Novel connections between B-vitamins and microbial communities along biogeochemical gradients in a large temperate estuary

34 As B -vitamins are organic cofactors required by prokaryoGc and eukaryoGc planktonic 35 cells, their availability impact s aquaGc microbial communiGes and associated biogeochemistry. 36 Contrary to inorganic nutrients, measurements of B -vitamins from brackish systems are scarce 37 and relaGonships between B-vitamins and plankton composiGon in estuaries are unclear, limiGng 38 our understanding of estuary biology in general as well as how B -vitamins are distributed and 39 dispersed in marine systems . Here, we quanGfy mulGple B -vitamins and their vitamers in 40 parGculate and dissolved phases , and characterize microbial community composiGon, across 41 fresh to polyhaline zones of the Neuse River Estuary (NRE), N orth Carolina, USA. We uncover 42 elevated concentraGons of B-vitamins within the mid -estuary, Chlorophyll a max imum along with 43 a unique suite of dissolved B-vitamin associated with sporadic surges in pico- and microplankton 44 populaGons. The dynamics of both dissolved and parGculate B-vitamin concentraGons in space 45 and Gme were striking - from subpicomolar to high picomolar levels observed and strong short-46 term (weeks) variability. We find notable autochtonous B-vitamin producGon in the estuary, but 47 we expect the ability of the system to deliver these micronutrients to the ocean will depend on 48 flushing as well as changes in microbial community . We idenGfy vitamin B1, B12, psB12 49 (pseudocobalamin), and B3 as key explanatory variables for change in prokaryoGc and eukaryoGc 50 NRE plankton, providing new evidence of B-vitamin influence upon estuarine plankton 51 community composiGon . Our work reveals new complexiGes in B-vitamin producGon and 52 consumpGon within zones of estuaries while underscoring these micronutrients as key drivers of 53 microbial plankton composiGon . 54 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 3

55 Estuaries exhibit dynamic hydrological and physical-chemical properGes which alter 56 biology along the river to ocean conGnuum (Bianchi 2007). Microbial (prokaryoGc and eukaryoGc) 57 communiGes drive biogeochemical cycling and primary producGon in aquaGc environments – 58 including estuaries – and the diversity and physiology of these microbes can vary notably across 59 temporal and spaGal scales (Peierls et al. 2012; Hall et al. 2013; Wang et al. 2020). Accordingly, 60 idenGfying key factors that influence microbial communiGes over space and Gme is of significant 61 interest. Monitoring inorganic (macro)nutrients in aquaGc ecosystems and microbial community 62 responses has been the focus of many research and monitoring efforts (Paerl 2006). For example, 63 elevated nitrogen and phosphorus inputs are parGcularly impacmul to estuarine microbial 64 communiGes , leading to eutrophicaGon and promoGon of harmful algal blooms (HABs) 65 (Wurtsbaugh et al. 2019). 66 The Neuse River Estuary (NRE, North Carolina) is a major tributary of the 2 nd largest 67 estuary complex in the USA, the Albemarle Pamlico Estuarine System (APES) , characterized by 68 salinity and nutrient gradients, seasonality, and hydrological variaGon that impact s plankton 69 abundance and composiGon along the estuary (Pinckney et al. 1998; Peierls et al. 2012; Hall et 70 al. 2013; Gong et al. 2018) . In the NRE, moderate to low river flow leads to biological 71 heterogeneity – specifically a Chlorophyll a (Chl a) max mid -estuary and increased planktonic 72 biomass. Similar to other temperate estuary systems, the main phytoplankton groups in NRE are 73 chlorophytes, diatoms, dinoflagellates and cryptophytes (Pinckney et al. 1998; Gong et al. 2018, 74 2020) with high abundances of cyanobacteria during summer (Gaulke et al. 2010; Hall et al. 2013). 75 ProkaryoGc plankton community analyses in the NRE have focused on pathogens (e.g. Vibrio spp.) 76 (Froelich et al. 2019) and recently cyanobacterial populaGons (Sánchez-Gallego et al. 2025), and 77 dominant populaGons are expected to be similar to other temperate estuaries (Wang et al. 2020). 78 Changes in inorganic nutrients and/or hydrological properGes (e.g. salinity, discharge) explain 79 some, but not all, of the observed variability in microbial plankton in the NRE (Peierls et al. 2012; 80 Hall et al. 2013) and beyond. Micronutrients like tracemetals and B -vitamins are commonly 81 overlooked and the impact of B-vitamins on microbial communiGes within estuary has not been 82 robustly studied. 83 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 4 B-vitamins are a group of eight essenGal micronutrients that primarily funcGon as enzyme 84 cofactors in cells - but also as anGoxidants or riboswitch ligands (Lukienko et al. 2000; Miranda-85 Ríos et al. 2001). Vitamin B1 (B1, thiamin), vitamin B2 (B2, riboflavin), vitamin B3 (B3, niacin or 86 niacinamide), vitamin B5 (B5, pantothenate), vitamin B6 (B6, pyridoxine) and vitamin B12 (B12, 87 cobalamin) are essenGal for most cells – with the excepGon of some organism s lacking B12 88 requiring enzymes altogether (e.g. SAR11 bacterioplankton) (Carini 2013). Most prokaryoGc and 89 eukaryoGc plankton taxa are auxotrophic (unable to synthesize a vitamin they require for 90 metabolism de novo, termed auxotrophs) for one or more B-vitamins (Tang et al. 2010; Sañudo-91 Wilhelmy et al. 2014; Paerl et al. 2018). As a result, these taxa require an exogenous source of the 92 respecGve vitamin or vitamers (vitamin related com pounds, such as precursors for de novo 93 synthesis or degradaGon products) to meet cell needs. 94 Common NRE phytoplankton groups are likely to include auxotrophs and taxa with disGnct 95 vitamin requirements. Cyanobacteria are B1 prototrophs (organisms capable of de novo syntesis) 96 and are hypothesized to be dominant summerGme B1 synthesizers of coastal brackish 97 environments (Sañudo-Wilhelmy et al. 2014; Bi,ner et al. 2024) . Cyanobacteria also produce 98 pseudocobalamin (psB12), a cobalamin analog that is not biologically available to most other 99 plankton but can be made useful to other plankton groups aeer microbe -mediated chemical 100 remodeling (Helliwell et al. 2016; Heal et al. 2017; Bannon et al. 2024b). We hypothesize that the 101 availability of B-vitamins are important drivers of unresolved microbial plankton variability as 102 seen in other marine systems (Paerl et al. 2018; Joglar et al. 2020; Bannon et al. 2025). There is 103 also evidence of changes in microbial community composiGon during limiGng B -vitamin 104 condiGons, potenGally altering community composiGon of lower trophic levels and leading to 105 vitamin deficiency at higher trophic levels (e.g. seabirds, fish) (Balk et al. 2009; Joglar et al. 2020). 106 Therefore, quanGficaGon of B-vitamins and vitamers in aquaGc ecosystems is of considerable 107 interest as it potenGally alters producGvity, biomass, and/or the success of specific taxa. Only few 108 studies simultaneously quanGfied mulGple B-vitamins in the dissolved phase (Sañudo-Wilhelmy 109 et al. 2012; Heal et al. 2014; Suffridge et al. 2017; Bruns et al. 2022, 2023; Bannon et al. 2025), 110 and measurements of B -vitamins/vitamers in environmental plankton biomass are even more 111 limited (Suffridge et al. 2017, 2018; Bannon et al. 2025). 112 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 5 Knowledge on B-vitamins and vitamers dynamics along the freshwater, brackish to coastal 113 ocean gradient is lacking, and their connecGon to patchy biology are unknown - contrasGng with 114 well-described and modeled estuary macronutrient dynamics (Twomey et al. 2005). A few studies 115 have quanGfied select B-vitamins /vitamers in a few brackish systems, and shown that their 116 concentraGons vary by two orders of magnitude between systems (Sañudo-Wilhelmy et al. 2006; 117 Gobler et al. 2007; Koch et al. 2012; Heal et al. 2014; Tovar-Sánchez et al. 2016; Gómez-Consarnau 118 et al. 2018; Bruns et al. 2022, 2023; Möller et al. 2022; Bi,ner et al. 2024) . While some of this 119 observed variability may be due to analyGcal difficulGes and differences, o verall, data on 120 environmental concentraGons of vitamin s/vitamers are scarce and the effects of these 121 micronutrients on microbial plankton growth and community dynamics in estuarine systems are 122 not well understood . Thus, it remains unclear whether estuaries could be significant 123 allochthonous sources of B-vitamins to the coastal ocean. Understanding these potenGal sources 124 is important in a broader biogeochemical context, as the extremely low picomolar concentraGons 125 of dissolved B-vitamins in coastal and open ocean may limit microbial growth. 126 Here, we leverage substanGal ongoing biogeochemical monitoring within the NRE and 127 tackle the following research quesGons: (1) How do B-vitamin concentraGons change short-term 128 (weeks) along the salinity gradient of a temperate estuary (fresh to polyhaline)? (2) What are the 129 relaGonships between B-vitamins and size-differenGated planktonic communiGes? To address 130 these quesGons, we measured B -vitamins and vitamers in two parGculate size fracGons 131 (picoplankton 0.22 -3 µm; nano - /microplankton 3 -90 µm) and in the dissolved phase using 132 targeted liquid -chromatography mass spectrometry. Simultaneously, we characterized 133 picoplankton (0.22-3 µm) and nano-/microplankton (3 -90 µm) communiG es by 16S and 18S rRNA 134 gene sequence analyses. We hypothesized that (1) B-vitamin concentraGons are dynamic on short 135 Gme scales and exhibit disGnct pa,erns along the estuary relaGve to macronutrients; and (2) 136 plankton community composiGon is significantly impacted by B-vitamin /vitamer availability , 137 especially B1 and B12 . The results provide a high resoluGon into the short-term dynamics of 138 mulGple B -vitamins and vitamers across phases and along a salinity gradient of a temperate 139 estuary. 140 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 6

141 Sampling 142 Near surface water (0.5 m ) was collected on 26 July, 13 and 27 September, and 11 and 25 October 143 2021 from staGons NRE0, NRE50, NRE70, NRE100, NRE120, NRE180 in collaboraGon with the 144 University of North Carolina at Chapel Hill InsGtute of Marine Sciences (UNC-IMS) Neuse River 145 Estuary Modeling and Monitoring Project (MODMON; h,ps://paerllab.web.unc.edu/modmon/ ; 146 Fig. 1A). NRE160 was sampled once on 11 Oct as weather condiGons did not permit sampling at 147 NRE180. Neuse River flow data was obtained from USGS gauge 02091814 near Fort Barnwell 148 upstream of the NRE (USGS NaGonal Water InformaGon System Web Interface: 149 h,ps://waterdata.usgs.gov/nwis ; SupporGng InformaGon Fig. S1). 150 151 152 Fig. 1. Environmental condiGons at estuary sampling sites. Map of NRE staGons sampled and 153 distance along the river from staGon NRE0 is provided in parenthesis, and color gradient shows 154 median N:P raGo along the estuary from the sampling Gme points (A), based on inorganic nutrient 155 data from the MODMON monitoring program . Gradients in salinity ( B), temperature ( C), 156 parGculate organic carbon (POC, D) and dissolved organic carbon (DOC, E) across the staGons and 157 Gme points sampled . Data may be found in SupporGng InformaGon Data S1. 158 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 7 Environmental parameters 159 Temperature, salinity, and turbidity were measured by YSI 6600 mulG parameter quality 160 sonde (Yellow Springs Instruments, Ohio, USA). ParGculate Organic Carbon (POC) was measured 161 by Costech ECS 4010 Analyzer as described before (Peierls et al. 2003). Dissolved Organic Carbon 162 (DOC) was measured by Shimadzu TOC -5000A Analyzer as previously described (Crosswell et al. 163 2012). Extracted Chl a was determined fluorometrically with a Turner Trilogy Fluorometer (Peierls 164 et al. 2012) . Nitrate/nitrite (NO 3-+NO2-), ammonium (NH 4+), orthophosphate (PO 43-), total 165 dissolved nitrogen (TDN) and silica (SiO2) were measured with a Lachat QuickChem 8000 166 Automated Ion Analyzer (Paerl et al. 2010; Peierls and Paerl 2010). DetecGon limits were 0.05 167 µM, 0.50 µM, 0.21 µM and 1.17 µM for NO3-+NO2-, NH4+, PO43- and SiO2, respecGvely. Primary 168 producGvity was assessed by light/dark 14C bicarbonate incorporaGon (Paerl 2006; Gaulke et al. 169 2010). These measurements are provided in SupporGng InformaGon Data S1. 170 Bacterial and phytoplankton abundance 171 Whole water samples were kept in the dark on ice overnight and then fixed for flow 172 cytometry counGng the following day with glutaraldehyde (0.25% final) for 15 min in the dark at 173 room temperature and stored at -80°C (Paerl et al. 2020). Prior to counGng, samples were thawed 174 unGl they reached room temperature. Bacterioplankton and phytoplankton abundances were 175 determined using a Guava EasyCyte HT (Millipore) flow cytometer equipped with red and blue 176 excitaGon lasers. Phytoplankton were counted based on fluorescence (Paerl et al. 2020) with 177 addiGonal gaGng based on forward sca,er to enumerate single cells rather than aggregates 178 (SupporGng InformaGon Fig. S2). Bacterioplankton were counted using SYBR Green I staining and 179 without any heaGng step (Brussaard et al. 2010) . Abundances are provided in SupporGng 180 InformaGon Data S1. 181 Metabolite sample collection and analysis 182 Sampling bo,les, filtraGon units, and collecGon bo,les (amber, HDPE) were cleaned with 183 0.1 M HCl, methanol, and MilliQ water. Sampling bo,les were rinsed with sample water, filled 184 with water through a 90 µm Nitex mesh, and stored at near in-situ temperature in the dark. Water 185 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 8 was filtered within 10 h in a dark room through 3.0 µm polycarbonate (Isopore, Millipore) and 186 0.20 µm nylon filters (Millipore). The biomass from the nano - and microplankton fracGon that 187 passed through the 90 µm mesh and was retained on the 3.0 µm filters was for simplicity defined 188 as “microplankton ”, and the biomass fracGon that was sequenGally retained on the 0.2 µm filter 189 was defined as “picoplankton”. Three bo,les of ~250 mL filtrate per staGon were prepared and 190 stored at -20°C. 191 Metabolite extracGons were conducted in a dark room with a red headlamp as light 192 source. Dissolved metabolites were captured using C18 Solid Phase ExtracGon (SPE; waters, 10 g) 193 columns and triplicate extracGons were performed for each sample. Filtrates were thawed at 4°C, 194 pH adjusted to 6.5 and spiked with 13C-thiamin (thiamine-4-methyl-13C-thiazol-5-yl-13C3, Sigma -195 Aldrich, 75 pM final) . ParGculate metabolites were extracted from 3.0 and 0.20 µm pore -size 196 filters (Heal et al. 2014), and prior to extracGon, vials with sample filters were spiked with 10 pmol 197 13C-B1 (4,5,4 -methyl-13C3, 97%; Cambridge Isotope Laboratories), 1 pmol heavy B2 ( 13C4-15N2, 198 97%; Cambridge Isotope Laboratories), and 2 pmol cyano-cobalamin (Fisher BioReagents). Mean 199 percent recoveries of 13C-B1 are provided in SupporGng InformaGon Table S1. Eluted metabolites 200 were analyzed using a Dionex UlGmate-3000 LC system coupled to a TSQ QuanGva triple -stage 201 quadrupole mass spectrometer (ThermoFisher) operated in selected reacGon monitoring mode. 202 Matrix groups included a high salinity grouping ( NRE180, NRE160, NRE120, NRE100 ) and a low 203 salinity grouping (NRE70, NRE50, NRE00), as matrix differences were expected. For each matrix 204 group limits of detecGon (LOD) and limits of quanGficaGon (LOQ) were determined by calculaGng 205 (x3) and (x10) the standard deviaGon of the QC pool run between samples , respecGvely 206 (SupporGng InformaGon Data S 2). For further details see (Paerl et al. 2023a) or addiGonal 207

in the SupporGng InformaGon . 208 Data analysis was adapted from (Heal et al. 2014; Bannon et al. 2025) . For parGculate 209 samples, the peaks of B1 and B2 were normalized to the stable isotope internal standard to 210 reduce variability from the instrument and sample preparaGon. Metabolite measurements were 211 excluded if only one of the two analyGcal injecGons was above the LOD. The mean of the two 212 analyGcal injecGons was calculated before applying percent recoveries. ParGculate B1 was 213 corrected for percent recovery of the stable isotope internal standard as variable recovery was 214 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 9 observed (SupporGng InformaGon Table S1). Dissolved B1 was corrected for percent recovery of 215 the 13C-B1 spike in each sample. Dissolved HMP , HET, FAMP and cHET were corrected for percent 216 recovery previously determined from samples from the estuary (Paerl et al. 2023a), for matrix 217 groups high and low salinity recovery values from staGon NRE180 and staGon NRE0 were applied, 218 respecGvely (SupporGng InformaGon Table S1). The mean and standard deviaGon were calculated 219 from biological replicate samples that passed LOD and LOQ, or batch by batch criteria if between 220 LOD and LOQ (SupporGng InformaGon Data S3). A few samples were lost during sample 221 processing or excluded during analysis see SupporGng InformaGon Methods. 222 Metabolite data was analyzed and visualized with heatmaps (R package ComplexHeatmap 223 v2.20.0), LODs and LOQs were included with their calculated picomolar concentraGon. Metabolite 224 concentraGons (mean of biological replicates) are shown relaGve to the concentraGon range of 225 each compound (rows) . Each sampl e (columns) is made up of the different relaGve vitamin 226 concentraGons, referred to here as a vitamin profile, like a B-vitamin “fingerprint“ from that 227 sampl e. Columns and rows were clustered based on Euclidean distances corresponding to 228 differences between samples and relaGve metabolite concentraGons, respecGvely. To analyze the 229 phase parGGoning of vitamins, only measurements above LOD/LOQ were included and visualized. 230 DNA extraction and sequencing 231 Sampling bo,les were rinsed with sample water, filled with water through a 90 µm Nitex 232 mesh and stored at near in-situ temperature in the dark. Within six hours of sampling , 1 L of water 233 was sequenGally filtered onto a 3.0 µm pore-size membrane (MCE, Millipore) and a 0.22 µm pore-234 size Sterivex filter (PES, Millipore), which were stored at -20°C. DNA was extracted with the 235 DNeasy Blood and Tissue kit (Qiagen) with addiGonal lysozyme and proteinase K steps (Bi,ner et 236 al. 2024) and quanGfied (Qubit 3.0, Invitrogen). 237 ParGal 16S and 18S rRNA genes were PCR amplified from both size fracGons (0.22-3.0 µm, 238 3-90 µm) with the KAPA HiFi HotStart ReadyMix (Roche) and primer pairs 515F-Y/926R (Parada et 239 al. 2016) and a modified version of 565F/964R primers to avoid mismatches with haptophytes 240 (Lin et al. 2017), previously used for NRE plankton community analysis (Gong et al. 2020). Primer 241 sequences and PCR condiGons are provided in SupporGng InformaGon Table S2. Triplicate PCR 242 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 10 reacGons were pooled, amplicons were indexed and sequenced with MiSeq 2 x 300 bp v3 243 (Illumina) at the Rush Genomics and Microbiome Core Facility, Rush University, Chicago, IL, US A 244 (Naqib Ankurand Poggi 2018). 245 Sequence and data analysis 246 Reads were trimmed with cutadapt (v3.4) (MarGn 2011), quality filtered (16S: F 27 7, R 247 240; 18S: F 280, R 250), dereplicated, denoised, read pairs were merged (minOverlap = 100) and 248 chimeras removed in DADA2 (v1.22.0) (Callahan et al. 2016). For both 16S and 18S rRNA gene 249 fragments, Amplicon Sequence Variants (ASVs) were generated. Taxonomy was assigned to 16S 250 ASVs with ‘assignTaxonomy’ from DADA2 with the SBDI -curated version (v 7) (Lundin and 251 Andersson 2021) of 16S sequences of GTDB (r09-rs220) (Parks et al. 2018; Pascoal et al. 2024). 252 Taxonomy to 18S ASVs was assigned with ‘IdTaxa’ from DECIPHER (Murali et al. 2018) with the 253 PR2 (v 5.0.0) database (Guillou et al. 2012) . RelaGve abundances of dinoflagellates may be 254 overesGmated due to their high 18S rRNA copy number relaGve to diatoms (Gong and Marche‚ 255 2019). 256 Both 16S and 18S ASV tables were rarefied (n = 100) to the lowest sequencing depth (16S: 257 15,031 reads; 18S: 1,490 reads) with the ‘rrarefy’ funcGon from the vegan package (v2.6.8) 258 (Oksanen et al. 2022) , to account for varying sequencing depth in accordance with recent 259 recommendaGons (Schloss 2024). One sample from a 3.0 µm filter from staGon NRE180 from 25 260 Oct was removed due to insufficient sequencing depth prior to rarefacGon. Further processing 261 was carried out in phyloseq (v1.46.0) (McMurdie and Holmes 2013). Sequences without a domain 262 annotaGon and singletons were removed. Empty taxonomic levels were filled by the nearest 263 classified taxonomic level with ‘tax_fix’ from microViz (v0.12.10) (Barne, et al. 2021) . 264 ASV abundance data was Hellinger transformed, environmental and metabolite data was 265 z-score transformed prior to transformaGon -based Redundancy Analysis (tb -RDA) and R2 -266 adjustment with vegan. Explanatory variables for a constrained ordinaGon were selected by 267 forward selecGon by permutaGon (nperm = 999) of residuals under reduced model by 268 ‘forward.sel’ implemented in adespaGal (Dray et al. 2012). Significance of the constrained tb-RDA, 269 the axes and the explanatory variables were tested with ‘anova.caa’ from the vegan package. 270 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 11 Principal Coordinate Analyses (PCoA) were performed with vegan on Bray -CurGs distance 271 matrixes of Hellinger transformed abundance data . CorrelaGon analyses were conducted with 272 rstaGx (v0.7. 2, A. Kassambara: h,ps://github.com/kassambara/rstaGx ) and visualized with 273 corrplot (v0.92) (Wei and Simko 2021), the lower LOD (SupporGng InformaGon Data S1) was used 274 for measurements below LOD/LOQ. Linear models were fi,ed with the ‘lm’ funcGon implemented 275 in R. Data was analyzed and visualized in the R environment (v4. 4.1) with Gdyverse (v2.0.0) 276 (Wickham et al. 2019). 277

278 Hydrological and biochemical condi8ons in the estuary 279 NRE surface water salinity ranged from 0 to 17 – peaking at staGon NRE180 furthest down 280 the estuary (Fig. 1B). During high river discharge (SupporGng InformaGon Fig. S 1A) a notable 281 freshwater signature occurred further downstream, e.g. periods of salinity ~4 at NRE70 (Fig. 1B). 282 Discharge (Neuse River flow) was elevated between the samplings of 11 and 25 October 283 (SupporGng InformaGon Fig. S1A). Surface water temperature peaked at 29°C in July and declined 284 to 21°C during October (Fig. 1C), showing a seasonal shie from summer to fall. NOx 285 concentraGons were below detecGon limit (< 0.05 µM) , except at NRE0 where concentraGons 286 reached up to 57 µM on 13 September. Ammoni um was mostly below detecGon limit (< 0.50 µM) 287 but was elevated on 11 October (2.3-11.1 µM). Phosphate concentraGons ranged from 20) at NRE0 and decreased towards the middle of the estuary (NRE70; Fig. 1 290 A), with the execpGon of 25 October. Between 11 and 25 October, inorganic nutrients became 291 depleted at all staGons aeer a period of higher discharge, change was especially pronounced at 292 NRE0 (SupporGng InfromaGon Data S1) . Silica concentraGons followed the salinity gradient, with 293 higher concentraGons upstream (SupporGng InformaGon Fig. S 3B). Dissolved organic nitrogen 294 (DON) averaged 292 µg N L-1 with periodically higher concentraGons at NRE0 (e.g. 26 Jul 420 µg 295 N L-1). Dissolved organic carbon (DOC) ranged from 399 to 754 µM, both minimum and maximum 296 were observed at NRE0 (Fig. 1E). As is typical for the NRE (Gaulke et al. 2010), max imum Chl a 297 occurred mid -estuary (NRE50, 70) reaching 20-40 µg L -1 (SupporGng InformaGon Fig. S 3D). A 298 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 12 notable excepGon occurred on October 25th, where 90 µg Chl a L-1 occurred in NRE0 surface 299 water coinciding with peaks in turbidity (13.1 NTU), ParGculate Organic Carbon (POC, 3.9 mg C L-300 1, Fig. 1D) and primary producGvity (277 mg C m -3 h-1; SupporGng InformaGon Fig. S3). 301 Quantification of dissolved B-vitamins and vitamers 302 Twelve B-vitamin s and vitamers were quanGfied from 22 dissolved phase sampl ing events 303 (60 samples; full list found in SupporGng InformaGon Data S3 A, Fig. S4). The biologically acGve 304 forms of B12 (Ado -B12, Me-B12) and psB12 (Me-psB12) were not detected, as expected due to 305 in-situ photodegradaGon into OH-B12 and OH-psB12, respecGvely (Bannon et al. 2025). Dissolved 306 OH-B12 ranged from 0.9 pM up to 3.0 pM (Fig. 2A), and OH -psB12 concentraGons were notably 307 lower - approximately half (0.4 -1.4 pM) that of OH -B12 and occasionally below our detecGon 308 limit (0.3 pM). ConcentraGons of DMB, the alpha ligand of B12, were above the LOD of 2.8 pM 309 but not quanGfiable in most samples. Temporally and spaGally, concentraGons of dissolved B1 310 ranged from 19 to 74 pM , while concentraGons of B1 vitamers (HMP , AmMP , FAMP , cHET, HET) 311 were lower and ranged from (below limit of detecGon - cHET, HMP , AmMP; SupporGng 312 InformaGon Data S2A) to 82 pM for FAMP – a B1 degradaGon product. 313 Dissolved B2 concentraGons (12-96 pM) were comparable to B1 but with a notable 314 increase across the estuary on Oct 25 following high discharge. Similary to B2, B6 was detected 315 in all samples and ranged from 7-36 pM (mean 13.7 ± 6 pM). B5 was notably low (< LOD) at higher 316 salinity staGons (NRE100, 120, 180) , with a unique peak of 159 ± 21 pM at NRE180 on 13 317 September. With respect to variability , B1 pyrimidine vitamers (HMP , AmMP , FAMP) showed 318 moderate variaGon (2 to 4-fold change) while B5 (159 ± 21 pM peak; 17 ± 14 pM mean; 13 Sep 319 NRE180) and cHET (628 ± 153 pM peak; 58 ± 13 pM mean, 27 Sep NRE50) concentraGons were 320 stable, except occasional high peaks (10-fold higher; Fig. 2B). ConcentraGon pa,erns of dissolved 321 B1 and OH-B12 clustered together (clustering of rows), but relaGve concentraGons of dissolved 322 OH-psB12 concentraGons showed a unique pa,ern compared to the other vitamins with higher 323 concentraGons (∼4-fold) in the lower estuary (NRE100, 120, 180; Fig. 2). ConcentraGon pa,erns 324 of dissolved B2, B6 and B3 clustered together and were characterized by a maximum at NRE100 325 on 25 Oct. This signature of increased dissolved B2 was also detected upstream at NRE70 on 25 326 Oct. 327 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 13 328 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 14 Fig. 2. Dissolved B-vitamins and vitamers. Bar plot showing concentraGons of dissolved OH -B12, 329 OH-psB12 and B12 lower ligand DMB (A). Upper and lower dashed lines indicate the average LOQ 330 and LOD for DMB, respecGvely. The average LOD and LOQ for OH-B12 was 0.9 pM and 3.4 pM, 331 respecGvely. The average LOD and LOQ for OH -psB12 was 0.3 pM and 1.3 pM, respecGvely . 332 Samples below LOD are indicated; error bars show ± standard deviaGon of biological replicate 333 water samples. Heat map of dissolved B -vitamins and vitamer profiles ( B). Rows represent 334 concentraGon pa,erns across samples for each compound and columns represent B -vitamin 335 profiles for each staGon and sampling date (illustrated by the gray boxes). The mean m etabolite 336 concentraGons across water replicates are shown relaGve to the concentraGon range of each 337 compound; clustering is based on Euclidean distances. For measurements below LOD or LOQ, the 338 concentraGon limits are displayed (SupporGng InformaGon Data S2A). 339 340 Freshwater dissolved vitamin samples (NRE0) clustered together and showed overall 341 lower relaGve concentraGons compared to downstream staGons (Fig. 2B). InteresGngly, higher-342 salinity samples from NRE180 clustered with the freshwater staGon cluster and most NRE120 343 samples, whereas brackish mid -estuary staGons (NRE50, NRE70) formed a separate cluster with 344 some NRE100 and NRE120 samples – jointly poinGng to the middle estuary as site of unique 345 dissolved vitamin/vitamers composiGon . 346 Quan8fica8on of p ar8culate B-vitamins and vitamers 347 Sixteen B-vitamin and vitamers were detected from 23 parGculate sampling events in two 348 size fracGons (0.2-3 µm - 90 samples , 3-90 µm - 58 samples ; Fig. 3; full list found in SupporGng 349 InformaGon Data S3B, C). Mean parGculate B1 concentraGon (0.2-3 µm: 25 ± 18 pM, 3-90 µm: 21 350 ± 16 pM) and its range (LOD to ∼70 pM) were similar for both size fracGons. HET in the 0.2-3 µm 351 size fracGon was only quanGfiable at five staGons (NRE0, 50, 70, 100, 120) on 25 Oct, following 352 higher discharge, with a maximum of 3.6 ± 0.6 pM at NRE120. Peak HET concentraGon occurred 353 in the 3 -90 µm size fracGon on 25 Oct at NRE120 and NRE0 reaching ∼12.5 pM, otherwise 354 concentraGons were below ~1-2 pM . Traces of cHET were in both size fracGons but not 355 quanGfiable due to low signal to noise raGos and low concentraGons . FAMP concentraGon was 356 similar in both parGculate size fracGons (0.2 -3 µm: 2.8 -18.1 pM, 3 -90 µm: 4.3 -20.2 pM) . In 357 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 15 contrast, HMP was only quanGfiable in select samples from the 3-90 µm size fracGon from the 13 358 and 27 Sep with concentraGons of up to 14 ± 7 pM at NRE120 . ContrasGng with dissolved 359 concentraGons, AmMP reached upwards of 65 ± 10 pM (3-90 µm ) at NRE120 and NRE180 on 25 360 Oct with notable variaGon between sampl es (non-quanGfiable in other samples ; SupporGng 361 InformaGon Data S3). 362 Mean parGculate B2 varied two-fold between size fracGons (0.2-3 µm: 4.3 ± 2.9 pM, 3-90 363 µm: 9.0 ± 8.2 pM) with lowest concentraGons (< 2 pM) across the estuary occurring 11 Oct. B3 in 364 the 0.2-3 µm size fracGon was highly dynamic with variaGons of more than 20 -fold ranging from 365 14 pM to a of maximum concentraGon of 306 ± 42 pM at NR70 on 27 Sep. In contrast, B3 in the 366 3-90 µm size fracGon were less dynamic with a mean of 20 ± 17 pM, excluding a peak 367 concentraGon of 125 pM at NRE0 on 25 Oct. Generally, parGculate B5 in both size fracGons was 368 lowest at freshest staGon NRE0 (< 2 pM), except on 25 Oct where a maximum (50 ± 19 pM in the 369 3-90 µm size fracGon) was found at NRE0. 370 Three forms of B12 (Me -B12: methylcobalamin, Ado -B12: adenosylcobalamin, OH -B12: 371 hydroxycobalamin) and two forms of psB12 ( OH-psB12: hydroxy-pseudocobalamin, Me-psB12: 372 methyl-pseudocobalamin) could be resolved in the parGculate size -fracGon, but bioavailable 373 forms (Ado -B12, Me-B12, Me-psB12) were oeen below their respecGve LODs, especially in the 3-374 90 µm size fracGon. Mean OH-B12 concentraGon was 1.1 ± 0.5 pM and 1.3 ± 0.8 pM in 0.2-3 µm 375 and 3-90 µm size fracGon s, respecGvely. The maximum parGculate concentraGon for a ny B12-376 compound was Ado-B12 at NRE70 on 27 Sep with 7.6 ± 2.8 pM in the 0.2 -3 µm size fracGon, 377 highlighGng a biological acGve form of B12 can be twice the concentraGon of OH-B12. 378 ConcentraGons of DMB (mean 2.2 ± 1.4 pM) were patchier than B12 forms in the 0.2-3 µm size 379 fracGon and between the LOD and 1.7 pM in 3 -90 µm size fracGon (Fig. 3A, SupporGng 380 InformaGon Fig. S5) . Notably, on 25 Oct 2021 3.4 ± 0.6 pM Me -psB12 was detected in the 381 microplankton. 382 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 16 383 384 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 17 Fig. 3. ParGculate B-vitamins and vitamers. Bar plot showing concentraGons of parGculate 385 (0.2-90 µm) total -B12, total-psB12 and B12 lower ligand DMB ( A). Samples where total-psB12 386 was below the limit of detecGon of 0.1 pM is indicated; error bars show ± standard deviaGon of 387 biological replicate water samples. Heat map of size -fracGonated parGculate B-vitamins and 388 vitamer profiles (B). Rows represent concentraGon pa,erns across samples for each compound 389 and columns represent B -vitamin profiles for each staGon and sampling date. The mean 390 m etabolite concentraGons across biomass replicates are shown relaGve to the concentraGon 391 range of each compound; clustering is based on Euclidean distances. For measurements below 392 LOD or LOQ, the concentraGon limits are displayed (SupporGng InformaGon Data S2B, C). 393 394 Clustering based on Euclidean distances was performed to idenGfy similariGes and 395 differences across parGculate metabolites, staGons and sampling dates (Fig. 3B). FAMP , OH-B12, 396 and B1 exhibited similar variability in concentraGon pa,erns of parGculate vitamin concentraGons 397 (relaGve to the concentraGon range of the respecGve compound) across sampling dates, staGons, 398 and the two size fracGons (Fig. 3B; clustering of rows in heatmap). These three compounds (FAMP , 399 OH-B12, B1) were characterized by occasionally elevated concentraGons, contrasGng to other 400 compounds (e.g. B2), which had sporadic someGmes 10 -fold higher concentraGons. ParGculate 401 vitamin profiles from the two size fracGons did not exhibit clustering by staGon, sampling Gme, 402 or size fracGon (Fig. 3B). Only in four instances did vitamin profiles from the two size fracGons at 403 the same staGon and Gme cluster closely together (e.g., NRE0 on 27 Sep and 11 Oct, NRE120 on 404 11 Oct, NRE70 on 25 Oct). Overall, parGculate vitamin profiles from a given size fracGon were 405 more similar to profiles from the same size fracGon at other staGons than to the alternate size 406 fracGon at the same staGon (e.g. 0.2-3 µm profiles from NRE100 and NRE120 on 11 Oct). 407 When clustering the parGculate vitamin profiles per size -fracGon separately, the three 408 freshwater (NRE0) vitamin profiles cluster together in the 0.2-3 µm size fracGon as they were 409 characterized by overall lower concentraGons contrasGng to samples from the middle estuary 410 (NRE70, 100) (SupporGng InformaGon Fig. S6). Highest parGculate vitamin concentraGons in the 411 0.2-3 µm size fracGon were measured at NRE70 in September. The parGculate vitamin profile of 412 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 18 the 3-90 µm size fracGon at NRE0 from 25 Oct, was unique with peak concentraGons of B5, psB12 413 and HET grouping separately (Fig. 3). 414 B-vitamins and abio8c estuary condi8ons 415 A Kendall’s rank correlaGon analysis was performed for metabolites with each other and 416 a suite of associated measurements (Fig. 4A, SupporGng InformaGon Fig. S7). To invesGgate if 417 rivers could be a source of vitamins to coastal systems, B -vitamin and vitamer concentraGons 418 were analyzed for correlaGons with salinity, and results were vitamin-specific (Fig. 4). Dissolved 419 B2 and B6 were negaGvely correlated with salinity. Dissolved OH-psB12 was posiGvely correlated 420 with salinity and higher concentraGons aligned with higher abundances of picocyanobacteria, 421 producers of pseudocobalamin (Bannon et al. 2024b) (Fig. 4B, SupporGng InformaGon Fig. S8). In 422 the 0.2-3 µm parGculate size fracGon B5 and Me-psB12 were posiGvely correlated to salinity. B1 423 and B3 within the 3-90 µm parGculate size fracGon were negaGvely correlated with salinity. 424 Next, we examined if pa,erns in B-vitamins/vitamers were connected to pa,erns of 425 measured dissolved and parGculate nutrients (Fig. 4A). MulGple dissolved and parGculate 426 vitamins (e.g. B1) displayed negaGve correlaGons to dissolved inorganic nitrogen forms, whereas 427 mulGple parGculate B -vitamin concentraGons were posiGvely correlated to POC and DOC 428 concentraGons. ParGculate B-vitamins (B2, B3 and B12) of the 0.2 -3 µm size fracGon were 429 posiGvely correlated to DOC concentraGons. B2 was posiGvely correlated to parGculate nitrogen 430 and Chl a in the dissolved and both parGculate phases. ParGculate B-vitamins (B1, B2, B3, B5, OH-431 B12) from the 3-90 µm size-fracGon showed strong posiGve correlaGons to ParGculate Nitrogen 432 (PN) concentraGons. 433 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 19 434 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? −0.59 −0.46 −0.33 −0.2 −0.07 0.06 0.18 0.31 0.44 0.57 0.7SalinityTemperatureTurbiditydissolved O2POCPNDOCDICNH4NO3+NO2PO4SiO2Chlorophyll aPrimary productivityPC −SYN PE −SYN PEUKLEUKcell abundance B1 cHET HET HMP FAMP B2 B3 B5 B6 OH−B12 OH−psB12 B1 FAMP B2 B3 B5 OH−B12 Ado−B12 Me−B12 OH−psB12 Me−psB12 DMB B1 HET FAMP B2 B3 B5 OH−B12 OH−psB12 DMB particulate 3−90 µm particulate 0.2−3 µm dissolved < 0.2 µm A B 1−180 2−70 2−100 2−120 2−180 3−0 3−50 3−70 3−100 3−120 3−180 4−0 4−50 4−100 4−120 4−180 5−0 5−50 5−70 5−100 5−120 5−180 R² = 0.51 p−value = 0.00018 0.0 0.5 1.0 0 5 10 15 salinity OH−pseudoB12 (pM) PC−SYN+PE−SYN (cells/mL) 3e+04 1e+05 3e+05 1e+06 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 20 Fig. 4. CorrelaGon matrix of significant (p < 0.05) Kendall’s rank correlaGons between metabolites 435 and environmental, nutrient and biological measurements ( A). Linear relaGonship between 436 dissolved OH -psB12 and salinity and corresponding cell abundances of Synechococcus-like 437 phytoplankton (B). Numbers next to points indicate sampling and staGon number. ParGculate 438 organic carbon (POC), parGculate nitrogen (PN), dissolved organic carbon (DOC), dissolved 439 inorganic carbon (DIC) are abbreviated. Abundances of Synechococcus-like phycocyanin (PC)-rich 440 cells (PC -SYN), Synechococcus-like phycoerythrin (PE) -rich cells (PE -SYN), picoeukaryoGc 441 phytoplankton cells (PEUK), larger eukaryoGc phytoplankton cells (LEUK) and bacterial cell 442 abundance (bacterial abundance) were determined by flow cytometry. For abbreviaGon of 443 metabolites names see SupporGng Data S2. Metabolites below LOD or LOQ were replaced by the 444 lower LOD value prior to correlaGon analysis. Gray shading indicates the 95% confidence interval 445 around the linear regression. 446 Par88oning of B -vitamins across dissolved and par8culate phases 447 To invesGgate vitamin connecGvity between the dissolved and parGculate pool, the 448 concentraGons of vitamins from both parGculate size fracGons were summed. RaGos of 449 parGculate to dissolved B1, B3 and B5 varied over Gme and fell on both sides of the idenGty line, 450 a phase parGGoning of 1:1 (Fig. 5A). ParGculate total B12 concentraGons were higher than 451 dissolved OH-B12 concentraGons in most samples, whereas concentraGons of B2, FAMP and HET 452 were higher in the dissolved pool than the parGculate pool. When parGculate vitamin/vitamer 453 concentraGons were normalized to POC and dissolved concentraGons to DOC, pa,erns shieed 454 towards higher values in the parGculate size fracGon (SupporGng InformaGon Fig. S9). 455 When comparing parGculate B-vitamin and vitamer concentraGons between the two size 456 fracGons, B1, B2, B5 and FAMP showed temporal dynamics, but overall measurements fell around 457 the idenGty line, where the concentraGon of a compound would be equal in both size fracGons 458 (Fig. 5B). DisGnctly, total-B12, DMB and B3 concentraGons in the 0.2-3 µm size fracGon were 459 higher compared to the 3-90 µm size fracGon, except B3 on 25 Oct at NRE0. HET concentraGons 460 were oeen elevated in the 3-90 µm size fracGon, whereas OH-psB12 were typically higher in the 461 0.2-3 µm size fracGon. 462 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 21 463 1 10 100 0.1 1.0 10.0 100.0 pM (total particulate) pM (dissolved) sampling dates 26 Jul 13 Sep 27 Sep 11 Oct 25 Oct compound B1 HET HMP FAMP B2 B3 B5 OH−B12 vs. total B12 OH−psB12 vs. total psB12 A 0.1 1.0 10.0 100.0 0.1 1.0 10.0 100.0 pM (0.2−3.0 µm particulate) pM (3−90 µm particulate) sampling dates 26 Jul 13 Sep 27 Sep 11 Oct 25 Oct compound B1 HET FAMP B2 B3 B5 total B12 total psB12 DMB B .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 22 Fig. 5. Sca,er plots of phase parGGoning of B-vitamins and vitamers, dissolved (< 0.2 µm) 464 vs. parGculate (0.2-90 µM) ( A), parGculate (3-90 µm) vs. parGculate (0.2 -3.0 µm; B). Axes are 465 log10 scale and show picomolar concentraGons. Shape of points indicates sampling Gme point 466 and color corresponds to metabolite compound. Gray dashed line indicates a phase parGGoning 467 of 1:1. Points lee of the line indicate an enrichment in the phase of the y-axis while points right 468 of the line indicate an enrichment in the phase on the x -axis. Only metabolites with 469 measurements in both phases were included; measurements below LODs were excluded. 470 ParGculate total B12 is the sum of Ado -B12, Me-B12 and OH-B12. ParGculate total psB12 is the 471 sum of OH -psB12 and Me-psB12. 472 473 B1 was the only compound exhibiGng a spaGal pa,ern across the two parGculate pools 474 (SupporGng Data 3B, C ). ParGculate B1 concentraGons were spaGally disGnct with higher 475 concentraGons (31 ± 21 pM) in the lower estuary (NRE180, 120, 100) in the 0.2-3 µm size fracGon 476 compared to the 3-90 µm size fracGon (18 ± 17 pM); however, in the upper estuary (NRE0, 50, 477 70) B1 was higher in the 3 -90 µm fracGon (25 ± 13 pM). The most notable temporal feature 478 observed was a sharp increase in B2, B3, B5, and HET concentraGons in the 3–90 µm size fracGon 479 compared to the 0.2 –3 µm fracGon at NRE0 on 25 Oct Oct relaGve to the other Gme points 480 sampled. 481 As an indicator of potenGal connecGons between metabolite pools , we analyzed 482 correlaGons between quanGfied metabolites with Kendall’s rank correlaGon analysis (SupporGng 483 InformaGon Fig. S7). Dissolved B1 was posiGvely correlated with HET, while dissolved B3 posiGvely 484 correlated with dissolved B5. Most dissolved B-vitamins and vitamers were not correlated to their 485 parGculate concentraGon of either size fracGon. In the 0.2-3 µm parGculate size fracGon some 486 metabolites were posiGvely correlated with each other (e.g. B3 and OH-B12) and m etabolites of 487 the two parGculate size fracGons exhibited only few significant (p < 0.05) correlaGons with each 488 other. 489 Plankton abundances 490 Four small phytoplankton morphotypes were detected in NRE surface water based on 491 flow cytometry (SupporGng InformaGon Fig. S 2, S3E-I). Synechococcus-like phycoerythrin (PE)-492 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 23 rich (PE -SYN) cells ranged from 7.29 x 10 1 to 1.23 x 10 5 cells mL -1, while Synechococcus-like 493 phycocyanin (PC)-rich cells (PC-SYN) were more abundant (2.31 x 104 to 1.62 x 106 cells mL-1). PE-494 SYN was more abundant at higher salinity staGons and peaked at NRE180, whereas PC-SYN cells 495 were abundant throughout the estuary, except NRE0. PicoeukaryoGc phytoplankton (PEUK) 496 surged at NRE100 on 25 October, reaching 1.58 x 105 cells mL-1. Abundance of larger eukaryoGc 497 phytoplankton cells (LEUK, SupporGng InformaGon Fig. S2; example plot with gates) peaked at 498 NRE70 on 27 September with 2.66 x 105 cells mL-1. Bacterioplankton abundance was on average 499 1.52 x 10 7 cells mL -1 across staGons and samples , except NRE0, where cell abundances were 500 typically 10 x lower, with 4.24 x 106 cells mL-1. 501 Eukaryo8c plankton community composi8on 502 A total of 1,640 eukaryoGc ASVs were retained for analysis, with 19% unique to the 503 microplankton size fracGon and 32% unique to the picoplankton size fracGon. The major 504 eukaryoGc plankton divisions present were Stramenopiles, Alveolata, Cryptophyta and 505 Chlorophyta (Fig. 6A, SupporGng InformaGon Fig. S8). Stramenopiles were abundant in both size 506 fracGons (0.2-3 µm: 22 ± 15% relaGve abundance, 3-90 µm: 34 ± 19% relaGve abundance). The 507 diatom genus Cyclotella was highly abundant, especially in September, accounGng for up to 53% 508 and 70% of relaGve abundance in the pico - and microplankton, respecGvely (SupporGng 509 InformaGon Fig. S10), in line with peak cell abundances of the large eukaryote phytoplankton 510 morphotype measured by flow cytometry (up to 2.7 x 10 5 cells/mL; SupporGng InformaGon Fig. 511 S3E). The Alveolata subdivision Dinoflagellata showed higher relaGve abundances in the 512 microplankton size fracGon (0.2-3 µm: 11 ± 11%, 3-90 µm: 26 ± 18%) and contributed up to 63% 513 in relaGve abundance, whereas Ciliophora showed similar relaGve abundances in the two size 514 fracGons (0.2-3 µm: 10 ± 9%, 3-90 µm: 9 ± 10%) with peak relaGve abundances at NRE0 (mean 25 515 ± 11%). The Dinoflagellata genera Polykrikos, Levanderina and Gyrodinium were among the top 516 10 assigned genera (48% of taxa were not assigned a genus) based on relaGve abundance across 517 the dataset and occurred predominantly in the microplankton. Peak relaGve abundances for 518 Polykrikos, Levanderina and Gyrodinium were 42% (NRE180), 32% (NRE70) and 20% (NRE120) on 519 26 July, respecGvely (SupporGng InformaGon Fig. S9). 520 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 24 The dominant genus of Cryptophyta was Cryptomonas and occurred across staGons and 521 both size fracGons but showed highest relaGve abundances at staGons NRE50 and NRE70 on 27 522 September with 42 and 35% relaGve abundance, respecGvely (SupporGng InformaGon Fig. S11). 523 Chlorophyta were present in higher relaGve abundances in the picoplankton (33 ± 17%) than in 524 the microplankton (6 ± 3%; Fig. 6A). The genera Bathycoccus (e.g. 57% at NRE180 25 Oct) and 525 Micromonas showed higher relaGve abundances in the lower estuary compared to Ostreococcus, 526 which peaked in the upper estuary (e.g. 46% at NRE50 25 Oct, SupporGng InformaGon Fig. S11). 527 Spikes in Fungi occurred in single samples (picoplankton size fracGon), e.g. Opisthokonta (21% 528 relaGve abundance) at NRE100 on 11 October and Rhizophydiales (13% relaGve abundance) at 529 NRE0 on 25 October. 530 Prokaryo8c plankton community composi8on 531 In total, 4,451 prokaryoGc ASVs were detected, 47% were unique to the microplankton 532 size fracGon, and 15% were unique to the picoplankton size fracGon. Overall, a high number of 533 1,484 ASVs (33%) could not be classified beyond the order level. Within the prokaryoGc 534 picoplankton (0.2-3 µm) the bacterial phyla Pseudomonadota (mean 59 ± 13%), AcGnomycetota 535 (mean 14 ± 8%), Bacteroidota (average 13 ± 6%) and Cyanobacteriota (average 10 ± 8%) were 536 dominant based on relaGve abundance ( SupporGng InformaGon Fig. S 12). Pseudomonadota 537 (mean 17 ± 5%) and AcGnomycetota (mean 7 ± 2%) were less abundant in the microplankton (3-538 90 µm) size fracGon, instead Cyanobacteria (mean 29 ± 11%) and Bacteroidota (mean 23 ± 8%) 539 were higher in relaGve abundance. 540 Pelagibacterales were high in relaGve abundance in the picoplankton across the estuary 541 (mean 4 4 ± 13%), with lower relaGve abundances detected at NRE0 (mean 21 ± 11%, Fig. 6B, 542 SupporGng InformaGon Fig. S13). Four genera of Pelagibacterales dominated the picoplankton, 543 Pelagibacter, SYDM01 and IMCC9063 were abundant across all brackish staGons, whereas 544 Fonsibacter was predominantly present at NRE0. The dominant cyanobacterial genus was 545 Vulcanococcus (Synechococcus-like) with 15 ± 9% and 8 ± 7% of relaGve abundance in the 546 microplankton and picoplankton size fracGon, respecGvely. On October 25 at NRE0, the class 547 Cyanobacteriia contributed 31% of relaGve abundance to the microplankton and the potenGally 548 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 25 cyanotoxic cyanobacterial species Planktothrix agardhii accounted for 11.5 % in relaGve 549 abundance. 550 551 Fig. 6. EukaryoGc (A) and prokaryoGc (B) plankton community composiGon based on 18S and 16S 552 rRNA genes at staGons NRE0, NRE70 and NRE160/180. Taxonomy shown on division and order 553 level for 18 and 16S rRNA, respecGvely. The sample NRE180 from the microplankton size fracGon 554 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 26 was removed due to low sequencing depth for the 18S rRNA gene amplicons. Taxonomic profiles 555 for all sampled staGons are provided in SupporGng InformaGon Fig. S10 and S13. 556 557 Principal Coordinate Analyses (PCoA) of prokaryoGc and eukaryoGc plankton community 558 showed a clustering of NRE0 samples (SupporGng InformaGon Fig. S14). At staGon NRE0 33% and 559 18% of the ASVs of eukaryoGc and prokaryoGc plankton were unique, respecGvely. The 560 prokaryoGc plankton community at staGon NRE70 in July was more similar to the NRE0 561 communiGes sampled across summer and fall , in line with lower salinity and higher river 562 discharge (SupporGng InformaGon Fig. S2, S13, S14). Across all samples, community composiGon 563 clustered by size fracGon rather than sampling Gme. The prokaryoGc community at mid -estuary 564 staGon NRE50 was usually disGnct compared to up- and downriver staGons. 565 Connec8vity between B-vitamins and plankton communi8es 566 PotenGal connecGons between B-vitamin/vitamer pools and the bacterioplankton and 567 phytoplankton abundances were examined by correlaGon analysis (Fig. 4, SupporGng InformaGon 568 Fig. S1 5). Dissolved OH-psB12 strongly and posiGvely correlated with PC-SYN and PE -SYN 569 abundance, addiGonally parGculate OH -psB12 was posiGvely correlated to PC -SYN. Similarly, 570 dissolved B1 posiGvely correlated with PC-SYN abundance (Fig. 4, SupporGng InformaGon Fig. 571 S8B). Dissolved and parGculate OH-psB12, and parGculate B2 and B5 of the 0.2-3 µm size fracGon 572 posiGvely correlated with bacterioplankton abundances. The (pico-)cyanobacterial order PCC-573 6307 was posiGvely correlated with dissolved OH-psB12, parGculate B1 and total psB12 of the 574 picoplankton size fracGon (SupporGng InformaGon Fig. S 8, S 15). RelaGve abundances of of 575 picocyanobacteria (0.2-3 µm) were addiGonally posiGvely correlated to DMB. The Chlorophyta 576 genus Bathycoccus (0.2-3 µm) was posiGvely correlated to dissolved OH -psB12 and total 577 parGculate psB12. In the 3 -90 µm parGculate size fracGon the relaGve abundance of the 578 Dinoflagellate genus Levanderina was posiGvely correlated to total B12 . In both size fracGons 579 Polykrikos was posiGvely correlated to dissolved B1 and B3. 580 Redundancy analys is was used to examine if B-vitamin and vitamer concentraGons 581 significantly (p < 0.05) helped explain observed variability in plankton community composiGon . 582 Select B-vitamins and vitamer s together explained 30.6% of variaGon in eukaryoGc plankton 583 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 27 composiGon and 42% of the observed variance in prokaryoGc plankton (SupporGng InformaGon 584 Table S3). Specifically, parGculate B3 and FAMP, and dissolved B1 and OH -psB12 were highly 585 significant (p ≤ 0.005) explanatory variables for plankton communit y composiGon . AddiGonal 586 significant explanatory variables for the eukaryoGc plankton community were dissolved B6 and 587 parGculate OH-B12. The strongest explanatory variable for the prokaryoGc community was 588 dissolved B1 and parGculate B3 for eukaryoGc plankton. Overall, mulGple B-vitamins , B1- and B12-589 vitamers showed a staGsGcally significant effect on plankton composiGon . 590 Salinity was the main driving environmental variable for eukaryoGc and prokaryoGc 591 community composiGon, explaining 6.9 and 8.5%, respecGvely (SupporGng InformaGon Table S3). 592 Measured environmental variables (aside from B-vitamins/vitamers) explained 13.9% and 20.4% 593 of the observed variability in eukaryoGc and prokaryoGc plankton community composiGon, 594 respecGvely. In both eukaryoGc and prokaryoGc communiGes, B -vitamin/vitamer concentraGons 595 cumulaGvely explained a greater proporGon of variance, compared to hydrological and nutrient 596 measurements during our summer –fall sampling. 597

598 Concentra8on paZerns of B-vitamins and vitamers 599 Here, we uncover elevated concentraGons of B-vitamins mid -estuary (Chl a max region) 600 along with unique dissolved B-vitamin profiles associated with higher phytoplankton bioma ss. 601 Key bacterio- and phytoplankton groups observed and their potenGal role as B-vitamin producers 602 or consumers are disscued below and summarized in a conceptual figure in relaGon to the 603 dissolved B-vitamins observed across the freshwater to polyhaline gradient of the NRE in summer 604 and fall 2021 (Fig. 7). Our study idenGfie s B1, B3 , and B12 compounds as key compounds 605 influencing prokaryoGc and eukaryoGc plankton communiGes. The simultaneous quanGficaGon 606 of the two dissolved B12 forms, OH-B12 and OH-psB12, show that on average OH-psB12 (mean 607 0.9 ± 0.3 pM) was at about 47% of the concentraGon of OH -B12 (mean 1.9 ± 0.6 pM ; Fig. 2A ), 608 supporGng the hypothesis that psB12 could be a crucial source for B12-remodellers, especially in 609 B12 limited systems. Dissolved OH-psB12 emerged as a key driver of both prokaryoGc and 610 eukaryoGc community composiGon in the NRE, strengthening its importance as an indirect B12 611 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 28 source even for organisms thought to be unable to remodel the compound. EsGmates suggest up 612 to 17% of bacteria can salvage B12 (Shelton et al. 2019), and the geneGc capability to remodel 613 psB12 or use other cobalamin intermediates might be of high importance for planktonic 614 microorganisms in the NRE and beyond. While OH -B12 is expected to be the dominant form of 615 B12 in the sun -lit aquaGc environments (Bannon et al. 2024a) , OH -psB12 synthesized by 616 cyanobacteria, could be an addiGonal important source of B12 to organisms capable of 617 remodeling psB12 (Helliwell et al. 2016) . Since the 1950s/1960s evidence has accumulated 618 highlighGng B12 as an essenGal metabolite for aquaGc microorganisms , however fewer than 40% 619 of prokaryotes are predicted to synthesize B12 de novo (Shelton et al. 2019) and only now we are 620 beginning to elucidate microbial sources, cycling, and transformaGons of the disGnct B12 forms 621 (Soto et al. 2023; Bannon et al. 2024a; Wienhausen et al. 2024). 622 623 Fig. 7 Conceptual map of key bacterio- and phytoplankton groups observed and their potenGal 624 role as B-vitamin producers or consumers across the freshwater to polyhaline gradient of the NRE 625 in summer and fall 2021. The color gradient of the estuary indicates the relaGve total dissolved 626 Low High B-vitamin concentrations Burkholderiales B1 B12 Picocyanobacteria (PC- SYN, Vulcanococcus sp.) B1 psB12 Picocyanobacteria (PE- SYN, Synechococcus sp.) B1 psB12 Cyclotella (Diatoms) B1 B12 Larger phytoplankton: Planktothrix agardhii B1 psB12 Ostreococcus B1-vitamers B1 Levanderina (Dinoflagellate) B1 B12 Cryptophya B1 B12 Pelagibacterales B1-vitamers B1 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 29 concentraGons of measured B-vitamins and vitamers. Yellow or blue arrows in the taxa bubbles 627 indicate potenGal producGon (parGculate & dissolved) or consumpGon (dissolved) of B -628 vitamins/vitamers, respecGvely, based on findings from culture studies and/or the correlaGon 629 analyses presented and discussed here. 630 631 Tracking of B-vitamins/vitamer dynamics across the parGculate and dissolved phase of an 632 estuary shows that some of the compounds produced within an estuary are enriched in the 633 dissolved phase compared to the parGculate phase (FAMP , HET, OH -psB12; Fig. 5A). We 634 hypothesize that these degradaGon compounds (FAMP , HET) and OH-psB12 are not readily used 635 by most bacterio- and phytoplankton groups, due to a lack of sufficiently sensiGve transporter or 636 vitamin salvage pathways, and therefore accumulate. These compounds require further salvaging 637 and remodeling acGvity to fulfil the cellular vitamin requirement and could provide an advantage 638 for microorganisms with these metabolic pathways (Paerl et al. 2023a). For further discussion on 639 B1, B3 and more details on the parGGoning of B-vitamins and vitamers across the pico- and nano-640 /microplankton phases and the dissolved phase see the SupporGng InformaGon. 641 Puta8ve eukaryo8c microbial sources and transforma8ons of B-vitamins 642 Our data highlights a Gght connecGon between dominant phytoplankton and specific 643 vitamin profiles in the parGculate pool, while we are sGll working to understand the vitamin 644 availability resulGng from algal blooms (and more broadly the exometabolomes of blooms over 645 Gme and space). The phytoplankton growth in the middle of the estuary includes the biosynthesis 646 of a phytoplankton -specific suite of vitamin/vitamers, and these micronutrients in turn can 647 support growth of auxotrophic phytoplankton, including potenGal HABs, and bacterioplankton. 648 The dominant eukaryoGc phytoplankton groups idenGfied (with 18S rRNA gene 649 sequencing) and their dynamics are typical for late summer, early fall in the NRE, consisGng of 650 diatoms, dinoflagellates, cryptomonas, and chlorophyta (Gong and Marche‚ 2019; Gong et al. 651 2020). The highly abundant Cyclotella (diatom) could have a B12 requirement and is likely 652 prototrophic for B1, based on findings of experiments with one isolate (Tang et al. 2010; Bertrand 653 and Allen 2012), however B12 auxotrophy has been shown to be strain specific. ContrasGngly, 654 Cryptophytes can have a requirement for both exogenous B1 and B12 (Tang et al. 2010), making 655 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 30 them likely consumers of B1, B12, and their respecGve vitamers. Thus, surges of Cryptophyta 656 (Cryptophytes) in NRE could be governed by the availability of B1 and B12. Similarly, the 657 abundance of Chlorophyta taxa Ostreococcus, Micromonas, Bathycoccus could be regulated by 658 the availability of B1 vitamers, as they are B1 auxotrophs but can salvage B1 from vitamers (Paerl 659 et al. 2015, 2023b) . Thereby, these taxa simultaneously consume vitamers and potenGally 660 funcGon as a source for ‘regenerated’ vitamin in the upper estuary (NRE50, 70; 25 Oct). 661 Levanderina (Dinoflagellate) recurrently surged during summer/fall ( SupporGng 662 InformaGon Fig. S9) and, based on elevated microplankton parGculate B1 concentraGons during 663 higher abundances of Levanderina (e.g. 28.1 pM B1 and 21.8% relaGve abundance of Levanderina 664 fissa NRE70 25 Oct), these dinoflagellates are potenGal B1 producers and sources of B1 to higher 665 trophic levels. Previously, B1 biosynthesis transcripts ( thiC, thiE) were abundant during a 666 dinoflagellate bloom (Levanderina fissa) in the NRE (Gong et al. 2017) , further supporGng that 667 some bloom -forming Dinoflagellates in the NRE are significant B1 producers. 668 Insight into B-vitamin cycling along an estuary con8nuum 669 Our measurements have revealed several new perspecGves on vitamin cycling within 670 (temperate, long residence Gme) estuaries: (1) freshwater input indirectly fuels increased levels 671 of vitamins mid -estuary by promoGng algal growth and (2) B-vitamins (dissolved and parGculate) 672 are likely uGlized quickly in the lower estuary and transferred to higher tropic levels in the benthos 673 or downstream (Fig. 7), similar to macronutrients. During extended periods of high discharge, the 674 mid estuary peak of planktonic biomass (Chl a, bacterial abundance, POC) could shie further 675 downstream or not occur and accordingly affect vitamin supply to the larger Pamlico Sound (Paerl 676 et al. 2010; Hall et al. 2013) . Surprisingly, B-vitamin/vitamer concentraGons were not overtly 677 elevated for all compounds in the freshest region of the NRE as hog and poultry operaGons are 678 extensive within the Neuse River watershed (Lebo et al. 2012) and both are established sources 679 of nutrients and presumably B-vitamins (Lune,a et al. 2022) . 680 While we find evidence that estuaries can be a source of B -vitamins /vitamers to 681 downstream coastal waters (Fig. 2, 4), the extent of supply is likely dependent on the freshwater 682 discharge and vitamin -specific (Fig. 7), as some B -vitamins or vitamers appear to accumulate in 683 the dissolved phase (e.g. psB12, FAMP, Fig. 5A). These accumulated compounds could funcGon as 684 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 31 more ‘recalcitrant’ vitamin/vitamers for the phytoplankton and bacterioplankton downstream. 685 Overall, the oligohaline water of the NRE was lower in B-vitamin s and characterized by disGnct 686 plankton and B -vitamin/vitamer profiles compared to the brackish part of the estuary (see 687 SupporGng InformaGon ). Prior work found inverse correlaGons between B-vitamin concentraGon 688 and salinity suggesGng rivers and groundwater as sources of B-vitamins to coastal systems (Gobler 689 et al. 2007). Data from coastal regions influenced by the Amazon (Brazil; dissolved B1 and B6) and 690 Moulouya (Morocco; dissolved B1, B2, B6, B12) rivers found that estuaries were not a major 691 source to the coastal ocean (Barada et al. 2013; Tovar-Sánchez et al. 2016). While we find negaGve 692 correlaGons of dissolved B2 and B6 with salinity, posiGve correlaGons are evident with dissolved 693 compounds (HMP , B5 and OH-psB12), indicaGng that a general pa,ern for dissolved B -vitamins 694 with salinity might not exist in this system . Importantly, salinity is a driving factor for bacterio- 695 and phytoplankton community structure in the NRE, as shown here and in previous studies (Paerl 696 et al. 2020; Sánchez -Gallego et al. 2025) , thereby salinity indirectly affects B -vitamin 697 concentraGons. Overall, our data indicate that the NRE delivers B -vitamins and vitamers into 698 Pamlico Sound (massive component of the APES) but the extent is expected to depend on flushing 699 (antecedent precipitaGon) and changes in microbial community composiGon. Moreover, growth 700 and proliferaGon of HAB spp. within the Pamlico Sound may be impacted by B-vitamin/vitamer 701 delivery via the NRE – making the process impacmul to food web, water quality, and ecosystem 702 health. 703 Temporal dynamics of B-vitamins and vitamers 704 B vitamin concentraGons across the estuary were highly dynamic - from sub picomolar to 705 high picomolar levels - revealing strong short-term variability driven by synthesis, uptake, and 706 degradaGon processes, as well as sporadic surges in pico- and microplankton populaGons. Two 707 modes of B -vitamin dynamics were observed: 1) moderate concentraGon dynamics (2 to 4 -fold 708 change) within a compound specific range (e.g. dissolved HMP; parGculate FAMP) including 709 occasionally elevated concentraGons and 2) strong changes in concentraGons (10-fold or more) 710 characterized by occasionally high peaks in concentraGon (e.g. dissolved B5 and cHET; parGculate 711 B2; Fig. 3, 4). We argue that the dynamic pa,erns observed with mode 1 reflect oscillaGons 712 between states of balance/imbalance for the planktonic communiGes due to vitamin/vitamer 713 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 32 producGon and consumpGon, resulGng in slight to moderate fluctuaGons. We hypothesize that 714 m ode 2 is driven by the strong rise and fall (or elevated acGvity) of specific planktonic populaGons 715 leading to disGnct vitamin profiles (e.g. October 25th NRE0, Planktothrix agardhii, peak parGculate 716 psB12). 717 During a bloom, parGculate B -vitamin and vitamer concentraGons can increase sharply, 718 potenGally disproporGonally to changes in the dissolved pool as observed for (microplankton) 719 parGculate B-vitamin and vitamer concentraGons at NRE0 on 25 Oct. These sporadic episodes of 720 rapid biomass accumulaGon likely reflect phases, during which intracellular vitamin synthesis (or 721 uptake and salvage) dominates over release. As phytoplankton become nutrient -limited or 722 physiologically stressed, however, extracellular release (and leakage) of vitamins can rise, a 723 pa,ern documented for DOM in both cultures and environmental samples. While the transfer 724 into the dissolved phase might be taxa specific (Sultana et al. 2023) , the observed vitamin 725 concentraGon pa,erns suggest overall minimal transfer of vitamin/vitamer to the communal 726 dissolved pool – outside of declines and death. 727 B-vitamins/vitamers shape microbial community composi8on 728 Here, we find evidence of B -vitamins - especially dissolved B1 and OH -psB12 and 729 parGculate B3 and FAMP - as key explanatory variables for both prokaryoGc and eukaryoGc 730 plankton estuarine communiGes ( SupporGng InformaGon Table S3). This is congruent with 731 widespread auxotrophy for B1 and B12 and the important role of B3 in cellular metabolism. 732 Together, this builds a greater perspecGve that these compounds are key ‘shapers’ of plankton 733 communiGes. The concentraGons of B-vitamins and vitamers collecGvely explained 42% and 31% 734 of the observed variance in the prokaryoGc and eukaryoGc planktonic community respecGvely - 735 highlighGng the overall importance of B-vitamins in shaping microbial community composiGon in 736 estuarine waters, complementary to previous observaGons from bo,le/mesocosm incubaGons 737 that have demonstrated vitamin -driven changes in plankton communiGes in certain habitats 738 (reviewed in: Bertrand & Allen 2012; Joglar et al. 2020). 739 In the NRE, Picocyanobacteria are likely de novo synthesizers of B1 and psB12 and reach 740 higher abundances in brackish waters away from the freshwater endmember of the NRE (e.g. 741 NRE0; Paerl et al. 2020) . The detecGon of peak parGculate Me-psB12 (3.4 ± 0.6 pM) in the 742 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 33 microplankton on 25 October coincides with a high relaGve abundance of the filamentous 743 cyanobacteria Planktothrix agardhii in the microplankton size fracGon. Congruently, we found a 744 link between dissolved B1, OH-psB12 concentraGons in the picoplankton , and dissolved OH-745 psB12 with the abundance of PC -SYN (Fig. 4 , SupporGng InformaGon Fig. S8, S15) highlighGng 746 them as sources. As dissolved OH -psB12 is not readily available to most plankton a Gghter 747 coupling between picocyanobacterial abundance and psB12 is observed compared to dissolved 748 B1, which is likely rapidly uGlized. Recent field data (Roskilde Fjord Denmark; Northwest AtlanGc) 749 also point to picocyanobacteria as important B1 and psB12 sources (Bannon et al. 2024b; Bi,ner 750 et al. 2024). These lines of evidence support that picocyanobacterial abundance could funcGon 751 as a proxy for psB12 concentraGons, and that psB12 provided by (pico -) cyanobacteria could 752 significantly promote taxa capable of cobalamin remodeling (Helliwell et al. 2016; Soto et al. 753 2023). Salvage of B12 from degraded cobalamin and DMB may represent an important yet 754 overlooked pathway in aquaGc systems, as cyanobacteria are already considered key planktonic 755 community members - especially through C and N -fixaGon - their role may extend further as 756 producers of key vitamins such as B1 and psB12. 757 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 34

758 This work was supported by Independent Research Fund Denmark (9040-00067B to LR, RWP , and 759 Anders F. Andersson). MJB received funding from the European Union’s Horizon 2020 research 760 and innovaGon programme under the Marie Skłodowska-Curie grant agreement No 801199. RWP 761 acknowledges support from NSF OCE NSF OCE award s Oand 2416286. EMB acknowledges 762 support from NSERC Discovery Grant RGPIN -2015-05009 and Simons FoundaGon Grants 504183 763 and 1001702. We are grateful to Jeremy Braddy and Amy Bartenfelder for assistance with water 764 collecGons. We thank the whole UNC -IMS MODMON team for analyzing and providing 765 hydrological, chemical, and biological data. We thank Malcolm Barnard for help with filtraGons. 766 We thank UNC -IMS for providing lab space and local logisGcal support. We thank Anders F. 767 Andersson for input on the study. 768 769 CONFLICTS OF INTEREST 770 No conflicts of interest. 771 772 DATA AVAILABILITY STATEMENT 773 The data that support the findings of this study are openly available in the SupporGng InformaGon 774 Data S1, S2 and S3 deposited at h,ps://doi.org/10.11583/DTU.31353040 . Raw sequence reads 775 from 16S and 18S rRNA genes are deposited at NCBI under accession PRJNA1175993. 776 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 35

777 Balk, L. and others. 2009. Wild birds of declining European species are dying from a thiamine 778 deficiency syndrome. Proceedings of the NaGonal Academy of Sciences 106: 12001–12006. 779 doi:10.1073/pnas.0902903106 780 Bannon, C. and others. 2025. Seasonal pa,erns in B-vitamins and cobalamin co-limitaGon in the 781 Northwest AtlanGc. Limnol. Oceanogr. 1–16. doi:10.1002/lno.70204 782 Bannon, C. C., E. M. Mudge, and E. M. Bertrand. 2024a. Shedding light on cobalamin 783 photodegradaGon in the ocean. Limnol. Oceanogr. Le,. 9: 135–144. 784 doi:10.1002/lol2.10371 785 Bannon, C. C., M. A. Soto, E. Rowland, N. Chen, A. Gleason, E. Devred, J. LaRoche, and E. M. 786 Bertrand. 2024b. ProducGon and uGlizaGon of pseudocobalamin in marine Synechococcus 787 cultures and communiGes. Environ. Microbiol. 26: 1–14. doi:10.1111/1462-2920.16701 788 Barada, L. P ., L. Cu,er, J. P . Montoya, E. A. Webb, D. G. Capone, and S. A. Sañudo-Wilhelmy. 789 2013. The distribuGon of thiamin and pyridoxine in the western tropical North AtlanGc 790 Amazon River plume. Front. Microbiol. 4: 1–10. doi:10.3389/fmicb.2013.00025 791 Barne,, D. J. m., I. C. w. Arts, and J. Penders. 2021. microViz: an R package for microbiome data 792 visualizaGon and staGsGcs. J. Open Source Soew. 6: 3201. doi:10.21105/joss.03201 793 Bertrand, E. M., and A. E. Allen. 2012. Influence of vitamin B auxotrophy on nitrogen 794 metabolism in eukaryoGc phytoplankton. Front. Microbiol. 3. 795 doi:10.3389/fmicb.2012.00375 796 Bianchi, T. S. 2007. Biogeochemistry of estuaries, Oxford University Press on Demand. 797 Bi,ner, M. J., C. C. Bannon, E. Rowland, J. Sundh, E. M. Bertrand, A. F. Andersson, R. W. Paerl, 798 and L. Riemann. 2024. New chemical and microbial perspecGves on vitamin B1 and vitamer 799 dynamics of a coastal system. ISME CommunicaGons 4. doi:10.1093/ismeco/ycad016 800 Bruns, S., G. Wienhausen, B. Scholz-Bö,cher, S. Heyen, and H. Wilkes. 2023. Method 801 development and quanGficaGon of all B vitamins and selected biosyntheGc precursors in 802 winter and spring samples from the North Sea and de novo synthesized by Vibrio 803 campbellii. Mar. Chem. 256: 104300. doi:10.1016/j.marchem.2023.104300 804 Bruns, S., G. Wienhausen, B. Scholz-Bö,cher, and H. Wilkes. 2022. Simultaneous quanGficaGon 805 of all B vitamins and selected biosyntheGc precursors in seawater and bacteria by means of 806 different mass spectrometric approaches. Anal. Bioanal. Chem. 414: 7839–7854. 807 doi:10.1007/s00216-022-04317-8 808 Brussaard, C. P . D., J. P . Payet, C. Winter, and M. G. Weinbauer. 2010. QuanGficaGon of aquaGc 809 viruses by flow cytometry. Manual of aquaGc viral ecology 11: 102–107. 810 Callahan, B. J., P . J. McMurdie, M. J. Rosen, A. W. Han, A. J. A. Johnson, and S. P . Holmes. 2016. 811 DADA2: High-resoluGon sample inference from Illumina amplicon data. Nat. Methods 13: 812 581–583. doi:10.1038/nmeth.3869 813 Carini, P . J. 2013. Genome-enabled invesGgaGon of the minimal growth requirements 814 andphosphate metabolism for Pelagibacter marine bacteria. 815 Crosswell, J. R., M. S. Wetz, B. Hales, and H. W. Paerl. 2012. Air-water CO 2 fluxes in the 816 microGdal Neuse River Estuary, North Carolina. J. Geophys. Res. Oceans 117: 1–12. 817 doi:10.1029/2012JC007925 818 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 36 Dray, S. and others. 2012. Community ecology in the age of mulGvariate mulGscale spaGal 819 analysis. Ecol. Monogr. 82: 257–275. doi:h,ps://doi.org/10.1890/11 -1183.1 820 Froelich, B., R. Gonzalez, D. Blackwood, K. Lauer, and R. Noble. 2019. Decadal monitoring 821 reveals an increase in Vibrio spp. concentraGons in the Neuse River Estuary, North Carolina, 822 USA I. Karunasagar [ed.]. PLoS One 14: e0215254. doi:10.1371/journal.pone.0215254 823 Gaulke, A. K., M. S. Wetz, and H. W. Paerl. 2010. Picophytoplankton: A major contributor to 824 planktonic biomass and primary producGon in a eutrophic, river-dominated estuary. Estuar. 825 Coast. Shelf Sci. 90: 45–54. doi:10.1016/j.ecss.2010.08.006 826 Gobler, C., C. Norman, C. Panzeca, G. Taylor, and S. Sañudo-Wilhelmy. 2007. Effect of B-vitamins 827 (B1, B12) and inorganic nutrients on algal bloom dynamics in a coastal ecosystem. AquaGc 828 Microbial Ecology 49: 181–194. doi:10.3354/ame01132 829 Gómez-Consarnau, L. and others. 2018. Mosaic pa,erns of B-vitamin synthesis and uGlizaGon in 830 a natural marine microbial community. Environ. Microbiol. 20: 2809–2823. 831 doi:10.1111/1462-2920.14133 832 Gong, W., J. Browne, N. Hall, D. Schruth, H. Paerl, and A. Marche‚. 2017. Molecular insights 833 into a dinoflagellate bloom. ISME Journal 11: 439–452. doi:10.1038/ismej.2016.129 834 Gong, W., N. Hall, H. Paerl, and A. Marche‚. 2020. Phytoplankton composiGon in a eutrophic 835 estuary: Comparison of mulGple taxonomic approaches and influence of environmental 836 factors. Environ. Microbiol. 22: 4718–4731. doi:10.1111/1462-2920.15221 837 Gong, W., and A. Marche‚. 2019. EsGmaGon of 18S Gene Copy Number in Marine EukaryoGc 838 Plankton Using a Next-GeneraGon Sequencing Approach. Front. Mar. Sci. 6: 1–5. 839 doi:10.3389/fmars.2019.00219 840 Gong, W., H. Paerl, and A. Marche‚. 2018. EukaryoGc phytoplankton community 841 spaGotemporal dynamics as idenGfied through gene expression within a eutrophic estuary. 842 Environ. Microbiol. 20: 1095–1111. doi:10.1111/1462-2920.14049 843 Guillou, L. and others. 2012. The ProGst Ribosomal Reference database (PR2): a catalog of 844 unicellular eukaryote Small Sub-Unit rRNA sequences with curated taxonomy. Nucleic Acids 845 Res. 41: D597–D604. doi:10.1093/nar/gks1160 846 Hall, N. S., H. W. Paerl, B. L. Peierls, A. C. Whipple, and K. L. Rossignol. 2013. Effects of climaGc 847 variability on phytoplankton community structure and bloom development in the 848 eutrophic, microGdal, New River Estuary, North Carolina, USA. Estuar. Coast. Shelf Sci. 117: 849 70–82. doi:10.1016/j.ecss.2012.10.004 850 Heal, K. R. and others. 2017. Two disGnct pools of B 12 analogs reveal community 851 interdependencies in the ocean. Proceedings of the NaGonal Academy of Sciences 114: 852 364–369. doi:10.1073/pnas.1608462114 853 Heal, K. R., L. T. Carlson, A. H. Devol, E. V. Armbrust, J. W. Moffe,, D. A. Stahl, and A. E. Ingalls. 854 2014. DeterminaGon of four forms of vitamin B 12 and other B vitamins in seawater by 855 liquid chromatography/tandem mass spectrometry. Rapid CommunicaGons in Mass 856 Spectrometry 28: 2398–2404. doi:10.1002/rcm.7040 857 Helliwell, K. E. and others. 2016. Cyanobacteria and EukaryoGc Algae Use Different Chemical 858 Variants of Vitamin B12. Current Biology 26: 999–1008. doi:10.1016/j.cub.2016.02.041 859 Joglar, V., A. Prieto, E. Barber-Lluch, M. Hernández-Ruiz, E. Fernández, and E. Teira. 2020. SpaGal 860 and temporal variability in the response of phytoplankton and prokaryotes to B-vitamin 861 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 37 amendments in an upwelling system. Biogeosciences 17: 2807–2823. doi:10.5194/bg-17-862 2807-2020 863 Koch, F., T. K. Ha,enrath-Lehmann, J. A. Goleski, S. Sañudo-Wilhelmy, N. S. Fisher, and C. J. 864 Gobler. 2012. Vitamin B1 and B12 Uptake and Cycling by Plankton CommuniGes in Coastal 865 Ecosystems. Front. Microbiol. 3: 1–11. doi:10.3389/fmicb.2012.00363 866 Lebo, M. E., H. W. Paerl, and B. L. Peierls. 2012. EvaluaGon of Progress in Achieving TMDL 867 Mandated Nitrogen ReducGons in the Neuse River Basin, North Carolina. Environ. Manage. 868 49: 253–266. doi:10.1007/s00267-011-9774-5 869 Lin, Y ., N. Cassar, A. Marche‚, C. Moreno, H. Ducklow, and Z. Li. 2017. Specific eukaryoGc 870 plankton are good predictors of net community producGon in the Western AntarcGc 871 Peninsula. Sci. Rep. 7: 1–11. doi:10.1038/s41598-017-14109-1 872 Lukienko, P . I., N. G. Mel’nichenko, I. V. Zverinskii, and S. V. Zabrodskaya. 2000. AnGoxidant 873 properGes of thiamine. Bull. Exp. Biol. Med. 130: 874–876. doi:10.1007/BF02682257 874 Lundin, D., and A. Andersson. 2021. SBDI SaGva curated 16S GTDB database. Swedish 875 Biodiversity Infrastructure (SBDI). doi:10.17044/scilifelab.14869077.v7 876 Lune,a, R. S., R. G. Greene, and J. G. Lyon. 2022. Modeling the DistribuGon of Diffuse Nitrogen 877 Sources and Sinks in the Neuse River Basin, p. 119–149. In GeospaGal InformaGon 878 Handbook for Water Resources and Watershed Management, Volume II. CRC Press. 879 MarGn, M. 2011. Cutadapt removes adapter sequences from high-throughput sequencing 880 reads. EMBnet. J. 17: 10. doi:10.14806/ej.17.1.200 881 McMurdie, P . J., and S. Holmes. 2013. Phyloseq: An R Package for Reproducible InteracGve 882 Analysis and Graphics of Microbiome Census Data. PLoS One 8. 883 doi:10.1371/journal.pone.0061217 884 Miranda-Ríos, J., M. Navarro, and M. Soberón. 2001. A conserved RNA structure (thi box) is 885 involved in regulaGon of thiamin biosyntheGc gene expression in bacteria. Proceedings of 886 the NaGonal Academy of Sciences 98: 9736–9741. doi:10.1073/pnas.161168098 887 Möller, K., B. Krock, and F. Koch. 2022. Method opGmizaGon of the simultaneous detecGon of 888 B12 congeners leading to the detecGon of a novel isomer of hydroxycobalamin in seawater. 889 Rapid CommunicaGons in Mass Spectrometry 36. doi:10.1002/rcm.9401 890 Murali, A., A. Bhargava, and E. S. Wright. 2018. IDTAXA: a novel approach for accurate 891 taxonomic classificaGon of microbiome sequences. Microbiome 6: 140. 892 doi:10.1186/s40168-018-0521-5 893 Naqib Ankur and Poggi, S. and W. W. and H. M. and K. K. and G. S. J. 2018. Making and 894 Sequencing Heavily MulGplexed, High-Throughput 16S Ribosomal RNA Gene Amplicon 895 Libraries Using a Flexible, Two-Stage PCR Protocol, p. 149–169. In N. Raghavachari Nalini 896 and Garcia-Reyero [ed.], Gene Expression Analysis: Methods and Protocols. Springer New 897 York. 898 Oksanen, J. and others. 2022. vegan: Community Ecology Package. 899 Paerl, H. W. 2006. Assessing and managing nutrient-enhanced eutrophicaGon in estuarine and 900 coastal waters: InteracGve effects of human and climaGc perturbaGons. Ecol. Eng. 26: 40–901 54. doi:10.1016/j.ecoleng.2005.09.006 902 Paerl, H. W., K. L. Rossignol, S. N. Hall, B. L. Peierls, and M. S. Wetz. 2010. Phytoplankton 903 Community Indicators of Short - and Long-term Ecological Change in the Anthropogenically 904 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 38 and ClimaGcally Impacted Neuse River Estuary, North Carolina, USA. Estuaries and Coasts 905 33: 485–497. doi:10.1007/s12237-009-9137-0 906 Paerl, R. W., E. M. Bertrand, A. E. Allen, B. Palenik, and F. Azam. 2015. Vitamin B1 ecophysiology 907 of marine picoeukaryoGc algae: Strain-specific differences and a new role for bacteria in 908 vitamin cycling. Limnol. Oceanogr. 60: 215–228. doi:10.1002/lno.10009 909 Paerl, R. W., N. P . CurGs, M. J. Bi,ner, M. R. Cohn, S. M. Gifford, C. C. Bannon, E. Rowland, and E. 910 M. Bertrand. 2023a. Use and detecGon of a vitamin B1 degradaGon product yields new 911 views of the marine B1 cycle and plankton metabolite exchange S.J. Giovannoni [ed.]. mBio 912 14: e00061-23. doi:10.1128/mbio.00061 -23 913 Paerl, R. W., N. P . CurGs, M. J. Bi,ner, M. R. Cohn, S. M. Gifford, C. C. Bannon, E. Rowland, and E. 914 M. Bertrand. 2023b. Use and detecGon of a vitamin B1 degradaGon product yields new 915 views of the marine B1 cycle and plankton metabolite exchange S.J. Giovannoni [ed.]. 916 mBio. doi:10.1128/mbio.00061 -23 917 Paerl, R. W., J. Sundh, D. Tan, S. L. Svenningsen, S. Hylander, J. Pinhassi, A. F. Andersson, and L. 918 Riemann. 2018. Prevalent reliance of bacterioplankton on exogenous vitamin B1 and 919 precursor availability. Proceedings of the NaGonal Academy of Sciences 115: E10447–920 E10456. doi:10.1073/pnas.1806425115 921 Paerl, R. W., R. E. Venezia, J. J. Sanchez, and H. W. Paerl. 2020. Picophytoplankton dynamics in a 922 large temperate estuary and impacts of extreme storm events. Sci. Rep. 10: 1–15. 923 doi:10.1038/s41598-020-79157-6 924 Parada, A. E., D. M. Needham, and J. A. Fuhrman. 2016. Every base ma,ers: assessing small 925 subunit rRNA primers for marine microbiomes with mock communiGes, Gme series and 926 global field samples. Environ. Microbiol. 18: 1403–1414. doi:10.1111/1462-2920.13023 927 Parks, D. H., M. Chuvochina, D. W. Waite, C. Rinke, A. Skarshewski, P . A. Chaumeil, and P . 928 Hugenholtz. 2018. A standardized bacterial taxonomy based on genome phylogeny 929 substanGally revises the tree of life. Nat. Biotechnol. 36: 996. doi:10.1038/nbt.4229 930 Pascoal, F., P . Duarte, P . Assmy, R. Costa, and C. Magalhães. 2024. Full-length 16S rRNA gene 931 sequencing combined with adequate database selecGon improves the descripGon of ArcGc 932 marine prokaryoGc communiGes. Ann. Microbiol. 74: 29. doi:10.1186/s13213-024-01767-6 933 Peierls, B. L., R. R. ChrisGan, and H. W. Paerl. 2003. Water quality and phytoplankton as 934 indicators of hurricane impacts on a large estuarine ecosystem. Estuaries 26: 1329–1343. 935 doi:10.1007/BF02803635 936 Peierls, B. L., N. S. Hall, and H. W. Paerl. 2012. Non-monotonic responses of phytoplankton 937 biomass accumulaGon to hydrologic variability: A comparison of two coastal plain north 938 carolina estuaries. Estuaries and Coasts 35: 1376–1392. doi:10.1007/s12237-012-9547-2 939 Peierls, B. L., and H. W. Paerl. 2010. Temperature, organic ma,er, and the control of 940 bacterioplankton in the Neuse River and Pamlico Sound estuarine system. AquaGc 941 Microbial Ecology 60: 139–149. doi:10.3354/ame1415 942 Pinckney, J. L., H. W. Paerl, M. B. Harrington, and K. E. Howe. 1998. Annual cycles of 943 phytoplankton community -structure and bloom dynamics in the Neuse River Estuary, 944 North Carolina. Mar. Biol. 131: 371–381. doi:10.1007/s002270050330 945 Sánchez-Gallego, J., N. P . CurGs, H. W. Paerl, and R. W. Paerl. 2025. New perspecGves on 946 picocyanobacteria and understudied cyanobacterial diversity in the Albemarle Pamlico 947 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 39 sound system, North Carolina, USA. Front. Microbiol. 16: 1–16. 948 doi:10.3389/fmicb.2025.1539050 949 Sañudo-Wilhelmy, S. A. and others. 2012. MulGple B-vitamin depleGon in large areas of the 950 coastal ocean. Proceedings of the NaGonal Academy of Sciences 109: 14041–14045. 951 doi:10.1073/pnas.1208755109 952 Sañudo-Wilhelmy, S. A., C. J. Gobler, M. Okbamichael, and G. T. Taylor. 2006. RegulaGon of 953 phytoplankton dynamics by vitamin B 12. Geophys. Res. Le,. 33: L04604. 954 doi:10.1029/2005GL025046 955 Sañudo-Wilhelmy, S. A., L. Gómez -Consarnau, C. Suffridge, and E. A. Webb. 2014. The Role of B 956 Vitamins in Marine Biogeochemistry. Ann. Rev. Mar. Sci. 6: 339–367. doi:10.1146/annurev-957 marine-120710-100912 958 Schloss, P . D. 2024. RarefacGon is currently the best approach to control for uneven sequencing 959 effort in amplicon sequence analyses K. McMahon [ed.]. mSphere 9: 960 10.1128/msphere.00354-23. doi:10.1128/msphere.00354-23 961 Shelton, A. N., E. C. Seth, K. C. Mok, A. W. Han, S. N. Jackson, D. R. Hae, and M. E. Taga. 2019. 962 Uneven distribuGon of cobamide biosynthesis and dependence in bacteria predicted by 963 comparaGve genomics. ISME J. 13: 789–804. doi:10.1038/s41396-018-0304-9 964 Soto, M. A., D. Desai, C. Bannon, J. LaRoche, and E. M. Bertrand. 2023. Cobalamin producers and 965 prokaryoGc consumers in the Northwest AtlanGc. Environ. Microbiol. 25: 1300–1313. 966 doi:10.1111/1462-2920.16363 967 Suffridge, C., L. Cu,er, and S. A. SA. Sañudo-Wilhelmy. 2017. A New AnalyGcal Method for Direct 968 Measurement of ParGculate and Dissolved B-vitamins and Their Congeners in Seawater. 969 Front. Mar. Sci. 4: 1–11. doi:10.3389/fmars.2017.00011 970 Suffridge, C. P ., L. Gómez-Consarnau, D. R. Monteverde, L. Cu,er, J. Arístegui, X. A. Alvarez-971 Salgado, J. M. Gasol, and S. A. Sañudo-Wilhelmy. 2018. B Vitamins and Their Congeners as 972 PotenGal Drivers of Microbial Community ComposiGon in an Oligotrophic Marine 973 Ecosystem. J. Geophys. Res. Biogeosci. 123: 2890–2907. doi:10.1029/2018JG004554 974 Sultana, S., S. Bruns, H. Wilkes, M. Simon, and G. Wienhausen. 2023. Vitamin B12 is not shared 975 by all marine prototrophic bacteria with their environment. ISME J. 17: 836–845. 976 doi:10.1038/s41396-023-01391-3 977 Tang, Y . Z., F. Koch, and C. J. Gobler. 2010. Most harmful algal bloom species are vitamin B 1 and 978 B 12 auxotrophs. Proceedings of the NaGonal Academy of Sciences 107: 20756–20761. 979 doi:10.1073/pnas.1009566107 980 Tovar-Sánchez, A. and others. 2016. Nutrients, trace metals and B-vitamin composiGon of the 981 Moulouya River: A major North African river discharging into the Mediterranean Sea. 982 Estuar. Coast. Shelf Sci. 176: 47–57. doi:10.1016/j.ecss.2016.04.006 983 Twomey, L. J., M. F. Piehler, and H. W. Paerl. 2005. Phytoplankton uptake of ammonium, nitrate 984 and urea in the Neuse River Estuary, NC, USA. Hydrobiologia 533: 123–134. 985 doi:10.1007/s10750-004-2403-z 986 Wang, H., C. Zhang, F. Chen, and J. Kan. 2020. SpaGal and temporal variaGons of 987 bacterioplankton in the Chesapeake Bay: A re-examinaGon with high-throughput 988 sequencing analysis. Limnol. Oceanogr. 1–14. doi:10.1002/lno.11572 989 Wei, T., and V. Simko. 2021. R package “corrplot”: VisualizaGon of a CorrelaGon Matrix. 990 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint 40 Wickham, H. and others. 2019. Welcome to the Tidyverse. J. Open Source Soew. 4: 1686. 991 doi:10.21105/joss.01686 992 Wienhausen, G. and others. 2024. Ligand cross-feeding resolves bacterial vitamin B12 993 auxotrophies. Nature 629: 886–892. doi:10.1038/s41586-024-07396-y 994 Wurtsbaugh, W. A., H. W. Paerl, and W. K. Dodds. 2019. Nutrients, eutrophicaGon and harmful 995 algal blooms along the freshwater to marine conGnuum. WIREs Water 6: 1–27. 996 doi:10.1002/wat2.1373 997 998 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 27, 2026. ; https://doi.org/10.64898/2026.02.26.707256doi: bioRxiv preprint