Diazotrophic cyanobacteria are critical for driving inputs of new nitrogen (N) into the oligotrophic ocean and supporting carbon fixation (Gruber and Galloway, 2008; Zhang et al., 2019), particularly where marine primary production is limited by nitrogen (Moore et al., 2013). However, the composition and distribution of diazotrophs, as well as their contribution to dinitrogen (N₂) fixation, in the western South Atlantic remain largely unknown, and this presents a gap in global nitrogen fixation estimates and hampers related modeling efforts (Coles and Hood, 2007; Sohm et al., 2011; Snow et al., 2015; Landolfi et al., 2018; Tang and Cassar, 2019; Wang et al., 2019).

Much of what is known about marine nitrogen fixation is based on studies from the North and South Pacific (e.g., Church et al., 2009; Benavides et al., 2016; Zhang et al., 2019; Zehr and Capone, 2020), and the tropical and subtropical regions of the North Atlantic (e.g., Orcutt et al., 2001; Capone et al., 2005; Montoya et al., 2019; Tang et al., 2019; Palter et al., 2020), with comparatively few observations in the South Atlantic (e.g., Fernández et al., 2010; Sohm et al., 2011; Snow et al., 2015; Wasmund et al., 2015). Visible surface “slicks” of Trichodesmium have been observed in the southwestern Atlantic Ocean (Carvalho et al., 2008; Detoni et al., 2016), which are usually associated with warm surface water of the poleward-flowing Brazil Current (BC), and its meanders at the shelf-break region. Nitrogen fixation is thought to be largely attributable to Trichodesmium in this region (Moore et al., 2014), yet nitrogen fixation by unicellular diazotrophic cyanobacteria Crocosphaera watsonii, host-associated groups like UCYN-A, and diatom-diazotroph associations (DDAs) in other areas of the global ocean can be equal to or greater than those of Trichodesmium (Church et al., 2005a; Foster et al., 2007; Church et al., 2009; Fernández et al., 2010; Wu et al., 2018; Zehr and Capone, 2020).

The distribution and abundance of diazotrophic cyanobacteria can be tracked with measurements of nifH gene copies for different phylotypes, among other approaches (e.g., Karlusich et al., 2021). Database compilations of diazotroph distributions (Luo et al., 2012; Tang and Cassar, 2019) provide critical resources for studying the factors which drive these organisms’ distribution and for modeling nitrogen fixation. However, these observations are regionally skewed, and Tang and Cassar (2019) noted the limitations posed by the lack of observations in the South Atlantic. In the present study, we addressed this knowledge gap by examining diazotroph community structure in the western South Atlantic.

Samples were obtained from 18 stations during a cruise aboard the R/V Alpha Crucis (February 2 – 17, 2017) along the Vitória-Trindade Ridge (21°S, 1200 km transect), extending from the continental slope to the western limb of the South Atlantic subtropical gyre (Figure 1). The first three stations were located in meridional positions, following the 1000 m isobath, and subsequent stations followed the deeper isobaths along approximately the same latitude (around 20°S).

Recent synthesis and modeling efforts (Tang and Cassar, 2019; Karlusich et al., 2021) have highlighted that diazotroph distribution and community structure are under-sampled in the South Atlantic. Herein, we addressed this knowledge gap with a western South Atlantic survey of nifH genes representing six diazotroph phylotypes. The transect extended 1200 km, roughly along 20°S, traversing three sectors (West, Middle and East). These sectors were distinguished by temperature, which was significantly higher in the West relative to the other sectors. This likely reflects the influence of tropical water associated with the BC. Temperature significantly positively co-varied with the Trichodesmium, Crocosphaera and Het-2 nifH gene copy abundances, and average Trichodesmium nifH gene copies were highest in the West sector. Water temperature is a well-known factor controlling the distribution of diazotrophs (Luo et al., 2014), and typically their distribution has been thought to be largely restricted to warm, stratified, high solar radiation regions in the tropics and subtropics.

The average nifH gene distribution along the transect was dominated by Crocosphaera signals, and the average mixed layer nifH gene copies (~100 x 10³ nifH copies L^-1) were in the general range of Crocosphaera nifH gene abundance found in other regions such as the North Pacific (10³ to 10⁴ nifH copies L^-1; Church et al., 2005a) and the western tropical South Pacific (10² to 10⁶ nifH copies L⁻¹; Stenegren et al., 2018). Recent imaging work by Karlusich et al. (2021) identified Crocosphaera sp. that were clustered together, or associated with large diatoms (e.g., Climacodium), in addition to solitary cells. This highlights the potential for disparate ecologies within the phylotype which may, in part, support the broad distribution observed here. The apparent dominance of the Crocosphaera phylotype in this dataset suggests that it is an important contributor to nitrogen fixation in this region. However, the activity of the phylotypes cannot be accurately extrapolated from these results.

The average mixed layer Trichodesmium nifH gene copies were the second most abundant on the transect (~20 x 10³ nifH copies L^-1). This average is within the range observed in other systems such as the western tropical South Pacific (10² to 10⁶ nifH gene copies L^-1; Stenegren et al., 2018). The distribution of Trichodesmium is known to be highly variable, ranging from concentrations so high they are visible in pigmented surface slicks to non-detectable in this region (Lima et al., 2019). Here, the concentration of Trichodesmium nifH gene copies was significantly positively correlated with temperature, with a stronger correlation coefficient than that observed for Crocosphaera. Crocosphaera occupies a broader temperature range (15-30°C) than Trichodesmium (Fu et al., 2014), which might, in part, underpin the differences in the distribution of these two phylotypes in this study.

A recent study identified the widespread occurrence of a non-diazotrophic Trichodesmium species reconstructed from a global metagenomic survey (Delmont, 2021). By tracking nifH genes, this, and other studies that use this approach, may under-count Trichodesmium spp. Although the presence of the nifH gene for this phylotype assay suggests these Trichodesmium species have the ability to fix nitrogen, we can’t exclude the possibility that some portion of the Trichodesmium population lacks the full complement of genes required for nitrogen fixation. Thus, the values reported here could be an underestimate of Trichodesmium abundance.

The host-associated diazotrophs (UCYN-A, Het-1, and Het-2) had lower average abundances of nifH genes than Crocosphaera or Trichodesmium in the mixed layer. Although UCYN-A is increasingly recognized as a widely-distributed and abundant diazotroph (Martínez-Pérez et al., 2016), the UCYN-A nifH gene copy number observed here was not the dominant signal. This pattern is consistent with a study by Stenegren et al. (2018) in the western tropical South Pacific, where UCYN-A1 and A2 ecotypes were low to non-detectable and the system was dominated by Trichodesmium and Crocosphaera signals. Also, Karlusich et al. (2021) found low abundances of the UCYN-A nifH gene in the South Atlantic Ocean, making up only 3-4% of the bacterioplankton population.

The low copy number, or lack of detection, of nifH genes from the other host-associated diazotrophs Richelia (Het-1 and Het-2) and Calothrix (Het-3) may be, in part, driven by the silica requirement and physiological ecology of their diatom hosts. These DDAs are often observed in the transitional waters of river plumes (Foster et al., 2009; Montoya et al., 2019), although they can be observed in other open ocean systems (Karl et al., 2012; Stenegren et al., 2018). Overall, the low abundances of Het-1 nifH gene copies (10²-10³ nifH gene copies L^-1) among samples over the transect were consistent with studies from the North Atlantic, which have generally found low abundances of Het-1 (~10² nifH copies L^-1) (Benavides et al., 2016), and are in the range of Het-1 nifH gene copies found in the western tropical South Pacific (10¹-10⁵ nifH gene copies L^-1). We note that the lack of Het-3 nifH gene copies detected in this study is also consistent with the findings of previous work in the South Pacific and North Atlantic where Het-3 signals were low or non-detectable (Foster et al., 2007; Stenegren et al., 2018). In sum, the community composition of nifH phylotypes observed here is in the range of other systems, but requires more information across a broader swath of the South Atlantic to determine whether the patterns observed here are representative of this region.

A recent study by Tang and Cassar (2019) used a broad compilation of nifH phylotype data to model and then extrapolate phylotype distribution in the global ocean. They predicted that the western South Atlantic would transition from Trichodesmium dominance to UCYN-A dominance at roughly 30°S, but acknowledged the lack of data for this region (Tang and Cassar, 2019). With the dominance of the Crocosphera nifH phylotype, the observations in this study do not support this model output, but are limited in number. Expanding on the data presented here with a larger set of observations and related chemical and physical data would be valuable for future synthesis and modeling efforts.

Here we observed mixed diazotrophic assemblages with the distribution of colonial, free-living, and host-associated forms along the transect. Tracing diazotroph abundance, and the factors that drive both abundance and community structure, are critical to building robust models of marine nitrogen and carbon cycling. This is only becoming more pressing in a changing ocean. The data presented herein provides valuable information on the composition and abundance of the diazotrophic community in an under-sampled region in the western South Atlantic. Further expanding studies such as this in the South Atlantic would help to validate models of diazotroph biogeography and their effects on ecosystem processes such as primary production.

The original contributions presented in the study are included in the article/Supplementary Materials. Further inquiries can be directed to the corresponding author.

AMSD collected and processed samples. AMSD and AS analyzed the data. AMSD and PHRC contributed to the study implementation. STH and STD supported and advised the molecular methods. All authors discussed the results and contributed to the final manuscript.

We are grateful for the project support from the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Projeto ILHAS Processes: 458583/2013-8). Postdoctoral support to AD was funded by the Coordination for the Improvement of Higher Education Personnel Foundation (CAPES, Proejto REMARSUL processes: 23038.004299/2014-53) for one year at the Lamont-Doherty Earth Observatory – Columbia University. This project was partly supported by the Gordon and Betty Moore Foundation Grant #4886 and US National Science Foundation grant 1737128 to AS.

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.