Bioavailable nitrogen (N) is the limiting factor for phytoplankton growth throughout the majority of the (sub)tropical oceans (Moore et al., 2013). Diazotrophs, N₂ fixing microbes, supply new bioavailable N to such environments, thereby increasing the productivity of the whole community (Mahaffey, 2005). In regions where N₂ fixation is enhanced, dissolved inorganic phosphorus (DIP) is typically strongly drawn down (Martiny et al., 2019), often becoming growth-limiting for diazotrophs and co-limiting for the rest of the phytoplankton community (Moore et al., 2013).

In addition to DIP, P is also available as dissolved organic P (DOP), which accounts for up to 70–90% of the total dissolved P in surface waters of low DIP regions (Karl and Björkman, 2002). DOP utilization is mediated by enzymes that liberate DIP (Dyhrman et al., 2006); these include phosphoester hydrolyzing enzymes [e.g., alkaline phosphatase (AP), phosphodiesterase, and 5’ nucleotidase] and phosphonate catabolic enzymes (e.g., C-P lyase, substrate-specific enzymes including phosphonoacetate hydrolase and phosphonatase). Among these, several are metalloenzymes, requiring metal cofactors (Duhamel et al., 2021). Specifically, (i) AP enzymes require zinc (Zn) and magnesium (PhoA; Kim and Wyckoff, 1991), or calcium and iron (Fe) (PhoD/X; Yong et al., 2014; Rodriguez et al., 2014); (ii) phosphonoacetate hydrolases require Zn (McGrath et al., 2013); and (iii) C-P lyases contain Zn and Fe (Stosiek et al., 2020).

Alongside playing a role in regulating the rate of P acquisition from the DOP pool (Mahaffey et al., 2014; Browning et al., 2017), the trace element requirement for P releasing enzymes will also control the quotas of these elements in phytoplankton as a function of P limitation status. Altered trace element quotas should, in turn, feedback to the inventories and cycling of these elements in seawater (Duhamel et al., 2021). Clear evidence for this is, however, lacking. In this study, we use a new, trace-metal-clean culturing approach to examine the effects of different P sources on the growth of the diazotrophic cyanobacteria Trichodesmium. Our findings provide evidence for changes in micronutrient requirements, which are consistent with the diverse array of mechanisms used by marine microbes to alleviate P stress.

We tested for interactions between P limitation, DOP availability, and cellular elemental quotas in Trichodesmium ISM101 under two P supply treatments: (i) 5-μM DIP and (ii) 5-μM MPA, a phosphonate synthesized by marine microbes (Metcalf et al., 2012). Experiments were conducted at high volume (2 L) with triplicate replication under trace-metal-clean conditions using an EFB culture approach (Fischer et al., 2014; Marki et al., 2020), which allows for regular harvesting of sufficient biomass required for bulk phytoplankton elemental analysis. Trichodesmium biomass [particulate organic carbon (POC)] within the DIP treatment increased 1.4-fold over 18 days and showed no significant changes from day 12 (Quade post hoc test; p < 0.05; Figure 1A), after which we assumed that growth had reached an approximate steady-state matching the dilution rate of the culture (0.1 day^–1), and biomass concentration was dictated by the concentration of the most deficient nutrient in the culture medium (Fischer et al., 2014). The biomass of the MPA treatment showed a more complex trend, with POC ultimately decreasing to 0.7 times the initial value within 21 days, after which no significant changes were observed, and a steady state was assumed (Quade post hoc test; p < 0.05; Figure 1A).

Concentrations of DIP for both treatments were fully depleted at a steady-state (0.03 ± 0.01 and 0.02 ± 0.01 μM; Figure 1B), implying that for both sets of cultures, the low P availability limited biomass. Nevertheless, three lines of evidence suggested that (i) the level of P stress in the MPA treatments was higher than that of the DIP treatment, and (ii) that this was due to a situation approaching P-Fe co-limitation in the DIP treatment. Firstly, APA was fourfold higher in the MPA treatment, implying stronger P stress in this treatment (Figure 1D; Mahaffey et al., 2014). Secondly, the apparent photochemical PSII efficiency (F_v/F_m) was significantly lower in the DIP treatment (Supplementary Figure 2B). F_v/F_m has been previously demonstrated to remain high under steady-state P limitation (Kruskopf and Flynn, 2006), whereas under steady-state Fe limitation, large decreases are observed (Schrader et al., 2011). Thirdly, mean steady-state POC per filament in the DIP treatment was 45% lower than the MPA treatment (pairwise Wilcox test, p < 0.05), which we hypothesize was related to filament length, as Trichodesmium filaments have been shown to be significantly smaller under limitation by Fe in comparison with P (Tzubari et al., 2018). Fourth, the mean steady-state particulate Fe concentration in DIP treatment was 46 ± 16 nM, indicating that all of the supplied Fe in the medium (40 nM) was taken up by Trichodesmium. In turn, stronger P limitation in the MPA treatment could be ascribed to a lower accessibility of this P source (Dyhrman and Haley, 2006; Sosa et al., 2019), with dissolved MPA concentrations in this treatment remaining at around 40% of the supplied concentration (2.05 ± 0.93 μM; Figure 1C).

Differences in elemental composition between the two treatments at steady state were minimal for all elements apart from Zn (Figure 2 and Supplementary Figure 1), which in contrast showed 3.5-fold elevated Zn:C in the MPA treatment (0.057 ± 0.018 mmol/mol C in comparison with 0.016 ± 0.009 mmol/mol C in the DIP treatment). This finding is consistent with a recent evaluation of trace element requirements for mechanisms to cope with P limitation (Duhamel et al., 2021): of all the trace elements assessed, Zn stands out as having the greatest demand, with elevated requirements in comparison with all other metals across a diversity of transferases, hydrolases, lyases, isomerases, ligases, and translocases (Duhamel et al., 2021). With our set of observations, we cannot decouple the specific cellular sinks for Zn that occurred within the MPA treatment. However, notable was that APA in MPA treatment increased from day 6 and stabilized from day 9 (Figure 1D), whereas the increase in Zn:C was later, from day 12, and did not reach a clear plateau over the experiment duration (30 days; Figure 1E). The latter observation suggests that the Zn enhancement was not simply due to the well-characterized enhanced production of AP enzymes under greater P stress (and specifically, Zn-requiring PhoA; Mahaffey et al., 2014) but rather to a diversity of strategies to cope with stronger P limitation (Wei et al., 2018; Duhamel et al., 2021). Furthermore, the lack of increase in Fe requirements under stronger P limitation (Figure 1F) is inconsistent with the expectation of more dominant roles of PhoX and PhoD over PhoA (Luo et al., 2011) but is consistent with (i) the Fe requirement for these enzymes being relatively minor in comparison to overall cellular Fe demands and (ii) the lower Fe—but dominant Zn—requirement in a variety of other, non-AP, enzymes produced in response to low P (Duhamel et al., 2021).

Our observations of enhanced Zn quotas under stronger P limitation in culture are consistent with available field observations (Tovar-Sanchez et al., 2006; Twining et al., 2010). Tovar-Sanchez et al. (2006) found that Zn:P quotas of Trichodesmium in the low P eastern Subtropical North Atlantic were ∼8.6 mmol mol^–1, very close to the mean steady-state Zn:P quota that we observed in the MPA treatment (8 ± 3.7 mmol mol^–1). Additionally, Twining et al. (2010) found that the Zn quotas of Synechococcus in Sargasso Sea eddies with lower DIP supply were roughly 20 times higher than those with higher DIP supply. Together, our observations in culture and these findings in the field suggest a crucial level of interaction between the cycles of these two nutrients in the ocean. The impacts of such an interaction might be exacerbated in a future ocean, where increasing atmospheric N inputs could drive a shift toward more widespread and stronger P stress (Moore et al., 2013).

The original contributions presented in the study are publicly available. This data can be found here: doi: 10.5281/zenodo.5607662.

TB, MG, and XW designed the study. XW performed culture work, data collection, analysis, and wrote the manuscript. TB, EA, and MG contributed editorial comments and approved the final version of the manuscript. All authors contributed to the article and approved the submitted version.

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.