oxygen, a large expansion of low-oxygen zones, additional water column denitrification of 38 Tg N/y, and the loss of fixed nitrogen and carbon production in the ocean. We then present direct chemical measurements of [Formula: see text] using a Pacific Ocean meridional transect crossing all major surface biome types. The observed [Formula: see text] has a positive correlation with temperature, and regional mean values differ significantly from Redfield proportions. Finally, an independent global inverse model analysis constrained with nutrients, oxygen, and carbon concentrations supports a positive temperature dependence of [Formula: see text] in exported organic matter. We provide evidence against the common assumption of a static biological link between the respiration of organic carbon and the consumption of oxygen. Furthermore, the model simulations suggest that a changing respiration quotient will impact multiple biogeochemical cycles and that future warming can lead to more intense deoxygenation than previously anticipated.

Decomposition responses to climate depend on microbial community composition.

Bacteria and fungi drive decomposition, a fundamental process in the carbon cycle, yet the importance of microbial community composition for decomposition remains elusive. Here, we used an 18-month reciprocal transplant experiment along a climate gradient in Southern California to disentangle the effects of the microbial community versus the environment on decomposition. Specifically, we tested whether the decomposition response to climate change depends on the microbial community. We inoculated microbial decomposers from each site onto a common, irradiated leaf litter within "microbial cages" that prevent microbial exchange with the environment. We characterized fungal and bacterial composition and abundance over time and investigated the functional consequences through litter mass loss and chemistry. After 12 months, microbial communities altered both decomposition rate and litter chemistry. Further, the functional measurements depended on an interaction between the community and its climate in a manner not predicted by current theory. Moreover, microbial ecologists have traditionally considered fungi to be the primary agents of decomposition and for bacteria to play a minor role. Our results indicate that not only does climate change and transplantation have differential legacy effects among bacteria and fungi, but also that bacterial communities might be less functionally redundant than fungi with regards to decomposition. Thus, it may be time to reevaluate both the role of microbial community composition in its decomposition response to climate and the relative roles of bacterial and fungal communities in decomposition.

Global biogeography of microbial nitrogen-cycling traits in soil.

Microorganisms drive much of the Earth's nitrogen (N) cycle, but we still lack a global overview of the abundance and composition of the microorganisms carrying out soil N processes. To address this gap, we characterized the biogeography of microbial N traits, defined as eight N-cycling pathways, using publically available soil metagenomes. The relative frequency of N pathways varied consistently across soils, such that the frequencies of the individual N pathways were positively correlated across the soil samples. Habitat type, soil carbon, and soil N largely explained the total N pathway frequency in a sample. In contrast, we could not identify major drivers of the taxonomic composition of the N functional groups. Further, the dominant genera encoding a pathway were generally similar among habitat types. The soil samples also revealed an unexpectedly high frequency of bacteria carrying the pathways required for dissimilatory nitrate reduction to ammonium, a little-studied N process in soil. Finally, phylogenetic analysis showed that some microbial groups seem to be N-cycling specialists or generalists. For instance, taxa within the Deltaproteobacteria encoded all eight N pathways, whereas those within the Cyanobacteria primarily encoded three pathways. Overall, this trait-based approach provides a baseline for investigating the relationship between microbial diversity and N cycling across global soils.

Emergence of soil bacterial ecotypes along a climate gradient.

The high diversity of soil bacteria is attributed to the spatial complexity of soil systems, where habitat heterogeneity promotes niche partitioning among bacterial taxa. This premise remains challenging to test, however, as it requires quantifying the traits of closely related soil bacteria and relating these traits to bacterial abundances and geographic distributions. Here, we sought to investigate whether the widespread soil taxon Curtobacterium consists of multiple coexisting ecotypes with differential geographic distributions. We isolated Curtobacterium strains from six sites along a climate gradient and assayed four functional traits that may contribute to niche partitioning in leaf litter, the top layer of soil. Our results revealed that cultured isolates separated into fine-scale genetic clusters that reflected distinct suites of phenotypic traits, denoting the existence of multiple ecotypes. We then quantified the distribution of Curtobacterium by analysing metagenomic data collected across the gradient over 18 months. Six abundant ecotypes were observed with differential abundances along the gradient, suggesting fine-scale niche partitioning. However, we could not clearly explain observed geographic distributions of ecotypes by relating their traits to environmental variables. Thus, while we can resolve soil bacterial ecotypes, the traits delineating their distinct niches in the environment remain unclear.

A simple nutrient-dependence mechanism for predicting the stoichiometry of marine ecosystems.

It is widely recognized that the stoichiometry of nutrient elements in phytoplankton varies within the ocean. However, there are many conflicting mechanistic explanations for this variability, and it is often ignored in global biogeochemical models and carbon cycle simulations. Here we show that globally distributed particulate P:C varies as a linear function of ambient phosphate concentrations, whereas the N:C varies with ambient nitrate concentrations, but only when nitrate is most scarce. This observation is consistent with the adjustment of the phytoplankton community to local nutrient availability, with greater flexibility of phytoplankton P:C because P is a less abundant cellular component than N. This simple relationship is shown to predict the large-scale, long-term average composition of surface particles throughout large parts of the ocean remarkably well. The relationship implies that most of the observed variation in N:P actually arises from a greater plasticity in the cellular P:C content, relative to N:C, such that as overall macronutrient concentrations decrease, N:P rises. Although other mechanisms are certainly also relevant, this simple relationship can be applied as a first-order basis for predicting organic matter stoichiometry in large-scale biogeochemical models, as illustrated using a simple box model. The results show that including variable P:C makes atmospheric CO2 more sensitive to changes in low latitude export and ocean circulation than a fixed-stoichiometry model. In addition, variable P:C weakens the relationship between preformed phosphate and atmospheric CO2 while implying a more important role for the nitrogen cycle.

Stoichiometry of Prochlorococcus, Synechococcus, and small eukaryotic populations in the western North Atlantic Ocean.

In the North Atlantic Ocean, we found that natural populations of Prochlorococcus adhered to Redfield ratio dimensions when comparing cell quotas of carbon to nitrogen, but had flexible composition under nutrient and light stress, allowing for a broad range of cellular carbon- and nitrogen-to-phosphorus ratios. Synechococcus populations also exhibited a wide range of elemental stoichiometry, including carbon-to-nitrogen ratios and increased their carbon-to-phosphorus ratios in response to low dissolved phosphorus availability. Small eukaryotic populations tended to have lower carbon-to-phosphorus ratios than single cell cyanobacterial groups, with the exception of one group of samples, which highlights the importance of community composition when determining how biological diversity influences bulk particle stoichiometry. The ratio of dissolved nitrogen:phosphorus fluxes into the euphotic zone was not correlated to nitrogen:phosphorus cellular quotas. The lack of a homeostatic relationship implies that other mechanisms, such as species-specific adaptation to oligotrophic phosphorus concentrations, control elemental particle ratios.

Glycoside Hydrolases across Environmental Microbial Communities.

Across many environments microbial glycoside hydrolases support the enzymatic processing of carbohydrates, a critical function in many ecosystems. Little is known about how the microbial composition of a community and the potential for carbohydrate processing relate to each other. Here, using 1,934 metagenomic datasets, we linked changes in community composition to variation of potential for carbohydrate processing across environments. We were able to show that each ecosystem-type displays a specific potential for carbohydrate utilization. Most of this potential was associated with just 77 bacterial genera. The GH content in bacterial genera is best described by their taxonomic affiliation. Across metagenomes, fluctuations of the microbial community structure and GH potential for carbohydrate utilization were correlated. Our analysis reveals that both deterministic and stochastic processes contribute to the assembly of complex microbial communities.

Impact of ocean phytoplankton diversity on phosphate uptake.

We have a limited understanding of the consequences of variations in microbial biodiversity on ocean ecosystem functioning and global biogeochemical cycles. A core process is macronutrient uptake by microorganisms, as the uptake of nutrients controls ocean CO2 fixation rates in many regions. Here, we ask whether variations in ocean phytoplankton biodiversity lead to novel functional relationships between environmental variability and phosphate (Pi) uptake. We analyzed Pi uptake capabilities and cellular allocations among phytoplankton groups and the whole community throughout the extremely Pi-depleted western North Atlantic Ocean. Pi uptake capabilities of individual populations were well described by a classic uptake function but displayed adaptive differences in uptake capabilities that depend on cell size and nutrient availability. Using an eco-evolutionary model as well as observations of in situ uptake across the region, we confirmed that differences among populations lead to previously uncharacterized relationships between ambient Pi concentrations and uptake. Supported by novel theory, this work provides a robust empirical basis for describing and understanding assimilation of limiting nutrients in the oceans. Thus, it demonstrates that microbial biodiversity, beyond cell size, is important for understanding the global cycling of nutrients.

Present and future global distributions of the marine Cyanobacteria Prochlorococcus and Synechococcus.

The Cyanobacteria Prochlorococcus and Synechococcus account for a substantial fraction of marine primary production. Here, we present quantitative niche models for these lineages that assess present and future global abundances and distributions. These niche models are the result of neural network, nonparametric, and parametric analyses, and they rely on >35,000 discrete observations from all major ocean regions. The models assess cell abundance based on temperature and photosynthetically active radiation, but the individual responses to these environmental variables differ for each lineage. The models estimate global biogeographic patterns and seasonal variability of cell abundance, with maxima in the warm oligotrophic gyres of the Indian and the western Pacific Oceans and minima at higher latitudes. The annual mean global abundances of Prochlorococcus and Synechococcus are 2.9 ± 0.1 × 10(27) and 7.0 ± 0.3 × 10(26) cells, respectively. Using projections of sea surface temperature as a result of increased concentration of greenhouse gases at the end of the 21st century, our niche models projected increases in cell numbers of 29% and 14% for Prochlorococcus and Synechococcus, respectively. The changes are geographically uneven but include an increase in area. Thus, our global niche models suggest that oceanic microbial communities will experience complex changes as a result of projected future climate conditions. Because of the high abundances and contributions to primary production of Prochlorococcus and Synechococcus, these changes may have large impacts on ocean ecosystems and biogeochemical cycles.

The ocean as a global reservoir of antibiotic resistance genes.

Recent studies of natural environments have revealed vast genetic reservoirs of antibiotic resistance (AR) genes. Soil bacteria and human pathogens share AR genes, and AR genes have been discovered in a variety of habitats. However, there is little knowledge about the presence and diversity of AR genes in marine environments and which organisms host AR genes. To address this, we identified the diversity of genes conferring resistance to ampicillin, tetracycline, nitrofurantoin, and sulfadimethoxine in diverse marine environments using functional metagenomics (the cloning and screening of random DNA fragments). Marine environments were host to a diversity of AR-conferring genes. Antibiotic-resistant clones were found at all sites, with 28% of the genes identified as known AR genes (encoding beta-lactamases, bicyclomycin resistance pumps, etc.). However, the majority of AR genes were not previously classified as such but had products similar to proteins such as transport pumps, oxidoreductases, and hydrolases. Furthermore, 44% of the genes conferring antibiotic resistance were found in abundant marine taxa (e.g., Pelagibacter, Prochlorococcus, and Vibrio). Therefore, we uncovered a previously unknown diversity of genes that conferred an AR phenotype among marine environments, which makes the ocean a global reservoir of both clinically relevant and potentially novel AR genes.

Genomic potential for polysaccharide deconstruction in bacteria.

Glycoside hydrolases are important enzymes that support bacterial growth by enabling the degradation of polysaccharides (e.g., starch, cellulose, xylan, and chitin) in the environment. Presently, little is known about the overall phylogenetic distribution of the genomic potential to degrade these polysaccharides in bacteria. However, knowing the phylogenetic breadth of these traits may help us predict the overall polysaccharide processing in environmental microbial communities. In order to address this, we identified and analyzed the distribution of 392,166 enzyme genes derived from 53 glycoside hydrolase families in 8,133 sequenced bacterial genomes. Enzymes for oligosaccharides and starch/glycogen were observed in most taxonomic groups, whereas glycoside hydrolases for structural polymers (i.e., cellulose, xylan, and chitin) were observed in clusters of relatives at taxonomic levels ranging from species to genus as determined by consenTRAIT. The potential for starch and glycogen processing, as well as oligosaccharide processing, was observed in 85% of the strains, whereas 65% possessed enzymes to degrade some structural polysaccharides (i.e., cellulose, xylan, or chitin). Potential degraders targeting one, two, and three structural polysaccharides accounted for 22.6, 32.9, and 9.3% of genomes analyzed, respectively. Finally, potential degraders targeting multiple structural polysaccharides displayed increased potential for oligosaccharide deconstruction. This study provides a framework for linking the potential for polymer deconstruction with phylogeny in complex microbial assemblages.

Nitrogen Cycling Potential of a Grassland Litter Microbial Community.

Because microorganisms have different abilities to utilize nitrogen (N) through various assimilatory and dissimilatory pathways, microbial composition and diversity likely influence N cycling in an ecosystem. Terrestrial plant litter decomposition is often limited by N availability; however, little is known about the microorganisms involved in litter N cycling. In this study, we used metagenomics to characterize the potential N utilization of microbial communities in grassland plant litter. The frequencies of sequences associated with eight N cycling pathways differed by several orders of magnitude. Within a pathway, the distributions of these sequences among bacterial orders differed greatly. Many orders within the Actinobacteria and Proteobacteria appeared to be N cycling generalists, carrying genes from most (five or six) of the pathways. In contrast, orders from the Bacteroidetes were more specialized and carried genes for fewer (two or three) pathways. We also investigated how the abundance and composition of microbial N cycling genes differed over time and in response to two global change manipulations (drought and N addition). For many pathways, the abundance and composition of N cycling taxa differed over time, apparently reflecting precipitation patterns. In contrast to temporal variability, simulated global change had minor effects on N cycling potential. Overall, this study provides a blueprint for the genetic potential of N cycle processes in plant litter and a baseline for comparisons to other ecosystems.

Phylogenetic constraints on elemental stoichiometry and resource allocation in heterotrophic marine bacteria.

The objective of this study was to evaluate the contribution of evolutionary history to variation in the biomass stoichiometry and underlying biochemical allocation patterns of heterotrophic marine bacteria. We hypothesized that phylogeny significantly constrains biochemical allocation strategy and elemental composition among taxa of heterotrophic marine bacteria. Using a 'common-garden' experimental design, we detected significant interspecific variation in stoichiometry, macromolecule allocation and growth rate among 13 strains of marine Proteobacteria. However, this variation was not well explained by 16S rRNA phylogenetic relationships or differences in growth rate. Heterotrophic bacteria likely experience C-limitation when consuming resources in Redfield proportions, which consequently decouples growth rate from allocation to rRNA and biomass P content. Accordingly, overall bacterial C : nutrient ratios (C : P = 77, C : N = 4.9) were lower than Redfield proportions, whereas the average N : P ratio of 17 was consistent with the Redfield ratio. Our results suggest that strain-level diversity is an important driver of variation in the C : N : P ratios of heterotrophic bacterial biomass and that the potential importance of non-nucleic acid pools of P warrants further investigation. Continued work clarifying the range and controls on the stoichiometry of heterotrophic marine bacteria will help improve understanding and predictions of global ocean C, N and P dynamics.

Development and bias assessment of a method for targeted metagenomic sequencing of marine cyanobacteria.

Prochlorococcus and Synechococcus are the most abundant photosynthetic organisms in oligotrophic waters and responsible for a significant percentage of the earth's primary production. Here we developed a method for metagenomic sequencing of sorted Prochlorococcus and Synechococcus populations using a transposon-based library preparation technique. First, we observed that the cell lysis technique and associated amount of input DNA had an important role in determining the DNA library quality. Second, we found that our transposon-based method provided a more even coverage distribution and matched more sequences of a reference genome than multiple displacement amplification, a commonly used method for metagenomic sequencing. We then demonstrated the method on Prochlorococcus and Synechococcus field populations from the Sargasso Sea and California Current isolated by flow cytometric sorting and found clear environmentally related differences in ecotype distributions and gene abundances. In addition, we saw a significant correspondence between metagenomic libraries sequenced with our technique and regular sequencing of bulk DNA. Our results show that this targeted method is a viable replacement for regular metagenomic approaches and will be useful for identifying the biogeography and genome content of specific marine cyanobacterial populations.

Intraspecific variation in antibiotic resistance potential within E. coli.

Intraspecific genomic diversity brings the potential for an unreported and diverse reservoir of cryptic antibiotic resistance genes in pathogens, as cryptic resistance can occur without major mutations and horizontal transmission. Here, we predicted the differences in the types of antibiotics and genes that induce cryptic and latent resistance between micro-diverse Escherichia coli strains. For example, we hypothesize that known resistance genes will be the culprit of latent resistance within clinical strains. We used a modified functional metagenomics method to induce expression in eight E. coli strains. We found a total of 66 individual genes conferring phenotypic resistance to 11 out of 16 antibiotics. A total of 14 known antibiotic resistance genes comprised 21% of total identified genes, whereas the majority (52 genes) were unclassified cryptic resistance genes. Between the eight strains, 1.2% of core orthologous genes were positive (conferred resistance in at least one strain). Sixty-four percent of positive orthologous genes conferred resistance to only one strain, demonstrating high intraspecific variability of latent resistance genes. Cryptic resistance genes comprised most resistance genes among laboratory and clinical strains as well as natural, semisynthetic, and synthetic antibiotics. Known antibiotic resistance genes primarily conferred resistance to multiple antibiotics from varying origins and within multiple strains. Hence, it is uncommon for E. coli to develop cross-cryptic resistance to antibiotics from multiple origins or within multiple strains. We have uncovered prospective and previously unknown resistance genes as well as antibiotics that have the potential to trigger latent antibiotic resistance in E. coli strains from varying origins.IMPORTANCEIntraspecific genomic diversity may be a driving force in the emergence of adaptive antibiotic resistance. Adaptive antibiotic resistance enables sensitive bacterial cells to acquire temporary antibiotic resistance, creating an optimal window for the development of permanent mutational resistance. In this study, we investigate cryptic resistance, an adaptive resistance mechanism, and unveil novel (cryptic) antibiotic resistance genes that confer resistance when amplified within eight E. coli strains derived from clinical and laboratory origins. We identify the potential of cryptic resistance genes to confer cross-resistance to antibiotics from varying origins and within multiple strains. We discern antibiotic characteristics that promote latent resistance in multiple strains, considering intraspecific diversity. This study may help detect novel resistance genes and functional genes that could become responsible for cryptic resistance among diverse strains and antibiotics, thus also identifying potential novel antibiotic targets and mechanisms.

Intraspecific trait variation modulates the temperature effect on elemental quotas and stoichiometry in marine Synechococcus.

Diverse phytoplankton modulate the coupling between the ocean carbon and nutrient cycles through life-history traits such as cell size, elemental quotas, and ratios. Biodiversity is mostly considered at broad functional levels, but major phytoplankton lineages are themselves highly diverse. As an example, Synechococcus is found in nearly all ocean regions, and we demonstrate contains extensive intraspecific variation. Here, we grew four closely related Synechococcus isolates in serially transferred cultures across a range of temperatures (16-25°C) to quantify for the relative role of intraspecific trait variation vs. environmental change. We report differences in cell size (p<0.01) as a function of strain and clade (p<0.01). The carbon (QC), nitrogen (QN), and phosphorus (QP) cell quotas all increased with cell size. Furthermore, cell size has an inverse relationship to growth rate. Within our experimental design, temperature alone had a weak physiological effect on cell quota and elemental ratios. Instead, we find systemic intraspecific variance of C:N:P, with cell size and N:P having an inverse relationship. Our results suggest a key role for intraspecific life history traits in determining elemental quotas and stoichiometry. Thus, the extensive biodiversity harbored within many lineages may modulate the impact of environmental change on ocean biogeochemical cycles.

Global Phylogeography and Microdiversity of the Marine Diazotrophic Cyanobacteria Trichodesmium and UCYN-A.

Cyanobacterial diazotrophs, specifically the genera Trichodesmium and UCYN-A, play a pivotal role in marine nitrogen cycling through their capacity for nitrogen fixation. Despite their global distribution, the microdiversity and environmental drivers of these diazotrophs remain underexplored. This study provides a comprehensive analysis of the global diversity and distribution of Trichodesmium and UCYN-A using the nitrogenase gene ( nifH ) as a genetic marker. We sequenced 954 samples from the Pacific, Atlantic, and Indian Oceans as part of the Bio-GO-SHIP project. Our results reveal significant phylogenetic and biogeographic differences between and within the two genera. Trichodesmium exhibited greater microdiversity compared to UCYN-A, with clades showing region-specific distribution. Trichodesmium clades were primarily influenced by temperature and nutrient availability, and particularly frequent in regions of phosphorus stress. In contrast UCYN-A was found in regions of iron stress. UCYN-A clades demonstrated a more homogeneous distributions, with a single sequencing variant within the UCYN-A1 clade dominating across varied environments. The biogeographic patterns and environmental correlations of Trichodesmium and UCYN-A highlight the role of microdiversity in their ecological adaptation and reflect their different ecological strategies. This study underscores the importance of characterizing the global patterns of fine-scale genetic diversity to better understand the functional roles and distribution of marine nitrogen-fixing cyanobacteria.