Ulva fasciata is a common wild species in southern China. In November 2022, U. fasciata was collected and transferred to a laboratory. The algae were washed several times with filtered seawater and placed in a light incubator containing artificial seawater (Berges et al., 2001) with 50 mmol/L (NO₃⁻:NH₄⁺ = 2:1) nitrogen. Artificial seawater was prepared following the method described by Berges et al. (2001), and all reagents were sourced from Sigma-Aldrich (St. Louis, MO, USA). U. fasciata was acclimated in artificial seawater for 1 week. The environmental parameters were as follows: temperature = 20 ± 1°C, light intensity = 800 lux (photoperiod L:D was 12 h:12 h), salinity = 30 PSU, and pH = 8.0.

Then, 20 g of U. fasciata was cultivated in nine closed plastic incubators of 28 × 27 × 18 cm. The nitrogen concentrations in each group in the three incubators were 10 (low concentration), 50 (control), and 200 mmol/L (high concentration). The nitrogen concentrations (10, 50, and 200 mmol/L) used in this experiment were selected based on ecological relevance and findings from previous studies on Ulva species. The control concentration of 50 mmol/L (NO₃⁻:NH₄⁺ = 2:1) was set following the protocol Berges et al. (2001) described, which simulates nitrogen levels commonly found in nutrient-enriched coastal waters. The low concentration of 10 mmol/L represents conditions with limited nitrogen availability, similar to those in oligotrophic environments, where nutrient limitation can affect algal growth (Li et al., 2019). The high concentration of 200 mmol/L was selected to simulate eutrophic conditions, where excessive nutrient input, especially nitrogen, can lead to algal blooms, such as green tides (Zhang et al., 2015). During the culture, the artificial seawater was replaced every 2 days. The experiment lasted 2 weeks in the same light incubator as the abovementioned conditions. Two grams of U. fasciata were sampled to determine the contents of soluble protein, sugar, phycocyanin, and Chlorophyll a (Schwenzfeier et al., 2011). Furthermore, seawater was taken to measure pH, salinity, dissolved oxygen, PO₄³⁻, NO₂⁻, NH₄⁺, NO₃⁻, and DSi₂⁺ (Demirtas et al., 2008). Samples were collected at 7-time points (days 0, 2, 4, 6, 8, 10, and 12) and stored at −80°C for DNA extraction.

Ulva fasciata was kept in a 50-mL centrifuge tube, and an appropriate amount of sterile seawater was subsequently added. The samples were treated with ultrasound (1 min, 50 kHz, 5 s/5 s) to isolate bacteria from the surface of U. fasciata. After removing the U. fasciata with sterile tweezers, the sample was centrifuged at 4000 g for 10 min to collect the cell pallet, which was then used for bacterial DNA extraction. Total genomic DNA was extracted using the CTAB method (Zhou et al., 1996). All chemicals used for the CTAB extraction were purchased from Thermo Fisher Scientific Inc. (Waltham, MA, USA), and the procedure was carried out using an Eppendorf 5804R centrifuge (Eppendorf, Berlin, Germany). We used the PRISM^® 7500 rapid real-time PCR System to amplify high-quality DNA because it can achieve precise and efficient amplification of DNA in real-time monitoring (Huang et al., 2014). The PCR reactions used the TaKaRa Ex Taq™ DNA polymerase (TaKaRa Bio Inc., Shiga, Japan). The extracted DNA was sent to Shanghai Meiji Co., Ltd. (Shanghai, China) for high-throughput 16S rRNA measurement and epiphyte determination. Illumina MiSeq sequencing and universal primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) were used to amplify the V3-V4 region of bacterial 16S rRNA gene.

After demultiplexing, sequence quality was assessed, with the removal of low-quality sequences using fastp (version 0.19.6) followed by merging with FLASH (version 1.2.11). The DADA2 plugin in the QIIME2 pipeline (version 2024) was then used with the recommended parameters for de-noising the sequences, yielding resolution at the single-nucleotide level according to the sample error profiles. The resulting amplicon sequence variants (ASVs) were annotated and classified using the SILVA 16S rRNA database (version 138) with the Vsearch consensus taxonomy classifier in QIIME2. To mitigate the impact of sequencing depth on diversity measures, the sequence count of individual samples was rarefied to 26,609, while maintaining an overall good coverage rate of 99.90%. Alpha diversity indices, including the observed ASVs and the Chao1, Shannon, and ACE indices, were calculated with Mothur (version 1.30.1). Similarities among the microbial communities in the different samples were assessed by principal coordinate analysis (PCoA) based on the Bray–Curtis dissimilarity using the vegan package in R (version 2.5–3). Monte Carlo permutation tests (9,999 permutations) were used for forward selection. In this analysis, the values of the x- and y-axes and the lengths of the corresponding arrows represented the significance of the different parameters in explaining taxon distributions across communities. Relationships within communities were explored by the construction of co-occurrence networks. These networks were analyzed using Cytoscape software, facilitating the visualization and exploration of complex interactions within the community. Spearman’s correlation coefficients were considered significant if above 0.6 or below −0.6, with p < 0.01.

The surface of U. fasciata was gently scraped with a sterile knife, and the scraped materials were sequentially diluted and spread on the marine agar plate 2216E and placed in an incubator at 25°C for 3–4 days. Single colonies were selected, streaked, purified, and inoculated in marine broth 2216E. PCR amplification of the 16S rDNA gene of these strains isolated from the surface of U. fasciata was performed using 16S-27F (3′-AGAGTTTGATCCTGGCTCAG-5′) and 16S-1492R (3′-TACGGCTACCTTGTTACGACTT-5′) primers (Mincer et al., 2002). Then, the samples were sent to Bioengineering Co., Ltd. (Shanghai, China) for sequencing. These strains were cultured in marine broth 2216E until they reached the logarithmic growth phase. The bacterial suspension was mixed with 40% sterilized glycerol in a 1:1 ratio for cryopreservation, ensuring the viability of the bacteria during freezing and long-term storage. A total of 228 pure strains were isolated from the surface of U. fasciata. PCR amplification of the 16S rRNA gene of the isolated strains from U. fasciata was performed using the primers 16S-27F (3′-AGAGTTTGATCCTGGC TCAG-5′) and 16S-1492R (3′-TACGGCTACCTTGTTACGACTT-5′), following the protocol described by Mincer et al. (2002). The amplified 16S rDNA samples were sent to Bioengineering Co., Ltd. (Shanghai, China) for sequencing to enable strain identification and phylogenetic analysis. Given the predominant relative abundance of Rhodobacteraceae and Flavobacteriaceae within the colony, along with their strong correlation as illustrated in the network diagram, these taxa were selected for constructing an evolutionary tree utilizing the adjacency method implemented in MEGA 7.0 software.

The N₂ fixed marker gene (nifH) was confirmed by PCR assay using the following primer pairs: nifH-F (AAAGGYGGWATC GGYAARTCCACCAC) and nifH-R (TTGTTSGCSGCRTACATSG CCATCAT) (Rösch et al., 2002). As indole-3-acetic acid (IAA) regulates plant growth and development, we also tested the strains’ capacity to produce IAA (Ahemad and Kibret, 2014). Salkowski’s colorimetric method was used to screen for IAA production by strains according to Bric et al.’s (1991) research. The strains were cultured in 50 mL of 2,216 E medium at 25°C for 3 days, and the bacterial culture medium was centrifuged at 10000 rpm for 10 min at 4°C. The cell pellet was washed twice with phosphate-buffered saline (PBS), supplemented with 1 g/L of L-tryptophan, and then oscillated in an oscillating incubator for 7 days. After centrifugation, 1 mL of the supernatant was added to 2 mL of Salkowski reagent (1 mL of 0.5 mol/ L FeCl₃ and 49 mL of 35% HClO₄). The mixture was incubated in the dark at 25°C for 30 min, and the absorbance was measured at 530 nm using a spectrophotometer. The concentration of IAA was estimated using the standard IAA curve. All IAA measurements were performed in triplicate. To determine phosphorus solubilizing activity, Ca₃(PO₄)₂ was utilized as an insoluble inorganic phosphate source. Members of the Rhodobacteraceae and Flavobacteriaceae families that could activate and solubilize phosphate were screened. The strain was incubated at 25°C for 3 days, and then, the phosphorus solubilization capacity was screened in a solid NBRIP medium. In a 250-mL conical flask, 50 mL of NBRIP medium was mixed with 100 μL of strain culture at neutral initial pH, then incubated for 7 days in a shaker at 25°C, and repeated inoculation three times. The culture was centrifuged at 5000 rpm for 20 min. The precipitation of P₂O₅ was determined by the molybdenum-blue method (Galhardo and Masini, 2000).

The changes in bacterial abundance on the surface of U. fasciata under different nitrogen concentrations and the essential nodes in the network indicated that Rhodobacteraceae and Flavobacteriaceae may be potential bacterial strains to improve nutrient absorption efficiency and growth of U. fasciata. Therefore, 29 strains of Rhodobacteraceae and Flavobacteriaceae were combined into SynCom1. Based on the above functional screening, 5 strains with the nifH gene, high phosphorus solubilization capacity, and high IAA yield were selected from 29 strains as candidates of SynCom2. SynCom was co-cultured with sterilized U. fasciata without bacteria on the surface without nitrogen.

The preparation method of U. fasciata without surface bacteria (WSB U. fasciata) was summarized as follows: U. fasciata was collected from the field and transported back to the laboratory on the same day. The surface of U. fasciata was thrice washed with sterile seawater and cleaned with ultrasonic (10 min, 50 W, 5 s/5 s) to remove bacteria. Antibiotics, including streptomycin sulfate (final concentration of 100 μg/mL), amphotericin (0.25 μg/mL), ampicillin (100 μg/mL), and imidazole (0.1 μg/mL), were added and incubated for 4 days. The exact amount of antibiotics was added again on the third day. Next, a sterile cotton swab was dipped in a multi-enzyme cleaning solution (3 M) and used to quickly scrub the surface of U. fasciata. Then, U. fasciata was rinsed with a large volume of sterile seawater to obtain WSB U. fasciata. The water from the last rinse was applied to the solid medium 2216E and cultured for 1 week. The absence of bacteria was indicative of WSB U. fasciata. SynCom strain was activated at 25°C. After 3 days of activation, the cell pellet was collected by centrifugation (at 4000 × g) and twice washed with sterile seawater. It was, after that, resuspended to a concentration of 10⁷ strains per mL. After mixing equal volumes (1:1 ratio), the final volume in each SynCom group was adjusted to 5 mL. The mixed bacteria were added into an airtight bottle and co-cultured with WSB U. fasciata in artificial seawater without nitrogen addition. The treatment of WSB U. fasciata without adding bacteria served as the control. On the last day of the experiment (day 7), the co-cultured U. fasciata was washed with a large amount of sterile seawater, and the surface strains of U. fasciata were obtained by ultrasonic shock, which was coated in marine medium 2216E. The growth of the strain in the medium was the same as that initially added, indicating that the co-culture was unpolluted. During the experiment, each SynCom group was repeated at least three times.

To assess the effects of SynCom on the growth of U. fasciata, samples of U. fasciata with similar growth and size were collected from the same batch using a multi-spot sampling method. Surface moisture was absorbed using sterile filter paper, and the samples were weighed for wet weight. Three samples were dried in an oven at 60°C until they reached a constant dry weight, while the remaining six samples were sterilized for the co-culture experiment. After 1 week of co-culturing SynCom with WSB U. fasciata, the wet and dry weights of the U. fasciata samples were measured. The contents of soluble sugar, protein, phycocyanin, and Chlorophyll a were also determined. Nitrogen content in the samples was calculated according to the method described by Farina et al. (1991), while phosphorus concentration was determined using phosphomolybdenum blue spectrophotometry and a UV–vis spectrophotometer.

About 0.5 g U. fasciata sample (wet weight) was homogenized in liquid nitrogen with 5 mL of 0.2 M HEPES buffer (pH 7.9). The homogenate was centrifuged for 20 min, and the supernatant was used to determine GS activity based on the method proposed by Slawky and Rodier (1988).

Total RNA of co-cultured U. fasciata was extracted by TRIzol reagent under low nitrogen conditions (Ma et al., 2011). The culture of uninoculated WSB U. fasciata was used as a negative control. DNA enzyme was used to remove possible DNA, and gel electrophoresis was used to confirm the DNA removal and ensure the RNA’s purity. RNA quantification was performed using a spectrophotometer, and RNA was used to synthesize the first strand of cDNA. The analysis was conducted in a 96-well plate using the SYBR Green method, and the U. fasciata β-actin gene served as an internal control. The effect of the synthetic community (SynCom) on U. fasciata at nitrogen-free concentration was confirmed by quantifying the expression levels of three key genes involved in photosynthesis (rbcL), lipid biosynthesis (accD), and growth hormone precursor (wrkY) pathways. The primer sequences used in this study are listed in Supplementary Table S1. Quantitative reverse transcription polymerase chain reaction (RT-qPCR) reactions were carried out at Each reaction was carried out in triplicate, and the 2^−△△CT formula was used to calculate the RT-qPCR amplification data.

A total of 2,741,338 high-quality sequences were obtained from U. fasciata samples (n = 54). The relative abundance of bacterial communities was analyzed at the genus level, and a Chi-square test was used (Figure 1). The results revealed that the bacterial community changed dramatically during the culture at low nitrogen concentration.

The relative abundance of the family Rhodobacteraceae (observed range: approximately 8.6–31.3%, p < 0.01) and Flavobacteriaceae (approximately 3.2–12.5%, p < 0.01) increased significantly, while the relative abundance of members of the families Pseudomonadaceae (decreasing from approximately 36.1 to 2.8%, p < 0.01) and Alcaligenaceae (decreasing from around 23.9 to 2.5%, p < 0.01) declined significantly during the growth of U. fasciata under low nitrogen levels. Under normal nitrogen levels, the relative abundance of Rhodobacteraceae (ranging from approximately 14.8 to 16.1%, p < 0.01) and Flavobacteriaceae (ranging from approximately 2.1 to 31.6%, p < 0.01) increased, whereas the relative abundance of Pseudomonadaceae (from 31.3 to 2.8%, p < 0.01) and Alcaligenaceae (from 25.4 to 1.3%, p < 0.01) showed a marked decrease. Similarly, when U. fasciata was cultured under high nitrogen concentrations, the relative abundance of Rhodobacteraceae increased from 17.5 to 18.7% (p < 0.01), and Flavobacteriaceae from 2.9 to 33.5% (p < 0.01). In contrast, the relative abundance of Pseudomonadaceae (decreasing from 27.2 to 18.9%, p < 0.01) and Alcaligenaceae (decreasing from 23.7 to 2.7%, p < 0.01) significantly declined. The relative abundance of Rhodobacteraceae likely reflects its direct involvement in nitrogen transformations, such as ammonification and nitrification, processes critical for regulating nitrogen availability within the ecosystem. Different forms of nitrogen, such as ammonium, nitrate, or organic nitrogen, might selectively influence the growth of Flavobacteriaceae, which could also be acting in concert with other nitrogen-cycling microbes to balance organic matter degradation and nutrient recycling. Under treatments with varying nitrogen levels, both microbial families may exhibit changes in abundance as a response to nitrogen limitation or surplus, adjusting their metabolic strategies to conserve or exploit available nitrogen sources for survival and growth (Figure 1).

In addition, the abundance of Rhodobacteraceae was elevated during the culture of U. fasciata with the low nitrogen concentration, while it decreased at a high N concentration. In addition, ANCOM statistical analysis revealed that common nitrogen-fixing bacteria, such as Methylophaga, and members of Flavobacteriaceae (e.g., Cellulophaga and Mesonia) and Rhodobacteraceae (e.g., Pseudophaeobacter), exhibited significant differences under varying nitrogen conditions (Figure 2). Moreover, the PCoA based on Bray–Curtis dissimilarity revealed that high concentration of nitrogen had a greater effect on the bacterial community than its low concentration (Figure 3). The PCoA presented in this study accurately reflects the variation in microbial communities (or relevant metric) across different nitrogen concentration treatments. The first principal coordinate (PC1) accounted for 43.38% of the variation, while the second principal coordinate (PC2) explained 18.56%. These values indicate that the first two components capture a significant proportion of the variation in the dataset. The observed separation between the groups treated with different nitrogen concentrations suggests nitrogen plays a substantial role in shaping the community structure. The permutational multivariate analysis of variance (PERMANOVA) was conducted to assess this separation’s significance statistically, confirming that the differences between nitrogen treatment groups were significant (p = 0.01). The correlation between nitrogen concentration and PC1 was also calculated, showing a strong positive correlation (r = 0.78), further supporting nitrogen’s influence on the community variation. Shannon index demonstrated that the diversity of bacterial community harboring the surface of U. fasciata cultured at low nitrogen concentration was similar to that at high concentration (Table 1).

RDA of environmental factors and microbiota was performed to identify the factors influencing changes in the bacterial community during the growth of U. fasciata (Figure 4). The results showed that the arrows representing NO₃⁻ were the longest compared with other influential factors, in which NO₃⁻ had a greater influence on the surface microbiota of U. fasciata cultured under different nitrogen concentrations. In this experiment, it was also found that DSi₂⁺ had a greater impact on most microbiota. The results revealed that nitrogen played a pivotal role in bacterial community changes.

The network was constructed to understand the relationship among bacterial communities during the growth of U. fasciata. The nodes in the network were the top 13 species on the surface of U. fasciata with the highest total abundance (Figure 5). The strong network presence of Rhodobacteraceae might indicate its importance in stress resilience and adaptability within fluctuating environmental conditions, possibly acting as a stabilizing force in dynamic ecosystems. The positive correlations associated with Flavobacteriaceae indicate potential relationships with other microbial taxa within the community, as detected by correlation analysis. The combination of both positive and negative correlations for Rhodobacteraceae, compared with the primarily positive correlations for Flavobacteriaceae, suggests that these families play complementary roles in shaping microbial diversity and supporting the community’s overall metabolic efficiency. The clustering coefficients of Rhodobacteraceae and Flavobacteriaceae in the network analysis were 0.76 and 0.50, respectively. The members of these families were determined as key nodes in the network analysis. Although this was not the highest in the network analysis, changes in bacterial abundance under different nitrogen concentrations (Figure 1) and the species of pure bacteria obtained were considered (Hu et al., 2016). Therefore, Rhodobacteraceae and Flavobacteriaceae served as the basis of synchronization, and their evolutionary relationship is illustrated in Figure 6.

As mentioned earlier, Rhodobacteraceae and Flavobacteriaceae were selected as potential nitrogen-fixing strains. A total of 29 strains belonging to the families Rhodobacteraceae and Flavobacteriaceae were isolated from the surface of U. fasciata. Among them, 12 strains had nifH, 15 produced IAA, and 10 acquired Pi from insoluble phosphate (Pi) (Table 2). Some strains had two or three of these potential functions. A total of five strains (including Ruegeria atlantica 11, Donghicola eburneus 5–35, Aliiroseovarius crassostreae Y11, Hyunsoonleella ulvae HU1-3, and Croceivirga radicis 6) with high IAA and Pi productivity and the presence of nifH genes were selected. Four (removing H. ulvae HU1-3) were combined into the SynCom2 and co-cultured with WSB U. fasciata to determine the effects of nitrogen fixation and plant growth promotion.

In SynCom, it is generally accepted that the combined function of several microbial strains is more effective than that of a single strain (Falkowski, 1997). Two SynCom systems were structured, and SynCom1 was based on the potential functions of members of Rhodobacteraceae and Flavobacteriaceae. SynCom2 was established using four strains with beneficial functions. These SynCom systems were first co-cultured with WSB U. fasciata to examine the physiological parameters, including soluble sugar, protein, phycocyanin and Chlorophyll a, wet weight, dry weight, and nitrogen and phosphorus contents. The results indicated that, compared to the control, the physiological parameters of U. fasciata—including soluble sugar, soluble protein, phycocyanin, Chlorophyll a, wet weight, dry weight, and nitrogen and phosphorus contents—changed within the following ranges: −8.4–10.1%, 1.6–16.8%, 23.9–49.2%, −6.7–18.3%, −6.0–5.1%, −6.6–10.1%, −11.7–7.5%, and −4.4–3.3%, respectively (Figures 7A–H). In addition, two SynCom systems significantly increased glutamine synthesis activity, which was almost undetectable in the control group (Figure 8A). The superiority of SynCom2 over SynCom1 in facilitating nitrogen and phosphorus acquisition is evident from the experimental data, where SynCom2 promoted more efficient nutrient uptake, enhanced U. fasciata growth, and increased nitrogen fixation compared to SynCom1. Moreover, SynCom2’s stability was demonstrated by the successful identification of all four strains after the experiment, whereas identifying specific species in SynCom1 was challenging due to varying growth rates of strains.

Since SynCom2 was highly correlated with the growth of U. fasciata, the U. fasciata co-cultured with SynCom2 was further analyzed by RT-qPCR to investigate the underlying mechanisms of several key genes associated with nitrogen fixation and plant growth. The expression levels of the three key genes involved in photosynthesis (rbcL), lipid biosynthesis (accD), and growth hormone precursor (wrkY) pathways were determined by adding U. fasciata to SynCom2 in nitrogen-free conditions. The increase in gene expression observed in SynCom2 compared to the control group refers to the upregulation of functional genes involved in nitrogen fixation, phosphorus acquisition, and other metabolic processes that support the growth of U. fasciata. These functional genes were activated to varying degrees in response to co-culturing with SynCom2, leading to enhanced nutrient uptake and growth efficiency. The expression levels of rbcL, accD, and wrkY were elevated by 1.04-, 11.21-, and 9.54-fold, respectively.