Fieldwork and sampling were conducted at the coastal experimental station of Shantou University, located on Nan’ao Island (23.48°N, 117.11°E, Figure 1 ), Guangdong, China: Shantou. Nan’ao Island features multiple bays with typical tropical and subtropical characteristics. Its surrounding waters exhibit complex hydrological conditions (Cai et al., 2011; Shu et al., 2018) due to the influence of surface runoff from the Hanjiang and Huanggang Rivers and upwelling from the Taiwan Shoal, which usually occurs in summer (June to September) (Jiang & Wang, 2018; Huang et al., 2021). These unique current conditions promote rich marine life. Enrichment experiments were conducted in autumn (November 2020) and spring (April 2021) to capture diverse initial ecological conditions.

Surface water samples (1 m depth) for enrichment experiments were collected using Niskin metal-free bottles. Chlorophyll α (Chl α), B vitamins, and microbial plankton community samples were obtained at the beginning (day 0). Portable hand-held water quality meters (HACH, USA) measured physicochemical indices such as pH, DO, temperature, and salinity. Subsequently, 300 mL of seawater was filtered through a 200 μm mesh to remove larger zooplankton and debris and added to photosynthetically active radiation (PAR) and ultraviolet radiation (UVR) transparent, sterile, and nontoxic (Whirl-Pak) bags along with nutrients or B vitamins, as shown in Table 1 . The bags and nutrient concentrations were consistent with those used in previous enrichment experiments (Martínez-García et al., 2010; Joglar et al., 2020). The added amounts of B₁, B₂, B₆, and B₁₂ were approximated to the maximum concentrations observed in earlier studies (Sañudo-Wilhelmy et al., 2006; Sañudo-Wilhelmy et al., 2012; Heal et al., 2014). Each treatment had five replicates. Experimental bags were incubated in situ in coastal floating cages at a depth of 1 m for 5 days. Untreated groups served as control treatments (C) alongside B vitamin addition treatments, while inorganic nutrient addition was set as another control treatment (I) with inorganic nutrients and B vitamin addition treatments.

The Chl α concentration was measured initially and after a 5-day incubation. Seawater samples (300 mL) were sequentially filtered through 3 and 0.2 μm Millipore polycarbonate (PC) filters, which were then frozen at -20 °C until analysis. Chl α extraction involved 90% acetone and 5 minutes of vortexing, followed by overnight storage in darkness at 4 °C. Finally, samples were analyzed using a spectrophotometer (General Administration of Quality Supervision et al., 2007, China).

Samples were filtered through 0.22 μm polycarbonate filters, immediately frozen in liquid nitrogen, and stored at -80 °C until analysis. Phytoplankton and zooplankton communities were assessed by sequencing the entire 18S rRNA gene (18S rDNA). DNA extraction was performed using the PowerSoil DNA Isolation Kit (MoBio Laboratories Inc., CA, USA) following the kit’s instructions. Eukaryotes were amplified using the primers “Euk-A: AACCTGGTTGATCCTGCCAGT and Euk-B: GATCCTTCTGCAGGTTCACCTAC”(Countway et al., 2005). The KOD One PCR Master Mix (TOYOBOLife Science) was used to perform 25 cycles of PCR amplification, with initial denaturation at 95°C for 5 min, followed by 25 cycles of denaturation at 95°C for 30 s, annealing at 50°C for 30 s, and extension at 72°C for 1 min, and a final step at 72°C for 7 min. Amplified regions were sequenced with the PacBio platform at Biomarker Technologies (Beijing, China). CCS sequences were derived from the original data, and chimeras were removed to obtain effective CCS. Effective CCS sequences were clustered/denoised, OTUs/ASVs (features) were classified, and species classification was obtained according to feature sequences. The coverages of the samples ranged from 0.9936 to 0.9998. Finally, the SILVA reference database was used for taxonomic assignment of 18S ASVs through the Naive Bayes classifier combination comparison method.

Response ratios (RRs) were calculated to intuitively illustrate the responses of phytoplankton and bacteria to B vitamin supplementation (Barber-Lluch et al., 2019; Joglar et al., 2020). Phytoplankton and bacterial biomass of B vitamin amendment treatments were divided by control treatment values, and when combined with inorganic nutrients and B vitamins, treatments were divided by inorganic nutrient (I) treatment values. RR values >1 indicate a positive response, RR values = 1 imply no response, and RR values <1 signify a negative response; the further RR values deviate from 1, the more pronounced the response. One-way ANOVA and t-tests were employed to test significant differences between treatments. Principal coordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) were used to display community composition differences across seasons at the OTU level. Pearson/Spearman correlation analysis and redundancy analysis (RDA) were employed to assess correlations between specific copepods and phytoplankton. Detrended correspondence analysis (DCA) for biological data was applied to determine whether to use linear or unimodal ordination methods. Correlation networks were constructed using OmicStudio tools at https://www.omicstudio.cn/tool. All statistical analyses were considered significant at the 0.05 significance level.

The initial environmental conditions in November and April were compared to assess changes in hydrographic and biochemical conditions during the in situ amendment experiments ( Table 4 ). Differences in pH, temperature, dissolved oxygen, and salinity were observed between the two months, with November having slightly lower temperature salinity. Phytoplankton biomass was 1.16 times higher in April, particularly for smaller phytoplankton, which were 2.44 times higher in April. The B vitamin content in the experimental field exhibited seasonal variation, with B₁ and B₂ concentrations 37.08% and 80.14% lower in November than in April, while B₆ and B₁₂ concentrations were 3.03 times and 5.92 times higher in November, respectively.

The total chlorophyll α (Chl α) was measured to assess phytoplankton biomass ( Figure 2A ). After the end of the experiments, the phytoplankton biomass of all treatments increased significantly in November compared with the initial biomass (T0), and the total Chl α ranged from 12.10-29.52 μg L^-1, while that in April was only 0.64-4.58 μg L^-1. In 87.5% of the eight treatments supplemented with B vitamins, total Chl α showed significant positive responses compared to the control treatment (ANOVA, p < 0.05). In terms of response rate (RR) ( Figure 2B ), the range of positive RR of treatments in November was 1.22-2.08, while in April, the highest was as high as 7.12, and the average positive RR was 3.75. The positive response ratios in April were 1.54-8.96 times higher than those in November ( Figure 2B ). Only 50% of treatments comprising a mixture of B vitamins and inorganic nutrients had a slight secondary response, and there were IB₁ and IB₆ treatments with RRs slightly greater than 1. The response rates of IB₂ and IB₁₂ were less than 1, indicating that there was no secondary response of phytoplankton to vitamin B in these two treatments.

Analysis of different particle sizes of phytoplankton revealed that smaller phytoplankton (0.22-3 μm) exhibited strong positive responses to B vitamins in November treatments and mixed inorganic nutrients and B vitamins treatments in April, regardless of inorganic nutrient levels. The average Chl α (0.22-3 μm) RR of the treatments supplemented with B vitamins reached 4.27 in November, and that of the mixed inorganic nutrients and B vitamins treatments in April was 3.83 ( Figure 2C ). Conversely, larger phytoplankton (> 3 μm) showed stronger positive responses in the April than in the November experiments, with a maximum Chl α (> 3 μm) RR of 12.18 in April ( Figure 2D ). These findings suggest that B₁, B₂, B₆, and B₁₂ vitamins may ultimately limit phytoplankton growth in the marine environment.

Community composition analysis revealed the impact of vitamin B supplementation on phytoplankton communities. The initial community compositions differed between November and April, with diatoms dominating in November and Picochlorum dominating in April. In both seasons, B vitamin supplemented treatments showed increases in the relative abundance of several phytoplankton compared to the control treatments, such as Thalassiosirales, Odontella in B₆ and B₁₂ treatments, Prymnesiales, Thalassiosirales and Ditylum in B₁ treatment, Minutocellus in B₂ treatment in November ( Figure 3A ), and then Picochlorum in both of the B vitamin supplemented treatments in April ( Figure 3B ). Principal coordinate analysis (PCoA) and nonmetric multidimensional scaling (NMDS) based on Bray−Curtis distance at the endpoint of each experiment confirmed that the community composition of phytoplankton in each treatment was obviously different from that in the control treatment after B vitamin supplementation ( Figures 4A-1, 2 ). The points of each B vitamin treatment in PCoA and NMDS were far from the points represented by the control treatment. These results demonstrate that B vitamins have a substantial impact on phytoplankton community composition.

The initial community compositions of the identified zooplankton were similar in the two experimental seasons, with Calanoida (Copepoda) predominating, which was not the case with phytoplankton. In addition, their relative abundances were at low levels, at 0.2 in November and less than 0.02 in April. However, compared with the C treatment, the relative abundance of identified zooplankton in 87.5% of the treatments supplemented with B vitamins had negative responses ( Figure 3 ). Calanoida showed a decrease in the B₁ and B₆ treatments, which decreased to almost undetectable levels in November. The decline also occurred in April, when the relative abundance of Cyclopoida and Paracalanus sharply decreased in all B vitamin treatments, and the maximum reduction was in the B₁ treatment, decreasing to less than 0.02, while that of the C treatment was greater than 0.8.

We also used PCoA and NMDS analysis to show the differences in identified zooplankton communities between experiments ( Figures 4B-1, 2 ). The first principal coordinate explained 53.08% and 58.47% of the total variation in November and April, respectively, and the second principal coordinate explained 25.53% and 26.28% of the total variation in November and April, respectively. Both stress values of the NMDS analysis were less than 0.01. Accordingly, significant differences between B vitamin treatments and C treatments were observed. Besides, we used RDA analysis to show the relationship between specific phytoplankton and zooplankton as the first axis lengths of gradient in the DCA analysis is less than 3.0. The RDA results showed that different patterns of response in phytoplankton and zooplankton to B-vitamin amendments appeared to be mostly explained by the certain phytoplankton and zooplankton ( Figure 5 ). In details, Calanoida dominated the plankton community changes in the vitamin B₂ treatments, while Thalassiosirales, Ditylum, Odontella, Chaetoceros dominated the plankton community changes in the vitamin B₁, B₆, and B₁₂ treatments in November. The changes of plankton community composition in the B vitamins treatments in April were mostly dominated by Paracalanus, Protodinium and Picochlorum. At the same time different correlations were observed between different species of zooplankton and phytoplankton, most phytoplankton species showed negative correlation with zooplankton, such as Calanoida with Thalassiosirales, Ditylum, Odontella, Prymnesiales and Chaetoceros in November ( Figure 5A ), Paracalanus with Protodinium and Picochlorum in April ( Figure 5B ).

Our study highlights the importance of B vitamins, including B₂ and B₆, in addition to B₁ and B₁₂, in stimulating phytoplankton growth. These findings challenge the conventional belief that only B₁ and B₁₂ are significant to phytoplankton (Barber-Lluch et al., 2019; Joglar et al., 2020). Additionally, our results demonstrate distinct response patterns among phytoplankton of different sizes in relation to B vitamins. For example, since the inorganic nutrient content was higher in spring and lowest in autumn in the experimental area, the probability of nitrogen restriction in autumn is higher than that in spring (Ke et al., 2019; Chen et al., 2020). Therefore, smaller phytoplankton (0.22-3 μm) exhibited strong positive responses to B vitamins in the autumn when inorganic nutrition was relatively limited, suggesting that they may have an advantage in the absorption and assimilation of B₁, B₂, B₆, and B₁₂ under nutrient-limited conditions (Koch et al., 2012; Fridolfsson et al., 2019). In contrast, larger phytoplankton (> 3 μm) were mainly limited by inorganic nutrients under nutrient-limited conditions, exhibiting less pronounced responses to B vitamin supplementation. However, when inorganic nutrients were abundant, larger phytoplankton outcompeted smaller species, and vitamin B limitation emerged as the primary constraint (Gobler, 2007). The observed limitation of single B vitamins in spring implies that organic micronutrient reserves may be insufficient during more productive seasons (Barber-Lluch et al., 2019).

Previous studies have found that vitamin B₁ (Barber-Lluch et al., 2019; Joglar et al., 2020) and B₁₂ (Sañudo-Wilhelmy et al., 2006; Koch et al., 2011; Barber-Lluch et al., 2019) supplementation increased phytoplankton biomass. There is widespread auxotrophy for B vitamins among phytoplankton (Giovannoni, 2012; Sañudo-Wilhelmy et al., 2014). In fact, most phytoplankton (80%) are auxotrophic for B₁ (Croft et al., 2006; Paerl et al., 2015), and 50% of sequenced phytoplankton, including most diatoms, possess a pathway for B₁ synthesis (Sañudo-Wilhelmy et al., 2014). Our study observed an increased relative abundance of diatom species (Thalassiosirales, Chaetoceros, and Ditylum) and Prymnesiales following B₁ supplementation. However, eukaryotic phytoplankton are widely believed to be incapable of de novo B₁₂ synthesis (Bertrand and Allen, 2012; Helliwell, 2017), and more than half of marine phytoplankton require B₁₂, indicating widespread B₁₂ auxotrophy in the ocean (Tang et al., 2010). Chaetoceros, Ditylum, and Thalassiosirales have also been reported as B₁₂ auxotrophs (Croft et al., 2006), which accounts for the increase in the relative abundance of these species following B₁₂ supplementation observed in this study and another (Koch et al., 2011).

Limited information exists about B₂ and B₆ auxotrophy, but our study demonstrated changes in phytoplankton community composition following their supplementation. We hypothesize that Chaetoceros, Ditylum, and Thalassiosirales may also be B₆ auxotrophs, given the similarity of their relative abundance trends in the B₁ and B₁₂ treatments. In contrast, the relative abundance of Minutocellus decreased in most treatments compared to the control, which suggests that they might not have significant B vitamin dependence. The increase in Picochlorum abundance in April suggests an obvious stimulation by B vitamins, particularly B₁, B₆, and B₁₂. Picochlorum, a member of the Chlorophyta phylum, is a group of small, fast-growing nanoplanktonic algae known for high oil-producing potential and applications in wastewater remediation, biomass production, and aquaculture feedstock (Henley et al., 2004; De la Vega et al., 2011; Chen et al., 2012b; Zhu and Dunford, 2013; Wang et al., 2016; Watanabe and Fujii, 2016; Kumar et al., 2017). However, the B vitamin auxotrophy of Picochlorum remains unclear. Our findings highlight the need for further research on the relationship of Picochlorum with B vitamins to better understand their ecological role and potential applications.

Our study identified a decrease in the relative abundance of Copepoda, including Calanoida, Paracalanus, and Cyclopoida, following B vitamin amendments. A network diagram based on Spearman’s rank correlation analysis ( Figure 6 ) and redundancy analysis (RDA) ( Figure 5 ) revealed negative correlations between Copepoda and several phytoplankton species, such as Thalassiosirales, Chaetoceros, Ditylum, Picochlorum, and Protodinium.

The addition of B vitamins stimulated the abundance of specific diatoms (Odontella, Thalassiosirales, Ditylum, and Chaetoceros), which may be auxotrophic for B vitamins. These diatoms displayed significant negative correlations with Copepoda, with Spearman Rho values greater than 0.8 (p < 0.05 or 0.01) ( Figure 7 ). Copepoda were known to feed on phytoplankton. Fridolfsson et al. (2018) found in their study on marine vitamin B₁ that the presence of filamentous cyanobacteria could negatively affect copepods reproduction through phytoplankton and copepods vitamin B₁ content. Previous research has demonstrated that high concentrations of certain diatoms can inhibit Copepoda reproduction (Carotenuto et al., 2002; Miralto et al., 2003; Pierson et al., 2005), with two prevailing hypotheses: the “nutritional deficiency hypothesis”, suggesting that diatom nutrient composition is insufficient for Copepoda growth and reproduction (Irigoien et al., 2002), and the “toxicity hypothesis”, proposing that diatoms produce toxic secondary metabolites that directly affect Copepoda (Ianora et al., 2003). Thalassiosirales have been shown to produce mitogenic aldehydes at peak levels, which then affect copepods hatching and survival (Halsband-Lenk et al., 2005; Ask et al., 2006). At the same time, laboratory studies have reported negative effects of specific diatoms on Copepoda spawning and/or hatching (Ceballos and Ianora, 2003; Poulet et al., 2007; Dutz et al., 2008). Based on the above discussion, we believed that vitamin B induced changes in phytoplankton community composition may negatively affect Copepoda through food inhibition.

In addition to some certain diatoms, Protodinium and Picochlorum also showed significant negative correlations with Copepoda, with Spearman Rho values greater than 0.8 (p < 0.05) ( Figure 7 ). Previous studies have reported that Copepoda biomass did not increase with phytoplankton biomass in some coastal waters (Vadstein et al., 2004; Hong et al., 2013). Copepoda were known to exhibit adaptive food selection to meet their growth and reproduction requirements (Chen et al., 2012a), and their ingestion rate was linearly related to food concentration (Frost, 1972). Copepods increase their ingestion rate when food concentration is low, but when it exceeds a certain threshold, their feeding rate decreases. Copepoda typically preferred phytoplankton that were nutritious and within a suitable size range (Liu et al., 2010; Lee et al., 2012; Fridolfsson et al., 2019). Picochlorum has been reported to be too small for Copepoda to graze effectively, making top-down control of their bloom less likely (Ma et al., 2021). Thus, our study suggests that B vitamin-induced changes in phytoplankton community composition may also negatively affect Copepoda through feeding inhibition.