Biological N₂ fixation has evolved early in Earth’s history (Falkowski, 1997), when the ocean was void of oxygen (O₂) and fixed N but rich in dissolved iron (Fe²⁺). The appearance of this process marked the beginning of the modern ocean about 2.5 billion years ago (Canfield et al., 2010). It is thus not surprising that the nitrogenase enzyme complex, which catalyzes the energy-demanding reduction of the inert N₂ to NH₄⁺ (Postgate, 1982), functions only under strictly anaerobic conditions and has an elevated Fe²⁺ requirement (Kustka et al., 2003). As aerobic N₂ fixers evolved and ocean chemistry changed, diazotrophs developed numerous strategies to protect nitrogenase from O₂, including elevated respiration rates and temporal or spatial separation of oxygenic photosynthesis from N₂ fixation (Berman-Frank, 2001).

The ability to fix N₂ is associated with additional energetic costs that are generally understood to yield lower growth rates of N₂ fixing phytoplankton as compared to their non-N₂ fixing phytoplankton competitors. Key costs are the breaking of the triple bond and reduction of N₂ to NH₄⁺, and include the indirect costs required to maintain a functioning nitrogenase complex, e.g., an anaerobic intracellular environment (Großkopf and La Roche, 2012). While diazotrophs are generally considered facultative (Zehr and Capone, 2020), nitrate utilization seems to be beneficial only at rather high concentrations (above 7 μmol L^–1, Holl and Montoya, 2005; Pahlow et al., 2013). We restrict our analysis here to conditions where diazotrophs fix N₂. The slow-growth assumption has long guided our understanding of the competitive ability of diazotrophs (Redfield et al., 1963). It has been the basic tenet of resource competition theory (RCT) to explain coexistence patterns in idealized systems (Tilman, 1980), to understand N-inventory regulatory mechanisms on long time scales (Tyrrell, 1999; Gruber, 2004), and to simulate diazotroph activity in state-of-the-art global biogeochemical models (Ward et al., 2013; Dutkiewicz et al., 2014; Landolfi et al., 2015, 2017, 2018; Wang et al., 2019; Pahlow et al., 2020). Elevated growth rates have been reported recently for several diazotrophic species (Turk-Kubo et al., 2018), further questioning our conceptual understanding of diazotrophy. Nevertheless, the competitive advantage of N₂ fixation in slowly growing autotrophic diazotrophs with elevated Fe-requirements should be restricted to N-limited and/or Fe-replete regions, leading to the traditional resource or “bottom-up” control paradigm of their ecological niche. This bottom-up control of autotrophic diazotrophs can be visualized using the simplified RCT graphical approach with nitrogen (N) and phosphorus (P) as limiting resources, in the absence of grazing pressure (Figure 1A).

Novel emerging data suggest that diazotrophs are distributed far more widely than the bottom-up control by N, Fe, and P predicts, spanning a wide range of ocean environments from surface to deep sea (Benavides et al., 2018) and hydrothermal vents (Mehta and Baross, 2006), from warm tropical to cold polar regions (Moisander et al., 2017; Shiozaki et al., 2017; Harding et al., 2018; Mulholland et al., 2019), from permanently oligotrophic waters of the subtropical gyres to high-NO₃^– oceanic and coastal waters (Knapp, 2012; Turk-Kubo et al., 2018) to aphotic oxygen minimum zones (Löscher et al., 2014) and hypoxic basins (Hamersley et al., 2011; Farnelid et al., 2013). These oceanic environments are populated by genetically and morphologically diverse diazotrophs reflecting a large ecological diversity, from unicellular (Zehr et al., 1998) free-living (UCYN-B) and symbiont species (UCYN-A, UCYN-C) and diazotroph-diatom associations, DDA (Foster et al., 2011) to multicellular species with heterocysts (Nodularia and Aphanizomenon) and colonial forms (Trichodesmium), including also non-cyanobacterial diazotrophs (Riemann et al., 2010; Moisander et al., 2017). Can recent observations be reconciled with theoretical predictions?

Laboratory experiments (Knapp, 2012) as well as RCT analyses and results from numerical models of different complexity (Pahlow et al., 2013; Landolfi et al., 2015; Inomura et al., 2018) collectively suggest that autotrophic diazotrophs can still coexist with non-N₂ fixing phytoplankton where fixed forms of N (e.g., NO₃^–, NH₄⁺, and NH₃) are available, which may be due to a combination of co-limitation by N and P (and possibly Fe or other micro-nutrients), greater N requirements for nutrient acquisition (Klausmeier et al., 2004) and the high competitive ability of diazotrophs for P (Pahlow et al., 2013). In models, with explicit representation of physiologically costly nutrient acquisition strategies, coexistence among organisms that compete for limited resources occurs for a wide range of nutrient-supply conditions (Pahlow et al., 2013, 2020; Landolfi et al., 2015; Inomura et al., 2018; Chien et al., 2020). Insight from these recent works highlights the key role of competitive interactions in setting the ecological niche of diazotrophs.

There is a growing understanding of how environmental factors affecting diazotroph physiology, with regard to temperature optima (Moisander et al., 2010), Fe requirements (Saito et al., 2011), NO3- tolerance (Knapp, 2012; Inomura et al., 2018), O₂ inhibition (Stal, 2009), and competition for P (Pahlow et al., 2013; Landolfi et al., 2015), shape competition patters. There is still little knowledge, however, of how ecological interactions, such as selective mortality by zooplankton grazing and/or viral-lysis, modulate the ability of diazotrophs to compete for resources in relation to their main competitors and thereby influence their ecological niche.

Top-down control by grazers (Prowe et al., 2012a; Vallina et al., 2014) and mortality by viral infection (Suttle, 2007; Weitz et al., 2015) are thought to exert a major control on plankton diversity and coexistence, driving adaptation and evolution. For example, grazing can drive changes in cell size and morphology (Fenchel, 1980; Branco et al., 2020) as well as defense mechanisms (Lürling, 2021). Selective mortality, via grazing or viral-lysis, leads to increased diversity and coexistence (Thingstad, 2000), contributing, together with the spatial and temporal heterogeneity of the environment, to maintain plankton diversity (Hutchinson, 1961) on seasonal to centennial time scales (Barton et al., 2010; Tsakalakis et al., 2018; Dutkiewicz et al., 2020).

To better understand how selective mortality affects phytoplankton competition for resources and coexistence patterns, we use a simple model with one nutrient (N), a fast-growing and a slowly growing phytoplankton (Phy_fast and Phy_slow) and one zooplankton (Z), and interpret our results within the RCT theoretical framework. The RCT predicts the outcome of competition and co-existence of autotrophic phytoplankton based on the R^∗ hierarchy. R^∗ is the resource concentration required by each phytoplankton type to reach equilibrium growth (growth = mortality). At equilibrium, each phytoplankton type draws down the ambient nutrient concentrations to its own equilibrium requirements R^∗. It follows that different phytoplankton types can coexist only if each has at least one resource for which its R^∗ is less than that of all others. The R^∗ hierarchy depends on the combinations of physiological characteristics (maximal growth rate, μ_max, and half saturation constants, k) and mortality terms [specific mortality m and grazing gZ, where g is the grazing rate per unit biomass (m³ mmol^–1 day^–1) and Z (mmol m^–3) is zooplankton biomass] of each phytoplankton type, as expressed by Eq. 1 (see appendix for derivation).

Traditionally, RCT considers the simplest case of similar mortality terms for all phytoplankton types, assuming constant specific mortality m and negligible effects of grazing (gZ = 0; Tilman, 1980). Under these simplified assumptions, the phytoplankton with the largest maximal growth rate (μ_max; or lowest nutrient half saturation constant k) will have the lowest resource requirement R^∗ at equilibrium (Figure 1D, dashed blue line), and will be the superior competitor, out-competing the slower-growing competitor (Figures 1D,E, dashed red line; Tilman, 1980). Recognizing that mortality terms may differ among phytoplankton and thereby relaxing the original RCT assumption of negligible grazing effects, we find that differential mortality, in our example mediated by selective top-down processes, but which could also include viral-mediated processes, can allow coexistence for a wider range of conditions. If zooplankton graze equally on two phytoplankton types (i.e., gZ_fast = gZ_slow) the more slowly growing one will be competitively excluded as in the simplified gZ_fast = 0 case (Figure 1E, dashed lines). However, selective grazing on fast-growing phytoplankton (i.e., gZ_fast > gZ_slow, Figure 1F), can change the hierarchy of R^∗ (Figure 1D, solid lines). This allows the inferior competitor, otherwise out-competed in the absence of selective grazing, to survive and coexist (Figure 1E, compare red lines) in top-down controlled systems (Prowe et al., 2012b; analytical derivation in Supplementary Material). The relative importance of top-down control on R^∗ is not fixed but varies, increasing with resource levels, or “fertility” of the system, as a larger zooplankton biomass can be supported (Figure 1G). With selective grazing, the effect of top-down control on R^∗ increases with system “fertility.” It should be noted that physiologically costly defense strategies (e.g., morphological and size changes, toxicity, etc.) have associated trade-offs, implying a reduction of the maximum growth rate that would affect R^∗ (Eq. 1) and the resulting R^∗ hierarchy. The extended RCT framework illustrates that selective top-down control can prevent the fastest-growing organism from exploiting all of the limiting resource, expanding the niche of slower-growing species in top-down controlled regions, effectively providing for novel coexistence regimes compared to bottom-up control and non-selective mortality. Although simple, the RCT principles can help interpret phytoplankton biogeographies emerging in complex global ecological-biogeochemical models that include multiple limiting nutrients, complex grazing functions and loss terms, and circulation and mixing (Dutkiewicz et al., 2009; Ward et al., 2012, 2014).

In the specific case of autotrophic diazotrophs (D) competing with faster-growing non-fixing phytoplankton (Phy), we now extend the traditional bottom-up paradigm within the RCT graphical framework (Figure 1A) to include selective grazing on Phy (Figures 1B,C). This effectively expands the diazotrophs’ region of co-existence by increasing the minimum nutrient requirements, N_eq and P_eq, of Phy (Figures 1B,C). Also, when accounting for interdependent N and P requirements (Sterner and Elser, 2002) as in chain models (e.g., Pahlow et al., 2013), selective grazing on Phy enlarges the niche of diazotrophs relative to bottom-up control only (Supplementary Figure 2). While competition for resources (P and Fe) has been central for explaining the diazotrophs’ spatiotemporal distribution in models (Landolfi et al., 2013, 2015, 2017; Ward et al., 2013; Dutkiewicz et al., 2014), the role of top-down control in modulating the space of diazotroph-non-diazotroph coexistence remains insufficiently explored (Wang et al., 2019). Our expanded resource-competition analysis suggests that physiological characteristics determine competitive outcomes in bottom-up controlled (nutrient scarce) environments, but as environmental resource levels increase, competition patterns become modulated by selective mortality. This can expand the ecological niche of autotrophic diazotrophs. A more comprehensive mechanistic understanding of the links between phytoplankton traits, environmental factors and ecological interactions (competition, predation, defense strategies, and mortality) is required.

The remarkable variety of growth strategies that allows diazotrophs to flourish in waters ranging from warm oligotrophic regions to cold, nutrient-rich, and highly productive systems prompts the question: What are the underlying advantages of fixing N₂, given the associated additional energetic costs? This remarkable and widely distributed ability demands a better understanding of interdependencies between physiology and ecological dynamics (competition and predation) that set the broad ecological niches of diazotrophs. While our comprehension of physiological constraints on diazotrophs and their environmental (bottom-up) sensitivity is growing, the role of selective mortality mediated by top-down processes and/or viral-lysis in shaping their ecological niche is less clear. Relaxing the RCT’s equal mortality assumption, we identify selective mortality of faster-growing competitors as a key process for expanding the niche of autotrophic diazotrophs in nutrient rich regions. However, to date, observational evidence is limited and insufficient to support the occurrence of selective grazing against diazotrophs. Identifying traits and trade-offs associated with selective top-down control (changes in size and morphology, nutritional quality, toxicity, and DOM release) and linking them in a multi-trait perspective remains a fundamental challenge for elucidating the mechanisms that allow the ecological complexity needed for insightful model applications. Whether future environmental and ecological changes will introduce benefits or disadvantages for N₂ fixers will depend on how these changes affect the competitive ability of diazotrophs in relation to their main competitors and predators on seasonal to centennial timescales. This suggests that we need to move beyond correlative relationships and instead establish mechanistic links between physiologically costly traits and their function in ecological dynamics. Understanding and resolving these links is key to making ecological complexity and its impact on, and interaction with, marine nitrogen fixation emerge in biogeochemical models and allowing for more reliable predictions of the future ocean.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

This is a contribution from the “N₂-fixation pathfinder group” at the GEOMAR Biogeochemical Modelling group. AL has written the original draft and provided the resource competition analysis. All the authors have discussed and commented-on various stages of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.