A total of 639 metagenome-assembled genomes (MAGs) were recovered from 20 shallow sediment samples from 5 sites (4.1, 8.1, 13, 21, 24) spanning 4 seasons (July and October of 2011, January and May of 2012; Fig. 1) in SFB (see Lee and Francis, 2017 for sampling details)11. These genomes were obtained from a transect of surface sediments from North to South SFB, including Suisun Bay, San Pablo Bay, and the northern portion of South Bay. The five sites span two distinct parts of the estuary characterized by different freshwater inputs, salinities, and other environmental characteristics (Fig. 1, Supplementary Table 1)11. Our sampling scheme captures microbial communities in distinct parts of the estuary along the major salinity gradient of SFB.
To determine the identity of the MAGs, we performed a combination of phylogenomic analyses using 37 marker proteins and taxonomic assignment using GTDB-Tk v1.0.2, r207 (Fig. 2, Supplementary Table 2)18. This revealed 81 archaeal and 558 bacterial MAGs. Most archaea are Nitrososphaeria (referred to as Thaumarchaeota in text, 37 MAGs) and Bathyarchaeia (34 MAGs), while bacteria are dominated by Gammaproteobacteria (221 MAGs), Desulfobacterota (55 MAGs), and Actinobacteriota (51 MAGs). Next, we examined the metabolic capacity of these bacteria and archaea by comparing their predicted proteins to a variety of functional databases and manually curated protein phylogenies. A broad diversity of taxa and metabolic pathways were obtained in this study. We first focused on key players in the estuary according to relative abundance, which was determined by calculating the relative abundance of each MAG (see methods, MAG abundance), summing the abundance of each taxonomic group, and selecting the top 10 most abundant taxa per sample (Fig. 1). We observed that these abundant organisms captured important N and S cycling metabolisms (Supplementary Tables 3 and 4). On average, nearly 40% of SFB reads recruit to the 639 MAGs across all samples (Extended Data Fig. 1). In addition to dominant taxa, we identified organisms that are part of the rare biosphere (< 1% relative abundance; Extended Data Fig. 2). The SFB rare biosphere includes up to 21 phyla (of 31 in SFB) in each of the 5 sites, including 6 archaeal and 16 bacterial phyla.
Gammaproteobacteria Generalists Broadly Distributed Across The Estuarine Salinity Gradient
“Generalist” microorganisms have broad environmental tolerances and may drive microbial speciation and evolution19. In SFB, Gammaproteobacteria appear to be generalists, as the most broadly distributed and abundant class across the estuary. In particular, Woeseiales is the only Gammaproteobacteria order present at all sites. These MAGs contain a variety of metabolic pathways for S-cycling that are maintained over space and time, including reverse dissimilatory sulfite reductase genes (rdsr, Extended Data Fig. 3, Supplementary Table 5) and the SOX system for sulfur oxidation (Fig. 3). One of the most common N-cycling abilities in SFB Gammaproteobacteria is nirS (cytochrome-cd1 nitrite reductase) for nitrite reduction to NO, present in 41% of these MAGs. In contrast, copper-containing nitrite reductase nirK, which catalyzes the same reaction as nirS, is only present in 6% of Gammaproteobacteria MAGs (Extended Data Fig. 4). These findings are consistent with a previous qPCR-based study in SFB on the same samples11. Many SFB Gammaproteobacteria are capable of both N- and S-cycling, such as three SOX-encoding Woeseiales (genus JAACFE01) that also encode nrfAH for nitrite reduction to ammonium. In addition, SFB MAGs in orders SZUA-229 and UBA6429 often encode dissimilatory nitrate reduction pathways (napAB and nirBD), nosZ for nitrous oxide (N2O) reduction, as well as SOX and rdsrAB for sulfur oxidation. Finally, three Gammaproteobacteria (orders UBA6429 and AKS1) from site 8.1 are the only predicted complete denitrifiers in the entire dataset besides two Bacteroidetes from the same site. These complete denitrifiers encode rdsr and the SOX system for sulfur oxidation (underlined taxa in Fig. 3). In support of this, the closest cultured representative to these MAGs are Sulfuriflexus mobilis and Thiohalophilus thiocyanatoxydans, known to be capable of nitrogen compound reduction paired to thiosulfate and sulfur oxidation20,21.
North Bay Site 4.1
USGS Site 4.1 is located in the North Bay near the head of the estuary, and has the lowest salinity (average 2.3 PSU, Fig. 1) of sites sampled. In total, 163 MAGs were reconstructed here, predominantly belonging to Nitrospirota (28 MAGs) and Actinobacteriota (23 MAGs). Three archaeal MAGs were also reconstructed from July (one Lokiarchaeota, Bathyarchaeia, and Thermoplasmatota). Nitrospirota are among the most abundant MAGs in all months, while Actinobacteriota and Gemmatimonadota are abundant in all months except July (Fig. 1). Several community members are uniquely abundant in July, including Chloroflexota, SAR324, and Spirochaetota. SAR324 and Spirochaetota genomes were only obtained from sites 4.1 and 8.1, and Chloroflexota is abundant at 4.1 and 8.1, suggesting these bacteria may prefer low-mid salinity sites (0.23–16.7 PSU, Supplementary Results). Additionally, Krumholzibacteriota is an abundant taxon in the month of October, but not dominant in any other samples. This phylum has one cultured representative and is not well described in estuaries or other environments22.
The predominance of Nitrospirota at this station may be due to anthropogenic inputs of dissolved inorganic nitrogen from the nearby Sacramento-San Joaquin Delta and wastewater treatment plants23. Nine Nitrospira MAGs were identified as comammox bacteria, encoding amoABC, hao, and nxrAB (Extended Data Figs. 5, 6, and 7, Supplementary Tables 3–5), and were present in all months. Comammox are capable of the complete oxidation of ammonia to nitrate (both steps of nitrification), which has historically been thought of as a two-step process catalyzed by two distinct guilds, ammonia oxidizers and nitrite oxidizers24. At site 4.1, nitrite-oxidizing Nitrospira were identified in the same months as comammox Nitrospira (Supplementary Results).
We observed seasonal changes in sulfate-reducing bacteria (SRB) at site 4.1 (Fig. 3). In July, Desulfobacterota and Nitrospirota SRB are abundant, and shift to Gemmatimonadota SRB in October and January, followed by Desulfobacterota SRB in May. One SRB Desulfobacterota MAG is within an understudied class (SM23-61) that was only reconstructed at site 4.1 in July, suggesting this organism may prefer low salinity environments. Seasonal changes also occur in sulfur oxidizing bacteria, where rdsr-encoding Nitrospirota, SAR324, and Spirochaetota (described in more detail below) and SOX-encoding SAR324 are abundant in July, and shift to Gammaproteobacteria in later months.
Northern Station 8.1
USGS site 8.1 is within a narrow tidal strait connecting low-salinity Suisun Bay to the more marine-influenced San Pablo Bay (2.71–16.7 PSU). This site yielded the most archaea (48 MAGs), which are predominantly abundant Bathyarchaeia (top 3 or more abundant taxa in July, October, and January). Abundant community members also include Nitrospirota, Actinobacteriota, Thaumarchaeota, Nitrospinota, Bacteroidota, and Alphaproteobacteria. Heimdallarchaeota (LC-2 group) were identified here that encode a putative Cu-containing nitrite reductase (nirK in 8_1_Jan_SF_Bin58 and 8_1_May_SF_Bin13). These nitrite reductases are related to a Heimdallarchaeota archaeon in group LC-3 (OLS27132.1; Extended Data Fig. 4 and Supplementary Results)25. While nitrite reductases have recently been reported in Heimdallarchaea, they are not well studied26.
Intriguingly, all three canonical nitrifiers (AOA, AOB, and NOB) as well as comammox bacteria were present at site 8.1. Three comammox Nitrospirota MAGs were identified here, and appear to coexist with AOA Nitrosopumilus (Thaumarchaeota), AOB Nitrosomonas (Gammaproteobacteria, formerly Betaproteobacteria), and nitrite-oxidizing bacteria (NOB) Nitrospirota and Nitrospinota. The Nitrosomonas MAGs at site 8.1 are unique in that they are the only amoABC-encoding members of Gammaproteobacteria reconstructed from SFB (Extended Data Fig. 5). One Gemmatimonadota MAG (8_1_May_SF_Bin5) encodes a nitrate reductase/nitrite oxidoreductase that is closely related to Nitrospirota nxrA (Extended Data Fig. 7; Supplementary Results). While this MAG contains the beta subunit of nitrate reductase/nitrite oxidoreductase and the conserved residues in this gene, experimental evidence is needed to determine the direction of this enzyme.
Most Nitrospinota MAGs were recovered at site 8.1 (9/11 MAGs) and are present in two classes, Nitrospinia (3 MAGs) and UBA7883 (6 MAGs). These organisms have some sulfur cycling abilities, where all UBA7883 encode rdsrAB genes for sulfur oxidation and all Nitrospinota code glpE (K02439), a thiosulfate sulfurtransferase or Rhodanese, to catalyze the transfer of sulfane from thiosulfate to cyanide and form thiocyanate and sulfite27. Very few glpE genes were found in MAGs generated from SFB and are only present in 6 MAGs total in the entire dataset (1 Desulfobacterota and 2 Enterobacterales Gammaproteobacteria).
Spirochaetota are abundant at site 8.1 in October, and interestingly, two of these MAGs encode rdsrAB (classes Leptospirae and UBA6919). To our knowledge, rDsr has not been identified in Spirochaetota, though reductive DsrAB was recently identified in this phylum28 and sulfur oxidation has been shown in Spirochaeta perfilievii29. One rdsrAB-encoding MAG in class Leptospirae (8_1_Oct_SF_Bin1) also encodes sat, aprAB, dsrC, dsrEFH, and qmoABC, and lacks dsrD, which aligns with previously outlined genomic-based evidence for sulfur oxidation30. In total, there are five Leptospirae MAGs that contain rdsrAB genes on contigs up to 46 kb in length and have neighboring proteins with homology to other Leptospirae. Therefore, this is not a result of binning contamination. Three rdsr-encoding Spirochaetota MAGs from the same classes were identified in site 4.1 July, indicating these organisms may prefer low-mid salinity environments.
In addition to rdsrAB-encoding Spirochaetota, four additional taxa encoding these genes were present in the month of October, including Alpha- and Gammaproteobacteria, as well as Nitrospinota and SAR324. Some rdsrAB-encoding Gammaproteobacteria and SAR324 also encode genes for DNRA, suggesting metabolic flexibility in N- and S-cycling (underlined taxa in Fig. 3). Gammaproteobacteria and Nitrospinota with rdsrAB are present in all months, while rdsrAB-encoding SAR324 are limited to July.
San Pablo Bay
With closer proximity to the ocean, USGS station 13 has a higher salinity compared to 4.1 and 8.1 (15.9–26.2 PSU) and is dominated by Thaumarchaeota. All Thaumarchaeota MAGs here are in the Nitrosopumilus genus and almost all encode amoABC (12/14 MAGs). Interestingly, several of these MAGs are closely related to Nitrosopumilus salaria BD31, an AOA originally isolated from sediments of San Pablo Bay31. Station 13 is unique in that Gemmatimonadota are among the most abundant taxa, and some are N2O reducers (5/12 MAGs code nosZ) and contain sqr genes (type III; Extended Data Fig. 8)32. Gemmatimonadota also encodes group 1f NiFe hydrogenases (7/12 MAGs) for the oxidation of atmospheric hydrogen (Supplementary Table 5 and Extended Data Fig. 9)33. Gemmatimonadota have experimentally been shown to oxidize hydrogen using group 2a NiFe hydrogenases34, and thus validation is needed to determine the exact role of group 1f NiFe hydrogenases in this phylum.
In contrast to low-salinity sites, station 13 is the first sampling location where Desulfobacterota MAGs emerge as abundant community members. These MAGs consist of six classes that appear to catalyze distinct components of the nitrogen cycle from each other. While classes Desulfobulbia and Desulfuromonadia uniquely fix nitrogen, Desulfobulbia and MBNT15 are the only Desulfobacterota classes capable of napAB-mediated nitrate reduction, and Syntrophobacteria and Desulfuromonadia are capable of narGH-mediated nitrate reduction. MBNT15 is the only one of these classes that encodes norBC for nitric oxide (NO) reduction. These divisions suggest Desulfobacterota are important community members for mediating nitrogen-related metabolic handoffs.
Compared to sites 4.1 and 8.1, site 13 has fewer S cycling taxa. However, the greatest number of SRB Desulfobacterota MAGs were reconstructed here (12 MAGs), and these organisms are abundant in October and January, but absent in July and May. Alphaproteobacteria encoding rsdr for sulfur oxidation are present in January and May, and one of these organisms (13_Jan_SF_Bin10) is within the same family as cultured representative Magnetovibrio blakemorei, known to oxidize sulfur35. M. blakemorei also reduces nitrate and fixes CO2 via RubisCO. We did not identify N cycling genes in the Alphaproteobacteria MAG; however, we did identify a type II RubisCO (Supplementary Tables 3 and 5, Extended Data Fig. 10).
South Central Bay Station 21
Ninety-three MAGs were reconstructed from USGS station 21 (27.24–30.41 PSU), including abundant Actinobacteriota and Thaumarchaeota. Actinobacteriota in class Acidimicrobiales, which are abundant throughout all four seasons (Fig. 1), have been predicted to use numerous energy sources and electron acceptors36, encoding Sqrs, rhodopsins, Ni-Fe hydrogenases, reductive dehalogenase (rdhA), and nitrite reductases (nirS and nirK). In site 21 Acidimicrobiales, we identified nirK (2/3 MAGs), NiFe Group 3b hydrogenases (3/3 MAGs, Extended Data Fig. 9), type II SQR (3/3 MAGs, Extended Data Fig. 8), and rdhA (2/3 MAGs encode rdhA). Site 21 Thaumarchaeota encode amoABC (11/14 MAGs) for ammonia oxidation and nirK (12/14 MAGs) for nitrite reduction, and in the month of July, coexist with a nitrite-oxidizing bacteria (NOB) Nitrospirota. Among S cyclers, SOX- and rdsrAB-encoding Gammaproteobacteria are present at all months, and SOX- and rdsrAB-encoding Alphaproteobacteria occur in January and July. SRB Desulfobacterota are abundant in all months except January.
A single representative of the Tectomicrobia phylum (21_Jan_SF_Bin1, family Entotheonellaceae) was identified at station 21, and this is the only Tectomicrobium MAG reconstructed in SFB. The Tectomicrobium plays a role in nitrogen cycling, with genes for nitrite reduction (nirK and nirBD) and NO reduction to N2O, a potent greenhouse gas (norBC). Few other site 21 MAGs encode norBC, except for one Gammaproteobacteria and three Myxococcota MAGs. No other norBC-encoding Myxococcota were identified in SFB, and thus this may be a unique role for these organisms at this site, where they are abundant in all months. In addition, the Tectomicrobium encodes dmsAB for DMSO reduction, and only one Alphaproteobacteria at site 24 also has this gene. As previous studies have found37, the Tectomicrobium has a large genome size, the largest of all MAGs in this dataset, at 7.13 MB (Supplementary Table 2).
Southernmost Station
The lowest number of MAGs were reconstructed at South Bay USGS station 24 (53 MAGs total, 25.9–29.3 PSU). Thaumarchaeota are abundant in all months, yet no Thaumarchaeota MAGs were reconstructed at this site. This could be caused by microdiversity, where closely related Thaumarchaeota limit their assembly and binning38. Desulfobacterota and Actinobacteriota are also abundant in all months, while Zixibacteria is among the most abundant community members in January (Supplementary Results). One Desulfobacterota MAG (24_Jan_SF_Bin10, order Desulfobacterales) encodes both a reductive dsrA, as well as an oxidative, or rdsrA, and this is the only Desulfobacterota in SFB that has a rdsrA. Only one other MAG in the dataset has both a reductive and oxidative dsrA, a Nitrospirota from site 4.1 in July (4_1_July_SF_Bin47, class Thermodesulfovibrionia). Both Dsr-dependent and sulfur-oxidizing and reducing genes have previously been identified in Nitrospirota28 and Desulfobacterota39 but organisms that encode genes for both pathways have not been well described.
While there are few complete N pathways in one organism at site 24, it is important to note that at this site, and all others, incomplete pathways are common. Alpha-, Gammaproteobacteria, and Desulfobacterota encode partial steps of DNRA and denitrification, which are often present in specific orders of these classes. For example, nirBD-mediated nitrite reduction is only present in Gammaproteobacteria order UBA9214, and nrfAH-mediated nitrite reduction is present solely in Desulfobacterota order Desulfobulbales for the same function. One Gammaproteobacteria MAG encodes norBC for the reduction of NO to N2O, and thus this step may present a bottleneck in denitrification at site 24. These divisions further highlight the role of metabolic handoffs among SFB microbial communities.