Microbiomes of pathogenic Vibrio species reveal environmental and planktonic associations

Background Many species of coastal Vibrio spp. bacteria can infect humans, representing an emerging health threat linked to increasing seawater temperatures. Vibrio interactions with the planktonic community impact coastal ecology and human infection potential. In particular, interactions with eukaryotic and photosynthetic organism may provide attachment substrate and critical nutrients (e.g. chitin, phytoplankton exudates) that facilitate the persistence, diversification, and spread of pathogenic Vibrio spp. Vibrio interactions with these organisms in an environmental context are, however, poorly understood. Results We quantified pathogenic Vibrio species, including V. cholerae , V. parahaemolyticus , and V. vulnificus , and two virulence-associated genes for one year at five coastal sites in Southern California and used metabarcoding to profile associated prokaryotic and eukaryotic communities, including vibrio-specific communities. These Vibrio spp. reached high abundances, particularly during Summer months, and inhabited distinct species-specific environmental niches driven by temperature and salinity. Associated bacterial and eukaryotic taxa identified at fine-scale taxonomic resolution revealed genus and species-level relationships. For example, common Thalassiosira genera diatoms capable of exuding chitin were positively associated with V. cholerae and V. vulnificus in a species-specific manner, while the most abundant eukaryotic genus, the diatom Chaetoceros , was positively associated with V. parahaemolyticus. Associations were often linked to shared environmental preferences, and several copepod genera were linked to low-salinity environmental conditions and abundant V. cholerae and V. vulnificus . Conclusions This study clarifies ecological relationships between pathogenic Vibrio spp. and the planktonic community, elucidating new functionally relevant associations, establishing a workflow for examining environmental pathogen microbiomes, and relevant Vibrio-plankton interactions.

3 highlighting prospective model systems for future mechanistic studies.

Background
Coastal bacterial Vibrio species can cause severe human infections, which are an emerging international health concern linked to rising global temperatures. V. cholerae, the causative agent of the disease cholera, infects millions of people each year, killing thousands, and is typically spread through ingesting contaminated drinking water [1]. Two other species of major concern are V. parahaemolyticus and V. vulnificus, which can cause severe wound infections, septicemia, and gastroenteritis from ingesting Vibrio-colonized seafood [2]. Several particularly dangerous pandemic strains have been identified, and non-virulent strains may become virulent via horizontal gene transfer as many infectionrelated genes are mobile [3]. Additionally, at least a dozen other species can infect humans or animals, extending the threat to aquaculture operations. Climate change may exacerbate the extent of these infections. Increasing air and water temperatures can facilitate increased metabolic growth capacity and temporal and geographic range expansion of Vibrio spp. pathogens [4][5][6]. Furthermore, V. cholerae epidemics have been linked to global temperature rise on decadal scales-a representative case study for understanding the link between the environment and human disease [7,8].
Vibrio spp. interactions with the planktonic community have implications for both coastal ecology and human health. Coastal communities are highly productive environments; diverse and abundant populations of microbes and multicellular organisms are supported by primary productivity driven by ample nutrient availability. Vibrio spp. attach to and form biofilms on particles and eukaryotic organisms, living and dead, [9][10][11][12], presumably to better acquire carbon and nutrients and avoid environmental stress. These "close quarters" incite competition and enable cooperation and horizontal gene transfer with co-4 occurring bacterial and eukaryotic species. An important facet of these interactions involves chitin, an abundant polymer produced by many marine eukaryotes. In addition to providing nutrients and an attachment substrate, chitin facilitates bacterial competition and horizontal gene transfer in Vibrio spp. [13,14], which may spread virulence and antibiotic resistance genes among populations. Attachment also enables environmental persistence and dispersal; for example, Vibrio spp. attach to copepod exoskeletons by the thousands with high copepod abundances linked to cholerae epidemics. Vibrio spp. can also attach to chitin-producing diatoms [15,16], though these dynamics and the ecological consequences are poorly understood.
Genus, species, and even strain-level distinctions likely modulate the functional characteristics of Vibrio-community interactions, but these are challenging to address because quantifying and characterizing environmental microbes is time and labor intensive and their identification often lacks clear criteria. Total quantities of Vibrio spp.
are frequently used to infer ecological associations and pathogenicity potential, however, pathogenic species often occupy distinct environmental niches driven by temperature, salinity, and other biotic and abiotic factors (reviewed in Takemura et al., 2014). Virulence mechanisms, while poorly understood, are also species dependent. Likewise, eukaryotes are often grouped into broad categories such as total copepods or in the case of algae, by bulk chlorophyll a concentration, total pigment concentration, or at a broad taxonomic level (e.g. diatoms, dinoflagellates). But physiological differences at lower taxonomical ranks may have functional consequences for interactions; for example, some diatom genera exude chitin while others may not [18,19], and algae are known to host distinct bacterial communities [20]. Thus, broad taxonomic groupings may be responsible for apparent conflicts and lack of consensus among prior studies.
Next generation sequencing based metabarcoding now enables large-scale community 5 analyses with improved taxonomic resolution that can be used to infer potential functional significance from observed associations and co-occurrence patterns. In our study, we sampled monthly for one year at 5 sites in Southern California (Fig. 1A-D), pairing environmental data and pathogenic Vibrio species and virulence-associated gene abundance obtained via digital droplet PCR with high resolution community composition data derived from three different DNA amplicons. We used 16S and 18S rRNA gene sequences derived from total RNA and cDNA to characterize the active communities of bacteria, archaea, and eukaryotes, including both phytoplankton and multicellular organisms. We further sequenced HSP60 gene sequences to characterize the vibriospecific community in both whole-community and vibrio isolate samples. Our findings establish a workflow for examining environmental pathogen microbiomes, offer insights into conflicting results from previous studies, elucidate new functionally relevant associations, and facilitate future environmentally realistic model laboratory studies.

Results
Environmental niche of pathogenic vibrio species and virulence-associated gene prevalence We quantified three pathogenic vibrio species with high human health relevance: Vibrio parahaemolyticus, V. vulnificus, and V. cholerae. Digital droplet PCR (ddPCR) was performed to absolutely quantify single copy number genes specific to each species in known filtration volumes using primers previously designed for qPCR (Additional File 1) [21 -25]. A wide range of temperatures (13.2-33 °C), and salinities (2.6-42.4 ppt) were sampled (Fig. 1E,F), with highly variable chlorophyll a concentrations (Fig. 1G). Both salinity and chlorophyll a were positively associated with temperature (Fig. 1H). Each species was detected at all sites, often simultaneously, and exhibited species-specific 6 environmental niches ( Fig. 2A-C). Target species were predominantly found above 20 °C.
V. parahaemolyticus was detected in 80% of samples (Additional File 2) and was most abundant at warm temperatures and high salinities (Fig. 1A), though only the association with temperature was significant (p < 0.05) (Fig. 2D). We observed V. parahaemolyticus at extremely high salinities (> 40 ppt, Fig. 1A), while V. vulnificus and V. cholerae were both significantly associated with low salinity but not temperature (Fig. 2D). V. vulnificus was most abundant at moderate and V. cholerae at low salinity sites (Fig. 2B,C). High numbers of V. cholerae (> 280,000 copies/ 100 mL) were detected at Los Peñasquitos Lagoon (LPL) from March through May, corresponding with low salinity caused by lagoon closure and subsequent urban freshwater accumulation (Fig. 1F, Fig. 2C). All three species peaked between March and July at the LPL and San Diego River (SDR) sites. At the Tijuana River Estuary (TJ) sites, V. parahaemolyticus was predominantly detected, occurring between February and October and peaking at > 33,000 copies/ 100 mL in September ( Fig. 2A).
We further screened for the presence of virulence-associated genes and quantified two genes associated with V. vulnificus: pilF and vcgC. We measured pilF and vcgC only in V.
vulnificus-positive samples (Figs. 2E,F). Both were most abundant at the SDR sites, with pilF reaching > 7,000 copies/ 100 mL (Figs. 2F). When V. vulnificus was detected, 50% of samples also tested positive for one or both of the virulence-associated genes (Fig. 2G).
The ratio of these targets to V. vulnificus copies (potentially reflective of number of virulence-associated gene copies per V. vulnificus cell) was often below 1, though sometimes closer to 2 as in the LPL April sample (Fig. 2H).

Vibrio community abundance and composition
We assessed relative abundance of Vibrio spp. in the context of the entire bacterial community using HSP60 DNA amplicon sequences, amplified using DNA as a template. We characterized Vibrio-specific community composition using sequences generated from 7 culture-dependent Vibrio communities isolated on CHROMagar Vibrio plates (CHROMagar) (Fig. 3A, Additional File 3) and also culture-independent whole community filtered samples ( Fig. 3B), which we then compared to Vibrio community composition determined via the 16S community data. Based on whole community sequencing, Vibrio spp. represented 0.03-4.9% of the 16S bacterial community (Mean = 0.44%, Additional Files 4 and 5) and 0-4.5% of the HSP60 community (Mean = 0.4%, Additional File 4). As in previous studies (e.g. Jesser and Noble, 2018), the composition of 16S Vibrio spp. sequences were poorly resolved at the species levels, with the majority identified only as "Vibrio sp." (Additional File 5B). Of the pathogenic species detected via ddPCR, only V. cholerae was detected by 16S, and both V. cholerae and V. parahaemolyticus were detected using HSP60. Nearly 4 times as many 16S sequences were obtained across whole-community-samples than the HSP60 sequences (Additional File 4), with a higher percentage of assigned reads, resulting in the apparent absence of Vibrio spp. in some HSP60 samples though they are known to be present at these sites based on 16S, ddPCR, and isolate community sequencing ( Fig. 3B). Additionally, in whole community samples, the HSP60 amplicon identified different but not substantially more Vibrio species (Fig. 3B).
Culture-based sequencing of the HSP60 amplicon from isolates revealed diverse Vibrio communities with higher taxonomic resolution than both whole 16S and HSP60 cultureindependent sequencing approaches. HSP60 isolate community sequencing, which produced substantially more reads overall and a higher percentage of assigned reads (Additional File 4), identified more than 100 unique ASVs and identified all three quantified species and additional potential Vibrio spp. pathogens (Fig. 3A). Alpha and Shannon diversity were elevated during the summer months sampled, May and July, which coincided with moderate temperatures (Fig. 3C-E). Assuming not all Vibrio spp. in the samples were cultured/culturable, this may be an underestimate of species diversity.

16S and 18S Community Composition and Diversity
For 16S rRNA gene sequences, which were sequenced using RNA as a template, ~ 29K exact sequence variants (ASVs) were identified after removing eukaryotic, mitochondrial, and chloroplast sequences. A few major bacterial classes dominated community composition, including Gammaproteobacteria (encompassing Vibrio spp.), Bacteroidia, and Alphaproteobacteria (Fig. 4A). LPL and SDR sites had sizeable populations of Oxyphotobacteria (i.e. cyanobacteria), while other prominent classes included Campylobacteria and Verrucomicrobia. In contrast, 18S communities comprising ~ 17K ASVs were more diverse, with many rare species. For example, > 50% of classes at some sites were < 5% of relative abundance (Fig. 4B). Diatoms were the most common eukaryotes, comprising ~ 28% of 18S reads (Fig. 4A), and while common at the LPL and SDR sites, they were particularly abundant at the TJ sites, frequently representing > 75% of 18S reads. Other abundant groups included unicellular Spirotrichea ciliates, photosynthetic Cryptophyceae, and chitin-producing Arthropoda organisms.
Both sampling site and month influenced community diversity, and samples collected at nearby sites or close in time were more similar to each other than to other communities. parahaemolyticus (e.g. members of the genera Glaciecola and Marinobacterium unidentified at the species level) while alphaproteobacterial ASVs including a member of the SAR11 group, a Salinihabitans sp., and a Litorimicrobium sp. were positively associated; the former two were also positively associated with temperature and chlorophyll a. Bacterial taxa from both of these classes were also significantly associated with the other Vibrio spp. and genes quantified, indicating ASV-level associations and revealing shared environmental preferences.
Eukaryotic taxa also exhibited ASV-level associations unapparent at higher taxonomic ranks. The most abundant class, Bacillariophyta (containing diatoms), was positively associated with V. parahaemolyticus and temperature, salinity, and chlorophyll a ( potentially overlooking species shifts in response to surrounding environmental community changes. By screening all three species, we capture some of these dynamics.
We also present the first quantification and ecological analysis of pathogenic Vibrio spp. in the Southern California coastal region, and area of emerging risk due to warm coastal seawater temperatures, high residential and tourism recreational water use, and seafood cultivation. Vibrio spp. infections in Southern California have increased in recent years [30], particularly in San Diego County; the most recent year assessed, 2018, showed the highest number of infections ever reported and an infection rate substantially higher than both the California and US infection rates [31].
All three species were detected at all sites, occasionally simultaneously. This suggests either a continuous presence at all times, sometimes below detectable concentrations, or a temporal residence in the sediments or a viable but non-culturable (VBNC) [32] state until conditions become ideal for proliferation in the water column [33]. Peak abundance among the three species varied noticeably with salinity ( Fig. 1A-C), with V. cholerae highest at the lowest salinity sites (~ 2-5 ppt), V. vulnificus highest at moderate salinities 13 (~ 17-25 ppt), and V. parahaemolyticus highest at high salinity sites (> 30 ppt). The distribution patterns of these three species were also linked to site, with V. cholerae most abundant at LPL, V. vulnificus most common at the SDR sites, and V. parahaemolyticus most abundant at the TJ sites. It is unclear whether those sites happened to present an ideal ecological niche at a given time, or if other factors such as biotic interactions limit concentrations of species that would otherwise be abundant.
V. cholerae and V. vulnificus were significantly associated with low salinity and were most abundant in warm temperatures though there was no significant temperature association across all samples (Fig. 1B,C). While V. cholerae has been reported in high salinity conditions, it is most common in low salinities, hence it's tendency to contaminate drinking water. Likewise, V. vulnificus grows poorly at salinities higher than 25 ppt, preferring the range of 10-18 ppt [34,35]. Both species peaked during warm summer months, typically a month or two before the peak temperature, and high abundances (i.e. > 1000 copies/ mL) were only found from March through July (Additional File ). As temperature was associated with high salinity, intermediate conditions where temperatures are warm, but salinity is low or moderate may be ideal. Both species were abundant only above 20 °C, a temperature above which human Vibrio infections are a serious concern [35][36][37].
V. parahaemolyticus abundance was significantly associated with high temperatures, but not salinity, suggesting V. parahaemolyticus may be a more halotolerant species. This is supported by a meta-analysis finding that in contrast to V. cholerae, V. parahaemolyticus was distributed across a broader salinity range of 3-35 ppt, with a warmer, more narrow temperature range [17]. The abundant V. parahaemolyticus populations we observed at extremely high salinities (> 40 ppt) ( Fig. 2A) were out of the reported range in the metaanalysis and for other prior studies we examined, perhaps suggesting unique high-salinity 14 adaptations. However, the fundamental ecological niche of many Vibrio species, particularly in terms of salinity, is often larger than realistic environmental conditions [38]. Comparative analyses examining specific physiological properties of these populations, other high-salinity V. parahaemolyticus strains if identified, and those with "typical" salinity ranges may shed light on whether the strains we observed were atypical.
As the high-salinity populations were found at moderate temperatures, salinity tolerance may allow V. parahaemolyticus to take advantage of fortuitous warm temperatures, though other site-specific factors are likely to be involved. In general, our study supports the well-established temperature and salinity preferences previously observed in these species in a different geographic region.
Some of our findings suggest local Vibrio populations may be of regional concern, perhaps necessitating a monitoring and surveillance plan. The potentially pathogenic V. vulnificus targets we examined appeared to contain a high percentage of virulence-associated genes as measured by ddPCR (Fig. 2G,H). Half of the samples that tested positive for V.
vulnificus also tested positive for one or both of the virulence-associated genes tested. For example, the vcgC gene, which is a marker more common in clinical V. vulnificus strains than environmental counterparts, was detected in 20% of these samples. Along the North Carolina Coast Williams et al. found that 5.3% of the V. vulnificus examined possessed the vcgC gene [39]. The pilF gene, which based on human serum sensitivity is highly correlated with pathogenicity potential [40], was also detected in 45% of these samples.
Additionally, we detected high concentrations of V. cholerae (> 2800 cells/ mL) at the LPL site. While it is unclear whether these strains possess virulence-associated genetic markers such as the ctxA toxin-associated gene, V. cholerae can infect immunocompromised humans without these virulence genes and Vibrio communities lacking these virulence genes can acquire them rapidly via viral infection [41] or other 15 horizontal gene transfer events.
Culture dependent and independent sequencing methods highlight Vibrio diversity and additional pathogenic species of interest We gained insight into Vibrio community diversity by sequencing the HSP60 gene in both whole community samples (culture independent) and Vibrio isolate communities (culture dependent) (Fig. 3). The HSP60 amplicon is better at phylogenetically resolving certain bacterial taxa, including Vibrio spp. [29,42,43], compared to the more commonly used 16S amplicon. Consistent with these prior findings, in our study, for both culture dependent and independent HSP60 sequencing methods, more Vibrio ASVs were identified at the species level than the 16S rRNA gene sequencing which identified the large majority of Vibrio spp. only at the genus level only (Fig. 3B, Additional File 5). In our HSP60 whole community analysis we did not observe as many species as Jesser and Noble 2018, which introduced this method for profiling Vibrio communities; this could be the result of differences in sequencing depth and/or regional differences in Vibrio spp. abundance as that study was conducted in the Neuse River Estuary of North Carolina.

However
Sequencing Vibrio isolate communities from 24 sites allowed us to identify nearly twice as many species as the 16S or HSP60 amplicons, including all 3 pathogenic species detected using ddPCR and additional human pathogenic species including V. alginolyticus, V. fluvialis, V. furnissi (also identified by HSP60), and V. metschnikovi [ 44]. With a large number of sequence reads obtained, the majority of which belonged to the Vibrio genus (Additional File 4) the diversity of these communities was better captured than with either the HSP60 or 16S rRNA gene amplicon -particularly regarding intraspecific diversity (Fig. 3A). This method does present some limitations; sequences are limited to cultivated organisms, and this method precludes an analysis of relative abundance within the 16 community as some species likely outgrow others. Nonetheless, our results reveal an impressive, and likely underestimated, picture of Vibrio diversity.
Combining these methods, we substantially expanded our understanding of what other Vibrio bacteria exist with abundant pathogenic species populations. For example, an interesting species, V. antiquarius (formerly known as Vibrio sp. Ex25), was a highly abundant member of the Vibrio community in both culture-dependent and independent analyses. This species is closely related to V. parahaemolyticus and V. alginolyticus and was originally isolated from deep-sea hydrothermal vents [45]. Base on genome sequencing, its predicted to possesses both the functional potential to survive in extreme conditions and factors potentially involved in human disease caused by coastal Vibrio spp.
More recent studies have identified isolates in heat-shocked oysters, confirming that it inhabits diverse environments [46], though given its similarity to other Vibrio species its taxonomic designation is still being explored [47]. In our study, this species co-occurred with highly abundant pathogenic Vibrio spp., and while the ecological role and pathogenicity potential of V. antiquaries is unknown, its close phylogenetic relationship to and copresence with the pathogenic species at these sites suggests they may be interacting and potentially even horizontally sharing genes.
Pathogenic Vibrio spp. are commonly associated with prokaryotic and eukaryotic community members, including chitin producers Our study elucidates links between three pathogenic Vibrio species, the environment, and the planktonic community. In conducting these analyses, it is clear that taxonomic resolution plays an important role in defining potential functionally relevant relationships, and in establishing an ecologically relevant context for prior and future studies. Using Vibrio spp. abundance to assess community interactions, ecological niche, and even health risk is a common practice. However, these results can be conflicting or misleading, unsurprisingly since Vibrio spp. occupy distinct ecological niches and possess unique physiological capabilities, including virulence mechanisms and modes of infection. This applies to community associations as well; demonstrating the potential importance of species-level associations, Turner et al. [16,48] found that while total Vibrio spp. bacteria were negatively correlated with copepods in a particular size fraction (63- Alphaproteobacteria) (Fig. 4A, Fig. 5A). Notably, we observed associations between members of the class Oxyphotobacteria (Cyanobacteria), which was negatively associated with V. vulnificus, V. cholerae, and pilF (Fig. 5A). Jesser and Noble 2018 also found a negative association between Cyanobacteria and V. vulnificus using a comparative relative abundance approach. In our dataset only one ASV, a Prochlorothrix sp., was among the 20 most abundant and appeared to drive this pattern. Multiple less abundant ASVs had more variable associations, as revealed by LEfSe, which may guide future studies. A prior laboratory study investigated the response of Synechococcus sp. WH8102 to co-culture with V. parahaemolyticus, finding significant transcriptional changes including evidence of possible phosphate stress and utilization of specific nitrogen sources [49]. While we didn't observe this organismal pairing in our dataset, future transcriptomic studies may identify similarly important Vibrio-cyanobacterial interactions based on our dataset.
A primary focus of our study was assessing pathogenic Vibrio spp. interactions with eukaryotes in the community, including those that produce chitin and organisms crucial to ecosystem function such as primary producers and grazers which are frequently overlooked in environmental microbiome studies. Of particular interest are diatoms (class Bacillariophyta) and copepods (class Arthropoda) as both of these groups are capable of chitin production and have been shown to interact with pathogenic Vibrio spp. in laboratory studies. For most known Vibrio spp., chitin serves as a nutrient source and a substrate for biofilm formation and subsequent protection from environmental stressors and predation [12,50]. It also induces a well-studied suite of cellular interactions initiating bacterial competition via the Type VI secretion system (T6SS) and natural competence, which may be the mechanism for how non-virulent populations become virulent (Meibom, 2005 The most abundant eukaryotic organisms found in our samples (> 28% of 18S sequences) were diatoms, which in prior studies were frequently associated with high Vibrio spp.
concentrations [16,55,56] but are often analyzed as a single group or under the "bulk" algae category even though individual species host distinct microbial communities [20], release unique dissolved organic matter substrates [57][58][59], have variable susceptibility to viral or bacterial infection [60-63], or exude chitin. In our study the class Bacillariophyta was positively associated with V. parahaemolyticus, temperature, salinity, and chlorophyll a (Fig. 5C). However, looking at the more resolved genus level, this affect appears to be driven by the most abundant diatom genus Chaetoceros (Additional File 6B). In particular, two ASVs most closely related to Chaetoceros pumilus comprised the majority of Chaetoceros diatoms (Fig. 5D). Though many diatom genera contain chitin synthesis genes or full pathways, and may potentially produce chitin as a component of the cell wall [64,65] (though to the best of our knowledge this hasn't yet been reported), only two have been shown to date to actually exude chitin: Thalassiosira and Cyclotella [ 18]. These chitin producing diatoms were highly abundant in our samples, with 3 ASVs among the most abundant 20 eukaryotic taxa (Fig. 5D). Notably, they exhibited ASV-specific relationships with the pathogenic attaching to the chitin-producing diatom T. weissflogii, suggesting an unexplored mechanism of environmental persistence [15]. In addition to revealing new environmental associations, our findings establish an ecologically relevant foundation for studying interactions between chitin-producing diatoms, pathogenic Vibrio spp., and the environment.
The interaction between pathogenic Vibrio spp. and planktonic copepods is an important, well-studied coastal phenomenon with demonstrated human health implications. Huq et al. 1983 found that V. cholerae 01 and non-01 serovars attached to living but not dead Acartia tonsa, Eurytemora affinis, and Scottolana spp. copepods from natural samples [9]. Another laboratory study investigating these same living copepod species found that V. cholerae preferentially attached to Acartia tonsa copepods over Eurytemora affinis, and that individual V. cholerae strains exhibited different attachment efficiencies [67]. In contrast, an O1 V. cholerae serovar (strain N16961) and two non-O1/O139 V. cholerae isolates, were found to preferentially attach to dead, rather than living, Tigriopus californicus copepods, as well as dinoflagellates [11]. It is unclear whether this difference is due to experimental methodology, the copepod species, or the Vibrio strains.
Environmental studies accounting for copepod taxonomy are rare and inconclusive: one found no association between V. cholerae and co-occurring Diaptomus and Cyclops genera copepods [68], and while others have reported qualitative associations in field samples [69,70] relationships based on quantitative data and copepod specificity are rare and consequently poorly defined.
We observed positive correlations between pathogenic Vibrio spp., particularly those found in lower salinities, and several copepod genera (Additional File 6D). The most abundant copepod was identified as Pseudodiaptamus inopinus, an invasive species originating in Asia [71,72], which was not significantly associated with any Vibrio species across all samples but was highly abundant during the months where the highest levels of V. cholerae and V. vulnificus were detected at LPL and the SDR sites (Fig. 4B, Fig. 5D).
Other abundant copepods were the Harpacticoid genera Canuella and Tigriopus, both positively associated with V. vulnificus and the virulence-associated gene pilF (Fig. 5C, D).
In laboratory studies, the type IV pilus (containing the pilF subunit) has been shown to be involved in chitin attachment to Vibrio spp. [12]. Tigriopus was also positively associated with V. cholerae. Though we could not obtain species-specific taxonomic resolution for Tigriopus, it is a well-established laboratory model genus with gene-silencing capabilities and full or partially assembled genomes for several species [73][74][75]. Thus, Tigriopus and Canuella spp. may be good candidate genera for future laboratory studies involving ecologically relevant Vibrio-plankton interactions.

Conclusions
Our study quantifies for the first time potentially pathogenic Vibrio spp. in Southern California while using a metabarcoding approach to identify community diversity and potential interactions. In addition to observing abundant populations of V. parahaemolyticus, V. vulnificus, and V. cholerae that conform to previously observed temperature and salinity niches, culture-dependent and culture-independent sequencing approaches revealed diverse Vibrio communities including additional potentially pathogenic species. High abundances in previously unstudied areas with high potential for human exposure, along with the detection of multiple genes associated with human infection, suggest that future sampling and risk modelling for these areas may be appropriate.
In characterizing the microbial and eukaryotic communities co-occurring with these individual Vibrio species, we identified ASV-specific relationships with potential functional  (Fig. 1A-D). For intra-site comparisons, two different sites at SDR (SDR1 and SDR2) and TJ (TJ1 and TJ2) were sampled, totaling 5 sampling sites.

22
Temperature and salinity were measured between 12 pm and 1 pm using a YSI Pro 30 field instrument (YSI Inc.). Unfiltered water samples were collected in 4 L opaque bottles and processed in lab beginning no more than 2 hours after collection. These samples were kept in a cool area at roughly room temperature rather than at 4 °C to prevent a viable but non-culturable (VBNC) state in Vibrio bacteria [76].
Water samples were gently filtered and flash-frozen in the lab for downstream processing.
For chlorophyll a quantification, 10-100 mL samples were collected on GF/F filters (Whatman) and stored at -20 °C. Samples were later extracted in 90% acetone overnight and measured on a 10AU fluorometer (Turner), followed by addition of HCL and remeasurement to account for the chlorophyll a degradation product pheophytin [77]. Initial sample lysis buffer resuspension and vortexing was completed manually, the remainder using an epMotion liquid handling system (Eppendorf). RNA was reversetranscribed into cDNA using the SuperScript III First-strand cDNA Synthesis System (Invitrogen). gDNA was quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen) and RNA using the Quant-iT RiboGreen RNA Assay Kit. Nucleic acid integrity was confirmed using an Agilent 2200 TapeStation (Agilent). Duplicate filters were extracted for all RNA and DNA samples with the exception of SDR1 December and April, for which two RNA but only one DNA sample was extracted. Genomic DNA was extracted from Vibrio isolate pellets using a DNeasy Blood and Tissue Kit (Qiagen), with subsequent quantification and quality control as described above.
Vibrio digital droplet and end-point PCR Select pathogenic Vibrio species and virulence genes were quantified using the QX200 digital droplet PCR (ddPCR) System (BioRad), following the manufacturer's protocols and recommended reagents. Previously published assays based on qPCR were optimized for ddPCR, including running temperature gradients for each target to establish optimum reaction temperature. Results from technical replicates were merged for analysis, and more than 19,000 droplets were measured per sample. Target-specific gBlocks (Integrated DNA Technologies) were used as positive controls for all ddPCR and end-point PCR targets.
Single copy-number gene targets for the species V. parahaemolyticus, V. vulnificus, and V.
cholerae were quantified and used to approximate cell number per 100 mL of sample (Additional File 1). We targeted toxR for V. parahaemolyticus [ 23], vvhA for V. vulnificus [24], and ompW for V. cholerae [ 25]. We also quantified the virulence-associated V.

Amplicon library construction and sequencing
Amplicon libraries were constructed and sequenced using cDNA or DNA template for whole community and Vibrio "isolate community" samples to characterize composition of multiple co-occurring communities. RNA template was used to characterize biologically active community members using 16S and 18S amplicons, while DNA was used to 24 characterize vibrio-specific diversity using the HSP60 amplicon. The 16S rRNA gene small subunit (SSU-rRNA) V4-5 region was targeted to characterize the prokaryotic bacterial and archaeal community using primers 515F-926R [78]. The V9 region of the 18S rRNA gene was targeted for eukaryotic community composition using primers 1389F and 1510R [79].
The universal regions of heat shock protein 60 (HSP60), also known as chaperonin 60 (cpn60), sequence was amplified and sequenced as described in Jesser and Noble 2018 using primers identified in previous studies [42,80] Demultiplexed sequences were analyzed using the Qiime2 [82] pipeline and additional analyses and visualizations were conducted using the R package phyloseq [83] and the web-based tool MicrobiomeAnalyst [84]. Sequences were quality filtered, chimeric sequences were removed, and exact amplicon sequence variants (ASVs) [85] were defined using dada2 [86] with a maximum expected error cutoff rate of 2 for 16S and 18S rRNA gene amplicons and 5 for the HSP60 amplicon. For 16S and 18S amplicons, replicate samples were merged using the "qiime feature-table group" function. Taxonomy was assigned using Silva [87] version 132 for bacterial and archaeal 16S sequences, and PR2 [88]  Authors' contributions: RED, AEA, JAS, and JFG were responsible for the conception and design of the study. RED, HZ, and AR were responsible for acquiring and analyzing study data, RED and AEA were responsible for data interpretation.